Modelling anthropogenic impacts on the growth of tropical rain forests

- using an individual-oriented forest growth model for analyses of logging and fragmentation in three case studies

Peter K¨ohler Der Andere Verlag, Osnabr¨uck, Germany, 2000 ISBN 3-934366-99-6 Zugl.: Kassel, Univ., Diss., 2000

Cover: Dawn in Danum Valley, Sabah (Borneo), Malaysia October 1997 taken by P. K¨ohler Modelling anthropogenic impacts on the growth of tropical rain forests

- using an individual-oriented forest growth model for analyses of logging and fragmentation in three case studies

Modellierung anthropogener Einflusse¨ auf das Wachstum tropischer Regenw¨alder - Analyse von Holznutzung und Fragmentierung in drei Fallstudien unter Verwendung eines individuen-orientierten Waldwachstumsmodells

Inaugural-Dissertation zur Erlangung des akademischen Grades eines Doktors der Naturwissenschaften (Dr. rer. nat.) im Fachbereich Physik der Universit¨at Gesamthochschule Kassel

vorgelegt von Dipl.-Phys.

Peter K¨ohler

aus Kassel Kassel, den 01.11.2000 Als Dissertation vom Fachbereich Physik angenommen am 01.11.2000

Erster Gutachter: Prof. Dr. Hartmut Bossel Zweiter Gutachter: Prof. Dr. Burkhart Fricke Drittes Mitglied der Promotionskommission: Dr. habil. Andreas Huth Viertes Mitglied der Promotionskommission: Prof. Dr. Albrecht Goldmann

Tag der m¨undlichen Pr¨ufung: 01.11.2000 ”If everyone complains bitterness, then to whom is the world?”

Zainal Abidin Jaafar

Overview

For answering questions concerning anthropogeneous impacts on tropical forest develop- ment the individual-oriented and process-based forest growth model Formind2.0 was developed. It simulates the spatio-temporal dynamics of uneven-aged mixed forest stands in areas of one hectare to several km2. The model describes forest dynamics as a mosaic of interacting forest patches of 20 m2×20 m2 in size. Within these patches trees are not spatial-explicitly distributed, and thus all compete for light and space following the gap model approach. Tree species diversity is aggregated into 5-20 plant functional types (PFT) on the basis of species maximum tree height and successional status. The carbon balance of each individual tree incl. photosynthesis and respiration is modelled explicitly. Thus, we can match measured diameter increment for different PFT, size and light condi- tions accurately. Allometric relationships connect above-ground biomass, stem diameter, tree height and crown dimensions. Beside increasing mortality through self-thinning in dense plots one of the main processes of mortality is gap creation by falling of large trees. This process as well as seed dispersal from mature trees interlinks neighbouring plots with each other. The model was parametrised for three different sites in South-East Asia and south- America: Sabah (Malaysia), Venezuela, and French Guiana. Model accuracy is tested with growth data from permanent sampling plots in Sabah. Sensitivity of various result variables on variation of most parameter values is tested and gives important insights into general model behaviour. Two examples of anthropogeneous impacts on tropical forest dynamics are management practise and fragmentation, both of major concern. Following applications are performed: Growth and yield of Venezuelan rain forest under various logging methods, intensities and cycles are analysed for their sustainability. Effects of logging (methods and cycles), fragmentation and recruitment assumptions on forest dynamics in Sabah are discussed. Finally, fragmentation impacts on mortality and recruitment are simulated and their effects on forest dynamic and biomass loss are evaluated for a forest site in French Guiana.

Keywords: abandoned land; basal area; dipterocarp forest; edge effects; forest growth model; Formind; fragmentation; French Guiana; functional groups; individual-oriented model; logging impacts; logging scenarios; Malaysia; maximum height; model; mortal- ity; plant functional types; rain forest; recruitment; secondary succession; simulation; successional status; sustainable timber harvest; tropical rain forest.

Contents

1 Introduction 15

2 An introduction to tropical rain forests 21

3 Concepts for the aggregation of tropical tree species into functional types and the application to Sabah’s lowland rain forests 25

4 The model Formind2.0 35

5 Comparison of measured and simulated growth on permanent plots in Sabah’s rain forests 51

6 Sustainable timber harvesting in Venezuela: a modelling approach 65

7 The effects of logging, fragmentation and recruitment on growth of di- pterocarp forest 85

8 Long-term response of tropical rain forests to the effects of fragmenta- tion: a simulation study 111

Summary 133

Zusammenfassung 139

Bibliography 145

AInventory data 165

B Lists of tree species 171

Danksagung 215 Account

Chapters of this thesis have been published as follows:

Chapter 3:

K¨ohler, P., Ditzer, T., & Huth, A. 2000b. Concepts for the aggregation of tropical tree species into functional types and the application on Sabah’s lowland rain forests. Journal of Tropical Ecology, 16(4), 591-602.

Chapter 5:

K¨ohler, P., Ditzer, T., Ong, R. C., & Huth, A. 2001. Comparison of mea- sured and simulated growth on permanent plots in Sabah’s rain forests. Forest Ecology and Management, 142(1-3), in press.

Chapter 6:

Kammesheidt, L., K¨ohler, P., & Huth, A. 2000. Sustainable timber harvest- ing in Venezuela: a modelling approach. Journal of Applied Ecology, in press.

Chapter 7:

K¨ohler, P., Ditzer, T., & Huth, A. 2000c. The effects of logging, fragmen- tation and recruitment on growth of dipterocarp forest. Journal of Ecology, submitted.

Chapter 8:

K¨ohler, P., Chave, J., Riera, B., & Huth, A. 2000a. Long-term response of tropical rain forests to the effects of fragmentation: a simulation study. To be submitted. Other publications by the author related to the topics of this thesis:

Ditzer, T., Glauner, R., F¨orster, M., K¨ohler, P., & Huth, A. 2000. The process-based stand growth model FORMIX3-Q applied in a GIS-environment for growth and yield analysis in a tropical rain forest. Tree Physiology, 20, 367–381.

K¨ohler, P. 1996. Ein individuenbasiertes Wachstumsmodell zur Simulation tropischer Regenw¨alder. Diploma thesis, University of Kassel, Germany.

K¨ohler, P. 1997. An individual based rain forest model: Formind. in Hahn-Schilling, B. (editor), Forest management with growth models. Malaysian-German Technical Cooperation Project, Forest Department of Sarawak, Malaysia, Kuching, Malaysia.

K¨ohler, P. 1998. Parameter research for the tropical rain forest growth model FORMIX4. Research report P9801, Center for Environmental Systems Research, University of Kassel, Germany.

K¨ohler, P. & Huth, A. 1998a. The effect of tree species grouping in tropical rain for- est modelling - Simulation with the individual based model Formind. Ecological Modelling, 109, 301–321.

K¨ohler, P. & Huth, A. 1998b. An individual based rain forest model - concepts and sim- ulation results. In: Kastner-Maresch, A., Kurth, W., Sonntag, M., & Breckling, B. (editors), Individual-based structural and functional models in ecology, number 52 in Bayreuther Forum Okologie.¨ Bayreuther Institut f¨ur Terrestrische Okosystem-¨ forschung, Bayreuth, 35–51.

Chapter 1

Introduction

Introduction Goldammer 1999; Laurance & Fearnside 1999; Nepstad et al. 1999; Hashimotio et al. The use of natural resources change our en- 2000). vironment directly and indirectly through With 36000 000 km 2 of forests, cov- effects which are not fully understood so far. ering a quarter of the total land sur- Climate change and an increase in mean face on earth, beside the oceans forests global temperature, the amount of carbon- are the biggest ecosystems on our planet. dioxide in the atmosphere, or rising sea lev- About 475 to 825 billion tons of carbon els are some examples of occurring effects are bound in the forests and thus they are (Fan et al. 1998; Chavez et al. 1999; Malhi & the biggest above-ground carbon storages Grace 2000). These anthropogeneous influ- (Murphy 1975; Enquete-Kommission 1990; ences will change our environment for cen- Grace et al. 1995; Fan et al. 1998; Pren- turies. Plants might react adaptivly to their tice & Lloyd 1998; Alexandrov et al. 1999b, changing environment (Pastor & Post 1988; 1999b). A further reduction of woodland Friend 1997; Cao & Woodward 1998; Tian and, following this, an increasing release of et al. 1998; DeLucia et al. 1999; Pounds carbon in the form of carbondioxide would et al. 1999; Stil et al. 1999; Hashimotio et al. certainly intensify climate changing effects. 2000). Huge ecosystems like forests might Currently, annual release of carbon and its buffer changes, caused for example by ex- input in the atmosphere are estimated at traction of timber woods over a long period. seven billion tons. About 20 % of the release But if natural catastrophes occur in ecosys- is caused by global deforestation (Enquete- tems, which have already been weakened, Kommission 1994; Houghton et al. 2000). damage is more dramatic than ever thought There are various reasons which make before (Phillips & Gentry 1994; Laurance forests worth protecting and sustainable et al. 1997; Phillips et al. 1998; Peres 1999; management desirable. Forests produce Gascon et al. 2000). Thus, forest fires in timber, firewood and food, act as lo- the Amazonian rain forest and the Indo- cal climate regulator, prevent erosion, and Malayan archipelago in the years 1997/1998 are important water storages. Addition- spread very fast. El Ni˜no, the Great South- ally, tropical rain forests are remarkable ern Oscillation, caused serious dry periods for their enormous biological species diver- (Leighton & Wirawan 1986; Walsh 1996; sity (Tuomisto et al. 1995; Hubbell 1997; Hartshorn & Bynum 1999), in which human Tilman 1999). It is assumed that 50-75 % made fires for land clearing could spread of all existing species are found in the tropi- easily to adjacent areas. These forest were cal wet forests (Terborgh 1993). In a survey already highly disturbed through logging to identify global extinction threats tropi- and forest management, and available dead cal regions were endangered the most (Sisk wood fuelled the fires further (Brown 1998; et al. 1994). For the conservation of as Kellman et al. 1998; Cochrane et al. 1999; 16 Chapter 1 many different species as possible the ap- detailed planning effort it might be possi- proach of biodiversity hotspots is proposed ble to manage tropical forests in a way that (Myers 1989; 1990; Reid 1998 Myers et al. ecosystems have a realistic chance for sus- 2000; Cincotta et al. 2000). Thus, areas tainable regeneration. On the global scale with highest diversity are considered most this is especially of interest, as case stud- valuable for protection. ies have shown that forests under reduced- Tropical rain forests covered roughly impact management will act as a sink for 18 000 000 km2 in the year 1990, corre- carbondioxide, compared to those under sponding to 13 % of earth’s land surface. commercial logging (Putz & Pinard 1993; Characteristics of tropical climate are a con- Pinard & Putz 1996, 1997). However, stant high temperature with monthly aver- the most important motivation for sus- ages above 18 ◦C and high precipitation (> tainable management are economic profit 100 mm per month) with no, or only short, on a short time scale (Plumptre 1996). dry seasons. Areas with those climatic Economic studies have verified increasing conditions are found in a belt of 40◦ lati- profit in well planned management (Bar- tude around the equator (Whitmore 1998). reto et al. 1998). Certification of trop- There are three regions where tropical rain ical timber wood is one of the possibil- forests occur: South and Central America, ities to support sustainable management central Africa and South-East Asia. strategies (Hahn-Schilling et al. 1994; Boot & Gullison 1995). Non-governmental or- Logging of timber, land clearing, slash- ganisations like the World Wildlife Found and-burn cultivation, high population pres- for Nature (WWF) are promoting this ap- sure and ongoing forest fragmentation are proach (Liedeker 1999; Forest-Stewardship- threatening tropical forests (Aiken & Leigh Council 2000). The controversial discussion 1993; Cannon et al. 1998; Brown 1998; about criteria of sustainability is certainly Grainger 1998; Foster et al. 1999; Rosen- not finished (Johns 1985, 1997; F¨olster et al. zweig 1999; Hughes et al. 2000). Reduc- 1986; Brown & Lugo 1990, 1994; Bruenig ing those impacts and changing to sus- 1996; Ong et al. 1996; Putz & Viana 1996; tainable development is needed to stop Weidelt 1996; Rice et al. 1997; Bowles et al. the extinction of various animal and plant 1998). species (Terborgh 1993; Laurance et al. 1997; Bowles et al. 1998; Whitmore 1998). For an estimate of annual allowable cuts The Food and Agriculture Organisation of (AAC) knowledge on the main processes the United Nations (FAO) estimated the an- of forest dynamics is essential. In tem- nual loss of tropical forests at 169 000 km2 perate forests, management planning was in 1990 with increasing tendency (Riswan based on modelling and computer simula- & Hartanti 1995; Laurance 1999b). If these tions over some decades (e.g. Botkin et al. trends of deforestation continue most trop- 1972; Shugart 1984, 1998; Battaglia & ical forest will be destroyed within the 21st Sands 1998; Borgesa & Hoganson 2000). century. Thus, long-term tendencies of forest dynam- ics can be estimated under changing condi- From an ecological point of view it might tions. be desirable to declare as many forest ar- eas as possible as protection zones with a Modelling of tropical rain forests started total ban of timber logging (Whitmore & in the early nineties with models of various Sayer 1992). Very often those idealistic con- complexity (e.g. Adlard et al. 1989; Alvarez- servation ideas conform with public opin- Buylla & Garcia-Barrios 1991, 1993; Bossel ion and policy making in developed coun- & Krieger 1991, 1994; Alder 1992; Howard tries, but neglect local needs for fuel, tim- & Valerio 1992; Vanclay 1994; Osho ber for construction and labour. With a 1995, 1996; Albers 1996; Alder & Silva Introduction 17

2000). Available data sets from long- (e.g. the model Formal (K¨urpick et al. term ecological research plots, (Nakashizuka 1997) needs the maximum tree age as an in- et al. 1999; Smithsonian-Tropical-Research- put parameter, which can be estimated only Institute 2000) and simulation studies in roughly) others come up with elegant math- the context of international climate protec- ematical equations (Kaspar 1996). Models, tion programmes (IGBP 1990) led to an in- which are not only used for testing ecologi- creasing interest in rain forest models in re- cal hypothesis, but for model application in cent years (Liu & Ashton 1998, 1999; Chave forestal management planning are of special 1999a, b; Pinard & Cropper 2000). Another interest (Ditzer 1999; Ditzer et al. 2000). motivation for developing models for trop- Problems of these approaches arise be- ical forest growth was management plan- cause they are mostly based on an aggre- ning, very often with the cooperation of Eu- gation of tree species in a few (3-5) species ropean developing projects and local gov- groups, which are easy to parametrise, but ernmental institutes, e.g. the British gov- simplify ecological processes of rain forests ernment in Indonesia (van Gardingen & greatly. Concepts of species grouping in Phillips 1999) or the German Gesellschaft tropical rain forests based on a more sys- f¨ur Technische Zusammenarbeit (gtz) in tematic approach were developed only re- Malaysia (Ong & Kleine 1995, 1996; Ditzer cently (Gitay et al. 1999; K¨ohler et al. 1999; Ditzer et al. 2000; Huth & Ditzer 2000b; Phillips et al. 2000). They are 2000a,b). The focus of these schemes was in the most important for all further work of South-East Asia, where deforestation rates this thesis. Furthermore, analysis in Kas- were highest in the last decade (Plumptre sel showed the largest potential for data 1996). Current rates of forest loss in Latin- based model development in individual- −1 America (7.4 million ha y )arenearly based models (Huston et al. 1988; DeAn- twice as high as those in Asia (3.9 million ha gelis & Gross 1992; Judson 1994; Liu & −1 y ) (FAO 1997). Modelling approaches de- Ashton 1995), as computation time was not pend on available field data used for model a limiting factor anymore, because of im- Symfor development. Thus, the model provements in computer capabilities. was developed in tight cooperation with log- Formind ging companies and for instant application The model developed by the in forest management planning and depends author in previous studies (K¨ohler 1996; mainly on inventory data collected by the K¨ohler & Huth 1998a, 1998b) was the ba- companies (Young & Muezelfeldt 1998; van sis of the further research and development Formind Gardingen & Phillips 1999), while Chave presented in this thesis. is an and colleagues (Chave et al., unpublished individual-based model, while the parallel Formix3-Q manuscript) are interested in long-term de- development of (Ditzer et al. velopment of rain forest migration and try 2000) is still based on a simple matrix ap- to understand seed dispersal patterns found proach, incl. transition rates between classes in paleoecological research. of different tree size. As result of the chosen approach, the model structure of Formind Various projects of the research group was more flexible and an application with Ecosystem Modelling at the Center of En- different numbers of species groups was easy vironmental Research, University of Kassel, to perform (K¨ohler & Huth 1998a). Germany, for the Deramakot Forest Reserve showed dependency of simulation results Two main targets are the focus of on the chosen modelling approach (Haupt this work. First, no existing model of 1995; Kaspar 1996; K¨ohler 1996; K¨urpick rain forests growth was applied to tropical 1 et al. 1997; Ditzer 1999; Huth 1999). While forests in various regions . This work tries some models are difficult to parametrise 1Ditzer (1999) was developing a concept of site 18 Chapter 1 to show that, with a model structure cov- only Chapters 2 and 4 have not been pub- ering all main processes, sites in various re- lished or submitted for publication. The gions can be simulated. Second, most pre- Chapters are arranged in chronological or- vious work was based on a forest recruit- der to allow a comprehensible understand- ment module covering only simple princi- ing of model improvement. ples. Ongoing forest fragmentation will de- An introduction to rain forest dynamics termine recruitment as one limitating factor is given in Chapter 2. A general approach in rain forest dynamics (Ribbens et al. 1994; to tree species grouping, based on avail- da Silva & Tabarelli 2000). Thus, beside able data sets, follows thereafter (Chap- general model improvement and enhance- ter 3, K¨ohler et al. 2000b). The model For- ment the development of new approaches mind2.0 used in this thesis is completely for modelling of recruitment is one of the described in Chapter 4. Chapter 5 consists main focuses of this work. The resulting of a validation of the model in its version Formind2.0 new model will be used to an- Formind1.1 with field data from Sabah swer various questions: (K¨ohler et al. 2001). Additional analysis of the same data with current versions of the 1. Is there a general approach for clas- model close this Chapter. Besides an inten- sifing several hundred tree species in sive sensitivity analysis of model behaviour, different rain forest sites into a few various logging methods and intensities in groups? a Venezuelan rain forest are analysed in 2. Does simulated tree growth match Chapter 6(Kammesheidt et al. 2000). An measured data sets with acceptable ac- application of the model to a rain forest site curacy? in Sabah (Malaysia) is performed in Chap- ter 7. The influence of various recruitment 3. Which logging method and rotation modules and their impacts on timber log- length can be called sustainable de- ging are analysed in detail (K¨ohler et al. pending on the forest site? 2000c). The model application to French Guiana contains an analysis of the effects 4. How does recruitment determine forest of forest fragmentation on further forest dy- growth and yield? namics (Chapter 8, K¨ohler et al. 2000a). 5. Can tropical rain forests buffer the ef- Finally, the methods and most important fects of ongoing fragmentation? results related to questions posed are sum- marised incl. an outlook. This summary is For this purpose three different rain written in both German and English. forest sites, in South-East Asia (Sabah, Data collections (inventory data and lists Malaysia) and South-America (Venezuela of tree species), which were needed for sim- and French Guiana), were parametrised ulations, are found in the Appendix. (Fig. 1.1)2 Besides the introduction, this thesis con- sists of seven further chapters from which depending parametrisation, but was restricted to dipterocarp lowland rain forests in South-East Asia. 2At the time of planning this research project a cooperation with a project in Kenya (Africa) ex- isted.Thus, it seemed possible to apply the model to all three global rain forest regions.Unfortu- nately the leader of the gtz-project was shot dead two days before cooperation started and the project was closed thereafter. Introduction 19

Caparo Venezuela

Deramakot Sabah (Malaysia) Piste de Saint-Elie French Guiana

Figure 1.1: Global distribution of forests, including research plots used in this thesis.The map is based on data collected between 1980 and 1990 (from WWF 1997).

Chapter 2

An introduction to tropical rain forests

In the following chapter some fundamen- is based on systematic combination of infor- tal characteristics of tropical rain forests are mation on temperature and water availabil- described. I focus on processes which are ity (Terborgh 1993; Shugart 1998). Cur- important to understand growth dynamics rently, a framework for a worldwide com- of the forest trees and their species compo- parison of tropical woody vegetation types sition. A more detailed description of the is developed (Blasco et al. 2000). ecology of tropical forests is found in sev- Classically, the term rain forest describes eral informative text books (Richards 1952, evergreen tropical lowland wet forest up to 1996; Whitmore 1984, 1993, 1998; Jacobs an elevation of 750 m. Those are closed 1988; Lieth & Werger 1989; Terborgh 1993; large growing forests found in latitudes be- MacKinnon et al. 1996; Huth 1999). The tween 10◦ north and 10◦ south with high objective of the current chapter is not to go precipitations without seasonal dry peri- into the details of the ecological processes, ods. Evergreen tropical wet forests covered but to explain some basic facts about the about 7 million km2 of land surface in 1993, ecology of tropical rain forests. Thus, the mainly in the Amazon-Orinoco area (Amer- following introduction will be rather brief. ican rain forest formation), at the Gulf of Guinea and in the water catchment of the Congo river (African rain forest formation), Evergreen lowland rain forest in Sri Lanka, Western India, Thailand, In- dochina, on the Philippines, in Malaysia, In- The tropics are mostly defined by their cli- donesia, New Guinea (Indo-Malaysian rain mate conditions. In tropical regions daily forest formation), and on the east coast of temperature fluctuations exceed average an- Australia (Enquete-Kommission 1994). nual temperature variability. Thus, trop- ical regions are extended north and south Lowland rain forests are by far the most of the equator until daily and annual tem- diverse plant communities on earth. Up to perature variability match each other (Lam- 400 different tree species are found in one precht 1986; Enquete-Kommission 1990). hectare (Terborgh 1993). The largest trees reach heights of 45 to 60 m, in a few cases up The most important site factors for vege- to 70 m. The tree crowns of those large in- tation formations are temperature, precipi- dividuals, called emergents, rise above the tation, light, and soil conditions. For an ex- closed forest canopy, which reaches about plicit differentiation of several tropical for- 30 m in height. Depending on light condi- est formations, climate, soil water, soil qual- tions small trees and saplings are found be- ity and elevation are considered (see Ta- low the canopy. Ground vegetation is rare ble 2.1). In Central America vegetation is in dense closed forests and consists mainly classified after a scheme of Holdridge, which 22 Chapter 2. An introduction to tropical rain forests

Table 2.1: Classification of tropical wet forests (from Whitmore 1998).

Climate Soil water Soils Elevation Forest formation

Seasonally Strong annual shortage Monsoon forests (various dry formations) Slight annual shortage Semi-evergreen forest Everwet Dryland Zonal Lowlands Lowland evergreen (perhumid) (mainly rain forest oxisols, (750) 1200-1500 m Lower montane rain forest ultisols) (600) 1500-3000 m Upper montane rain forest 3000 m to tree line Subalpine forest Podzolized Mostly lowlands Heath forest sands Limestone Mostly lowlands Forest over limestone Ultrabasic Mostly lowlands Forest over ultrabasics rocks Water Coastal salt- Beach vegetation, Man- table water grove forest, Brackish wa- high ter forest (at least Inland fresh- Oligotrophic Peat swamp forest periodi- water peats cally) Eutrophic ±Permanently wet Freshwater swamp forest (muck and mineral Periodically wet Freshwater periodic soils) swamp forest of recruitment of young trees. Shrubs and into two or three ecological classes (Denslow bushes are found in single areas (Whit- 1987; Whitmore 1998; Thomas & Bazzaz more 1998). This sort of layer structure 1999). Pioneers and climax species are the is controversially discussed in the literature extreme positions in a more or less contin- (Richards 1936; Terborgh & Petren 1991). uous spectrum. While pioneers establish While a model for light distribution in for- early in succession of available areas, climax est canopies tries to explain the structure or late successional species follow last in a (Terborgh 1993), new mathematical analy- succession process. Most important charac- sis of different vertical forest structures for teristics of pioneers and climax species are tropical and temperate regions found no dif- summarised in Table 2.2. ferences between them and only a few dis- Seeds of climax species germinate and es- tinct layers in both regions (Baker & Wilson tablish in the shade of the own mature com- 2000). munity. Therefore, they are called shade- All other rain forest formations differ tolerant. They are the dominant plant from this type through simpler structures, species in undisturbed primary forests and lower species diversity and a smaller spec- contribute mainly to the main canopy of trum of life forms. For means of simplicity a rain forest. The largest individuals nor- we address evergreen tropical lowland rain mally belong to those species (Whitmore forest by the short term rain forest. 1998). The second category are the pioneers. Their seeds depend on light and can only Tree species germinate and establish in forest gaps. Height growth is fast, and thus shade- Tree species in rain forests can be distin- tolerant competitors are suppressed. Pio- guished, after their successional behaviour, neer tree species are seldom found in ma- 23

Table 2.2: Most important characteristics of pioneer and climax species in tropical rain forests (from Whitmore 1998).

Pioneers Climax

Common alter- Light-demander, (shade-) intolerant, sec- Shade-bearer, (shade-) tolerant, primary native names ondary Germination Only in canopy gaps open to the sky Usually below canopy which receive some full sunlight Seedlings Cannot survive below canopy in shade, Can survive below canopy, forming a never found there ”seedling bank” Seeds Usually small, produced copiously and Often large, not copious, often produced more or less continuously, and from early annually or less frequently and only on in life trees that have (almost) reached full height Soil seed bank Many species Few species Dispersal By wind or animals, often for a consider- By diverse means, including gravity, able distance sometimes only a short distance Dormancy Capable of dormancy commonly abun- Often with no capacity for dormancy, sel- dant in forest soil as a seed bank dom found in soil seed bank Growth rate Carbon fixation rate, unit leaf rate, and these rates lower relative growth rates high Light compensa- High Low tion point Height growth Fast Often slow Branching Sparse, few orders Often copious, often several orders Leaf life Short, one generation present, viz.high Long, sometimes several generations turn-over rate present so slow turn-over rate Wood Usually pale, low density Variable, pale to very dark, low to high density Longevity Often short Sometimes very long

ture primary forests, but are most domi- dynamic equilibrium, gap creation and re- nant in secondary forest following regrowth growth within them balance each other of abandoned land, or in highly disturbed (Shugart 1984, 1998; Brokaw 1985; Brokaw forests after logging or catastrophic events. & Scheiner 1989; Platt & Strong 1989; Bel- The canopy of those forests is not closed and sky & Conham 1994). light demanding plants dominate the sites. Gaps are first filled with pioneers. In Tree species of medium characteristics, a second growth cycle, climax seedlings es- called mid successional species, are also dis- tablish themselves underneath the pioneers. tinguished. They are neither pioneer, nor After the death of the short-living pioneer climax (Whitmore 1998). species, established climax trees grow and fill the gap. It takes between several decades and some centuries until trees of sizes simi- Succession and gap dynamics lar to mature forest dominate those former gap areas (Whitmore 1998). A forest gap is a not-fully-closed canopy within a mature forest. Gaps are created This growth cycle is called succession. It by the falling of large trees, often causing is essential for the simultaneous coexistence the destruction of several other, i.e. smaller of tree species with different successional trees. As mature forest stands are in a behaviour in forests. 24 Chapter 2. An introduction to tropical rain forests

In tropical rain forests annual mortal- show that even in areas with dry seasons ity rates of trees with a diameter ≥10 cm of a few months some small, but effective are about 1-3 % (Swaine 1989; Phillips & model improvements will lead to acceptable Gentry 1994). Mortality rates cover dead results (Chapter 6). It should be mentioned standing trees, fallen individuals and those that dry periods as caused regularly by the which were smashed by collapsing trees. Great Southern Oscillation, El Ni˜no, will The causes of tree falls are wind, heavy rain result in significantly higher tree mortality fall and others. Field data show that an- rates (Walsh 1996). nually up to 1.5 % of standing trees fall Soil investigations show that two thirds over and thus 90 % of mortality is connected of all tropical soils have average to very low with gap creating events (van der Meer & fertility. Generally, agriculture can only be Bongers 1996). performed for a very short period of a few In the literature the definition of a forest years before soils become infertile. It has gap is widely discussed (Vandermeer 1994; been shown that above-ground growth of van der Meer et al. 1994). For example, tropical forests depends little on soil fer- Brokaw (1982) defined a gap as a missing tility (F¨olster et al. 1986; Terborgh 1993). canopy, which reaches down to 2 m above Endemic species are very well adapted to the forest floor. Others (van der Meer & nutrient-poor conditions. Plant growth de- Bongers 1996) define it as a canopy gap pends on very effective and fast decompo- reaching down to 20 m above the floor. For sition processes in the top soil layer. Most comparing field studies, gap definition is nutrients are bound in the living biomass, crucial. Thus, the range of gap numbers and and only about 20 % are depleted and reen- gap sizes varies widely (Barden 1989; Run- ter through precipitation and mineral rock kle 1989). With the second definition given erosion. Heavy disturbances of those cycles above, a neo-tropical rain forest in Panama through clearing, erosion or damage of the would have a gap fraction of 34 % (Hubbell humus layer might lead to massive nutrient & Foster 1986a). depletion. Thus, in these soils forests might Disturbance of forests by gap creation not grow to their former complexity and size can be distinguished in three different areas. (Terborgh 1993). In the region of the roots of the falling tree, Dependence of forest dynamics on soil the forest floor is damaged. Light intensity conditions and slope was analysed in other is increased through the missing tree crown studies (Biehounek 1999; Clark et al. 1999a; above. Beside the trunk of the falling tree Ditzer 1999; Glauner 1999; Ditzer et al. the disturbance is weak. The crown of the 2000), and is not the subject of the current falling tree destroys most trees, especially in thesis. We assume in the following more or the understorey (Hubbell & Foster 1986a). less undisturbed nutrient cycles. Investiga- tions of nutrient inputs through air and rain on Borneo support this approach (Weidelt Water and nutrient cycles 1993). The implications of this simplifica- tion are discussed in Chapter 4. As precipitation in the tropics is high and regular (e.g. Sabah on Borneo, Malaysia: 3505 mm per year, Schlensog 1997) without distinct dry seasons, water is not a limit- ing factor in tree growth (Friend 1993). An explicit description of water cycles within the model is therefore not necessary for ac- curate modelling results. Applications will Chapter 3

Concepts for the aggregation of tropical tree species into functional types and the application to Sabah’s lowland rain forests

Peter K¨ohler, Thomas Ditzer and Andreas Huth

Center for Environmental Systems Research, University of Kassel Kurt-Wolters-Str. 3, D-34109 Kassel, Germany

Abstract

For analysing field data as well as for modelling purposes it is useful to classify tree species into a few functional types.In this paper a new aggregation of tree species of the dipterocarp rain forests in Sabah (Borneo), Malaysia, is developed.The aggregation is based on the two criteria successional status and potential maximum height.Three classes of successional status (early, mid and late successional species), five classes of potential maximum heights (≤5 m, 5–15 m, 15–25 m, 25–36 m, > 36 m) and their systematic crossing lead up to 15 functional types.The criteria of the developed classification are chosen to fit applications with process-based models, such as Formix3 and Formind, which are based on photosynthesis production as the main process determining tree growth.The concept is universal and can easily be applied to other areas.With this new method of grouping a more realistic parametrisation of process-based rain forest growth models appears possible.

Keywords: dipterocarp forest, Malaysia, maximum height, model, plant functional types, successional status, tropical rain forest Journal of Tropical Ecology (2000) 16(4), 591-602. 26 Chapter 3

Introduction (2) Classification based on differences in potential maximum height. Richards (1936) was the first to derive a grouping concept Tropical rain forests are known for their in tropical rain forest research when he de- great tree species diversity with up to sev- scribed the layering structure of rain forest eral hundred different tree species in one canopy and distinguished tree species ac- hectare (Groombridge 1992). Their ecology cording to potential canopy layers. This ap- and physiology have been increasingly stud- proach was developed further by various re- ied in the last decades (e.g. Bazzaz & Pick- searchers (Hubbell & Foster 1986a; Swaine ett 1980; Mooney et al. 1980; Leigh et al. & Whitmore 1988; Poker 1995; Condit et al. 1985; Mulkey et al. 1996; Whitmore 1988, 1996; Denslow 1996). 1995, 1998). For generalization of ecologi- cal results for single species as well as for (3) Intensive statistical data analysis of modelling purposes different concepts have diameter growth pattern, for a specific site been developed for aggregating tree species to derive groups with significant different diversity in tropical forests into few plant diameter increment behaviour (Host & Pre- functional types (PFTs). gitzer 1991; Vanclay 1991; Ong & Kleine 1995). The principles of species aggregation into PFTs have been discussed widely (Botkin (4) Approaches which combine several 1975; Smith et al. 1993, 1997; Box 1996; Gi- concepts together. Lieberman et al. (1985 tay & Noble 1997; Lavorel et al. 1997). As 1990) combine diameter growth analysis pointed out by Gitay & Noble (1997) there with maximum size, Acevedo et al. (1995), is no universal classification or concept for Condit et al. (1996) shade-tolerance with the development of PFTs, the type of classi- maximum height, Shugart (1984, 1997) gap fication depends on the context of the per- requirements for regeneration with maxi- formed aggregation. PFTs are often used mum size. Kammesheidt’s (2000) classifi- in global vegetation models (Cramer 1997; cation is based on all available data con- Leemans 1997) and climate change analy- cerning different criteria as growth form, es- sis (Bugmann 1996a). For forest ecosystems tablishment, phenology, etc. In single case the following conceptual approaches can be studies pioneer species are distinguished distinguished: from other tree species, which are fur- ther subdivided (Manokaran & Kochum- (1) Grouping based on physiological cri- men 1987; Manokaran & Swaine 1994; teria such as shade tolerance at different Bossel & Krieger 1991; K¨ohler & Huth life stages (Hubbell & Foster 1986b; Whit- 1998a, b). more 1988, 1989, 1998). This concept varies from the rough distinctions whether Within the context of modelling, group- species are early or late successional ones ing concepts become important for integrat- (Shugart 1997) to more exact differentia- ing field data in terms of parameter values tions of several aspects of plant behaviour in models and for comparing typical simu- and growth strategies for light demanding lation results with observations (Vanclay & pioneer species and shade-tolerant climax Skovsgaard 1997). Interpretation of results species (Whitmore 1989). While Swaine & is easier with a small number of functional Whitmore (1988) state that it is difficult to types, where by with increasing number of distinguish more than the mentioned two PFTs accuracy increases as well. groups, Kammesheidt (2000) distinguishes Approaches already published are unsat- early, mid and late successional species. isfactory for the purpose of process-based However, Swaine & Whitmore suggest to modelling for two reasons. First, the bal- subdivide the two major ecological groups ance between adequate and necessary ac- into further sub-groups. Plant functional types in Sabah’s rain forests 27

Table 3.1: Successional status (SS) of 468 Swaine & Whitmore 1988). In the context of Sabah’s lowland tree species.No: Num- of modelling we define different successional ber of species per SS.Ab: Abundance of trees status as (a) different light requirements for with diameter > 10 cm in forest inventories the establishment of seedlings, (b) different in Deramakot, Lingkabau, Kalabakan and Ulu growth rates in a given light regime for trees Segama. of similar size, and (c) differences in mortal- ity rates. While early successional species Successional status SS No Ab [%] grow fast they tend to build low-density stems, in contrast to the slow growing late successional species which have higher wood Early successional spp. 1 31 24.8 densities. Based on the correlation between Mid successional spp. 2 317 63.4 wood density and growth rate a data set Late successional spp. 3 120 11.9 of Ong & Kleine (1995) on wood density covering 468 tree species was used to de- rive species successional status. Apart from typical pioneers (classified as early succes- sional spp. in our context), Ong & Kleine curacy has so far not been dealt with sat- distinguish timber groups of light, medium isfactorily. Most approaches use very few and heavy hardwood species. We classify (e.g. five) or many (20–50) PFTs, where those light and medium hardwood species 10 to 20 PFTs seems to be more appropri- as mid-successional, and heavy hardwoods ate, if both interpretation and accuracy is as late successional species. In a few of concern. Second, no approach is generic cases (including an undefined group, called in its concept and easily applicable to differ- OTHERS), grouping differs due to addi- ent forest sites using available data to derive tional knowledge on successional behaviour the appropriate number of PFTs. We there- (Rundi, pers. comm.). The quality of fore develop a universal approach, based on the timber group classification is verified the systematic combination of well estab- through a literature survey on wood den- lished classifications into different succes- sity (Meijer & Wood 1964; Burgess 1966; sional status and maximum height at matu- rity to derive about 10–20 PFTs, and apply Table 3.2: Aggregation of 468 of Sabah’s low- the concept to tropical lowland rain forests land tree species into five height groups (HG). in Sabah, Malaysia. Corresponding canopy layer.H: Height range at maturity.No: Number of species per HG. Ab: Abundance of trees with diameter > 10 cm in forest inventories in Deramakot, Lingkabau, Methods Kalabakan and Ulu Segama.

Criteria for the development of Canopy layer H [m] HG No Ab [%] plant functional types

We choose as grouping criteria successional Shrubs 0- 5 1 15 5.7 status (as defined in detail below) and at- Understorey 5-15 2 97 13.5 tainable maximum height. Lower canopy 15-25 3 119 32.9 We distinguish early, mid and late suc- Upper canopy 25-364 117 21.9 cessional species. We are aware of several Emergents >365 120 26.0 different nomenclatures for these classes (e.g. pioneers, non-pioneers), but find this the most appropriate (for alternatives see 28 Chapter 3

Table 3.3: Autecological characteristics of 13 plant functional types (PFTs) of Sabah’s lowland tree species.Height at maturity.SS: related successional status (Table 3.1).HG:related height group (Table 3.2). No: number of species per PFT. Ab: Abundance of trees with diameter > 10 cm in forest inventories in Deramakot, Lingkabau, Kalabakan and Ulu Segama.

Plant functional type Height [m] PFT SS HG No Ab [%]

Shrub mid successional spp. 0-5 1 2 1 15 5.7

Understorey early successional spp. 5-15 2 1 2 5 0.4 Understorey mid successional spp. 5-15 3 2 2 28 4.7 Understorey late successional spp. 5-15 4 3 2 65 8.3

Lower canopy early successional spp. 15-25 5 1 3 14 19.0 Lower canopy mid successional spp. 15-25 62 3 92 13.6 Lower canopy late successional spp. 15-25 7 3 3 13 0.3

Upper canopy early successional spp. 25-368 1 4 10 4.1 Upper canopy mid successional spp. 25-369 2 4 89 16.0 Upper canopy late successional spp. 25-3610 3 4 18 1.8

Emergent early successional spp. >3611 1 5 3 1.2 Emergent mid successional spp. >3612 2 5 93 23.3 Emergent late successional spp. >3613 3 5 24 1.5

Fox 1970; Cockburn 1980; Keating & Bolza determined using the literature (Meijer & 1982; PROSEA 1994). Wood 1964; Burgess 1966; Fox 1970; Whit- more & Ng 1972; Cockburn 1980; Keat- The maximum potential height hmax of trees is grouped into five classes for Sabah’s ing & Bolza 1982; PROSEA 1994). In rain forests. The classes can be named some cases, where no data on maximum according to their canopy layers as emer- height were available maximum girth or di- ameter was used to determine maximum gents (hmax > 36m), upper main canopy height by using height-to-diameter func- (25 m shrub species was occupied by a smaller needs of model development and its appli- number of species (15). The abundance cation and the existence or availability of of trees in the FMU inventories however field data (Gitay & Noble 1997). However, was not equally distributed. Lower main for the proposed groupings there are possi- canopy species were most abundant (33%), bilities to estimate the quality of the classifi- followed by emergent species (26%), up- cation. In particular, different successional per main canopy species (22%), understorey status can be checked against field data in species (14%) and shrub species (6%). the criteria we defined above for our classi- fication. Specifically, we (1) check growth rates in terms of diameter increment and Plant functional types photosynthesis measurements, and (2) dis- cuss mortality rates, the latter for both suc- An independent application of both clas- cessional status and grouping of potential sification criteria led to a maximum of 15 maximum heights. PFTs. We did not distinguish the shrub species in their successional status, because Mortality and diameter increment rates of the lack of information on them. The calculated from permanent sampling plots resulting 13 PFTs and their autecologi- (PSPs) in three different forests reserves cal characteristics are shown in Table 3.3. in Sabah (Garinono, Segaliud Lokan and Numbers of species per PFT varied from Sepilok) are taken to validate the resulting 3 to 93. Again the FMU inventories as grouping. The PSPs differ in size (2 to 7 one possible species distribution for Sabah ha), observations length (9 to 20 y), num- showed different abundances between 0.4% ber of (3–10) and time between recordings and 23.3%. From the characteristics of the (1–5 y). Analysis of these data can be found aggregation one can easily identify various 30 Chapter 3 important species, e.g. most dipterocarps ences between groups. The data set of PSP are within PFT 9, 10, 12 and 13, Macaranga is not appropriate for this analysis for sev- spp. are found in PFT 5. eral reasons. Time between two enumera- tions was 5 y in most cases, and thus light regimes might change extremely within this Discussion time step. Low growth rates of early suc- cessional spp. in Garinono can be explained with the high density of this stands (basal Successional status area BA = 31.3 m2 ha−1, all other stands BA < 30 m2 ha−1). Thus, average light For 259 species (of 468 species in total), regimes might have suppressed growth of data on wood density were found. This in- early successional spp. Plotting frequency completeness was the reason to group the distribution of diameter increment data for species list not according to values on den- different successional status is another pos- sity but on the more general concept of sibility to gain comprehensive information quality differences in wood density as done (not shown). It can be seen, in individuals by Ong & Kleine (1995). Another reason showing high diameter increment rates, that for not using detailed values as the basis for early successionals had the highest share, grouping is the fact that often wide ranges late successional species the lowest. of densities are found in the literature for in- Data of photosynthesis measurement for dividual species. Therefore dividing up the testing our PFTs are rare (Eschenbach et al. range of wood densities would become dif- 1998) or not appropriate to our problems ficult if not impossible. However the data (Moad (1992) measured only photosynthe- available enable us to validate the timber sis of dipterocarp seedlings, all belonging to grouping of Ong & Kleine (1995). Because PFTs 12 and 13 in our concept and thus of the incomplete data set, comparing me- giving no information about differences be- dians (not means) seems to be the appropri- tween PFTs; the measurements of Koyama ate statistic. We found wood density ρ median (1981) were on excised leaves). Testing of 455, 657 and 921 kg m−3 for early, mid, of nine species against photosynthesis data and late successional species, respectively. of Eschenbach et al. (1998) supported that Thus, the concept of timber groups by Ong early successional species have a higher sat- & Kleine (1995) seems to be based on wood uration point in photoproduction. From the densities. nine measured species, one was an early, Growth rates in terms of diameter incre- six mid, and two were late successional ments show trends, expected by our group- species. Their average maximum assimila- ing concept (Table 3.4). In Sepilok and Se- tion rates (ranges in parentheses) were 19.4, galiud Lokan 1 & 2 average growth rates 9.3 (5.7–16.7) and 6.8 (6.8) µmol m−2 s−1 re- are highest in early successional, and lowest spectively estimated by fitting a rectangu- in late successional species, although differ- lar hyperbola (Thornley 1976). Recent re- ences are not significant (χ2 test; P > 0.7). search on the relationship of photosynthesis In Garinono, average growth rate of early characteristics and asymptotic tree heights successionals is lower than growth rates of (Thomas & Bazzaz 1999) found significant mid successionals. It is not surprising to differences in photosynthetic production for find no significance in the differences of trees of the same successional status but growth rates. They are averaged over both with different maximum heights. different light regimes and diameter ranges. As a last test of the classification of suc- A more detailed analysis considering growth cessional status we analyse mortality rates rates as function of PFT, light regime and measured in PSPs. The average annual diameter range might find significant differ- Plant functional types in Sabah’s rain forests 31

Table 3.4: Diameter increment rates [mm y−1] for different successional status SS (early (1), mid (2) and late (3) successional spp.). N: sample size. P-values of χ2-test.

Location SS N χ2 P 123

Garinono 3.3 4.1 2.8 7694 0.33 0.85 Sepilok 4.8 3.9 2.9 6435 0.43 0.81 Segaliud Lokan1 5.0 4.9 4.2 6132 0.11 0.95 Segaliud Lokan2 6.6 5.4 4.6 2213 0.47 0.79

Table 3.5: Average annual mortality rates [% y−1] for different successional status SS (A: early (1), mid (2) and late (3) successional spp.). Sample size see Table 3.4. P-values of χ2-test.B only distinguishes between early (1) and non-early (4) successional spp.

AB Location mean SG SG 123χ2 P4χ2 P

Garinono 2.6 3.8 1.9 2.7 0.75 0.69 3.2 0.68 0.41 Sepilok 5.1 6.6 4.7 7.5 1.6 0.45 5.0 2.68 0.10 Segaliud Lokan1 5.1 8.4 4.4 3.9 2.54 0.28 6.4 2.48 0.12 Segaliud Lokan2 6.3 9.8 3.4 2.9 5.15 0.08 4.8 2.24 0.13

mortality rates differ in the different PSPs tween mid- and late successional species (Garinono: 2.6% y −1; Segaliud Lokan1: easily. However, differences between early 5.1 % y−1; Segaliud Lokan2: 6.3 % y−1 and and non-early successional spp. are seen Sepilok: 5.1 % y−1) and over time, indi- clearly (Table 3.5B). Differences between cating changes with dry years as observed groups increase as analysis is focused on two in Sabah in 1982/83 (Walsh 1996). Mor- groups only. tality is unexpectedly high in all observa- The discussion (by Hubbell et al. (1999)) tions, compared to normally observed an- about recruitment limitations and abun- −1 nual mortality rates of 1–2 % y in tropi- dances of seedling in canopy gaps cannot cal rain forests (e.g. Manokaran & Swaine be broadened to include our concepts yet, 1994). Mortality rates decline from early because data available on recruitment pat- to late successional species in the two ar- terns (FMU inventories) lack information eas in Segaliud Lokan, whereas in Garinono on canopy structure. and Sepilok, beside highest mortality rates in early successional spp., mid-successionals have lowest rates (Table 3.5A). The group- ing might not resolve the differences be- 32 Chapter 3

Table 3.6: Average annual mortality rates [% y−1] of mid- and late successional spp.for different height groups (shrubs (1), understorey(2), lower main canopy (3), upper main canopy(4) and emergent (5)).N: sample size.P-values of χ2-test.

Location mean Height group N χ2 P 12345

Garinono 2.0 3.4 3.0 1.61.60.9 4867 2.28 0.68 Sepilok 5.0 9.2 7.0 4.1 4.4 4.3 5825 4.63 0.33 Segaliud Lokan1 4.4 6.2 3.7 2.9 3.4 4.9 4752 1.69 0.79 Segaliud Lokan2 3.3 6.3 3.6 2.7 3.5 2.9 952 38.05 0.56

Maximum potential height It should be noted finally that any PFTs defined lie on a continuum and dividing it up is a matter of convenience based on ar- Because the list underlying our classifica- bitrary divisions. tion concentrates on tree species it is not surprising to find very few shrub species in it. In our context, missing shrubs are Plant functional types unimportant. It might be necessary, how- ever, to consider those shrubs for analysis of In previous model applications (Huth et al. slash-and-burn-techniques practised by in- 1998; K¨ohler & Huth 1998a; Ditzer et al. digenous people (Whitmore 1998). 2000) the non-existence of a principal ap- The height limits chosen were al- proach to grouping has been a crucial ready used (with small differences) in the point. Thus, within only five groups, Formix3 model (Huth et al. 1998; Ditzer which were distinguished by maximum tree et al. 2000). Thus, model application and heights, one contained all early successional former data analysis have shown them to be species. This implied that all mid or late very practical. Nevertheless, one might de- successional species with similar maximum fine a different number of height groups at heights were grouped to slightly incorrect different height limits. height groups. From this experience, the optimal number of derived PFTs was be- As some verification of the height group tween 10 and 20. At the upper end, classification we again analyse trends in the parametrization already becomes difficult, mortality rates for different groups. We but modelling the complex system of the only consider differences between height tropical rain forest with less than ten PFTs groups of mid and late successional species, might include assumptions leading to biased knowing that early successional species have results. higher mortality rates. Taller-growing trees, in general, should have longer life-times than shorter-growing trees (Manokaran & Conclusions Swaine 1994). This tendency is found in their mortality rates (Table 3.6), although We presented a generic concept for the ag- differences from the average are not signif- gregation of tree species into plant func- icant (χ2 test; P > 0.3). Again the test tional types which can be applied to forests verifies our classification. in different regions. The concept was de- Plant functional types in Sabah’s rain forests 33 veloped in the context of process-based Additional remarks, not modelling of forest dynamics and therefore was focused on criteria which are essen- included in the article tially influencing tree growth in the models Formix3 and Formind: successional sta- We think that the aggregation of tree tus and potential height. In the application species to plant functional types is abso- for Sabah’s lowland rain forests successional lutely necessary in the modelling of tropical status was classified into three groups, po- forest dynamics. In the meantime a model Formosaic tential height into five groups. Thirteen called was developed (Liu & plant functional types in total were dis- Ashton 1998, 1999), which tried to quan- tinguished for this application, a number tify the dynamic of species richness for the which we consider as very practical for fur- large and long-term research area of Pa- ther forest growth analysis. Within this soh, Malaysia (50 ha, inventoried for now Formosaic concept it is and will be difficult to rely 15 years). The approach of was on published data sets for all species in to model the dynamic of individual species, this respect. In this case it is important but the abundance of many of these tree to test the classification with all available species was too low - even in this area of field data. We have shown different possibil- 50 ha - to gain statistically well supported ities for testing using field data on diameter results for recruitment, growth and mortal- increment, mortality rates, photosynthesis ity. production and wood densities. Simultaneously to the development of As a consequence of the final PFTs de- our grouping concept different approaches rived in this paper a new parametriza- were proposed for tropical rain forest in tion of the forest growth models Formix3 Ghana, Africa (Atta-Boateng & Moser and Formind will be elaborated. Simu- 1998), Australia (Gitay et al. 1999) and lations and model analysis with the new Kalimantan, Indonesian Borneo (Phillips parametrization will show whether and in et al. 2000). The first approach was fo- what ways the quality and accuracy of the cused on commercial tree species and based results are improved. on typical diameter increment rates with the emphasis on model construction. In the second case, various theoretical con- siderations about principle differences of Acknowledgements ecological characteristics, which might be used for the identification of plant func- We would like to thank all members of tional groups were discussed (Pillar 1999; the Malaysian–German Sustainable Forest Weiher et al. 1999). To identify timber Management Project, and the Forest Re- groups was the main targets in the species search Center, Sandakan, Sabah, Malaysia, grouping in Indonesia. However, applica- responsible for data collection, for their kind tion and validation possibilities of the con- cooperation, especially M. Rundi for shar- cepts were also of central interest (McIntyre ing his knowledge on the successional status et al. 1999b). More general considerations of trees, R. Ong and M. Kleine for making concerning different applications in global data available for us and R. Glauner for in- vegetation models and for analysing field formation on height-to-diameter-relations. data were found in a special issue of Jour- We also owe a debt of gratitude to D. New- nal of Vegetation Science (McIntyre et al. bery and an anonymous reviewer for very 1999a) and in the standard text book of helpful comments on a former version of the Smith and colleagues (Smith et al. 1997). manuscript. Thanks to L. Kammesheidt for critical reading.

Chapter 4

The model Formind2.0

Because model descriptions in articles with a model, which initially might be ver- need to be very brief, a complete descrip- bal describing main interactions. How- tion of the model used is contained in this ever, to obtain quantitative answers to ques- chapter. tions of interest, mathematical equations An individual-oriented cohort model are needed. From the first more general de- (Uchma´nski & Grimm 1996) is described, scription of interactions, qualitative conclu- which is able to simulate growth dynam- sions about modelled systems can be drawn. ics in mixed tropical rain forests. The Two general types of models can be distin- model includes all important growth pro- guished: Those describing behaviour and cesses. Thus, the model can be applied those explaining the system. Descriptive to different rain forest sites, if parametri- models try to match system behaviour with sation is adapted to specific conditions. Af- model behaviour. Very often regression ter some more general thoughts about mod- functions are used in this context. Explana- els, the principles of the modelling approach tory models try to extract essential struc- are described. Spatial and temporal resolu- tures of the system correctly and rebuild tions are described. Individual physiolog- them in the model. The advantage of the ical submodel and their mathematical im- latter approach is the possibility to study plementations are explained in detail. Fol- systems with different environmental con- lowing this, the main differences to a former ditions (Bossel 1992). version of Formind (K¨ohler 1996; K¨ohler Often models describe very complex sys- & Huth 1998a, 1998b) are discussed. Simi- tems. Thus, it is necessary to reduce the lar features of various versions of the model number of modelled processes. One in- Formix3 (Ditzer et al. 2000; Huth & Ditzer evitably has to make simplifying assump- 2000a) are also mentioned. A discussion of tions which will not enable all possible ques- the chosen model approach closes the chap- tions to be answered with the same mod- ter. elling approach. To gain an overview over the quality of a model, several criteria in respect to the aim General concepts about of modelling should be fulfilled. The dy- namic behaviour of the model should qual- models itatively fit to that of the real system. Nu- merical and logical model results should cor- One of the basic approaches in physics is respond to those of the original if environ- the description of physical phenomena with mental or boundary conditions are compa- mathematical models. rable. Differences should be explainable through assumptions made during model Modelling of ecological systems uses a building. Simulation results should be use- similar approach. A system is described 36 Chapter 4 ful with respect to potential applications of the model and with respect to the aim of modelling.

Basic structure of model ha Patch 100 m In the following section spatial and tempo- 20 m ral resolutions are described. Furthermore, the technical realisation of the model as Figure 4.1: Spatial resolution of simulated an individual-oriented cohort model is ex- area. plained. 4. the extent to which variability of indi- viduals of the same age is considered. The individual-oriented ap- proach Formind2.0 uses simple assumptions concerning nutrient and water cycles. For Individual-based modelling is one of the higher computing efficiency, small individu- main concepts in recent theoretical ecology als are packed together into cohorts. Thus, (DeAngelis & Gross 1992; Judson 1994; Liu models are highly flexible because cohorts & Ashton 1995; Grimm 1999; Lett et al. canbeaddedandremovedeasily.Accord- 1999; Haefner & Dugaw 2000). However, ing to Vanclay (1994) the three main com- there are various researchers who empha- ponents of tree growth are modelled in the size the importance of different modelling following way within a cohort model: approaches. Thus, it is desirable to combine the advantages of different concepts (Bolker 1. Diameter increment is modelled by in- et al. 1997; Uchma´nski & Grimm 1996). crementing the size of a representative Advantages of individual-based modelling tree; with those of the cohort approach (Vanclay 1994) are optimised and unified in the cur- 2. mortality is simulated by reducing the rent study. Thus, our approach is called expansion factor (the number of trees individual-oriented according to the rather represented by each cohort); and narrow definition of Uchma´nski & Grimm 3. recruitment is accommodated by initi- (1996). ating new cohorts from time to time. The biological criteria, which underlie this classification of different approaches are Within one cohort the growth of one tree is as follows (Uchma´nski & Grimm 1996): modelled, which interacts through complex functional relationships with trees of its own 1. the degree to which complexity of in- and the other cohorts. dividual’s life cycles is reflected in the A strict individual-based model corre- model; sponds to a cohort model with an expan- 2. whether or not the dynamics of the re- sion factor of one. Aggregating trees to sources (e.g. food, space) is explicitly cohorts is an effective optimisation of com- taken into account; puting time. For initialisation, trees of the same species group, same commercial status 3. the use of numbers of individuals or and within the same spatial subunit are ag- densities in representing the size of gregated into cohorts in diameter classes of populations; and 5 cm. Within the cohorts of the small trees The model Formind2.0 37

Figure 4.2: Different boundary conditions and their ecological meaning.Left: Open boundaries represent a highly fragmented situation.Right: With periodic boundaries the simulated area is assumed to be surrounded by similar forest structure.Falling trees and seed dispersal are shown as possible interlinking processes between neighbouring patches.

some hundred individuals might be packed cess, crowns of large trees might extend into together. In general, trees with a diameter neighbouring areas if they exceed the size of larger than 20 cm are simulated individually their own patch. as a result of mortality and self-thinning. We distinguish between open and pe- riodic (or toroidal) boundary conditions. Spatial resolution Open boundaries correspond to forest is- lands in a heavily fragmented landscape. A simulation area of one hectare of rain Thus, e.g. seed dispersal leaving the sim- forest is divided into 5 × 5 subareas (so- ulation area will get lost. We do not con- called patches), each 20 m × 20 m in size sider any migration effects, which might en- (Fig. 4.1). Within the patches trees do not ter simulation in open boundaries. Peri- have explicit positions. This corresponds to odic boundaries assume the area of interest the spatial resolution of available inventory to be embedded in a larger area of similar data and has been proven practical in var- structure. Any action leaving the simula- ious different so-called gap models for tem- tion area will thus enter it on the opposite perate forests (Botkin et al. 1972; Shugart side (Fig. 4.2). 1984, 1998; overview at recent applications Several field inventories were performed in Liu & Ashton 1995). In recent years spa- with the same spatial resolution of 400 m2. tial explicit rain forest models have been de- Distribution of trees to single patches was veloped (Liu & Ashton 1998; Chave 1999b; thus easily possible. In other cases only av- van Gardingen & Phillips 1999), but they all erage tree densities, without spatial infor- use much simpler formulations for describ- mation, was recorded. In the later case, ing tree growth compared to Formind2.0. trees are randomly distributed into single All individuals in one patch are consid- patches. ered as direct neighbours and compete for light and space. Interlinking processes be- tween neighbouring patches are the falling of trees and an explicit modelling of seed dispersal. In a third minor interacting pro- 38 Chapter 4

Respiration Photoproduction

Biomass Competition for Light

Tree physiognomy

Recruitment of Seedlings Tree number

Cohort i Mortality & Competition for Space i+1 i+2 new cohorts

Figure 4.3: Overview of the interactions of the submodels and the dependencies on the main variables biomass B and tree number N.Arrows indicate whether the results of a submodel influence the calculations of another submodel.

Temporal resolution olution of dt = 1 y is chosen here. Solv- ing the system of differential equations with dt = 1 y will lead in most cases to minor, Temporal resolution is of importance in inrelevant, numerical errors (investigated in dynamic models, as accuracy of modelled K¨ohler 1996). dt is critical for the numeri- processes, numerical solution of differential cal solution only for small trees and will be equations and computing time depend on it. reduced in this case (explained after model Various forest growth models include description). yearly and daily variations in sun an- gle in their description of light conditions (e.g. Treedyn3, cf. Bossel 1994). Rain for- Description of plant- est growth models in general do not consider a temporal resolution smaller than a year physiological submodels for other than numerical reasons (Vanclay 1994). This is possible because of relatively The model is divided into the submodels small variations of daily and seasonal pa- • rameters. Day length changes only slightly tree physiognomy, within a year, dusk and dawn are very short, • light competition, irradiance is mostly diffuse and not direct as a result of rainy and cloudy weather condi- • tree growth (photosynthesis, respira- tions (Schlensog 1997). Data for a valida- tion), tion of a more detailed description of pro- • mortality and spatial competition, and cesses like photosynthesis are not available. Furthermore, all relevant data of basic pro- • recruitment. cesses (growth, mortality, recruitment) are very rarely measured with a resolution be- An overview on interactions between dif- low one year. Therefore, a temporal res- ferent submodels and the main variables The model Formind2.0 39 biomass and tree number within one cohort crown diameter dc can be seen in Fig. 4.3. Photoproduction and respiration are calculated for individ- ual trees. They change their biomass. Tree crown projection area

physiognomic assumptions allow transform- c f ing biomass into various other variables like diameter, tree height, leaf area, crown di- mensions, and bole volume. Light competi- tion and mortality are interacting processes between trees of the same and different co- crown length horts, and they directly affect individual h photoproduction and tree number, respec- tively. Competition for light regulates, via available irradiance at the forest floor, po- height tential recruitment of seedlings. New re- cruits will be added to a new cohort. In the following, parameters with indices s or h depend on species grouping as out- lined in Chapter 3, specifically on succes- sional behaviour or maximum tree height, respectively. Parametrisations for different sites are found in the corresponding Chap- stem diameter d ters 5–8. Variables are calculated for each Figure 4.4: A typical tree of the model is cohort i. shown including the relevant physiognomic di- mensions.Stem diameter is measured at breast height (h = 1.3 m.) Tree physiognomy

For calculating competition processes and τ1, τ2: parameters. For sites in South- growth variables such as stem diameter di, America we take average values for τ. height hi, crown length ci and crown pro- A form factor γi covers differences of con- jection area fi are required (Fig. 4.4). They ical stem from a cylindric form (Fig. 4.6A): can be calculated from biomass Bi as fol- · · γ2 lows. γi = γ0 exp(γ1 di ), (4.3) The stem of each tree is assumed to be γ0, γ1,andγ2: parameters. of conical shape. Its aboveground biomass Tree height hi is calculated from diame- Bi is calculated via ter (Fig. 4.6B). · π 2 ρs γi(di) di Bi = di · hi(di) · . (4.1) hi = , (4.4) 4 τ 1 + di h0h h1h with ρs: wood density, γi: form factor, h0h,andh1h: parameters. and τ: fraction of stem biomass to total aboveground biomass. Several field studies A tree in the model has a cylindrical in South-East Asia (Kato et al. 1978; Ya- crown shape. Its crown diameter dci is cal- makura et al. 1986) indicate a relation be- culated from stem diameter: f2 tween τ and tree height for reference size dc =(f0 + f1 · d ) · d, (4.5) d120 (d =120 cm) at a certain site, also and its circular crown projections area fi called site index (Ditzer 1999, Fig. 4.5): π 2 i follows from crown diameter (f = 4 dci), τ = τ1 + τ2 · h(d120), (4.2) with f0, f1,andf2: parameters. 40 Chapter 4

[-] 1.0 r2 = 99.6%

0.9 [-] 0.6 Sebulu 0.8 0.55 0.7 0.5 0.6 Pasoh 0.45 Form factor 0.5 0.4

Stem biomass fraction 40 45 50 55 60 65 70 50 Site index [m] [m] 40 h 30 Figure 4.5: Depending of stem wood fraction 20 τ on site index.Field data were taken in Pasoh Height 10

(Kato et al.1978) and Sebulu (Yamakura et al. ] 2 0 1986).

[m leaf area

f, l 600 6 Total leaf area li of one tree i is a function crown area of stem diameter di with an upper thresh- 400 4 old LAIM to avoid unrealistic high values LAI [-] (Ashton 1978). 200 2 LAI 2 3 l1 · di + l2 · di + l3 · di li =max (4.6) 0 0 · Crown/leaf area LAIM fi 0 50 100 150 Stem diameter d [cm] Thus, leaf area index LAIi of an in- dividual tree is obtained from total leaf area lI divided by crown projection area Figure 4.6: Functions depending on stem di- (Fig. 4.6AC): ameter plotted for parametrisation of Sabah, Malaysia.Top: Form factor γ(d) (Kato et al. li 1978; Yamakura et al.1986). Middle: Height- LAIi = (4.7) fi to-diameter-relations h(d).Height is only plotted for achievable diameters of different

Crown length ci is proportional to tree height groups (Forestal-International-Limited 1973; K¨ohler 1998).Bottom: Leaf area l(d) height hi (Burgess 1961; Poker 1993): (closed line), crown projection area f(d) (bro- ken line),andleaf area index LAI(d) (dotted ci = cP · hi, (4.8) line) (Kato et al.1978; Yamakura et al.1986). cP: parameter. with Stem volume Vi can be calculated out of    biomass using τ and ρs (Vi = τ/ρs ·Bi). For 3 3 growth and yield studies only the volume xi =1− (1 − c ) − 3fi − . (4.10) p 2 4 Vbi below the crown is of interest (called bole volume). Using geometric relations of for the frustum of a cone (Bronstein & Se- Light competition mendjajew 1991; Ditzer 1999) Vbi is calcu- lated as: Knowledge about the distribution of leaf area within the canopy is necessary for cal- 1 2 Vbi = (1 + xi + xi )(1 − cP)Vi, (4.9) 3fi culating the light climate in a patch. The The model Formind2.0 41 15 description of a single tree crown was al- Early SS ready given in the previous subsection. s)] 2 In a vertical direction the canopy is di- 10

)/(m Mid SS vided with equidistant steps of the width 2 ∆z in a finite number of layers. It is now

(CO 5 Late SS necessary to calculate which leaf areas are [

found in the various layers between z and P z +∆z (called z in the following). It is as- 0 sumed that leaves are distributed homoge- 0 500 1000 neously within the tree crown. 2 Irradiance I0 [ (photons)/(m s)] The contribution of each cohort i to crown coverage F (z)ofthelayerz is calcu- [-] q lated out of stem number Ni and the crown 0.8 - Early SS projection area fi of an individual tree nor- 0.6 malised to patch area A: - Mid SS 0.4 - Late SS fi F (z)= Ni · , for some i. (4.11) i A 0.2

The leaf area L of layer z follows from the Growth limitation 0.0 0 50 100 150 summation of all leaf area indices LAIi(z) of cohorts belonging to layer z multiplied Stem diameter d [cm] with their crown coverage : fi Figure 4.7: Functional dependencies of tree L(z)= Ni · · LAIi(z) , for some i. i A growth at the site Sabah, Malaysia.Top: Light (4.12) response curve (Eschenbach et al.1998). Bot- tom: Growth limitation.For each of the 13 Now, for each individual tree i the leaf plant functional types one graph is plotted. area Li has to be calculated, which is found Height groups have different maximum diame- above its crown and which partially absorbs ter.For one height group graphs are labelled the incoming irradiance: according to their successional status. Li = L(z), with z>hi. (4.13) z as a function of total leaf area index L.Pho- Tree growth tosynthetic Active Radiation (PAR) is fre- quency dependent and called irradiance or For the determination of changes in biomass light intensity I in the following. first The light response curve of photosynthe- sis Pi is of Michaelis-Menten type, a typical • photosynthetic production, then saturation relationship between light inten- • respiration sity and production: αs · Ii is calculated. Pi(Ii)= αs , (4.14) i 1+ PMs I

with P s as maximum photoproduction at Photosynthetic production M saturation, and αs as quantum use effi- Photosynthetic production is based on the ciency. work of Monsi & Saeki (1953). Incoming ir- Within the canopy vertical light absorp- radiance is partially absorbed in the canopy tion after the law of Beer-Lambert (Gerth- 42 Chapter 4 sen et al. 1989) is assumed (Fig. 4.7A): Respiration

· −k·Li Ii(Li)=I0y e . (4.15) All biomass losses are summed up under

Individual available light Ii of each tree, as what we call respiration. They are com- function of its total leaf area index above posed of root decay, litter-fall and respi- ration of tree organs and leaves at night. tree crown Li,iscalculated. I0y corresponds to average light intensity of the photoactive Daily leaf respiration is implicitly included time of day, taking seasonal differences into in the light response curve (Eq. 4.16). account. Absorption coefficient k is esti- We distinguish between a biomass- mated out of detailed field studies about the dependent maintenance respiration Rmi microclimate of light within tropical forests (Kira 1978; Yoda 1983), and growth respi- (Kira & Yoda 1989). ration Rgi, depending on actual photopro- Actual photoproduction P˜i is calculated duction. by integrating Pi over the shading canopy · 2/3 · under the assumption of totally closed Rmi = r0s Bi + r1s Bi, (4.20) canopy layers: r0s and r1s: parameters, Li ˜ Pi = Pi(Ii(L))dL. (4.16) · · ˜ 0 Rgi = RG qi Pi, (4.21) Solving the integral (Eq. 4.16) leads to: RG: parameter. Details on respiration are − PMs αskIi + PMs (1 m) found in the work of Ditzer (Ditzer 1999; P˜i = ln , (−kLAIi) k αskIie + PMs(1 − m) Ditzer et al. 2000). (4.17) with m as transmission coefficient of the Changes in biomass leaves (Thornley 1976).

Different length of wet/dry seasons SSy With our assumptions concerning photo- and seasonal dependent length of photoac- synthesis and respiration the time depen- tive daytime SDy determine length of pho- dent changes in biomass are calculated as toactive time and thus contribute to the follows in our main growth equation: photoproduction. dBi Furthermore, a size-dependent growth = P˜i · q(1 − R ) − R i (4.22) dt G m limitation q caused by water transport deficits is assumed (Fig. 4.7B). Limitation is chosen in a way that trees will stop grow- Mortality and spatial competi- ing if they reach their maximum diameters tion DM (Ditzer 1999). Other models incorpo- rate similar limitation factors, but call them Tree mortality in undisturbed tropical rain age limitation (as done for cs in Landsberg forest lies on average between 1 and 3 % & Waring 1997): per year (Swaine 1989; Phillips & Gentry 1994; Condit et al. 1995b; van der Meer & − − · di 2 q =1 (1 qDM ) ( ) (4.18) Bongers 1996). DM Different types of mortality are included Parameter qD corresponds to growth lim- M in the model. itation at maximum diameter and is cal- culated internally from the following con- Normal mortality: If field data indicate dition: functional relationships between tree mor- dd(DM) tality and tree size M = f(d) (Okuda et al. = 0 (4.19) D dt 1997), or diameter growth MI = f(dinc) The model Formind2.0 43

[-] patches. The number of trees NF destroyed

D 0.4 from total number Np in target patch p is M calculated from crown projection area fF of 0.3 the falling tree relative to patch size A: 0.2 fF N = Np , (4.25) 0.1 F A

Mortality rate 0.0 Individuals are chosen randomly with the 01020304050 restriction that only trees smaller than the Stem diameter d [cm] one falling can be destroyed, and contribute with different NFi to tree losses in the co- horts of target patch. Figure 4.8: Size-depending mortality rate Self-thinning: In sites with a high tree MD at the site Sabah, Malaysia.For trees with adiameterd>10cm MD =0. density, mortality is significantly increased. This phenomenon is called self-thinning. In the model NT trees in patches with crown (Swaine 1989) they can be added to the ba- closure (F (z) > 1) are randomly extracted sic mortality MBs,h. Well known significant until crown coverage F (z) decreases below differences in mortality rates between differ- its maximum value of 1.0. ent successional status and maximum tree As second main equation covering height are covered in MB. Thus, early suc- changes in tree numbers of each cohort we cessionals have shorter lifetimes, and tree obtain: mortality is lower in high growing trees. dNi The basic equation for mortality rates is = −(Mi · Ni + N i + N i). (4.26) dt T F Mi = MBs,h + MD + MI. (4.23) In cohorts with high tree numbers (Ni ≥ Mortality of small trees is significantly 100) and small individuals (d<10 cm) de- higher than average (Fig. 4.8): terministic mortality takes place. Thus, M − M /M · d : dSouth America 90 % of total mortality approach consists of assuming that an in- was caused by falling trees (van der Meer & tact forest surrounding the simulation area Bongers 1996). is supporting a constant seed input rate NSs Dying trees with a diameter d ≥ 10 cm within the simulated patches. can fall over (probability W ) and cre- (2) Scenario seed tree: The second ap- ate gaps of different sizes in neighbouring proach takes into account the dispersal of 44 Chapter 4 0.016 from the mother tree (Clark 1998; Clark 0.014 XR = 100m

[-] XR = 75m et al. 1998a)].

f 0.012 XR = 50m 0.01 Different dispersal agents (e.g. wind, 0.008 birds, mammals) are not directly distin- 0.006 guished in our model, but the resulting av- 0.004 erage dispersal distance XRs depends upon Probability 0.002 the tree species and should match with the 0.0 parameter set. From the dispersal kernels 0 100 200 300 discussed by Clark et al. (1999) we use the Distance r [m] Gaussian distribution (as used by Chave 1999b). Assuming rotation symmetry, the Figure 4.9: Seeds dispersal kernels for a probability density f ofseedstobedis- Gaussian distribution with different average persed at the distance r from the mother dispersal distances XR.Crown diameter was tree is fixed at cd=20 m. 2 2r − r f(r)= 2 exp 2 , (XR + cd/2) (XR + cd/2) seeds produced within each patch from lo- (4.27) cal parent trees, which are trees exceeding a with cd, the crown diameter (see Fig. 4.9). certain diameter D h. Recruitment strate- Thus, 99% of the seeds are dispersed in a R × gies are highly variable in rain forests, with distance less than 2.14 (XR+cd/2). The interspecific differences in the fruiting pe- actual dispersal distance r is randomly riod, seed sizes varying in a six-fold range drawn from this probability distribution, (Westoby, 1995), dispersal strategies, dis- and the direction is drawn uniformly. The persal agents, dispersal distances, seed sur- resulting seed shadow is the product of the vival, germination probabilities and matur- rate of seed production and the dispersal ing size of seed dispersers (Denslow 1987; kernel (Clark et al. 1999b). Garwood 1983; Whitmore 1983). Thus, For both recruitment mechanisms, in- some fundamental assumptions on the most coming seeds will be added to a seed pool, important trends have to be drawn. taking into account the average seed mor- Flowering, fruiting and seed produc- tality rate MSs across functional groups (cf. tion vary in duration and frequency across Garwood 1983, 1989). These seeds corre- species, some species fruiting after sev- spond to the reproductive success and are eral years of unfecundity (Garwood 1983). those which can potentially be established Other species flower and fruit continu- as seedlings at the minimum diameter of ally throughout the year in Malaysian rain 1 cm (Ribbens et al. 1994; Chave 1999b). forests (Putz 1979). Seasonal differences in Seed loss due to predators is implicitly in- seed production are not taken into consid- cluded in relative low seed production rates. eration. The actual seed germination depends upon

The rate of seed production N s varies R • a minimum light intensity at the forest widely among species (Whitmore 1998). floor I ≥ I s (Whitmore 1998), and Various studies have analysed different dis- F S persal strategies and lengths (review in • a not fully closed lowest canopy layer. Clark et al. 1999b). A major result is that migration velocity found in paleoeco- If conditions for ingrowth are fulfilled ad- logical records can only be explained with a ditional cohorts are created. The state vari- seed dispersal kernel which allows a reason- ables biomass Bi and tree number Ni are able amount of seed establishments far away initialised with biomass corresponding to The model Formind2.0 45

seedlings diameter DS and number of seeds Improvements to former of PFT available for ingrowth in seed pool of current patch. versions of the model

Main differential equa- tions In comparison to former versions For- mind1.0 (K¨ohler & Huth 1998a) and For- mind1.1 (K¨ohler et al. 2001) all submod- Finally, resulting differential equations are els of the model were revised and improved explained in detail: on the basis of current understanding of For each patch l(l =1, ..., m) a certain the ecological processes in tropical rain number n of time (t) dependent cohorts forests. Tree physiology was formulated in i(i =1, ..., n(l, t)) exist. For each cohort more general functional relationships and i, changes in above-ground biomass Bi,l(t) parametrised on the basis of new field data of an individual tree and number of trees available only recently. Assumptions in the Ni,l(t) belonging to that cohort are calcu- mortality submodel were simplified and a lated. density dependent self-thinning rule was im- plemented. The recruitment submodel was enlarged with an alternative site-dependent dBi approach, whose development and influence = P˜i · q(1 − R ) − R i dt G m on general model behaviour play a major dNi role in model analysis. Respiration was = −(Mi · Ni + N i + N i) dt T F modelled in greater detail, and a concept was developed to validate it on the basis of These equations are a system of 2 × m × available field data. Photosynthesis was re- n(l, t) coupled ordinary integro-differential vised slightly. Finally, a simply dependency equations, which are not solvable with an- of tree growth on potential dry periods was alytic methods. Equations are coupled implemented and thus the range of applica- through the interactions of trees of same tions was enlarged. and different cohorts and patches. They In the end, only the concepts of spa- can be called integro-differential equations, tial resolution and general formulations because a differentiation over time and an of photosynthesis and respiration were integration over leaf area (as function of taken from previous versions and from the biomass) are incorporated. Formix3-models, the latter especially from For numerical solutions of the system Formix3-Q (Ditzer 1999). of equations an Euler-Cauchy algorithm is The main conceptual difference to used with time step dt of one year. The Formix3 is the individual-oriented ap- solution of the first equation ( dBi ) is criti- dt proach. Formix3 so far works with a sim- cal for small Bi. Therefore, for small trees ple concept of matrix models incl. tran- (Bi < 1t)dt is reduced to dt =0.01 y. sition rates between different size classes, ThemodelwascodedinC++usingthe which are difficult to parametrise. With its simulation software XSiSi/SiSi, which was more general formulation of processes For- developed in Kassel (http://www.usf.uni- mind2.0 differs from its predecessors also kassel.de/ reinhard/sisi). Simulations were in its basic concept of species grouping, on run on a PC (400 MHz, system Linux), tak- which the whole parametrisation is based. ing on average 9 sec to simulate 100 years This allows a relatively fast model applica- and 1 ha of rain forest growth. tion to different forest sites. 46 Chapter 4

Discussion especially the case if secondary succession or external disturbances (e.g. logging of trees) are of interest. The modelling approach used in For- mind2.0 enables the developer or user to In contrast, the model used in this study change each submodel and replace it with is able to simulate an area of any size, a currently more practical one. Sometimes restricted only by computing capabilities. research and new available field data for a This is useful for various reasons. First, certain site will suggest such modifications. the fraction of patches at the border of the We discuss each submodel on its own. simulated area will decrease with increasing size if the shape of area remains quatratic The size of the model area in which trees (from64%at1hato15%at25ha).Bor- compete with each other is important, as der areas are sensitive as assumptions on the dynamics of succession processes de- boundary conditions (e.g interactions leav- pend on it. The impact of one dominant ing the simulation area) will affect them tree on the light climate in the patches is strongly. Furthermore, results are more in- too strong in too small areas. Thus, all dependent of stochasticity, included in mor- other plants are repressed more than in re- tality and recruitment, the larger the area ality. Only after the death of the domi- is. nant tree does growth in the recruitment take place. In too large simulated areas, The individual-oriented cohort-approach leaf area in relation to patch area is too is coupled with the chosen spatial resolu- small. Death of even large trees will change tion of 20 m × 20 m. Only the non- the light climate only slightly. In both cases explicit spatial position causes an aggrega- the real dynamic of succession in forest gaps tion of individuals into cohorts. The res- will not be met with acceptable accuracy. olution of available data at the time of the From these considerations it emerges that model development was the main reason for the most appropriate size of a patch used that approach. Until now only a few re- in models should be that of the crown of search plots with explicit tree positions for a typical large-canopy dominant tree of the each individual larger than 1 cm in stem stand. Typical sizes lie between 400 m2 and diameter have been inventorised (Condit 800 m2. In forests of the tropical regions et al. 2000; Smithsonian-Tropical-Research- patch size can be chosen at the lower end Institute 2000). In a few cases data are of the range. This may be due to the fact freely available, e.g. for a neo-tropical rain that the steeper sun angles at low latitudes forest in Costa Rica (Clark & Clark 2000). allow light to reach the forest floor in rela- The gap-model approach, used for spatial tively small gaps (Shugart 1998). The patch resolution and calculation of competition size of 400 m2 chosen in Formind2.0 lies in situations in single patches, has been proven the given range taken from those theoretical practical in various case studies (overview considerations. in Liu & Ashton 1995). Tropical (Shugart et al. 1980; Doyle 1981; K¨urpick et al. 1997) A large number of forest growth models and temperate forests (Botkin et al. 1972; calculate forest dynamics only in one patch Shugart 1984, 1998; Botkin 1993) have been that is why they are called gap models (Liu modelled, and influences of climate and el- & Ashton 1995). Most of them emerged evation gradients were analysed (Bugmann from Jabowa (Botkin et al. 1972; Botkin 1996b, 1997). All gap-models known to the 1993) and use age dependent growth func- author calculate so-called potential natural tions. From our point of view a single patch vegetation (PNV), which reflects the steady- can never be representative of a whole rain state of a model at current parametrisation, forest. Normally, areas in different phase of if simulation was started from a clear-cut succession exist beside each other. This is The model Formind2.0 47 area. Thus, tendencies in species composi- cept for germination is not needed. For tion as a response to assumed changes in en- temperate forests detailed models of light vironmental conditions can be analysed, but climate are available which can be used in a comparison of simulated dynamics with spatial-explicit models (Brunner 1998). In field data of long term observation areas was this context, an error propagation analy- missing. This comparison is an important sis undertaken in the spatial-explicit model validation method (Vanclay & Skovsgaard Sortie has to be discussed (Deutschmann 1997), whose results, together with an es- et al. 1999). It was analysed how model timate about the quality of the model, are dynamics depend on a more detailed reso- more important than technical details, e.g. lution of incoming light (in 1, 16, 48, or 216 which modelling approach has been chosen. light rays). It transpires that most results The few available data sets with explicit (e.g. succession of tree species) achieved tree position raise the question if an appli- with one ray of light do not significantly cation of Formind2.0 to those sites is only differ from those of a more detailed descrip- possible and meaningful after considerable tion. Differences between 16and 216are model improvements. In this context only always negligible. the vertical light competition needs to be Comparisons of model results with diam- discussed. In tests with Formind (K¨ohler eter increment data from permanent sam- 1996) we analysed whether we gain any fur- pling plots without explicit tree positions ther information of reducing temporal res- have proven our light competition model to olution from steps of one year to months, be of acceptable quality (K¨ohler et al. 2001). days or even hours. A case study for French Formulations concerning tree physiog- Guiana uses a spatial-explicit forest growth nomy are at the current state of re- model (Chave 1999b). Complex interac- search. The available data sets for South- tions of individual trees with a three dimen- East Asia verified the used functional re- sional (3D) field vector of irradiance were lationships between different tree variables calculated. As a result a detailed distribu- (e.g. Forestal-International-Limited 1973; tion of irradiance in each part of the canopy Yamakura et al. 1986; Ashton & Hall 1992; can be calculated. This study and our own Poker 1993). Currently, research activities investigations have shown that computation in rain forests are concentrated on the neo- time mainly depends on calculation of this tropics (Condit 1995; Cook 1998; Holl & irradiance field. Thus, Chave was using Kappelle 1999; Peres 1999). Thus, pub- highly parallel computer systems available lished field data for South America will only in large research institutes to perform be improved in the near future. This will his study in an acceptable time frame. It is enable us to specify some of assumptions, questionable if the detailed field vector gets where regional differences were not cap- the model closer to the real system, as ele- tured so far (see applications in Chapter 6- vation on the forest floor or daily variations 8). Site-dependent relationships of sev- in light climate have not been considered so eral physiognomic variables in South-East far. Modelling of sun spots, important for Asia has improved parametrisation (Ditzer the germination of seeds (Hammond et al. 1999). The broadening of this approach to 1999), would be possible with daily varia- the sites in South America was so far not tions. Influence of sun spots on growth of possible because of a lack of field data. In Formind seedlings in was already analysed particular, analysis of leaf area or biomass (K¨ohler 1996, p. 31) and can be neglected. partitioning was only undertaken in a few As recruitment rates work with the concept sites in South-East Asia. A verification of of reproductive success covering also seed the upper boundary of individual tree’s leaf and seedling predation, a more detailed con- area index LAIM (Eq. 4.7) is of special in- 48 Chapter 4 terest. This boundary is, so far, caused by subject. The splitting of respiration and a model comparison with data from photo- more detailed description of the processes synthesis production. Self-shading in tree was only possible with the process-based ap- crowns with high values of LAI leads to proach used in the model. Ditzer (1999) has higher production rates of mid successional highlighted the importance of these details, species compared to early successionals for which can not be incorporated in simpler trees of the same size and for high light in- approaches, where tree growth is modelled tensities. This is inconsistent with observa- with one diameter growth function based on tions. In this context parametrisation of the regressions. light response curve (Eq. 4.14) does not dis- We considered light conditions as most tinguish different values for the light use effi- important for determining tree growth, ciency αs, as differences in field data are sta- an approach used in other studies before tistically weak (Eschenbach 1994; Eschen- (Bartelink 1998a, 1998b). One obstacle in bach et al. 1998), but results are sensitive including below-ground processes is the dif- to differences in αs. Other field studies use ficulty in measurement design and imple- different regression functions, where αs is mentation. However, there are research ac- not calculable, or determine only maximum tivities in this direction in tropical forests photosynthesis PM (e.g. Bazzaz & Pickett (Denslow et al. 1998; Hall & Matson 1999; 1980; Oberbauer & Strain 1984; Ellsworth Chambers et al. 2000) or on a global scale & Reich 1996; Barker et al. 1997). However, (Jackson et al. 2000). In temperate for- the principal differences in photosynthetic est growth models nutrient or water cy- characteristics of tropical trees along suc- cles were considered already (Aber et al. cessional gradients were confirmed (Strauss- 1982; Jansen et al. 1995; Bossel 1996b; Debenedetti & Bazzaz 1996), and were par- Endejan 1997; Friend et al. 1997; Kim- tially correlated with tree height (Davies mins et al. 1999; Thornley & Cannell 2000). 1998; Thomas & Bazzaz 1999). Analysing corelations between soil nuitri- The main improvements of the tree ents and forest stockings led to corrections growth submodel were the size-dependent factors of tree growth in the rain forest growth limitation and splitting of respira- model Formix3-Q (Glauner 1999; Ditzer tion into maintenance and growth respira- et al. 2000). We are therefore aware of tion. Limitation of tree growth has been the effects and the dependency of the tree found in field datas (Koch et al. 1994; growth on soil matters. Nevertheless, for Maruyama et al. 1997) and was already used average site conditions measured diameter in forest grwoth modelling (Valentine 1988 growth was matched with simulation results 1990; Landsberg & Waring 1997). Thus, (Chapter 5). growth is limited, even if there is evidence A main improvement of the mortality that large emerging trees do not stop grow- submodel is to model tree death with- ing (Chambers et al. 1998). Species richness out growth dependent mortality, which was Formind2.0 is aggregated in PFTs in and used in a former version or the Formix3 only the growth of average trees is modelled. model. Some field data (Swaine & Whit- Limitation is a very useful and proper con- more 1988) give a hint of that relation- cept. ship. But as tree size was not consid- Respiration in tropical trees is one of ered in their analysis it is difficult to gen- the processes of which very little is under- eralise results. Otherwise it would promote stood (Kira 1978; Medina & Klinge 1983; higher mortality in large trees, which nat- Yoda 1983; Oberbauer & Strain 1984; Ryan urally grow slower than smaller ones. This et al. 1997). The work of Ryan and co- is in contradiction with observations. The- authors mark important progress in this oretical considerations concerning the prod- The model Formind2.0 49 uct ω of growth rate g, maximum diameter • establishment of seedlings (Lang & dM and mortality rate M (ω = dM/g · M) Knight 1983; Whitmore 1983; Ri´era support our thesis. The number of large 1985; Hubbell & Foster 1986a, b; trees would be over- (ω<<1) or under- Schupp et al. 1989; Hartshorn 1989; Ve- estimated (ω>>1) if ω is very different blen 1989; Manokaran & Swaine 1994; from one (Chave 1999b). Enhanced mortal- Milton et al. 1994; Phillips & Gen- ity of young trees is the only effective regu- try 1994; Pinard et al. 1996; Poorter lation of ingrowing trees. This effect is well et al. 1996; Sheil & May 1996; Powers known, but confirmed by only a few stud- et al. 1997; Okuda et al. 1997; Tucker ies (Clark & Clark 1992; Kennedy & Swaine & Murphy 1997; van Gardingen et al. 1992; Okuda et al. 1997). Sensitivity analy- 1998; Diaz et al. 1999; Hammond et al. sis of size-dependent mortality and compar- 1999; Hubbell et al. 1999; Kyereh et al. ison of tree densities with field census make 1999; Xiong & Nilsson 1999), a site-dependent parametrisation possible. • and photosynthesis and growth of Density regulation through self-thinning seedlings (Chim & On 1973; Enright is a known phenomenon in forests of dif- 1978; Manokaran & Kochummen 1987; ferent latitude (White 1981; Westoby 1987; Brown 1990, 1993, 1996; Condit et al. Valentine 1988; Clark 1992a, 1992b; Pen- 1995a; Lee et al. 1996; Press et al. fold & Lamb 1999; Silva-Matos et al. 1996; Barker et al. 1997; Lee et al. 1999). Increasing basic mortality rates in 1997; Zipperlen & Press 1997; Ko- dense patches, as done in previous ver- hyama & Takada 1998; Agyeman et al. sions (K¨ohler & Huth 1998a; Ditzer 1999; 1999; Kobe 1999b; Nicotra et al. 1999; Huth & Ditzer 2000a) is far less restric- d’Oliveira 2000). tive. In Formix3 density regulation might work this way, as transition rates from Thus, it is important to incorporate de- one layer to the other limit the growth of tailed submodels of recruitment in forest saplings, and thus, residence times in lower growth models. Besides a theoretical ver- layers are prolonged. Adaptive changes ification of field data on recruitment, the in tree physiognomy (Wirtz 1998), espe- simulations of temporal dynamics of forests cially crown shape, are not considered in will be improved. the model. Thus, in self-thinning an instant extraction of trees with overlapping crowns It is one of the main improvements of the was included in Formind. model to be able to model explicit seed dis- persal depending on mother trees compared In experimental ecology, natural regener- to only constant input of seedlings. How- ation and all processes concerning recruit- ever, it is certainly true that the establish- ment as main mechanisms determining fu- ment of individual trees of the huge number ture forest compositions were of central in- of tree species in tropical forests follow more terest in the last years. Research can be complex patterns. A controversial study distinguished in concerning recruitment on Barro Colorado • seed production (Putz 1979; Garwood Island, a nature reserve on an island in the 1983, 1989; Charles-Dominique 1993), Panama Canal (Hubbell et al. 1999; Chaz- don et al. 1999; Hubbell 1999; Kobe 1999a; • seed dispersal (Fox 1972; Hubbell Brokaw & Busing 2000) found that ”being et al. 1991; Kennedy 1991; Wunderle at the right place at the right time”ismore 1997; Clark 1998; Clark et al. 1998a, important for recruitment success than any 1998b; Higgins & Richardson 1999; individual strategy. Hubbell et al. (1999) Martinez-Garza & Gonz´alez-Montagut did not find significant differences in the re- 1999; Robinson et al. 1999), cruitment of different successional groups 50 Chapter 4 in closed forest and canopy gaps. Fur- thermore, the limiting number of available seedlings determined success more than any other environmental conditions. So far, it is not possible with the model to verify stud- ies which identify tree species diversity as function of different impacts (Cannon et al. 1998, 1999; Sheil et al. 1999). Enhancing Formind2.0 in a way that each individ- ual tree is correlated with one specific tree species while still modelling forest growth with a reduced set of plant functional types might address these questions in the future. In particular, recruitment could be coupled to trees of individual species. A more detailed description of recruit- ment than done so far does not seem to be useful. Necessary assumptions and data concerning seed predators, wind direction and seed etc. would be highly speculative, and parameters difficult to determine. Interactions of disturbed animal, bird, or insect populations, which act as seed dis- persers and are important for future forest development, are in the focus of current eco- logical interest (Redford 1992; Curran et al. 1999; Law & Lean 1999; Lynam & Billick 1999; Price et al. 1999; Cullen et al. 2000; da Silva & Tabarelli 2000). As dispersal agents of trees are complex, and mostly not depending on one species, the effects of ex- tinction of a single animal species on plant dynamics is difficult to estimate. However, it is definitely correct that hunting pressure and the collection of seed bearing fruits in now intact forest communities will alter re- cruitment capabilities of these in the future (Redford 1992). We take this into account through scenario analysis, where those in- teractions are considered as main effects (Chapter 8). Chapter 5

Comparison of measured and simulated growth on permanent plots in Sabah’s rain forests

Peter K¨ohler∗, Thomas Ditzer∗,RobertC.Ong+ and Andreas Huth∗

*: Center for Environmental Systems Research, University of Kassel Kurt-Wolters-Str. 3, D-34109 Kassel, Germany +: Forest Research Centre, Forestry Department Sabah P.O.Box 1407, 90008 Sandakan, Sabah, Malaysia

Abstract

In this paper previously unpublished field data from 25 ha of permanent sampling plots (PSPs) in Sabah, Malaysia, in four different forest reserves are analysed for mortality rates and basal area development.Field data of an observation length of nine to 20 years were available.These data then form the basis of several benchmark tests for the evaluation of the individual-oriented tropical rain forest growth model Formind.Anew version of Formind is presented.The model in its version Formind1.1 includes enhanced submodels for mortality and tree growth.The model evaluation is focused on the model components for tree growth, competition and mortality.Data for tree recruitment were not available.Results show a good agreement between simulation and field data for the main output variables basal area and stem number indicating a reasonable behaviour of the model components we focused on.Furthermore the results show that differences in site conditions influence tree growth and mortality.Site characteristics should be included in the model in the future. Keywords: forest growth model; tropical rain forest; dipterocarp forest; mortality; Malaysia; basal area; Formind Forest Ecology and Management (2001) 142(1–3), in press. 52 Chapter 5

Introduction model Formind. Formind was devel- oped following an individual-oriented ap- proach (Huston et al. 1988; Judson 1994; Evaluation of forest growth models is an im- Liu & Ashton 1995; Uchma´nski & Grimm portant procedure of model development. 1996) and used to validate the approach of Vanclay & Skovsgaard (1997) discussed the more aggregated process-based model range and importance of model evaluation. Formix3 (Huth et al. 1998). One impor- An evaluation of tropical rain forest models tant feature of both models is the use of is difficult due to a lack of adequate field species grouping into PFTs. A detailed data. Besides a comparison of model out- model description and some results of For- put in a steady state with primary rain for- mind have already been presented in K¨ohler est data (e.g. Bossel & Krieger 1994; Huth and Huth (1998a, b). Several submod- et al. 1994; Huth et al. 1998; K¨ohler & els (tree growth, competition, mortality) of Huth 1998a, 1998b; K¨urpick et al. 1997) few Formind were modified in the meantime permanent sampling plot data exist which due to new available datasets, research ac- are suitable for testing rain forest growth tivities and model analysis. These improve- models. The field data used in this pa- ments are documented in the following. per were not available to the authors at the time of model development and can there- fore serve for testing of model results. Be- Area description cause those field data are used here for elab- orating parameter values of mortality, they can not be seen as fully independent, but The permanent sampling plots (PSPs) in- semi-independent. vestigated in this study were established and inventorised by the Forest Research The data used in the following were col- Centre and Forestry Department Sabah, lected in the forest reserves Garinono, Gu- Malaysia. They are all located in the nung Rara, Segaliud Lokan and Sepilok in lowland dipterocarp rain forest of Sabah, Sabah, Malaysia by the Forestry Depart- Malaysia. The PSPs located in different ment. Analysis in terms of site and stand forest reserves across Sabah sum up to a characteristics, mortality and recruitment forest area of 25 ha (see Table 5.1). The data are unpublished and only available data set of Segaliud Lokan is split into two in several research reports (Ong & Kleine parts because of differing observation times 1995; K¨ohler 1998). Details of the field within the forest reserve. The number of data used here are therefore documented in PSPs in the different locations varies from this paper. Especially mortality rates are one to eleven ha, observation time from nine analysed as a function of time and different to 20 years with recordings in intervals be- plant functional types (PFTs). tween 1 and 5 years. Elevation is below 100 Comparison of simulation results and m, only Gunung Rara is located in a higher field data from permanent sampling plots region (200 m - 600 m). Site quality was (PSPs) is important especially when models analysed by Ong & Kleine (1995) on the are used for estimation of long term trends basis of landform and parent material. The of forest growth with or without anthropo- site quality of Gunung Rara differs signifi- gene influences as forest management (Huth cantly from that of the other reserves. et al. 1994, 1998; Riswan & Hartanti 1995) Each PSP covers an area of or climate change (Pastor & Post 1988; 100 m × 100 m, subdivided into 25 Overpeck et al. 1990; Shugart 1998). patches of 20 m × 20 m, which are further The simulation model investigated in this split up into 4 sub-patches of 10 m × 10 m. study is the tropical rain forest growth Within these sub-patches no further infor- Comparison of measured and simulated growth on permanent plots in Sabah 53

Table 5.1: Information about permanent sampling plots (PSPs) located in different forest reserves in Sabah, Malaysia.A: size of PSPs [ha]; B: number of trees at first enumeration; C: time of observation; D: length of observation [y]; E: number of enumerations; F: time between two enumerations [y]; G: time between last logging and first inventory [y]; H: site quality.

Location A B C D E F G H Elevation [m]

Garinono 2 871 1973-1982 9 10 1 45 good 40-80 Gunung Rara 11 4978 1981-1990 9 7 1-2 11-12 poor 200-600 Segaliud Lokan1 7 4258 1982-1992 10 3 5 25 good 40-100 Segaliud Lokan2 1 365 1972-1985 13 8 1-2 8 good 40-100 Sepilok 4 2218 1973-1993 20 5 5 19 good 20-50

mation about tree location were recorded. concepts for PFTs were proposed (Swaine & Trees with a diameter at breast height Whitmore 1988; Poker 1993). We use three (dbh) ≥ 10 cm are labelled. In regular growth characteristics for grouping (poten- inventories the dbh of all labelled trees were tial height, light demands for growth and recorded including ingrowing small trees. regeneration) and derive four PFTs for the Death of labelled trees was also recorded. dipterocarp lowland rain forests of Malaysia (Table 5.2, for details see K¨ohler & Huth 1998b). A fifth PFT for bushes and small Methods plants with heights below 1.3 m, which was used in former simulations (K¨ohler & Huth 1998b) is not necessary here, because PSP The rain forest growth model inventories were focused on trees with a Formind1.1 dbh ≥ 10 cm. For simulations a forest stand area of one hectare is divided into The Formind model was developed for the smaller patches. The model follows the simulation of tropical rain forest in Malaysia gap-model approach (Botkin et al. 1972; (see K¨ohler & Huth 1998a, b for further de- Shugart 1984; Botkin 1993) to modelling tails). It is a successor of the Formix3 tree competition by describing tree interac- model (Appanah et al. 1990; Bossel & tion on patches. These patches have the size Krieger 1991, 1994; Huth et al. 1994, 1998). typical of treefall-gaps as they are naturally As main processes the model includes tree created by dying larger trees (20 m × 20 m), growth, competition, mortality and regen- which is the same patch size as in the PSP eration (last is not included in this version inventories. In contrast to most gap-models because of a lack of field data). In the fol- (an exception is the ZELIG model by Smith lowing we will explain the approach used in & Urban 1988; Urban et al. 1991) we aim version Formind1.1. at picturing the shifting forest stand mosaic Species grouping and spatial structure: and we therefore simultaneously simulate Tropical forest stands are usually composed several patches explicitly in their neighbour- of a large number of species. For the pur- ing location within the stand. The patches pose of investigating forest dynamics it is themselves are pictured as homogeneous. useful to classify species into a small number of plant functional types (PFTs). Different 54 Chapter 5

Table 5.2: Characteristics of the aggregated plant functional types (PFTs) of lowland diptero- carp rain forest of Sabah, Malaysia.

PFT Maximum Light demand Species composition heights

1 >36m shade tolerant emerging species mainly dipterocarps 2 25-36m shade tolerant climax species dipterocarps and non- dipterocarps 3 15-25 m light demanding pioneer species mainly Macaranga spp. and Antocephalus chinensis 4 ≤15 m shade tolerant understorey non-dipterocarps species

Table 5.3: Parametrisation for a dipterocarp lowland rain forest in Sabah, Malaysia, used by the Formind1.1 model.Parameters concerning mortality are depending on the location and can be found in Table 5.4. Names are identical to those used in the detailed model description in K¨ohler and Huth (1998b).Index j indicates that parameter values differ for different plant functional types.

Name Description Unit Plant functional type 1234

a a0j Coefficient of height-diameter relation [m] 2.94 2.30 1.97 3.11 −1 a1j Coefficient of height-diameter relation [m cm ] 0.42 0.42 0.39 0.30 −2 a2j Coefficient of height-diameter relation [m cm ] -0.002 -0.002 -0.002 -0.001 −3 ρj Wood density [todm m ] 0.62 0.57 0.37 0.71 hMj Maximum potential height [m] 55 36 25 15 mgCO2 PMj Maximum photo-productivity [ 2· ] 10.9 11.6 29.1 18.8 dm h 2 mgCO ·m j 2 α Slope of light response curve [ dm2·h·W ] 0.36 0.20 0.20 0.30 total aboveground biomass)

τj Fraction of stemwood to total biomass [-] 0.7 sj Crown-to-stem-diameter-ratio [-] 25 LAIj Leaf area index of single tree [-] 2 −1 RPj Respiration (biomass losses relative to [y ]0.16 −2 I0 Light intensity above canopy [W m ] 335 k Light extinction coefficient [-] 0.7 W Probability for a dying tree to fall [-] 0.0

a 2 Height-diameter relation: h = a0j + a1j · d + a2j · d . Comparison of measured and simulated growth on permanent plots in Sabah 55

Individual tree growth: Within a sin- and of respiration of woody tree organs and gle patch the model calculates the devel- of leaves. Respiration is considered a func- opment of a forest stand based on cohorts tion of tree size and PFT (Ditzer 1999). A of trees of the same PFT. Such a cohort is water balance is not included in the model. characterised by the number of trees and The calculation of tree growth is performed by the size of one representative tree. Us- in annual time steps. ing allometric relations, the size of a tree Competition: Competition is modelled can equivalently be expressed in terms of in terms of competition for light as de- its above-ground biomass, height, or diam- scribed above and competition for space as eter at breast height. The crown projec- described below concerning mortality. tion area is calculated from stem diame- ter via the proportionality of stem diame- Mortality: Mortality is modelled on an ter and crown diameter (Rollet 1978; Whit- annual basis. In the current version it more 1984; Poker 1993). These relation- does not depend on any other processes ships between components of tree size (di- such as diameter increment. The mortal- ameter, height and crown dimensions) are ity rates used for the simulations in this based on average field data, and are impor- study are directly obtained from the anal- tant simplifications that makes the model ysis of PSP-data. The model includes an tractable, but they may reduce its accuracy. additional crowding mortality for trees in Emergent trees might have a crown projec- dense patches (crowns do not have enough tion area bigger than the patch size. Their space). In this case trees die to such an ex- crowns are then assumed to reach into the tent that crowding does not occur anymore. neighbouring four patches. Crown length Because of the short length of simulations ≤ is a function of tree height (Richards 1952; ( 20 years) we do not include processes of Burgess 1961; Poker 1993). With these re- falling trees and the creation of canopy gaps lations the distribution of individual tree by these trees. crowns in the canopy can be calculated. As- Regeneration: The Formind model in- suming a fixed leaf area index (LAI) of in- cludes also a submodel for regeneration. dividual trees the leaf area distribution in Seedling establishment was not measured in the forest can be calculated. The growth − of the individual tree is based on a carbon Table 5.4: Average mortality rate m [% y 1] balance. Calculations include photoproduc- for different plant functional types (PFT) cal- tion of the trees and assimilate losses due to culated from permanent sampling plot data in respiration and renewal. Photoproduction different locations and used as parameter val- is calculated from the tree’s leaf area and ues for simulations.For Segaliud Lokan2 we only used the data recorded between 1972-1982 its specific productivity. The latter depends and the parameter values used in simulations on the local irradiance for each tree (Monsi therefore differ from the average value. & Saeki 1953; Thornley 1976). Within a patch light attenuation downwards in the canopy is calculated with respect to the Location average PFT absorption of higher located tree crowns. 12 3 4 The dependence of specific photosynthetic productivity on irradiance is modelled us- ing a Michaelis-Menten-type light response Garinono 2.59 2.40 0.62 3.86 2.54 curve parametrised for each PFT (Eschen- Gunung Rara 0.24 0.31 0.26 0.12 0.16 Segaliud Lokan1 5.10 4.48 2.89 12.03 3.46 bach et al. 1998). Assimilate losses are es- Segaliud Lokan2 (6.34) 0.0 0.0 0.0 0.0 timated in relation to tree biomass (Kira Sepilok 5.09 5.49 3.76 5.89 2.58 1978; Yoda 1983). Losses are composed of renewal of roots, above-ground litter fall 56 Chapter 5

Table 5.5: Basal area (BA) and stem number of all species (Nall) and the different plant functional types (N1,N2,N3,N4) at the beginning of observation for trees with d≥10cm in different forest reserves.

Location BA Nall N1 N2 N3 N4 [m2 ha−1][ha−1][ha−1][ha−1][ha−1][ha−1]

Garinono 28.3 435.5 288.0 27.5 95.5 24.5 Gunung Rara 17.4 450.4 205.9 21.6 168.6 56.0 Segaliud Lokan1 31.3 608.3 422.4 41.0 95.1 45.4 Segaliud Lokan2 12.0 365.0 133.0 10.0 199.0 22.0 Sepilok 24.6 554.5 462.0 44.0 16.3 28.5

the PSPs. Estimation of recruitment rates forest growth models. This is called bench- as an alternative to the use of field data mark test. Basal area and stem number is not considered, because uncertainties in- were chosen for comparison of simulation cluded in the estimation will lower the qual- results with field data because these vari- ity of the evaluation. Therefore all tests are ables can directly be derived from the PSP done without considering regeneration. inventory data. For each PSP a simula- Formind1.1 Model parametrisation: A detailed de- tion with was performed over scription of literature sources of the param- the same time period for which data were eter values used for the lowland dipterocarp available. In cases where data of more than rain forests of Sabah, Malaysia, is presented one hectare were available data were aver- in K¨ohler & Huth, (1998b). Table 5.3 con- aged after simulation. Two different kinds tains the parametrisation used in the test of comparison were undertaken. First, basal undertaken for this paper. Values of param- area and stem number for different PFTs eters in Table 5.3 are similar to those used at the end of the simulations were com- in previous studies (K¨ohler & Huth 1998a, pared with those measured in the PSPs. b) with the exception of mortality rates (see Second, temporal development of basal area Table 5.4), missing regeneration parameters and stem number over simulated/observed and the probability W of dying trees to fall. time was analysed. Initialisation: From the stem-diameter We represent results in the following way: distribution of the first enumeration of each PSP trees are aggregated into different co- xsimulated(tend) horts regarding their PFT, diameter (in = f(xmeasured(tend), PFT, FR) xmeasured(tend) diameter classes with a width of 5 cm) (5.1) and location in the stand (in patches of and 20 m × 20 m).

x simulated = f(t, PFT, FR), (5.2) x Benchmark tests measured

As outlined by Vanclay & Skovsgaard with x: basal area or stem number, tend: (1997) a comparison of simulated data with last year of inventory/simulation, FR: forest field data not used for model development reserve, t: time and PFT: plant functional is an appropriate method for evaluation of type. Comparison of measured and simulated growth on permanent plots in Sabah 57

Garinono Gunung Rara Sepilok

] 7 ] 1.0 ] 10 -1 6 -1 -1 9 0.8 8 5 7 4 0.6 6 5 3 0.4 4 2 3 0.2 2 1 1 Mortality rate [% y 0 Mortality rate [% y 0.0 Mortality rate [% y 0 1974 1976 1978 1980 1982 1982 1984 1986 1988 1990 1978 1983 1988 1993 Time [y] Time [y] Time [y]

Segaliud Lokan1 Segaliud Lokan2

] 6 ] 40 -1 5 -1 35 30 4 25 3 20 2 15 10 1 5 Mortality rate [% y 0 Mortality rate [% y 0 1987 1992 1974 1978 1982 1985 Time [y] Time [y] Figure 5.1: Mortality rates as function of time in permanent sampling plots (PSPs) in different locations (Garinono, Gunung Rara, Segaliud Lokan and Sepilok) in Sabah.Detailed information about PSPs in Table 5.1. Dotted Line: Average mortality rate between first and last enumer- ation.Circles: Mortality rate between actual and previous enumeration.Circles also describe when enumerations took place.

Results where n0 is the number of trees at the first enumeration, n1 is the number of trees at the second enumeration t years later with- Permanent sampling plot anal- out considering any new trees growing in ysis between the two enumerations. Mortality rates differ widely for differ- The structure and stocking of the PSPs ent forest reserves. Table 5.4 lists average varies widely as seen in Table 5.5. Where mortality rates for different PFTs over the the stocking of Segaliud Lokan2 and Gu- whole time of observation. Average values nung Rara is very low (basal area (BA) of range from 0.24 % y−1 in Gunung Rara to 12.0 and 17.4 m2 ha−1 respectively) the rela- 6.34 % y−1 in Segaliud Lokan2. Mortality tive fraction of pioneer species of (PFT 3) is rates of pioneer species (PFT 3) are with the very high indicating that these stands were exception of Gunung Rara generally higher heavily disturbed by logging. In contrast than of non-pioneer species (e.g. Segal- to that PSPs in Segaliud Lokan1, Garinono iud Lokan1: m =5.10 % y−1 for PFT 1, and Sepilok are well stocked (BA = 31.3, m =12.03 % y−1 for PFT 3). 2 −1 28.3 and 24.6m ha , respectively) with a The temporal development of the average lower fraction of pioneer species. Thus the mortality rates (Fig. 5.1) shows high fluctu- data represent a wide range of forest stock- ations in most forest reserves. Especially in ing. Gunung Rara, Segaliud Lokan2 and Sepi- Annual mortality rates m were calcu- lok there is a constant increase in mortality lated in the following way (Manokaran & rate to the end of the observation period. Swaine 1994): m =(logen0 − logen1)/t, In Segaliud Lokan2 a very high increase of 58 Chapter 5 the mortality rate was analysed (m =0 Gunung Rara. Sepilok and Garinono even %y−1 for eight years, m =36%y−1 in come closer to measured values the longer the last year). Trends like this cannot be we simulate. explained with the present version of the Stem number (Fig. 5.2C&D): model which is based on constant mortal- ity rates and therefore the last three years Stem number can be simulated more pre- of data in Segaliud Lokan2 are not consid- cisely than basal area (maximum deviation: ered in our tests. 25 %). This is a result of the mortality rates used in the simulation which were derived from the observations in the PSPs. The de- Evaluation of Formind1.1 viation of the total stem number after total simulated/observed time is in all forest re- In the simulation studies shown in this serves below 6%. There is a tendency of un- article we concentrate on the comparison derestimating stem number in simulations. of results with the data from the perma- Again deviation in plots with a higher stem nent sampling plots. Other tests like the number is smaller, highest deviation occurs long term tendency of mature forest stands from PFT 3, which represent the pioneer incl. species composition were performed in species. K¨ohler & Huth (1998b). The deviation in total stem number plot- The results of the benchmark tests fol- ted against simulation/observation time in- lowing Eq. 5.1 and 5.2 are documented in dicates always an underestimation of simu- Fig. 5.2. First we analyse the results for the lated stem number. The deviation is even basal area, then the findings for the stem stabilising with longer simulation time for numbers. Sepilok from 10 % to nearly 0 %. Basal area (Fig. 5.2A&B): Results in stem number and basal area have to be analysed together. Development Simulations show good agreement with of stem number is considered as a result of the field observations. The deviation of sim- the simulated mortality processes, but the ulation results range between 0 % and 30 development of basal area is the product %, in only one case 50 %. There is no PFT of mortality, growth and competition pro- where our simulations show a trend of per- cesses in their interaction in a forest stand. manent over- or underestimation. PFT 1 seems to be the most critical PFT with the highest deviation of nearly 50%. The high- est deviation was observed for the forest re- Discussion serve Gunung Rara, which seems not to be simulated accurately with this version of the Mortality rates model. PFTs which have a basal area below 2 −1 5m ha tend to be simulated with lower Typical values of the average tree mortal- values than measured. The total basal ar- ity rates in primary tropical rain forests are eas are matching the measured values more 1-2 % of stem number per year (Putz & precisely in stands with a higher stand basal Milton 1982; Lang & Knight 1983; Swaine area. Again the highest variation is found et al. 1987a, b; Manokaran & Swaine 1994; in the Gunung Rara simulation. Milton et al. 1994; Phillips & Gentry 1994; Deviations in total basal area plotted Condit 1995, 1998; Condit et al. 1995a) against simulation/observation time show with a significant higher mortality rate that nearly all forest reserves stabilise for pioneer species (Primack & Lee 1991; within the simulation time at an acceptable Manokaran & Swaine 1994). Manokaran & error range (± 20%) with the exception of Swaine (1994) analysed mortality rates in Comparison of measured and simulated growth on permanent plots in Sabah 59

1.5 1.5 Sum Garinono 1.4 PFT 1 1.4 Gunung Rara PFT 2 Segaliud Lokan1 1.3 PFT 3 1.3 Segaliud Lokan2

[-] PFT 4 [-] Sepilok 1.2 1.2 1.1 1.1 measured measured 1.0 1.0 /BA /BA 0.9 0.9

simulated 0.8 simulated 0.8

BA 0.7 BA 0.7 0.6 0.6 0.5 A 0.5 B 0 5 10 15 20 25 30 02468101214161820 Field data: basal area [m2 ha-1] Time [y] 1.5 1.5 Sum Garinono 1.4 PFT 1 1.4 Gunung Rara PFT 2 Segaliud Lokan1 1.3 PFT 3 1.3 Segaliud Lokan2 PFT 4 Sepilok [-] 1.2 [-] 1.2 1.1 1.1 measured 1.0 measured 1.0 /N /N 0.9 0.9

simulated 0.8 simulated 0.8 N N 0.7 0.7 0.6 0.6 0.5 C 0.5 D 0 100 200 300 400 500 02468101214161820 Field data: stem number [ha-1] Time [y] Figure 5.2: Benchmark tests.Relative variation in basal area BA (A, B) and stem number N (C, D) of simulation against field data.A, C: Final variation after maximum simulation time (= length of observation) as a function of field measurement in last enumeration.For each of the permanent sampling plots (PSPs) data for each plant functional type PFT 1-4 and sum are plotted.B, D: Variation as function of simulation/observation time.Plotted are total basal areas and total stem numbers for PSPs in Garinono, Gunung Rara, Segaliud Lokan and Sepilok. For information on PSPs see Table 5.1. secondary tropical rain forest and find no 10cm does not differ from average mortal- significant differences. The fact that all our ity. The increasing mortality with time analysed average mortality rates do not fall in three forest reserves indicates that older in this range has to be discussed. trees might die faster than the average rate. We did not consider ingrowth of trees af- However, the typical fluctuations in mortal- ter the first enumeration for the reason of ity as seen in Garinono show no trend at evaluating our model without recruitment. all. Another reason for mortality increase This was also done assuming that the mor- might be an eight month long drought with tality rate for small trees with a dbh around no rainfall at all in the years 1982/83 in parts of Sabah (Leighton & Wirawan 1986; 60 Chapter 5

Richards 1996). processes give us very few arguments on Gunung Rara’s very low mortality of 0.25 how to parametrise them. %y−1 over ten years seems very unreal- The spatial resolution used in the inven- istic. As mentioned earlier this forest re- tories (each PSP plot has an area of 1 ha serve lies on poor sites and in higher eleva- divided in 25 patches of 20 m × 20 m) and tion, and one might expect a mortality rate in the model is the same. For that reason even higher than average. It might be that competition processes for light and space within the process of enumeration tree la- are simulated as accurately as possible in bels of dying trees were used several times Formind. However nature is not as homo- leading to an underestimation of mortality geneous in tree distribution as we assume in rates. Mortality rate in Segaliud Lokan2, the model. Shading processes might there- which was zero over ten years, seems rea- fore have a more significant influence on in- sonable, because of the small area of only 1 dividual tree growth. The aggregation of ha. On this scale extreme values might oc- field inventory data into diameter classes cur. Pioneer species show higher mortality with a width of 5cm as done in the ini- rates as expected. tialisation results in slight overestimation of Even if the mortality rates are question- basal area at the beginning of the simula- able in comparison with literature, they are tions (time = 0 y) as seen in Fig. 5.2. a result of the data analysis of the PSPs and The deactivation of the recruitment sub- it is reasonable to use them as parameter model has only a small influence on the sim- values for simulations performed for bench- ulated stand dynamic, if short time scales mark testing. are considered as in this paper. With the approximation of an upper diameter incre- mentof1cmy−1 for non-pioneer species Model Evaluation without light competition (Ong & Kleine 1995; Huth et al. 1998) ingrowing trees with a dbh of 10 cm will not exceed a dbh of 30 The model in the version documented here cm within 20 years. In all competition pro- is more complex in terms of competition and cesses trees are only influencing other trees tree growth processes than any other rain of approximately the same size or smaller. forest growth model known to the authors The ingrowth would, if activated, not effect (e.g. Kohyama 1993; Ong & Kleine 1995; the growth of the big trees in stand simula- K¨urpick et al. 1997; Huth et al. 1998; Liu tion and would therefore lead to only small & Ashton 1998). However one might find differences in the simulated stand develop- models for temperate forests (e.g. Bugmann ment. Because PSPs data were analysed 1996a) or even monocultures (e.g. Bossel without recruitment as well (only trees la- 1996a), which enhance certain features not belled during first enumeration were consid- included in Formind (e.g. soil properties, ered further), accuracy of the comparison nutrient circles, weather, daily resolution, should not be weakened. climate gradients). One might therefore think the model itself is simpler than to- Considering mortality without the pro- day’s forest growth models. For that reason cess of falling trees influences only the spa- we like to highlight the general differences tial distribution of tree mortality. Because in complexity between growth models for mortality effects of gap creating falling trees temperate and tropical forests and problems are implicitly included in the field data, av- arising with a more detailed model struc- erage mortality is parametrised correctly. ture. Beside the very high number of tree Locally high mortality rates would effect species in the tropics (over 400 per hectare recruitment pattern in this area, but may in Sabah) the non existing data on those be ignored due to inactive recruitment sub- Comparison of measured and simulated growth on permanent plots in Sabah 61 model. ments might be one reason for the unex- Applying the model with four of the five pected data. Additionally, as our main ob- PFT, as indicated in the model description, jective in this benchmark testing was to val- has no effects on model results. As only idate our growth model, an assumed av- trees with a dbh ≥ 10 cm are considered erage mortality for the whole observation  −1 in the results, these fifth PFT would not time =0.0%y in Segaliud Lokan2 would change simulated basal area or stem num- largely cover tree growth effects. ber directly, trees of the fifth PFT have a The good agreement between simulation maximum diameter of 2 cm. The only con- and field data are first hints that For- sequences might be indirect competition ef- mind1.1 is an adequate tool for simulating fects on small ingrowing saplings of other the growth of tropical rain forest not only PFTs. on a short time scale of some decades but The fixed allometric relations between also for long time forest development. Re- different variables like tree height and crown sults over simulation periods of 100 years length in the model do not allow the trees and more were already published in K¨ohler to adapt crown structures to their spe- & Huth (1998a, b). cific individual environment. Crowns might Given the wide variation in calculated overlap with those of neighbouring trees in mortality rates, one might ask how safe is dense patches. Therefore crowding mortal- it to use an average or published rate to ity is needed as a regulating process. This predict forest development. With a sensi- is an additional mortality which leads to tivity analysis, which highlights the influ- a constant underestimation of stem num- ences of different parameters, the impor- bers. Simulations without this regulation tance of mortality and the acceptable pa- end with higher deviation in basal area from rameter range can be analysed (Huth et al. measured data. The process of crowding 1998). Investigations show that mortality mortalitycoversonlyasmallpartofthe is important for model behaviour, but pa- total mortality (0.05-0.25 % out of 2-5 %) rameter values might vary reasonably with- but is important for a realistic simulation of out changing results in general. Thus, using basal area. more general independent data for bench- The simulations for Gunung Rara agree mark testing might influence the accuracy the least with field data. These deviations only slightly. might be caused by the higher elevation and Because of the variety of different stock- poor site conditions in Gunung Rara. Con- ings represented in the PSPs the tests show ditions found in Gunung Rara fall out of the that Formind1.1 is applicable within good present application range of Formind1.1. site conditions on every possible level of for- However with a more detailed description est degradation. Tests shown in this paper of individual tree growth as a function of together with results for the simulation of site conditions Formind has the potential primary forest (K¨ohler & Huth 1998a, b) to simulate stands like Gunung Rara with evaluate it as an accurate tool for estimating similar accuracy as the other forest reserves. the effects of logging operations on tropical To exclude data of the last years in Se- forest ecosystems (future work). galiud Lokan2 from further testing was mo- tivated by the small spatial scale of only 2 ha and the large temporal differences in − Conclusion mortality rates (0.0 % y 1 for 10 years and rates between 20 and 40% y−1 in the last years). We think errors in field measure- Beside Ong & Kleine (1995) and Liu & Ash- ton (1998), which all used data from per- 62 Chapter 5 manent sampling plots to parametrise their on individual people. This height ranged models, a detailed comparison of growth between 120 cm and 160 cm in a comprehen- data with model results has not been per- sive literature survey (Brokaw & Thompson formed so far. A comparison is limited to 2000). Considering the conical form a trunk the quality and observation period of avail- this might have also influenced our compar- able data sets and therefore the case study ison of modelled and measured tree growth. in this paper is limited to time periods up A comprehensive study, which assesses the to twenty years without consideration of re- growth of tropical rain forest trees and its generation. However, for the development issues for modelling was published recently of models, which estimate long term tenden- (Clark & Clark 1999). The authors high- cies in tropical rain forests with and without light the importance of growth models for anthropogeneous influences even those lim- management questions and possibilities of ited data are of importance for model eval- model evaluation with field data. uation. Thus, the benchmark test gave us The same comparisons of the model re- indications as to where the limits of model sults and the field data were repeated with application are. Only with this knowledge the new grouping concept (Chapter 3) and can an application of forest growth models the current version of the model, For- to questions of management practise be vi- mind2.0 (Fig. 5.3). The same concept than able. in Fig 5.2 was used for the visualisation of results. Now, only total values (A,C,D,F) or of species groups with different succes- Acknowledgement sional status (B,E) are plotted. Note dif- ference in y-axes between Fig. 5.2 and 5.3. We like to thank the Forest Research Overall simulation match field data slightly Centre, the Forestry Department Sabah, better, especially for total values (C,F) or Malaysia, and the Malaysian-German Sus- stem numbers (D-F). If single PFTs are con- tainable Forest Management Project and sidered (B,E) differences between modelled their members, especially M. Kleine. and measured growth was highest. The Thanks to R. Glauner for coordinating number of individuals per PFT is smaller between Malaysian and German projects. than in original analysis. Thus, the death Parts of the work for this paper were sup- of few large trees might lead to relative large ported by the Deutsche Forschungsgemein- errors in estimated biomass. schaft (DFG). We also owe thanks to two anonymous reviewers, who gave critical but very useful comments.

Additional remarks and comparison with new grouping - not included in the article

Additionally, it should be mentioned that field measurements of stem diameters might be biased due to the fact that the defini- tion of the height (called at breast height) at which measurement took place depended Comparison of measured and simulated growth on permanent plots in Sabah 63

2.5

[-] 2.0

data 1.5

/BA 1.0 sim 0.5 BA 0.0 A B C 0 5 10 15 20 25 0510 0 5 10 15 20 2 -1 2 -1 Time [y] BAdata [m ha ] BAdata [m ha ] 2.5 2.0 [-] 1.5 data

/N 1.0 sim

N 0.5 0.0 D E F 0 100 200 300 400 0 50 100 150 0 5 10 15 20 -1 -1 Time [y] Ndata [ha ] Ndata [ha ]

Figure 5.3: Benchmark tests.Relative variation in basal area BA (A-C) and stem number N (D-F) of simulation against field data.A,B,D,E: Final variation after maximum simulation time (= length of observation) as a function of field measurement in last enumeration.For each of the permanent sampling plots (PSPs) data of (A,D) total sum and partial sum of successional groups, (B,E) each PFT, (symbols distinguish only successional status) are plotted (total: ∗;1: ◦;2: ✷;3: ×).C,F: Variation as function of simulation/observation time.Plotted are total basal areas and total stem numbers for PSPs in Garinono (◦), Gunung Rara (✷), Segaliud Lokan (× and ) and Sepilok ( ).

Chapter 6

Sustainable timber harvesting in Venezuela: a modelling approach

Ludwig Kammesheidt∗, Peter K¨ohler† and Andreas Huth† ∗ Institut f¨ur Waldbau, Abt. II: Waldbau der Tropen, Universit¨at G¨ottingen B¨usgenweg 1, D-37077 G¨ottingen, Germany † Center for Environmental Systems Research, University of Kassel Kurt-Wolters-Str. 3, D-34109 Kassel, Germany

Abstract

Reliable data on growth and yield of logged-over forest needed to determine sustainable cutting cycles are widely missing for the tropics.We used the process-based model For- mind2.0 to analyse growth and yield of logged-over forest in Venezuela under different logging scenarios over a period of 240 years and compared results with unlogged stands. For model evaluation a detailed stability and sensitivity analysis was done.In the ab- sence of further logging, the logged-over stand approached the stand structure of mature forest in terms of bole volume and basal area after about 50-100 years.Thirty year cut- ting cycles with conventional logging methods and net extraction volumes of 45 and 60 m3 ha−1 cycle−1 did not provide sustainable yields under either of two minimum felling diameters (35 or 50 cm) applied.Only the 60 year cutting cycle provided sustainable yields under conventional and reduced-impact logging with the different minimum felling diameters and a range of net volumes extracted (30-60 m3 ha−1 cycle−1).With the longest cutting cycle (60 years), bole volume recovered to levels similar to the mature unlogged stand, but the species composition was very different.Scenarios with reduced-impact log- ging provided a significantly higher timber volume than under conventional logging.The conservation of forest resources will only be possible with long cutting cycles (at least 60 years) in combination with reduced-impact logging. Keywords: Formind2.0, logging scenarios, plant functional types, simulation, trop- ical forest. Journal of Applied Ecology (2000), in press. 66 Chapter 6

Introduction veloped into the individual-oriented For- mind2.0 model which provides more details about forest dynamics compared to its pre- The determination of sustainable cutting decessor (cf. K¨ohler & Huth 1998a; K¨ohler cycles and annual allowable cuts is crucial to et al. 2001). forestalling further degradation of tropical In this paper we test whether prescribed timber resources. In Latin America, where cutting cycles at intervals of 30 years in extensive areas of natural forest have been Venezuela provide sustainable timber yields granted to concession logging in the last few under the currently uncontrolled logging years (FAO 1997), little is known about the methods by modelling a logged-over stand long-term dynamics of logged-over stands. with Formind2.0. We also test whether The silvicultural system CELOS in Surinam controlled logging would allow cutting cy- is the only experiment to date giving a fairly cles to be shortened and still allow sus- comprehensive idea about the management tainable timber yields. Finally, we simu- tools needed to sustain timber production late the impact of different cutting cycles, in natural forests (de Graaf 1986). How- logging methods and extraction volumes on ever, the suggested cutting cycle of 20 years the ingrowth and mortality of commercial with a restricted net removal of 20 m3 ha−1 species, and species composition. The re- in the CELOS system is based on inten- sults are compared with stands that have ei- sive post-logging treatments. Another long- ther never been logged or logged only once. term study on growth and yield in logged- over and untreated plots in the Brazilian Amazon revealed a low volume increment of timber species, indicating that short cutting Material and methods cycles are unlikely to be sustainable (Silva et al. 1995, 1996). Study site Numerous forest models have been de- veloped to bridge the gap between gener- The 7000 ha ’Estaci´on experimental’ of ally short-term empirical data on forest dy- the Universidad de Los Andes is part of namics and the need for reliable long-term the ’Reserva Forestal de Caparo’ located in the western Venezuelan plains (7◦30’N, yield prediction. Vanclay (1989, 1994), ◦ for example, simulated logged-over forest 70 45’W; elevation 100 m). The mean in North-Queensland, Australia, with stand annual rainfall is 1750 mm, with a pro- models, based on tree density and basal nounced dry season from December to March (monthly precipitation < 50 mm). area. K¨urpick et al. (1997) used the gap- ◦ model Formal to simulate the stand de- The average annual temperature is 24.6 C. velopment of logged-over Malayan diptero- The soils are of alluvial origin and relatively carp forest. Process-based models, simu- fertile compared to other neotropical low- lating physiological processes under chang- land areas (Hase & F¨olster 1982). Inten- ing environmental conditions, are another sive sedimentation by river flooding has re- approach to forest modelling (Landsberg sulted in a fine-scaled microrelief, ranging & Gower 1997). The Formix model and from sandy levee sites to clay-rich depres- its successor Formix3 which were success- sions. Inundation of depressions lasts up to fully tested in Malayan dipterocarp and nine months, depending on soil texture, re- peat swamp forest, respectively (Bossel & lief and ground water level (Franco 1979). Krieger 1991; Huth et al. 1994, 1998; Ditzer Levee sites, by contrast, are well-drained et al. 2000) combine the advantages of throughout the year. conventional forestry models with process- The high-forest of Caparo, naturally based models. Formix was recently de- distributed on well-drained and poorly Sustainable timber harvesting in Venezuela 67 drained sites up to an inundation pe- 37 ha per 100 ha logging unit). There- riod of six months, is classified as ’moist after, 30 plots of 400 m2 (1.2 ha in each of semi-deciduous’ (Lamprecht 1989). On the logged stands) were put at systematic both sites, the palm Attalea maracaibensis distances along transect lines using tape (Mart.) Burret (Arecaceae) and the large and compass. In MF (total area ca. 30 tree Bombacopsis quinata (Jacq.) Dugand ha), 25 plots were established in the same (Bombacaceae) are the predominant com- way. In all stands, a roughly equal num- ponents in terms of basal area share (Franco ber of plots were laid out on well-drained 1979; Kammesheidt 1994). A total of and poorly drained sites, respectively. Trees 53 tree and palm species ≥ 10 cm d.b.h. and palms ≥ 10 cm d.b.h. were mea- have been recorded and both the well- sured; seedlings and saplings were sampled drained and poorly drained sites are similar in subplots nested in the major sampling in stand structure and species composition unit. Results refer to the stand 5 years (Kammesheidt 1994, unpubl. data). The after logging (LG5) because it represents number of deciduous trees increases from a damage level, judged by the proportion well-drained to poorly drained sites (Franco of landings in the overall area logged, be- 1979). tween those of the other two logged stands Logging, carried out on both site types, (cf. Kammesheidt 1994). started in the early 1970s. In the begin- ning, only B. quinata, Swietenia macro- Species grouping phylla King and Cedrela odorata L. (both Meliaceae), and Cordia thaisiana Agostini Shrub, tree and palm species (to- (Boraginaceae) were logged. Currently, tal number: 127 spp., species list more than 20 species are logged (L. Lugo, available online: http//:www.usf.uni- pers. comm.). The shift from few to kassel.de/archiv/dokumente.en.htm, and in many merchantable species and increasingly Appendix Table B.4) were assigned to 12 mechanised logging has resulted in an in- different plant functional types based on crease in the level of damage on residual successional status and maximum height stands compared with stands logged in the at maturity (Table 6.1), following an ap- 1970s. Although logging is carried out by a proach developed for rain forest in Malaysia local sawmill owner under the supervision of (K¨ohler et al. 2000b). The successional university staff, no efforts have apparently status of species was determined by their been made to reduce the impact of logging gap association at juvenile stage, spatial on residual stands. About one-third of the pattern and ability/inability to persist in individual logging unit (100 ha) is actually closed mature forest as adult individuals affected by logging; in this third, on aver- (Kammesheidt 2000; unpubl. data). Simi- age, 10 trees with a standing bole volume lar functional groups are defined by Swaine 3 −1 of 66.5 m ha are removed (Kammesheidt & Whitmore (1988), Manokaran & Swaine 1998). No post-harvest treatments are car- (1994) and Thomas & Bazzaz (1999). ried out.

Description of the model Data set Formind2.0 is an individual-oriented for- The data base for modelling was an inven- est growth model (K¨ohler & Huth 1998a; tory of logged (5-, 8-, 19 years after tim- K¨ohler et al. 2001) to simulate stand devel- ber harvest) and unlogged stands (hereafter opment under certain scenarios, e.g. differ- ’mature forest’: MF), made in late 1991. ent cutting cycles. The model includes tree The actual logged area was delimited (27- growth, competition, mortality and regen- 68 Chapter 6

Table 6.1: Autecological characteristics of plant functional types (PFTs) in the Caparo forest, Venezuela. Height and d.b.h. range at maturity. Height group (HG) and successional status (SS).Total number of species (No) in the different PFTs.Relative abundance of trees ( >10 cm d.b.h.) in the mature forest (MF) and the stand 5 years after logging (LG5).

Plant functional type Height D.b.h. PFT HG SS No MF LG5 (m) (cm) index (%) (%)

Mid successional shrub spp.1-10 2-10 1 1 2 18 - - Late successional shrub spp.1-10 2-10 2 1 3 7 - -

Small early successional spp.10-15 10-25 3 2 1 3 - 2.9 Small mid successional spp. 10-15 10-25 4 2 2 11 3.7 2.7 Small late successional spp. 10-15 10-25 5 2 3 17 0.3 4.6

Medium-sized early successional spp. 15-30 20-70 6 3 1 2 1.0 13.1 Medium-sized mid successional spp. 15-30 20-70 7 3 2 11 6.0 9.9 Medium-sized late successional spp. 15-30 20-70 8 3 3 26 17.4 14.2

Large mid successional spp. 30-40 60-100 9 4 2 16 14.9 8.3 Large late successional spp. 30-40 60-100 10 4 3 12 15.6 9.1

Small palm sp.1-15 5-10 11 5 4 1 - - Medium-sized palm spp. 15-30 10-50 12 6 4 3 41.0 35.1

eration. The main processes including im- Tree growth and competition provements on former versions are described Within a single patch, the model calculates below (functional relationships are found in stand development based on cohorts of trees Table 6.2). belonging to the same plant functional type. Spatial structure A cohort i is characterised by the number The model describes tree competition in of trees Ni and by the size of one represen- patches. These patches have the size typical tative tree. Using allometric relationships, of natural gaps created by the fall of large the size of a tree can be expressed in terms trees (=400 m2, van der Meer & Bongers of its above-ground biomass Bi,heighthi, 1996). The model follows the gap-approach or diameter at breast height di. A form (Botkin et al. 1972; Botkin 1993; Shugart factor is applied which takes the difference 1998) and is therefore spatially non-explicit. from an idealistic cylindrical stem into ac- Tree positions of falling trees are deter- count. Tree height is calculated from diam- mined randomly within single patches. In eter. The crown projection area f is calcu- contrast to most gap-models, Formind2.0 lated from the proportionality of stem di- simulates a shifting stand mosaic. Thus, ameter to crown diameter dc (Rollet 1978; several contiguous patches (5 × 5patches Poker 1993). Crown length is proportional per hectare, each a 20 × 20 m2 in size) are to tree height (Burgess 1961; Poker 1993), simulated simultaneously (Smith & Urban using a constant factor. Leaf area l is a 1988; Urban et al. 1991). function of diameter but must be corrected to avoid unrealistic high values in leaf area Sustainable timber harvesting in Venezuela 69

index LAI = f/l, which should not exceed tenance respiration Rm(B) and growth res- a certain value LAIM (Ashton 1978). The piration RG (Ditzer et al. 2000). This leads bole volume is calculated by the stem wood to our main growth equation fraction τ, wood density ρs,andthegeo- dBi metrical relation of a truncated cone from = P˜i · q(1 − R ) − R (Bi)(6.1) dt G m biomass Bi. Using these allometric rela- tionships the distribution of individual tree Water balance is not included in the crowns and their leaf area in the canopy model. The calculation of tree growth is are calculated in horizontal canopy layers performed in annual time steps. of 0.5 m. Competition is modelled in terms of com- The growth of an individual tree is based petition for light as described above and on its carbon balance. Calculations include competition for space as described below photoproduction of the trees and assimilate concerning mortality. losses due to respiration, litter-fall and fine root decay. Within a patch the light at- Mortality Mortality is modelled on an annual basis tenuation Ii downwards in the canopy is calculated from light intensity above the at a basic mortality rate MB.Tothisis added a diameter-dependent mortality MD, canopy I0 and the light extinction coeffi- cient k with respect to the absorption of tree which is zero above a threshold diameter crowns. The dependence of specific pho- dt = MD1. Thinning is assumed to occur in dense patches. Mortality is modelled as tosynthetic productivity Pi on irradiance ≥ is modelled using a Michaelis-Menten-type stochastic event. Senescent trees ( 10 cm d.b.h.) die and fall with probability W ; light response curve. Photoproduction P˜i is calculated from the tree’s leaf area and its they knock down smaller trees in neighbour- ing patches and create gaps. The number of specific productivity Pi by integrating down the canopy of the tree in question (Monsi & trees NF destroyed from the total number Saeki 1953). Differences between wet and Np in the target patch p is calculated from dry season y are considered in terms of dif- the crown projection area fF of the falling fF tree relative to patch size A (NF = Np A ). ferent light intensity I0y, the different length of daily photoactive period SDy,andthedif- Regeneration ferent length of seasons SSy. We assume an Seed germination depends on minimal light increasing limiting effect of water transport intensities IS on the forest floor. It is as- deficiencies with increasing tree height. Ac- sumed that intact forest surrounding the tual productivity is calculated by applying simulation area supports a constant seed in- a size-dependent limitation factor q(di)ac- put NS. Incoming seeds are added to a seed di 2 i − − D · cording to q(d )=1 (1 q M ) ( DM ) , pool, which takes into account the variance where DM is the maximum diameter of trees in the length of dormancy (MS) between and qDM is the limitation factor at max- plant functional types (cf. Garwood 1983, imum tree height (corresponds to the ag- 1989). ing factor cs of Landsberg & Waring 1997). With the condition of no tree growth at Parameters and initial condi- maximum diameter, qDM can be calculated from the parameter set. Assimilate losses tions are estimated in relation to tree biomass (Kira 1978; Yoda 1983). Losses are com- Table 6.3 contains the parameters used for posed of root decay, litter-fall and respira- the simulations. Data on the light environ- tion of tree organs and leaves. We distin- ment are drawn from Veillon (1989) and guish between a biomass-dependent main- Kammesheidt (unpubl. data). Most allo- metric relations (h = f(d),cP) are based 70 Chapter 6

Table 6.2: Description of parameters including functional relationships.

Parameter Description

Environmental parameters k Light extinction coefficient I0 Light intensity above canopy SD Daylength SS Length of wet/dry season

Establishment parameters DS Initial diameter of seedlings ISs Minimal light intensity for germination NSs Ingrowth rate of seeds into seed pool

Mortality parameters MB Basic mortality rate MS Mortality rate of seeds MDj Size dependent mortality rate (MD = MD0 − MD0/MD1 · d) W Probability of a dying tree to fall

Tree physignomic parameters DM Maximum diameter cP Crown length fraction τ Fraction of stemwood biomass to total aboveground biomass h0h and h1h Height = f(diameter) (h = d/(1/h0h + d/h1h)) γ2 γj Form factor = f(diameter) (γ = γ0 · exp(γ1 · d )) f2 fj Crown diameter = f(diameter) (dc =(f0 + f1 · d ) · d) 2 3 lj Leaf area = f(diameter) (l = l1 · d + l2 · d + l3 · d ) LAIM Maximal leaf area index of single tree

Biomass production parameters PM Photosynthetic capacity in light response curve α Photosynthetic efficiency in light response curve ρ Stem wood density 2/3 r0l and r1l Respiration = f(biomass) (Rm(Bi)=r0l · B + r1l · Bi) RG Specific growth respiration rate ss m Leaf transmittance g Conversation factor gCO2 to godm

on data derived from Kammesheidt (1994, Malaysia. Data on the photosynthetic re- unpubl. data). The form factor γ,leaf sponse of plant functional types to different and crown area to diameter relations are light levels are given in Oberbauer & Strain taken from measurements of Kato, Tadaki (1984). The wood density of species was de- & Ogawa (1978) and Kira (1978) in Pasoh, termined at the Institute of Wood Technol- Sustainable timber harvesting in Venezuela 71

Table 6.3: Parameter estimates for the simulation of the Caparo forest, Venezuela.Param- eters with subindex vary according to season(y), successional status (s), potential height (h) (corresponding to SS and HG in Table 6.1), or different functional coefficients (j).

Name Special Units Values

Environmental parameter k [-] 0.7 −2 −1 a I0y wet dry [µmol(p) m s ] 816.0 1005.0 SDy wet dry [h] 12.0 12.0 SSy wet dry [-] 0.67 0.33 Establishment parameter DS [m] 0.01 −1 −1 NSs s=1-4 [ha y )] 500 200 25 50 ISs s=1-4 [fraction of I0y] 0.05 0.01 0.01 0.01 Mortality parameter −1 MBs,h s=1; h=1-6 [y ] 0.00 0.12 0.08 0.00 0.00 0.00 −1 MBs,h s=2; h=1-6 [y ] 0.06 0.05 0.035 0.03 0.00 0.00 −1 MBs,h s=3; h=1-6 [y ] 0.05 0.04 0.03 0.025 0.00 0.00 −1 MBs,h s=4; h=1-6 [y ] 0.00 0.00 0.00 0.00 0.01 0.01 −1 MSs s=1-4 [y ] 0.1 0.5 1.0 1.0 −1 −1 MDj j=0-1 [y ,m ]0.40.2 W [-] 0.40 Tree physignomic parameter DMs,h s=1; h=1-6 [m] 0.10 0.25 0.70 1.00 0.25 0.40 DMs,h s=1; h=2-3 [m] 0.13 0.35 cp [-] 0.358 τ [-] 0.7 −1 h0h h=1-6 [cm m ] 1.63 1.63 1.41 1.50 0.22 0.22 −1 h1h h=1-6 [m ] 19.9 19.9 35.7 45.4 325.7 325.7 −1 γj j=0-2 [-, cm , -] 2.575 -1.409 0.0358 fj j=0-2 [-, -, -] 0.132 0.933 -0.6615 m m m lj j=1-3 [ cm , cm2 , cm3 ] 3.197 0.0684 -0.000379 LAIM [-] 2 Biomass production parameter −2 −1 a PMs s=1-4 [µmol(c) m s ] 27.7 11.3 6.8 6.8 −1 a αs s=1-4 [µmol(c) µmol(p) ] 0.043 0.043 0.043 0.043 −3 ρs s=1-4 [todm m ] 0.24 0.69 0.69 0.75 3/2 r0s s=1-4 [todm] 0.20 0.06 0.05 0.04 r1s s=1-4 [-] 0.60 0.02 0.015 0.04 RG [-] 0.25 m [-] 0.1 −1 g [godm gCO2] 0.63

a p: photons; c: CO2 ogy and Wood Biology of G¨ottingen Univer- literature (Swaine 1989; Condit et al. 1992; sity. Parameters for respiration processes 1995a, b; Carey et al. 1994; Phillips & Gen- (r0s and r1s) were investigated using param- try 1994; Silva et al. 1995; van der Meer & eter variation to gain realistic diameter in- Bongers 1996). Mortality M is correlated crement values for different size classes and to the diameter growth rate gd and maxi- light conditions. Mortality and ingrowth mum size dmax such that ω = dmax/gd · M rates correspond to typical values found in is roughly constant. Otherwise the num- 72 Chapter 6 ber of large trees would be overestimated caltofielddata. (ω<<1) or only small trees would occur The sensitivity of 28 result variables (ω>>1) (Chave 1999b). (same as for stability analysis) to parameter From data sets of the two stands cho- variations was investigated by varying the sen for simulations (MF, LG5), 25 patches 60 parameters in Table 6.3 within their re- (of 400 m2 each) — randomly chosen in alistic range (values found in the literature thecaseofLG5—wereclusteredtoform and within physical boundaries). Seven the initial data set for one hectare. The simulations over 4 hectares and 100 years in functional groups were then aggregated into the mature stand were performed for each different cohorts regarding their diameter varied parameter, including one simulation (d.b.h. class of 5 cm). To minimise stochas- with the standard value. Mean values (v) tic effects in tree mortality, each simulation of the chosen result variables were averaged was performed for an area of 25 hectares. over the seven simulations. The coefficient The model was written in the program- of variation (CV ) from the resulting average ming language C++. Simulations were run (a) was chosen as an indicator to find out on a PC (400 MHz, system Linux), taking whether the individual variable responded | − | × on average 9 sec to simulate the growth of sensitively (CV = v a /a 100). 1 ha of rain forest over 100 years. Logging scenarios

Stability and sensitivity analy- On the basis of the logging practices in sis the study area documented by Plonczak (1989) and Kammesheidt (1994) and data For model evaluation a stability analysis from the literature (Hendrison 1990), the of the long-term model dynamics was per- following logging scenarios were simulated: formed. It compares field data of the ma- (1) during conventional logging one-third of ture stand (MF) with its dynamic variabil- the area was converted into roads and log- ity following the theory of long-term sta- landings (complete removal of the residual bility in tropical mature forests (Whitmore stand). Beside felling damages which de- 1988) without consideration of climate or stroyed an area proportional to the crown evolutionary changes. Stability indices were projection area of the cut tree, 55% of trees calculated by averaging the values of 28 re- in the felling area were killed due to skid- sult variables (leaf area index LAI, succes- ding operation. (2) By applying reduced- sion stages (GAP phase: no trees with h ≥ impact logging, landing areas diminished 15m in patch, BUILDING phase: trees to 12% and skidding damage was limited with h ≥ 15m in patch, and MATURE to 25%. If possible, trees were felled into phase: trees with h ≥ 30m in patch), existing gaps under both logging scenar- basal area BA, bole volume V of the whole ios. In the initial phase after logging (year stand, relative basal area of the different 1-10), mortality was two and three times plant functional types, successional status, higher than the normal rate found in ma- or height groups, total (N)stemnumber ture forest for reduced-impact and conven- and as function of diameter in four size tional logging, respectively. Cutting cy- classes (N0−30, N30−60, N60−90, N90)over cles of 30, 40 and 60 years, respectively, the last 100 from 240 simulated years and with three net harvest volumes (30, 45 and normalising them with their initial values 60m3 ha−1 cycle−1, respectively), apply- (cf. Huth & Ditzer 2000a). A stability in- ing a minimum felling diameter (MFD) of dex of 1.0 corresponds to a stable variable, 35 and 50 cm, were simulated. The sim- whose long-term averaged values are identi- ulation time was 240 years. The simu- Sustainable timber harvesting in Venezuela 73

4.0 GEN PFT SS HG ND 3.5

3.0

2.5

2.0

Stability index 1.5

1.0

0.5 S1 SS 2 SS 3 SS 4 SS GAP G2 HG 3 HG 4 HG 6 HG BUILDING BA V MATURE N N LAI N N N 0.0 3 PFT 4 PFT 5 PFT 6 PFT 7 PFT 8 PFT 9 PFT 10 PFT 12 PFT 0-30 30-60 60-90 90

Result variable

Figure 6.1: Stability index for the simulation of a mature forest stand.A stability index of 1.0 corresponds to a time averaged stable variable. The simulations were run over 240 years in an area of 25 ha, stability was analysed over the last 100 years (mean±SD).Result vari- ables are classified according to their level of information: GEN: General information refers to: total leaf area index LAI, succession stages (GAP phase: no trees with h ≥ 15m in patch, BUILDING phase: trees with h ≥ 15m in patch, and MATURE phase: trees with h ≥ 30m in patch), basal area BA and bole volume V of the whole stand. PFT: relative basal area of the different plant functional types (cf.Table 6.1). SS: relative basal area of the different successional stages (cf.Table 6.1). HG: relative basal area of the different height groups (cf.Ta- ble 6.1). ND: Stem number as function of diameter (N: all trees) for trees between 0 and 30 cm d.b.h. (N0−30), 30 cm and 60 cm d.b.h. (N30−60), 60 cm and 90 cm d.b.h. (N60−90)and above 90 cm d.b.h. (N90).For the stability analysis only trees ≥10 cm d.b.h. are considered. Consequently, PFT1,PFT2,PFT11,HG1andHG5 are omitted. lated area comprised 25 ha. To obtain the cutting cycle and hence, timber yield was given net volume, 30% harvest loss had unsustainable. to be added to the logged volume corre- sponding to gross volumes of 43, 65 and 3 −1 −1 86m ha cycle , respectively. From all Data analysis medium-sized and large, mid and late suc- cessional species at minimum felling diam- The G-test was applied to test for dif- eter, trees were randomly chosen for log- ferences in the proportion of successional ging until the target harvest volume was groups in bole volume between logging sce- reached. Early successional species are not narios. The z-test was used to compare the merchantable and did not attain a diameter means of ingrowth and mortality, and bole ≥ 35 cm (Kammesheidt 2000). In case of volume between different logging methods, understocking (if harvestable standing vol- and the unlogged and once logged stand ume is lower than the target gross volume (Fowler et al. 1998). Two-way ANOVA was for logging), logging was suspended for one employed to test for differences in bole vol- 74 Chapter 6 ume prior to logging within and between tions total bole volume (V ), and relative different logging scenarios. share of mid and late successional species (SS2,SS3) did not react sensitively. The parameter values for the initial diameter of Results ingrowing seedlings DS, general and various single mortality rates MB generated sensi- tive responses over the whole range of result Model evaluation variables. The parameters of the mortality module had a greater influence on simula- The results of the stability analysis of a ma- tion results than those of other modules. ture forest are shown in Fig. 6.1. Most gen- eral result variables were stable with sta- bility indices between 0.85 and 1.0, except Stand development in the ab- for those relating to the proportion of for- est in the gap and building phase. In fact, sence of logging the initial stand comprised no area in the gap phase, keeping this value at zero and The logged stand (LG5) approached the influencing directly the proportion of forest stand structure of mature forest in terms of in the building phase. Relative basal areas both basal area and bole volume after about were stable in PFTs with different succes- 50-100 years (Fig. 6.3). In the equilibrium sional status and maximum heights. Only phase, mid-successional species dominated SS1andHG2 showed values fairly far from over late successional species in both stands. stability (1.7 and 2.8, respectively). In the Early successional species disappeared be- individual PFTs the forest was more sta- fore the equilibrium phase was reached. In ble in PFTs with higher indices, referring both stands, palm species showed a con- to larger trees and higher absolute values in stant high share in basal area. The slight basal area. PFT3 was not found in the ini- decline of the bole volume and basal area tial site and therefore its stability index was in mature forest over the simulation period zero. PFT4, PFT5andPFT7 were unsta- indicates that stand structure was crowded ble with values between 2.2 and 3.2. In the at the time of sampling and thinned there- stem diameter distribution, different diame- after. ter classes were stable, total stem number N Figure 6.2: (opposite page): Sensitivity anal- declined to 80%, while the number of large ysis.Analysed were the sensitivity of 28 re- trees (N90) increased by 50%. sult variables describing the state of the for- In sensitivity analysis, the parameters est (for abbreviations see Fig.6.1).The re- were grouped according to the different sult variables were grouped in five groups: gen- parts of the model they belonged to eral structure (GEN), abundance of species groups (PFT) and successional groups (SS), (Fig. 6.2). Results showed a sensitive be- abundance of trees in different height classes haviour of the forest in gap phase (GAP ), (HG) and diameter classes (NG).Each pa- relative share of early successional species rameter was varied within the given range.The ≥ (SS1) and number of trees 90 cm (N90). grey scale of boxes indicates how sensitive a Their values varied throughout nearly the certain result variable reacts on variations of whole parameter range (CV>50%). Rela- a model parameter (for abrevations see Table tive share of some PFTs (PFT3 − PFT7) 6.3). Black: high sensitivity (CV > 50 %); were medium sensitive over a wide range grey: medium sensitivity (10 %

GEN PFT SS HG ND range

k 0.6-0.8 I00 666-966 I01 855-1155 I0 600-1200 SD0 6-18 SD1 6-18 Environment SD 6-18 SS 0.25-1.00 DS 0.01-0.30 NS1 30-120 NS2 20-80 NS3 4-16 NS4 5-20 IS1 0.1-0.4 I 0.01-0.08 Establishment S2 IS3 0.00-0.04 IS4 0.00-0.04 MB 0.00-0.05 MBs,1 0.00-0.05 MBs,2 0.00-0.05 MBs,3 0.00-0.05 MBs,4 0.00-0.05 MBs,5 0.00-0.05 MB1,h 0.01-0.20 MB2,h 0.00-0.05 M 0.00-0.05 Mortality B3,h MS1 0.01-2.80 MS2 0.1-1.0 MS3 0.1-1.0 MS4 0.1-1.0 W 0.01-1.00 D 0.02-0.20 Parameter Ms=1,1 DMs=1,2 0.1-0.4 DMs=1,3 0.4-1.0 DMs=1,4 0.7-1.3 DMs=1,5 0.1-0.4 DMs=1,6 0.2-2.8 c 0.2-0.5 Physignomy P 0.55-0.85 LAIM 1-3 PM1 12.7-42.7 PM2 5.3-17.3 PM3 3.8-9.8 1 0.03-0.05 2 0.03-0.05 3 0.03-0.05 1 0.1-0.5 2 0.55-0.85 3 0.55-0.85 4 0.60-0.95 r01 0.05-0.40 r 0.05-0.09 Production 02 r03 0.02-0.08 r04 0.03-0.06 r11 0.3-0.9 r12 0.01-0.04 r13 0.01-0.03 r14 0.03-0.06 RG 0.1-0.4 m 0.05-0.15 S1 SS 2 SS 3 SS 4 SS GAP BUILDING BA V G2 HG 3 HG 4 HG G6 HG MATURE N N LAI N N N F 3 PFT 4 PFT 5 PFT 6 PFT 7 PFT 8 PFT 9 PFT 10 PFT 12 PFT 30-60 0-30 90 60-90

Result variable

Figure 6.2: Caption are found on opposite page. 76 Chapter 6

Table 6.4: Bole volume of trees ≥ 30 (50) cm d.b.h after and prior to logging applying different logging scenarios (CON = conventional logging, RIL 0 reduced-impact logging) over a simulation period of 240 years.Minimum felling diameter (MFD) either 35 or 50 cm.Values for the mature forest (MF) are given as reference.If the individual gross bole volume was not reached, logging was omitted.Mean values and standard deviation for bole volume after/prior to logging were taken from the different number of logging events (n=4-8).Mean annual bole volume increment was averaged over the simulation period.

Mean annual Net volume Bole volume (m3 ha−1) Times bole volume increment Logging Cutting extracted after logging prior to logging logging (m3 ha−1 y−1) method cycle (y) (m3 ha−1 cycle−1)mean±SD range mean±SD range omitted mean±SD

MFD=35cm; trees ≥35 cm d.b.h.

MF 250±14 231–274 0.0±1.2

CON 30 30 33±16 21–71 77±16 65–116 – 3.5±5.9 CON 30 45 43±34 0–90 109±34 65–157 2 3.2±5.3 CON 30 60 60±17 29–71 147±18 116–159 4 3.2±5.0

RIL 30 30 93±11 71–101 138±11 116–147 – 2.7±2.4 RIL 30 45 54±5 45–61 121±6 112–128 – 3.3±2.4 RIL 30 60 25±5 14–30 112±5 101–117 – 3.8±2.6

CON 40 30 79±12 70–105 125±12 115–151 – 3.0±4.8 CON 40 45 44±18 30–84 111±18 97–151 – 3.5±5.3 CON 40 60 19±20 6–64 106±20 93–151 – 3.6±5.5

RIL 40 30 119±7 105–127 165±7 151–172 – 2.3±2.2 RIL 40 45 89±4 83–94 157±5 150–162 – 2.8±2.3 RIL 40 60 64±3 59–67 151±3 146–155 – 3.2±2.3

CON 60 30 133±9 123–146 178±9 168–191 – 2.2±4.0 CON 60 45 108±9 100–123 176±9 168–191 – 2.6±4.0 CON 60 60 80±13 72–103 169±13 161–191 – 2.9±4.2

RIL 60 30 152±4 146–156 199±5 191–202 – 1.8±2.1 RIL 60 45 128±3 123–131 196±3 191–199 – 2.1±2.0 RIL 60 60 109±4 103–113 196±4 191–202 – 2.4±2.1

MFD=50cm; trees ≥50 cm d.b.h.

MF 208±12 192–232 0.0±1.3

CON 30 30 5±4 2–16 49±5 44–62 – 1.8±1.9 CON 30 45 70±4 67–77 138±4 135–145 4 2.2±3.1 CON 30 60 45±7 39–57 134±7 127–145 4 2.4±3.4

RIL 30 30 51±16 16–65 96±16 62–111 – 2.1±1.8 RIL 30 45 31±19 17–77 99±19 84–145 1 2.4±1.7 RIL 30 60 42±29 2–72 130±30 88–161 3 2.2±2.2

CON 40 30 27±8 19–45 72±8 64–90 – 1.6±1.8 CON 40 45 58±41 10–104 124±42 75–171 2 2.0±2.8 CON 40 60 52±33 5–77 139±35 90–165 3 1.8±2.9

RIL 40 30 66±10 45–75 112±11 90–121 – 1.7±1.4 RIL 40 45 39±9 23–46 106±9 90–114 – 2.1±1.5 RIL 40 60 27±32 5–91 114±33 90–180 1 2.1±2.0

CON 60 30 98±4 93–104 144±4 138–150 – 1.9±2.9 CON 60 45 71±7 65–82 139±7 133–150 – 2.2±3.2 CON 60 60 45±10 37–62 133±10 126–150 – 2.4±3.4

RIL 60 30 119±10 104–131 165±10 150–177 – 1.6±1.9 RIL 60 45 91±6 82–97 160±6 150–165 – 1.9±2.0 RIL 60 60 69±5 62–74 157±5 150–163 – 2.1±2.1 Sustainable timber harvesting in Venezuela 77

40 ] -1

ha 30 2

20

10 Basal area [m 0 A B ] -1 250 ha 3 200 150 100 50 Bole volume [m 0 C D 0 100 200 0 100 200 Time [y] Time [y]

Figure 6.3: Development of the basal area (m2 ha−1) and bole volume (m3 ha−1) in unlogged (A, C) and logged stands (B, D) by plant functional type over a simulation period of 240 years in 25 hectares for trees ≥10 cm d.b.h. Total (solid bold line), early successional spp. (solid line), mid-successional spp.(broken dotted line) , late successional spp.(broken line) , palm spp.(long broken line).

Yield prediction under different a significant increase in commercial stock logging scenarios (z =37.1,P <0.01). By applying reduced- impact logging, timber harvest had to be suspended less often. Providing conventional logging with a MFD of 35 cm and 30 year cutting cycles, logging In logging scenarios with a sustainable would not take place two and four times, re- timber supply over the 240 year simula- spectively, over a simulation period of 240 tion period differences in bole volume af- years, if the net volume extracted (NVE) ter logging, corresponding to the different were either 45 or 60 m3 ha−1 cycle−1 (Ta- NVE diminished until the next cutting cy- ble 6.4). All other logging scenarios with cle (Table 6.4). This trend is largely due to a MFD of 35 cm provide a merchantable the higher bole volume increment with in- volume in any of the individual cutting cy- creasing NVE in most scenarios. However, cles. Conventional logging methods with within-group differences in the 40 year cut- a MFD of 50 cm and 30 year cutting cy- ting cycle prior to logging remained signif- cles are unlikely to be sustainable even with icant (F2,30 =8.9,P < 0.01), while differ- low NVEs of 30 m3 ha−1 because the over- ences were insignificant in the 60 year cut- all volume prior to logging was on aver- ting cycle for both MFDs applied (F2,30 < age only slightly higher than the minimum 1.8,P > 0.05). Differences between log- levelof43m3 ha−1 needed as gross volume ging methods and NVE prior to logging (Table 6.4, Fig. 6.4). An extension of the were significantly different in all scenarios length of cutting cycle to 40 years — but (F1,30 > 31.4,P <0.01). otherwise unchanged conditions — led to Even under the longest cutting cycle with 78 Chapter 6

] 3 -1 -1 3 -1 -1 3 -1 -1

-1 CON 30y 30m ha c CON 30y 45m ha c CON 30y 60m ha c 250 ha 3 200 150 100 50 Bole volume [m 0

] 3 -1 -1 3 -1 -1 3 -1 -1

-1 RIL 30y 30m ha c RIL 30y 45m ha c RIL 30y 60m ha c 250 ha 3 200 150 100 50 Bole volume [m 0

] 3 -1 -1 3 -1 -1 3 -1 -1

-1 CON 60y 30m ha c CON 60y 45m ha c CON 60y 60m ha c 250 ha 3 200 150 100 50 Bole volume [m 0

] 3 -1 -1 3 -1 -1 3 -1 -1

-1 RIL 60y 30m ha c RIL 60y 45m ha c RIL 60y 60m ha c 250 ha 3 200 150 100 50 Bole volume [m 0 0 100 200 0 100 200 0 100 200 Time [y] Time [y] Time [y]

Figure 6.4: Development of bole volume (m3 ha−1) for a range of logging scenarios by plant functional types over a simulation period of 240 years in 25 heactares for trees ≥10 cm d.b.h. The subheadings indicate logging conditions (method, cycle in years, intensity in m3 ha−1 cycle−1) with RIL: reduced-impact logging; CON: conventional logging.Minimum felling diameter was 50 cm.Total (solid bold line), early successional spp.(solid line), mid-successional spp.(broken dotted line), late successional spp.(broken line) . Sustainable timber harvesting in Venezuela 79

Table 6.5: Average annual ingrowth and mor- Ingrowth and mortality tality (%) of trees ≥ 35 or 50 cm d.b.h., respec- tively in unlogged (MF), one time (LG5), and Regular logging operations kept the stand several times logged stands under different log- under all logging scenarios in a building ging scenarios over a simulation period of 240 phase, indicated by the fact that ingrowth years.Minimum felling diameter (MFD) is ei- ther 35 or 50 cm.Logging methods are desig- dominated over mortality (Table 6.5). In- nated as either conventional (CON) or reduced- growth was significantly higher than mor- impact logging (RIL).Data are mean ±SD of tality in trees ≥ 50 cm d.b.h. under all log- time averaged values of scenarios with differ- ging scenarios (z>2.6,P <0.01), while no ent extraction intensities (n =3). significant differences were found in trees ≥ 35 cm d.b.h. (z<1.4,P > 0.05). Dif- ferences in ingrowth rates between logging Logging Cutting Ingrowth Mortality methods were insignificant, except for trees method cycle (y) ≥35 cm d.b.h. in the 40 year cutting cycle (z =2.36,P < 0.05). Although annual in- growth and mortality was higher under con- MFD=35cm; trees ≥35 cm d.b.h. ventional logging, significant differences be- tween logging methods were only found in a MF 2.7±0.5 2.7±0.3 few cases. Overall, the rate of ingrowth and LG5 3.0±2.5 2.8±2.6 mortality declined with longer cutting cy- cles irrespective of logging methods. How- CON 30 6.2±9.7 5.6±9.9 ever, even under a 60 years cutting cycle RIL 30 5.1±3.5 4.3±4.5 and reduced-impact logging, ingrowth and CON 40 6.0±9.4 5.5±9.8 mortality were significantly higher than in RIL 40 4.5±2.9 3.9±3.8 both mature forest and LG5 (z>2.82,P < CON 60 4.5±7.0 4.4±7.1 0.01). RIL 60 3.8±2.63.5 ±2.8

MFD=50cm; trees ≥50 cm d.b.h. Successional groups of species ± ± MF 1.6 0.8 1.6 0.5 Logging methods had a significant influence ± ± LG5 2.3 2.8 1.7 0.6 on the proportion of different successional groups in bole volume (G>6.6,P < 0.05) ± ± CON 30 5.7 7.9 3.5 9.1 (Table 6.6). The extension of the length of ± ± RIL 30 4.6 4.3 2.6 3.6 cutting cycles resulted in a decline of the ± ± CON 40 4.8 7.3 3.1 7.6 proportion of early successional species and ± ± RIL 40 4.4 4.8 2.5 3.4 an increase in late successional species un- ± ± CON 60 4.1 5.1 2.9 6.8 der both conventional and reduced-impact ± ± RIL 60 3.3 3.2 2.2 2.5 logging. This is also illustrated by the development of these successional groups of species along the 240 years simulation period (Fig. 6.4). Over the whole sim- reduced-impact logging and a low timber ulation period, the proportion of succes- 3 −1 extraction of 30 m ha , the average bole sional groups in LG5 differed insignificantly volume prior to logging was significantly from that of mature forest (G =2.4,P > lower compared to mature forest for both 0.05), whereas the bole volume was still diameter limits (z>44.0, P<0.01). significantly lower (z =13.1,P < 0.01). Reduced-impact logging kept the standing stock in any of the cutting cycles and MFDs 80 Chapter 6

Table 6.6: Proportion (mean±SD) of plant functional types summarised into successional groups (early (1), mid (2) and late (3) successional spp.; ≥10 cm d.b.h.) in bole volume (m3 ha−1,mean±SD) in unlogged (MF), one time (LG5) and several times logged stands under different logging scenarios over a simulation period of 240 years.Minimum felling diameter (MFD) is either 35 or 50 cm.Logging methods are desingated as either conventional (CON) or reduced-impact logging (RIL).Bole volume is given as mean annual value out of the average of different net volumes extracted.Data are mean ±SD of time averaged values of scenarios with different extraction intensities (n =3).

Successional groups (%)

Logging Cutting m3 ha−1 123 methods cycles (y)

MF 270±14 1±056±444±2 LG5 235±39 3±456±642 ±2

MFD≥ 35 cm d.b.h.

CON 30 144±44 29±18 57±17 14±12 RIL 30 154±3610 ±570±12 21±10 CON 40 144±39 28±18 56±1616 ±14 RIL 40 176±38 7±468±12 25±10 CON 60 179±44 15±13 63±15 22±13 RIL 60 201±42 5±465±11 30±9

MFD≥ 50 cm d.b.h.

CON 30 161±45 20±14 61±14 19±13 RIL 30 172±37 8±468±12 25±11 CON 40 173±47 16±13 65±15 19±12 RIL 40 181±40 7±468±12 25±10 CON 60 179±45 15±13 63±15 23±13 RIL 60 200±42 5±465±12 30±9

on a significantly higher level than conven- species composition to LG5 declined to an tional logging (z>1.83,P < 0.05,P < insignificant level (G =4.0,P >0.05). The 0.01). Within the same logging method, mean bole volume of all logging scenarios the lower MFD applied led to a significantly was significantly lower than in both LG5 lower standing stock in short cutting cycles and mature stand MF (z>9.0,P <0.01). (z>3.2,P < 0.01), except for reduced- impact logging with a 40 year cutting cy- cle. Only with the longest cutting cycle and reduced-impact logging, differences in Sustainable timber harvesting in Venezuela 81

Discussion elled one. Field data of a larger sampling area might be needed to reflect the accu- This study has been a two-step approach racy of the modelled composition of plant to forecast growth and yield of logged-over functional types. From our model analysis, forest, consisting of testing the model under we know that recruitment is the most im- undisturbed conditions and then simulating portant factor for species composition and logging scenarios. Following this, we discuss needs to be modelled in more detail in fu- the two working steps ture applications. The decline of the values of general variables could be explained by the missing gap phase. The initial data set Model performance under suggests that the analysed mature forest is undisturbed conditions a well-structured stand showing high values in basal area, bole volume and leaf area in- Although the structure and species compo- dex. The stability index of general variables sition of mature forest shift continuously on showed changes below 15%, indicating that a small scale (Wiens 1999), we think that the model and its parametrisation is stable a stability analysis is worth performing to and a suitable tool for further analysis of evaluate model performance. In fact, all logging scenarios in the Caparo forest. forest growth models known to the authors The sensitivity analysis was undertaken (cf. Liu & Ashton 1995) implicitly accept in realistic parameter ranges. This implies a so-called ”potential natural vegetation” that the response in the result variables (PNV), which represents the steady state as should not be too sensitive, otherwise a a product of model structure and parametri- model with that many parameters is not sation. None of them have tested the PNV suitable for a sound analysis. The results in detail against field data of mature forest. showed low sensitivity for most parameter The critical point of the stability test was variations. The sensitivity of the model whether the site we categorised as mature to variations in the mortality parameters forest was representative. In a few aspects, raises the question of whether these values the mature forest lacked this representativ- were chosen properly. However, simulated ity. For example, no area was in gap phase mortality rates in the mature forest are and consequently early successional species within the range observed in other neotrop- were rare. We consider these points cru- ical forests (Condit et al. 1992; Carey et al. cial for the stability analysis. However, a 1994). field inventory represents only one condition in space and time. A comparison with the time averaged dynamic as done in this anal- Logging scenarios ysis suggests that these points have a minor influence on the quality of the model results. Net timber volumes in the range of 30-60 For instance, the time averaged fraction of m3 ha−1 cycle−1 assumed in the scenarios gap area in mature forest or the proportion for the second and subsequent cutting cycles of early successional species showed small may be perceived as high having in mind differences to inital values (cf. Table 6.6). the traditionally low intensity of wood re- Two further trends need to be discussed: moval in Latin America compared to South- the decrease of most general variables be- East Asia (Plumptre 1996). In fact, the low the 1.0 level (cf. Fig. 6.1), and the in- first cutting cycle is selective focusing on crease of most PFTs indices far apart from the most valuable species i.e. B. quinata, S. this level. The latter trend results from macrophylla and C. odorata in areas where the fact that the initial species composition they occur clumped, resulting in low har- was fairly distinct from the averaged mod- vest volumes, if referring to the whole log- 82 Chapter 6 ging unit. As individuals of these species are moval of 20-84%. By simulating 240 years, found rarely below the minimum felling di- early successional species accounted on av- ameter MFD (Plonczak 1989; Kammesheidt erage with 83-95% to the basal area incre- 1994, 1998), logged-over stands are com- ment at any initial phase (1-10 yr) after posed of potential commercial species with logging in 30-years intervals, while Lozada a considerably lower market value (cf. Cen- (1998) found a mean percentage of 45. This teno 1995). Further depletion of the most suggests that by shorting cutting cycles over valuable species might result in an increase a longer period of time early successional of log prices of formerly unlogged or rarely species can become predominant after dis- logged species. However, even an increase of turbance. log prices for less acceptable species would Ingrowth and mortality declined with in- hardly offset the loss of valuable timber creasing length of cutting cycles (cf. Ta- species so that a higher volume must be har- ble 6.5). In the 60 year cutting cycle vested to keep the cost/benefit ratio of the with reduced-impact logging both param- first cutting cycle. To include all species eters reached average values in the range above the legal size in the present logging of 2.2-3.8% which, is similar to turn-over scenario is reasonable because this is being rates of trees ≥ 30 cm d.b.h. in old-growth done already in other concession areas in forest in Panama (Condit et al. 1992). We the western plains (J. Duque, pers. comm.). assumed in our simulation a continuous in- Also, the currently applied MFD of 50 cm put of seeds. This might have been too for medium-hardwood species, most of them optimistic in case of short cutting cycles. with a mid-successional status constituting In combination with a low minimum felling the bulk of commercial volume might be re- diameter at least some commercial species duced if sawable logs above this size should will be harvested before they have attained become scarce. Overall, harvest volume and their reproductive stage. Thus, stand com- MFD may vary within the range given in position will either shift to common species this study depending on the composition of capable of early reproduction or overall re- commercial volume in the individual logging generation will decline. unit. We simulated the spatial pattern of dis- Unlike in other neotropical regions, some turbance associated with logging. Differ- empirical data on long-term growth and ences between logging methods in terms yield are available to evaluate our results. of area damaged could be even more pro- An average annual bole volume increment nounced taking into account that cut trees 3 −1 −1 ≥ of 3.8 m ha y (SD = 4.2; trees 10 are expected to fall swiftly to the ground if cm d.b.h.) for the 30 years cutting cycle vines are cut well before logging operation with conventional logging is a conservative as assumed in our model. Conventional log- growth rate compared with Veillon’s (1985) ging methods do not consider vine cutting; 3 −1 −1 mean figure of 4.4 m ha y (SD = 0.5) this results in tree tangles which extend the measured in a stand 15-32 years after light canopy gap area (cf. Johns et al. 1996). logging. The basal area increment in the Vine cutting may become an even more im- first 10 years after logging over the simula- portant measure in logged-over stands be- tion period under conventional or reduced- cause these areas support a proliferation of impact logging with 30 years cutting cycles lianas (Kammesheidt 1999). With reduced- 3 −1 −1 andaremovalof30and60m ha cycle , impact and conventional logging, respec- 2 −1 −1 respectively was 0.3 and 1.1 m ha y . tively 23-35% and 50-73% of the simulated Lozada (1998) found similar basal area in- area was damaged (i.e. log-landings, felling crement rates in the first ten years in exper- and skidding areas), corresponding to a imentally cut stands with a basal area re- basal area removal of 2.5-5 m2 ha−1 cycle−1. Sustainable timber harvesting in Venezuela 83

By contrast, Hendrison (1990) found in Conclusions Suriname with a basal area removal of 4 − − m2 ha 1 cycle 1 that 22 and 36%, respec- Both the stability and sensitivity analy- tively of the forest area was damaged under sis showed that Formind2.0 simulates the controlled and uncontrolled logging. The stand dynamics of Caparo forest within re- much higher damage level with conventional alistic limits. The model’s capability to logging in Venezuela highlights the careless simulate the spatial heterogeneity of stands logging methods in the study area. with high resolution makes the model useful The data set for simulation combined for simulating growth and yield of logged- well-drained and poorly drained sites. The over forest. Whether cutting cycles identi- latter sites might show a slower rate of fied as sustainable in terms of timber yield, tree establishment and lower diameter in- are economically viable in the long run will crement rates than the well-drained site ow- strongly depend on species composition and ing to different soil water availability in the log quality of merchantable trees. Reliable rainy and dry season. This may result in a forecasts to this end will offer new chal- different speed of succession. To date, how- lenges to forest modelling. ever, no study has tested this hypothesis. Simulations were made under the as- sumption of no disturbance other than log- Acknowledgements ging and gap creation owing to tree fall. Fire, for example, is a real hazard because We thank F. Hapla, G¨ottingen University, of the considerable increase in fuel mass af- who determined the wood density of se- ter logging (cf. Nepstad et al. 1999) which lected tree species. H. Bossel, J. Chave, is easily inflammable during the pronounced A.R. Watkinson, and an anonymous re- dry season. Particularly short cutting cycles viewer provided helpful comments on the with conventional logging methods, leaving manuscript. P. K¨ohler was funded by the large tracts of open forest, increase the sus- Otto-Braun-Foundation of the University of ceptibility to fire. Kassel.

Chapter 7

The effects of logging, fragmentation and recruitment on growth of dipterocarp forest

Peter K¨ohler, Thomas Ditzer and Andreas Huth Center for Environmental Systems Research, University of Kassel Kurt-Wolters-Str. 3, D-34109 Kassel, Germany

Abstract

As deforestation through logging continues, danger to residual stands increases.Overuse of natural resources and a shortage in recruitment rates as indicated in recent research warrant having serious attention to those processes.We contibute to current discussion with a comprehensive simulation study for assessing various impacts on tropical forest dy- namics.The effects of different recruitment assumptions, size of observed area, boundary conditions, logging methods and harvesting intervals on growth and yield of two different tropical forest stands in Deramakot (Sabah), Malaysia were analysed.For this purpose, results of simulations of 70 different scenarios with the process-based forest growth model Formind2.0 were analysed. Formind2.0 is based on calculations of the carbon balance of individual trees belonging to 13 different plant functional types.Simulations suggest that natural recruitment acts as a strong buffer in response to disturbances.Thus, especially in fragmented forests, standing bole volume or number of saplings cannot indicate whether recruitment is sufficient to prevent forest degeneration and species loss.Recruitment rates at small diameters (1 cm) must become a focus of observations.From the detailed de- scription of different logging damages, simple regression functions emerge, which enable us to assess logging impacts.Disturbances in fragmented rain forests lead to shifts in abundances of species and to species loss.Reduced-impact logging methods produce up to three times higher yields than conventional methods.In the latter, yield increases with the length of the harvesting intervals, whereas reduced-impact methods produce maximum yield for a cycle period of 40 years.Results are based on optimistic assumptions, but they show that current logging practice in South-East Asia overuses forests to a significant degree. Keywords: forest growth model; Formind2.0; individual-oriented model; Malaysia; simulation; tropical rain forest; Journal of Ecology (2000), submitted. 86 Chapter 7

Introduction since many of their assumptions on forest recruitment are relevant in the context of forest management decisions as discussed Timber harvesting in tropical forests is a by Ditzer et al. (2000). Thus, we anal- widely discussed topic (Pinard & Putz 1996; yse in this work various logging scenarios Whitmore 1998). Damages to the resid- for a forest stand in Deramakot (Sabah), ual forest vary considerably for different Malaysia with a focus on the recruitment logging techniques and cycles. In addi- pattern. Moreover, as stated by Plumptre tion to its function as global carbon sink (1996), economy plays an important role in (Putz & Pinard 1993; Pinard & Putz 1997) making forest management decisions. Thus, a forest logged according to methods of timber yield will be analysed as well. reduced-impact logging is also expected to increase economic profit compared to a con- The simulation model used in this ventional logged forest (Barreto et al. 1998, study is the process-based forest growth cost/benefit analysis in Korpelainen et al. model Formind2.0. Formind (K¨ohler & 1995). Studies show convincingly that only Huth 1998a) was developed following an economically sound approaches will lead individual-oriented approach (Huston et al. to conservation and sustainable practices 1988; Judson 1994; Liu & Ashton 1995; (Plumptre 1996). In this context, criteria Uchma´nski & Grimm 1996). It was used for sustainability and ecological certification to evaluate the approach of the more aggre- of timber are subject of much current de- gated model Formix3 (Huth et al. 1998). bate (Johns 1985, 1997; F¨olster et al. 1986; One important feature of both models is Brown & Lugo 1990, 1994; Bruenig 1996; grouping of species into plant functional Ong et al. 1996; Putz & Viana 1996; Wei- types PFT (K¨ohler et al. 2000b). A care- delt 1996; Rice et al. 1997; Bowles et al. ful comparison of measured and modelled 1998). Computerised simulation models growth with Formind for different for- aiming to estimate growth and yield of trop- est stands in Sabah, Malaysia has been ical rain forest should become a useful tool used to validate the growth and compe- to broaden this discussion (Boot & Gullison tition processes as included in the model 1995; Clark & Clark 1999). (K¨ohler et al. 2001). A model applica- tion to simulate growth and yield in rain This discussion has motivated a lot of forests of Venezuela was performed recently recent work on the simulation of tropical (Kammesheidt et al. 2000). forest growth. Chave (1999) simulated the forest dynamics in French Guiana without In this study we want to answer the anthropogeneous influences. Liu & Ashton following questions for a stand of low- (1999) concentrated on the consequences land dipterocarp forest in Sabah, Malaysia. of timber harvest on tree species diversity (1) How does the size of the area, the under different seed dispersal assumptions boundary conditions and the recruitment in Malaysia. Ditzer et al. (2000), Huth pattern influence forest development? (2) Is & Ditzer (2000), and van Gardingen & there a lower threshold of forest fragmenta- Phillips (1999) focused on supporting for- tion, below which natural recruitment will est management decisions in Malaysian and fail? (3) How are forest structure and Indonesian Borneo, while Pinard & Crop- species composition modified by logging as per (2000) simulated the effects of logging function of the length of cutting cycle, the on carbon storage in dipterocarp forests. logging method and assumptions on recruit- ment? (4) Is there an optimal combina- Our aim is to bridge the gap between tion of the logging parameters (method, some of these studies. In our view, the re- length of cutting cycle, recruitment assump- sults of Liu & Ashton (1999) need to be tions) which maximises yields and min- generalised to forest management aspects, Growth & yield of rain forest in Sabah 87 imises changes in the forest structure? with few pioneers, two dominated by pio- neers) and the last one was a recently (one year prior) logged stand using methods of Methods reduced-impact logging. For the purpose of our simulations, we choose the recently logged stand (plot 4 in Area Description the work of Schlensog, labelled L1 in Huth et al. 1998 and here) as an example of for- The study area is Deramakot Forest Re- est structure after logging, which also cor- serve (DFR) situated in Sabah (North Bor- responds to most of the forests in DFR, and neo, Malaysia, 117◦30’ E, 5◦25’ N, 130- one of the primary forest stands (plot 1, re- 300 m asl.). Deramakot has a perhumid ferred to as P1 here) for reference. climate typical for the inner tropics. Mean annual temperature is 27◦ with little sea- sonal variations. Average annual precip- Species grouping itation is about 3500 mm, with no pro- nounced dry season. The geology of De- Shrub and tree species (total number: ramakot is characterised by tertiary sedi- 468 species) were assigned to 13 different ments, mostly sandstones. The soils are plant functional types (PFT) based on nutrient-poor and prone to erosion once left their successional status and maximum devoid of tree cover. The prevailing forest height at maturity (Table 7.1). The suc- type is lowland dipterocarp forest (Schlen- cessional status (early, mid, or late) was sog 1997). The forest remained essentially determined by their growth rates under undisturbed until this century. Commer- various light regimes, as well as through cial logging started in 1956. The intensity asurveyofwooddensities,whichare of logging and of logging damages varies good indicators of growth rates for most widely (Kilou et al. 1993). In 1991 the species. Species list including grouping Sabah Forestry Department carried out a is available online (http://www.usf.uni- terrestrial inventory. All trees with a diame- kassel.de/usf/archiv/dokumente.en.htm, ≥ ter 10 cm in 0.25 ha sample plots regularly Appendix of thesis, Table B.2). A detailed × distributed in a 1 1 km grid over the whole description and validation of the grouping reserve of 55,000 ha were recorded. Average concept and its application to Sabah was 2 −1 2 basalareawas20.9m ha (SD=9.2 m published elsewhere (K¨ohler et al. 2000b). −1 2 −1 ha ; range: 1.3-57.8 m ha ), indicating Similar grouping concepts are found in logged-over forest compositions (Kilou et al. Swaine and Whitmore (1988), Manokaran 1993; K¨ohler 1998). & Swaine (1994), Thomas & Bazzaz (1999), Within the Deramakot Forest Reserve, and Kammesheidt (2000). In addition, eight research plots with different degrees of a subgrouping into commercial and non- disturbance were analysed for tree species commercial species is performed for all composition and forest structure (Schlen- mid and late successional species. Since sog 1997). All trees with a diameter at detailed information of the commercial breast height d≥30 cm were measured in status was not available at the individual plots of 90×90 m, small trees (d≥10 cm), tree level, 80 % of mid and late successional saplings (height≥1.5 m and d<10 cm) and species are considered as commercial timber seedlings (height<1.5m)innestedplotsof (Sabah-Forestry-Department 1994). A suf- 30×30 m, 30×5 m, and 59 plots of 1×1m, ficient number of PFTs is essential for the respectively. Out of the eight plots, three accuracy of the output in the simulation of were primary forest stands possibly never highly diverse rain forests. However, in the logged, four were logged-over stands (two analysis and for the sake of simplicity, we 88 Chapter 7

Table 7.1: Autecological characteristics of 13 plant functional types (PFTs) of Sabah’s lowland tree species.H: Height at maturity.SS: successional status.HG: height group.COM: fraction of commercial species in PFT.No: Number of species per PFT (total 468 spp.).P1, L1: Abundance of trees with diameter >10 cm in plot P1 and L1, respectively.Sum of abundances might not match 100 % due to rounding errors.(Modified from K¨ohler et al.2000b.)

Plant functional type H [m] PFT SS HG COM [%] No P1 [%] L1 [%]

Shrub mid succ. spp. 0-5 1 2 1 0 15 0.0 0.0

Understorey early succ. spp. 5-15 2 1 2 0 5 0.0 4.6 Understorey mid succ. spp. 5-15 3 2 2 80 28 6.9 1.3 Understorey late succ. spp. 5-15 4 3 2 80 65 6.4 0.8

Lower canopy early succ. spp. 15-25 5 1 3 0 14 0.7 65.8 Lower canopy mid succ. spp. 15-25 6 2 3 80 92 18.8 2.1 Lower canopy late succ. spp. 15-25 7 3 3 80 13 0.2 0.8

Upper canopy early succ. spp. 25-36 8 1 4 0 10 0.0 4.6 Upper canopy mid succ. spp. 25-36 9 2 4 80 89 6.6 4.6 Upper canopy late succ. spp. 25-36 10 3 4 80 18 3.6 2.5

Emergent early succ.spp. >36 11 1 5 0 3 0.0 0.4 Emergent mid succ.spp. >36 12 2 5 80 93 37.0 11.7 Emergent late succ.spp. >36 13 3 5 80 24 19.5 0.0

distinguish results only between the three a mosaic of interacting forest patches of different successional status (early, mid, 20 m2×20 m2 in size. Within these patches and late). trees are not spatially explicit distributed, and thus all compete for light and space following the gap model approach (Botkin Description of the model 1993; Shugart 1998). Allometric relation- ships connect above-ground biomass, stem Formind2.0 is an individual-oriented diameter, tree height, stem volume and process-based forest growth model (K¨ohler crown dimensions. Using these allometric & Huth 1998a; K¨ohler et al. 2001) to relationships, the distribution of individ- simulate spatial and temporal development ual tree crowns and their leaf area in the of uneven-aged mixed forest stands. A canopy is calculated in horizontal canopy complete description including all the layers with a depth of 0.5 m. relevant functional relationships of the The growth of an individual tree is based model version Formind2.0 was published on a carbon balance. Calculations include elsewhere (Kammesheidt et al. 2000). We photoproduction of the trees and assimilate concentrate in the following after a short losses due to respiration, litter-fall and fine general description on the recruitment root decay. Within a patch, vertical light submodel. Basic functions are shown in the attenuation in the canopy is calculated from Appendix (of article). light intensity above the canopy with re- The model describes forest dynamics as Growth & yield of rain forest in Sabah 89 spect to the absorption of tree crowns. The dispersal strategies, agents, and distances, dependence of specific photosynthetic pro- seed survival, germination probabilities and ductivity on irradiance is modelled using a maturing size of seed disperser (Garwood Michaelis-Menten-type light response curve. 1983; Whitmore 1983; Denslow 1987) some Photoproduction P˜ is calculated from the fundamental assumptions on the most im- tree’s leaf area (Monsi & Saeki 1953). We portant trends have to be made. assume an increasing limitation effect of wa- Flowering, fruiting and seed produc- ter transport deficiencies with growing tree tion vary in duration and frequency across height (Ryan et al. 1997). Thus, actual pro- species, some species fruiting after several ductivity is calculated by applying a size- years of unfecundity (Garwood 1983; Cur- dependent limitation factor q(d) (according ran & Leighton 2000). Other species flower to the aging factor of Landsberg & Waring and fruit continually throughout the year 1997). Assimilate losses are estimated in re- in Malaysian rain forests (Putz 1979). Sea- lation to tree biomass B (Kira 1978; Yoda sonal differences in seed production are 1983). We distinguish between a biomass- not taken into consideration. The rate of dependent maintenance respiration Rm and seed production varies widely among species growth respiration RG (Ditzer et al. 2000). (Whitmore 1998). Various studies have Our main time-dependent growth equation analysed different dispersal strategies and for one tree i is lengths (review in Clark et al. 1999b). Dif- dBi ferent dispersal agents (e.g. wind, birds, = P˜i · q(di)(1 − R ) − R (Bi) (7.1) dt G m mammals) are not directly distinguished in our model, but the resulting average disper- Tree growth is calculated in annual time sal distance XR depends upon the species steps. and should match with the parameter set. Competition is modelled in terms of com- From the dispersal kernels discussed by petition for light and space, the latter re- Clark et al. (1999) we use the Gaussian dis- sulting in self-thinning. tribution (as used by Chave 1999b). As- suming rotation symmetry, the probability For small trees (diameter < 10 cm a density f of seeds to be dispersed at the diameter-dependent mortality is added to distance r from the mother tree is a basic mortality rate. Trees resulting in crown closure are eliminated to avoid 2r r2 crowding (self-thinning). Mortality is mod- f(r)= exp − , (X + cd )2 (X + cd )2 elled as stochastic event. Senescent trees (≥ R 2 R 2 10 cm d.b.h.) die and collapse with a certain (7.2) probability, knocking down smaller trees in with cd, the crown diameter (see neighbouring patches thereby creating gaps Fig. 7.1). Thus, 99% of the seeds of a size that depends upon their crown size. are dispersed in a distance less than × Two different recruitment mechanisms 2.14 (XR+cd/2). The actual dispersal were incorporated in the model. The sim- distance r is randomly drawn from this plest approach consists in assuming that an probability distribution, and the direction intact forest is supporting a constant seed is drawn uniformly. The resulting seed input rate. The second takes into account shadow is the product of the rate of seed the dispersal of seeds produced from local production and the dispersal kernel (Clark mother trees, i.e. trees exceeding a certain et al. 1999b). diameter DR. As recruitment strategies are For both recruitment mechanisms, in- highly variable in rain forests, with interspe- coming seeds update a seed pool, tak- cific differences in fruiting period, number ing into account the dormancy variability of seeds, seed sizes (Leishman et al. 1995), across functional groups (cf. Garwood 1983, 90 Chapter 7 0.016 piration processes were estimated by sensi- 0.014 XR = 100m

[-] XR = 75m tivity analysis to obtain realistic diameter

f 0.012 X = 50m 0.01 R increment values for different size classes 0.008 and different light conditions. Mortality 0.006 rates correspond to typical values found 0.004 in the literature (Manokaran & Kochum- Probability 0.002 men 1987; Swaine 1989; Condit et al. 1992, 0.0 1995a, 1995b; Carey et al. 1994; Manokaran 0 100 200 300 & Swaine 1994; Phillips & Gentry 1994; Distance r [m] Silva et al. 1995; van der Meer & Bongers 1996). Seed production rate NR per mother tree and constant seed input rate NS were Figure 7.1: Seeds dispersal kernels for a estimated from a sensitivity analysis of Gaussian distribution with different average long-term runs of the mature stand P1, as- dispersal distances X .Crown diameter was R suming a dynamic equilibrium of the forest fixed at cd=20 m. structure and a given species composition. In both stands (P1, L1), trees were dis- 1989). These seeds correspond to the repro- tributed randomly to the 5×5=25 patches ductive success and are those which can po- of one hectare. P1 contained nearly no early tentially be established at the minimum di- successional species, abundance was high- ameter of 1 cm (Ribbens et al. 1994; Chave est in PFT 12, 13 and 6with 37, 19.5, and 1999b). Seed loss due to predators is implic- 18.8 % respectively. Site L1 was dominated itly included in relative low seed production by medium-size pioneers (PFT 5: 65.8 %), rates. The actual seed germination depends and contained nearly no late successional upon understorey light intensities. species (Table 7.1). The seed pool was filled with average seed numbers from long term simulations. Parameters and initialisation

Parametrisation used in our simulations is found in the Appendix (of article). Data Model evaluation on light environment are taken from Schlen- sog (1997). Allometric relations of tree One of the main results of a previous Formind2.0 crowns were found in Rollet (1978), Poker analysis with applied to a (1993), and Schlensog (1997). The form Venezuelan rain forest was that the mor- factor, leaf and crown area to diameter tality parameters contributed to most of relations are taken from measurements of the variance in the simulation outputs Kato et al. (1978) and Kira (1978) in (Kammesheidt et al. 2000). While the bole Pasoh, Malaysia. Height-to-diameter re- volume, basal area and total leaf area index lations are obtained from an inventory were quite insensitive, the density of early of Sabah’s forests (Forestal-International- successional species was highly sensitive to Limited 1973). Data on photosynthetic most parameter changes. The present appli- response of functional groups to different cation for rain forests in Sabah differed only light intensities are given in Eschenbach in the value of the parameters, the sensitiv- et al. (1998). The wood density of species ity analysis was expected to yield the same was determined from various sources (Mei- results as the previous study. jer & Wood 1964; Burgess 1966; Fox 1970; In the present work, we mainly focus Cockburn 1980; Keating & Bolza 1982; on the influence of the recruitment pa- PROSEA 1994). Parameter values for res- rameters on the results. Sensitivity of Growth & yield of rain forest in Sabah 91

28 output quantities to variations of para- Table 7.2: Contribution of constant seeds in- meter values was investigated in 231 sim- put and local recruitment to different recruit- ulations (cf. Vanclay & Skovsgaard 1997). ment scenarios [%].ST: Seed trees only, MS: mixed seeds, and SP: seed pool. Each of the twelve parameters for the seed pool submodel and each of the 21 param- eters for the within-patch fecundity mod- Recruitment scenariosa ule (see parameters in Appendix) was varied Input process ST MS1 MS2 MS3 SP seven times, from the standard value (StV), using an identical scheme spanning two orders of magnitude (0.1×StV; 0.5×StV; Constant input 0 2 10 50 100 0.9×StV; StV; 1.1×StV; 2×StV; 10×StV). Local recruitment 100 98 90 50 0 In a few cases physical boundaries restricted the range of the parameter values, thus a slightly reducing the total number of simu- As number of seeds dispersed from each mother tree must not be a fraction, seed numbers are lations performed (e.g. mortality rates can- rounded up to natural numbers. not exceed values of 1.0). Simulations over 4 hectares and 1000 years in the mature stand (P1) were performed for each varied dition to a constant seed shadow seems to parameter. Mean values of the chosen result be the most realistic situation. Five scenar- variables were averaged over the seven sim- ios with different recruitment pattern were ulations. The coefficient of variation (CV) defined from strict local seed dispersal to a from this average was chosen as an indicator constant seed input without local influences to find whether a result variable was sensi- (seed tree, mixed seeds 1-3, seed pool). The tive to parameter changes or not. More de- number of dispersed seeds per mother tree tails on variations are found in caption of NRs and constant seedling input NSs are Fig. 7.2. modified from standard values using the fac- tors given in Table 7.2. An extreme sce- nario was the forest development without Simulation scenarios any recruitment, which was simulated for both stands P1 and L1. Once our model is validated, we designed 70 different simulation scenarios for this study. Size of the simulated area. - To anal- The scenarios differ in at least one factor: yse the effects of size we ran simulations on recruitment pattern, size, boundary condi- three different scales: 1 ha, 9 ha, 25 ha, tions, logging cycle, and logging method. all square in shape. After an analysis of size effects on forest development without Recruitment pattern. - In former stud- timber harvest, logging was performed on a Formind ies using , recruitment was mod- medium size of 9 ha. elled on the basis of constant input of seeds over the whole simulation area. This was Boundary conditions. - By simulat- justified assuming an intact forest structure ing relative small areas of a few hectares, surrounding the simulated area. As forest boundary effects are worth considering. In fragmentation has increased in the tropics an 1 ha area with 25 patches 64% of them (Laurance et al. 1997; Gascon et al. 2000) are at an outer border. This is reduced to these assumption might be too simplistic for 15% in an area of 25 ha. Especially when ex- the future. Explicit seed dispersal was mod- plicit seed dispersal is modelled, boundary elled in an alternative recruitment model. conditions are important for forest develop- Our simulation areas were still small (up to ment. Two different conditions were exam- 25 ha). Thus, a mixture of explicit seed ined: toroidal (or periodic) boundaries and dispersal from the simulated stand in ad- open boundaries. With the toroidal bound- 92 Chapter 7 ary conditions it was assumed that the right Table 7.3: Different logging impacts to resid- boundary is connected to the left boundary, ual stand if reduced-impact logging (RIL) or and the top to the bottom. Thus, falling conventional logging (CON) is applied.Skid- ding damages were percentages of stem num- trees or seeds leaving the simulation area bers, area losses were percentages of simulation reenter at the opposite site. With open area. boundaries, everything leaving the simu- lated area is lost and nothing enters the system. Open boundary scenarios can be Impact RIL CON understood as a strongly fragmented land- scape without any interactions as immigra- tion between different forest islands. In con- Felling damage ∼ crown size trast, with toroidal boundaries the simu- Felling direction to gaps random lated area can be understood as nested in Skidding damage 25% 55% a larger forest of a similar structure. Log- Area loss 12% 33% × × ging scenarios are performed with toroidal Mortality 10 y after 2 3 logging boundary conditions. Logging method and cycle length. - Our modelling of logging practices was moti- vated by several studies (Hendrison 1990; complete removal of the residual stand in Crome et al. 1992;Cannon et al. 1994; Johns 12% (RIL) and 33% (CON) of randomly et al. 1996; Pinard & Putz 1996; Johns chosen patches. Mortality in the years af- 1997; Bertault & Sist 1997; Sist et al. ter logging was twice (RIL) and three times 1998; Ditzer et al. 2000; Kammesheidt (CON) higher than normally, counting for et al. 2000). Two methods were distin- damaged but not instantly destroyed trees. guished: highly damaging conventional log- The time between two harvesting opera- ging (CON) which generally makes use of tions was varied in steps of 20 y between heavy machinery, unskilled workers and lit- 20 y and 80 y. Stand L1 was recorded di- tle planning effort, and reduced-impact log- rectly after a logging operation, therefore ging (RIL), where substantial planning for the first logging starts right after one cy- road construction, felling directions etc. was cle length. All commercial trees of the mid performed. In RIL, tree removal usually and late successional species with a mini- implies the use of winching cables or air- mumdiameterof60cmwereremovedina borne cable systems. Modelled differences logging operation. of the two methods were damages to the residual stand. We distinguished (1) dam- ages through tree felling, (2) skidding dam- Results ages in the patch of a felled tree, (3) area loss due to road construction and log land- ings, and (4) increased mortality rates for Model evaluation ten years after logging (Table 7.3). Tree-fell damage was proportional to crown size and We used the analysis of variance (ANOVA) method-independent. The felling direction to detect significant differences in our simu- was random in CON, but directed towards lation results (Graf et al. 1987; Sachs 1997; neighbouring gaps in RIL, whenever possi- Fowler et al. 1998). Across scenarios the ble. In the patches where felled trees were standard error is generally small, it is nor- situated, 25% and 55% of the stems were mally not shown in the figures. Further- killed through skidding in RIL and CON, more, tests of statistical differences at a 1% respectively. Area loss was simulated by level were performed as t-test for matched pairs (Fowler et al. 1998), but cannot be Growth & yield of rain forest in Sabah 93

GEN PFT SS HG ND range

IS-all 0.1-10.0 MS-all 0.1-10.0 NS-all 0.1-10.0 IS1 0.02-1.00 IS2 0.0-0.4 IS3 0.00-0.1 MS1 0.01-1.00 MS2 0.05-1.00 Seed pool MS3 0.01-1.00 NS1 15-1500 NS2 63-6250 NS3 5-500

IS-all 0.1-10 MS-all 0.1-10.0 NR-all 0.1-10.0 XR-all 0.1-10.0 IS1 0.02-1.00 IS2 0.0-0.4

Parameter IS3 0.0-0.1 MS1 0.01-1.00 MS2 0.05-1.00 MS3 0.01-1.00 DR1 0.0-0.4 D 0.01-1.00 Seed tree R2 DR3 0.02-1.50 DR4 0.04-1.50 DR5 0.05-1.50 NR1 10-1000 NR2 2-200 NR3 1-40 XR1 10-1000 XR2 7.5-750.0 XR3 5-500 S1 SS 2 SS 3 SS GAP BUILDING BA V G1 HG 2 HG 3 HG 4 HG 5 HG MATURE LAI N N N N N F 1 PFT 2 PFT 3 PFT 4 PFT 5 PFT 6 PFT 7 PFT 8 PFT 9 PFT 10 PFT 11 PFT 12 PFT 13 PFT 0-30 60-90 30-60 90

Result variable

Figure 7.2: Sensitivity analysis.Each parameter of the recruitment module (different ap- proaches Seed pool and Seed tree are distinguished) was varied seven times in the given range (one simulation with standard parameter value).Range stretched from 1/10 to 10 × standard parameter value.Simulations were made for 4 ha over 1000 years of mature forest (P1).De- scription of labels and units is found in Appendix (of article).Time averaged values ( v) vary from average (a) with coefficient of variation (CV =| v − a | /a × 100).White: low sensi- tivity (CV ≤ 10 %); grey: medium sensitivity (10 % 50 %).Parameters varied for all groups, where indicated (e.g. IS-all).Result variables are: GEN: General information refers to (LAI) total leaf area index, succession stages ((GAP ) gap phase: no trees with h≥15 m in patch, (BUIDLING) building phase: trees with h≥15 m in patch, and (MATURE) mature phase: trees with h≥30 m in patch), (BA) basal area, and (V ) bole volume of the whole stand. PFT, SS, HG: relative bole volume of the 13 PFTs, 3 successional status groups, or 5 height groups, respectivly (cf.Tab.7.1). ND: Stem number as function of diameter (N: all trees) for trees between 0 and 30 cm dbh (N0−30), 30 cm and 60 cm dbh (N30−60), 60 cm and 90 cm dbh (N60−90) and above 90 cm dbh (N90). 94 Chapter 7

2.0 1.5 1.0

SS1 [%] 0.5 0 80 60 40 SS3 [%] 20 0 IN 0.1 1 10 15 150 1500 63 625 6250 5 50 500 -1 -1 -1 -1 -1 -1 NS-all [-] NS1 [ha y ] NS2 [ha y ] NS3 [ha y ]

Figure 7.3: Sensitivity analysis of constant seeds input rate NSx (x=all, or s=1,2,3: succes- sional status, Table 7.1) on abundance of early (SS1) and late (SS3) successional species. Further description see Fig.7.4.

2.0 1.5 1.0

SS1 [%] 0.5 0.0 80 60 40 SS3 [%] 20 0 IN 0.1 1 10 10 100 1000 2 20 200 14 40 -1 -1 -1 -1 -1 -1 NR-all [-] NR1 [tree y ] NR2 [tree y ] NR3 [tree y ]

Figure 7.4: Sensitivity analysis of seed production rates NRx (x=all, or s=1,2,3: successional status, Table 7.1) on abundance of early (SS1) and late (SS3) successional species. Circles mark time averaged values ± SD.Parameters were varied over two orders of magnitude with standard value inbetween.Simulations were made for 4 ha over 1000 years of mature forest (P1) with toroidal boundaries.Solid line: average of all variations.Doted line: coefficient of variation (CV) of single values to average.Values on the y-axes (labeled IN in lower left subfigure) correspond to inventory of P1.Note logarithmic scale of x-axis.Further information in Fig.7.2and text. Growth & yield of rain forest in Sabah 95 ]

-1 all PFTs PFT 8 PFT 13 30

20 mean 1 mean 2 10 OTLAI = 0 OTLAI <= 2 OTLAI > 2 0

Diameter incr. [mm y 0 50 100 0 50 100 0 50 100 Diameter [cm]

Figure 7.5: Diameter increment as function of diameter, light competition and plant functional type (PFT).Analysis of annual increments of all trees in simulations of 4 ha and 1000 y of mature forest (P1).Left: Average over different plant functional types; middle: PFT 8 as an example for early successional species; right: PFT 13 as an example of late successional species. Light competition in overtopping leaf area index (OTLAI) from OTLAI=0: no competition to OTLAI≥2: strong competition.Mean 1: average over both diameter and OTLAI; mean 2: average over different OTLAI. shown in the figures, where up to 40 dif- index were insensitive to most parameter ferent scenarios were compared. They were changes, while species composition varied reported in the Figure captions. widely. Size structure of the forest remained Diameter increment rates as function of insensitive as well. For the most uncertain tree diameter, PFT, and light intensity were parameters seed input rate NS in SP and analysed in average and for PFTs (Fig. 7.5). seed production rate NR in ST more de- The overall increment is about 5 mm y−1, tailed analysis of species composition were mainly caused by late successional species, undertaken (Fig. 7.3, 7.4). as abundance of early succ. spp. is low in mature stands. There are obvious differ- ences between early (e.g. PFT 8: mean Influence of area size, boundary 18 mm y−1) and late (e.g. PFT 13: 4 mm conditions and recruitment pat- y−1) successional species. Light competi- tern on forest structure tion reduced both the maximum reachable diameter, and diameter increment rates. There were no significant effects of the area Small (d<20 cm) trees of PFT 13 were never size on the total bole volume. Though found in unshaded position, whereas most ANOVA indicates significant differences of of PFT 8 are exposed to full or near to full boundary conditions or recruitment as- light. sumptions (Table 7.4A), those were hardly An overview of the sensitivity analysis seen when the total bole volume (469- 3 −1 concerning the two recruitment scenarios 499 m ha ) was plotted for all the scenar- seed pool (SP) and seed tree (ST) is shown ios (Fig. 7.6). However, species composi- in Fig. 7.2. Results in ST were more sensi- tion varied significantly when the boundary tive than in SP. While minimum light inten- conditions, recruitment assumptions (early and late succ. spp.), or area size (only early sity for germination IS produces sensitive behaviour in SP and ST, similar strong re- succ. spp.) (Table 7.4B,C) were changed. sponse were achieved by varying minimum Abundance of early successional species were below 2 % (0.5-1.3 %), while fraction diameter of seed producing trees DR in ST. Total bole volume, basal area and leaf area of late successional species increased from about 25 % in the ST scenario to 37 % 96 Chapter 7

Table 7.4: ANOVA for the main and interactive effects of area size (1 ha, 9 ha, 25 ha), boundary conditions (open, closed), and recruitment assumptions (scenarios SP, MS3, MS2, MS1, ST) on (A) total bole volume, (B) relative share of early successional species, (C) relative share of late successional species.No logging was performed.Simulation of primary forest at site P1, simulation time was 1000 y (n = 5).SS: sum of squares; df: degree of freedom; MS: mean of squares; P: probability (details in Fowler et al.1998).

Effects SS df MS F ratio P

A) Total bole volume

Size 357.68 2 178.84 4.66 >0.01 Boundary 810.02 1 810.02 21.10 <0.001 Recruitment 1010.48 4 252.62 6.58 <0.001 Size × boundary 60.52 2 30.26 0.79 >0.1 Size × recruitment 834.65 8 104.33 2.72 <0.01 Boundary × recruitment 1143.01 4 285.75 7.44 <0.001 Size × boundary × recruitment 1013.93 8 126.74 3.30 <0.005 Error 4606.69 120 38.39

B) Share of early successional species

Size 2.063 2 1.032 41.11 <0.001 Boundary 2.137 1 2.137 85.18 <0.001 Recruitment 2.602 4 0.651 25.93 <0.001 Size × boundary 0.690 2 0.345 13.75 <0.001 Size × recruitment 0.087 8 0.011 0.43 >0.5 Boundary × recruitment 0.113 4 0.028 1.13 >0.1 Size × boundary × recruitment 0.399 8 0.050 1.99 >0.05 Error 3.011 120 0.025

C) Share of late successional species

Size 67.64 2 33.82 2.51 >0.05 Boundary 190.83 1 190.83 14.19 <0.001 Recruitment 3704.63 4 926.16 68.86 <0.001 Size × boundary 41.63 2 20.81 1.55 >0.1 Size × recruitment 213.16 8 26.64 1.98 >0.05 Boundary × recruitment 36.66 4 9.16 0.68 >0.5 Size × boundary × recruitment 40.89 8 5.11 0.38 >0.5 Error 1614.03 120 13.45 Growth & yield of rain forest in Sabah 97

] Seed tree Seed pool -1 Mixed seeds 1 Mixed seeds 2 Mixed seeds 3 600 ha

3 500 400 300 200 100 0

Bole volume [m 1925 1925 1925 1925 1925 Area size [ha]

Figure 7.6: Effects of area size (1 ha, 9 ha, 25 ha), boundary conditions (toroidal (left, grey), open (right, black)) and recruitment scenarios (seed tree; mixed seeds 1-3; seed pool) on standing bole volume.Simulations of primary forest at site P1 without logging.Results are means (n=5) of 1000 simulated years.SE was always <6m3 ha−1 and thus not shown.T-test for matched pairs found significant differences at 1 % level for 20% of tested pairs.

Seed tree Mixed seeds 1 Mixed seeds 2 Mixed seeds 3 Seed pool 100 80 60 40 20

Bole volume [%] 0 1925 1925 1925 1925 1925 Area size [ha]

Figure 7.7: Effects of area size (1 ha, 9 ha, 25 ha), boundary conditions (toroidal (left), open (right)) and recruitment scenarios (seed tree; mixed seeds 1-3; seed pool) on species composition (early (white), mid (grey) and late (black) successional species).Simulations of primary forest at site P1 without logging.Results are means (n=5) of 1000 simulated years.SE not shown for technical reasons (SE: early succ spp.<0.13 m3 ha−1; late succ spp.<3.38 m3 ha−1).T-test for matched pairs found significant differences at 1 % level for 34% and 35 % of tested pairs of abundance early and late successional species, respectively. 98 Chapter 7

Seed tree Mixed seeds 2 Seed pool

500 400 1ha 300 toroidal 200 100 0 500

] 400 25ha -1 300 toroidal ha

3 200 100 0 500 400 1ha 300 open Bole volume200 [m 100 0 500 400 25ha 300 open 200 100 0 0 500 0 500 0 500 1000 Time [y]

Figure 7.8: Development of bole volume (m3 ha−1) for different scenarios.Simulation of 1000 y of mature forest at site P1, not averaged over n=5 runs to show dependency on stochastics. Variation of area size (1 ha, 25 ha), boundary conditions (toroidal, open) and recruitment scenarios (seed tree; mixed seeds 2; seed pool).Total (solid bold line), early successional spp. (solid line), mid successional spp.(broken dotted line) , late successional spp.(broken line) . Growth & yield of rain forest in Sabah 99

Seed tree Mixed seeds 2 Seed pool

500 400 1ha 300 toroidal 200 100 0 500

] 400 25ha -1 300 toroidal ha

3 200 100 0 500 400 1ha 300 open Bole volume200 [m 100 0 500 400 25ha 300 open 200 100 0 0 500 0 500 0 500 1000 Time [y]

Figure 7.9: Development of bole volume (m3 ha−1) for different scenarios.Simulation of 1000 y of logged forest at site L1, not averaged over n=5 runs to show dependency on stochastics. Variation of area size (1 ha, 25 ha), boundary conditions (toroidal, open) and recruitment scenarios (seed tree; mixed seeds 2; seed pool).Total (solid bold line), early successional spp. (solid line), mid successional spp.(broken dotted line) , late successional spp.(broken line) . 100 Chapter 7

] P1 L1 species, was reduced in open boundary con- -1 ditions. However, as a most realistic recruit- ha

3 500 ment scenario, local supply with seeds (SP) 400 will be insufficient in logged small forests 300 and will lead to a loss of tree species, even if 200 the standing bole volume remains at a com- 100 parable high level of about 500 m3 ha−1.In 0 dramatic events, e.g. a total loss of seed sup-

Bole volume [m 0 100 200 300 0 100 200 300 Time [y] ply, possible through enhanced seed preda- tion coupled with low seed production (Cur- ran & Leighton 2000), forest fragmentation Figure 7.10: Development of bole volume (m3 will lead to a collapse of the forest struc- ha−1) without seed supply.Simulation of 1 ha ture, but only after about 300 and 200 y for and 350 y of mature (P1) and logged forest (L1) stand P1 and L1, respectively (Fig. 7.10). under open boundary conditions.Total (solid In the latter case the bole volume increased bold line), early successional spp.(solid line), first from 280 to 400 m3 ha−1 before volume mid successional spp.(broken dotted line) , late eventually decreased. successional spp.(broken line) . in SP (Fig. 7.7). This fraction was always Modification of forest struc- higher in scenarios with open than with toroidal boundary conditions, which is a re- ture and species composition sult of lower average seed dispersal distance through logging in late compared to early and mid succes- sional species. The temporal development The total bole volume was significantly of the bole volume (total and for differ- affected (ANOVA) by all logging cycles, ent successional types) showed high fluctua- methods, and recruitment assumptions (Ta- tions in simulations of small areas (Fig. 7.8). ble 7.5A). Average total bole volume var- To visualise the temporal dependence of for- ied from 226to 395 m 3 ha−1. Short log- est structure on initial stand composition, ging cycles always resulted in lower bole vol- the same dynamics are shown for simula- ume than longer cycles, and likewise con- tions with logged stand L1. (Fig. 7.9). ventional logging had greater impacts than Here, few and only small trees of late succes- reduced-impact logging, if all other param- sional species persist in the inventory, which eters were held constant (Fig. 7.11). Dif- led to an extinction of those species in sce- ferences between recruitment scenarios were nario ST, but substantial regeneration in small. Logging damage was correlated to scenario SP. harvesting intensities. Thus, basal area of residual, logged, and damaged trees were plotted as function of removed basal area Forest fragmentation (Fig. 7.12). The percentage of residual standing basal area (denoted yrs)andof Following the above analysis, open bound- damaged trees (ydt) were linear functions ary conditions (forest islands with no im- the basal area x. In the RIL scenario − migration) favoured species with lower seed we found: yrs =83.3 3.2x and ydt = dispersal distance (e.g. late vs. mid suc- 15.9+0.7x, while in the CON scenario: − cessional species), as less seeds are lost in yrs =70.7 5.1x and ydt =28.0+2.6x. the surrounding. Furthermore, in border Higher damages in conventional method are patches, the number of gap creating tree clearly seen. fall events, which favour early successional Changes of abundance of early and late Growth & yield of rain forest in Sabah 101

Table 7.5: ANOVA for the main and interactive effects of logging cycle (20 y, 40 y, 60 y, 80 y), logging method (conventional (CON), reduced-impact (RIL)) and recruitment assump- tions (scenarios ST, MS1, MS2, MS3, SP) on (A) Total bole volume, (B) relative share of early successional species, (C) relative share of late successional species, and (D) harvest yield.Sim- ulation of logged forest at site L1 covering 9 ha for 240 y (n = 5) with toroidal boundaries. Columns are explained further in Table 7.4.

Effects SS df MS F ratio P

A) Total bole volume

Cycle 297993.34 3 99331.11 9485.88 <0.001 Method 40431.62 1 40431.62 3861.12 <0.001 Recruitment 19639.96 4 4909.99 468.89 <0.001 Cycle × method 3317.78 3 1105.93 105.61 <0.001 Cycle × recruitment 11383.34 12 948.61 90.59 <0.001 Method × recruitment 1615.02 4 403.75 38.56 <0.001 Cycle × method × recruitment 1166.07 12 97.17 9.28 <0.001 Error 1675.44 160 10.47

B) Share of early successional species

Cycle 2628.66 3 876.22 2118.25 <0.001 Method 7740.53 1 7740.53 18712.66 <0.001 Recruitment 210.22 4 52.56 127.05 <0.001 Cycle × method 272.94 3 90.98 219.94 <0.001 Cycle × recruitment 147.61 12 12.30 29.74 <0.001 Method × recruitment 73.93 4 18.48 44.68 <0.001 Cycle × method × recruitment 46.46 12 3.87 9.36 <0.001 Error 66.18 160 0.41

C) Share of late successional species

Cycle 60.39 3 20.13 233.69 <0.001 Method 86.87 1 86.87 1008.44 <0.001 Recruitment 2427.15 4 606.79 7044.16 <0.001 Cycle × method 1.46 3 0.49 5.63 <0.005 Cycle × recruitment 88.57 12 7.38 85.68 <0.001 Method × recruitment 82.52 4 20.63 239.51 <0.001 Cycle × method × recruitment 1.40 12 0.12 1.36 >0.1 Error 13.78 160 0.09

D) Harvest yield

Cycle 189696.41 3 63232.14 233.67 <0.001 Method 475111.67 1 475111.67 1755.73 <0.001 Recruitment 1047346.36 4 261836.59 967.59 <0.001 Cycle × method 85767.21 3 28589.07 105.65 <0.001 Cycle × recruitment 20074.29 12 1672.86 6.18 <0.001 Method × recruitment 67639.75 4 16909.94 62.49 <0.001 Cycle × method × recruitment 19516.92 12 1626.41 6.01 <0.001 Error 43297.14 160 270.61 102 Chapter 7

] Seed tree Seed pool -1 Mixed seeds 1 Mixed seeds 2 Mixed seeds 3 500 ha 3 400 300 200 100 0

Bole volume [m 20 40 60 80 20 40 60 80 20 40 60 80 20 40 60 80 20 40 60 80 Cycle length [y]

Figure 7.11: Effects of logging cycle (20 y, 40 y, 60 y, 80 y), method (reduced-impact (grey), conventional (black)) and recruitment scenarios (seed tree; mixed seeds 1-3; seed pool) on stand- ing bole volume.Simulations of 9 ha with toroidal boundaries.Results are means (n=5) of 240 simulated years.SE was always <2m3 ha−1 and thus not shown.T-test for matched pairs found significant differences at 1 % level for 89% of tested pairs. successional species in relation to logging Optimising harvest results cycles, methods, and recruitment assump- tions were significant (Table 7.5B,C). Abun- Timber harvest varied significantly across dance of early successional species varied the three scenario parameters (Table 7.5D). − between 4.9 % and 29.2 %, late succes- Gross yield ranged from 0.8 to 2.2 m3 ha 1 − sional species between 1.3 % and 13.9 % y 1. Net yield, covering an additional 30 % (Fig. 7.13). The more the recruitment loss due to hollow stems and skidding dam- depended on local seed production (ST), ages (Pulkki 1997), varied between 0.6to − − the rarer the late successional species were. 1.4 m3 ha 1 y 1 (Fig. 7.15). While differ- Conventional logging favoured abundance ences between recruitment scenarios were of early successional species and suppressed again negligible, the species origin of tim- development of late successional species. ber might be worth considering, as yield Temporal forest development showed dras- in ST scenarios was mostly achieved from tic increases of early successionals after con- mid successional species. SP scenarios sus- ventional logging, and short logging cycles tained a higher share of late successional did not give the forest the possibility to re- species. Reduced-impact logging achieved cover (Fig. 7.14). Furthermore, in scenario more than double yield compared to con- ST and conventional logging, a constant de- ventional scenarios in short cycles. Yield cline of total bole volume resulting from a was maximised in cycles of 40 y in RIL, missing seed supply was detectable. In con- where conventional logging achieved higher ventional logging with short cycles and lo- timber yields in long logging cycles. In cy- cal seed production, an increase in the num- cles length of 80 y timber yields converged ber of suspended logging operations was ob- towards similar yields independent of the served, because of a total lack of marketable logging method (Fig. 7.15). timber. Conventional logging with long cy- cles led to increasing fraction of early suc- cessional species. Extinction of late succes- Discussion sionals was detectable in all ST scenarios with CON after 100 simulated years, in RIL A comparison of simulated forest growth a very small fraction <1 % survived. with data from permanent plots was per- formed with a former version of the model, including a different parameter set (K¨ohler et al. 2001). The same tests were under- Growth & yield of rain forest in Sabah 103

residuals damaged logged

100 2 y=83.3-3.2x;. r =98.4% y=28.0+2.6x; r2=77.8% y=0.6+2.6x; r2=99.7% ...... 80 ...... 60 ...... 40 ...... 20 ...... 2 .... 2 ...... Basal area [%] y=70.7-5.1x; r =85.2% y=15.9+0.7x; r =77.9% ...... 0 ...... 0246810 0246810 024681012 Basal area removed [m2 ha-1]

Figure 7.12: Impacts of different logging methods (reduced-impact (dots), conventional (crosses)) and intensity on the forest structure.Basal area (residuals, damaged, logged) is plotted as function of removed basal area. taken with the current version (not shown). 1999; Curran & Leighton 2000; Curran & Thus, growth data and the simulation re- Webb 2000). In tests (not shown in the sults of 25 ha for 9 to 20 y in four differ- present study) we have compared the sen- ent forest reserves in Sabah with different sitivity of model output on regular mast- site and stocking conditions were compared. fruiting events. Populations of seed preda- The ratio of simulated data to field data of tors were not modelled explicitly, thus re- total basal area lay in the range of 0.9 to sults did not depend on mast-fruitings. The 1.25. The accuracy of these results is there- dynamics of stem numbers of small trees fore slightly better than in previous simula- (d<10 cm) is affected, but effects were tions. smoothed out for larger trees through self- The present knowledge of recruitment thinning. There were only minor fluctu- and recruitment rates in tropical forests is ations in stem volume and species com- still limited (ter Steege et al. 1995; Clark position. We therefore relied on the ap- et al. 1999b; Duncan & Chapman 1999; proach without mast-fruiting, but are aware Nicotra et al. 1999; Webb & Peart 1999). that seed predation might be the crucial Sensitivity analyses of simulation models of- bottleneck of current recruitment scenar- fer the possibility to evaluate the impor- ios. New findings in seed dispersal limi- tance of different processes on forest devel- tations through extinction of key dispersal opment. bird species in Brazil (da Silva & Tabarelli 2000) are another example of how im- Analysis of the effects of different param- portant fauna-flora interactions in tropical eters of the recruitment modules highlights forests are. But seeds in dipterocarp forests the fact that the influence of average seed are mostly wind dispersed, and those limita- dispersal distances XR on results is weaker tions might not be important in South-East than that of absolute seed production rates Asia. NS and NR (Fig. 7.2). In addition, the sensi- tivity analysis of the recruitment rates sug- Different assumptions on recruitment led gests that mid successional species have a to changes in species composition (Fig. 7.7), key role in the recruitment process, as their as analysed in greater details by Liu & Ash- recruitment strongly influences the specific ton (1999). However, the level of total bole composition of the forest. volume was reasonably constant (Fig. 7.6, 7.8, and 7.9). As a consequence it was im- Recent research highlights the impor- portant to isolate the effect of regeneration tance of mast-fruiting for recruitment suc- on forest development, e.g. by simulations cess in dipterocarp forests (Curran et al. 104 Chapter 7

Seed tree Mixed seeds 1 Mixed seeds 2 Mixed seeds 3 Seed pool 100 80 60 40 20

Bole volume [%] 0 20 40 60 80 20 40 60 80 20 40 60 80 20 40 60 80 20 40 60 80 Cycle length [y]

Figure 7.13: Effects of logging cycle (20 y, 40 y, 60 y, 80 y), method (reduced-impact (left), conventional (right)) and recruitment scenarios (seed tree; mixed seeds 1-3; seed pool) on species composition (early (white), mid (grey) and late (black) successional species).Simulations of 9 ha with toroidal boundaries.Results are means (n=5) of 240 simulated years.SE was always < 0.5 m3 ha−1 and thus not shown.T-test for matched pairs found significant differences at 1 % level for 86% and 80 % of tested pairs of abundance early and late successional species, respectively. without recruitment input (Fig 7.10). It The size of the simulated forest area turned out that changes in the recruitment plays a minor role in determining future for- rates will affect standing volume only with a est composition, if undisturbed forest de- time-lag. Thus, standing bole volume, num- velopment was considered. Even in small bers of saplings or mother trees are no ap- simulation areas the forest stand was able propriate indicators to detect early stages to supply itself with sufficient recruitment. of forest dieback. Important variables are This conclusion has to be seen in the con- the recruitment and mortality rates. Natu- text of the assumptions used. We are aware ral recruitment thus acts as a buffer sys- that field studies indicate dramatic changes tem for the stand development (Holling in mortality and recruitment rates in frag- 1973; Warner & Chesson 1985). Changes mented forests (Benitez-Malvido 1998; Lau- in seed production rates are buffered over rance et al. 1997; Mesquita et al. 1999), several orders of magnitude. It is not obvi- which was not considered in our studies so ous whether a threshold of seed production far. However, disturbances resulted in a rates exists below which the forest collapses loss of species within a short time-period, or degenerates to a large degree. Analysis if an external seed supply was missing, as of data from long-term research plots (Con- seen in the development of logged-over for- dit 1998) might disentangle this issue, and est (Fig. 7.9). Liu & Ashton (1999) found indicate for which values recruitment rates similar effects. They propose the establish- are low enough to draw our attention to ment of a seed zone around logging areas, a possible dieback in a given tropical for- which enables natural recruitment of the est. To study long-term trends in future disturbed forest. forest inventories we suggest paying more We found that the detailed description attention to recruitment rates, and record- of logging damages was important for a re- ing not only standing volume and estimat- alistic simulation of logging impacts. Log- ing recruitment potential not through exist- ging damages were depending upon logging ing young trees and saplings. In this con- intensity and were non-uniform in space. text the work of Curran and colleagues on It seemed important to distinguish differ- seed predation is remarkable and important ent types of logging damages, including one (Curran & Leighton 2000; Curran & Webb damage class which is proportional to the 2000). size of the felled tree. Other forest models Growth & yield of rain forest in Sabah 105

Seed tree Mixed seeds 2 Seed pool

500 400 20y 300 RIL 200 100 0 500

] 400 80y -1 300 RIL ha

3 200 100 0 500 400 20y 300 CON Bole volume [m 200 100 0 500 400 80y 300 CON 200 100 0 0 100 200 0 100 200 0 100 200 Time [y]

Figure 7.14: Development of bole volume (m3 ha−1) for different scenarios.Simulation of 9 ha over 240 y (n=5) of logged forest at site L1 with toroidal boundaries.Variation of logging cycle (20 y; 80 y), method (RIL: reduced-impact; CON: conventional) and recruitment scenarios (seed tree; mixed seeds 2; seed pool).Total (solid bold line), early successional spp.(solid line), mid successional spp.(broken dotted line) , late successional spp.(broken line) . assume only one type of damage (Boscolo Pinard & Putz 1996; van der Meer & et al. 1997; Liu & Ashton 1999; Huth & Bongers 1996; Weidelt 1996; Johns et al. Ditzer 2000a) or use regression equations 1996; Bertault & Sist 1997; Pulkki 1997; (Howard & Valerio 1992; Vanclay 1995). Brown 1998). For example, in a study in Resulting light climate in our logging sim- Kalimantan (Indonesian Borneo) damages ulations was more realistic and growth of to the residual stand varied from 30 % in early successional species in canopy gaps reduced-impact logging to 48 % in conven- benefit from spatial differences in damages. tional logging at an extraction intensity of As emerging property of the modelled log- 87 m3 ha−1 (Bertault & Sist 1997). ging methods the resulting linear relation- For conventional logging the highest ships between logging damages and logging yields were obtained by a logging cycle of intensity (Fig 7.15) was achieved. They cor- 80 y. In logging scenarios with short cycles respond well to field measurements (Jonkers the forest was overused, and had not enough 1987; Hendrison 1990; Cannon et al. 1994; 106 Chapter 7

Seed tree Mixed seeds 1 Mixed seeds 2 Mixed seeds 3 Seed pool ] -1 ]

2.5 y -1 . .

y . . . 1.5 -1 ...... 2.0 ...... -1 ha . . . . . 3 ha 1.5 . . . . . 1.0 . . . 3 . .

. . 1.0 . . . 0.5 0.5

Yield [m 0.0 0.0

20 40 60 80 20 40 60 80 20 40 60 80 20 40 60 80 20 40 60 80 Net yield [m Cycle length [y]

Figure 7.15: Effects of logging cycle (20 y, 40 y, 60 y, 80 y), method (reduced-impact (grey bars), conventional (black bars)) and recruitment scenarios (seed tree; mixed seeds 1-3; seed pool) on harvest yields.Yield (left scale) was reduced to net yield (right scale) by 30 % losses due to skidding etc.Simulations of 9 ha with toroidal boundaries.Results are means ± 1SE (n=5) of 240 simulated years.T-test for matched pairs found significant differences at 1 % level for 74% of tested pairs. time to regenerate. For reduced-impact log- Yield losses were mainly caused by high ging the highest yields were observed for abundance of early successional species af- a cycle of 40 y. The yield obtained with ter each logging event. Reducing light con- RIL methods were always higher than those ditions which favour those species can be obtained with conventional methods. For achieved by reducing the numbers and the short logging cycles the yield was doubled sizes of log landings and roads. Switch- in RIL compared to CON. Many of the cur- ing from the use of heavy and destructive rently practised selective logging systems in machinery as caterpillars to skyline yarding South-East Asia are based on logging cycles or cable supported systems will also reduce between 20 and 40 years and conventional damages and thus make shorter cycles more logging methods (Whitmore 1998). Thus, economical. our results suggest that these systems are Recruitment assumptions have only little not appropriate for the recruitment capabil- impact on yield, but strongly influence the ities of the forests and will lead to resource species composition. In fragmented forests depletion. without external seed input, the late succes- Yields obtained in the present study are sional species have nearly disappeared after similar to previous studies on the same the first logging operation. With an exter- area (K¨urpick et al. 1997; Ditzer et al. nal seed input it takes more than 200 years 2000; Huth & Ditzer 2000a). Maximal before these species recover to their natu- net extracted volumes ranged between 0.8- ral abundance. Thus, disturbances such as 1.5 m3 ha−1 y−1. Huth & Ditzer (2000) ob- logging in fragmented forests change species tain similar results for reduced-impact and composition drastically. conventional logging in cycles of 60 years These findings correspond to observa- and longer. Differences between the stud- tions in forest fragments. Growth of trees ies result from different assumptions on log- at the edge of fragmented forests is highly ging damages and different logging intensi- disturbed due to microclimatic changes and ties. In the present study, lower yields were elevated wind turbulence (Laurance et al. obtained than in the previous ones, espe- 1997). Regeneration rates are biased to- cially in conventional logging. This is an wards families of early successional species important result, all the more so because lo- (Laurance et al. 1998). Liu & Ashton (1999) cal recruitment was considered for the first showed in simulations that logging in frag- time here. mented forests will also reduce the number Growth & yield of rain forest in Sabah 107 of tree species. Nevertheless, undisturbed Acknowledgement forest fragments show also shifts in species composition (Turner 1996; Turner & Cor- Thanks to H. Bossel, J. Chave, R. Glauner lett 1996). As consequence of species group- and L. Kammesheidt for very helpful com- ings, our simulations suggest that these ments and improving suggestions on for- shifts occur inside the early, mid and late mer versions of the manuscript. P. K¨ohler successional species groups, a hypothesis was funded by the Otto-Braun-Foundation which should be tested. of the University of Kassel, Germany. The simulation results in this study may be optimistic regarding harvesting impact, because the model assumed that the soil in the logged forest provided suitable condi- tions for seeds to germinate and establish. In reality, soil in parts of the logged for- est may be compacted and, therefore es- tablishment and germination of seeds re- duced (Cannon et al. 1994; Pinard & Putz 1996; Johns 1997; Frederickson & Mostacedo 2000; Guariguata 2000; Pinard et al. 2000). Heavily compacted soils may loose also their nutrients due to erosion pro- cesses (Malmer 1996). Another problem might be the extraction of nutrient due to harvesting. In logged dipterocarp forest it will take 20 to 60 years to restore the normal nutrient amount in the ecosystem (Ruhiyat 1989; Bruijnzeel 1992; Glauner 1999). In other regions nutrient input due to precip- itation or rock decomposing may be much lower (Golley 1983; Bruijnzeel 1991). More- over, we have no knowledge yet which nu- trient level trees need (Whitmore 1998). In other words, logging impacts might be more serious than those shown in the simula- tions. Nevertheless, the model provides a first assessment of impacts of different har- vest regimes. Even if we might have under- estimated the impacts of timber harvesting on recruitment capabilities, simulated im- pacts are still large enough to warrant at- tention to present tropical forest manage- ment. 108 Chapter 7 Appendix

Table 7.6: Short description of parameters incl.functional relationships (modified from Kammesheidt et al.2000).

Parameter Description

Environmental parameters k Light extinction coefficient I0 Light intensity above canopy SD Day length

Establishment parameters DS Initial diameter of seedlings IS Minimal light intensity for germination NS Ingrowth rate of seeds into seed pool NR Seed dispersal rate of mother trees XR Average seed dispersal distance DR Minimum diameter of mother trees

Mortality parameters MB Basic mortality rate MS Mortality rate of seeds MD Size dependent mortality rate (MD = MD0 − MD0/MD1 · d) W Probability of a dying tree (d>10 cm) to fall

Tree physiognomic parameters HM Maximum height cp Crown length fraction τj Site dependent fraction of stemwood biomass to total above-ground biomass (τ = τ1 + τ2 · h(d = 120cm)) h0 and h1 Height = f(diameter) (h = d/(1/h1 + d/h2)) γ2 γj Form factor = f(diameter) (γ = γ0 · exp(γ1 · d )) f2 fj Crown diameter = f(diameter) (dc =(f0 + f1 · d ) · d) 2 3 lj Leaf area = f(diameter) (l = l1 · d + l2 · d + l3 · d ) LAIM Maximum leaf area index of single tree

Biomass production parameters PM, α Photosynthetic capacity and efficiency in light response curve (Pi(Ii)= α · i i α I /(1 + PM I )) ρ Stem wood density r1 Maintenance respiration = f(biomass) (Rm(Bi)=r1 · Bi) RG Growth respiration as part of biomass m Leaf transmittance g Conversation factor gCO2 to godm Growth & yield of rain forest in Sabah 109

Table 7.7: Parametrisation for Sabah (Malaysia).Short description of parameters found found in Table 7.6. Parameters with subindex vary with successional status (s), potential height (h) (corresponding to SS and HG in Table 7.1, respectively), or different functional coefficients (j).

Parameter Subindex Units Values

Environmental parameters k [-] 0.7 µmol(photons) I0 [ m2s ] 642.0 SD [h] 12.0

Recruitment parameters DS [m] 0.01 ISs s=1-3 [fraction of I0] 0.20 0.04 0.01 −1 −1 NSs s=1-3 [ha y ] 150 625 50 −1 −1 NRs s=1-3 [tree y ] 100 20 4 XRs s=1-3 [m] 100 75 50 DRh h=1-5 [m] 0.04 0.10 0.18 0.40 0.50

Mortality parameters −1 MBs,h s=1; h=1-5 [y ] 0.00 0.12 0.10 0.08 0.06 −1 MBs,h s=2; h=1-5 [y ] 0.06 0.05 0.04 0.03 0.025 −1 MBs,h s=3; h=1-5 [y ] 0.00 0.04 0.03 0.02 0.015 −1 MSs s=1-3 [y ] 0.1 0.5 1.0 −1 −1 MDj j=0-1 [y ,m ]0.20.1 W [-] 0.40

Tree physiognomic parameters HMh h=1-5 [m] 5 15 25 36 50 cp [-] 0.358 − τj j=0-1 [-, m 1] -0.035 0.0139 −1 h0h h=1-5 [cm m ] 1.24 1.18 0.97 1.08 1.33 −1 h1h h=1-5 [m ] 38.5 43.6 88.6 57.3 70.5 − γj j=0-2 [-, cm 1, -] 2.575 -1.409 0.0358 fj j=0-2 [-, -, -] 0.132 0.933 -0.6615 m m m j l j=1-3 [ cm , cm2 , cm3 ] 3.197 0.0684 -0.000379 LAIM [-] 2

Biomass production parameters µmol(CO2) PMs s=1-3 [ m2s ] 19.4 9.3 6.8 µmol(CO2) αs s=1-3 [ µmol(photons) ] 0.043 0.043 0.043 −3 ρs s=1-3 [todm m ] 0.37 0.55 0.75 r1s s=1-3 [-] 0.12 0.05 0.02 RG [-] 0.25 m [-] 0.1 −1 g [godm gCO2]0.63

Chapter 8

Long-term response of tropical rain forests to the effects of fragmentation: a simulation study

Peter K¨ohler+,J´erˆome Chave∗, Bernard Ri´era§ and Andreas Huth+ +: Center for Environmental Systems Research, University of Kassel Kurt-Wolters-Str. 3, D-34109 Kassel, Germany *: Department of Ecology and Evolutionary Biology, 216 Eno Hall Princeton University, Princeton NJ 08544-1003, USA §: Laboratoire d’Ecologie G´en´erale, CNRS–MNHN, UMR 8571, 4 avenue du Petit Chˆateau, 91800 Brunoy, France

Abstract

The impacts of forest fragmentation on the residual stands have only recently come to attention.We combined the current knowledge of a higher tree mortality rate at the forest edges and a lower recruitment capability of tree species in forest fragments in a comprehensive simulation study for a rain forest site in French Guiana using the individual- oriented and process-based forest growth model Formind2.0.Simulations were based on the aggregation of the tree species into 19 different plant functional types.Specially, we investigated the spatial and temporal effects of the fragment size (1–100 ha) on forest dynamics, species composition, and future recruitment potentials in terms of the number of seed producing trees.Biomass in small scale forest fragments declined after 300 years below 10 % of the primary forest level in the most extreme scenario.In particular, the decreasing recruitment capability limited forest growth.Furthermore, different intensities and pattern of forest clearing were analysed in the simulation of a forest/non-forest landscape of 1 km2 over several centuries.The cleared areas were abandoned after 300 simulated years.Thus, the regrowth of a secondary succession as function of size and shape of forest fragments was also investigated. Keywords: fragmentation; modelling; rain forest; secondary succession; simulation. To be submitted. 112 Chapter 8

Introduction ies show clear differences in secondary veg- etation dynamics depending on disturbance The fraction of forested area being clear- type (Uhl et al. 1988; Aide et al. 1995). cut is still high around 1 % per year on However, direct data on secondary regrowth average (FAO 1997; Whitmore 1998; Malhi are rarely older than 50 years (Finegan & Grace 2000). Tropical forests may soon 1996). have shrunk to quite small patches of iso- While long-term assessements are obvi- lated remnants. Indeed, from the small- ously required, our current knowledge can scale slash-and-burn agriculture, to large- be gathered into an integrative modelling scale clearing for monocultures (sugar cane, approach, in order to relate the recent find- oil palm, ecalyptus or hevea) the overall im- ings of field ecology to conservation issues pact on the ecosystems should be very dif- (Kremen et al. 2000). Recently, simple ferent. There has been a strong effort in try- models on forest fragmentation have been ing to understand the ecology of these frag- published. However, they only take into mented forests (for a review, see Saunders account global, spatially-implicit variables, et al. 1991; Murcia 1995; Laurance 1999a). such as the core area of a forest frag- Observations in Australia (e.g. Crome 1991; ment left undisturbed (Laurance & Yensen Law & Lean 1999), Thailand (e.g. Ly- 1991), or persistence probability of plant nam 1997; Lynam & Billick 1999), French species within a land clearing (Laurance Guiana (e.g. Thiollay 1989; Thiollay 1992; et al. 1999). Moreover, the long term con- Cosson et al. 1999), the Brazilian Atlantic sequences of fragmentation over large scale forest (e.g. da Fonseca 1985; Tabarelli et al. patterns in tropical ecosystems are still 1999) and central Amazon (e.g. Lovejoy poorly understood at the community-level et al. 1984; Gascon et al. 1999) have en- (Brown et al. 1995, Chave et al. 2000c). In riched our current knowledge of processes in the temperate zone, a modelling approach animal and plant communities within for- has been used to analyse the effects of frag- est fragments and with respect to scale. mentation on tree species diversity (Malan- These studies have shown that a forest com- son 1996; Malanson & Armstrong 1996). In munity, when restricted to small discon- the present paper, we provide a more de- nected fragments, shows dramatic changes tailed analysis for the tropical zone. Our in its dynamics. For example, the mor- approach combines the current knowledge tality in emergent mature trees increases, of mortality and recruitment patterns in establishment of light demanding pioneer fragmented rain forests in the framework plants and potentially invading plants is of the individual-oriented and process-based promoted, local microclimatic conditions forest growth model Formind (K¨ohler & are changed, leading to higher daily tran- Huth 1998a). This model describes the spiration (Nepstad et al. 1999). Likewise, spatio-temporal dynamics of a mixed for- the age structure of animal populations is est stand for an area up to several km2 altered (Woodroffe & Ginsberg 1998; Cur- based on a carbon cycle model for different ran et al. 1999; da Silva & Tabarelli 2000). plant functional types. The model was ap- These factors result in a significant dimi- plied to the rain forests of Malaysia (K¨ohler nuition of species diversity and a modifica- & Huth 1998a; K¨ohler et al. 2001) and tion of the species composition (Laurance the neotropical moist forests of Venezuela & Bierregaard 1997; Tabarelli et al. 1999). (Kammesheidt et al. 2000) to address forest Thedisturbedlandscapeshaveapotential management issues such as the long-term to recover from clearings, but abandoned impacts of logging on the forest dynamics. pastures are rapidly invaded by regrowing Here, we address the issue of fragmentation secondary forests (Saldarriaga et al. 1988; with a new parameterization of our model, Brown & Lugo 1990). Existing field stud- Regeneration of fragmented rain forest in French Guiana 113 for the rain forest of French Guiana. southwards. A pronounced dry season of We shall assess the generality of our re- 2 months is recorded from September to sults with respect to the forest type. November and a short dry season in March. The average temperature is 25.8 ◦Cwithan We shall here focus on the influence of annual amplitude of 2 ◦C and daily ampli- the scale and of the shape of the disturbed tudes of 7 ◦Cintherainyseason(10◦C dur- area, a topic that has quite a long history in ing the dry season). Geology is typical of the context of plant ecology (Levin & Paine the Guiana Shield with a central pediplain 1974). However, our aim is not only theo- and sparse rugged mountains of Precam- retical, and we aim to come up with reason- brian metamorphic and granitic rocks. The able suggestions for land-use and landscape- altitude is less than 500 m above sea level. scale management of wet tropical areas. The observed forest of La Piste de Saint- Specially, we want to address the follow- ◦ ◦ ing questions: (1) How do the size and the Elie (5 30’ N, 53 00’ W) is located 16km shape of a fragmented forest influence both south of Sinnamary and has been much forest dynamics and species composition? studied since 1976(Lescure et al. 1983, (2) Do different patterns of fragmentation Lescure & Boulet 1985, Puig et al. 1990, in a landscape influence spatio-temporal dy- Pelissier & Ri´era 1993, Roggy et al. 1999). namics? (3) What is the forest structure Annual rainfall is slightly above 3 000 −1 after clear-cut areas, such as pastures, are mm y . The forest mostly grows on a abandoned? schist mantle covered by a sandy clayey soil (Lescure & Boulet 1985). A 5 ha forest inventory where all the 2740 trees ≥ 10 cm in dbh have been tagged, Methods measured and mapped has been used for this study. 261 species (or morphospecies) Study area have been found in a sample of 2475 indi- viduals (Tab. 8.1). Fisher’s α diversity pa- Our study focuses on the French Guiana rameter gives α =73.3 which is quite large rain forest. This forest is reminiscent of (Gentry 1988; Valencia et al. 1994; Leigh-Jr. the ’Wallaba’ forest type (Richards 1996). & de Lao 2000). The forest is dominated by Dominant families are Lecythidaceae and Lecythidaceae (22 spp., 30.0 % of the trees, Caesalpiniaceae (Mori 1990; Poncy et al. 22.4 % of the basal area), Caesalpiniaceae 1998). (19 spp., 18.4 % of the trees, 31.4%ofthe basal area) and Chrysobalanaceae (24 spp., The forest in French Guiana is one of 9.3 % of the trees, 10.1 % of the basal area). the best protected in South America, due The average basal area is 30.2 m2 ha−1. to low anthropogenic pressure and high la- bor cost, having made this forest unattrac- tive for logging companies, for the last 60 years at least. French Guiana lies be- The model tween 2◦10’ North and 5◦45’ North and 51◦40’ West and 54◦30’ West in Northeast- Formind2.0 is an individual-oriented and ern South America. Of the region, 97 % is process-based forest growth model which covered by a pristine lowland wet tropical simulates the spatial and temporal dynam- forest, which extends, northwards, to the ics of uneven-aged mixed forest stands. Surinam, Guyana and western Venezuela, A complete model description has been and, southwards, to Brazil. Annual rain- published in previous papers (K¨ohler & fall is between 1650 and 4000 mm, with a Huth 1998a; Kammesheidt et al. 2000; decreasing gradient from the coastal area K¨ohler et al. 2000c). Nonetheless, since we 114 Chapter 8

Table 8.1: Density and basal area of the major tree species at the Piste de Saint Elie Research Station.Of 2740 trees ≥ 10 cm dbh in the 5 ha plot, 2475 (90.3 %) were identified to the species.The actual number of species recorded was 261, and the total diversity extrapolated using Fisher’s α is 267 species.

Rank Species No. % Basal area % trees (m2 ha−1)

1 Lecythis idatimon Aubl. (Lecyth.) 241 9.7 1.78 6.3 2 Eperua falcata Aubl. (Caesalp.) 204 8.2 1.07 3.8 3 Lecythis persistens Sagot (Lecyth.) 133 5.4 0.79 2.6 4 Eschweilera micrantha (Berg) Miers (Lecyth.) 119 4.8 0.8 3.0 5 Eschweilera sagotiana Miers (Lecyth.) 81 3.3 1.2 4.3 6Licania alba (Bernoulli) Cuatrec. (Chryso.) 77 3.1 1.1 4.0 ... 11 Eperua grandiflora (Aubl.) Benth. (Caesalp.) 52 2.1 1.1 3.7 ... 15 Dicorynia guianensis Amsh. (Caesalp.) 40 1.60.8 3.0

Total 947 38.3 8.7 30.6 Grand total 2475 100 28.4 100

have parametrized our model specifically for self-thinning in dense patches or through the Guiana rain forest, basic informations falling of large trees (gap formation). The about the model and the species groups can dispersal of seeds produced by mature trees be found in the Appendix. is, beside gap formation, the other major The model describes a forest stand as a cause of spatial correlation in the model. mosaic of interacting patches of 20 m × The seed production rates of mature trees 20 m in size. Within these patches trees correspond to the reproductive success at are not spatially-explicit distributed, and dbh 1 cm. Thus, the rates include fecun- thus compete for light and space following dity, seed survival, germination and possi- the gap model approach (Botkin 1993; Liu ble predation upon young seedlings (Chave & Ashton 1995; Shugart 1998; Bugmann 1999b). & Solomon 2000). The carbon balance of each individual tree is modelled explicitly including as main processes photosynthe- Species grouping, parametrisa- sis and respiration. Allometric relationships tion and initialisation relate the above-ground biomass, the stem diameter, the tree height and the crown di- A species list covering 1022 tree and shrub ameter with each other. Growth process species found in French Guiana was con- equations and physiological parameters are structed (Chave 1999a), extending the work Formix3-Q taken from a related model of Favrichon (1995), using the data of van (Ditzer et al. 2000). Beside normal mortal- Roosmalen (1985) and some new informa- ity, death of trees can occur either through tion (Chave & Ri´era, in preparation). Nine- Regeneration of fragmented rain forest in French Guiana 115

Table 8.2: Autecological characteristics of 19 plant functional types (PFT) of French Guiana’s tropical rain forest tree species.Height at maturity.SS: successional status.HG: height group. No: Number of species per PFT.Ab: Abundance of trees with dbh ≥ 10 cm in research plots in Nouragues (from the AUBLET database), Piste de Saint-Elie, and Paracou (from Favrichon 1995).

Plant Functional Type Height [m] PFT SS HG No Ab [%]

Shrub savanna spp. 0-5 1 0 1 7 0.0 Shrub early successional spp. 0-5 2 1 1 17 0.1 Shrub mid successional spp. 0-5 3 2 1 29 1.0 Shrub late successional spp. 0-5 4 3 1 83 2.0

Understorey savanna spp. 5-15 5 0 2 31 0.0 Understorey early successional spp. 5-15 61 2 74 1.4 Understorey mid successional spp. 5-15 7 2 2 763.5 Understorey late successional spp. 5-15 8 3 2 152 5.4

Lower canopy savanna spp. 15-25 9 0 3 7 0.0 Lower canopy early successional spp. 15-25 10 1 3 48 2.8 Lower canopy mid successional spp. 15-25 11 2 3 84 11.4 Lower canopy late successional spp. 15-25 12 3 3 122 18.1

Upper canopy savanna spp. 25-3613 0 4 2 0.0 Upper canopy early successional spp. 25-3614 1 4 38 1.6 Upper canopy mid successional spp. 25-3615 2 4 6722.2 Upper canopy late successional spp. 25-36163 4 103 21.0

Emergent early successional spp. >3617 1 5 12 0.9 Emergent mid successional spp. >3618 2 5 38 6.4 Emergent late successional spp. >3619 3 5 32 2.2

teen different plant functional types (PFT) the species into shrubs (0–5 m), understorey based on successional status and maximum (5–15 m), lower canopy (15–25 m), upper tree height at maturity were assigned (Ta- canopy (25–36m), and emergents (more ble 8.2). The grouping of tree species uses than 36m) similarly to Sabah’s rain forest an approach already described for a rain for- species grouping (K¨ohler et al. 2000b). est community in Sabah, Malaysia (K¨ohler Parameter values (see Appendix) were et al. 2000b). The species list (available gathered from a former case study in French from the authors) contains savanna species, Guiana (Chave 1999b), and from param- which are considered as extreme pioneers. eter variations and sensitivity analysis to In addition, we consider three classes of suc- match typical diameter increment pattern cessional behaviour for forest species (early, (Gourlet-Fleury 1997; Gourlet-Fleury & mid and late successional species), and five Houllier 2000). Model structure and its sen- classes of maximum tree height. We classify sitivity to parameter variations were inves- 116 Chapter 8

3.0 1.0 2.5 0.8 0.6 2.0 0.4 1.5 0.2 0.0 1.0 Reduction factor [-] Correction factor [-] 0 20 40 60 80 100 120 1 49 100 Distance from edge [m] Size [ha] Figure 8.2: Reduction factor of recruitment Figure 8.1: Correction factor of tree mortal- rates in small forest fragments as function of ity rates at forest edges.Black: correction the system size.In fragments larger than factor for large trees with d ≥ 60 cm in the 100 ha recruitment is not altered. neighbourhood of cultivated lands; grey: cor- rection factor for trees with d<60 cm in the neighbourhood of cultivated lands; white: cor- the type of edge effect (Mesquita et al. rection factor for trees in the neighbourhood of 1999). Overall, emergent trees (d≥60cm) secondary succession for the first 20 y.Mortal- are most sensitive to edge effects both be- ity in patches more than 100 m inwards is not affected.All distances between patches were cause of a larger impact of winds and be- computed between patch centers. cause of changes in microclimatic conditions (Laurance et al. 2000). We have included these mechanisms into a correction factor tigated in detail and reported in previous for mortality rates. Tree mortality rates reports (Kammesheidt et al. 2000; K¨ohler are explicit functions of the distance to the et al. 2000c). nearest forest edge, of the type of surround- ing vegetation and of tree size (Fig. 8.1). By simulating the long-term dynamics of the 5 ha inventory (only trees with dbh Fragmentation also leads to a drastic al- ≥ 10 cm) with constant recruitment input, teration of the recruitment. Once the forest we generate a stem-dbh distribution of an matrix has been sufficiently fragmented, the undisturbed forest. It reflects for trees with density of various animal and bird popula- dbh ≥10 cm the characteristics of field data. tions, the main seed dispersers of neotrop- This generated forest stand at equilibrium ical plants, was found to decrease signifi- is then used as the initial stand for our sim- cantly, resulting in a reduction of the seed ulations. dispersal, as well as on the predation rates on seedlings (Laurance et al. 1998; Lawton et al. 1998; Law & Lean 1999; Lynam & Fundamental mechanisms in Billick 1999; Price et al. 1999; da Silva & forest fragments Tabarelli 2000). Thus, the number of in- growing seedlings per year (at a 1 cm dbh) The dynamics of the forest relies strongly depends critically upon the size of forest upon any fragmentation. Tree mortality fragments. We have therefore introduced a was observed to increase at the edge of the hypothetical reduction factor on seed pro- fragmented remnants. Fragmentation re- duction rate as function of fragment size (Benitez-Malvido 1998) and we have as- sults in microclimatic changes as well as en- 2 hanced wind disturbances (Laurance et al. sumed that areas larger than 1 km were 1997; Laurance et al. 1998). Trees af- unaffected by these effects (Fig. 8.2). fected by edge effects were found up to 100 m inwards, and the nature of surround- ing ecosystems also had a great impact on Regeneration of fragmented rain forest in French Guiana 117

Table 8.3: Description of different fragmentation scenarios.Mortality is only altered in a border zone of 100 m (Fig.8.1),and, if switched on, affect all size classes ( a) or can depend on the size-class (b).In scenario fragmented 3, the recruitment is affected over the whole area (Fig.8.2).

Scenario Boundary Increase in Reduction in condition mortality recruitment (global)

periodic periodic no no open 1 open yesa no open 2 open yesb no open 3 open yesa yes

Scenarios Within a landscape of 1 km2 CF =0%to 90 % of the total area was cleared and con- To address the issue of scale and shape us- verted into non-forest areas (values of CF ing our model, we have performed several every 10 %). Different pattern of fragmen- computer experiments. tation were investigated. The null-model Scale: Seven different fragment sizes — of fragmentation (Fahrig 1992), which con- from 1 to 100 ha —, and four different sists in removing patches at random (sce- fragmentation assumptions (scenarios non- nario random), was compared with simula- fragmented, fragmented 1 to fragmented 3) tions with some spatial correlation in the were matched, which resulted in possible fragmentation process. Weakly, medium 7 × 4 = 28 scenarios. The assumptions and highly clustered fragmentation scenar- on fragmentation included boundary condi- ios were investigated (scenarios clustered 1 × tions and alterations of mortality and re- to clustered 3, respectively), giving 9 4= cruitment rates (Table 8.3). We have used 36scenarios. A mathematical interpreta- two types of boundary conditions: periodic, tion of this spatial correlation can be found or toroidal condition (opposite ends are in Moloney & Levin (1996). The regrowth wrapped), and open boundaries (free flow of a secondary succession after the cleared of seeds outwards). The former assumes landscape has been abandoned at year 300 that the simulated area is embedded in a was analysed for another 300 years. Seed larger forest with a similar structure. With dispersal was made possible throughout the the latter, the landscape outside the simu- landscape. For this set of scenarios, the lated area is different from a mixed tropical edge correction of mortality rate was not de- rain forest (agriculture, pasture, monocul- pending on tree size, and assumed reduced tural forests). The sensitivity of forest dy- edge effects in the first 20 y of secondary namics to boundary conditions was inves- succession (Fig. 8.1, Mesquita et al. 1999). tigated recently (K¨ohler et al. 2000c). For The nutrient budgets of the abandoned ar- all sizes the runs contained of 300 simulated eas were kept constant. These scenarios in- years and were replicated (n =5). cluded periodic boundary conditions. Shape: Moreover, we have studied the Theanalysisofvariance(ANOVA)was spatio-temporal dynamics of fragmented restricted to the special case with n =1 landscapes to assess the long-term impact (Graf et al. 1987). of fragmentation on the forest dynamics. 118 Chapter 8

In our analysis we have used four differ- only slightly affected (Fig. 8.3). The to- ent indicators to keep track of the various tal above ground biomass was almost un- effects of a fragmentation experiment on a affected by the system size in the non- control pristine forest. We have monitored fragmented scenario; it fluctuated around (i) the standing above-ground biomass, (ii) 445 ± 15 Mg ha−1. The biomass fraction the abundance of early and late successional of early successional species was negligible species, as indicator for species composi- (around 4 %), while that of late succes- tion, (iii) the number of large trees, capable sional species was 54 ± 2 % of the to- for seeds production (for the height groups tal biomass. Finally, the density of mature upper canopy and emergents), as an indica- trees was slightly less than 20 ha−1 (upper tor of size distribution and potential recruit- canopy) and 17 ha−1 (emergents). ment limitation, (iv) the spatial distribution The forest dynamics of selected scenarios of dominant trees in fragmented landscapes. are plotted in Fig. 8.4. The three scenar- Successional status and height of dominant ios of fragmented forest situations raised the trees in each patch is shown in those fig- same global pattern, albeit a stronger influ- ures. As simulations were undertaken with ence of finite size effects. Below a system 19 PFTs, aggregation of results is necessary size of about 50 ha, the total above ground for interpreting them. biomass was below 300 Mg ha−1.Atmaxi- mum it reached 400 Mg ha−1 for the largest simulated fragment (100 ha). Early succes- Results sional species took advantage of edge dis- turbances and reached values around 10 % of the total biomass. The density of ma- Influence of fragmentation in- ture trees was also slightly reduced (21 % tensity and of system size on decrease for upper canopy species, and 6 − forest dynamics 19 % decrease for emergents). For the 1 ha runs, standing above-ground biomass was Fragmentation affected the simulated area reduced to 30 % compared with the con- in various ways. For example, mortality trol run after 300 y (scenario fragmented 1), was increased in the border zone, whereas 25 % (scenario fragmented 2), and 7 % (sce- recruitment decreased over the whole area. nario fragmented 3). Mature trees com- Therefore, nontrivial correlations might be pletely disappeared from the 1 ha plot in unveiled when the size of the simulated scenario fragmented 3. Thus, species com- area and fragmentation assumptions were position and size structure in small scale changed. fragments were altered dramatically. When recruitment is altered through fragmenta- We found that the influence of both sys- tion (fragmented 3) the community was still tem size and fragmentation method had a not in a steady state after 300 y, and abun- significant influence on the total biomass dance of early successional species was still and on the fraction of early successional increasing for all sizes <100 ha. As abun- species (ANOVA: P<0.001, see Table 8.4). dance of savanna species was always be- Late successional species were significantly low 1 %, each shift in species composition affected by size (P<0.01), but not by the strongly influences mid-successional species fragmentation type (P>0.1). The number as well. of seed trees was significantly affected by all scenarios (P<0.001). Fragmentation mostly influences forest edges. Overall, the fraction of patches con- In the control run (scenario non- taining trees larger than 30 m in height de- fragmented), size influenced the variance of clined between the control run (from 91 − diagnostics, but time averaged values were Regeneration of fragmented rain forest in French Guiana 119

] non-fragmented fragmented 1 fragmented 2 fragmented 3 -1 500

. . .

. . .

. ha

.

.

. .

.

.

.

.

.

. 400 .

.

.

. odm

.

. 300 . . total

200 .

.

100 . 0 Biomass [Mg 80 60 early ss 40

.

.

.

.

20 .

.

. .

.

.

.

. . .

. .

. . .

.

.

. . . .

.

. 0 . 80

. Biomass [%]

.

.

.

......

. . .

. . . 60 . .

.

.

.

.

. . late ss 40 20 0

.

.

. .

. . 20 .

.

. .

.

.

.

.

.

. .

.

.

.

.

15 .

.

. ] upp. can. -1 10 .

.

.

5 . 0 20

.

. .

.

.

.

. Seed trees [ha

.

.

.

.

.

. .

. .

15 .

.

.

.

. . emerg.

.

. 10 .

.

. 5 . 0 94981949819498194981 Size [ha]

Figure 8.3: Above-ground biomass (row 1-3) and number of seed trees (row 4-5) as function of different fragmentation scenarios (non-fragmented and fragmented 1–3, see Table 8.3 for details) and spatial scale (1–100 ha).Row 1: total biomass; row 2: fraction of early successional species; row 3: fraction of late successional species; row 4: number of seed producing trees of upper canopy species; row 5: number of seed producing trees of emergent species.Bars show the time averaged values, including standard deviation.Crosses correspond to values at t = 300 y.Data are averages over n = 5 runs. 120 Chapter 8

Table 8.4: Forest dynamics in a single forest fragment.ANOVA for the main and interactive effects of size (1 ha to 100 ha) and fragmentation assumptions (boundary conditions, increased edge mortality and reduced recruitment) on the following monitored quantities: (A) total above- ground biomass, (B) fraction of early successional species on biomass, and (C) fraction of late successional species on biomass.Simulation time was 300 y ( n =5).

Effects SS df MS F ratio P

A) Biomass

Size 544801.02 6 90800.17 3699.61 <0.001 Fragmentation 397447.43 3 132482.48 5397.94 <0.001 Size × fragmentation 196128.34 18 10896.02 443.95 <0.001 Error 2748.83 112 24.54

B) Early successional species

Size 1024.70 6170.78 11.05 <0.001 Fragmentation 1381.77 3 460.59 29.80 <0.001 Size × fragmentation 776.51 18 43.14 2.79 <0.01 Error 1730.97 112 15.46

C) Late successional species

Size 275.66 6 45.94 3.06 <0.01 Fragmentation 62.48 3 20.83 1.39 >0.1 Size × fragmentation 645.10 18 35.84 2.39 <0.01 Error 1682.04 112 15.02

93 % for all system sizes) and fragmented by the fragmentation shape (P<0.001). regimes. Again, the impact was quite lim- However, the abundance of late successional ited in large areas (100 ha), which were only species was not significantly influenced by slightly altered, while in 1 ha simulations the shape of the forest fragments (P>0.1, in most extreme scenarios few to no plots Table 8.5). The biomass per ha, as well as were dominated by large trees (32 %, 16%, the fraction of this biomass corresponding and 0 % for fragmented 1, fragmented 2,and to early successional species was compare- fragmented 3, respectively). able with the control run. When the clearing fraction (CF)in- creased close to 50 %, the abundance of Influence of the fragmentation early successional species saturated at 33 % shape at the landscape scale for the case of fragmentation without clus- tering (null-model). The total biomass loss The total biomass and the abundance of also reached a maximum around CF =0.5. early successional species were both signifi- Differences in the shape of the fragmenta- cantly affected by the clearing fraction and tion procedure only led to slight changes in Regeneration of fragmented rain forest in French Guiana 121

non-fragmented fragmented 1 fragmented 2 fragmented 3 500 400 300 200 1ha 100 0

] 400 -1 300 ha 200 25ha odm 100 0 400 300 Biomass [Mg 200 64ha 100 0 400 300 200 100ha 100 0 0 100 200 0 100 200 0 100 200 0 100 200 300 Time [y]

Figure 8.4: Time variation of the above-ground biomass for selected fragmentation scenarios (non-fragmented and fragmented 1–3, see Table 8.3 for details) and spatial scale (1-100 ha). Total biomass (solid bold line), biomass for early successional species (solid line), for mid-successional species (broken dotted line), and for late successional species (broken line).Savanna species are always below 1 Mg ha−1 and are not shown.

the dynamics. At maximum fragmentation > 30 m decreased from 91 % (control run) to (CF = 90 %) scenario clustered 3 achieved 28−34 % at year 300 and with CF =90%. highest value in total biomass (51 % of Large dominant trees were most sensitive to reference case compared to 32 − 37 % in fragmentation patterns in clearing scenarios other scenarios). Abundance of early suc- of medium intensity (e.g. CF =50%,dom- cessional species increased from 4 ± 1% inance varied between 39 − 70 %). in the control run, to 27 − 33 % at CF = 90 %. Late successional species were rela- tively unaffected in all scenarios (43−56%) Regeneration and secondary (Figs. 8.5, 8.7). succession in abandoned lands Undisturbed forest remnants persisted in the center of forest fragments (Fig. 8.8 top). Besides the short-term impact of framen- The fraction of patches dominated by trees tation upon the ecology of the forest rem- 122 Chapter 8

Table 8.5: Forest dynamics in partially fragmented landscape.ANOVA for the main effects of clearing fraction (CF =0− 90 %) and shape of the fragmentation pattern (random, three types of clustering) on the following monitored quantities: (A) total above-ground biomass, (B) fraction of early successional species on biomass, and (C) fraction of late successional species on biomass.Simulation time was 300 y, simulated area was 1 km 2.Sample size n = 1 and no cross-correlation.

Effects SS df MS F ratio P

A) Biomass

Clearing 185746.38 9 20638.49 36.86 <0.001 Pattern 30521.83 3 10173.94 18.17 <0.001 Error 15118.98 27 559.96

B) Early successional species

Clearing 545.04 9 60.56 15.97 <0.001 Pattern 164.62 3 54.87 14.47 <0.001 Error 102.37 27 3.79

C) Late successional species

Clearing 64.91 9 7.21 3.75 <0.005 Pattern 6.00 3 2.00 1.04 >0.1 Error 51.94 27 1.92

nants, our model allowed us to investigate cf. Fig. 8.7). In all scenarios a steady long-term trends. Of central interest is the state was not reached after 300 simulated question of how previously clear-cut areas years, but the abundance of early succes- regenerate once they are abandoned by hu- sional species declined to closed forest lev- man exploitation. els (around 5 %) in all cases except for ≥ In the time-averaged pattern of sec- CF 80 % and clustered 3, were this frac- − ondary succession, all monitored variables tion ranged between 7 12 %. — total biomass, abundance of early and In the final years of the simulation, the of late successionals — were significantly important aspects of the dynamics were affected by the fraction of clear-cut land, the height growth of established long living CF (ANOVA: P<0.001). None of the trees and species shift between mid to late monitored quantities was sensitive to the successional species. Abundance of late suc- shape of the disturbed area (P>0.01, cessionals ranged between 15−56% at year P>0.5, P>0.01, Tab. 8.6), but the dom- 300 (Fig. 8.6). The total standing biomass inance of early successional species lasted reached 264-306 Mg ha−1 or 59 − 68 % of longer in highly clustered landscapes (200 y the closed forest level reached in the control in scenario clustered 3 versus only 100 y simulation (448 t ha−1), for CF =90%. in scenario random, both for CF =90%, Even at medium clearing intensity (CF = Regeneration of fragmented rain forest in French Guiana 123

] random clustered 1 clustered 2 clustered 2 -1 500

. .

. . ha

.

.

.

.

.

. .

.

.

. . 400 .

.

.

.

.

. odm

.

.

.

.

.

.

.

.

.

300 . .

.

.

.

. . total .

. 200 . 100 0 Biomass [Mg 80 60 early ss 40

.

.

.

.

.

.

.

.

.

.

.

.

. 20 .

.

.

.

.

.

.

.

.

.

. . .

.

.

. .

.

. . . .

. 0 . . . . 80

. Biomass [%]

.

.

......

......

. .

. . . . . 60 .

. . . . .

. .

. . .

. .

. . late ss 40 20 0 0306090030609003060900306090 Clearing fraction CF [%]

Figure 8.5: Patterns with a fraction of landscape being cleared (t =0− 300 y).Above-ground biomass as function of cleared fraction (CF) and pattern formation (random, clustered 1 − 3). Bars show time averages, (including SD), crosses correspond to values at t = 300 y.Total biomass (1st row), and fraction of early and late successional species (2nd and 3rd row).

50 %) the final biomass was not totally re- Discussion covered after 300 years (82 − 84 % of the control run). The simplifications of the modelled regen- A landscape with one intact forest rem- eration processes might need further dis- nant (clustered 3, CF = 90 %), will cussion. While usual pollinators and seed- not regrow as quickly as forests with dispersers of the species found in the pris- other spatial distribution of forest frag- tine forest have disappeared, they are often ments. Indeed, long distance seed disper- replaced by invasive animals of the neigh- sal favour early successional species in the boring areas which might over-predate the latter case (Fig. 8.8 bottom). Large tree relict flora (Curran et al. 1999; Curran (height > 30 m) dominating patches in & Leighton 2000; Curran & Webb 2000). the final year were fewer in scenario clus- There is also evidence that large carnivores tered 3 (42 %), compared with others sce- which prey upon herbivores, and which narios (49 − 60 %, all scenarios with CF = may stabilize the food web by prevent- 90 %). Again, in medium clearing intensity ing the over-grazing are threatened with (CF = 50 %) differences across patterns are extinction even inside large protection ar- negligible (78 − 79 %). eas (Woodroffe & Ginsberg 1998). Animal species are further threatened with extinc- tion if wildlife harvest is taken into account (Redford 1992; Robinson et al. 1999; Cullen 124 Chapter 8

Table 8.6: Regeneration and secondary succession in abandoned lands.ANOVA for the main effects of clearing fraction (CF =0− 90 %) and shape of the fragmentation pattern (random, three types of clustering) on the following monitored quantities: (A) total above-ground biomass, (B) fraction of early successional species on biomass, and (C) fraction of late successional species on biomass.Simulation time was 300 y, simulated area was 1 km 2.Sample size n = 1 and no cross-correlation.

Effects SS df MS F ratio P

A) Biomass

Clearing 211957.38 9 23550.82 474.47 <0.001 Pattern 705.84 3 235.28 4.74 >0.01 Error 1340.18 27 49.64

B) Early successional species

Clearing 4666.19 9 518.47 55.86 <0.001 Pattern 74.49 3 24.83 2.68 >0.5 Error 250.62 27 9.28

C) Late successional species

Clearing 6490.55 9 721.17 211.60 <0.001 Pattern 48.57 3 16.19 4.75 >0.01 Error 92.02 27 3.41

et al. 2000). All these effects are obviously (9–64 ha); (iii): rain forest with a species too complex to be taken into account ex- composition of an undisturbed forest (81– plicitely into our model. However, the ap- 100 ha). A size of at least 80 ha is necessary proach of the reproductive success in the re- for maintaining a total biomass and species generation rates as used in the model is well composition, which is similar to a primary established and was used before (Ribbens forest. It might be that even in these large et al. 1994; Chave 1999b). scale areas a shift in the species composi- tion within the different species groups oc- curs. Turner & Corlett (1996) found in a The multiple consequences of rain forest fragment of 100 ha in Southeast- forest deforestation Asia no change in standing biomass over 50 years, but strong shifts in the species com- The simulation of forest fragments of differ- position. ent sizes showed three different directions in So far, little modelling effort has been which forest remnants might develop (ex- focused towards attempting to understand amples for size were taken from results of the interplay between fragmentation pro- scenario fragmented 3): (i) collaps of the cesses and natural regeneration of the for- forest structure (1 ha); (ii): a forest with est. A probable explanation for this is that a high fraction of early successional species Regeneration of fragmented rain forest in French Guiana 125

] random clustered 1 clustered 2 clustered 2 -1 500

. .

. . ha

.

.

.

.

. .

.

.

.

.

400 .

.

.

.

. odm

. . .

.

. .

.

.

.

.

.

. 300 .

.

.

.

.

.

.

. total 200 . 100 0 Biomass [Mg 80 60 . early ss

.

.

.

. 40 .

.

.

.

.

.

.

.

.

. .

.

.

.

.

. . 20 .

.

.

.

.

.

. . .

.

.

. . . 0 . . . . 80 Biomass [%]

. .

. .

.

.

.

60 .

.

.

.

.

.

.

.

. .

.

. . late ss

.

.

.

.

.

.

.

.

.

40 .

.

.

.

.

.

.

.

.

.

20 . 0 0306090030609003060900306090 Clearing fraction CF [%]

Figure 8.6: Patterns with a fraction of the landscape being regrowed with secondary succession after abandoned of cleared ares (same scenarios as Fig.8.5but t = 301 − 600 y).Above-ground biomass as function of cleared fraction (CF) and pattern formation (random, clustered 1 − 3). Bars show time averages (including SD), crosses correspond to values at t = 600 y.Total biomass (1st row), and fraction of early- and late-successional species (2nd and 3rd row).

the problem of fragmentation is spatially ex- der area seems to overestimate the effects plicit. Laurance et al. (1998) estimated with largely. the model of Laurance & Yensen (1991) Therefore, it requires forest growth simu- when forest fragments were dominated by lators able to describe the spatial structure edge effects. According to their analyses of the landscape. Such quantitative models strong and moderate effects should rise once have only recently become available (Bossel the fragment size falls below 90–500 ha. Af- & Krieger 1994, Pacala et al. 1996, K¨ohler ter our simulations strong effects on stand- & Huth 1998a, Chave 1999b). ing biomass and species composition are only seen at the lower end of this range. The structure of the forest matrix has Ferreira & Laurance (1997) stated that even been repeatedly shown to be an important in forest fragments of 1000 ha a substantial indicator of the faunal diversity (Gascon impact of fragmentation will be expected, et al. 1999; Hanski & Ovaskainen 2000; Jor- because 22 − 42 % of the border area was dan 2000). Indeed, the three-dimensional influenced by edges. With our simulations spatial heterogeneity of forest communities the structural characteristics of the forest is the main mechanism which allows for the are not expected to change substantial for observed diversity of heterotrophic species. that large areas. The estimate of impacts The species composition is strongly affected of fragmentation over the ratio of the bor- by the invasion of ecotonal species, which 126 Chapter 8

random clustered 1 clustered 2 clustered 3

400 300 200 CF=10% 100 ]

-1 0 ha 400 odm 300 200 CF=50% 100 0 Biomass [Mg 400 300 200 CF=90% 100 0 0 200 400 0 200 400 0 200 400 0 200 400 600 Time [y]

Figure 8.7: Time variation of the above-ground biomass for selected fragmentation scenarios (random, clustered 1−3) and cleared fraction (CF =0−90 %).Total biomass (solid bold line), biomass for early successional species (solid line), for mid-successional species (broken dotted line), and for late successional species (broken line).Savanna species are always below 1 Mg ha −1 and are not shown.Land-uses were abandoned at year 300.The biomass was computed only within the forest plots from t =0ytot = 300 y, and over the whole landscape after t = 300 y, whence the apparent discontinuity in the biomass curve.

might outcompete the species of the closed Chave et al. 2000c). At the Piste de rain forest. Our simulations show that Saint Elie 5 ha dataset, Brown’s (1997) the indicators of forest structure such as allometric equation gave an estimate of the within-plot maximal canopy height are 210.7 Mg ha−1, Higuchi-Santos’s (1998) indeed strongly affected by fragmentation. equation yielded a high 358.9 Mg ha−1,and The underlying mechanism in our simula- the fit proposed in Chave et al. (2000) gave tions is the increased mortality rate for large edge trees. Since the ecotone area increases Figure 8.8: (opposite page): Spatial distri- quickly with fragmented fraction, the strong bution of dominant trees as function of clear- positive correlation between canopy height ing fraction (CF) and of spatial clustering − and fragmented fraction is unsurprising. (random, clustered 1 3) of selected scenar- ios for different times.Top: time t = 300 y, Above-ground biomass estimates for just before meadows are abandoned; bottom: neotropical forests still rely upon scarce t = 600 y.Simulation area was 1 km 2, each data, and the models used to relate the pixel corresponds to one patch (20 m × 20 m). dbh with the tree biomass (e.g. Brown Pixels inform about size and successional group 1997; Higuchi et al. 1998) have a lim- of the dominat tree of the patches according to ited predictive power (Brown et al. 1995; legend. Regeneration of fragmented rain forest in French Guiana 127

t = 300 y random clustered 1 clustered 2 clustered 3 1.0

0.5 CF=10%

0.0

0.5 CF=50%

y direction [km] 0.0

0.5 CF=90%

0.0 t =600y random clustered 1 clustered 2 clustered 3 1.0

0.5 CF=10%

0.0

0.5 CF=50%

y direction [km] 0.0

0.5 CF=90%

0.0 0.0 0.5 0.0 0.5 0.0 0.5 0.0 0.5 1.0 x direction [km]

undefined early successional sp. mid successional sp. late successional sp. H<=10m H = 10-30m H = 10-30m H = 10-30m H > 30m H > 30m H > 30m Figure 8.8: Captions are found on opposite page. 128 Chapter 8

245.2 ± 30 Mg ha−1. Aninsitubiomass fied to the species level). experiment (0.25 ha plot where all trees were felled and weighed) predicted values above 400 Mg ha−1 more consistent with Which strategy is best? Higuchi-Santos’s predictions, but probably skewed by the choice for the experiment Our results show that the extent of forest plot. Therefore, although it is quite diffi- boundaries should be minimized within a cult to assess the quality of the values pre- fragmented landscape. However, scenarios dicted by our model (445 ± 15 Mg ha−1 which minimize the edge-to-area ratio (clus- in the control run), it is probable that the tered scenarios) also lead to a longer regen- real number is smaller. We have found eration. that after 300 years the regenerating forest In the random model of fragmentation, reached significantly smaller figures than in the forest structure is only weakly affected the control run (59 − 68 %). This suggests by the cleared fraction up to about CF = a long-lasting effect of disturbances upon 50 %. Below this level, clear-cut plots are the carbon pool of forests, as observed in surrounded by an intact matrix which pro- field experiments (Saldarriaga et al. 1988). vides favorable microenvironmental condi- Another explanation is that natural distur- tions and seeds as for treefall gaps. How- bances operating over large areas (such as ever, above this level most of the fragments large treefalls) maintain the system below are isolated and seed dispersal becomes the its expected biomass capacity. major issue for the regeneration. A possi- Plant functional diversity was strongly ble strategy would be to preserve large ma- affected by fragmentation in all the sim- ture trees in a sufficient density to favour a ulations. The fraction of early succes- subsequent regeneration of the forest. This sional species which were maintained was strategy is however quite difficult to apply significantly larger over the whole land- in practice because large trees are usually scape, which had to be expected because of valuable timber species. the increased area available for this PFT. However, the increase was also significant Limitations of the present mod- within the forest fragments (up to 30 % for CF = 90 %). Regeneration after land elling approach use change showed that the forest recovered Our study provides a first attempt to as- only after 100 years. In clustered fragmen- sess the long-term impact of fragmenta- tation scenarios, this transient regime was tion on tropical rain forest, and our results longer (> 200 y), which is a consequence are consistent with available field studies in of seed-dispersal limitation for late succes- South America. We believe that the mod- sional species. A recent study has found elling approach is an efficient tool to address that species composition took even longer landscape-scale issues. Moreover, it makes to recover (Ferreira & Prance 1999). Undis- it possible to relate our findings with satel- turbed forest had 147 species per ha while lite imagery information. the secondary forest had a mere 89 species per ha, but the estimated biomass was com- For example, data at 1 km2 spatial res- parable. This picture constrasts with that olution are available, with a percentage of Saldarriaga et al. (1988), in part because of forest cover from 0 % to 100 % (De- the disturbance type were clearly different Fries et al. 2000). Using this dataset, it in the two forests, but also because of an is easy to compute the fraction of rain for- unsufficient taxonomic effort in the latter est edge pixels in South America (1 km2 survey (only 75 % of the trees were identi- plots of rain forest at the edge of the population). Overall, these sites represent Regeneration of fragmented rain forest in French Guiana 129

0.60 million km2, out of 8.37 million km2 Mart.) in Brazil, which impedes the succes- of rain forest (7.17 %). Of this edge area, sion. 2 0.219 million km (37 %) is a boundary be- Our analysis was restricted to forest sizes tween forest and agriculture/grasslands. up to 100 ha. Recent studies identify indi- We may want to address the generality rect edge effects of fragmentation on larger of these patterns across forest types. Our scales up to several thousands of hectars model was validated for two other forests, in (Curran & Leighton 2000). Fauna-flora in- Venezuela and in Sabah, Malaysia. Never- teractions on recruitment in terms of seed theless, we think that the model application dispersal and predation might be far more to the French Guiana rain forest was more important for the long-term forest dynamics appropriate than to Venezuela or Malaysia. as assumed today. Animals migrate to the Most field data on fragmentation were col- remnant forests, if land clearing destroys lected in different research plots in South their old habitats. Thus, even large for- America. Moreover, the current knowledge est fragments will be affected by anthopo- on forest growth and published data was geneous impacts. A size above which frag- wider for the French Guiana site than for mentation effects are negliglible might not that of Venezuela. Furthermore, available exist on scales still found in tropical rain satellite imagery data sets already men- forests. tioned allow us to validate our approach for forests in French Guiana (future project). However, even if this case study was re- Acknowledgements stricted to a rain foret in French Guiana, we think that the general pattern of how We thank Paul Moorecroft for giving us ref. fragmentation will influence forest dynam- Moorecroft et al. (2000) prior to publica- ics will be similar in different sites. tion. J. Chave was supported by grants to We have investigated scenarios assum- S.A. Levin, from the Andrew W. Mellon ing that the major mechanisms were the Foundation and from the David and Lu- increased mortality of trees near edges cille Packard Foundation (grant 99-8307). (≤ 100 m) and a reduced recruitment P. K¨ohler was funded by the Otto-Braun- rate. Although these assumptions are real- Foundation at the University of Kassel. istic they compound several distinct mech- H. Bossel provided helpful comments to the anisms, such as, for the mortality gradi- manuscript. ent, microclimate changes (higher tempera- ture and lower moisture), higher probablity of tree uprooting, and explicit competition with ecotone species. It would be difficult, yet valuable, to construct a model which would take these as separate mechanisms. A similar comment should be made for land-used areas. The soil properties are usually radically modified by agricultural activities, with a rapid loss of nutrients (Sal- darriaga 1986; Uhl et al. 1988; Buschbacher et al. 1988; Mackensen & F¨olster 2000), that might considerably slow down the forest re- generation pace. Also, possibly invasive species can take over abandoned pastures, as for the palm baba¸cu (Orbignya phalerata 130 Chapter 8 Appendix

Table 8.7: Short description of parameters including functional relationships (modified from K¨ohler et al.2000c).

Parameter Description

Environmental parameters k Light extinction coefficient I0 Light intensity above canopy SD Day length

Establishment parameters DS Initial diameter of seedlings ISs Minimal light intensity for germination NSs Ingrowth rate of seeds into seed pool NRs Seed dispersal rate of mother trees XRs Average seed dispersal distance DRh Minimal diameter of mother trees

Mortality parameters MBs,h Basic mortality rate MSs Mortality rate of seeds MDj Size dependent mortality rate (MD = MD0 − MD0/MD1 · d) W Probability of a dying tree to fall

Tree physiognomic parameters HM Maximum height cp Crown length fraction τj Site dependent fraction of stemwood biomass to total aboveground biomass (τ = τ1 + τ2 · h(d = 120cm)) h0 and h1 Height = f(diameter) (h = d/(1/h0 + d/h1)) γ2 γj Form factor = f(diameter) (γ = γ0 · exp(γ1 · d )) f2 fj Crown diameter = f(diameter) (dc =(f0 + f1 · d ) · d) 2 3 lj Leaf area = f(diameter) (l = l1 · d + l2 · d + l3 · d ) LAIM Maximal leaf area index of single tree

Biomass production parameters PM, α Photosynthetic capacity and efficiency in light response curve (Pi(Ii)= αs·Ii αs ) 1+ Ii PMs ρ Stem wood density r1l Maintenance respiration = f(biomass) (Rm(Bi)=r1l · Bi) RG Growth respiration as part of biomass m Leaf transmittance g Conversation factor gCO2 to godm Regeneration of fragmented rain forest in French Guiana 131

Table 8.8: Parametrisation for French Guiana.Short description of parameters in Table 8.7. Parameters with subindex vary with successional status (s), potential height (h) (corresponding to SS and HG in Table 8.2, respectively), or different functional coefficients (j).

Name Special Units Values

Environmental parameters k [-] 0.7 −2 −1 a I0y wet dry [µmol(p) m s ] 642.0 694.0 SDy wet dry [h] 12.0 8.0 SSy wet dry [-] 0.75 0.25 Establishment parameters DS [m] 0.01 ISs s=0-3 [fraction of I0y] 0.2 0.1 0.04 0.01 −1 −1 NSs s=0-3 [ha y )] 25 125 500 100 −1 −1 NRs s=0-3 [ha y )] 50 100 15 10 XRs s=0-3 [m] 100 50 40 20 DRh h=1-5 [m] 0.028 0.103 0.20 0.35 0.56 Mortality parameters −1 MBs,h s=0; h=1-5 [y ] 0.18 0.16 0.12 0.10 0.00 −1 MBs,h s=1; h=1-5 [y ] 0.16 0.12 0.10 0.08 0.06 −1 MBs,h s=2; h=1-5 [y ] 0.07 0.06 0.05 0.04 0.03 −1 MBs,h s=3; h=1-5 [y ] 0.06 0.05 0.04 0.03 0.02 −1 MSs s=0-3 [y ] 0.01 0.1 0.5 1.0 −1 MDj j=0-1 [y , cm] 0.2 0.1 W [-] 0.40 Tree physignomic parameters HMs,h h=1-5 [m] 5.0 15.0 25.0 36.0 40.0 cp [-] 0.358 τ [-] 0.7 −1 h0 [cm m ] 1.96 −1 h1 [m ] 49.0 −1 γj j=0-2 [-, cm , -] 2.575 -1.409 0.0358 −1 −2 fj j=0-2 [m cm ,mcm , -] 0.132 0.933 -0.6615 −1 −2 −3 lj j=1-3 [m cm ,mcm ,mcm ] 3.197 0.0684 -0.000379 LAIM [-] 2 Biomass production parameters −2 −1 a PMs s=1-4 [µmol(c) m s ] 27.7 27.7 11.3 6.8 −1 a αs s=1-4 [µmol(c) µmol(p) ] 0.043 0.043 0.043 0.043 −3 ρs s=1-4 [todm m ] 0.83 0.62 0.75 0.81 r1s s=1-4 [-] 0.08 0.08 0.04 0.03 RG [-] 0.25 m [-] 0.1 −1 g [godm gCO2] 0.63

a p: photons; c: CO2

Summary

Background 3. Which logging method and rotation length can be called sustainable de- Deforestation and degradation of tropical pending on the forest site? rain forests threaten these ecosystems all 4. How does recruitment determine forest over the world. Rain forests, which were growth and yield? heavily disturbed through timber extraction or fragmentation, are endangered further if 5. Can tropical rain forests buffer the ef- exposed to hurricanes or forest fires. fects of ongoing fragmentation? Currently, these human impacts on trop- ical forests are unsustainable and will cer- tainly continue for the near future. Field Modelling framework studies in various research activities try to analyse short term impacts on forests, but The process-based and individual-oriented cannot address questions of forest develop- forest growth model Formind2.0 was de- ment and the threat of species loss in the veloped to answer these questions. It long term. simulates the spatio-temporal dynamics of uneven-aged mixed forest stands in areas of one hectare to several km2. The model de- Research goals and objec- scribes forest dynamics as a mosaic of in- teracting forest patches of 20 m2×20 m2 in tives size. Within these patches trees are not spatially-explicit distributed, and thus all The research objective was to analyse the compete for light and space following the effects of timber logging and forest fragmen- gap model approach. A concept for ag- tation as two important anthropogenic im- gregating tree species diversity into 10-20 pacts on growth and yield of tropical rain plant functional types (PFT) on the basis forests, using a newly developed process- of species maximum tree height and suc- based forest growth model. Current prac- cessional status was developed and applied tices were analysed for their sustainability, to all study sites. The carbon balance of and suggestions concerning possible man- each individual tree including photosynthe- agement improvement were stated, where sis and respiration is modelled explicitly. possible. The following research questions Thus, we can model measured diameter in- were defined: crement for different PFT, size and light conditions accurately. Allometric relation- 1. Is there a general approach for clas- ships connect above-ground biomass, stem sifing several hundred tree species in diameter, tree height and crown dimensions. different rain forest sites into a few Besides increasing mortality through self- groups? thinning in dense patches, one of the main 2. Does simulated tree growth match processes of mortality is gap creation by the measured data sets with acceptable ac- falling of large trees. This process, as well as curacy? seed dispersal from mature trees, interlinks 134 Summary neighbouring patches with each other. The rainy season (10 ◦C during the dry sea- modelling approach was discussed in detail son). A species list covering 1022 tree and compared with several others. Com- and shrub species was aggregated into parison of modelled and measured growth 19 PFTs. Inventories of five hectares of data were used to validate model perfor- mature forest was used for simulations. mance and accuracy.

Growth & yield of forest Study sites in Venezuela One of the key aspects was model applica- tions to several different rain forest areas. Stability and sensitivity analysis of the Thus, data from three study sites were used model behaviour was tested in detail for the for further analysis. They are: application in Venezuela. Mortality rates were the most crucial for model dynamics. The stability of primary forest was anal- • Capara Forest Reserve (Venezuela, ysed. Most result variables were stable. 7◦30’N, 70◦45’W; elevation 100 m). Only biomass of small growing tree species The mean annual rainfall is 1750 mm, showed higher fluctuations. with a pronounced dry season from December to March (monthly precip- Typical logging practices in Venezuela itation < 50 mm). The average an- were analysed for their sustainability and nual temperature is 24.6◦C. 127 tree impact on further forest dynamics. Thus, and palm species were grouped into 12 logging methods (conventional, or reduced- PFTs. Data of a mature forest and a impact logging), length of cutting cycles forest logged five year prior to enumer- (30, 40, or 60 years), and logging intensities ation (each 1 ha in size) were used for in terms of extracted net bole volume (30, 3 −1 −1 model applications. 45, 60 m ha cycles ), under two differ- ent minimum felling diameters (35, 50 cm) • Deramakot Forest Reserve, (Sabah, were analysed. Malaysia, 117◦30’E, 5◦25’N, 130-300 m ◦ Conventional logging methods did not asl.). Mean annual temperature is 27 provide sustainable timber yields under with little seasonal variations. Aver- most logging concepts. Only with the age annual precipitation is about 3500 longest (60 years) cutting cycles, standing mm, with no pronounced dry season. A bole volume recovers similarly to mature tree species list of 468 species was ag- unlogged stands. However, species compo- gregated into 13 PFTs. Inventory data sition varies significantly from mature for- for a mature and a recently logged-over est for all logging impacts. Abundance of forest (each 0.81 ha in size) were taken fast growing early successional species in- for model initialisation. creased up to 29 % (mature forest: 1 %), • Piste de Saint-Elie (French Guiana, while late successional species declined re- 5◦30’N, 53◦00’W; elevation < 500 m). markably (down to 14 % compared to 44 %). Annual rainfall is slightly above 3 000 Scenarios with reduced-impact logging mm y−1. A pronounced dry season of provided a significantly higher timber vol- 2 months is recorded from September ume than under conventional logging. If to November and a short dry season long logging cycles were provided they could in March. The average temperature is be called sustainable with respect to achiev- 25.8 ◦C with an annual amplitude of able timber yields, but impact on species 2 ◦C and daily amplitudes of 7 ◦Cinthe composition was still large. Summary 135 Forest growth in Sabah Forest fragmentation in French Guiana In ongoing forest fragmentation seed dis- persal will be one of the most important The number of field studies in forest frag- factors determining forest dynamics. Thus, ments has increased greatly in the last model assumptions on recruitment were en- years. It has been shown that mortality hanced. Five alternative recruitment sce- rates are increased at forest edges, and that narios, which cover the range of possible recruitment rates in fragments will differ fragmentation effects from single island to a from the ones in undisturbed forests. closed forest, were analysed. Additionally, These findings were incorporated in four impacts of boundary effects and fragment different scenarios to analyse the effects of size on forest dynamics were investigated. fragmentation and future forest dynamics. Sensitivity analyses of all parameters of The size of fragmented forest was a second recruitment processes show that seed pro- variable. Biomass was reduced to 7 % of the duction rates influenced forest dynamics mature forest level after 300 y in the most and species composition the most. Others, extreme scenario (forest fragment of one like average seed dispersal distances, were hectare, alteration of mortality and recruit- of minor importance. ment rate through fragmentation). Species composition was shifted towards light de- To isolate the effects of recruitment from manding early successional species. other parts of the model, scenarios without any recruits were computed, showing the In a second study, forest clearing and re- buffer capacity of standing forests. Only growth of secondary vegetation in a land- after 50 simulated years did the stand- scape context was analysed. Clearings were ing biomass begin to decline. This shows clustered at different degrees, and the frac- the importance of field studies on ingrowth tion of cleared area varied between 0 % (no rates of saplings at about 1 cm in diameter. clearing) and 90 %. Stable and unchanged forest structures in large forest islands were The impacts of logging (reduced-impact easy to identify. However, in the process and conventional methods, cycle length of of secondary succession distribution of seed- 20-80 years) as one main disturbance in producing mother trees was important for Sabah were analysed. In fragmented forests the succession process. Thus, in landscapes these impacts led to immediate shifts in with highly clustered forest fragments, sec- species composition and species loss. Yields ondary succession of abandoned areas took did not depend on recruitment assump- longest. The danger of species loss was high. tions, and thus fragmentation status. But Immigration was not considered and species the fraction of late successional species in composition and recruits depend solely on logged timber was declining in forest frag- local abundance. ments. Timber was highest with reduced- impact methods and medium cutting cycles of 40 years. In short cycles the timber gain depend strongly on logging methods. Conclusions Further impacts of logging on forest floor and nutrient content were omitted, and thus The following conclusions are drawn in re- the results were optimistic. But with re- lation to the five research questions stated spect to current logging cycles of 10-40 years in Chapter 1: and conventional methods, the simulations demenstrate the unsustainability of com- Tree species grouping: mon practices. A general approach for the aggregation 136 Summary

of several hundred tree species into a forest fragments (∼ 100 ha) stand- few (10-20) plant functional types was ing biomass was reduced compared to developed. Grouping was based on the closed forest structures. Thus, the po- two independent criteria successional tential of tropical forests for carbon se- status and maximum tree height. Ap- questration declined as well. plication of the approach was discussed in detail for the rain forest in Sabah, and was applied additionally to the two Recommendations con- other sites (Venezuela, French Guiana). cerning the modelling Simulated vs. measured tree growth: approach Simulated tree growth was compared with measurements from permanent For further applications, Formind2.0 sampling plot data in Sabah (25 ha, ob- couldbeimprovedandextendedinseveral servation period 9-20 y). The current ways: application was limited to good site conditions. For single species groups • stochasticity in mortality influenced re- In the context of climate change re- sults widely. The total basal area and search, tropical rain forests act as stem number agreed with good accu- global carbon sink of different poten- racy. tials depending on anthropogenic im- pacts. Analysing these potentials is im- Sustainability of logging methods: portant for questions concerning car- Dependent on forest site, criteria for bon sequestration and might be possi- sustainability changed. However, cur- ble with a simulation study using For- rent logging practices in Sabah and mind. Venezuela overuse forests to a great de- • Various site dependent factors (slope, gree. This is all the more important soil) might be incorporated, as done in as results were based on optimistic as- Formix3-Q (Ditzer 1999). sumptions concerning recruitment ca- pabilities. Reduced-impact logging • Testing of the current version with methods produced higher timber yields spatially-explicit forest inventory data than conventional methods. Yield was may suggest improvements for model- also higher in longer logging cycles. ing the light competition process.

Assumptions on recruitment: • The coupling of recruitment with a A detailed description of recruitment model for animal diversity (Reinhard processes was necessary in fragmented 1999) would be a first step to model or highly disturbed forests. Species seed dispersal and predation more ac- composition depended on a detailed curately. description of recruitment, while total biomass did not. • Modelling of individual tree species might be of interest for further model Forest fragmentation: applications. Using the current ap- The species composition in fragmented proach of plant functional types to forests was changed. If regrowth of for- derive parameter sets, but relating mer human-use areas was a possibility each individual tree to one specific for forest conservation, species loss was species would improve complexity sig- limited, but depended on the length nificantly. In particular, recruitment of the human impacts. Even in large processes would depend on a single Summary 137 species and thus analysis of species composition would be possible. This might contribute to answering the question, of what is the key process in determining high species diversity in tropical rain forests (Cook 1998; Hubbell et al. 1999).

Zusammenfassung

Hintergrund mehrere hundert Baumarten an un- terschiedlichen Standorten in wenige Entwaldung und Degradierung bestehen- Gruppen zusammenzufassen? der tropischer Regenw¨alder bedrohen die- 2. Stimmen simuliertes und gemessenes se Okosystem¨ weltweit. Die durch Baumwachstum mit akzeptabler Ge- Holzentnahme oder Fragmentierung stark nauigkeit uberein?¨ gest¨orten Regenw¨alder sind noch tiefgrei- fender gef¨ahrdet, wenn Wirbelsturme¨ oder 3. Welche Managementmethoden und Waldbr¨ande uber¨ sie herziehen. Nutzungszyklen k¨onnen, abh¨angig Momentan sind menschliche Eingriffe in vom jeweiligen Standort, nachhaltig tropische W¨alder weit entfernt von einem genannt werden? nachhaltigem Konzept, und werden es si- 4. Inwieweit bestimmt die Natur- cher auch in der nahen Zukunft sein. Feld- verjungung Waldwachstum und untersuchungen in unterschiedlichen For- ¨ Erntekapazit¨at? schungsaktivit¨aten versuchen die kurzfristi- gen St¨orungen auf die W¨alder zu untersu- 5. Kann tropischer Regenwald die Aus- chen. Es wird jedoch nicht m¨oglich sein, wirkungen der voranschreitenden Fragen der langfristigen Waldentwicklung Waldfragmentierung abpuffern? und die Gefahr des Aussterbens etlicher Spezies zu beantworten. Modellierungskonzept

Forschungsziel und Frage- Das prozess-basierte und individuen- stellungen orientierte Waldwachstumsmodell For- mind2.0 wurde entwickelt, um diese Fragen Ziel der Forschungsarbeit war es, die zu beantworten. Es simuliert die raum- Auswirkungen der Holzentnahme und der zeitliche Dynamik eines ungleichaltrigen Waldfragmentierung - zwei Beispiele an- gemischten Waldbestandes auf Fl¨achen thropogener Einflusse¨ auf die Wuchsdyna- zwischen einem Hektar und mehreren mik tropischer Regenw¨alder - mit Hilfe eines Quadratkilometern. Das Modell beschreibt neu entwickelten prozess-basierten Wald- die Walddynamik als ein Mosaik interagie- 2× 2 wachstumsmodells zu analysieren. Momen- render Waldfl¨achen von 20 m 20 m .Die tane Nutzungspraktiken wurden auf ihre B¨aume sind innerhalb dieser Teilfl¨achen Nachhaltigkeit hin untersucht, und Vor- nicht r¨aumlich-explizit verteilt und kon- schl¨age zu einer m¨oglichen Verbesserung kurrieren somit nach dem Ansatz fur¨ des heutigen Managements gemacht, so- Gap-Modelle allesamt um Licht und Raum. weit m¨oglich. Die folgenden Fragestellun- Ein Konzept zur Aggregation der gen wurden definiert: Baumartenvielfalt in 10-20 pflanzenfunk- tionale Typen (PFT) wurde entwickelt. Es 1. Gibt es einen generellen Ansatz, um basiert auf maximaler Baumh¨ohe und dem 140 Zusammenfassung

Sukzessionsstatus einzelner Baumarten des funf¨ Jahre nach einem Holzernte- und wurde in allen Untersuchungsgebieten eingriff (jeweils 1 ha groß) wurden fur¨ angewendet. die Modellanwendung benutzt.

Die Kohlenstoffbilanz jedes Einzelbau- • Deramakot Forest Reserve, (Sabah, mes, einschließlich Photosynthese und Malaysia, 117◦30’ E, 5◦25’ N, 130- Respiration, wird explizit modelliert. 300 m uber¨ NN). Die mittlere j¨ahrliche Somit kann der Durchmesserzuwachs in Temperatur betr¨agt 27◦ mit geringen Abh¨angigkeit des PFT, der Baumgr¨oße saisonalen Unterschieden. Mittlerer und der Lichtbedingungen genau modelliert j¨ahrlicher Niederschlag ohne nennens- werden. Allometrische Zusammenh¨ange werte Trockenperiode bel¨auft sich auf verknupfen¨ oberirdische Biomasse, Stamm- 3500 mm. Eine Baumartenliste mit 468 durchmesser, Baumh¨ohe und Kronendimen- Arten wurde in 13 PFT zusammenge- sionen. Neben einer erh¨ohten Mortalit¨at fasst. Inventurdaten eines ausgewach- durch Ausdunnungsprozesse¨ in dichtstehen- senen und eines kurz zuvor genutzten den Fl¨achen ist die Luckenbildung¨ durch Bestandes (jeweils 0.81 ha) wurden fur¨ umfallende große B¨aume ein wichtiger die Modellinitialisierung verwendet. Mortalit¨atsprozess. Diese Luckenbildung¨ sowie Samenverbreitung ausgehend von • Piste de Saint-Elie (Franz¨osisch Gua- Mutterb¨aumen verknupfen¨ benachbarte yana, 5◦30’N, 53◦00’W; Erhebungen < Teilfl¨achen miteinander. 500 m). Der j¨ahrliche Niederschlag Der Modellierungsansatz wurde detai- liegt etwas uber¨ 3000 mm. Eine aus- liert diskutiert und mit verschiedenen an- gepr¨agte zweimonatige Trockenperiode deren verglichen. Mittels eines Vergleichs wird zwischen September und Novem- von modellierten und gemessenen Wuchsda- ber registriert sowie eine weitere kurze im M¨arz. Die durchschnittliche Tem- ten wurden Modellverhalten und Genauig- ◦ keit validiert. peratur liegt bei 25.8 Cmitj¨ahrlichen Amplituden von 2 ◦C und t¨aglichen Amplituden von 7 ◦CinderRegen- ◦ Untersuchungsgebiete zeit (10 C in der trockenen Jahreszeit). Eine Artenliste mit 1022 Baum- und Straucharten wurde aggregiert zu 19 Eine der Grundideen dieser Arbeit war die PFT. Inventuren von funf¨ Hektar aus- Modellanwendung in unterschiedlichen Re- gewachsenem Regenwald wurden fur¨ genwaldgebieten. Es wurden Daten aus die durchgefuhrten¨ Simulationen ver- drei Untersuchungsgebieten fur¨ die weite- wendet. ren Analysen verwendet. Im einzelnen sind dies: Wuchs und Ernte von ve- • Capara Forest Reserve (Venezuela, 7◦30’N, 70◦45’W; Erhebung 100 m). nezuelanischem Wald Der durchschnittliche j¨ahrliche Nieder- schlag betr¨agt 1750 mm, mit einer aus- Das Modellverhalten wurde in der An- gepr¨agten Trockenperiode von Dezem- wendung in Venezuela detailliert in Stabi- ber bis M¨arz (monatlicher Niederschlag lit¨ats- und Sensitivit¨atsuntersuchungen ge- < 50 mm). Die durchschnittliche mitt- testet. Die Modelldynamik wurde am lere Jahrestemperatur ist 24.6◦C. 127 meisten durch die Mortalit¨atsraten beein- Baum- und Palmenarten wurden in 12 flusst. Die Stabilit¨at eines Prim¨arwaldes PFT zusammengefasst. Daten eines wurde analysiert, wobei die meisten Ergeb- ausgewachsenen Waldes und eines Wal- nisgr¨oßen stabiles Verhalten zeigten. Nur Zusammenfassung 141 die Biomasse kleinwuchsiger¨ Baumarten un- zur Naturverjungung¨ erweitert. Funf¨ alter- terlag hohen Fluktuationen. native Verjungungsszenarien¨ wurden analy- Typische Holzentnahmepraktiken in Ve- siert, die das gesamte Spektrum m¨oglicher nezuela wurden auf ihre Nachhaltigkeit und Fragmentierung abdecken, von einzelnen ihre Auswirkungen auf die weitere Wald- Bauminseln bis hin zu einem geschlossenen entwicklung hin untersucht. Die Analy- Wald. Zus¨atzlich wurden die Auswirkun- se erstreckte sich auf Erntemethoden (kon- gen von Randbedingungen und Fragmentie- ventionelle oder schadensreduzierte Nut- rungsgr¨oße auf die Walddynamik hin unter- zung), L¨ange der Nutzungszyklen (30, 40, sucht. oder 60 Jahre), Nutzungsintensit¨aten in Eine Sensitivit¨atsanalyse aller Parame- Form von entnommenem Nettostammvolu- ter der Verjungungsprozesse¨ verdeutlichte, men (30, 45, 60 m3 ha−1 Zyklus−1) und un- dass Samenproduktionsraten die Waldent- terschiedlichen minimalen Entnahmegr¨oßen wicklung und Artenzusammensetzung am (Durchmesser=35, 50 cm). meisten beeinflussen. Andere Parameter, Konventionelle Nutzungsmethoden erga- wie z.B. durchschnittliche Distanzen der Sa- ben keinen nachhaltigen Ernteertrag un- menverbreitung, waren von geringerer Be- ter den meisten Nutzungskonzepten. Nur deutung. bei den l¨angsten (60 Jahre) Nutzungszy- Um die Auswirkungen der Verjungung¨ klen erholt sich das stehende Stammvo- von den anderen Modellteilen zu trennen, lumen zu einem Niveau vergleichbar mit wurden Szenarien ohne jeglichen Jungwuchs ungenutzten Best¨anden. Doch auch dort durchgefuhrt,¨ die die Pufferkapazit¨at des wie bei allen anderen Nutzungen ver¨andert bestehenden Waldes verdeutlichen. Erst sich die Artenzusammensetzung signifikant nach 50 simulierten Jahren gab es einen im Vergleich zum ausgewachsenen Wald. Ruckgang¨ in der stehenden Biomasse. Die Die Abundanz der schnell wachsenden Bedeutung von Feldstudien zu Einwuchs- Fruhsukzessionsarten¨ stieg auf bis zu 29 % raten von kleinen B¨aumen von etwa 1 cm an (ausgewachsener Wald: 1 %), w¨ahrend Stammdurchmesser wird hierdurch unter- das Vorkommen der Sp¨atsukzessionsarten strichen. merklich zuruckging¨ (auf 14 % im Vergleich Die Auswirkungen von Holznutzung zu 44 % im ausgewachsenen Wald). (schadensreduzierte und konventionelle Me- Szenarios mit schadensreduzierter Nut- thoden, Zyklusl¨angen von 20 bis 80 Jah- zung ergaben signifikant h¨ohere Volumener- ren) als eine der bedeutensten Eingriffe tr¨age als konventionelle Verfahren. Unter in Sabah’s Regenw¨aldern wurden analy- langen Nutzungszyklen k¨onnen sie im Be- siert. In fragmentierten W¨aldern fuhrten¨ zug auf erreichbare Ernteertr¨age hin nach- diese Eingriffe zu sofortigen Verschiebun- haltig genannt werden. Doch auch hier sind gen in der Artenzusammensetzung und zu die St¨orungen der Artenzusammensetzung einem Verlust etlicher Baumarten. Ern- noch immer groß. teertr¨age hingen nicht von den Annahmen zur Verjungung¨ und somit dem Fragmen- tierungsgrad ab. Die Ertr¨age waren am h¨ochsten mit schadensreduzierten Metho- Waldwuchs in Sabah den und einem Nutzungszyklus von 40 Jah- ren. In kurzen Zyklen hing die Ernte maß- In der immer weiter fortschreitenden Wald- geblich von der gew¨ahlten Erntemethode fragmentierung wird die Samenverbreitung ab. einer der wichtigsten Faktoren fur¨ die Weitere Auswirkungen der Holz- Bestimmung zukunftiger¨ Walddynamiken nutzung auf den Waldboden und sein. Daher wurden die Modellannahmen 142 Zusammenfassung den N¨ahrstoffhaushalt wurden nicht geclusterten Landschaft am l¨angsten. Die berucksichtigt,¨ weshalb die Ergebnisse als Gefahr der Extinktion einzelner Arten war optimistisch eingesch¨atzt werden k¨onnen. hoch. Immigration wurde nicht betrachtet, Die Simulationen zeigten dennoch die feh- so dass die Artenzusammensetzung nur von lende Nachhaltigkeit heutiger Erntezyklen der lokalen Verjungung¨ abhing. von 10-40 Jahren unter konventionellen Methoden. Schlussfolgerungen

Waldfragmentierung in Die folgenden Schlusse¨ k¨onnen als Antwor- ten auf die im Kap. 1 gestellten Fragen ge- Franz¨osisch Guayana zogen werden:

Baumartengruppierung: Die Anzahl von Feldstudien zur Waldfrag- Ein genereller Ansatz zur Aggregati- mentierung hat in den vergangenen Jahren on mehrerer hundert Baumarten in we- deutlich zugenommen. Anhand dieser Stu- nige (10-20) pflanzenfunktionale Ty- dien wurde gezeigt, dass die Mortalit¨atsrate pen wurde entwickelt. Die Grup- an Waldr¨andern erh¨oht ist und dass die pierung basierte auf den beiden un- Verjungungsraten¨ in Waldfragmenten sich abh¨angigen Kriterien Sukzessionstatus von denen in ungest¨orten W¨aldern unter- und maximale Baumh¨ohe. Dieser scheiden. Ansatz wurde fur¨ einen Regenwald Diese Ergebnisse wurden in vier un- in Sabah ausfuhrlich¨ diskutiert, und terschiedlichen Szenarien integriert, um zus¨atzlich in den beiden anderen Un- die Auswirkungen der Fragmentierung auf tersuchungsfl¨achen angewendet (Vene- zukunftige¨ Walddynamiken hin zu unter- zuela, Franz¨osisch Guayana). suchen. Die Gr¨oße der Waldfragmen- te war eine zweite Variable. Die ste- Simuliertes vs. gemessenes Baumwachs- hende Biomasse wurde in dem extrem- tum: sten Szenario (Waldfragment von einem Simuliertes Baumwachstum wurde mit Hektar und einer Ver¨anderung von Mor- Messungen aus permanenten Aufnah- talit¨ats- und Verjungungsraten)¨ nach 300 mefl¨achen in Sabah verglichen (25 ha, Jahren auf 7 % eines ausgewachsenen Wal- Beobachtungszeitraum 9-20 Jahre). des reduziert. Die Artenzusammenset- Die momentane Modellanwendung war zung verschob sich zugunsten lichtliebender beschr¨ankt auf Standorte mit guten Fruhsukzessionsarten.¨ Bedingungen. Fur¨ einzelne Artengrup- pen beeinflussten stochastische Prozes- In einer zweiten Studie wurden se des Mortalit¨atsmodells die Ergebnis- Waldrodung und der Wuchs einer Su- se deutlich. Die Gesamtstammgrund- kund¨arvegetation untersucht. Rodungs- fl¨ache und die Gesamtstammzahl wur- flachen wurden in unterschiedlichem Maße ¨ den jedoch mit guter Genauigkeit wie- geclustert, der Anteil der gerodeten Flache ¨ dergegeben. variierte zwischen 0 % (keine Rodung) und 90 %. Stabile und unver¨anderte Wald- Nachhaltigkeit von Holznutzungsmethoden: strukturen konnten in großen Waldinseln In Abh¨angigkeit von der Untersu- identifiziert werden. Da jedoch im Zuge chungsfl¨ache ver¨anderten sich die Kri- der Sekund¨arsukzession die Verteilung terien einer nachhaltigen Holznutzung. samenproduzierender Mutterb¨aume von Unabh¨angig davon ubernutzen¨ mo- Bedeutung war, dauerte die Sukzession mentane Praktiken in Sabah und Ve- in brachliegenden Fl¨achen der hochgradig nezuela den Wald zu großen Teilen. Zusammenfassung 143

Dies ist umso besorgniserregender, da der Bindung des atmosph¨arischen unsere Ergebnisse auf optimistischen Kohlenstoffes und mag durch Simu- Annahmen zum Verjungungspotential¨ lationsstudien mit Formind m¨oglich basierten. Schadenreduzierende Nut- sein. zungsmethoden erzielten h¨ohere Ern- • teertr¨age als konventionelle Methoden. Mehrere standortabh¨angige Faktoren Die Ernte war weiterhin in l¨angeren Zy- (Hangneigung, Fertilit¨at des Bodens) klen am gr¨oßten. k¨onnen in das Modell eingebaut wer- den, wie bereits in Formix3-Q (Ditzer Annahmen zur Verjungung:¨ 1999). In fragmentierten oder hochgradig • gest¨orten W¨aldern war eine detaillierte Ein Test der momentanen Modellver- Beschreibung der Verjungungsprozesse¨ sion mit r¨aumlich-expliziten Waldin- notwendig. Die Artenzusammen- venturdaten mag ergeben, inwieweit setzung hing maßgeblich hiervon der Mechanismus zur Beschreibung ab, w¨ahrend die Gesamtbiomasse der Lichtkonkurrenz verbessert werden unabh¨angig davon blieb. kann. • Waldfragmentierung: Eine Kopplung des Verjungungs-¨ Die Artenzusammensetzung in frag- modelles an ein Modell zur Beschrei- mentierten W¨aldern ¨anderte sich. bung der Tierartenvielfalt (Reinhard Durch den Wuchs einer Sekund¨ar- 1999) w¨are ein erster Schritt, um vegetation in ehemals genutzten Samenverbreitung und Samenverlust Fl¨achen konnte der Verlust einzelner genauer zu beschreiben. Arten eingeschr¨ankt werden. Dieser • Die Modellierung einzelner Baumar- Verlust hing jedoch von der L¨ange ten w¨are fur¨ weitere Untersuchungen der Nutzung ab. Selbst in großen von Interesse. Hierbei k¨onnten mit Waldfragmenten (∼ 100 ha) reduzierte dem momentanen Ansatz der pflanzen- sich die stehende Biomasse im Ver- funktionalen Typen die notwendigen gleich zu geschlossenen ungest¨orten Parameters¨atze erzeugt werden. Je- Waldstrukturen. Somit verringert sich dem Einzelbaum wird jedoch eine spe- auch das Potential tropischer Walder ¨ zifische Baumart zugeordnet werden. zur Bindung von Kohlenstoff. Insbesondere die Verjungungsprozesse¨ k¨onnten somit auf der Ebene einzelner Baumarten modelliert werden. Somit Erweiterungen des Model- mag ein Beitrag zur L¨osung der Frage lierungsansatzes gegeben werden, welches die entschei- denden Prozesse sind, die die hohen Ar- Fur¨ weitere Anwendung kann Formind2.0 tenvielfalt in tropischen Regenw¨aldern in unterschiedlichen Aspekten verbessert bestimmen (Cook 1998; Hubbell et al. und erweitert werden: 1999).

• Im Zusammenhang der Klimafolgenfor- schung werden tropische Regenw¨alder als globale Kohlenstoffsenken dis- kutiert. Ihr Senkenpotential h¨angt hierbei vom Grad anthropogener St¨orungen ab. Die Analyse dieses Potentials ist wichtig fur¨ die Fragen

Bibliography

Aber, J.D.,Melillo, M.J.,& Federer, C.A.1982. and how they may be interpreted? Ecological Predicting the effects of rotation length har- Modelling, 117, 361–369. vest intensity and fertilization on fiber yield from Northern hardwood forests in New Eng- Alexandrov, G. A., Yamagata, Y., & Oikawa, T. land.Forest Science, 28, 31–45. 1999b.Towards a model for projecting Net Ecosystem Production of the world forests. Acevedo, M.F.,Urban, D.L.,& Ablan, M.1995. Ecological Modelling, 123, 183–191. Transition and gap models of forest dynamics. Ecological Applications, 5, 1040–1055. Alvarez-Buylla, E.R.& Garcia-Barrios, R.1991. Seed and forest dynamics: a theoretical frame- Adlard, P.G.,Spilsbury, M.J.,& Whitmore, T.C. work and an example form the neotropics. 1989.Current thinking on modelling the trop- American Naturalist, 137, 133–154. ical moist forest. In: Mohd, W.R.W.,Chan, H.T.,& Appanah, S.(editors), Proceedings Alvarez-Buylla, E.R.& Garcia-Barrios, R.1993. of the seminar on growth and yield in tropical Models of patch dynamics in tropical forests. mixed/moist forest.Forest Research Institute Trends in Ecology and Evolution, 8, 201–204. of Malaysia, Kuala Lumpur, Malaysia, 6–20. Appanah, S., Weinland, G., Bossel, H., & Krieger, Agyeman, V. K., Swaine, M. D., & Thompson, H.1990. Are tropical rain forests non- J.1999. Responses of tropical forest tree renewable? An enquiry through modelling. seedlings to irradiance and the derivation of Journal of Tropical Forest Science, 2, 331–348. light response index.Journal of Ecology, 87, 815–827. Ashton, P.S.1978. Crown characteristics of trop- ical trees. In: Tomlinson, P.B.& Zimmer- Aide, T. M., Zimmerman, J. K., Herrera, L., mann, M.H.(editors), Tropical trees as living Rosario, M., & Serrano, M. 1995. Forest systems.Cambridge University Press, Cam- recovery in abandoned tropical pastures in bridge, U.K., 591–616. Puerto Rico.Forest Ecology and Manage- ment, 77, 77–86. Ashton, P.S.& Hall, P.1992. Comparison of structure among mixed dipterocarp forests of Aiken, S.R.& Leigh, C.H.1993. Vanish- north-western Borneo.Journal of Ecology, 80, ing Rainforests: The Ecological Transition in 459–481. Malaysia.Clarendon, Oxford. Atta-Boateng, J.& Moser, J.W.1998. A method Albers, H.J.1996. Modeling ecological con- for classifying commercial tree species of an staints on tropical forest management: Spa- uneven-aged mixed species tropical forest for tial interdependence irreversibility and uncer- growth and yield model construction.Forest tainty.Journal of Environmental Economics Ecology and Management, 104, 89–99. and Management, 30, 73–94. Alder, D.1992. Simple methods for calculating Baker, P.J.& Wilson, J.S.2000. A quan- minimum diameter and sustainable yield in titative technique for the identification of mixed tropical forest. In: Miller, F.M.& canopy stratification in tropical and temper- Adam, K.L.(editors), Wise Management of ate forests.Forest Ecology and Management, Tropical Forests.Oxford Forestry Institute, 127, 77–86. Oxford, 189–200. Barden, L.S.1989. Repeatability in forest gap re- Alder, D.& Silva, J.N.M.2000.An empirical co- search.Ecology, 70, 558–559. hort model for manangement of Terra Firme Barker, M.G.,Press, M.C.,& Brown, N.D.1997. forests in the Brazilian Amazon.Forest Ecol- Photosynthetic characteristics of dipterocarp ogy and Management, 130, 141–157. seedlings in three tropical rain forest light en- Alexandrov, G. A., Oikawa, T., & Esser, G. 1999a. vironments: a basis for niche partioning? Oe- Estimating terrestrial NPP: what the data say cologia, 112, 453–463. 146 Bibliography

Barreto, P., Amaral, P., Vidal, E., & Uhl, C. 1998. Boscolo, M., Buongiorno, J., & Panayotou, T. 1997. Costs and benefits of forest management for Simulating options for carbon sequestration timber production in eastern Amazonia.For- through improved management of a lowland est Ecology and Management, 108, 9–26. tropical rainforest.Environment and Devel- opment Economics, 2, 241–263. Bartelink, H.H.1998a. Radiation interception by forest trees: a simulation study on effects of Bossel, H.1992. Modellbildung und Simulation. stand density and foliage clustering on absorp- Konzepte Verfahren und Modelle zum Ver- tion and transmission.Ecological Modelling, halten dynamischer Systeme.Vieweg, Braun- 105, 213–225. schweig. Bartelink, H.H.1998b. Simulation of growth and Bossel, H.1994. Process models of environmental competition in mixed stands of Dougla-fir and systems: Simulation of forest development dy- beech.PhD thesis, University of Wageningen, namics as an example.Annals of Operations Wageningen, The Netherlands. Research, 54, 175–202. Bossel, H.1996a. Ecosystems and society: Impli- Battaglia, M.& Sands, P.J.1998. Process-based cations for sustainable development.World forest productivity models and their applica- Futures, 47, 143–213. tion in forest management.Forest Ecology and Management, 102, 13–32. Bossel, H.1996b. TREEDYN3 forest simulation model.Ecological Modelling, 90, 187–227. Bazzaz, F.A.& Pickett, S.T.A.1980.Physiolog- ical ecology of tropical succession: a compar- Bossel, H.& Krieger, H.1991.Simulation model of ative review.Annual Review of Ecology and natural tropical forest dynamics.Ecological Systematics, 11, 287–310. Modelling, 59, 37–71. Belsky, A.J.& Conham, C.D.1994. Forest Gaps Bossel, H.& Krieger, H.1994.Simulation of multi- and Isolated Savanna Trees.BioScience, 44,. species tropical forest dynamics using a verti- cally and horizontally structured model.For- Benitez-Malvido, J.1998.Impact of forest fragmen- est Ecology and Management, 69, 123–144. tation on seedling abundance in a tropical rain Botkin, D.B.1975. Functional groups of organ- forest.Conservation Biology, 12, 380–389. isms in model ecosystems. In: Levin, S.A. Bertault, J.G.& Sist, P.1997. An experimental (editor), Ecosystem Analysis and Prediction. comparison of different harvesting intensities Philadelphia Society of Industrial and Applied with reduced-impact and conventional logging Mathematics, 98–102. in East Kalimantan Indonesia.Forest Ecology Botkin, D.B.1993.Forest Dynamics: An Ecologi- and Management, 94, 209–218. cal Model.Oxford University Press, Oxford. Biehounek, A.1999.Modellierung von Wasser- und Botkin, D. B., Janak, J. F., & Wallis, J. R. 1972. N¨ahrstoffstress im tropischen Tieflandregen- Some ecological consequences of a computer wald - eine Erweiterung des Regenwaldmod- model of forest growth.Journal of Ecology, ells Formix3.Diploma thesis, University of 60, 849–872. Kassel, Germany. Bowles, I. A., Rice, R. E., Mittermeier, R. A., & Blasco, F., Whitmore, T. C., & Gers, C. 2000. A da Fonseca, G.A.B.1998.Logging and trop- framework for the worldwide comparison of ical forest conservation.Science, 280, 1899– tropical woody vegetation types.Biological 1900. Conservation, 95, 175–189. Box, E.O.1996.Plant functional types and climate Bolker, B. M., Deutschman, D. H., Hartvigsen, G., at the global scale.Journal of Vegetation Sci- & Smith, D.L.1997. Individual-based mod- ence, 7, 309–320. elling: what is the difference? Trends in Ecol- Brokaw, N.& Busing, R.T.2000. Niche ver- ogy and Evolution, 12, 111. sus chance and tree diversity in forest gaps. Boot, R.G.A.& Gullison, R.E.1995.Approaches Trends in Ecology and Evolution, 15, 183– to developing sustainable extraction systems 188. for tropical forest products.Ecological Appli- Brokaw, N.& Thompson, J.2000.The H of DBH. cations, 5, 894–903. Forest Ecology and Management, 129, 89–91. Borgesa, J.G.& Hoganson, H.M.2000. Struc- Brokaw, N.V.L.1982.Treefalls: frequency timing turing a landscape by forestland classification and consequences. In: Leigh, E.G.,Rand, and harvest scheduling spatial constraints. S.A.,& Windsor, D.M.(editors), The Ecol- Forest Ecology and Management, 130, 269– ogy of a Tropical Forest.Smithsonian Institu- 275. tion Press, Washington, DC, 101–108. Bibliography 147

Brokaw, N.V.L.1985.Gap-Phase regeneration in Bugmann, H.1996a. Functional types of trees in a tropical forest.Ecology, 66, 682–687. temperate and boreal forests: Classification and testing.Journal of Vegetation Science, 7, Brokaw, N.V.L.& Scheiner, S.M.1989. Species 359–376. composition in gaps and structure in a tropical forest.Ecology, 70, 538–541. Bugmann, H.1997. An efficient method for esti- mating the steady-state species composition Bronstein, I.N.& Semendjajew, K.A.1991. of forest gap models.Canadian Journal of Taschenbuch der Mathematik.Teubner, Forest Research, 27, 551–556. Stuttgart, Leipzip, Germany. Bugmann, H.K.M.1996b. A simplified forest Brown, I.F.,Martinelli, L.A.,Thomas, W.W., model to study species composition along cli- Moreira, M.Z.,Ferreira, C.A.C.,& Victo- mate gradients.Ecology, 77, 2055–2074. ria, R.A.1995. Uncertainty in the biomass of Amazonian forests: an example from Bugmann, H.K.M.& Solomon, A.M.2000. Rondˆonia, Brazil.Forest Ecology and Man- Explaining forest composition and biomass agement, 75, 175–189. across multiple biogeographical regions.Eco- logical Applications, 10, 95–114. Brown, N.1993. The implications of climate and gap microclimate for seedling growth condi- Burgess, P.F.1961. The structure and composi- tions in a Bornean lowland rain forest.Jour- tion of lowland tropical rain forest in North nal of Tropical Ecology, 9, 153–168. Borneo.Malayan Forester, 24, 66–80. Brown, N.1996. A gradient of seedling growth Burgess, P.F.1966.Timbers of Sabah. Number 6 from the centre of a tropical rain forest canopy in Sabah Forest Record.Sabah Forestry De- gap.Forest Ecology and Management, 82, partment, Sandakan. 239–244. Buschbacher, R., Uhl, C., & Serrao, E. A. S. 1988. Brown, N.1998.Out of control: fires and forestry in Abandoned pastures in eastern Amazonia.II: Indonesia.Trends in Ecology and Evolution, Nutrient stocks in the soil and vegetation. 13, 41. Journal of Ecology, 76, 682–699. Brown, N.D.1990. Dipterocarp regeneration in Cannon, C. H., Peart, D. R., & Leighton, M. 1998. tropical rain forest gaps of different sizes.PhD Tree species diversity in commercially logged thesis, University of Oxford, Oxford, U.K. Bornean rainforest.Science, 281, 1366–1368. Brown, S.1997. Estimating biomass and biomass Cannon, C. H., Peart, D. R., Leighton, M., & change of tropical forests.A Forest Re- Kartawinata, K.1994. The structure of low- sources Assessment publication, volume 134 land rainforest after selective logging in West of Forestry Papers.FAO, Rome. Kalimantan Indonesia.Forest Ecology and Management, 67, 49–68. Brown, S.& Lugo, A.E.1990.Tropical secondary forests.Journal of Tropical Ecology, 6, 1–32. Cao, M.& Woodward, F.I.1998. Dynamic re- sponses of terrestrial ecosystem carbon cyling Brown, S.& Lugo, A.E.1994. Rehabilitation of to global climate change.Nature, 393, 249– tropical lands: a key to sustaining develop- 252. ment.Restoration Ecology, 2, 97–111. Carey, E.V.,Brown, S.,Gillespie, A.J.R.,& Lugo, Bruenig, E.F.1996. Conservation and Manage- A.E.1994. Tree mortality in mature low- ment of Tropical Rainforests: An Integrated land tropical moist and tropical lower mon- Approach to Sustainability.CAB Interna- tane moist forests of Venezuela.Biotropica, tional, Wellington. 26, 255–265. Bruijnzeel, L.A.1991.Nutrient Input-Output Bud- Centeno, J.C.1995. Estrat´ egia para el desarrollo gets of Tropoical Forest Ecosystems: A Re- forestal de Venezuela.Fondo Nacional de In- view.Journal of Tropical Ecology, 7, 1–24. vestigaci´on Forestal, Caracas, Venezuela. Bruijnzeel, L.A.1992. Managing tropical wa- Chai, D., Kilou, P., & Kleine, M. 1991. Field tersheds for production: where contradictory manual for the medium-term Forest Man- theory and practis co-exist. In: Miller, F.R. agement Planning Inventory.GTZ re- & Adam, K.L.(editors), Wise management port 18, Malysian-German Forestry Research of tropical forests.Oxford Forestry Institute, Project, Forestry Department Sabah, San- Oxford, 37–75. dakan, Sabah, Malaysia. Brunner, A.1998. A light model for spatially ex- Chambers, J.Q.,Higuchi, N.,& Schimel, J.P.1998. plicit forest stand models.Forest Ecology and Ancient trees in Amazonia.Nature, 391, 135– Management, 107, 19–46. 136. 148 Bibliography

Chambers, J. Q., Higuchi, N., Schimel, J. P., Fer- Clark, D.B.& Clark, D.H.A.2000.Tree growth, reira, L.V.,& Melack, J.M.2000. Decom- mortality, physical condition, and microsite in position and carbon cycling of dead trees in old-growth lowland tropical rain forest.Ecol- tropical forests of the central Amazon.Oe- ogy, 81, 294. cologia, 122, 380–388. Clark, D.B.,Palmer, M.W.,& Clark, D.A.1999a. Charles-Dominique, P.1993.Speciation and coevo- Edaphic factors and the landscape-scale dis- lution: an interpretation of frugivory phenom- tributions of tropical rain forest trees.Ecol- ena.Vegetatio, 107/108, 75–84. ogy, 80, 2662–2675. Chave, J.1999a. Dynamique spatio-temporelle Clark, J.S.1992a. Density-Independent Mortal- de la foret tropicale.PhD thesis, Universite ity Density Compensation Gap Formation and d’Orsay, Paris, France. Self-thinning in Plant Populations.Theoreti- cal Population Biology. Chave, J.1999b. Study of structural, successional and spatial pattern in tropical rain forests us- Clark, J.S.1992b.Relationships among individual ing TROLL, a spatially explicit forest model. plant growth and the dynamics of populations Ecological Modelling, 124, 233–254. and ecosystems. In: deAngelis, D.S.& Gross, L.J.(editors), Individual-based models and Chave, J., Dubois, M. A., & Ri´era, B.2000a. Wet approaches in ecology.Chapman Hall, 421– tropical forest response to long term climate 454. changes I: Theoretical implications of seed dis- persal limitations.Unpublished manuscript. Clark, J.S.1998. Why trees migrate so fast: con- fronting theory with dispersal biology and the Chave,J.,Dubois,M.A.,&Ri´era, B.2000b. Paleorecord.American Naturalist, 152, 204– Wet tropical forest response to long term cli- 224. mate changes II: Applications to a neotropi- cal rain forest of French Guiana.Unpublished Clark, J. S., Fastie, C., Hurtt, G., Jackson, S. T., manuscript. Johnson, C., King, G. A., Lewis, M., Lynch, J., Pacala, S., Prentice, C., Schupp, E. W., Chave, J., Ri´era, B.,& Dubois, M.A.2000c.Eval- Webb, T., & Wyckoff, P. 1998a. Reid’s para- uation of biomass in a neotropical forest of dox of rapid plant migration.BioScience, 48, French Guiana: spatial and temporal variabil- 13–24. ity.Journal of Tropical Ecology, submitted. Clark, J. S., Macklin, E., & Wood, L. 1998b. Stages Chavez, F.P.,Strutton, P.G.,Friedrich, G.E., and spatial scales of recruitment limitation Feely, R.A.,Feldman, G.C.,Foley, D.G.,& in southern Appalachian forests.Ecological McPhaden, M.J.1999.Biological and chemi- Monographs, 68, 213–235. cal response of the equatorial Pacific Ocean to Clark, J. S., Silman, M., Kern, R., Macklin, E., the 1997-98 El Ni˜no.Science, 286, 2126–2131. & HilleRisLambers, J.1999b. Seed disper- Chazdon, R.L.,Colwell, R.K.,& Denslow, J.S. sal near and far: pattern across temerate and 1999.Comments on: Tropical tree rich- tropical forests.Ecology, 80, 1475–1494. ness and resource-based niches.Science, 285, Cochrane, M. A., Alencar, A., Schulze, M. D., 1459a. Jr., C. M. S., Nepstad, D. C., Lefebvre, P., Chim, L.T.& On, W.F.1973. Density recruit- & Davidson, E.A.1999. Positive feedbacks ment mortality and growth of Dipterocarp in the fire dynamic of closed canopy tropical seedlings in virgin and logged-over forests. forests.Science, 284, 1832–1835. The Malaysian Forester, 36, 3–15. Cockburn, P.F.1980. Trees of Sabah I and II. Cincotta, R. P., Wisnewski, J., & Engelman, R. Sabah Forest Record No.10.Sabah Forestry 2000.Human population in the biodiversity Department, Sandakan. hotspots.Nature, 404, 990–992. Condit, R.1995.Research in large long-term trop- ical forest plots.Trends in Ecology and Evo- Clark, D.A.& Clark, D.B.1992. Life his- lution, 10, 18–21. tory diversity of canopy and emergent trees in a neotropical rain forest.Ecological Mono- Condit, R.1998. Tropical Forest Census Plots. graphs, 62, 315–344. Springer, New York. Clark, D.A.& Clark, D.B.1999. Assessing the Condit, R., Ashton, P. S., Baker, P., Bunyave- growth of tropical rain forest trees: Issues for jchewin, S., Gunatillke, S., Gunatillke, N., forest modeling and management.Ecological Hubbell, S. P., Foster, R. B., Ioth, A., Applications, 3, 981–997. LaFrankie, J. V., Lee, H. S., Losos, E., Bibliography 149

Manokaran, N., Sukumar, R., & Yamakura, Curran, L.M.& Leighton, M.2000. Vertebrate T.2000.Spatial pattern in the distribution of responses to spatiotemporal variation in seed tropical tree species.Science, 288, 1414–1418. production of mast-fruiting Dipterocarpaceae. Ecological Monographs, 70, 101–128. Condit, R., Hubbell, S. P., & Foster, R. B. 1992. Short-term dynamics of a neotropical forest Curran, L.M.& Webb, C.O.2000. Experimen- change with limits.BioScience, 42, 822–828. tal tests of the spatiotemporal scale of seed predation in mast-fruiting Dipterocarpaceae. Condit, R., Hubbell, S. P., & Foster, R. B. 1995a. Ecological Monographs, 70, 129–148. Demography and harvest potential of Latin American timber species: data from a large da Fonseca, G.A.B.1985.The vanishing Brasilian permanent plot in Panama.Journal of Tropi- Atlantic forest.Biological Conservation, 34, cal Forest Science, 7, 599–622. 17–34. Condit, R., Hubbell, S. P., & Foster, R. B. 1995b. da Silva, J.M.C.& Tabarelli, M.2000.Tree species Mortality rates of 205 neotropical tree and impoverishment and the future flora of the At- shrub species and the impact of a severe lantic forest of northeast Brazil.Nature, 404, drought.Ecological Monographs, 65, 419– 72–74. 439. Davies, S.J.1998. Photosynthesis of nine pioneer Condit, R.,Hubbell, S.P.,& Foster, R.B.1996.As- Macaranga species from Borneo in relation to sessing the response of plant functional types life history.Ecology, 79, 2292–2308. to climatic change in tropical forests.Journal of Vegetation Science, 7, 405–416. de Graaf, N.R.1986. A silvicultural system for natural regeneration of tropical rain forest in Cook, S.1998. A diversity of approaches to the Suriname.PhD thesis, Agricultural Univer- study of species richness.Trends in Ecology sity of Wageningen, The Netherlands. and Evolution, 13, 340–341. DeAngelis, D.L.& Gross, L.J.1992. Individual- Cosson, J. F., Ringuet, S., Claessens, O., de Mas- Based Models: Concepts and Approaches sary, J. C., Dalecky, A., Villiers, J., Granjon, in Ecology; Populations Communities and L.,& Pons, J.M.1999.Ecological changes in Ecosystems.Proceedings of a Sympo- recent land-bridge islands in French Guiana, sium/Workshop Held in Knoxville, Tennessee, with emphasis on vertebrate communities.Bi- May 16-19, 1990.Chapman and Hall, London. ological Conservation, 91, 213–222. Cramer, W.1997. Using plant functional types in DeFries, R.S.,Hansen, M.C.,Townshend, J.R.G., a global vegetation model. In: Smith, T.M., Janetos, A.C.,& Loveland, T.R.2000.A new Shugart, H.H.,& Woodward, F.I.(editors), global 1-km dataset of percentage tree cover Plant Funcitional Types.Cambridge Univer- derived from remote sensing.Global Change sity Press, Cambridge, 271–288. Biology, 6, 247–254. Crome, F.H.J.1991. Wildlife conservation DeLucia, E. H., Hamilton, J. G., Naidu, S. L., and rain forest management - Examples from Thomas, R.B.,J.A.,Finzi, A.A.,Lavine, north-east Queensland. In: Gomez-Pompa, M., Matamala, R., Mohan, J. E., Hendrey, A., Whitmore, T. C., & Hadley, M. (editors), G.R.,& Schlesinger, W.H.1999. Net pri- Rain Forest Regeneration and Management. mary production of a forest ecosystem with Parthenon Press, Paris, France, 407–418. experimental CO2 enrichment.Science, 284, 1177–1179. Crome, F.H.J.,Moore, L.A.,& Richards, G.C. 1992.A study of logging damage in upland Denslow, J.S.1987. Tropical rainforest gaps and rainforest in North Queensland.Forest Ecol- tree species diversity.Annual Review of Ecol- ogy and Management, 49, 1–29. ogy and Systematics, 18, 431–451. Cullen, L., Bodmer, R. E., & P´adua, C.V.2000. Denslow, J.S.1996.Functional group diversity and Effects of hunting in habitat fragments of the responses to disturbances. In: Orians, G.H., Atlantic forests.Brzil. Biological Conserva- Dirzo, R., & Cushman, J. H. (editors), Biodi- tion, 96, 49–56. versity and Ecosystem Processes in Tropical Forests.Springer, Berlin, 127–152. Curran, L. M., Caniago, I., Paoli, G. D., Astianti, D., Kusneti, M., Leighton, M., Nirarita, C. E., Denslow, J. S., Ellison, A. M., & Sanford, R. E. & Haeruman, H.1999.Impact of El Ni˜no and 1998.Treefall gap size effects on above- and logging on canopy tree recruitment in Borneo. below-ground processes in a tropical wet for- Science, 286, 2184–2188. est.Journal of Ecology, 86, 597–609. 150 Bibliography

Deutschmann, D.H.,Levin, S.A.,& Pacala, S.W. Enright, N.J.1978. The effects of logging on the 1999.Error propagation in a forest succession regeneration and nutrient budget of Araucaria model: The role of fine-scale heterogeneity in cunninghamii dominated tropical rainforest in light.Ecology, 80, 1927–1943. Papua New Guinea.The Malaysian Forester, 41, 303–318. Diaz, I., Papic, C., & Armesto, J. J. 1999. An as- sessment of post-dispersal seed predation in Eschenbach, C.1994. Photosynthesis rates of temperate rain forest fragments in Chilo´eIs- selected tree species.GTZ report 189, land Chile.OIKOS, 87, 228–238. Malaysian-German Sustainable Forest Man- Ditzer, T.1999. Struktur und Dynamik agement Project, Sandakan, Malaysia. nat¨urlicher und bewirtschafteter Diptero- Eschenbach, C., Glauner, R., Kleine, M., & Kap- carpaceenw¨alder: Eine Fallstudie der Modell- pen, L.1998.Photosynthetic rates of selected bildung und Simulation f¨ur die Bewirtschaf- tree species in lowland Dipterocarp rain forest tung nat¨urlicher Ressourcen unter Unsicher- of Sabah Malaysia.Trees, 12, 356–365. heit.Number 131 in G¨ ottinger Beitr¨age zur Land- und Forstwirtschaft in den Tropen und Fahrig, L.1992.Relative importance of spatial and Subtropen.Erich Goltze Verlag, G¨ottingen, temporal scales in a patchy environment.The- Germany. oretical Population Biology, 41, 300–314. Ditzer, T., Glauner, R., F¨orster, M., K¨ohler, P., Fan, S., Gloor, J., Mahlman, J., Pacala, S., & Huth, A.2000. The process-based stand Sarmiento, J., Takahashi, T., & Tans, P. 1998. growth model Formix3-Q applied in a GIS- A large terrestrial carbon sink in north amer- environment for growth and yield analysis in ica implied by atmospheric and oceanic diox- a tropical rain forest.Tree Physiology, 20, ide data and models.Science, 282, 442–446. 367–381. d’Oliveira, M.V.N.2000.Artifical regeneration in FAO 1997.State of the world‘s forests 1997.FAO, gaps and skidding trails after mechanised for- Rom. est exploration in Acre, Brazil.Forest Ecology Favrichon, V.1995. Modele Matriciel Determin- and Management, 127, 67–76. iste en Temps Discret.PhD thesis, Universite Doyle, T.W.1981. The role of disturbance in the Claude Bernard, Lyon, France. gap dynamics of montane rain forest: an ap- plication of a tropical forest succession model. Ferreira, L.V.& Laurance, W.F.1997. Effects of In: West, D. C., Shugart, H. H., & Botkin, forest fragmentation on mortality and damage D.B.(editors), Forest Succession: Concepts of selected trees in central Amazonia.Conser- and Application.Springer, New York, 56–73. vation Biology, 11, 797–801. Duncan, R.S.& Chapman, C.A.1999. Seed dis- Ferreira, L.V.& Prance, G.T.1999. Ecosystem persal and potential forest succession in aban- recovery in terra firme forests after cutting doned agriculture in tropical Africa.Ecologi- and burning: a comparison on species rich- cal Applications, 9, 998–1008. ness, floristic composition and forest structure in the Ja´u National Park, Amazonia.Botan- Ellsworth, D.S.& Reich, P.B.1996. Photosyn- ical Journal of the Linnean Society, 130, 97– thesis and leaf nitrogen in five Amazonian 110. tree species during early secondary succession. Ecology, 77, 581–594. Finegan, B.1996. Pattern and process in neotrop- ical secondary rain forests: the first 100 years Endejan, M.1997. Entwicklung und Analyse eines of succession.Trends in Ecology and Evolu- Modells zur Beschreibung von Wasserstress tion, 11, 119–123. und dessen Wirkung auf die Wachstumsdy- namik eines Fichtenbestandes.Diploma the- F¨olster, H., v. F¨urstenberg, P., Lamprecht, H., Pe- sis, Hochschule Vechta, Germany. ters, K. J., Pretzsch, J., Rehm, S., Riechert, Enquete-Kommission (editor) 1990.Vorsorge zum C., Roth, H. H., Steinlin, H. J., & Wei- Schutz der Erdatmosphaere des Deutschen delt, H.J.1986. Erhaltung und nachhaltige Bundestages - Schutz der Tropenwaelder.Eco- Nutzung tropischer Regenw¨alder.Num- nomica Verlag, Bonn. ber 74 in Forschungsberichte des Bundesmin- isteriums f¨ur wirtschaftliche Zusammenarbeit. Enquete-Kommission (editor) 1994.Schutz der Weltforum Verlag, M¨unchen. Gr¨unen Erde - Klimaschutz durch umwelt- gerechte Landwirtschaft und Erhalt der Forest-Stewardship-Council 2000. W¨alder.Economica, Bonn. http://www.fscoax.org. www-pages. Bibliography 151

Forestal-International-Limited 1973.Sabah for- richness in tropical forest remnants.Biological est inventory 1969-1972.Research re- Conservation, 91, 223–229. port, Sabah Forestry Department, Sandakan, Sabah, Malaysia. Gascon, C., Williamson, G. B., & da Fonseca, G. A.B.2000.Receding forest edges and vanish- Foster, D. R., Fluet, M., & Boose, E. R. 1999. ing reserves.Science, 288, 1356–1358. Human or natural disturbance: landscape- scale dynamics of the tropical forests of Puerto Gentry, A.H.1988. Tree species richness of up- Rico.Ecological Applications, 9, 555–572. per Amazonian rainforests.Proceedings of the National Academy of Sciences USA,, 85, 156– Fowler, J., Cohen, L., & Jarvis, P. 1998. Practical 159. statistics for field biology.Wiley and Sons, Chichester, U.K. Gerthsen, C., Kneser, H. O., & Vogel, H. 1989. Physik.Springer, Berlin. Fox, J.E.D.1970. Preferred Check-List of Sabah Trees.Number 7 in Sabah Forest Record.Bor- Gitay, H.& Noble, I.R.1997. What are func- neo Literature Bureau, Sandakan, Malaysia. tional types and how should we seek them? In: Smith, T.M.,Shugart, H.H.,& Wood- Fox, J.E.D.1972.The natural vegetation of Sabah ward, F.I.(editors), Plant Functional Types. and natural regeneration of the dipterocarp Cambridge University Press, Cambridge, 3– forests.PhD thesis, University of Wales, U.K. 19. Franco, W.1979.Die Wasserdynamik einiger Wald- Gitay, H., Noble, I. R., & Connell, J. H. 1999. De- standorte der West-Llanos Venezuelas und riving functional types for rain-forest trees. ihre Beziehung zur Saisonalit¨at des Laub- Journal of Vegetation Science, 10, 641–650. falls.Number 61 in G¨ottinger Bodenkundliche Berichte.Erich Goltze Verlag, G¨ ottingen, Glauner, R.1999. Die Bedeutung der Boden- Germany. und Standortkunde in der Forsteinrich- tung von bewirtschafteten Naturw¨aldern. Frederickson, T.S.& Mostacedo, B.2000.Regener- Ein Beispiel aus Sabah/Malaysia.Number ation of timber species following selection log- 130 in G¨ottinger Beitr¨age zur Land- und ging in a Bolivian tropical dry forest.Forest Forstwirtschaft in den Tropen und Subtropen. Ecology and Management, 131, 47–55. Erich Goltze Verlag, G¨ottingen, Germany. Friend, A.D.1993. The prediction and physiolog- ical significance of tree hight. In: Solomon, Goldammer, J.G.1999. Forests on fire. Science, 284 A.M.& Shugart, H.H.(editors), Vegetation , 1782–1783. Dynamics and Global Change.Chapman and Golley, F.B.(editor) 1983. Tropical rain forest Hall, New York, 101–115. ecosystems (Ecosystems of the World 14A). Friend, A.D.1997. Hybrid v4.1. In: of the Eu- Elsevier, Amsterdam. ropean Communities, C.(editor), Long-term Gourlet-Fleury, S.1997. Mod´ elisation individuelle effects of CO2-increase and climate change on spatialement explicite de la dynamique d’un European forests (LTEEF).Wageningen, 62– peuplement de forˆet dense tropicale humide. 81. PhD thesis, Universit´e Claude Bernard, Lyon, Friend, A. D., Stevens, A. K., Knox, R. G., & Can- France. nell, M.G.R.1997. A process based terres- Gourlet-Fleury, S.& Houllier, F.2000. Modelling trial biosphere model of ecosystem dynamics diameter increment in a lowland evergreen (Hybrid v3.0).Ecological Modelling, 95, 249– rain forest in French Guiana.Forest Ecology 287. and Management, 131, 269–289. Garwood, N.C.1983. Seed dormancy in a sea- Grace, J., Lloyd, J., McIntyre, J., Miranda, A. C., sonal tropical forest in Panama: a community Meir, P., Miranda, H. S., Nobre, C., Mon- study.Ecological Monographs, 53, 159–181. crieff, J., Massheder, J., Malhi, J., Wright, I., Garwood, N.C.1989. Tropical soil seed banks: & Gash, J.1995. Carbon dioxide uptake by a review. In: Pickett, S.T.& McDonald, an undisturbed tropical rain forest in South- M.J.(editors), Ecology of soil seed banks. west Amazonien 1992 to 1993.Science, 270, Academic Press, New York, 149–209. 778–780. Gascon, C., Lovejoy, T. E., Bierregaard, R. O., Mal- Graf, Henning, H.P.,Stange, & Wilrich, P.T. colm, J.R.,Stouffer, P.C.,Vasconcelos, H.L., 1987.Formeln und Tabellen der angewan- Laurance, W. F., Zimmerman, B., Tocher, M., deten mathematischen Statistik.Springer, & Borges, S.1999.Matrix habitat and species Berlin, Heidelberg, New York. 152 Bibliography

Grainger, A.1998. Modelling tropical land use Hashimotio, T., Kojima, K., Tang, T., & Sasaki, change and deforestation. In: Goldsmith, S.2000. Changes in carbon storage in fallow F.B.(editor), Tropical Rain Forest: A forests in the tropical lowlands of Borneo.For- Wider Perspective.Chapman and Hall, Lon- est Ecology and Management, 126, 331–337. don, 303–344. Haupt, W.1995. Simulationsmodell zur Op- Grime, J. P., Hodgson, J. G., Hunt, R., Thomp- timierung von nachhaltigen Nutzungsstrate- son, K., Hendry, G. A. F., Campbell, B. D., gien - am Beispiel eines logistisch wachsenden Jalili, A., Hilier, S. H., Diaz, S., & Burke, M. Waldes.Diploma thesis, University of Kassel, J.W.1997.Functional tpes: testing the con- Germany. cept in Northern England. In: Smith, T.M., Shugart, H.H.,& Woodward, F.I.(editors), Hendrison, J.1990. Damage-controlled logging in Plant Functional Types.Cambridge Univer- managed rain forest in Suriname.Agricultural sity Press, Cambridge, 122–152. University Wageningen, The Netherlands. Grimm, V.1999. Ten years of individual-based Higgins, S.I.& Richardson, D.M.1999. Predict- modelling in ecology: what have we learned ing plant migration rates in a changing world: and what could we learn in the future? Eco- the role of long-distance dispersal.American logical Modelling, 115, 129–148. Naturalist, 153, 464–475. Groombridge, B.1992.Global Biodiversity.Chap- Higuchi, N., Santos, J., Ribeiro, R. J., Minette, man and Hall, London. L., & Biot, Y. 1998. Biomassa da parte ´aerea da vegeta¸c˜ao da floresta tropicalumida ´ Guariguata, M.R.2000. Seed and seedling ecol- de terra-firme da Amazˆonia brasileira.Acta ogy of tree species in neotropical secondary Amazˆonica, 28, 153–166. forests: management implications.Ecological Applications, 10, 145–154. Holl, K.D.& Kappelle, M.1999. Tropical forest recovery and restoration.Trends in Ecology Haefner, J.W.& Dugaw, C.J.2000. Individual- and Evolution, 14, 378–379. based models solved using fast Fourier trans- forms.Ecological Modelling, 125, 159–172. Holling, C.S.1973.Resilience and stability of eco- logical systems.Annual Review of Ecology Hahn-Schilling, B., Heuveldop, J., & Palmer, J. and Systematics, 4, 1–23. 1994.A comparative study of evaluation sys- tems for sustainable forest management. In: Host, G.E.& Pregitzer, K.S.1991. Ecological Heuveldop, J.(editor), Assessment of Sus- Species Groups for Upland Forest Ecosystems tainable Tropical Forest Management.Wiede- of Northwestern Lower Michigan.Forest Ecol- busch, Hamburg, 1–35. ogy and Management, 43, 87–102. Hall, S.J.& Matson, P.A.1999. Nitrogen oxide Houghton, R. A., Skole, D. L., Nobre, C. A., Hack- emissions after nitrogen addition in tropical ler, J.L.,Lawrence, K.T.,& Chomentowski, forests.Nature, 400, 152–155. W.H.2000. Annual fluxes of carbon from deforestation and regrowth in the Brazilian Hammond, D.S.,Brown, V.K.,& Zagt, R.1999. Amazon.Nature, 403, 301–304. Spatial and temporal pattern of seed attack and germination in a large-seeded neotropical Howard, A.F.& Valerio, J.1992. A diameter tree species.Oecologia, 119, 208–218. class growth model for assessing the sustain- ability.Commonwealth Forestry Review, 71, Hanski, L.& Ovaskainen, O.2000. The metapop- 171–177. ulation capacity of a fragmented landscape. Nature, 404, 755–758. Hubbell, S.P.1997. A unified theory of biogeog- raphy and relative species abundance and its Hartshorn, G.& Bynum, N.1999. Tropical forest application to tropical rain forest and coral synergies.Science, 286, 2093–2094. reefs.Coral Reefs, 16, 9–21. Hartshorn, G.S.1989.Application of gap theory to Hubbell, S.P.1999. Answers to comments: Trop- tropical forest management: Natural regener- ical tree richness and resource-based niches. ation on strip clear-cuts in the Peruvian Ama- Science, 285, 1459a. zon.Ecology, 70, 567–569. Hubbell, S. P., Condit, R., & Foster, R. B. 1991. Hase, H.& F¨olster, H.1982.Bioelement inventory Presence and absence of density dependence of a tropical (semi-) evergreen seasonal for- in a neotropical tree conservation.Philosiph- est on eutrophic alluvial soils Western Llanos ical Transactions of the Royal Society of Lon- Venezuela.Acta Oecologica, 4, 331–346. don, 330, 269–281. Bibliography 153

Hubbell, S.P.& Foster, R.B.1986a. Common- Jackson, R. B., Schenk, H. J., Jobbagy, E. G., ness and rarity in a neotropical forest: Im- Canadell, J., Colelo, G. D., Dickinson, R. E., plications for tropical tree conservation. In: Field, C. B., Friedlingstein, P., Heimann, M., Soule, M.E.(editor), Conservation Biology: Hibbard, K., Kicklighter, D. W., Kleidon, A., the science of scarcity and diversity.Sinauer Neilson, R. P., Parton, W. J., Sala, O. E., & Associates, Sunderland, Massachusetts, 205– Sykes, M.T.2000.Belowground consequences 231. of vegetation change and their treatment in models.Ecological Applications, 10, 470–483. Hubbell, S.P.& Foster, R.B.1986b. The spatial context of regeneration in a neotropical forest. Jacobs, M.1988. The Tropical Rain Forest. In: Gray,A.,Crawley,M.J.,&Edwards,P.J. Springer, Berlin. (editors), Colonization, Succession and Stabil- ity.Blackwell Scientific, Oxford, 395–412. Jansen, M., Martin, P. G., Sonntag, M., Bossel, H., & Ditzer, T. 1995. Modellierung von Be- Hubbell, S. P., Foster, R. B., O’Brian, S. T., Harms, standeswachstum Stoff- und Energieums¨atzen K. E., Condit, R., Wechsler, B., Wright, S. J., mit TREEDYN am Beispiel Solling F1.Num- & de Lao, S.L.1999.Light-Gap Distribution, ber B45 in Berichte des Forschungszentrums Recruitment Limitation and Tree Diversity in Wald¨okosysteme.Selbstverlag, G¨ottingen. a Neotropical Forest.Science, 283, 554–557. Johns, A.D.1985. Selective logging and wildlife Hughes, R.F.,Kauffman, J.B.,& Jaramillo, V.J. conservation in tropical rain-forest: problems 2000.Ecosystem-scale impacts of deforesta- and recommendations.Biological Conserva- tion and land use in a humid tropical region tion, 31, 355–375. in Mexico.Ecological Applications, 10, 515– 527. Johns, A.D.1997. Timber Production and Biodi- versity Conservation in Tropical Rain Forests. Huston, M., DeAngelis, D., & Post, W. 1988. New Cambridge University Press, Cambridge. computer models unify ecological theory.Bio- Science, 38, 682–691. Johns, J. S., Barreto, P., & Uhl, C. 1996. Logging Huth, A.1999. Modellierung des Wachstums und damage during planned and unplanned log- der Nutzung von tropischem Regenwald.Ha- ging operations in the eastern Amazon.Forest bilitation thesis, University of Kassel, Ger- Ecology and Management, 89, 59–77. many. Jonkers, W.B.J.1987. Vegetation structure log- Huth, A.& Ditzer, T.2000a.Long-term impacts of ging damage and silviculture in tropical rain logging in a tropical rain forest - a simulation forest in Suriname.Agricultural University study.Forest Ecology and Management, in Wageningen, The Netherlands. press. Jordan, F.2000. A reliability-theory approach to Huth, A.& Ditzer, T.2000b. Simulation of the corridor design.Ecological Modelling, 128, growth of tropical rain forests - Formix3. 211–220. Ecological Modelling, in press. Judson, O.P.1994.The rise of the individual-based Huth, A., Ditzer, T., & Bossel, H. 1998. The model in ecology.Trends in Ecology and Evo- rain forest growth model Formix3 - Model lution, 9, 9–14. description and analysis of forest growth and logging scenarios for the Deramakot forest re- Kammesheidt, L.1994. Bestandesstruktur serve (Malaysia).Number 124 in G¨ ottinger und Artendiversit¨at in selektiv genutzten Beitr¨age zur Land- und Forstwirtschaft in den Feuchtw¨aldern der westlichen Llanos Venezue- Tropen und Subtropen.Erich Goltze Verlag, las unter besonderer Ber¨ucksichtigung einiger G¨ottingen, Germany. aut¨okologischen Merkmale wichtiger Bau- marten.Number 100 in G¨ ottinger Beitr¨age Huth, A., Hahn-Schilling, B., & Bossel, H. 1994. zur Land- und Forstwirtschaft in den Tropen Untersuchung der Auswirkungen von ver- und Subtropen.Erich Goltze Verlag, G¨ottin- schiedenen Nutzungsstrategien auf das Wach- gen, Germany. stum von Moorregenwald - Simulationen mit Formix.Zeitschrift f¨ ur Okologie¨ und Kammesheidt, L.1998.Stand structure and spatial Naturschutz, 3, 217–225. pattern of commercial species in logged and unlogged Venezuelan forest.Forest Ecology IGBP 1990.The International Geosphere- and Management, 109, 163–174. Biosphere Programme: A study of global change - The initial core projects.Global Kammesheidt, L.1999. Liana infestation of trees: Change Report 12, International Geosphere some observations in a neotropical lowland Biosphere Programme. forest.Ecotropica, 5, 217–220. 154 Bibliography

Kammesheidt, L.2000. Some autecological char- Kira, T.& Yoda, K.1989. Vertical stratification acteristics of early to late successional tree in microclimate. In: Lieth, H.& Werger, M. species in Venezuela.Acta Oecologica, 21, J.A.(editors), Tropical Rain Forest Ecosys- 37–48. tems.Elsevier, Amsterdam, 55–71. Kammesheidt, L., K¨ohler, P., & Huth, A. 2000. Kobe, R.K.1999a. Comments on: Tropical tree Sustainable timber harvesting in Venezuela: a richness and resource-based niches.Science, modelling approach.Journal of Applied Ecol- 285, 1459a. ogy, in press. Kobe, R.K.1999b. Light gradient partitioning Kaspar, F.1996. Untersuchungen zur Koex- among tropical tree species through differen- istenz von Baumarten in malaysischen Re- tial seedling mortality and growth.Ecology, genw¨aldern mit einem Waldwachstumsmod- 80, 187–201. ell.Diploma thesis, University of Kassel, Ger- many. Koch, G.W.,Amthor, J.S.,& Goulden, M.L. 1994.Diurnal patterns of leaf photosynthesis Kato, R., Tadaki, Y., & Ogawa, H. 1978. Plant conductance and water potential at the top biomass and growth increment studies in Pa- of a lowland rain forest canopy in Cameroon- soh Forest.Malaysian National Journal, 30, measurements from the Radeau-Des-Cimes. 211–224. Tree Physiology, 14, 347–360. Keating, W.G.& Bolza, E.1982. Characteristics K¨ohler, P.1996. Ein individuenbasiertes Wach- properties and use of timbers in South-East stumsmodell zur Simulation tropischer Re- Asia North Australia and the Pacific.Inkarta genw¨alder.Diploma thesis, University of Kas- Press, Melbourne. sel, Germany. Kellman, M., Tackaberry, R., & Rigg, L. 1998. K¨ohler, P.1998. Parameter research for the trop- Structure and function in two tropical gallery ical rain forest growth model Formix4.Re- forest communities: implications for forest search report P9801, Center for Environmen- conservation in fragemented systems.Journal tal Systems Research, University of Kassel, of Applied Ecology, 35, 195–206. Germany. Kennedy, D.J.1991.The role of colonizing species K¨ohler, P., Chave, J., Ri´era, B., & Huth, A. 2000a. in the regeneration of dipterocarp rain forest. Long-term response of tropical rain forests PhD thesis, Aberdeen University, U.K. to the effects of fragmentation: a simulation Kennedy, D.N.& Swaine, M.D.1992.Germination study.To be submitted. and growth of colonizing species in artificial K¨ohler, P., Ditzer, T., & Huth, A. 2000b. Concepts gaps of different sizes in dipterocarp rain for- for the aggregation of tropical tree species est. In: Swaine, M.D.(editor), Tropical rain into functional types and the application on forest: Disturbance and recovery, number 335 Sabah’s lowland rain forests.Journal of Trop- in Philosophical Transactions of the Royal So- ical Ecology, 16, 591–602. ciety of London. Royal Society, London, U.K., 357–367. K¨ohler, P., Ditzer, T., & Huth, A. 2000c. The effects of logging, fragmentation and recruit- Kilou, P., Lagan, P., Glauner, R., & Kleine, ment on growth of dipterocarp forest.Journal M.1993. Medium term forest manage- of Ecology, submitted. ment planning inventory.GTZ report 174, Sabah Forestry Department, Malaysian- K¨ohler, P., Ditzer, T., Ong, R. C., & Huth, A. German Sustainable Forest Management 2001.Comparison of measured and simu- Project, Sandakan, Sabah, Malaysia. lated growth on permanent plots in Sabah’s rain forests.Forest Ecology and Management, Kimmins, J. P., Mailly, D., & Seely, B. 1999. Mod- 142, in press. elling forest ecosystem net primary produc- tion: the hybrid simulation approach used in K¨ohler, P.& Huth, A.1998a. The effect of tree FORECAST.Ecological Modelling, 122, 195– species grouping in tropical rain forest mod- 224. elling - Simulation with the individual based model Formind.Ecological Modelling, 109, Kira, T.1978. Community architecture and or- 301–321. ganic matter dynamics in tropical lowland rain forests of Southeast Asia with special ref- K¨ohler, P.& Huth, A.1998b. An individ- erence to Pasoh Forest West-Malaysia. In: ual based rain forest model - concepts and Tomlinson, P.B.& Zimmermann, M.H.(edi- simulation results. In: Kastner-Maresch, tors), Tropical Trees as Living Systems.Cam- A., Kurth, W., Sonntag, M., & Breckling, bridge University Press, Cambridge, 26–30. B.(editors), Individual-based structural and Bibliography 155

functional models in ecology, number 52 in Laurance, W.F.1999b. Reflection on the tropical Bayreuther Forum Okologie.Bayreuther¨ In- deforestation crisis.Biological Conservation, stitut f¨ur Terrestrische Okosystemforschung,¨ 91, 109–117. Bayreuth, Germany, 35–51. Laurance, W.F.& Bierregaard, R.O.(editors) Kohyama, T.1993. Size-structured tree popula- 1997.Tropical Forest Remnants.Univ.of tions in gap-dynamic forest.The forest archi- Chicago Press, Chicago. tecture hypothesis for the stable coexistence of species.Journal of Ecology, 81, 131–143. Laurance, W. F., Delamonica, P., Laurance, S. G., Vasconcelos, H.L.,& Lovejoy, T.E.2000. Kohyama, T.& Takada, T.1998.Recruitment rates Rainforest fragmentation kills big trees.Na- in forest plots: Gf estimates using growth ture, 404, 836. rates and site distribution.Ecology, 86, 633– 639. Laurance, W.F.& Fearnside, P.M.1999.Amazon Korpelainen, H., Adjers, G., Kuusipalo, J., burning.Trends in Ecology and Evolution, Nuryanto, K., & Otsamo, A. 1995. Profitabil- 14, 457. ity of rehabilitation of overlogged dipterocarp Laurance, W.F., Ferreira, L.V., Rankin- forest: a case study from South Kalimantan deMerona, J.M.,& Laurance, S.G.1998. Indonesia.Forest Ecology and Management, Rain forest fragmentation and the dynamics 79, 207–215. of Amazonian tree communities.Ecology, 79, Koyama, H.1981. Photosynthetic rates in low- 2032–2040. land rain forest trees of Peninsular Malaysia. Japanese Journal of Ecology, 31, 361–369. Laurance, W. F., Gascon, C., & Rankin-deMerona, J.M.1999. Predicting effects of habitat de- Kremen, C., Niles, J., Dalton, Daily, M. G., struction on plant communities: a test of a Ehrlich, P., Fay, J., Grewal, D., & Guillery, model using Amazonian trees.Ecological Ap- R.P.2000. Economic incentives for rain for- plications, 9, 548–554. est conservation across scales.Science, 288, 1828–1832. Laurance, W. F., Laurance, S. G., Ferreira, L. V., K¨urpick, P., K¨urpick, U., & Huth, A. 1997. The in- Rankindemerona, J. M., Gascon, C., & Love- fluence of logging on a Malaysian Dipterocarp joy, T.E.1997. Biomass collapse in Ama- rain forest: a study using a forest gap model. zonian forest fragments.Science, 278, 1117– Journal of Theoretical Biology, 185, 47–54. 1118. Kyereh, B., Swaine, M. D., & Thompson, J. 1999. Laurance, W.F.& Yensen, E.1991.Predicting the Effect of light on the germination of forest impacts of edge effects in fragmented habitats. trees in Ghana.Journal of Ecology, 87, 772– Biological Conservation, 55, 77–92. 783. Lavorel, S., McIntyre, S., Landsberg, J., & Forbes, Lamprecht, H.1986. Waldbau in den Tropen. T.D.A.1997.Plant functional classifications: Parey, Hamburg. from general groups to specific groups based Lamprecht, H.1989. Silviculture in the tropics. on response to disturbance.Trends in Ecology Deutsche Gesellschaft f¨ur Technische Zusam- and Evolution, 12, 474–478. menarbeit (GTZ), Eschborn, Germany. Law, B.S.& Lean, M.1999.Common blosom bats Landsberg, J.J.& Gower, S.T.1997. Applica- (Syconycteris australis) as pollinators in frag- tions of Physiological Ecology to Forest Man- mented Australian tropical rainforest.Biolog- agement.Academic Press, San Diego. ical Conservation, 91, 201–212.

Landsberg, J.J.& Waring, R.H.1997.A general- Lawton, J. H., Bignell, D. E., Bolton, B., Bloe- ized model for forest productivity using sim- mers, G. F., Eggleton, P., Hammond, P. M., plified concepts of radiation-use efficiency car- Hodda, M., Holt, R. D., Larsen, T. B., Mawd- bon balance and partitioning.Forest Ecology sley, N.A.,Stork, N.E.,Srivastava, D.S.,& and Management, 95, 209–228. Watt, A.D.1998.Biodiversity inventories in- Lang, G.E.& Knight, D.H.1983. Tree growth dicator taxa and effects of habitat modifica- mortality, recruitment and canopy gap for- tion in tropical forest.Nature, 391, 72–76. mation during a 10-year period in a tropical Lee, D. W., Baskaran, K., Mansor, M., Mohamad, moist forest.Ecology, 64, 1075–1080. H., & K., Y. S. 1996. Irradiance and spectral Laurance, W.F.1999a.Introduction and synthesis. quality affect Asian tropical rainforest tree Biological Conservation, 91, 101–107. seedling developement.Ecology, 77, 568–580. 156 Bibliography

Lee, D. W., Oberbauer, S. F., Krishnapilay, B., Lieberman, D., Lieberman, M., Hartshorn, G., & Mansor, M., Mohamad, H., & Yap, S. K. 1997. Peralta, R.1985. Growth rates and age-size Effects of irradiance and spectral quality on relationships of tropical wet forest trees in seedling development of two Southeast Asian Costa Rica.Journal of Tropical Ecology, 1, Hopea species.Oecologia, 110, 1–9. 97–109. Leemans, R.1997. The use of plant functional Liedeker, H.(editor) 1999. G¨ utesiegel FSC: Der type classifications to model global land cover neue Weg zur Rettung der W¨alder.Umwelt- and simulate the interactions between the ter- stiftung WWF-Deutschland, Frankfurt, Ger- restrial biosphere and the atmosphere. In: many. Smith, T.M.,Shugart, H.H.,& Woodward, F.I.(editors), Plant Functional Types.Cam- Lieth, H.& Werger, M.J.A.(editors) 1989.Trop- bridge University Press, Cambridge, 289–317. ical Rain Forest Ecosystems.Biogeographical and Ecological Studies.Elsevier, Amsterdam. Leigh, E.G.,Rand, A.S.,& Windsor, D.M.(edi- tors) 1985.The Ecology of a Tropical Forest: Liu, J.& Ashton, P.1998. FORMOSAIC: Seasonal Rhythms and Long-Term Changes. an individual-based spatially explicit model Smithonian Institute Press, Washington, U. for simulating forest dynamics in landscape S.A. mosacis.Ecological Modelling, 106, 177–200. Leigh-Jr.,E.G.& de Lao, S.L.2000. Fisher’s Liu, J.& Ashton, P.S.1995.Individual-based sim- alpha: measuring tree diversity.Research re- ulation models for forest succession and man- port, Inside CTFS, publication of the Center agement.Forest Ecology and Management, for Tropical Forest Science, Summer 2000, 6- 73, 157–175. 7. Liu, J.& Ashton, P.S.1999.Simulating effects of Leighton, M.& Wirawan, N.1986. Catastrophic landscape context and timber harvest on tree Draught and Fire in Borneo Tropical Rain species diversity.Ecological Applications, 9, Forest Associated with the 1982-1983 El Nino 186–201. Southern Oszillation Event. In: Prance, G.T. Lovejoy, T. E., Rankin, J. M., Bierregaard-Jr., (editor), Tropical Rain Forests and the World R. O., Brown-Jr., K. S., Emmons, L. H., & Atmosphere.Westview, Boulder, 75–102. van der Voort, M.E.1984.Ecosystem decay of Leishman, M. R., Westoby, M., & Jurado, E. 1995. Amazon forest fragments. In: Nitecki, M.H. Correlates of seed size variation - A compari- (editor), Extinctions.University of Chicago son among 5 temperate floras.Journal of Ecol- Press, Chicago Ill, 295–325. ogy, 83, 517–529. Lozada, J.R.1998. Impacto de diferentes intensi- Lescure, J.-P.& Boulet, R.1985.Relationships be- dades de aprovechamiento forestal sobre la ca- tween soil and vegetation in a tropical rain pacidad de recuperaci´on de la masa remanente forest in French Guiana.Biotropica, 17, 155– en la estaci´on experimental de Caparo.Msc 164. thesis, Universidad de Los Andes, Facultad Lescure, J. P., Puig, H., Ri´era, B., Leclerc, D., de Ciencas Forestales y Ambientales, M´erida, Beekman, A., & Beneteau, A. 1983. La phy- Venezuela. tomasse ´epig´ee d’une forˆet dense en Guyane Lynam, A.J.1997.Rapid decline of small mammal Fran¸caise.Acta Ecologica, Ecol.Gener., 4, diversity in monsoon evergreen forest frag- 237–251. ments in Thailand. In: Laurance, W.F. Lett, C., Silber, C., & Barret, N. 1999. Compar- & Bierregaard-Jr., R.O.(editors), Tropi- ison of a cellular automata network and an cal Forest Remnants - Ecology, Management, individual-based model for the simulation of and Conservation of Fragmented Communi- forest dynamics.Ecological Modelling, 121, ties.Chicago University Press, Chicago Ill, 277–293. 222–239. Levin, S.A.& Paine, R.T.1974. Disturbance, Lynam, A.J.& Billick, I.1999. Differential re- patch formation, and community structure. sponses of small mammals to fragmentation Proceedings of the National Academy of Sci- in a Thailand tropical forest.Biological Con- ences USA, 71, 2744–2747. servation, 91, 191–200. Lieberman, D., Hartshorn, G. S., Lieberman, M., & Mackensen, J.& F¨olster, H.2000. Cost-analysis Peralta, R.1990.Forest dynamics at La Selva for a sustainable nutrient management of fast Biological Station 1969-1985. In: Gentry, growing-tree plantations in East-Kalimantan, A.H.(editor), Four Neotropical Rain forests. Indonesia.Forest Ecology and Management, Yale University Press, New Haven, 509–521. 131, 239–253. Bibliography 157

MacKinnon, K., Hatta, G., Halim, H., & Mangalik, Meijer, W.& Wood, G.H.S.1964. Diptero- A.1996.The Ecology of Kalimantan Indone- carps of Sabah.Number 5 in Sabah For- sian Borneo.Number III in The Ecology of est Record.Sabah Forestry Department, San- Indonesia Series.Periplus Edition. dakan, Malaysia. Malanson, G.P.1996.Effects of dispersal and mor- Mesquita, R. C. G., Delamonica, P., & Laurance, tality on diversity in a forest stand model. W.F.1999. Effect of surrounding vegetation Ecological Modelling, 87, 103–110. on edge-related tree mortality in Amazonian Malanson, G.P.& Armstrong, M.P.1996.Disper- forest fragmentation.Biological Conservation, sal probability and forest diversity in a frag- 91, 129–134. mented landscape.Ecological Modelling, 87, Milton, K., Laca, E. A., & Demment, M. W. 1994. 91–102. Successional patterns of mortality and growth Malhi, Y.& Grace, J.2000.Tropical forests and at- of large trees in a Panamanian lowland forest. mospheric carbon dioxide.Trends in Ecology Journal of Ecology, 82, 79–87. and Evolution, 15, 332–337. Moad, A.S.1992. Dipterocarp Juvenile Malmer, A.1996.Nutrient losses from Dipterocarp Growth and Understory Light Availability in forests - a case study of forest plantation es- Malaysian Tropical Forest.PhD thesis, Har- tablishment in Sabah Malaysia. In: Schulte, vard University, Cambridge. A.& Sch¨one, D.(editors), Dipterocarp For- Moloney, K.E.& Levin, S.A.1996. The effects est Ecosystems: Towards Sustainable Man- of disturbance architecture on landscape-level agement.World Scientific, Singapore, 52–73. population dynamics.Ecology, 77, 375–394. Manokaran, N.& Kochummen, K.M.1987. Re- ¨ cruitment growth and mortality of tree species Monsi, M.& Saeki, T.1953. Uber den Lichtfak- in a lowland Dipterocarp forest in Peninsular tor in den Pflanzengesellschaften und seine Malaysia.Journal of Tropical Ecology, 3, 315– Bedeutung f¨ur die Stoffproduktion.Japanese 330. Journal of Botany, 14, 22–52. Manokaran, N.& Swaine, M.D.1994. Population Mooney,H.A.,Bj¨orkman, O., Hall, A. E., Medina, Dynamics of Trees in Dipterocarp Forests of E., & Tomlinson, P. B. 1980. The study of Peninsular Malaysia.Number 40 in Malayan the physiological ecology of tropical plants - Forest Records.Forest Research Institute of current status and needs.BioScience, 30, 22– Malaysia, Kuala Lumpur. 26. Martinez-Garza, C.& Gonz´ alez-Montagut, R. Moorecroft, P., Hurtt, G., & Pacala, S. W. 2000. A 1999.Seed rain from forest fragements into method for scaling vegetation dynamics: the tropical pastures in Los Tuxtlas.Plant Ecol- ecosystem demography model (ED).Ecologi- ogy, 145, 255–265. cal Monographs, accepted. Maruyama, Y., Toma, T., Ishida, A., Matsumoto, Mori, S.A.1990.Diversity of Lecythidaceae in the Y.,Morikawa,Y.,Ang,L.H.,Yap,S.K.,& Guianas. In: Holm-Nielsen, L., Nielsen, I., & Iwasa, M.1997.Photosynthesis and water use Balslev, H.(editors), Tropical Forests: Botan- efficiency of 19 tropical tree species.Journal ical Dynamics, Speciation and Diversity.Aca- of Tropical Forest Science, 9, 434–438. demic Press. McIntyre, S., Diaz, S., Lavorel, S., & Cramer, W. Mulkey, S. S., Kitajima, K., & Wright, S. J. 1996. 1999a.Plant functional types and distur- Plant physiological ecology of tropical forest bances - Introduction.Journal of Vegetation canopies.Trends in Ecology and Evolution, Science, 10, 604–608. 11, 408–412. McIntyre, S., Lavorel, S., Landsberg, J., & Forbes, Murcia, C.1995.Edge effects in fragmented forests: T.D.A.1999b.Disturbance response in vege- Implications for conservation.Trends in Ecol- tation - towards a global perspective on func- ogy and Evolution, 10, 58–62. tional traits.Journal of Vegetation Science, 10, 621–630. Murphy, P.G.1975. Net primary productivity in tropical terrestral ecosystems. In: Lieth, H.& Medina, E.& Klinge, H.1983.Productivity of trop- Whittaker, R.H.(editors), Primary Produc- ical forests and tropical woodlands. In: Lange, tivity of the Biosphere.Springer, New York, O. L., Nobel, P. W., Osmond, C. B., & Ziegler, 217–231. H.(editors), Physiological Plant Ecology IV, Encyclopedia of Plant Physiology.Springer, Myers, N.1989. Threatened biotas:Hotspots in Berlin, 281–303. tropical forests.Environmentalist, 8, 1–20. 158 Bibliography

Myers, N.1990. The biodiversity challenge: ex- Osho, J.S.A.1996. Modelling the tree popula- panded hotspots analysis.Environmentalist, tion dynamics of the most abundant species 10, 243–256. in a Nigerian tropical rain forest.Ecological Modelling, 89, 175–181. Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G.A.B.,& Kent, J.2000. Bio- Overpeck, J. T., Rind, D., & Goldberg, R. 1990. diversity hotspots for conservation priorities. Climate-induced changes in forest disturbance Nature, 403, 853–858. and vegetation.Nature, 343, 51–53. Nakashizuka, T., Kohyama, T., Whitmore, T. C., Pacala, S. W., Canham, C. D., Saponara, J., Silan- & Ashton, P.S.1999. Tree diversity and dy- der, J.A.,Kobe, R.K.,& Ribbens, E.1996. namics of western Pacific and eastern Asian Forest models defined by field measurements: forests: An introduction.Journal of Vegeta- Estimation error analysis and dynamics.Eco- tion Science, 10, 765–766. logical Monographs, 66, 1–43. Nepstad, D. C., Verissimo, A., Alencar, A., No- Pastor, J.& Post, W.M.1988.Response of north- bre, C., Lima, E., Lefebvre, P., Schlesinger, ern forests to CO2-induced climate change. P., Potter, C., Moutinho, P., Mendoza, E., Nature, 334, 55–58. Cochrane, M., & Brooks, V. 1999. Large-scale Pelissier, R.& Ri´era, B.1993. Dix ans de dy- impoverishment of Amazonian forests by log- namique d’une foret dense humide de Guyane ging and fire.Nature, 398, 505–508. Francaise.Revue d’Ecologie (Terre Vie), 48, 21–33. Nicotra, A.B.,Chazdon, R.L.,& Iriarte, S.V.B. 1999.Spatial heterogeneity of light and woody Penfold, G.C.& Lamb, D.1999. Species co- seedlings regeneration in tropical wet forests. existence in an Australian subtropical rain Ecology, 80, 1908–1926. forest: evidence for compensatory mortality. Journal of Ecology, 87, 316–329. Oberbauer, S.F.& Strain, B.S.1984. Photosyn- thesis and successional status of Costa Rican Peres, C.1999.Tropical forest disturbance and dy- rain forest trees.Photosynthesis Research, 5, namics in Southeast Asia.Trends in Ecology 227–232. and Evolution, 14, 217–218. Okuda, T., Kachi, N., Yap, S. K., & Manokaran, Phillips, O.L.& Gentry, A.H.1994. Increasing N.1997.Tree distribution pattern and fate of turnover through time in tropical forests.Sci- juveniles in a lowland tropical rain forest - im- ence, 263, 954–958. plications for regeneration and maintenance Phillips, O. L., Malhi, Y., Higuchi, N., Laurance, of species diversity.Plant Ecology, 131, 155– W. F., Nunez, P. V., Vasquez, R. M., Lau- 171. rence, S. G., Ferreira, L. V., Stern, M., Brown, S., & Grace, J. 1998. Changes in the Carbon Ong, R.& Kleine, M.1995.DIPSIM: a dipterocarp Balance of Tropical Forests: Evidence from forest growth simulation model for Sabah. Long-Term Plots.Science, 282, 439–442. Number 2 in FRC Research Papers.Sabah Forestry Department, Sandakan, Malaysia. Phillips, P. D., Yasman, I., & van Gardingen, P. R. 2000.Grouping tree species for analysis of for- Ong, R.C.& Kleine, M.1996. DIPSIM: Dipte- est data in Kalimantan (Indonesian Borneo). rocarp forest growth simulation model - a Forest Ecology and Management, submitted. tool for forest-level management planning. In: Schulte, A.& Sch¨one, D.(editors), Diptero- Pillar, V.D.1999.On the identification of optimal carp Forest Ecosystems - Towards Sustain- plant functional types.Journal of Vegetation able Management.World Scientific, Singa- Science, 10, 631–640. pore, 228–246. Pinard, M. A., Barker, M. G., & Tay, J. 2000. Soil Ong, R. C., Lagan, P. M., Glauner, R., Kleine, M., distribution and post-logging forest recovery & Uebelh¨or, K.1996. Examples of sustain- on bulldozer paths in Sabah, Malaysia.Forest ability criteria for dipterocarp forest manage- Ecology and Management, 130, 213–225. ment. In: Schulte, A.& Sch¨one, D.(editors), Pinard, M.A.& Cropper, W.P.2000. Simulated Dipterocarp Forest Ecosystems: Towards Sus- effects of logging on carbon storage in dipte- tainable Management.World Scientific, Sin- rocarp forest.Journal of Applied Ecology, 37, gapore, 274–292. 267–283. Osho, J.S.A.1995. Optimal sustainable har- Pinard, M. A., Howlett, B., & Davidson, D. 1996. vest models for a Nigerian tropical rain for- Site conditions limit pioneer tree recruitment est.Journal of Environmental Management, after logging of dipterocarp forests in Sabah 45, 101–108. Malaysia.Biotropica, 28, 2–12. Bibliography 159

Pinard, M.A.& Putz, F.E.1996.Retaining forest Powers, J.S.,Haggar, J.P.,& Fisher, R.F.1997. biomass by reducing logging damage.Biotrop- The effect of overstory composition on under- ica, 28, 278–295. story woody regeneration and species richness in 7-year-old plantation in Costa Rica.Forest Pinard, M.A.& Putz, F.E.1997. Moni- Ecology and Management, 99, 43–54. toring carbon sequestration benefits associ- ated with a reduced-impact logging project in Prentice, I.C.& Lloyd, J.1998. C-quest in the Malaysia.Mitigation and Adaptation Strate- Amazon Basin.Nature, 396, 619–610. gies for Global Change, 2, 203–215. Press, M.C.,Brown, N.D.,Barker, M.G.,& Zip- Platt, W.J.& Strong, D.R.1989. Tree fall gaps perlen, S.W.1996.Photosynthetic responses and forest dynamics - Gaps in forest ecology. to light in tropical rain forest tree seedlings. Special feature.Ecology, 70, 535–576. In: Swaine, M.D.(editor), The Ecology of Tropical Tree Seedlings.Parthenon Publish- Plonczak, M.1989. Struktur und Entwicklungs- ing Ltd., Carnforth, 41–54. dynamik eines Naturwaldes unter Konzes- sionsbewirtschaftung in den westlichen Llanos Price, O.F.,Woinarski, J.C.Z.,& Robinson, D. Venezuelas.Number 43 in G¨ottinger Beitr¨age 1999.Very large area requirements for fru- zur Land- und Forstwirtschaft in den Tropen givorous birds in monsoon rainforests of the und Subtropen.Erich Goltze Verlag, G¨ottin- Northern Territory, Australia.Biological Con- gen, Germany. servation, 91, 169–180. Primack, R.B.& Lee, H.S.1991. Population Plumptre, R.A.1996. Links between utilisation dynamics of pioneer (Macaranga) trees and product marketing and forest management in understorey (Mallotus) in selectively logged tropical moist forest.Commonwealth Forestry Bornean rain forests.Journal of Tropical Ecol- Review, 75, 316–324. ogy, 7, 439–458. Poker, J.1993. Struktur und Dynamik des PROSEA 1994.Plant resources South-East Asia 5: Bestandesmosaiks tropischer Regenw¨alder. 1 Timber trees: major commercial timbers; Number 174 in Mitteilungen der Bunde- 2 Timber trees: minor commercial timbers. sanstalt f¨ur Forst- und Holzwirtschaft.Kom- Bogor, Indonesia. missionsverlag Max Wiedebusch, Hamburg, Germany. Puig, H., Ri´era, B.,& Lescure, J.P.1990. Phy- tomasse et productivit´eenforˆet guyanaise. Poker, J.1995. Vegetation Pattern Structure and Bois et Forˆets des Tropiques, 220, 25–32. Dynamics in Different Types of Humid Trop- ical Evergreen Forests of Southeast Asia. In: Pulkki, R.E.1997. Literature synthesis on Box, E.O.(editor), Vegetation Science in logging impacts in moist tropical forests Forestry.Kluwer, 557–571. (Global fibre supply study).Technical Report GFSS/WP/06, FAO, Rom, Italy. Poncy, O., Ri´era, B., Larpin, D., Belbenoˆıt, P., Julien, M., Hoff, M., & Charles-Dominique, Putz, F.E.1979. Aseasonality in Malaysian tree P.1998. The permanent field research sta- phenology.The Malaysian Forestier, 42,1– tion les Nouragues in the tropical rainforest of 24. French Guiana: current projects and prelim- Putz, F.E.& Milton, K.1982.Tree mortality rates inary results on tree diversity, structure and on Barro Colorado Island. In: jr.Leigh, E.G., dynamics. In: Dallmeier, F.& Comiskey, J. Rand, A.S.,& Windsor, D.M.(editors), The (editors), Forest biodiversity in North, Cen- ecology of a tropical forest.Smithsonian Insti- tral and South America and the Carraibean: tution Press, Washington, DC, 95–100. research and monitoring.UNESCO, Paris and Parthenon Publishing Group, Lancs, UK, Putz, F.E.& Pinard, M.A.1993.Reduced-impact 398–414. logging as a carbon-offset method.Conserva- tion Biology, 7, 755–757. Poorter, L., Bongers, F., van Rompaey, R. S. A. R., Putz, F.E.& Viana, V.1996.Biological challenges & de Klerk, M.1996.Regeneration of canopy for certification of tropical timber.Biotropica, tree species at five sites in West African moist 28, 323–330. forest.Forest Ecology and Management, 84, 61–69. Redford, K.H.1992.The empty forest.BioScience, 42, 412–422. Pounds, J.A.,Fodgen, M.P.L.,& Campbell, J.H. 1999.Biological response to climate change on Reid, W.V.1998.Biodiversity hotspots.Trends in a tropical mountain.Nature, 398, 611–615. Ecology and Evolution, 13, 275–280. 160 Bibliography

Reinhard, K.1999.Modellierung von Biodiversit¨at Ryan, M. G., Binkley, D., & Fownes, J. H. 1997. in tropischen Regenw¨aldern.Diploma thesis, Age-related decline in forest productivity pat- University of Kassel, Germany. tern and process.Advances in Ecological Re- search, 27, 214–262. Ribbens, E.,Silander, J.A.,& Pacala, S.W.1994. Seedling recruitment in forests: calibrating Sabah-Forestry-Department 1994.Forest Manage- models to predict patterns of tree seedling dis- ment Plan, Forest Management Unit No.19. persion.Ecology, 75, 1794–1806. Report, Sabah Forestry Department, San- dakan, Sabah, Malaysia. Rice, R. E., Gullison, R. E., & Reid, J. W. 1997. Can sustainable management save tropical Sachs, L.1997.Angewandte Statistik.Springer, forests.Scientific American, 4, 34–39. Berlin, Heidelberg, New York. Saldarriaga, J.G.1986. Recovery following shift- Richards, P.W.1936. Ecological observations on ing cultivation.Case study no.2: A century the rain forest of Mount Dulit Sarawak Part I of succession in the upper Rio Negro. In: Jor- and II.Journal of Ecology, 24, 1–37, 340–360. dan, C.F.(editor), Amazonian rain forests - Richards, P.W.1952. The Tropical Rain Forest ecosystem disturbance and recovery.Springer- - an Ecological Study.Cambridge University Verlag, New York, Berlin, Heidelberg, Lon- Press, Cambridge. don, Paris, Tokyo, 24–33. Richards, P.W.1996.The tropical rain forest - an Saldarriaga, J.G.,West, D.C.,Tharp, M.L.,& ecological study (second edition).Cambridge Uhl, C.1988. Long-term chronosequence of Unversity Press, Cambridge. forest succession in the upper Rio Negro of Colombia and Venezuela.Journal of Ecology, Ri´era, B.1985.Importance des buttes de deracine- 76, 938–958. ment dans la regeneration forestiere en guyane francaise.Rev.Ecol.(Terre Vie), 40, 321–329. Saunders, D. A., Hobbs, R. J., & Margules, C. R. 1991.Biological consequences of ecosystem Riswan, S.& Hartanti, L.1995. Human impacts fragmentation: a review.Conservation Biol- on tropical forest dynamics.Vegetatio, 121, ogy, 5, 18–32. 41–52. Schlensog, M.1997. Experimentelle Untersuchun- Robinson, J. G., Redford, K. H., & Bennett, E. L. gen des Lichtklimas in Urwaldparzellen Nord- 1999.Wildlife harvest in logged tropical borneos.Number 117 in G¨ ottinger Beitr¨age forests.Science, 284, 595–596. zur Land- und Forstwirtschaft in den Tropen und Subtropen.Goltze, G¨ottingen. Roggy, J.C.,Pr´ evost, M.-F., Garbaye, J., & Dom- enach, A.M.1999. Nitrogen cycling in the Schupp, E., Howe, H., Augspurger, C. K., & Levey, tropical rain forest of French Guiana: com- D.J.1989. Arrival and survival in tropical parison of two sites with contrasting soil types treefall gaps.Ecology, 70, 562–564. 15 using δ N.Journal of Tropical Ecology, 15, Sheil, D.& May, R.M.1996.Mortality and recruit- 1–22. ment rate evaluation in heterogeneous tropical forests.Journal of Ecology, 84, 91–100. Rollet, B.1978.Organization. In: Unesco, UNEP, & FAO (editors), Tropical forest ecosystems Sheil, D., Sayer, J. A., & O’Brien, T. 1999. Tree - A state-of-knowledge report.UNESCO - species diversity in logged rainforests.Science, UNEP, Paris, 112–143. 284, 1587i. Rosenzweig, M.L.1999. Heeding the warning in Shugart, H.H.1984.A Theory of Forest Dynamics. biodiversity’s basic law.Science, 284, 276– Springer, New York. 277. Shugart, H.H.1997. Plant and ecosystem func- Ruhiyat, D.1989. Die Entwicklung der tional types. In: Smith, T. M., Shugart, H. H., stand¨ortlichen N¨ahrstoffvorr¨ate bei naturna- & Woodward, F.I.(editors), Plant Functional her Waldbewirtschaftung und im Plantagen- Types.Cambridge University Press, Cam- betrieb, Ostkalimantan, Indonesien.Num- bridge, 20–45. ber 35 in G¨ottinger Beitr¨age zur Land- und Shugart, H.H.1998. Terrestrial Ecosystems in Forstwirtschaft in den Tropen und Subtropen. Changing Environments.Cambridge Univer- Goltze, G¨ottingen, Germany. sity Press, Cambridge, U.K. Runkle, J.R.1989.Synchrony of regeneration gaps Shugart, H. H., Hopkins, M. S., Burgess, I. P., & and latitudinal differences in tree species di- Mortlock, A.T.1980. The development of versity.Ecology, 70, 546–547. a succession model for subtropical rain forest Bibliography 161

and its application to assess the effects of tim- Swaine, M.D.1989. Population dynamics of tree ber harvest at Wiangarree State Forest New species in tropical forests. In: Holm-Nielsen, South Wales.Journal of Environmental Man- L.B.,Nielsen, I.B.,& Balslev, H.(editors), agement, 11, 234–256. Tropical Forests: Botanical Dynamics, Speci- Silva, J.M.N.,Carvalho, J.O.P.,de Lopes, J. ation and Diversity.Academic Press, London, C. A., de Almeida, B. F., Costa, D. H. M., 101–110. de Oliviera, L.C.,Vanclay, J.K.,& Skovs- Swaine, M.D.,Hall, J.B.,& Alexander, I.J. gaard, J.P.1995.Growth and yield of a trop- 1987a.Tree population dynamics at Kade ical rain forest in the Brazilian Amazon 13 Ghana (1982-1986).Journal of Tropical Ecol- years after logging.Forest Ecology and Man- ogy, 3, 331–345. agement, 71, 267–274. Swaine, M.D.,Lieberman, D.,& Putz, F.E.1987b. Silva, J.N.M.,de Carvalho, J.O.P.,Lopes, J. The dynamics of tree populations in tropical C.A.,de Oliveira, R.P.,& de Oliveira, L.C. forest: a review.Journal of Tropical Ecology, 1996.Growth and yield studies in the Tapajos 3, 359–366. region Central Brazilian Amazon.Common- Swaine, M.D.& Whitmore, T.C.1988.On the def- wealth Forestry Review, 75, 325–329. inition of ecological species groups in tropical Silva-Matos, D.M.,Freckleton, R.P.,& Watkin- rain forests.Vegetatio, 75, 81–86. son, A.R.1999. The role of density depen- dence in the population dynamics of a tropical Tabarelli, M., Mantovani, W., & Peres, C. A. 1999. palm.Ecology, 80, 2635–2650. Effects of habitat fragementation on plant guild structure in the montane Atlantic forest Sisk, T. D., Launer, A. E., Switky, K. R., & Ehrlich, of southeastern Brazil.Biological Conserva- P.R.1994. Identifying extinction threats: tion, 91, 119–127. global analyses of the distribution of biodiver- sity and the expansion of human enterprise. ter Steege, H., Boot, R., Brouwer, L., Hammond, BioScience, 44, 592–604. D., van der Hout, P., Jetten, V. G., Khan, Z., Polak, A. M., Raaimakers, D., & Zagt, R. Sist, P., Nolan, T., Bertault, J., & Dykstra, D. 1995.Basic and applied research for sound 1998.Harvesting intensity versus sustainabil- rain forest management in Guyana.Ecological ity in Indonesia.Forest Ecology and Manage- Applications, 5, 904–910. ment, 108, 251–260. Terborgh, J.1993. Lebensraum Regenwald. Spek- Smith, T. M., Shugart, H. H., & Woodward, F. I. trum Akademischer Verlag, Heidelberg. (editors) 1997.Plant Functional Types.Num- ber 1 in International Geosphere-Biosphere Terborgh, J.& Petren, K.1991. Development Programme Book Series.Cambridge Univer- of habitat structure through succession in an sity Press, Cambridge. Amazonian flooplain forest. In: Bell, S.S., McCoy, E. D., & R., M. H. (editors), Habi- Smith, T.M.,Shugart, H.H.,Woodward, F.I.,& tat Structures: The physical arrangement of Burton, P.J.1993.Plant functional types. In: objects in space.Chapman and Hall, London, Solomon, A.M.& H.,S.H.(editors), Vege- 28–46. tation dynamics and global change.Chapman and Hall, New York, 272–292. Thiollay, J.-M. 1989. Area requirements for the con- servation of rain forest raptors and game birds Smith, T.M.& Urban, T.L.1988.Scale and reso- in French Guiana.Conservation Biology, 3, lution of forest structural pattern.Vegetatio, 128–137. 74, 143–150. Thiollay, J.-M. 1992. Influence of selective logging Smithsonian-Tropical-Research-Institute 2000. on bird species diversity in a Guianan rain for- http:// www.si.edu/ organiza/ centers/ stri/ est.Conservation Biology, 6, 47–63. forest/ ctfs.htm. www-pages. Thomas, S.C.& Bazzaz, F.A.1999. Asymptotic Stil, C.J.,Foster, P.N.,& Schneider, S.H.1999. height as predictor of photosynthetic charac- Simulating the effects of climate change in teristics in Malaysian rain forest trees.Ecol- tropical montane cloud forests.Nature, 398, ogy, 80, 1607–1622. 608–610. Thornley, J.H.M.1976. Mathematical Models in Strauss-Debenedetti, S.& Bazzaz, F.A.1996.Pho- Plant Physiology.Academic Press, New York. tosynthetic characteristics of tropical trees along successional gradients. In: Mulkey, Thornley, J.H.M.& Cannell, M.G.R.2000.Man- S. S., Chazdon, R. L., & Smith, A. P. (ed- aging forest for wood yield and carbon stor- itors), Tropical Forest Plant Ecophysiology. age: a theoretical study.Tree Physiology, 20, Chapman and Hall, 162–186. 477–484. 162 Bibliography

Tian, H., Melillo, J. M., Kicklighter, D. W., van der Meer, P. J., Bongers, F., Chatrou, L., & McGuire, A. D., Helfrich, J. V. K., Moore, Ri´era, B.1994. Defining canopy gaps in a B., & V¨or¨osmarty, C.J.1998.Effect of inter- tropical rain forest: effects on gap size and annual climate variability on carbon storage turnover time.Acta Oecologica, 15, 701–714. in Amazonian ecosystems.Nature, 396, 664– van Gardingen, P.R.,Clearwater, M.J.,Nifinluri, 667. T., Effendi, R., Rusmantoro, W., Noor, M., Tilman, D.1999. Diversity by default. Science, Mason, P. A., Ingleby, K., & Munro, R. C. 283, 495–496. 1998.Impacts of logging on the regenera- tion of lowland dipterocarp forest in Indone- Tucker, N.I.J.& Murphy, T.M.1997. The ef- sia.Commonwealth Forestry Review, 77, 71– fects of ecological rehabilitation on vegetation 82. recruitment: some observations from the Wet Tropics of North Queensland.Forest Ecology van Gardingen, P.R.& Phillips, P.D.1999.Appli- and Management, 99, 133–152. cation of growth and yield modelling as an in- dicator of sustainable forest management. In: Tuomisto, H., Ruokolainen, K., Kalliola, R., Linna, Laumonier, Y., King, B., Legg, C., & Ren- A., Danjoy, W., & Rodriguez, Z. 1995. Dis- nolls, K.(editors), International conference secting Amazonian Biodiversity.Science, 269, on data management and modelling using re- 63–66. mote sensing and GIS for tropical forest inven- Turner, I.M.1996. Species loss in fragments of tory, Jakarta, Indonesia, 26-29 October 1998. tropical rain forest: a review of the evidence. Rodeo International Publishers, Jakarta, 491– Journal of Applied Ecology, 33, 200–209. 502. Turner, I.M.& Corlett, R.T.1996.The conserva- van Roosmalen, M.G.M.1985. Fruits of the tion value of small isolated fragments of low- Guiana Flora.Institue of Systematics and land tropical rain forest.Trends in Ecology Botany, Utrecht, the Netherlands. and Evolution, 11, 330–333. Vanclay, J.K.1989. A growth model for North Uchma´nski, J.& Grimm, V.1996.Individual-based Queensland rainforests.Forest Ecology and modelling in ecology: What makes the differ- Management, 27, 245–271. ence? Trends in Ecology and Evolution, 11, Vanclay, J.K.1991.Aggregating tree species to de- 437–440. velop diameter increment equations for trop- Uhl, C., Buschbacher, R., & Serrao, E. A. S. 1988. ical rain forest.Forest Ecology and Manage- Abandoned pastures in eastern Amazonia.I: ment, 42, 143–168. Patterns of plant succession.Journal of Ecol- Vanclay, J.K.1994. Modelling Forest Growth ogy, 76, 663–681. and Yield: Applications to Mixed Tropical Urban, D.L.,Bonan, G.B.,Smith, T.M.,& Forests.CAB International, Wallingford. Shugart, H.H.1991. Spatial applications of Vanclay, J.K.1995. Growth models for tropical gap models.Forest Ecology and Management, forests: a synthesis of models and methods. 42, 95–110. Forest Science, 41, 7–42. Valencia, R.,Balslev, H.,& no, C.G.P.Y.M. Vanclay, J.K.& Skovsgaard, J.P.1997.Evaluating 1994.High tree alpha-diversity in Amazonian forest growth models.Ecological Modelling, Ecuador.Biodiversity and Conservation, 3, 98, 1–12. 21–28. Vandermeer, J.1994.Disturbance and neutral com- Valentine, H.T.1988. A carbon-balance model of petition theory in rain forest dynamics.Eco- stand growth: A derivation employing pipe- logical Modelling, 85, 99–111. model theory and self thinning rule.Annals of Botany (London), 68, 389–396. Veblen, T.T.1989.Tree regeneration responses to gaps along a transandean gradient.Ecology, Valentine, H.T.1990. A carbon-balance model of 70, 541–543. tree growth with a pipe-model framework. In: Dixon, R.K.,Meldahl, R.S.,Ruark, G.A., Veillon, J.P.1985. El crecimiento de algunas & Wareen, W.G.(editors), Process-Modeling bosques naturales de Venezuela en relaci´on of Forest Growth Responses to Environmental con los par´ametros del medio ambiente.Re- Stress.Timber Press, Oregon, 33–40. vista Forestal Venezolana, 29, 5–121. van der Meer, P.J.& Bongers, F.1996. Patterns Veillon, J.P.1989. Los bosques naturales de of tree-fall and branch-fall in a tropical rain Venezuela.Parte I: El medio ambiente.Insti- forest in French Guiana.Journal of Ecology, tuto de Silvicultura.Universidad de Los An- 84, 19–29. des, M´erida, Venezuela. Bibliography 163

Walsh, R.P.D.1996.Drought frequency changes in Whitmore, T.C.& Ng, F.S.1972. Tree flora of Sabah and adjacent parts of northern Borneo Malaya: a manual for foresters (1-4).Num- since the late nineteenth century and possible ber 26 in Malayan Forest Records.Forest Re- implications for tropical rain forest dynamics. search Institute, Kepong, London. Journal of Tropical Ecology, 12, 385–407. Whitmore, T.C.& Sayer, J.A.(editors) 1992. Warner, R.R.& Chesson, P.L.1985.Coexistence Tropical Deforestation and Species Extinc- mediated by recruitment fluctuations: a field tion.Chapman and Hall, London, U.K. guide to the storage effect.American Natu- ralist, 125, 769–787. Wiens, J.A.1999. Landscape ecology: Scaling from mechanims to management. In: Farina, Webb, C.O.& Peart, D.R.1999.Seedling density A.(editor), Perspectives in Ecology.A glance dependence promotes coexistence of Bornean from the VII International Congress of Ecol- rain forest trees.Ecology, 80, 2006–2017. ogy.Backhuys Publisher, Leiden, 13–24. Weidelt, H.J.1993.Some effects of selective logging Wirtz, K.W.1998. Modellierung von Anpas- on forest productivity and ecology. In: Lieth, sungsvorg¨angen in der Natur.PhD thesis, H.& Lohmann, M.(editors), Restoration of University of Kassel, Germany. Tropical Forest Ecosystems.Kluwer Academic Publishers, Dordrecht, 155–161. Woodroffe, R.& Ginsberg, J.R.1998. Edge ef- fects and the extinction of populations inside Weidelt, H.-J. 1996. Sustainable management of protected areas.Science, 280, 2126–2128. dipterocarp forests - opportunities and con- straints. In: Schulte, A.& Sch¨one, D.(ed- Wunderle, J.M.J.1997. The role of animal seed itors), Dipterocarp Forest Ecosystems: To- dispersal in accelerating native forest regener- wards Sustainable Management.World Scien- ation on degraded tropical land.Forest Ecol- tific, Singapore, 249–273. ogy and Management, 99, 223–235. Weiher, E., van der Werf, A., Thompson, K., Rod- WWF 1997.Forest for life.Global annual forest erick, M., Garnier, E., & Eriksson, O. 1999. report 97.Technical report, WWF - World Challenging Theophrastus: A common core Wide Fund for Nature and World Conserva- list of plant traits for functional ecology.Jour- tion Monitoring Centre, Surrey. nal of Vegetation Science, 10, 609–620. Xiong, S.& Nilsson, C.1999. The effects of plant Westoby, M.1987.The self-thinning rule.Advances litter on vegetation: a meta-analysis.Journal in Ecological Research, 14, 167–223. of Ecology, 87, 984–994. White, J.1981. The allometric interpretation of Yamakura, T., Hagihara, A., Sukardjo, S., & the self-thinnning rule.Journal of Theoretical Ogawa, H.1986. Aboveground biomass of Biology, 89, 475–500. tropical rain forest stands in Indonesian Bor- Whitmore, T.C.1983. Secondary succession from neo.Vegetatio, 68, 71–82. seed in tropical rain forests.Forestry Ab- Yoda, K.1983. Community respiration in a low- stracts, 44, 767–778. land rain forest in Pasoh Peninsular Malaysia. Whitmore, T.C.1984. Tropical Rain Forests of Japanese Journal of Ecology, 33, 183–197. the Far East.Oxford Science Publications, Young, A.C.& Muezelfeldt, R.1998. The SYM- Oxford. FOR tropical forest modelling framework. Whitmore, T.C.1988. The influence of tree pop- Commonwealth Forestry Review, 77, 19–23. ulation dynamics on forest species composi- Zipperlen, S.W.& Press, M.C.1997. Photosyn- tion. In: Davis,A.J.,Hutchings,M.J.,& thesis in relation to growth and seedling ecol- Watkinson, A.R.(editors), Population Biol- ogy of two dipterocarp rain forest tree species. ogy of Plants.Blackwell, Oxford, 271–291. Journal of Ecology, 84, 863–876. Whitmore, T.C.1989. Canopy gaps and the two major groups of forest trees.Ecology, 70, 536– 538. Whitmore, T.C.1995.Perspectives in tropical rain forest research. In: Lugo, A.E.& Lowe, C. (editors), Tropical Forests: Management and Ecology.Springer, New York, 397–407. Whitmore, T.C.1998.An introduction to tropical rain forests (2nd edition).Oxford University Press, Oxford.

Appendix A

Inventory data

Typical inventory data of field measurements are distributions of stem numbers over diameter, total and for single species or species groups. Most inventories used in the applications discussed in this thesis are compliled in the following Appendix. In detail they are:

• Sabah:

– Chapter 5: Data of permanent sampling plots in Garinono, Gunung Rara, Se- galiud Lokan, and Sepilok are taken from forest management inventories (Chai et al. 1991; Kilou et al. 1993; Ong, unpublished data), consist of 25 hectares and are therefore not plotted here. – Chapter 7: mature forest (P1, Table A.1) and recently logged over forest (L1, Table A.2) in Deramakot Forest Reserve, Sabah. They were complied from the data sets A1-A4 (P1) and D5-D8 (L1) in the appendix of Schlensog (1997).

• Venezuela: mature forest (MF, Table A.3) and logged over forest 5 years prior (LG5, Table A.4) in Caparo Forest Reserve, Venezuela, taken from work of Kammesheidt (Kammesheidt 1994, 1998).

• French Guiana: mature forest (Table A.5) at Piste de Saint Elie, French Guiana (B. Riera, unpublished data). 166 Appendix A

Table A.1: Inventory data for site P1 in Deramakot, Sabah.Stem Number [ha −1]indiameter classes D, total and per plant functional type PFT.(A1-A4 in Schlensog 1997).

D [cm] Total PFT 12345678910111213

0-10 4149 0 0 600 800 0 200 0 0 0 0 0 2082 467 10-20 322 0 0 33 33 0 66 0 0 22 0 0 89 79 20-30 110 0 0 0 0 0 22 0 0 0 11 0 55 22 30-40 53 0 0 1 1 4 12 0 0 5 3 0 25 2 40-50 28 1 0 0 0 0 3 1 0 5 5 0 13 0 50-60 16 0 0 4 1 0 0 0 0 1 0 0 10 0 60-70 5 0 0 0 0 0 0 0 0 1 1 0 3 0 70-80 5 0 0 0 0 0 0 0 0 0 0 0 5 0 80-90 1 0 0 0 0 0 0 0 0 0 0 0 1 0 90-100 3 0 0 0 0 0 0 0 0 1 0 0 2 0 100-110 2 0 0 0 0 0 0 0 0 1 0 0 0 1 110-120 1 0 0 0 0 0 0 0 0 0 0 0 0 1 120-130 1 0 0 0 0 0 0 0 0 0 0 0 0 1 130-140 1 0 0 0 0 0 0 0 0 0 0 0 0 1 140-150 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Total 4697 1 0 638 835 4 303 1 0 36 20 0 2285 574 Inventory data 167

Table A.2: Inventory data for site L1 in Deramakot, Sabah.Stem Number [ha −1]indiameter classes D, total and per plant functional type PFT.(D5-D8 in Schlensog 1997)

D [cm] Total PFT 1 2 3 4 5 6 7 8 9 10 11 12 13

0-10 3600 333 67 0 267 0 0 0 67 2333 0 0 533 0 10-20 88 0 11 0 0 66 0 0 11 0 0 0 0 0 20-30 78 0 0 0 0 78 0 0 0 0 0 0 0 0 30-40 33 1 0 1 1 10 3 0 0 3 3 1 10 0 40-50 22 0 0 2 0 3 1 2 0 5 1 0 8 0 50-60 5 0 0 0 1 1 0 0 0 0 1 0 2 0 60-70 6 0 0 0 0 0 1 0 0 2 0 0 3 0 70-80 3 0 0 0 0 0 0 0 0 1 0 0 2 0 80-90 3 1 0 0 0 0 0 0 0 0 1 0 1 0 90-100 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100-110 0 0 0 0 0 0 0 0 0 0 0 0 0 0 110-120 0 0 0 0 0 0 0 0 0 0 0 0 0 0 120-130 1 0 0 0 0 0 0 0 0 0 0 0 1 0 130-140 0 0 0 0 0 0 0 0 0 0 0 0 0 0 140-150 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Total 3840 335 78 3 269 158 5 2 78 2344 6 1 561 0 168 Appendix A

Table A.3: Inventory data for site MF in Caparo, Venezuela.Stem Number [ha −1]indiameter classes D, total and per plant functional type PFT.(PW or MF in Kammesheidt 1994, 1998).

D [cm] Total PFT 1 2 3 4 5 67 8 9101112

0-10 2400 176644 0 80 136 4 124 256 36160476308 10-20 161 0 0 0 15 0 1 15 35 15 21 0 59 20-30 102 0 0 0 0 1 3 7 23 4 17 0 47 30-40 760 0 0 0 0 0 2 614 60 48 40-50 22 0 0 0 0 0 0 0 3 3 5 0 11 50-60160000000286 00 60-709000000016 200 70-808000000006200 80-906000000002400 90-100 1 0 0 0 0 0 0 0 0 1 0 0 0 100-110 0 0 0 0 0 0 0 0 0 0 0 0 0 110-120 0 0 0 0 0 0 0 0 0 0 0 0 0 120-130 1 0 0 0 0 0 0 0 0 1 0 0 0 130-140 0 0 0 0 0 0 0 0 0 0 0 0 0 140-150 0 0 0 0 0 0 0 0 0 0 0 0 0

Total 2802 176644 0 95 137 8 148 326 96223 476473 Inventory data 169

Table A.4: Inventory data for site LG5 in Caparo, Venezuela.Stem Number [ha −1]indiameter classes D, total and per plant functional type PFT.(EW87 or LG5 in Kammesheidt 1994, 1998).

D[cm] Total PFT 123456789101112

0-10 3568 1304 272 112 288 360 136 176 188 152 48 248 284 10-20 220 0 0 11 9 1643 21 31 21 17 0 51 20-30 74 0 0 0 1 1 611 16 4 4 0 31 30-40490000004345033 40-50220000001105015 50-606 000000021201 60-701 000000000100 70-801 000000001000 80-900 000000000000 90-1000 000000000000 100-110 0 0 0 0 0 0 0 0 0 0 0 0 0 110-120 0 0 0 0 0 0 0 0 0 0 0 0 0 120-130 0 0 0 0 0 0 0 0 0 0 0 0 0 130-140 0 0 0 0 0 0 0 0 0 0 0 0 0 140-150 0 0 0 0 0 0 0 0 0 0 0 0 0

Total 3941 1304 272 123 298 377 185 213 241 183 82 248 415 170 Appendix A

Table A.5: Inventory data for 19 PFTs at site Piste de Saint Elie, French Guiana.Stem Number [ha−1] in diameter classes D, total and per plant functional type PFT.(B.Riera, unpublished data).

D[cm] Total PFT 123 456 7 8910111213141516171819

0-10 20000001000100000000 10-20322014154448310538640 829481202 20-30104011611 6 50 2 7230 212280 7 2 30-40 56000210 3 10 2 3 9 0 113170 3 1 40-50 29000100 1 00 1 2 2 0 1117 0 3 0 50-60 14000100 0 10 0 1 1 0 0 7 1 0 2 0 60-70 10000010 0 00 0 1 0 0 0 3 2 0 2 1 70-8040000000000000031000 80-9020000000000000011000 90-10010000001000000000000 100-110 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 110-120 1 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 120-130 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 130-140 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 140-150 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Total546025257561380105399012791051386 Appendix B

Lists of tree species

Model application in each site is based on a species grouping following the concept dis- cussed in Chapter 3. All available and relevant information on single tree species is found in the following three species list. Informations differ for each site, so labelling of columns is explained in legends before each species list. The species list for Sabah (Table B.2) is based on a compilation by Ong & Kleine (1995), updated with own informations (K¨ohler 1998). Venezuela’s species list (Table B.4) is based on work of Kammesheidt (1994, 2000), the one of French Guiana’s rain forest (Table B.6) was compiled by Chave (1999; Chave and Ri´era, unpublished manuscript), use of both with kind permission of the authors. A compressed version of species lists is available online (http://www.usf.uni- kassel.de/usf/archiv/dokumente.en.htm). 172 Appendix B

Table B.1: Legend of tree species lists of Sabah (Table B.2).

Column Description

Botanical name Genus and species name Local name Local name PFT Plant functional type according to Table 7.1 SS Successional status according to Table 7.1 HG Height group according to Table 7.1 CFRC Species code for identification of Forest Research Center, Sepilok CHQ Species code for identification of Head Quarter, Sandakan

TGRP Timber group after Ong & Kleine (1995) IGRP Diameter ingrowth group after Ong & Kleine (1995) F3 Plant functional type used by Formix3 GGRP Glauner group out of Canadian inventory, used for height calcula- tions Height Maximum height at maturity [m] out of literature. Special symbols (s: small; m: middle; l: large; b: bush; t: tree). Dia Maximum diameter at maturity [cm] out of literature H=f(d) Maximum height calculated out of Diameter −3 Den wood density [kgODM m ] SPHQ Number of species per C-HQ NDer Number of trees (diameter≥10cm) in Deramakot inventory NFMU Number of trees (diameter≥10cm) in inventories in Deramakot, Lingkabau, Kalabakan and Ulu Segama F Qualitative abundance according to literature (1: rare; 2: medium; 3: common) Lists of tree species 173 ohler 1998). Tree species list of Sabah (modified from Ong & Kleine 1995; K¨ Table B.2: Botanical nameAcronychia sp.Actinodaphne glomerataAdenanthera pavoniaAdina trichotomaAdinandra dumosa Local nameAfzelia Medang borneensis seraiAgathis dammaraAglaia argentea Saga LimauAglaia hutan cordataAglaia elliptica MengkeniabAglaia sp. BawingAilanthus integrifolia 6 Ipil PFTAlangium darat sp. SS MengilanAlbizzia sp. HG 2Aleurites CFRC moluccana 6 Koping-koping CHQAlphitonia 3 incana TGRPAlstonia Kalambio macrophylla IGRP 2 6 F3 ACGL Lantupak 9 TreeAlstonia jambu of sp. MDS heaven GGRPAmoora 3 NDLH rubiginosa Height 2 5 2 19Anacardiaceae Dia family 7 ACRO Kamiri Langsat-langsatAngelesia 3 6 splendens OTH 4 12 Kondolon 1 1 H=f(d)Anisophyllea OTHR disticha Pulai Den 3 14 daun ADTR PakuditaAnisoptera besar 3 2 SAGA 2 Batai costata MGB 9 3 OTH NDLHAnisoptera SPHQ grossivenia 3 6 1 OTHR 14 4 5 3 NDer 2Anisoptera BAWI 14 laevis Lantupak Rengas paya BW IDRTAnisoptera 3 15 NFMU 18 marginata AGDA 2 1 AGAR 3 2 F MGL NDLH IPD 1Anisoptera Tampaluan KOP Pulai reticulata Payung-payung NDLH 16 8 NDLHAnisoptera 2 3 NDMH 15 2 15 AGEL sp. 19 14 - - 15 PengiranAnonaceae OTH 3 kunyit family 2 Pengiran 1 2 AIIN kesat OTHR 1 AGCOAnonaceae 1 1 family 1 14 OTH 20 6Anthocephalus 30.5 15 9 TOH 2 OTHR Pengiran chinensis AGOD 4 2 kerangas 44.6 - 14 14 LLS OTHR 15 15Anthoshorea Pengiran 1 group durian 3 Pengiran 2 14 of 4 - gajah 2 - Shorea NDLHAntidesma ALMA 2 1 20.75 ghasemblica 20 1 OTH 14 15 4 12Antidesma 48.8 3 NDLH sp. 12 25 30 1 KOND 2 - - 3 4 KMRI Melapi KON 13Apocynaneae Laran 15 12 family 1 2 KMR NDLH 12 - 165.5 2 PAKU - Karai - 3Aporusa NDLH 2 15 Pengiran 20 2 14 AMRU grandistipulata 12 PAK 76.4 35 14 3 OTH 2 Pisang-pisangAporusa - 15 5 OTHR 2 nitida 12 5 OTHR 89 2 Tandoropis 10 2 BTAI 14 ANDI 2Aporusa 14 1 12 sp. 8 26.06 5 24 1 4.8 960 BTI RENG 897.1 OTH - 5 2Aquilaria PKUN 18.26 TAMP malaccensis RGS s OTHR 5 1 880 OTH 15 PY 2 NDLH 245 1 9.5Archidendron 465 1 PKST 14 15 NDMH Galang-galang 853 OTHR 0 PKER 15 5 89 19 7.64 Jelutong PS 0Ardisia 14 DLH PK PULA 30 15 sp. 5 15 1 999 610 1 2... PUL 9.59 12.2 Gerusih PDUR 1 1 22.3 - 1 DLH 2 1.2 DLH 1 8 PD NDLH 0 PGAJ 245 66.5 13 15 66.8 12 m - 1 - - 38.2 2 PJ 35 15 2 12 14.3 8 8 DLH Gaharu 0 15 3 - 1 1 2 Bagil - 999 8 2 0 38.23 3 89 2 DLH 19.14 2 0 800 7 4 1 1 - 1 - s-m 4 2 2 - 3350 5 752.9 89 370 15 2 - 1 - 8 Penatan 5 0 14.5 - 1 2 245 4 1 2 3 0 1 PGPG 2 Patai 5482 - keryong MELA - 4 PIS PENG - 11 1 MP 1 - 245 - 45 - 2 LARA PG 2 89 2 3 ANGH 999 - - NDMH 2 LRN PGPG OTH DLH 1 14 4 Serusop 300 60 309.9 30-50 2 3 DLH 329 KRY NDLH PION 999 - APGR NDLH - 14 - 2 0 - - GLG 2 245 5 13 - - 1 2 928 14 60 3 1004 OTHR 6 945 76.4 - 143.2 - 14 1 l JELU 3 8 999 1 - 2 1 480 89 4 - 9 1 40.31 0 JLT - 2 - 1 - 14.5 - 1 15 1 - 6 14 2 NDLH ANTI 8 432.5 3 801 2 1 3 2 3 15 116.5 245 - OTH 15 - 1 2 624 OTHR - 6 729 - 3 89 - 2 4 1 640 GAHA 14 0 1 89 50-60 GH 30 999 3 4 l 37 s 1 APOL PATA 1 95.5 - - 1 8 1 NDLH 632 245 PATA BGL OTHR 28.6 APEL - 12 245 600 2263 47.7 3 29 14 57.3 OTHR PTN 36.51 95 15 14 1714 - - 8 16.37 OTHR 999 7649 4 0 2 35.49 999 - 1.5-30 - 35.39 1 0 14 - 1 - 672 2 1 - 1 416 - 6 600 15 3 ARDI 15 1 3 12 - - OTH 15 89 0 0 0 1 1 - 1 OTHR 1 15 14 36 - - - 2 7 245 - - - 1 0 1 1156 0 20 3410 1 233 57.3 440 999 - 0 15 2440 0 - 23.43 1003 - 1 1 - 0 3 2852 1 3 - 3 8 - 600 3 18335 - 8 0 - - - 8 89 1 31.8 - - 119 200 17.34 - 245 260 112 - 1 - - 999 3 1 507 1 - 89 0 3 300 479 245 300 0 2366 999 - - - - 174 Appendix B Botanical name... Aromadendron sp.Artocarpus anisophyllusArtocarpus elasticusArtocarpus sp. Local nameArtocarpus sp.Artocarpus Terap tamaran ikalAzadirachta excelsa Kepayang ambokBaccaurea angulata TerapBaccaurea togop lanceolataBaccaurea sp.Baccaurea sp. Timbangan PFT TerapBarringtonia sp. SS 6 TerapBerrya Limpaga timadang cordifolia HG 6Bischofia Belimbing CFRC javanica 2 hutan Limpaung CHQBlumeodendron tokbrai TGRP 5 2 3Borneodendron IGRP enigmaticum F3Breynia patens 3 1 ARNU GGRPBridelia KAP glauca 3 Kunau-kunau Height Bangkau-bangkau 6 DMHBridelia Dia ARAN 4 3 stipularis Tampoi Tampalang Gangulang TRIBrownlowia 14 Mengkapang peltata 2 darat 2 Tungou OTHR AREL H=f(d)Bruinsmia 12 5 3 stracoides 14 Den TRO 1 2 4Buchanania sp. 3 NDLH 2 2 6 1Buchanania 15 SPHQ 1 sp. 15 AROD 3 ARTA NDerBurseraceae 5 TRT family 4 BAAN 4 TIMD 3 2 NDLH 15 Kubamban-kubamban 1 NDLHCallophyllum BBH NFMU sp. 15 2 15 F LIMP Balatotan OTHR Manik-manik/kutang 20Calophyllum Pingau-pingau TRAP inophyllum 3 3 3 14 3 15 Tingo-tingo 6 LM TRPCampnosperma BALA 30 auriculta 1 1 4 4 NDLH OTH 2Cananga BKAU NDLH 2 1 odorata 2 - 15 2 OTHR BB 17 4 45 15Canarium 9 15 14 decumanum 3 BACC 3 57.3 Kepala BECO 15 2Canarium tundang Penaga NDMH 3 1 odontophyllum KNU laut Kedondong OTH 19 1 Kepala 3 Terentang NDLHCarallia 2 tundang 1 2 OTHR sp. 2 - 66.8 t. m 23.43 15 TMPL 14 pendek 14 6 BLUM 15Cassia Bintangor 8 TNG nodosa 15 1 CG 2 739 4 15 BRPA 6 NDLH BACCCastanopsis 24.95 6 6 1 OTH 1 TMP 14 Kembayu 2 NDLH 9.5 Pomotodon -Casuarina 15 OTHR BRGL NDLH equisetifolia TUAI 24 490 - 14 1 14.3 2 14 OTH 14Celastraceae Bunga 50 7 15 6 15 2 TUN family gadong 2 1 3 OTHR NDMHCerbera 9.59 14 1 odollom 1 11.44 27 3 19 1 47.7 1 - 3Chaetocarpus 3 6 15 BRPE 2 castanocarpus 1 6 25 127.3 s-m 7 - - OTH -Chisocheton 15 1 BUAR 1 15 beccarianus 5 21.49 15 BRST OTHR BRST KPLT - 3 57.3 NDLH AruChisocheton 2 29 14 2 glomeratus OTH - 7 OTH 5 16 Kayu 612 15 2341 15 dusun -Cleistanthus OTHR 4 OTHR 1 Busuk-busuk paxii Meransi 21 BUSE - 23.43 s 1 3 1 14 9 - 14 3 - 1Cleistanthus 9 Perupok sp. KET 1 6 2 2162 1 1 600 860 Berangan 3Combretocarpus NDLH s Lisi-lisi 45 rotundatus CAIN - - 1 KDDG 1 15 2 14 28.6 1 4Cordia 3 15 2 KD 3 2 PGI dichotoma Burung Berindu - 150 TERA 0 - gagak 1 1 - -... 15 TRG 15 NDMH 4 NDLH 964 16.37 1 143.2 4 Perapat 18 BINT NDLH 3 - m paya 16 - - 233 27 15 0 678 - BIN - POMO 15 CADD 362 3295 960 POT 4 4 6 0 6 KBYU - 4 m Garu-garu - 2 NDLH 89 BUG KMY NDLH - 1 - 13 OTHR NDLH 1 38.2 - 15 - 15 14 600 - 89 1 18 3 Baubo9 2 10 6 0 27 13 768 15 - 9 245 29 200 - 3 6 4 19.14 - 1 Guma - 1 2 3 4 1 3 2 2 4 - 35 6 245 559 1 25 38.2 - 9 6 2 750 999 3 15 - 2 3 33 6 15 - CHAE CANO 15 5 4 3 - BSK - KAY 89 2 19.14 999 1 1 4 63.7 3 2 1466 NDMH 2 4575 3 NDLH 0 129 ARUX - 20 2 0 CLEI 55 MRSI 38.2 32 559 14 89 4 31 ARU 25 3 PERU 5205 - 1 24.49 BERA CEOD 3 MRSI NDHH BBO 245 - NDLH PEP 3 3 1 - OTH BER 19 19.14 100 1165 OTHR 1 1 14 CHBE 3 690 OTHR - NDLH 146.4 - NDMH 14 38.2 245 CORO - 0 89 57.3 OTH 20 14 432 20 999 CHGL PPP 15 3106 1 OTHR 15 - 108 2 BDU 1 NDMH 1 14 19.31 1 89 999 OTHR 18 23.43 89 1 1 - 1 0 4 1 - 14 14 245 - CLPA 682 15 3 12 3 - 608 15 1 25 OTH 4 9 15 - 8 512 245 3 OTHR 1 245 0 999 1 50 14 15 1 125 104 30 15 19.1 - 1 57.3 2 1 15 382 999 7 999 - 25 27 1 13.19 3 670 0 95.5 - 23.43 CODI 1214 66.8 25 0 1 - 2 1000 - 15 OTH 2511 - - 25 50.9 - - 3 47.7 3908 OTHR 24.95 14 1 2 10479 76.4 28.6 3 22.82 848 s 31.8 50 0 3 79 28.6 1 - 1 26.06 688 - 1010 16.37 801 1 0 16.37 750 - 15 560 1 1 0 - 1 560 1 4 - - 1 89 12 166 13 1 0 3 290 - 678 8 3 1 245 162 54 89 60.5 2502 100 0 1285 999 - 23.98 2 - 3 54 245 54 3 479 100 - 3 999 1.2 89 454 89 - - - 245 245 999 999 - 1 Lists of tree species 175 Botanical name... Cordia subcordataCotylelobium melanoxylonCrateva religiosaCratoxylon arborescens Local nameCratoxylon sp. Resak temporongCroton caudatusCroton heterocarpus AgutudCroton oblongus SerunganCroton sp.Crudia Pangos 13 reticulataCrypteronia griffithii 3 PFTCrypteronia griffithii SS BendakCtenolophon Geronggang Angguk-angguk parvifolius 5 HGCynometra CFRC sp. CHQ RETPDacrydium elatum Lokon TGRP RBG 4Dactylocladus stenostachys IGRP 2 DHH Rambai-rambai F3Dehassia Anggar-anggar incrassata Besi-besi Rambai-rambai 7 GGRP 3Dialium sp. 1 Height 4 Croton 4Dillenia Dia Jongkong borneensis 2 3 2 2Dillenia sp. 3 3 H=f(d)Dimocarpus 4 COSU longan Den Sempilor SERU 3 2 9 OTHDimorphocalyx Katong-katong 2 muriana 2 SERU 3 OTHR NDLH 9Diospyros 3 SPHQ Medang durionoides 2 14 13 sisek CRCA 2 NDer CRREDiospyros 4 sp. OTH 38 2 OTH 2 2 GERODipterocarpus OTHR 1 NFMU 3 applanatus 6 OTHR Simpor 4 SG 14 F gajahDipterocarpus 14 3 2 Obah caudatus CRHE 4 puteh 15 8 - 9Dipterocarpus NDLH OTH Mata CRGR Keranji 4 2 confertus kuching 1 4 18 KAM 1 OTHR 2 CRUDDipterocarpus Sabah CRGR DMH conformis 14 ebony OTH Keruing 1 RAM daun 3Dipterocarpus 15 15 NDLH 3 besar costulatus OTHR Simpor 12 14 15 CROT 3 4 3 42 - 14 14Dipterocarpus OTH 1 coudiferus 4 Keruing CTPA gasing OTHR 2 2Dipterocarpus 2 9 1 crinitus BSI Keruing 14 47.7 9 2 kobis 1 1 6 15 5 JONGDipterocarpus 6.4 Keruing exalatus beludu CROT Kayu 5 NDLH J malam KATO kuning 987 15 9 KeruingDipterocarpus OTH 2 21.49 1 14 kipas 4 geniculatus 15 KAT 15 6 OTHR 1 KeruingDipterocarpus 10 NDHH SPLR putih 7.88 12 globosus - 14 560 14 - 1 10 DEIN 2 15 NDLHDipterocarpus 1 12 SPL 3 45 gracilis 2 3 18 10 s 45 MDK 469Dipterocarpus Keruing NDMH 12 1 mempelas grandiflorus 3 NDLH 89 - 1 4 Keruing 14 2 - 15 Keruing tangkai 2Dipterocarpus 3 rapak 19 panjang SIMG 3 s hasseltii - 4 0 12 85.9 - 2 1 15 9 4Dipterocarpus SIG 8 KDBR 85.9 humeratus 5 1 DIMU MKUC Keruing - KDB 6 245 1 4 buahDipterocarpus s-l 2 15 MAT 3 12 NDMH bulat Keruing kerri DIMU 26.75 5 - DMH OTHR belimbing SEBY - DMH 19 2 26.75 KBKUDipterocarpus 14 12 14 Keruing - lamellatus s - 12 0 9 kesat SEB KRNJ KBK 4 - - 2 2 0 5 999 14 s-mDipterocarpus KGAS 2 - DMH lowii KJ NDHH - 1 - 4 2 - KGS Keruing Keruing 2 1Dipterocarpus 19 kerukap kerukup kecil oblongifolius 4 5 3 KKOB 480 37 DMH 1 - 2 2 12 1 - NDHH 12 KKODipterocarpus 15 SIMP - - 5 ochranceus 3 24 89 10 5 9 0 DMH 1 89 4 12 1 Keruing KKIP... KTPJ 1 SIM 9 jarang 2 1 KMLM 2 5 - 117.8 1 - 4 KMM 76.4 KPUT KTP KEK Keruing Keruing 12 NDMH 2 KMEM 1 NDLH neram gondol 15 36.5 KMP KPT 19 DMH 89 DMH 4.8 245 - - 5 1 19 245 5 DMH - 5 - DMH 12 12 3 4 42.68 Keruing 2 4 4 5 12 9 ranau 1 0 Keruing m 125 4 66.8 31 shol 1 4 - KRAP 5 89 999 KBBT 1 570 7.21 - 2 999 245 KRP 5 2 KBB KBEL 5 528 - 1075 DMH 24.95 1 1 8 l DMH 1062 89 - KB 4 15 1 5 612 - 2 1 - 12 l 1 0 KKKL 1 999 5 245 4 - 4 1 l 1 KKK DMH 5 5 1 12 3 85.9 5 DMH KKES 2 5 46 1 245 KKUK 4 - 20-38 s-m 2 999 KKS - 2 KKU 89 1 2 4 1 26.75 0 - 6 2 - - 2 DMH DMH 5 - 187 1 0 - m 37 999 0 12 - 958 175.1 4 4 5 5 - l 1 5 l 0 2 KJAR 1 2 245 - - - 1665 2 KEJ 237 - KGON 960 5 675 - 1 - - 108 - KNER KGD 0 1 1 5 3 DMH 529 - 999 - DMH KN - 37 5 37 - - 25 4 3 1 1 1 5 5 4 KRAN 649 - 802 DMH 1 KRN 42 - KSHO 4 1020 - - 725 - l DMH 89.1 4 KS - - 2 1 650 - - 1 1 - 4 1 483 344 0 l 46 DMH 41.2 136.9 1 1 5 1 4 1 5 4 - 758 - - 545 - 501 1 1 666 132 925 87.5 5 380 670 0 - 1528 100 1 55 1 5 - 3 3 1 4256 l 41.14 193 1 1 1144 - 1 790 0 6453 872 5 100 s 765 19607 - - - - 3 3 0 m 1 2 1 8 - 4 0 - 1 1343 790 0 55 - 2900 - - 45 755 57 1 100 233 37 - 0 74 - 3 98.7 1 - 3 125 3 65 925 - - 0 74 2 735 1 1 36 - 654 766 1 3 0 866 1 364 0 1 1 89 2 2 0 0 25 0 93 0 25 1 1 507 3 3 3 176 Appendix B Botanical name... Dipterocarpus palembanicusDipterocarpus spDipterocarpus stellatus Keruing palembangDipterocarpus tempehes Local nameDipterocarpus verrucosusDipterocarpus warburgiiDolichandrone spathacea 12 Keruing buluDracrontomelon sp. Keruing asam KeruingDryobalaonops Keruing merah 2 beccariiDryobalaonops Keruing keithii kasugoi 5 TuiDryobalaonops lanceolata PFTDryobalaonops rapa SS KPALDrypetes KPD microphylla HG Kapur DMH merahDuabanga Sengkuang/soronsob CFRC 12 (minyak) moluccana 9 12 Kapur CHQDuriograveolens paji 4 Kapur gumpait 9 TGRP 2Duriosp. 2 IGRP 2 12 F3Dyera costulata 1 12 5 6 2 5 GGRP KapurDyera 2 4 paya polyphylla Odopan Height putih 2 5 KBULElaeocarpus Dia 4 sp. 2 Magas KMRH 5 KBU KASM KMRElaeocarpus sp. DMH KA DMH 5 KKAS 12 6 H=f(d)Elaeocarpus 3 sp. KPMH Den 4 KK 4 Durian DMH 42 KPM merahElateriospermum KERU 12 tapos 2 DMH SENG KR DMH 4 2Endospermum SPHQ sp. SGK 5 2 1 1 4 NDer 5Ervatamia DMH - sp. NDLH 3 3 Jelutong bukit 12 18Erythrina Jelutong 1 NFMU 4 5 variegata paya 5 5 KPGM F 1 DurianErythroxylum KG 1 TUIX cuneatum 2 2 Tonop Perah 5 4 KPJI ikanEugenia - OTH sp. DMH 4 1 Kungkurad 5 12 OTHREuodia 5 KPG 2 61 11 l sp. 15 14 5 Kulibobok DMHEurycoma 5 longifolia 2 36 KPYA Sendok-sendok 1 DRMA 5Eusideroxilon Perepat 678 61 ODP 1 zwageri KY burung 12 117.8 31 1 36 5 NDLHEusideroxylon Dadap 6 - malagangai 5 DMH 14 - 15 m-l -Fagraea 2 1 1 racemosa Burut-burut 4 DUGR 95.5 5Fagraea - MAGA 2 DUGRNDLH 47.7 sp. MAG 1 5 4 17 - PIONFagraea 6 sp. - 24 49.29 12 Malangangai 3 3 731 21.49 Pahit-pahit 13 4 -Ficus 1 15 (tongkat 36.6 Belian fulva 8 DYCO ali 9 731 1 6 ) JLBFicus 2 3 - 2 600 Obah sp. DYPO 2 76 4 - 3 2 66.8 - NDLHFicus JLP 4 1 sp. 817 8 2 s-m 1 Pau-pau 5 2 15 3 4 2 642Flacourtia 1 NDLH 2 rukam 14 43.74 Tadapon 9 15 4Ganua 658 puak 146.4 4 3 1 motleyana DURI 3 ELAT - 1 37 2 14.5 - 3 ELAE 778 1 815Garcinia 46 PEI DRN forbesii - 1467 SEND 2 1 KUL ELST 2 2 - NDLH PEBR 7 0 1Garcinia 8 ELAE mangostana NDLH 116.5 SSB NDMH 17 PB KUK 2 1 2826 1 16 66.8Garcinia 14 112 TNP Tadapon nervosa 8 NDLH 4 - EULO NDLH putih 98.7 2 - 8 NDLH 3Garcinia 16 NDMH 16 OTH Tambusu parvifolia 736 1 1 2 14 19 43.74 2 66.8 8 0 1 OTHR 60... 224 ERMA 10 - Togung 14 3 OTH korop 767 2571 4 2 752 Rukam 2 8 ERVA m 2 NDLH 1 15 - 1 15 9 41.73 8 3 6 89 DDP Nyatoh 14 2 katiau MGAI 12145 1 Jiwit 248.3 Manggis NDLH 15 15 3 9992 0 3 400 1 MA 15 15 14 Ara - Bebata 4 2 2 - - s-m 1 - 21-46 15 7 245 1280.5 1 NDHH 2 1 2 1 12 8963 - BELI 1 40 4 - Kandis 3 - 15 Kandis 46 daun 2 - 0 besar FARA 3 999 B 10 2 1 4 465 6 15 - OTH OBAH 57.3 OTHR 7 454 OB OTHR - PAU 623 3 - 50.9 9 - - 14 - NDHH 13 OTHR - 129 23.43 1 2 - NDMH - 12 3 14 15 12 1433 TEMB 0 6968 4 22.18 - 1 840 2 6 TM 3 - 1 3 4 4 m-l 1 1 - 848 - 4 - NDHH 640 - 4 - 15 3 4 625 14 9 - 1 2 TAND 3 TEMB 0 13 15 OTH 3 9 57.3 1 1 TPP 1 2 GAMO OTHR 9 - 3 1 GAMOOTHR NDHH 1 2 14 - s 2 - 14 3 2 14 33 24.21 382 - 0 - - RUKA 31 - GARC 2 15 4 OTH 27 0 2 685 1 0 KAN 1 GAMA 347 1 OTHR 2 GAMANDMH 1 NDMH 1 200 19 14 2 - 20 4 - 1 KRAH 57.3 GAFO 15 9 - 200 1 s ARA 117.8 15 2 89 BBA - 2137 OTHR 1 25 200 1 JIWT 104 NDMH 1 23.43 14 - 19 0 1177 KNDS 1 OTH 1 18 308 - 89 15 m OTHR 15 KNDS - 245 40 NDMH 15 1 0 - 1 - 1 20 14 4234 - 1 - 1735 1038 0 - 999 245 3 15 s 15 1 12 1 - 82.8 21 - 1 - 0 0 1 - 800 - 15 15 999 28.96 m-l 18 - - 9.5 - - - 560 89 1 1957 - 1 0 985 - 30 l - 9.59 9.5 7740 1 - - 245 3401 - - 6081 960 22.3 2 3 9.59 1 - 3 - 999 40588 0 14.3 1 - 89 - 3 998 1 3 0 688 - - 0 1 245 0 89 1 1 4 448 999 - 1 3 0 245 0 0 - 325 104 3 999 0 89 0 2736 - 2 0 2 3 245 1045 - 2 - 999 - Lists of tree species 177 Botanical name... Geunsia pentandraGironniera sp.Glochidion litoraleGlochidion sp. Local nameGlochidion superbumGonystylus bancanusGordonia sp. TambungGuioa sp.Gymnacranthera contracta Saka-sakaHelicia Ampas sp. Gerumong tebu jantanHeritiera littoralis Ramin PFTHeritiera Oba Lunau simplicifolia SS nasiHibiscus tiliceus HGHomalium CFRC caryophyllaceum CHQ 5Homalium 6 sp. TGRP MelulokHopea IGRP aequalis F3 Takaliu 1Hopea 1 4 2 argentea Kembang/ mengkulang Dungun GGRPHopea Tanggir manuk beccariana Height 3 3 2Hopea 3 Dia dryobalanoides Kurunggu 9Hopea TAMB 6 1 dyeri GLSU Baru 1 2 TAMBOTHR H=f(d)Hopea 6 14 OTH ferruginea Den 2 OTHRHopea 2 2 GLLI latifolia GINE 14Hopea 2 1 SPHQ mengerawan Takaliu OTH OTH 4 4 3 1 Selangan NDer 1Hopea OTHR OTHR Selangan sama micrantha 1 daun 14 3 Selangan 14 15 kapur Selangan uratHopea KEMB NFMU penak 7 montana RAMN OBNA 3 KM F 2 R 4Hopea 15 ONA GYCO myrtifolia 1 1 NDLH NDMH OTHHopea 18 1 3 2 19 nervosa 14 OTHR 1 4 3 NDLHHopea 15 Selangan 15 14 nutans mata 18 18 kucing 3 GUIOHopea Selangan 1 2 Selangan GORD 1 pentanervia daun hitam - 2 halus OTH OTH 13 3 4 1Hopea Selangan sangal jongkong OTHR TLIU 13 s 4 OTHR Busch 9 1 38.2 6 15Hopea Selangan 14 14 HELI 3 semicuneata lunas - 2 7 TKU 15 7 3 3Hopea 15 NDHH sp. DUGNNDHH KGGU - 19.14 Selangan 14 1 bukit 2 1 19 KRGUNDLH 5 4Hopea SDKP Selangan tenuinervula Bsche 3 - beludu 2 45 14 5 3 - SLKHopea 18 vaccinifolia - 13 - 15 27 3 15 SSAM 1 1 13 DMH 3Hopea 3 Selangan HITI SPEN 3 wyatt-smithii jangkang 1 SLS Selangan - lima 3 133.7 7Hydnocarpus urat SLE SURT 3 15 sp. 9 Giam - 25.5 BRU DMH 2 SMKC 89 13 b 66.8 TLIU s SLTIlex 15 DMH - & OTHR Giam cissoidea. SMC - - t kulit 5 7 merah 14 5 DMHIlex TKU DMH 15.36 SDHS 2 cymosa 7 3 13 13 1 Gagil 24.95 - 6 OTHRIntsia 7 Selangan SDH 245 7 15 SJON bijuga 14 - - daun 9 b SHTM 1 serong & 2 675 - 2 3 3 10 DMHIntsia 5 t SJK SH 749 palembanica - 89 2 Selangan Selangan ribuIrvingia daun 7 1 999 - DMH 15 malayana 1016 2 bulat 2 2 3 31.8 2 5 5 76.4 - SLUN DMH 1 SelanganItea 1 2 10 macrophylla 7 89 89 SLN 13 s 15 3948 245 7 - 2Ixonanthes 4 2 3 4 SBUK Karpus 17.34 SBDU reticulata 28.03 1 2 DMH 12 3... SUK SU 3 - - - 928 795 2 - 7 SJKG 220 6 999 245 SLUR 245 DMH 410 2 37 2 s-l 4 Morogis DMH SJ 10 - SLU 5 m - m 7 19.1 2649 2 Merbau 2 1 1228 2 DHH 7 2 999 2 999 2 4078 4 HOSE 3 Bangkulatan - DMH SDSG - 95.5 - 89 5877 Pauh 13.19 GK s Ipil 13 7 kijang 3 laut 2 3 SDS - 3 7 - Inggir 2 - 1 - 4 burung 89 3 2 - - DHH 36.51 Marapid/kaintuhan 2 266.5 DMH 100 3 - 13 2 SDBL 245 2 1 - 7 GIAM 786 7 360 2 - 2 2 - SLB 5 G 245 650 3 - - - 718 6 2 DMH 999 1 - SRIB 2 1 3 - 2 GAGL 2 - 4 5 7 8 DHH 999 2 GL SLR - - 1000 4 1 6 2 - - 15 2 DMH - - 7 699 SELA 10 30 DMH 4 10 3 4 2 - - 31 29 2 3 7 3 S - 2 - 108 7 3 3 1 7 1 - 0 2 2 - 3 766 2 KARP 57.3 - 2 3 1 - 47.7 41 KAR 2 4 DMH 4 2 266.5 4 - ITMA 2 3 15 NDMH 2 715 - 1 29.17 18 BKLT 7 MORO OTH 360 0 0 0 26.2 MGS 2 - PAUH MERB BGN OTHR 108.2 m 4 3 IXON 2 OTHR 1104 PKI MER - 14 0 OTHR 787 1 3 4 14 l NDHH 14 IB - 704 1 - 2 NDHH 12 1 33 ILAT 851 10 1 0 0 - s 15 - 1 - 1 46 NDMH 1 IPL 2 1 0 14 4 - 0 1 1 15 - 316 NDHH 1008 66 15 15 114 14 15-30 9 - 1 2 117.8 - 0 15 - 2 1 m-l 1210 57.3 54 2 s - 0 - 1 114 1 m 0 15 s-m - 4417 23.43 40 - 55 15 0 - - 0 - - 0 - 700 3 s - - - 1 699 8 1056 146.4 146.4 24 - 1 1 0 124 - 1 1 - 1 - - - 0 - - 1 57.3 2 2417 1080 - 8 0 8 2 992 793 23.43 - 0 - - - 4952 1 838 1 1 - 0 0 8 0 1 1 - - 182 89 2 118 165 3 - 3 0 4239 245 0 0 4 1 421 813 - 999 - 3 2 125 0 218.5 100 1 - 3 300 3 - 178 Appendix B OTHR 14 1 15 24 44.6 20.75 432 1 0 0 3 Mallotus philipinensisMallotus sp.Mangifera pajangMangifera sp.Mangifera sp.Mangifera Mallotus sp. philipineMangitera sp.Mangostana sp.Meliaceae family BambanganMeliosma sumatrana 4 MelutosMemecylon sp.Mesua macrantha Bachang 3Microcos sp. DumpiringMicrocos sp. 2 PahuMilletia sp. Assam Manggis Gapas-gapas... MAPH Lantupak OTH 9 OTHR 14 2 Bintangor Nipis batu 4 kulit 1 4 6 6 3 15 MGPA Korodong 6 BBG 2 2 2 Korodong/damak-damak NDLH 10 6 6 17 6 2 3 6 3 Taroi-taroi 4 MALL 1 MTS 1 3 2 ASAM 2 2 ASAM NDLH 9.5 2 BC 3 2 DUM 1 14 NDLH 15 GPAS 3 3 3 19 NDLH 3 2 1 GP 9.59 1 2 19 GARC ASAM LANT ASAM 1 1 NDMH MGS PHU - MEMA KRDG LA 14 749 15 ASS BIB 1 1 NDMH KDG NDLH 19 OTHR NDLH 19 15 NDMH OTHR 2 4 14 19 MLAE 1 19 15 14 89 OTH s-m 1 - 1 OTHR 1 1 15 3 1 s-l 1 14 1 15 s-l 245 - 15 KRDG 15 2 9 15 DAMAOTHR 8 1 - 15 m - 999 - - MILL s-l 15 - b 1 s-l & OTH s t m 2 - OTHR - - 14 - 15 - b - - & - t - - 1 - - - b 1 & 600 - t - 15 - 600 - 1 - - - 1 - - 4 1 s-m 600 600 - 600 3970 - 0 - - - 1 - 1 4 1 429 12803 1 2 - 600 433 0 1 - 1 741 0 - 0 89 258 1 - 4 2236 - 7473 245 158 307 - 404 100 - 21692 875 218.5 999 283 - - - 3 - 89 975 - - - 245 999 3 Botanical name... Jackia ornataKleinhovia hospitaKoilodepus sp.Koompassia excelsa Local nameKoompassia malaccensisKoordersiodendron pinnatumLagerstroemia speciosaLansium Ranggu domesticum TimaharLapisanthes Kempas sp. SelumarLasianthus Mengaris sp.Lauraceae Kilas family Bungor PFTLeea sp. SSLeptospermum sp. Langsat HGLinociera CFRC sp. CHQLithocarpus sp. TGRPLitsea cubeba IGRP 6 9 F3 LapisanthesLitsea graciea 13 7 GGRP 12Litsea Medang Kopi-kopi odorifera Height 2 2 3Litsea Dia odorifera 2 Gelam 3 bukitMacaranga 3 4 conifera 6 4 5 H=f(d) 5Macaranga 3 hosei Den KLHO 4 RGGUMacaranga IMPS 2 sp. RGU 3 OTH Mali-mali MENG Mempening Bangkulat JAORMacaranga NDMH OTHR MEN SPHQ IMP sp. 9 18 3 14 SLR NDMH 3 NDerMacaranga 2 10 NDHH sp. Lindos/railos NDHH 10Macaranga NFMU 4 20 tanarius 2 Pengulobon 2 1 OTHR Medang 1 KILA F pawas 1Magnoliaceae Ludai OTH family 4 1 Medang 1 KLS OTHR pawas 8 LADOMallotus 4 1 15 mollissimus 14 2 12 LADOOTHR NDHHMallotus Lopokon 14 3 12 muticus 2 14 LAPI 15 1 1 20 12 30 OTH 2 1 80 1 Kubin 6 1 1 OTHR 9 55 KOPI 15 Lingkabong 2 Sedaman 35 14 6 LEPT MEDA 15 15 Cempaka OTH 28.6 2 Sedaman MD 2 47.7 6 222.8 OTH 6 OTHR 2 Dahu 5 1 95.5 OTHR 14 30.5 NDLH 2 57.3 16.37 - 3 14 1 31.8 19 7 2 15 4 MEMP 2 15 MEM 38.7 480 1 Mallotus 47.7 3 NDMH 23.43 5 paya LIPL LEEA 1 801 20 3 1 LICB 3 8 OTH 1114 912 BGT 19.1 15 827 21.49 LIOD 89 9.5 OTHR - LDS 15 1 OTHR LIGA 1 9 1 14 LIOD MDP 5 14 NDLH 674 1 1 13.19 NDLH 2 PGN 1 19 TWD 3 5 b 5 19 9.59 2 NDLH NDLH 245 & 1 1 - 12 t 1 1 4 19 89 19 5 - b 607 1 LUDA & - - 153 1 1 t 1 15 6 999 LUDA 0 NDLH 15 3 273 2 MAHO 1 1 4 1 OTH - - 13 - 15 1509 245 9 3 1 OTHR 3 573 - 2 MATA 1 2 14 9 9 981 b 3 OTH & - oft 1 3 2 3 t m 0 MACA OTHR MAGI 999 - SEDA 3 2 MAGN 14 MACA - - 1 OTH - SEDA - CP - m 2 13 0 2 14 MACA - SEDA m 13 m 3 MACA MAMU 525 1 NDMH MAMU 15 14 - 2 MAMO 20 3 OTH - 24 47.7 - - 3 OTHR 615 - - 0 15 1 8050 14 38.2 47.7 14 1 89 - 22 14 89 1 1 14 47.7 19.31 22 1 - 15 21 - 89 3 - - s-m 245 - 35.49 509 s-m - 15 245 s-m 7394 509 b - - 400 & 999 245 - t 1 1 - 999 89 1 32885 - s 1 - 1 - 2 - 999 2 - - - 1 245 2672 0 - 154 - - - 0 0 - 1033 - 23659 999 154 500 400 0 3 1562 25 400 8 - 400 3 0 - - 89 1 89 2 89 - 100 2 1 - 245 - 245 1 18555.5 - 64049.5 245 - 18555.5 999 999 64049.5 - 999 89 629 - - - 1008 245 - 999 3 Lists of tree species 179 Botanical name... Myristicaceae familyNauclea sp.Neesia sp.Nephelium glabrum Local nameNephelium maingayiNephelium mutabile Darah-darahNephelium sp.Nephelium sp.Notaphoebe obovataOchanostachys Satu amentacea inchi KelamondoiOctomeles Bangkal sumatrana PFT MaritamOmalanthus sp. SS DurianOsbornia 6 monyet octodonta HG PetalingOstodes CFRC sp. Lamau-lamau CHQOtophora 2 Meritam fruticosa TGRPPangium Rambutan edule IGRP Binuang F3 3Paranephelium sp. 6 4 GGRPParashorea malaanonan DARA Height 12 Gelam DRAParashorea Dia laut parvifolia 2 8 Ludai NDLH 3 9Parashorea 2 smythiesii 16 9 H=f(d)Parashorea 3 sp. 2 Balingasan 1 Den 7 5 2Parashorea 2 tomentella Urat 2 NEPH mata NEGLParastemon daun 4 SPHQ 9 Pait-pait urophyllum licin KDI DMYT 4 9 Membuakat 3 NEGL 15 DRM OTHR Pangi NDerParinari 4 OTHR NDLH 11 Urat 14 BKAL mataParinari 14 daun MERI 17 2 NFMU Urat 3 oblongifolia kecil 2 BKL mata F batu 1 MTM NOOBParishia s-l NDLH 9 insignis 1 NDLH 12 LMU 4 1 TGGL 1 14Parishia 4 Urat 14 NDLH sp. mata Mandailas PET 5 beludu 18 2 15Parkia 2 javanica 12 NDMH MERI 15 4 8 - RBTN 5 1 19 1Parkia Urat BINU sp. MTM mata RBTNOTHR 5 4 NDLH 2 4Peltophorum 14 4 BN racemosum s-m 15 14 3 15 1 1 Merbatu 35Pentace 12 6 UMDL adenophora 15 m-l 5 OSOC - PION UML 1 12Pentace 2 3 1 - laxiflora OTH 15 3 DLH 2 13 m-l 35 OTHRPentace 9 2 UMDK 50.9 sp. Layang-layang - 2 15 UMK 14 OTFR m 1 2 15 LUDAPentaspodon Timbarayong DLH 5 Bangkawang motleyii 3 OTH 591 30 LUDA 9 3 - NDLH 22.18Pericopsis 2 5 OTHR - mooniana 19.1 1 5 MEMB 16 Kupang 14 UMBT OTH 45Phaleria 1 11 Layang-layang - 45 - perrottetetiana UMB PAIT - OTHR 1 12 2 4 UMBL Takalis DMH 13.19 15Phyllanthus daun 57.3 14 emblica 2 bulat UMU PAIT 1 1 3 - OTHR DLHPithecellobium 5 2 - - 127.3 sp. 4 KEPA 55 14 127.3 23.43 Petai 9 15 1Planchonia PAN 9 15 1 Takalis 1 valida - Pelajau 3 3985 - - daun - halus 5 NDLH - PAURPlectronia 880 9 1 1 confertum Alig 14 61 15 MDS pagi 9 2 720 Ipil 194.2 18355 2Pleiocarpidia air 1 2 NDLH sandakanensis s 10 WHSY 1 6 9 3 15 WS 3 1Podocarpus 18 0 - 2 43 1 - blumeii Takalis 1 - 4 - 2 4 1 Laka 3 194.2 - s-mPodocarpus 3 DLH imbricatus 4 3 2 4 Buloh-buloh 15 9... PAOD 0 386.5 - 4 - LAYA m - 4 Jering 85.9 1 MEB 43 381 - 0 LAY 400 312 1 NDMH 1067.5 - TIMB 4 15 1 Grubai 3027 2 12 Putat NDMH 2 2 BKWG 61 paya - PEAD 20 OTH NAN 46.43 38 1092 1 - 0 PEAD OTHR NDLH 1 NDHH 10070 LAYA 194.2 531 1032 - 2 - 665 14 4 4 17 12 - 6 Lompoyou LAY 36 - Lampias 1 0 - 9 2 158 - 194.2 - 3 2 OTHR 2 85.9 386.5 PELA 1 20 15 KUPA 1 1 4 1 - 15 2 - 6 TKH KNG - 259 1067.5 2 - - 570 - 26.75 1 3 NDLH NDLH 9 - - 13432 15 1 9 15 14 16 l 3 89 678 40 660 4 2 3457 1432 - 60 2 6 - 0 2 506 15 2 - PELJ 1 2 9 2 - 3 1 PTAI l 89 3 7890 48 1 1 ALIG 245 89 30 82.8 PEL 2 4 2 - 4 PTI 60.5 1 8 9 OTH 9 NDLH 3 2 60 IAYR 25 OTHR 2 NDMH 19 PLSA 26.56 999 245 3 1 - 14 IPA TAKA 14 12 1033 3 23.98 1123 1075 245 66.8 - 0 4 TKS BLH 2 739 NDMH 1 - 560 4859 105 1 30 PYEM NDLH NDLH s 1 19 2 1562 - 2566 999 2 2 1 1 26.36 OTH 14 999 17 PUTP 4 0 OTHR JARG - 1 13103 PUT - 15 2 5 - 0 14 1 2 - PLCO 8 - 15 OTH - NDLH - 1 2 1 - POIM OTH OTHR - 315.1 18 0 14 OTHR POBL 15 LOM 1 14 36 0 15 NDLH 9 LPS - 543 810 s - 0 1 30 4 17 - 640 1 NDLH 15 - 1 1 17 30 66.5 1670 1 42 8 s 1 15 89 - m-l 79.6 360 - 15 359.5 0 3 36 3 1 2 14 47.7 - 94 20.13 2 40.84 - 1 0 b 245 30 14 - & t s 21.49 722 47.7 600 - 0 35 0 - 10725 999 800 1 95.5 37 21.49 - 1 1274 0 1 - - - 600 76.4 - 1 - - 42.2 2011 359.5 - 95.5 100 42.68 3 0 3 315 750 89 3 792 - - - 520 0 100 89 1303 1 - 1 245 0 1 3 3 1 245 - 619 0 999 2663 66 89 1 999 1 0 7985 0 2 2 89 865 245 - 2 0 0 0 245 1 999 999 - 0 2 - - 2 180 Appendix B Botanical name... Podocarpus rumphiiPolyosma integrifoliaPometia pinnataPongomia pinnata Local namePrunus javanicaPternandra coerulescens KayuPterocarpus china indicus BedaruPterocymbium tinctoriumPterospermum sp.Quassia Kasai Sirih-sirih borneensis MarabahaiRandia anisophylla Teluto PFTRyparosa sp. SS KelanusSandoricum Angsana maingayi HG 6Sandoricum CFRC mangyi CHQSapium indicum TGRP Bayor 2Sapotaceae 4 IGRP family Manunggal F3Saracca Bembalor sp. 3 GGRP 6 6Sarcotheca 3 Sentul diversifolia Height hutanSaurauia Dia PORU sp. 9 2 2 2 KCN SentolScaphium hutan affine 6 NDMH 6 Giwie H=f(d) 17Schima 9 wallichii BEDA 3 2 3 Den BEDScorodocarpus 2 borneensis Apid-apid 2 Nyatoh OTHR 1 2Serialbizzia Tabarus MHAI 4 SIRE SPHQ splendens 14 6 MHI 3 NDerSerianthes OTH dilmyi 3 14 4 5 KASA NDMH 4 9 OTHRShorea 18 NFMU 1 acuminatissima 2 KAS TELU 14 F Bawang PRJA hutanShorea NDMH ANGS agami TTO Gapis 1 9 19 KNS 15 3 m 2 4Shorea 3 ANG NDMH almon 1 14 Kembang NDLH NDLH semangkok Sokong-sokongShorea 3 andulensis Kungkur 14 17 Gatal-gatal 2 4 1 2 15 QUBOShorea 15 14 MGL angentifolia 1 - Seraya 6 OTHR 6 BAYOShorea kuning RAAN SAMA 1 15 1 4 angustifolia runcing 14 BY OTH STH 9 9 BataiShorea 4 9 laut atrinervosaq 20 OTHR - NDLH 9 2 15 8 2 SAMA NDLHShorea 14 1 beccariana 19 - STL 40 14 2 12Shorea 3 2 biawak 3 3 NDLH 4 2 - 85.9 1 15Shorea 1 30 15 47.7 bracteolata 1 2 Melapi 4 - 31 2 Seraya agama 4 76.4Shorea daun GIEW APID 4 Seraya coriacea merah 9 - Seraya 15 3 26.75 3 daun kerukup 15 GWI mas 21.49Shorea 1 9 OTH 5 KSMK 9 cristata m Seraya SADI NYAT OTHR kuning 66.8 26.06 KEM 1000 57.3 OTHR bukit BWHNShorea Selangan 76.4 NT 2 14 800 curtissii 2 batu NDLH BWH 14 2 OTH - hitam 15 SKRG NDMH 12 832Shorea 17 2 OTHR 45 19 89 24.95 dasyphylla 1 NDMH 24.35 SPC 3 Seraya 40.31 SAUR 14 - 21 m 4 langgai 1Shorea 2 1 12 1 domatiosa DLH - OTH 672 4 1 1Shorea 624 1 45 OTHR - exeliptica 3 Malapi 12 76.4 1 GTAL 1 15 2 245 12 2 2 GAPI pang 6 13 15 14Shorea 12 Selangan 0 faguetiana GT batu - KKUR 15 biawak 8 1 GAPI - 1Shorea 0 KUR 2 NDLH 2 15 2 26.06 falciferoides 2 3 5 9 76.4 2 1 999 NDLH 1 33 Seraya NDLH 14 tangkaiShorea 21 2 panjang 18 - fallax 14 20 0 - 5 5 45 Kawang SEDI 2 5 SDME... 4 daun 26.06 5 15 10 merah - 36 0 3 1 s Seraya SDR 12 6 Seraya 0 - 0 batu betul 111 BLT 1 s 1 SDMS 0 - - 12 MEAG - DLH SKBT SBHM 38.2 l SKER Selangan 3 NDLH MPA batu SDM 2 76.4 600 0 15 SBX SKT mata-mata 1 DLH SKE 14 s 15 DLH 15 - 3 76.4 2 3 Selangan DHH 12 DLH 19.14 87 batu 400 DLH l 4 - - 0 13 - 12 tembaga Seraya 5 2 2 26.06 kuning 89 0 8 Selangan siput 1 - 10 1 batu 5 448 40.31 6 14 2 laut 2 - 1 3 45 2 SBBK 560 30 0 1 SLGG - 1 905 13 SBW 15 1 STKP 1 SLG 5 1 - 89 - - - 245 2 DHH 2 5 5 2 1 - STP 1 1 0 DLH - 3 12 76.4 12 95.5 Seraya 1 2 7 12 daun 10 DLH MEPG 2 0 1990 kasar SBMM s-m KWDM 6 999 MPP 11 - KWM 2 12 DLH 245 SMM 5 2 2 DLH 26.06 27.03 7 2 14.5 DHH - - - 38 12526 1 1250 - 0 11 2 1 - 1 2 672 720 2 445 60 SBTM 5 60 5 3 10 999 53 2 5 - 15 SBZ 7 2194 - 2 61 5 - 2 1 SBET DHH SBAT 3 1 1 1 89 - 851 1 SKSP 1 2 2 3 - 143.2 127.3 650 SRB SRU 2 117.8 - SBLT 10 SSP 2 31.8 117.8 DLH DLH 31 7 - - 2 2 2 SBP - 89 DLH 55 245 - 2733 - 0 4 4217 1 21.42 8 DMH 9 1 2 1 43 6 - - 10 - - 853 17451 999 245 60 7 98.7 665 m-l 997 829 SEDK 1 1 2 - 150 4 527 638 1 2 SDK - 1 - 137 999 36.8 1 DLH 1 - 1 2 2 143.2 1 - - 60 7 6 3 1 1188 3 597 - 241 - 1 2 - 0 - 33 0 0 61 l - 143.2 709 - 930 1 2 - 63 1 98 - 1022 0 2127 325 62 699 1 2 146.4 4 0 1 - 146.4 647 729 278 828 2 - - 1 1 45.71 944 258 2 1 0 7 2 781 1 634 2 2 0 - 1 1 656 - 0 3 1 4 0 - 0 1 0 520 0 - 0 1050 4 3 33 1 - 0 1 1392 0 1 1 1 2 1 0 - 0 3 86 54 1 1 86 2 0 - 0 1 Lists of tree species 181 Botanical name... Shorea ferrugineaShorea flavifloraShorea foxworthyiiShorea gibbosa Local nameShorea glaucescensShorea gratissimaShorea havilandii SerayaShorea melantai hopeifolia kecilShorea hypoleuca Selangan Seraya daun batu besar bersisekShorea johorensisShorea kudatensis Selangan 12 batu PFT SerayaShorea laut kuning kunstleri SS 13 gajah MelapiShorea 2 laut lamentella HG 3 SelanganShorea CFRC 6 batu laxa pinang 5 CHQShorea Seraya leprosula kuning 5 TGRP jantan 12 Selangan 10Shorea 2 batu leptoderma IGRP SMKL kelabu F3 SerayaShorea SMK SBBS majau macrophylla 2 4 3 Seraya DLH GGRP 3 kuningShorea SBB kudat macroptera Height 12Shorea 5 Dia 4 11 DHH mecistopteryx 3 Seraya SDBR 10 sirap MelapiShorea 12 lapis 2 SDB multiflora 10 SKGH SBLX 1 H=f(d) 3Shorea 2 DLH myrionerva SGT Den 2 12 SBL 5Shorea 1 DLH nebulosa Seraya Selangan 11 2 tembaga 4 DMH SBPG batuShorea Kawang biabas 2 SPHQ 5 Seraya jantung obscura 12 kuning SKJN SPG 6 keladi 10 7 NDerShorea SBKB ochracea 1 Kawang Seraya DHH SJT burung melantai 5 MELT 2Shorea SBG NFMU oleosa - MPU DLH 1 10 F 2 DHH 2Shorea DLH 10 ovalis 9 SKKU 5 60 12 6Shorea Banjutan 3 SDD 10 7 ovata Seraya 2 1 6 3 urat 12 DLH banyakShorea SMAJ 12 2 2 parvifolia - m 2 SMShorea 1 143.2 7 2 4 2 parvistipulata 6 Seraya 1 kabut 2 9 12 5 4Shorea 35 patoiensis DLH 70 - 2 7 6 Selangan SBBI 5Shorea batu pauciflora 2 2 - 5 - tanduk Melapi MELP SSIR 2 9 2 2 daun 15Shorea besar SBI MPL SKKL pilosa 79.6 STEM 146.4 SSR DLH 5Shorea KWJT SLI 6 pinanga ST DHH 4 992 Seraya 35 KWJ 61 2 minyak DHH 1Shorea 45.71 53 9 45.86 DLH platycarpa 2 - 31.8 - Seraya DLH 10 SMEL Seraya lupah 3 DLH kepongShorea KWBR 10 Seraya 509 platyclodos 4 1 SML KWR punai 838 11 1 Seraya 6 DLH 79.6 punai 12Shorea 46 98.7 26.06 2 12 DLH bukit 3 polyandra 1 1 98.7 2 SUBK 1Shorea Seraya 11 1 kuning quandrinervis 1088 - 1 1 1 2 pinang 11 45.86 2 SBK 40.71 4Shorea Oba 7 2 8 2 36.8 quiso suluk 2 - 69 1 2 DLH 1 938 1Shorea 570 2 5 retusa 5 12 1 SBTA 6 11 BANJ 620Shorea 20 2 6 0 70 SBT revoluta 1 12 Kawang 12 2 Seraya SKAB Kawang 1 BJ bulu 117.8 31 2 paya 1 9 MEDBShorea pinang 16 2 46 DHH rubra - Seraya 55 MPB SKB bukit 0 12 1 46 1 2Shorea - 2 2 DLH rugosa DLH DMH 10 95 Seraya s 5 236 sudu 2 Seraya 61Shorea 233 kuning 0 2 - scaberrima 2 2 - quion 39 2 7 5 124 5 3 9 0 76.4 61... SMIN 1 98.7 638 0 4 - 0 2 5 2 499 SMY SKPG 98.7 SKEP SPBT - 33.74 136 12 1 2 DLH 76.4 7 36.8 220 1 PN SLUP - 38 98.7 SKP 9 SNB 1 - SPUN 428 848 Selangan 12 1 12 36.8 batu SLA 2 DLH DLH 3 SNI 1 merah 2 Seraya 6 DLH 2 33.74 9 0 Seraya 350 2 daun daun - 36.8 tumpul DLH 12 tajam 2 2 - 2 11 DLH 1 7 85.9 1 35 575 5 550 6 0 930 728 12 1 8 2 Seraya 1 2 540 Seraya mempelas 2244 4 bingkai 3 5 5 l 15 Seraya 35.36 1 2 OSUL 13 55 buaya 1 1 hantu 2 2 2 2 79.6 4 1 5 OS 1 10955 9 - KWPG 0 610 1 SKQN 1 KWBL KWP 3 2 0 1 9 KWB DLH SPQ 3 464 2 5 76.4 6 DLH SPAY DLH 45.86 127.3 SBKT DLH - 2 2 917 1 942 5 SYA SRI 11 2 61 70 922 2 11 - 5 0 SSUD 35.87 12 770 0 1 3519 DLH 9 6 61 2 DLH 4 SSU SBMH m 6453 1 38 2776 12 659 4 7 2 1 SBM 7132 1 DLH 35 - 5 2 - 0 194 1 2 SDTU 61 DHH 2 3 0 1 2 604 2 117.8 1 0 SBTA SDU 2 5 9 1 2 - 10 - 2 4 2 SRT DLH 2 - 6 1 63.7 5 137 98.7 1 DLH SBHA 0 539 - 3 1 SMMP 2 1 1 SRH 30 129 2 SMP 30.89 SBIN 122 - 7 36.8 137 - DLH 69 - DLH 1 - 7 2 509 SRK 499 1 174 1 1 7 478 413 DLH 79.6 11 468 46 2 - 55 146.4 - 1 1 2 8 784 174 - 100 34.33 - 1 38 1 2 - 1 2 2 - 57.3 419 175.1 1 2 2 1 112 2 840 0 - 30 - - 29.17 1 1951 - 3385 32 1 - 675 2 709 2 4072 10303 0 35 12993 54.1 - 0 409 1 63.7 - - 736 3 0 3 1 1635 - - l 28.23 63.7 1 30.89 2017 1 1643 - - 2 358 - 4 569 0 30.89 2 2 - 1 - 549 609 674 25 1 1 - 1 1 29 1 621 0 1 3 - 503 0 0 649 - 1 0 2 0 0 33 668 1 0 0 0 33 1 0 1 0 2 1 - 1 25 2 0 29 1 1 182 Appendix B Botanical name... Shorea scabridaShorea scrobiculataShorea seminisShorea slooteni Local nameShorea smithianaShorea sp.Shorea sp. (Eushorea group)Shorea Selangan sp. batu Seraya (Richetia kurap lop group)Shorea sp. (Rubroshorea group)Shorea Selangan superba batu Seraya SelanganShorea batu symingtonii Seraya terandak kuning PFT SerayaShorea Seraya 13 kepong s.d. timbau teysmanniana SS kasar besarShorea venulosa HG 3 13Shorea CFRC 12 virescens CHQShorea 3 5 waltonii TGRP Kawang 2 9Shorea 9 xanthophylla IGRP 13 F3 5 SBKPSindora Melapi irpicina 5 Seraya kuning SBS bunga 2 GGRP Selangan 3 (bunga) 12Stemonurus b. 2 corniculata Height SBTK daun DHH halusStemonurus SKUN Dia SBY scorpioides 4 5 2 12 4 DLH 10Sterculia DHH 12 macrophylla Seraya DLH kerangas 13 H=f(d)Sympetalandra SKEK SBTU 2 5 borneensis Melapi SLOP 10 Den sulang Seraya 2 1 SKG SB kuning salig 6Symplocos barun fasciculata SLP 3 Samala DLH Seraya 5 STIM kelabuSymplocos SPHQ 1 lateviridis DLH 5 7 DHH Katok SBU 9 12 NDerSymplocos 11 5 2 polyandra SRYA Merbau Sepetir DLH 3 10 lalat 7Symplocos MEBG NFMU SR Kelumpang sp. 2 9 MPK F SBDH 2 1 12 12 6Symplocos 1 DLH - sp. SBH 1 DLH 1Tectona 5 Jiak grandis 2 2 4 2 DHH 2 2 55Teijsmanniodendron 3 sp. 7 1 Poroi 2 - KWNG 10 untu 12 5Teijsmanniodendron 5 KW SBGA 4 sp. - Mogkulat 1 SRG 2Terminalia 117.8 DLH copelandii 40 1 2 1 SKGS MSSG DLH SKBATerminalia m-l 6 6 sp. - 2 MPS 31 11 SKA SKU -Tetramerista 2 5 7 6 7 DLH glabra - Buak-buak DLH DLH 6 79.6 53Thespesia 2 - 2 9 Kemenyan 1 populnea Buak-buak 2 jarietek SKBU 5 6 47.7Timonius 2 61 Lobo flavescens 2 921 34.33 Jati SKK 2 3 - 3 l 2Toona 63 2 146.4 sp. - Talisai DLH paya 26.2 - 3 1 6 2 1Trema 2 - 3 orientalis 1 STCO MLAL - 146.4 9 4 2Trigonopleura MBL OTH 6 malayana 558 KLPG 146.4 2 4 2 6 - m-l NDMH OTHR KTOK 2 - - TuyutTristania KPG clementis 19 1 KTK 14 SEPT 1 - l NDMH 1 1Unknown OTHR - 2 Talisai Baru 16 1 16 SPT laut 499 3 3 3 14 - Tapai-tapaiUpuna 1 1 borneensis 64 NDLH 1 60 26 2 2 1 3 - 520Vatica 14 2 0 2 SYMP Gambir albiramis 1 2 1 125 hutan 20 581 15 79.6 15 OTHVatica 1 0 12 JIAK bancana SYMP 2 OTHR 9 175.1 TEPT - 1 1 - -Vatica 2 OTH 15 14 dulitensis Randagong JAK 60 250 0 34.33 BUJ 1 2 1 OTHR - 4 2 -... 1 22 OTHR 1532 39 Pelawan-pelawan 9 14 9 NDMH BUAK 589 14 Surian 1454 16 0 1 5 BU 1 33 36 2 LOBO 4 5488 - - - - 1 2 0 8149 LOBOOTHR 194 76.4 15 1 57.3 NDMH 2 1 - 4 Upun 6 1 TECO 499 6 14 16 3 1 30 6 TLP 2 15 12 47.7 - 4 47.7 26.06 15 Resak 23.43 15 210 NDLH 20 putih 3 1 Unknown - 2 803 6 654 1 1 KEME 2 2 21 - 15 2 2 680 OTH JATI 0 21.49 1130 22 Resak 79.6 792 banka OTHR - Resak 9 2 2 3 bukit 15 15 3 15 s JTI 14 1 2 1000 1 3 2 5 1 89 1 28.52 - 702 19282 8 428 TIFL NDMH 560 TUYT 2 THPO 1 25 3 18 TRMA 1 TALI 598 8 1 TUY BRL b s GRH 12.7 - OTH & NDLH 245 1988 t 0 6 1274 - NDLH TLI 1 OTHR OTHR 4 2 PLWN 16 412 15 4 14 - 14 14 8 1 PP - 10.84 2 NDMH - 450 3466 4 18 999 l RAND 2 15 - 2 - 13 NDMH RANDOTHR 511 - 1 1 1 7 9 8 19 262 b - 2 14 1280.5 & 4 725 10 t 3 3 2 - 2 369 3 15 8963 - 15 15 15 2 l 3 1 - 2 578 79.6 1 3 8 2 TOON - 1 - - - 5 - 1249 SU - 3 15 - 4 14 18 40.84 27 10 4 89 - REPU - UPUN 1 NDLH RBT - - UP 430 REBA l UNKN 18 600 - REBU DHH 89 OTH REBA m DHH 27 - 38.2 1 RBK - 245 DLH OTHR 57.3 7 DHH 1 4 14 1 7 725 2 - 19.14 - 245 95.5 7 23.43 999 1 38.2 66.8 15 1 0 2 600 - - 3 2 720 0 42.2 999 0 1 19.14 1 41.73 2 15 3 625 89 1 30 3 791 - 1 - 730 3 0 - 3 - - 0 1 0 2586 245 66.8 1 8 0 24 0 1 1 46 1 1 27 1 24.95 999 89 3 0 3 - - 174 25.5 368 125 - 0 3 191 4 38.2 245 0 22.66 368 2 0 - 30.09 - 2867 893 - 999 - 2 2433 2 824 - 0 0 1 - 1 2 - 1 768 995 - 0 25 1 1 0 89 50 3 0 245 0 82 - 999 0 2 0 - 2 1 Lists of tree species 183 Botanical name... Vatica maritimaVatica oblongifoliaVatica odorataVatica sarawakensis Local nameVatica sp.Vatica/cotylelobium sp.Viburnum amplificatumVitex pubescens Resak Resak laut daunWeinmannia panjang blumei ResakWendlandia sarawak dasythyrsa Resak ResakWetria biabas macrophylla Ranuk PFTWikstroemia tenuiramis SS 10Xanthophyllum sp. HG 3Xanthophyllum CFRC sp. Malitap Resak bukit Sumu-silan CHQXerospermum degong sp. Kulimpapa 4 TGRP 7 7Xylosma sumatrana IGRP Tindot F3 RambaiZizyphus REDP hutan angustifolius 13 3 GGRP 3 RBPN.N. Height DHH 3N.N. Dia Minyak 3 10 3 berukN.N. 7 4 Minyak beruk 5 4 3 H=f(d)N.N. RESK RELT 10 Den RBW RBU Gurulau 4 LinauN.N. Monsit 3 REBI 2 DHH 2 4 3 DHH 3N.N. 4 SPHQ RBS 7 2 3 3 7 NDerN.N. DHH RESA 4 2 1 3N.N. 4 RB NFMU VIAM 6 7 2 F 2 RESAN.N. WEDA 2 2 OTH OTH 6 DHH 2 31 RBD OTHR 3N.N. OTHR WEBL 2 DHH 14 3 2 KULI 7 14 OTH 3N.N. WEMA 2 OTHR OTH 2 7 KULIN.N. 14 3 - OTHR NDHH 1 3 1 Buah-buah 14 14 2N.N. 3 6 TIND 15.3 19 N.N. XANT 3 1 15N.N. 2 6 15 OTH MNY 1 1 3 N.N. XAEL NDMH OTHRN.N. 2 - - - 20 XAEL 15 14 N.N. 3 2 OTHRN.N. - 2 15 15 14 s s N.N. 3N.N. 1 1 2 - N.N. 3 12.2N.N. - 1 858 50.9 OTHR s-l N.N. s 9.5 15 GU 15 MSITN.N. - - XYSU - 19.1 N.N. 6 15 36.22 MSTN.N. 1 OTH NDMH - N.N. 14 NDLH OTHR - -N.N. - - s-l 13.19 s 14 14 - N.N. 2N.N. - s-l - 1 - - N.N. 4N.N. 9 1 1 3 - - N.N. - - 1 15 1 10 - - - N.N. 15 15 OTHR 1 2 9 - - 89 BHN Adarah 3 129 9 OTHR - s-m - N.N. 4 800 0 - 14 0 2 9 s 18 - 4 - - N.N. 2 0 245 2 9 89 CEMP - 89 N.N. 1 1 4 DHHX CEMPOTHR 2 9 1 - DHHXDHH N.N. - 14 0 4 25 1 999 2 9 - - 15 DLHX - 89 Dryepetes 245 0 4 10 245 DLHX 2 9 DLH DMHX - 1 1448 N.N. 4 - DMHXDMH 2 8 1 2060 DURM 1 1 999 3 1 - N.N. 0 - 4 5 999 DURMOTHR 245 15 2177 2 - 9 89 14 EURO 1 1 N.N. 4 8934 EUROOTHR 9 1 9 15 - KARA - 3 N.N. 14 4 - 1 999 KARAOTHR 1 2 1 - 1424 KTUN - 0 - 14 245 - 4 KTUNOTHR 2 2 - 9 0 KURI 15 1 14 - 15 4 15 6838 - 9 KURI 1 MACX 1 999 4 OTHR 4 MACXMACA 9 2 9 15 14 14 - - MEDP 1 - 89 - - MEDPOTHR 15 - 2 1 9 0 - ADAR NDX - 14 - 4 OTH 2 1 15 1 2 0 - OTHR NDX 4 245 2 9 14 NDLH 1 ANJA - - 4 15 - 15 4 - 0 - 14 - 9 - OTH BNKL - - 4 OTHR 999 15 1 0 2 OTH 9 DRYP BRUN 14 - OTHR 1 OTH OTH - 2 - 8 DARU 14 - - 15 1 OTHR OTHR 4 100 - - OTH - 2 - 14 1 14 - - 15 - OTHR 4 - 1 1 14 KUWG - OTH 4 - 15 - - 1 - 1 OTHR LPDA - 1 125 4 - 15 - 14 - 1 - OTH - OTHX - 1 15 15 OTHR OTHXOTHR - PIOX 14 125 - 14 15 1 - - - - 0 PIOX - PION 1 - 1 1 - - 1 - 1 - 15 - 0 14 - - 1 15 - - 15 - 0 - 1 1 0 - - 0 0 - - 0 - 1 - 15 - - 0 - - - 1 1 - 0 0 0 - 0 - - - - 0 1 - - 0 - - - 0 0 89 0 - - - - - 1 0 - - 0 - - - - 89 245 - 0 0 - 89 0 - - 0 89 89 999 - 245 89 - - - 245 - - 245 0 245 999 89 - 245 999 - 89 1 999 999 - 245 - 999 - - - 1 245 999 0 - 999 - 0 0 - 0 - - 184 Appendix B

Table B.3: Legend of tree species lists of Venezuela (Table B.4).

Column Description

Family Family name Species Species name PFT Plant functional type after Table 6.1 SS Successional status after Table 6.1 HG Height group after Table 6.1 Code Species code for identification

Abund. Qualitative abundance in compartment in which species mature (VR: very rare; R: rare; C: common; VC: very common) Phen. Different phenologies (EV: evergreen; SD: semi-decidous; DE: de- ciduous) Dispersal a. Different dipersal agents (ZO: zoochor; ZB: zoochor or barochor; AN: anemochor; AB: anemochor or barochor) N-Trees Number of trees (diameter≥10cm) in 4.6ha in Caparo N-Sapl. Number of saplings (height≥130 cm; diameter<10cm) in 1.15 ha in Caparo N-Seedl. Number of seedlings (30cm≤Syagrus 3 Guana 4 sancona 2 PFT 4Bixaceae N.N. Jacaranda SS 2 R rhombifoliaBombacaceae Cojbe Tabebuia 6 Yatag rosea Gatea HGBombacaceae 4 R Code C VRBombacaceae Bombacopsis Mapor EV quinata 12 Bixa Joboo Abund.Phenol.DispersalBoraginaceae orellana R Ceiba a. pentandra 11 SD 4Burseraceae Ochroma N-Trees C SD 7 ZO lagopus SDCaesalpinaceae Cordia 4 apurensis 12 N-Sapl. 6 EVCaesalpinaceae AN 2 ZB Bauhinia ZB Protium sp.1 crenatum DE N-Seedl. 5Caesalpinaceae 4 9 Caesalpinia Palag granadillo 9 Density 3 ZOCaesalpinaceae Cassia grandis 2 Palcu 6 VC 3Capparidaceae ZO 2 Swartzia Sanro leptopetala 2Capparidaceae VC 9 1 1 1 Palsa Capparis 6 badocea EV VRCaricaceae 5 4 4 20 4 Crataeva tapia CChrysobalanaceae 2 10 2 EV Patav Hirtella 8 3 americana 1 Saqui 10 5Chrysobalanaceae 2 EV ZO Apama Licania VR 38 sp.7 3 4 1Clusiaceae Carica VR sp.1 C ZO 3 3 3 22 EV 2 76Cochlospermaceae 10 12 Cochlospermum AN 4 vitifolium EV Ceiba Tabaq 2Combretaceae - 3 2 4 9 Balso Onoto - VR 3 ZO VRElaeocarpaceae DE Rheedia 3 Parne 2 9 madrono 3 C 576 VC Terminalia ZB guianensis... Grana 1 Triac 2 61 Muntingia C 4 calabura 6 3 8 - 9 - SD VR Toson AN AN 5 3 C 2 Lecho SD 4 Orura R SD 1 3 - 403 1 VR SD - 3 DE AN VR 11 68 Canfi AN 822 AN EV - 2 3 ZB - Zalre 10 - 2 EV DE 356 VR ZB AN EV 16 - 8 9 Botot Guaja VC 496 1 ZO 3 Zorro - 13 - ZO ZO SD R R ZB 1 3 VR 4 2 - SD 44 111 - 33 5 3 - ZO 29 4 Guaya 24 - 1 EV SD DE ZB 2 C 20 Madro 6 Nigui - 15 - - VR 2 ZB AN ZB 1 1 40 VR 2 26 - SD - - - SD 12 - - SD - 3 AN - 2 3 ZO 20 - 12 11 - 2 43 ZO - - 6 1 6 161 222 2 - - - 38 - 27 16 - - 2 7 - - 1 - 190 17 - - 27 - - 19 5 - - - 67 - 254 18 - 1 - - - 186 Appendix B Family... Eleaocarpaceae SpeciesEuphorbiaceae Sloenea ternifloraEuphorbiaceae Acalypha diversifoliaEuphorbiaceae Adelia recinellaEuphorbiaceae Phyllantus suripaensisFlacourtiaceae Sapium stylareFlacourtiaceae Banara sp.1Flacourtiaceae Casearia nitidaFlacourtiaceae 1 8 Laetia americana 2Flacourticaceae 5 Xylosma sp.8Hypericaceae 2 3 Casearia sp.4 PFT 1Lauraceae 4 SS 3 1 3 3Lauraceae Vismia macrophylla Pelud HG 2Lecythidaceae 2 9 2 Code Palne 3 Picap RLecythidaceae Ocotea 2 glomerata Coroupita 2 VC Abund.Phenol.DispersalMalvaceae C guianensis Cirlo Pleurothryrium a. 4 2 Barti 5 aff.reflexum GustaveaMelastomaceae angustifolia N-Trees Clabl SD Clane VR VRMelastomaceae 4 EV 2 Micania 3 R N-Sapl. sp.1 SD 10 VRMelastomaceae 4 Malachra sp.1 Micania sp.1 ZO SD Leche N-Seedl.Meliaceae 2 SD AB 2 3 Mouriri barinensis 2 Density Meliaceae 2 SD C ZO 8 SD 2 2 Frupa 8Meliaceae 5 Ranch ZO 4 - 1 2 RMeliaceae VR 1 3 ZB Cedrela 1 AB odorata DE 3 LauamMeliaceae 3 Benao - Guarea Guapa trichiloides - C CardiMeliaceae 3 SD Melas R Swietenia VR macrophylla 27 3 SD 2Meliaceae ZO R Trichilia R maynasiana 1 Cocmo 8Mimosaceae Trichilia Laune - pallida SD Radef SD R 1 2Mimosaceae ZO 11 EV Trichilia palmerorum VR 3 R 3Mimosaceae DE 910 EV Trichilia 2 Albizia unifoliata caribaea ZO ZOMimosaceae 9 ZO SD Cassia 8 38 9 3 SD - alataMimosaceae ZO ZO Inga EV 71 marginata 5 -Mimosaceae 7 2 1757 Perhu Inga 7 3 2 ZO 6 sp.5 1Mimosaceae ZO Inga - R 8 ZO sp.5 3... 4 43 3 18 4 2 Inga 5 sp.5 3 - - - 69 Inga 2 8 3 - sp.8 3 Caoba - Tromp Cedro - 4 SD 2 3 49 9 3 VR - R - 6 Guram VR 3 2 - 3 - 3 2 24 R Gudec 7 2 2 31 ZO 2 EV 4 2 - 4 Cedri - Gublo VR 3 DE 3 EV 15 Cafec 3 Gucar - 18 VR - 4 C EV 2 AN Rejit 2 Guarm - 5 VR SD VR AN 70 2 ZB R - Carab 21 20 SD C 2 2 - 9 11 EV VR SD EV 5 ZO 713 - Guamo Mucut ZO SD VR DE - ZO 614 SD - 2 VR ZB ZO - - 1 34 7 - ZO 1 SD ZB - AB SD - 6 - 1 ZO 1 - - 28 ZB - 1 1 - 5 - 6 24 1 1 59 22 30 - 1 - - 7 2 13 - 23 93 78 3 1 42 7 - 4 - - - 1 - 118 54 - 4 - - - 3 ------Lists of tree species 187 Family... Mimosaceae SpeciesMimosaceaeMimosaceae Inga sp.8Mimosaceae Inga sp.8Mimosaceae Inga sp.8 3Mimosaceae Piptadenia peregrina 3Moraceae Pithecellobium 3 guachapele 3Moraceae Pithecellobium 3 sp.8Moraceae Guano 3Moraceae VR Guane Brosimum alicastrum 3 9Moraceae Guamc C Cecropia PFT peltataMoraceae 10 VR SS SD Chlorophora 3 tinctoraMyrsinaceae 2 Clarisia HG 3 bifloraMyrsinaceae SD SD Cazab Code Ficus ZO insipida 4Myrtaceae Rapanea 4 Trophis VR racemosa sp.5 10 Abund.Phenol.DispersalMyrtaceae a. ZO Stylogene Masag ZO venezolanaNyctaginaceae N-Trees Yopio 3 VR EV 3 8Papilionaceae N.N. VR 6 Torrubia N-Sapl. olfersianaPapilionaceae Myrcia 4 2 sp.5 2 SD ZB N-Seedl. 3 FissicalyxPapilionaceae fendleri DE 1 Density Chaam LonchocarpusPapilionaceae 31 - 10 pictus Cucha 3 C 3 5 Lonchocarpus ZBPapilionaceae 10 sericeus VR 3 10 ZB 3 LonchocarpusPapilionaceae 2 sp.8 2 Moraa 3 3 LonchocarpusPapilionaceae Yagru 3 sp.8 2 IM EV VR 4 31 5 Guato OrmosiaPapilionaceae macrocalyx 3 VC - 4 2 4 PlatymysciumPapilionaceae VR pinnatum 3 Chane 9 - 9 SD ZB - 3 PoeppigiaPiperaceae 10 3 procera Chabl 1 Morti SD R 22 Higue PterocarpusPiperaceae 3 SD acapulensis 57 VR 2 5 3 2 2 R Laubl ZO RPiperaceae 7 ZO 2 JebecPiperaceae Piper R 9 EV 4 aduncum 4 4 Casav ZB EV 7 1Piperaceae Piper R 2 48 - sp.1 SD - VR EVPiperaceae Piper Tasac 2 Zapca Cuesa - - sp.1 ZB 7 2 8 ZB SDPolygonaceae Piper 3 sp.1 C 2 R C 1 1 SD ZO - 4 ZB SD... Piper 127 sp.1 2 2 3 Paujo 84 Piper Coccoloba - ZB 1 1 sp.4 2 fallax 2 Roble - ZB R DE SD 1 DE 3 Pioni ZB 2 1 - Cobla R 5 1 1 2 1 101 2 4 VR Pixxx Sadra 1 R ZB ZB 3 ZB 4 93 EV Pipxx Frijo 2 R VR 42 - 3 SD 9 EV Pipex - VR 2 - VR - Pxxxx EV - ZO 8 VR SD 1 1 - EV VR AN ZB 5 EV 27 - 1 - 14 23 52 27 ZB Corne EV AN 2 EV ZB 3 VC ZB 7 - 2 - 25 ZB 4 - ZB 2 - 15 ZB 4 EV 1 8 12 - 38 4 - Uvero - 2 ZB - 30 R - - 12 2 - - - 33 32 - 3 - 30 2 - SD ------ZO - - - - 4 1 - - 5 - 3 13 1 - 5 2 - 1 - - - - 1028 - - 16 - - - - 9 - 188 Appendix B Family... Polygonaceae SpeciesPolygonaceaePolygonaceae Coccoloba padiformisRubiaceae Symmeria paniculataRubiaceae Triplaris caracasanaRubiaceaeRubiaceae N.N.Rubiaceae Chomelia 8 polianthaRutaceae Genipa 9 americanaRutaceae Hamelia 3 patens 7Sapindaceae Psychotria 2 carthagensis PFTSapotaceae 3 SS Zanthoxylum 2 sp.4Sapotaceae 4 Cupanea HG Zanthoxylum americana sp.7 CacaiSapotaceae 1 Code 3 2 ChrysophyllumSterculiaceae Palua auratum C 2 8 2 Pouteria Abund.Phenol.DispersalSterculiaceae anibaefolia 2 a. R Palma 2 SerjaniaTheophrastaceae N-Trees sp.8 Guazuma C tomentosa 3 2 3 3 Clavija 1Tiliaceae EV longifolia Herranea 1 Mapon sp.2 N-Sapl.Tiliaceae SD 8 3 VR Tachu 5 1 3 3 Garba N-Seedl.Tiliaceae EV AB 2 C 3 Density VR 3Ulmaceae 3 Hierr AN Carau SD 3 1 Goethalsia 10 meianthaUrticaceae 1 1 AN VR Heliocarpus Chubl VC 7 3 popayanensis SDUrticaceae SD 2 Coral 3 ZB Luecea R Manza cymulosaVerbenaceae Cacao Trema SD Caimi micrantha C 37 2 2 VR EV RpavaViolaceae 4 Urera C ZO caracasana AN 7 CVochysiaceae VR Urera Vitex sp.1 3 3 EV orinocensis 3 38 Churo ZO SDN.N. ZO 7 SD SD CN.N. Vochysia SD 1 lehmanii Guabl 2 55 Hybanthus 1 1 SD prunifolius ZON.N. - 2 C ZO - 1 24N.N. 1 ZO 2 Sancr ZB 92 7 EV ZO 3 3 - C - Pring 4 Majag 57 Cordia SD 1 sp. 2 Guaco ZO C 1 N.N. 7 R 3 2 39 VR 2 1 1 N.N. - 203 EV 3 AN 9 - - 2 - N.N. 2 19 3 1 2 751 DE 47 SD DE Guaci 1 2 ZO Bocho 3 1 29 C 59 Ortig - 576 VR AN ZO - - AN 4 21 Guaro R 1 Campa 32 20 24 11 1025 VC - R EV Murci SD - 725 - 11 VR 8 DE - DE ZO 4 ZO - - 42 SD 39 3 - - 1 9 1 DE 3 AN ZB - 1 53 ZB 4 3 77 ZB - 10 5 2 - 25 2 - - 28 12 3 Candi - - 2 1 VR 5 3 - - 2 37 7 Limon - 2 497 1 SD 56 - - Fruto R - Caven VR 46 ZO - R 399 SD 3 - - - 26 - - - SD - 2 373 - - 13 ZB 831 14 - - - - - 3 - 4 1 3 1 25 4 1 - 51 - - - - Lists of tree species 189

Table B.5: Legend of tree species lists of French Guiana (Table B.6).

Column Description

Family Family name Genus Genus name Species Species name PFT Plant functional type after Table 8.2 SS Successional status after Table 8.2 HG Height group after Table 8.2 Code Species code for identification

NNOU Number of trees with diameter >10cm in Nouragues NPSE Number of trees with diameter >10cm in Piste de Saint-Elie NPAR Number of trees with diameter >10cm in Paracou NTOT Total number of trees with diameter >10cm in Nouragues, Piste de Saint-Elie and Paracou VRI Van Roosmalen index of adundance (0: very rare; 1: rare; 2: fairly rare; 3: not common; 4: fairly common; 5: common; 6: very common). The average number of individuals per species should roughly follow Nexp(−a · VRI) Height Maximum height at maturity [m] Disp. Dispersion strategies (AN: anemochorous; HY: hydrochorous; Z: zoochorous; EZ: endozoochorous (seed eaten, then defecated); SZ: synzoochorous (seed transported)) by different processes (B: birds; M: monkeys; bat: bats; R: rodents; T:tortoises). Generally, the dis- persing capacity of each of these modes can be roughly estimated in terms of distance HYElaeis 2Arecaceae humilis 2 2 Euterpe NPSE 4Arecaceae 39 BACA Euterpe 3 ASTVArecaceae maraja NPAR Geonoma 0 1Arecaceae oligocarpa 0 3 1 4 NTOT Geonoma 3Arecaceae 86 simplicifrons 1 Geonoma VRI 1 BACBArecaceae Geonoma 3 oleifera 4 oleracea BACC 2 0Arecaceae 0 Geonoma BACD 0 precatoria HeightArecaceae 0 0 baculifera 1 Geonoma 4 3 1 8 0 3Arecaceae deversa Disp. Geonoma BACGArecaceae leptospadix BACE 0 Hyospathe 1 0 3 2Arecaceae 3 Density 0 0 maxima Jessenia 125 0 0 Architecture Arecaceae 12 2 BACH poiteauana 16 0 1 1 Manicaria 2 4Arecaceae stricta 0 Mauritia 3 0ArecaceaeOrbignyasagotii321ORBS000045SZMR-- 3 BACO 5 BACS 4 0 umbraculiformis BACN 1 Mauritiella 0 0 4 elegans 0Arecaceae 3 0 0 3 Maximiliana 0 0 4 0Arecaceae 1 4 4 3 4 Oenocarpus 1 bataua 0 3 sacciferaArecaceae EUTE 5 EUTP 12 Orbignya 0 ELAE 0 5Arecaceae 1 GEOB 0 flexuosa 3 10 0 3 1 martiana 0 3 0 0 maripaBignoniaceae 1 4 0 0 Scheleea 0 GEOLBignoniaceae 1 1 GEOD bacaba 0 4 1 SZ Socratea 6 0Bignoniaceae 1 3 0 0 9 1 Synagrus GEOU GEOP 0 5 GEOMBignoniaceae 7 0 0 polysticha Jacaranda 11 0 0 Synagrus 1 1 3 1 0 0Bignoniaceae 0 0 Jacaranda - 6 14 0Bignoniaceae EZB 2 1 Tabebuia 2 0 3 sp. GEOS 10 0 5 SZ 0Bixaceae Tabebuia 5 3 1 exorrhiza 0 0 1 3 HYOS 0 1 11 2 0 4... 0 - Tabebuia inajai 0 0 1 7 0 4 copaia - 4 4 0 Tabebuia stratincola 2 JESS MANA obtusifolia 2 - 0 EZB 3 0 3 11 0 4 MAUF EZB 3 MAUM capitata 0 1 5 14 5 3 0 EZB 1 MAXI 0 - 0 0 Cochlospermum 0 0 fluviatilis 0 - 7 1 4 0 - insignis OENO 1 0 - 6 - 8 0 EZB 8 serratifolia 10 orinocense 15 ORBI EZB 3 26 5 2 8 2 14 4 12 0 4 0 0 4 7 2 0 - 0 1 3 - EZB 0 3 14 2 - 1 - 13 2 11 3 20 0 - 1 0 3 2 25 2 EZB SOCR 27 EZB 1 4 - EZB 15 0 5 8 2 17 2 1 0 4 5 0 1 JACO SYNS 6 5 SCHN - 0 0 4 JACC EZBM - - 2 - SYNI 3 3 1 - 0 15 3 0 3 0 40 0 EZBM - 16 TABC 4 1 EZ 0 - TABF 5 - 3 0 EZ 5 2 2 3 0 0 1 3 14 0 TABS - 3 TABI EZ - 3 - COCO 0 EZ 0 28 11 4 - - 14 - 0 0 2 0 EZ 0 EZ 5 - EZ 0 5 6 0 - 2 0 - 6 4 42 20 0 0 0 0 - - - - - 0 2 EZ 0 11 35 0 18 - 6 EZ - 0 69 EZBM - EZ 20 0 0 4 4 - - - 0 - 0 EZ - 0 SZ SZMR - 0 5 5 - - 0 EZBM - 5 1 5 18 - 0.8 - - - 0 2 4 - 5 12 35 SZ - 5 - Corner 6 20 14 6 - - 3 5 12 EZBM AN - 35 18 AN - EZ 30 SZ EZ 0.45 15 40 AN - - HY - AN - Corner Scarrone - - AN AN - 01.01.00 - - Koriba - - 01.01.00 - - - - - Lists of tree species 193 Family... BixaceaeBombacaceae GenusBombacaceaeBombacaceaeCeibapentandra1715CEIB4004550AN0.4Massart Bombacaceae Bombacopsis CochlospermumBombacaceae CatostemmaBombacaceaeBombacaceae vitifolium Eriotheca nervosa SpeciesBombacaceae Eriotheca fragransBombacaceae EriothecaBombacaceae PachiraBombacaceae Pachira 8 crassaBoraginaceae Quararibea gobosa 13Boraginaceae PFT Quararibea surinamensis 15 3 SGBoraginaceae Quararibea 0 HGBoraginaceae Cordia aquatica 2 Code 2Boraginaceae guianensis Cordia 4 insignisBoraginaceae lasiocalyx 19 COCV Cordia 4 NNOU 11 BOMBBoraginaceae turbinata Cordia 0 12 NPSE CATO 3Boraginaceae 1 Cordia 2Boraginaceae 0 NPAR Cordia 3 7 11 5 bicolorBoraginaceae Cordia 3 NTOT 8 0 3 fulvaBoraginaceae 14 Cordia ERIS 2 0 VRI 2 8 ERIC goeldianaBurseraceae Cordia 13 ERIO 1 hirta 3Burseraceae 3 0 Cordia 2 0 Height laevifronsBurseraceae 3 0 Cordia 0 4 2 PACA 0 lomatolobaBurseraceae QUAG Disp. 69 Crepidospermum 8 2 nervosa 0Burseraceae PACH 1 QUAL 0 4 Dacryodes nodosa 0Burseraceae Density QUAT 0 12 rhoifolium 0 23 Protium 0 Architecture 3 panicularisBurseraceae 1 82 193 7 Protium 3 sagotiiBurseraceae 12 3 0 12 2 0 Protium 0 tetrandraBurseraceae 0 1 67 Protium 3 0 3 2Burseraceae 0 nitens 0 1 CORB 3 6 4 Protium 12 3Burseraceae 12 CORG 3 CORF 2 0 8 Protium altsonii 0 3Burseraceae 0 0 5 3 1 90 Protium 0 apiculatum 0 3Burseraceae CORM CORL 0 0 CORH 0 12 12 Protium 7 aracouchini 3Burseraceae 32 0 3 35 2 0 3 Protium 0 decandrum 0Burseraceae 0 3 1 2 4 Protium 3 0 demerarense CREP CORNBurseraceae 2 10 CORP 0 4 AN Protium 0 giganteum 8 0 EZMBurseraceae 193 3 0 CORO 1 AN 11 0 0 2 Protium 8 guianense 1 0Burseraceae 0 0 0 Protium 6 CORT heptaphyllum 0.7 16 5Burseraceae 25 3 - 2 40 0 CORS 16 3 4 0 Protium 20 - opacumBurseraceae 0 3 3 1 0 3 0 Protium 0 2 3 plagiocarpum 0... 0 DACR 16 3 12 2 - 0 Protium 0 0 11 polybotryum AN 25 AN 4 0 PROA 15 7 PRNZ Tetragastris 5 AN - 4 robustum 3 12 0 PROB - 0 35 2 10 Tetragastris 8 sagotianum 0 8 PROC 0 0 0 12 0 PROD 0 6 4 0.58 2 0.58 subserratum 0 SZR 16 11 0 23 3 3 0.58 SZ 23 16 tenuifolium SZ - 3 PROF 2 altissima 1 SZR 0 trifoliolatum 0 Rauh PROH 3 2 Rauh SZ 0 hostmannii 3 - 0 16 7 1 Rauh 0 16 0 2 4 PROG 32 5 - 23 16 1 4 3 0 12 0 - 0 - 0 3 PROQ 3 - 16 20 - 0 3 PROO PROR 0 8 0 0 8 1 4 - 0 0 5 0 4 3 6 16 25 11 0 0 3 0 EZBM 4 - 20 0 0 16 - - 3 PROS 0 PROT - EZBM 3 2 4 EZBM PROU 10 - 5 1 3 0 0 2 EZBM - 6 EZBM 38 0 25 5 - 20 10 0 4 3 PROV 0 EZBM 0 26 1 0.4 0 23 - PROW 0 0 - TETA TETH 2 Prevost 1 18 0 5 EZ 4 0 1 EZBM EZBM 17 EZBM 0 Prevost 7 10 31 9 Prevost Prevost 0 0 - 20 - 0.45 4 - Prevost 0 0 Prevost 15 - EZBM 8 0 0 10 EZBM 0 0 5 20 0 Prevost - - 20 EZ 0 0 35 - 27 Prevost Prevost EZBMR 3 0 0.7 0 - EZBMR EZBMR 0 1 0.6 7 2 20 0.7 29 0 - 30 Prevost EZBMR - 5 EZBMR 0.7 Prevost 0.7 Rauh 10 0 Rauh 9 Rauh EZBMR 17 - - 5 8 EZBMR 0.6 0 - Rauh 0.7 24 Rauh 30 EZBMR 5 0.7 4 - Rauh 30 Rauh 30 30 EZBMR 5 EZBMR EZBMR 0.7 - 0.7 Rauh 35 30 EZBMR 15 EZBMR EZBMR Rauh 0.7 0.7 0.7 25 Rauh - EZBMR 0.7 EZMT EZBMR Rauh Rauh Rauh 0.7 EZMT 0.85 - Rauh Rauh Rauh - 194 Appendix B 01.01.22 - CaesalpiniaceaeCaesalpiniaceaeCaesalpiniaceae CandolleodendronCaesalpiniaceae Cassia brachystachyumCaesalpiniaceae CassiaCaesalpiniaceae Cassia 8Caesalpiniaceae CassiaCaesalpiniaceae Chamaecrista 3Caesalpiniaceae CopaiferaCaesalpiniaceae cowanii Crudia 2Caesalpiniaceae fastuosa Crudia apoucouitaCaesalpiniaceae CANB grandis CrudiaCaesalpiniaceae spruceana Crudia 0 guianensisCaesalpiniaceae CynometraCaesalpiniaceae Cynometra 8 16 aromaticaCaesalpiniaceae Cynometra 12 0 glaberrimaCaesalpiniaceae Dialium 3 12 3 7 oblongaCaesalpiniaceae 3 Dicorynia 18 hostmanniana spicataCaesalpiniaceae 4 Dimorphandra marginata 2 3 3 2Caesalpiniaceae 2 0 Dimorphandra parviflora 12 CASQ CHAMCaesalpiniaceae 3 11 16 Elizabetha CASR 2 5 polyandraCaesalpiniaceae 0 0 Eperua 3 guianense CAST guianensis 0 pullei 2Caesalpiniaceae 3 CASS Eperua 8 COPA 0 12 3Caesalpiniaceae 0 Eperua 3 3 0 4 0 8Caesalpiniaceae Eperua 0 3 princeps 0 CRUA 3Caesalpiniaceae CRUG Heterostemon CYNH 0 15 0 2Caesalpiniaceae 0 15 3 3 2 18 Hymenaea 0 0 0 falcataCaesalpiniaceae 1 2 0 Lecointea 1 CYNM 2 CRUO grandiflora 2 0 0Caesalpiniaceae 2 19 mimosoides Macrolobium 0 jenmanii 4 0Caesalpiniaceae 0 CRUS 0 Macrolobium CYNP 15 4 16 5 rubiginosa 101 0 3Caesalpiniaceae 0 Macrolobium DIMP 0 0 courbaril 0 0Caesalpiniaceae DIAL Macrolobium DICO 0 0 3 acaciifolium 5 amazonica 0Caesalpiniaceae 15 8 0 Martiodendron 0 0 angustifolium 0 EZ 15 0 0 81Caesalpiniaceae 4 Paloue DIMQ 0 0 bifolium 2 16 1Caesalpiniaceae 2 0 0 15 Paloue 3 2 multijugum 0 ELIZ parviflorum 1 1Caesalpiniaceae 11 Peltogyne - 0 4 15 15 1 3 0 54 101 0Caesalpiniaceae 2 2 4 0 12 Peltogyne 1 0 0 1... 2 EPEG Recordoxylon 2 4 2 6 0 3 0 4 HETE EPEF 3 0 Sclerolobium 15 30 11 5 1 16 11 3 0 117 - 4 EPEJ guianensis 4 0 25 676 0 EPER 0 3 3 0 paniculata riparia 0 2 5 speciosum MACB 25 3 2 HYMC 0 MACC 75 venosa 40 2 15 0 EZ HY LECA 252 0 0 albiflorum 3 289 3 0 25 89 4 EZ 3 0 0 4 0 1 25 1 EZ 8 MACH 0 0 MART 27 MACD - EZM - - 0 EZ 18 489 3 0 4 15 - 0 136 AUT 0 226 0 0 0 3 4 0.75 8 2 0 AUT 15 20 - 0 12 16 2 5 SZR - - 1454 0 0 2 0 316 - 0 5 40 - 5 2 - 0 10 3 - 4 0 0 3 - 3 0 0 - PALO 5 SZR 30 PELP HY 0 4 0 2 - RECO 3 4 35 0 - 4 - AN 75 0 0 0 SZR SCLA 0 AUT - PALR 0 - PELV 0 0 - 50 3 0 EZM 32 - 0 0 1 0 EZM 0 0.78 - - 35 4 0 29 - 1 136 1 0 0 EZM - 01.05.00 0 5 SZ Troll 6 - 6 - 0 0 15 SZ 0 6 - - - EZM 5 30 0 1 - 30 82 0.85 4 4 17 - 35 30 EZ 0.92 0 0 25 4 SZ 25 - Troll 0 SZ 2 82 Troll HY 20 SZRM - HY 35 - 0.85 EZ 0 0 0.83 HY 10 0.92 3 3 - 6 - HY AN Troll Troll - - 0.75 Troll 5 6 2 - 30 8 - - 40 - - - 30 35 10 EZ - EZM HY - 0.9 AN EZM HY 0.95 - 0.84 0.55 - - Troll Troll - - - Family... BurseraceaeBurseraceae GenusBurseraceaeBurseraceae TetragastrisCaesalpiniaceae TrattinnickiaCaesalpiniaceae Trattinnickia Bauhinia Trattinnickia Bocoa panamensis Species burserifolia demerarae rhoifolia cinnamonea 12 6 3 PFT 14 prouacensis SG 1 3 HG 4 1 15 Code 2 TETP 4 2 3 16 TRAB NNOU 2 TRAD 4 NPSE 0 1 3 0 NPAR TRAR BAUH 4 2 0 NTOT 0 0 BOCO VRI 0 29 2 0 Height 0 0 Disp. 0 60 Density 6 0 0 Architecture 0 257 0 6 0 0 346 6 5 5 25 5 2 10 30 EZMT 35 35 5 EZ 0.85 EZ EZM+Szbat Rauh BA - EZ ------Lists of tree species 195 Family... CaesalpiniaceaeCaesalpiniaceae GenusCaesalpiniaceae SclerolobiumCaesalpiniaceae SclerolobiumCaesalpiniaceae SennaCaesalpiniaceae Senna guianenseCaesalpiniaceae Senna melinoniiCaesalpiniaceae Senna SpeciesCaesalpiniaceae SwartziaCaesalpiniaceae SwartziaCaesalpiniaceae 18 Swartzia multijugaCaesalpiniaceae Swartzia quinquangulata 15Caesalpiniaceae 2 Swartzia reticulataCaesalpiniaceae PFT 2 Swartzia arborescens silvestris 6 5 SGCaesalpiniaceae Swartzia benthamiana HG 4 8Caesalpiniaceae Swartzia grandifolia SCLG Code 1Caesalpiniaceae Swartzia guianensis SCLM 0 6Caesalpiniaceae 10 3 Swartzia laevicarpa NNOU 2 16 146Caesalpiniaceae Tachigali leblondii 8 NPSE 2 1Caricaceae SENQ 1 Tachigali longicarpa 3 12Caryocaraceae NPAR 0 Vouacapoua panacoco 16 SENM 3 0 3 11 2Caryocaraceae 4 polyphylla 3 NTOT 16 0Caryocaraceae 2 SWAA remiger 2 SENR SWAB VRI Caryocar 3Caryocaraceae myrmecophila Jacaratia 3 0 38 3 7 0 Caryocar 48 americana SENS 0Cecropiaceae paniculata 0 3 SWAG Caryocar HeightCecropiaceae 16 4 0 0 2 16 0 Caryocar SWAH 2 4Cecropiaceae Disp. 200 SWAK 3 3 0 1Cecropiaceae 0 1 Cecropia 3 0 0 2 glabrum 16 0Cecropiaceae spinosa Density 3 15 Cecropia 16 4 microcarpum 0 SWAL 0 Architecture 6Cecropiaceae 4 1 SWAM Cecropia 3 nuciferum 1Cecropiaceae 2 38 SWAP 3 0 1 0 Cecropia 4 villosum 4 SWAQ 0 0Cecropiaceae 4 1 Cecropia 0 TACV 4 11 0 4Cecropiaceae 0 latiloba Pourouma 0 30 0 18 16 SWARCecropiaceae 0 obtusa Pourouma VOUA TACW 10 2 1 4 48 31 4 18Cecropiaceae 40 palmata 0 Pourouma 6 0 2 460 5 0Cecropiaceae 3 0 peltata Pourouma 1 AN 6 18 0 2 0Cecropiaceae 1 5 sciadophylla Pourouma 0 bicolor 5 36 CARM 3Cecropiaceae AN 0 35 5 Pourouma 18 2 5 cecropiaefolia 6 6 CARK 6 1 0 0.55Cecropiaceae 17 0 Pourouma JACS guianensis 36 15 12Celastraceae 6 5 CARN 0 3 Pourouma 3 6 - melinonii 3 14 1 38Celastraceae 178 30 38 Petit Pourouma 0 0 1 23 minor CARV 2 6Celastraceae 8 23 EZ 1 Pourouma 0 0 2 1 36 mollis 1 1 0 10 HYChrysobalanaceae 10 Goupia 16 673 saulensis 10 18 2 -... 1 10 - SZRMbat1 1 20 0 CECL 4 2 Chrysobalanus Maytenus SZRMbat1 tomentosa 0 22 3 - 5 1 10 Maytenus - CECO EZ 1 velutina 0 1 1 2 0 1 CECS 36 CECP 30 4 0 - villosa HY 3 184 SZRMbat1 icaco SZRMbat0.9 1 POUC 0 42 CECQ 3 - Troll 0 10 4 Troll 0 - 12 POUB - 2 30 6 0 10 3 61 10 glabra myrsinoides 35 SZRMbat- - 0 POUG 5 - 0 2 1 35 25 0 Troll sp. - 5 Troll 1 1 POUM 1 0 6 1 38 SZRMbat1 3 30 SZRMbat1 SZRMbat01.01.00 0 0 - 3 0 SZRMbat1 0 3 Troll 3 2 0 14 SZR 30 4 0 - - 16 POUN 0 1 0 6 AN 0 POUO 6 211 POUT 4 POUP SZRMbat- 0 1 3 Troll 2 0.92 Troll 14 4 0 3 17 50 0 0 Troll 0 AN 1 0 4 25 0 4 - 186 POUV 0 1 1 Troll 4 49 11 2 20 0 POUW 0 MAYM Troll - 0 SZRbat 4 0 0 45 1 0 5 2 0 CHRA 0 2 SZRbat 0.79 3 25 - 0.79 0 GOUP 45 0 5 EZBM 9 3 0 3 11 SZRbat - - 0 - 0 0 2 0 15 - - 0 MAYT 0.79 5 - 10 SZRbat 0 211 0 30 0.79 5 0 0 15 - 17 EZBMbat0.35 4 1 - 0 0 0 EZBMbat0.35 15 - 3 25 - EZBMbat0.35 5 0 25 EZBMbat0.35 Rauh 54 0 Rauh 1 5 20 EZBMbat0.35 3 Rauh 0 EZBMbat0.4 1 25 EZBMbat0.4 Rauh 1 EZBMbat0.4 0 65 Rauh 3 20 EZBMbat0.4 Rauh 25 Rauh 5 EZBMbat0.4 1 11 Rauh 5 6 4 Rauh EZBMbat0.4 15 Rauh EZBMbat0.4 EZBMbat- 30 35 35 1 Rauh 6 EZBMbat0.4 Rauh EZBMbat0.4 EZBM - EZBM - Rauh 20 EZ 0.78 Rauh - EZBM Roux - 0.75 - - 196 Appendix B 0.980.98 Troll Troll 0.980.98 Troll Troll 0.98 Troll -- 0.980.980.98 Troll Troll Troll 0.98-- Troll ------0.980.98 Troll -- Troll -- 0.980.98-- Troll Troll -- 0.98 Troll 0.98 Troll 0.98 Troll 0.98-- Troll ChrysobalanaceaeChrysobalanaceae LicaniaChrysobalanaceae LicaniaChrysobalanaceae LicaniaChrysobalanaceae LicaniaChrysobalanaceae LicaniaChrysobalanaceae Licania albaChrysobalanaceae Licania apetalaChrysobalanaceae Licania canescensChrysobalanaceae Licania coriaceaChrysobalanaceae Licania davillifoliaChrysobalanaceae Licania densifloraChrysobalanaceae Licania discolorChrysobalanaceae Licania 18 11 16 ellipticaChrysobalanaceae Licania granvilleiChrysobalanaceae 11 11 2 2 Licania 3 heteromorphaChrysobalanaceae Licania 12 5 hypoleuca 2Chrysobalanaceae 3 2 4 Licania incanaChrysobalanaceae 19 3 Licania LIAC 3 LICA 3 LIAB 16 irwiniiChrysobalanaceae 12 Licania 12 kunthianaChrysobalanaceae 3 3 0 LICB LIDA 3 Licania 5 3 latistipulaChrysobalanaceae 3 Licania 11 3 LIDB 5 laxiflora 0Chrysobalanaceae 0 4 Licania 3 leptostachyaChrysobalanaceae 3 2 0 Licania LIDC 0 5 23 113 licaniiflora LIGBChrysobalanaceae LIEA 11 Licania LIGA 12 3 macrophyllaChrysobalanaceae 0 4 Licania 0 0 2 majusculaChrysobalanaceae 0 0 2 Licania 6 2 619 LIHA 3 44 5 membranaceaChrysobalanaceae 0 16 Licania 3 2 3 micranthaChrysobalanaceae 0 14 3 Parinari 15 0 1 0 0 minutiflora 3... 737 19 1 LIHB Parinari LIHK 0 70 22 LIHC octandra 16 1 39 2 2 16 0 4 LILA ovalifolia 0 0 3 0 1 4 3 persaudii LILC 4 0 0 3 LILB 0 0 12 1 3 83 robusta 0 19 44 LILD 4 campestris LIMA 0 4 5 0 3 0 0 excelsa 3 10 35 LIMC 0 115 0 0 0 LIMB 21 3 1 3 19 107 10 20 3 0 5 0 1 4 12 0 LIMD 40 7 3 116 11 0 0 SZR(Mbat) LIME 3 12 - 0 6 0 0 1 0 3 0 20 25 SZR(Mbat) 5 2 0 25 2 LIMF 3 0 4 3 19 SZR(Mbat) 0 1 LIMG 0 3 0 0 0 3 0 30 0 0 25 3 SZR(Mbat) SZR(Mbat) 40 LIMH 0 14 11 SZR(Mbat) PARC 0 20 LIMR 2 0 25 5 0 0 0 7 5 3 SZR(Mbat) 0 SZR(Mbat) 0 SZR(Mbat) PARE 56 1 0 0 - 5 SZR(Mbat) 0 SZR(Mbat) 0 5 12 10 4 0 1 10 25 70 0 0 25 3 4 - 0 4 0 3 0 1 8 48 SZR(Mbat) 30 SZR(Mbat) 5 SZR(Mbat) 0 0 30 30 5 SZR(Mbat) 30 1 0 10 30 48 SZR(Mbat) SZR(Mbat) 25 SZR(Mbat) 0 SZR(Mbat) 1 1 SZR(Mbat) - SZR(Mbat) 45 3 4 SZR(Mbat) 3 25 40 SZR(Mbat) - 2 25 SZR(Mbat) 25 SZR(Mbat) ZMRbat SZR(Mbat) 40 1 SZR(Mbat) ZMRbat 1 Troll Troll Family... ChrysobalanaceaeChrysobalanaceae Couepia GenusChrysobalanaceae CouepiaChrysobalanaceae CouepiaChrysobalanaceae CouepiaChrysobalanaceae ExellodendronChrysobalanaceae Hirtella bracteosaChrysobalanaceae Hirtella Species caryophylloidesChrysobalanaceae barbatum Hirtella guianensisChrysobalanaceae Hirtella parillo 11Chrysobalanaceae HirtellaChrysobalanaceae 11 Hirtella 2 bicornisChrysobalanaceae Hirtella 12 ciliata 18Chrysobalanaceae 2 PFT Hirtella 3 glandulosa SG Licania 3 3 2 hispidula HG COEC 10 Code obidensis 3 0 COEB 5 paniculata 15 1 NNOU 0 EXEL physophora 10 COEG NPSE 5 silicea 3 2 0 0 8 9 1 affinis NPAR COEP 4 12 0 1 6 NTOT 3 8 3 0 HIRB 3 VRI 0 2 5 HIRG 304 0 2 0 3 3 HIRC 86 0 8 Height 1 HIRH 13 2 HIRO 313 0 8 0 Disp. 33 HIRP 0 3 HIRQ 0 4 92 Density 2 3 0 4 0 2 0 Architecture 38 0 0 2 0 HIRS 1 0 3 LIAA 0 0 25 13 0 5 0 0 2 0 0 25 0 7 SZRbat 5 2 0 0 45 0.95 0 0 25 1 SZRbat 0 0 4 0.95 Troll 20 2 0 SZRbat 0 0 EZ 0.95 1 0 Troll 3 3 28 SZRbat Troll 25 6 0.95 - 0 3 12 1 15 EZ Troll 25 EZ 4 - 3 7 EZ 0.9 EZ 1 EZ 0.9 - EZ 10 EZ Troll - - Troll 15 - - EZ - - SZR(Mbat) - - - - - Lists of tree species 197 Family... ChrysobalanaceaeChrysobalanaceae Parinari GenusClusiaceae ParinariClusiaceaeClusiaceaeClusiaceaeClusiaceae Calophyllum montanaClusiaceae Caraipa Species rodolphiiClusiaceae CaraipaClusiaceae Caraipa brasilienseClusiaceae ClusiaClusiaceae Clusia 3Clusiaceae Clusia 18 densifoliaClusiaceae PFT Clusia punctulata 2 SGClusiaceae 14 2 Mahurea richardiana HGClusiaceae Mahurea 1 Code 1 5Clusiaceae fockeana MoronobeaClusiaceae PARM leprantha Platonia 16 NNOU 4 PARR 12Clusiaceae 0 melchiori Rheedia 12 NPSE 0Clusiaceae 3 CALB exstipulata minor Rheedia 3Clusiaceae NPAR palustris Rheedia coccinea 3 0 4ClusiaceaeTovomitagrata832TOVG0000110EZ-- 3 5 Symphonia NTOT 1 6Clusiaceae 3 Tovomita 1 insignis CARD VRI CAREClusiaceae 7 Tovomita 12 0 acuminata CARF 0 0Clusiaceae 1 0 Tovomita benthamiana Height 0Clusiaceae 0 3 2 8 2 globulifera 7 18 macrophylla 0 2Clusiaceae Disp. Tovomita 3 CLUA brevistaminea 2ClusiaceaeVismiaguianensis612VISG400449EZ0.55Roux 12 2 Vismia 0 CLUB 3 2 0 6 8 cephalostigma 15Combretaceae Density 0 Vismia MAHE 0 1 CLUC 0 choisyana 5 Architecture Combretaceae 12 2 1 2 Vismia 0 15 2 8 0Combretaceae 16 1 3 Vismia MONO 12 MAHU 0 Buchenavia CLUD 3 secundaCombretaceae 0 4 2 0 0 2 2 Buchenavia 0 0 -Combretaceae 3 0 0 3 3 1 Buchenavia 0 cayennensis PLAICombretaceae 4 28 RHEA RHEB 0 12 2 Buchenavia latifoliaCombretaceae 3 RHEM 5 0 0 1 SYMG tetraphylla 0 Conocarpus macrophylla 0Combretaceae 18 3 0 2 1 TOVB - 0 0 grandis TOVC Terminalia 10 sessilifoliaCombretaceae 40 5 7 0 0 macrophylla 6 0 Terminalia 3Connaraceae 0 nitidissima 27 Terminalia 3Connaraceae 94 2 7 10 0 2 TOVD 6 erecta 3 0 19 0 Terminalia 43Connaraceae 1 ZMRbat 0 0 ZMRbat 10 30 1 1 5 0 amazoniaConnaraceae 0 2 0 12 Connarus 2 3 3 2 EZ 1 dichotomaConnaraceae 112 0 1 18 25 Connarus 49 16 TOVS 18 5 229 guyanensisCyrillaceae 3 20 2 0 5 VISC Connarus 2 AN 0 4 1 3 lucida... 2 Troll 0 Troll 7 1 Connarus - 0 1 2 2 3 BUCT 0 0 VISM AN 1 Connarus 282 VISL 51 24 10 11 5 5 coriaceus 0.7 0 AN 11 5 11 BUCM 1 0 1 3 13 VIST erianthus Cyrilla 3 0 BUCG 2 0 0 185 20 - BUCN 0 0 5 40 - fasciculatus 2 15 3 0 5 4 0 5 - 0 - perrottetii 3 0 EZBM 2 2 1 3 9 EZBM punctuatus 10 0 3 197 6 - AN TERA 0 4 0 EZ 0 1 EZBM CONR - TERD 3 30 - 4 4 4 9 15 35 0 15 0 2 0 - - 0 0 0 TERG 4 racemiflora 0 AN 3 - 8 0 20 0.88 EZBM 3 6 3 8 3 - 0 0 EZ 0 1 15 - EZ EZM EZ - 25 9 0 0 TERL - 0 1 3 1 0 - - 0 0 20 3 COND 1 EZ 5 - 0.85 0 0 0.71 CONE 2 0 CONF 0 0 0.75 0.75 0 EZ 3 2 EZ 0 - 0 6 0 3 CONP 0 2 EZ - Massart 0.75 0 Roux 0 0 CONQ 3 Attims Attims 0 0 5 - 2 0 0 0.92 0 3 1 12 Attims 0.92 0 0 12 - CYRI 3 - 0 - 0 Fagerlind 50 0 0 12 0 - Fagerlind 0 6 EZ 0 18 20 EZ 0 0 5 50 - 5 EZBM 5 50 EZ - 0 0 0 EZ 0 0.78 EZBM - 0 0.55 EZBM 0 0.78 0 25 EZBM - 8 EZ 20 - 0.78 - Roux 0 - 2 0.78 0 0 - - 3 HY 1 0.55 - 1 HY HY - - - 4 20 0 3 Roux 0.75 5 SZ 0.75 - 5 4 Aubreville SZ 8 Aubreville 0.75 3 7 EZB - EZB EZB - Aubreville - EZB 6 - EZB - - - - - HY ------198 Appendix B Family... DichapetalaceaeDichapetalaceae GenusDichapetalaceae TapuraDilleniaceae TapuraEbenaceae TapuraEbenaceaeEbenaceae CuratellaEbenaceae Species amazonicaEbenaceae Diospyros capituliferaEbenaceae Diospyros singularisEbenaceaeDiospyrosmartinii832DIOL000047EZ0.8Massart DiospyrosEbenaceae Diospyros americana 8Elaeocarpaceae Diospyros 12 carbonariaElaeocarpaceae PFT Diospyros guianensis SGElaeocarpaceae 3 16 3 Sloanea lissocarpodes HGElaeocarpaceae Diospyros Sloanea Code 2 melinonii 5Elaeocarpaceae 3 3 Sloanea dichroaElaeocarpaceae 4 TAUA Sloanea NNOU TAUC 4 duckeiElaeocarpaceae 0 8 8 Sloanea NPSE 0 0Elaeocarpaceae 3 TAUS Sloanea 2 brevipes tetrandraElaeocarpaceae NPAR 3 Sloanea 3 12 0 grandiflora 1Erythroxylaceae CURA NTOT Sloanea guianensisErythroxylaceae 2 0 2 16 0 3 Sloanea DIOC 0 VRI latifoliaErythroxylaceae Erythroxylum 12 DIOL DIOG echinocarpaErythroxylaceae Erythroxylum 3 3 1 0 8 eichleri Height 4Erythroxylaceae 12 Erythroxylum 0 0 3 0 DIOM 4 0 garckeana amazonicumErythroxylaceae Erythroxylum 0 Disp. 11 3 3 parviflora amplumErythroxylaceae Erythroxylum 0 3 3 DIOD 15 0 94 rufa citrifoliumEuphorbiaceae 2 Density Erythroxylum 4 2 3 0 1 DIOE 1 0 Architecture kapplerianum 0Euphorbiaceae Erythroxylum 0 2 5 0 3 ligustrinumEuphorbiaceae DION SLOH 12 0 12 SLOB 3 1 Acalphya 94 macrophyllum 4Euphorbiaceae 0 SLOI 11 0 0 Alchornea 0 squamatum 6Euphorbiaceae 0 3 8 7 1 3 0 2 0 1 SLOE Alchornea 4 0Euphorbiaceae 0 2 2 - Alchorneopsis 3 6 0 SLOL 3 2 0Euphorbiaceae 0 1 2 1 1 8 Amanoa 3Euphorbiaceae 2 0 ERYD diversifolia SLOG 0 1 SLOF 8 Aparisthmium 1 6 0 12 2Euphorbiaceae 1 2 schomburgkii 1 20 1 0 5 0 SLOP floribunda Chaetocarpus 0 3Euphorbiaceae 0 triplinervia 35 0 1 ERYG 0 Conceveiba 2 ERYE 1 ERYF -Euphorbiaceae 1 0 cordatum 108 0 2 Croton 0 1 SZ 0Euphorbiaceae 5 0 1 SZ 0 0 8 ERYM ERYH 8 Croton 2 schomburgkianus 0 guianensis 6 436Euphorbiaceae SLOR SZ 0 0 0 Croton 0 0 4Euphorbiaceae 110 12 0 11 ERYS 11 3 - 0 guianensis 0 Croton 0.68 3 0 0.68Euphorbiaceae 0 0 1 10 0 436 6 Dodecastigma 0 4 0 3 2... 7 0.68 2 2 0 EZB 2 3 Drypetes 0 0 0 Cook 1 3 10 Cook cuneatus 18 3 0 3 Glycydendron 3 ALCS 0 ACAL 0 5 EZ 2 hostmanii 1 Cook - EZ integrifolia 0 1 - 0 1 7 EZ 0 ALDF 30 ALCT 0 matourensis 0 0 CHAE 0 2 18 0 EZ schiedeanus 0 0 0 amazonicum 0 0 25 3 0 0.8 1 0.8 2 0 0 18 variabilis APAR 0.8 8 - 0 EZ - 4 AMAN 0 - 0 0 12 0 1 1 0 2 EZBM 11 0.8 3 0 EZ 3 Massart 0 4 Massart 0 1 0 16 8 24 0.9 30 Massart 3 CONC EZBM 3 2 0.8 5 0 0 EZ 4 0 - Massart 1 0.9 3 EZBM 4 0 1 3 0 0 0 0.8 3 12 3 8 1 2 Aubreville 16 111 4 0.9 20 EZBM 0 - 1 4 Massart CROC DODI 0.8 2 CROM 3 16 CROH 1 Aubreville - 8 0 Massart 1 GLYC 4 10 EZBM 0 5 1 0 CROS 0 1 135 0 3 EZ Aubreville EZBM 41 EZBM 0 0.9 12 Massart 0 1 0 - 0.9 EZBM DRYP - 15 EZ - 4 EZ 0 - 0 64 - 8 0 EZ 10 0 0 3 0 43 Aubreville EZ 3 Aubreville 4 4 0 EZ EZBM - - - - 68 25 0 EZ 0 - 20 - 5 0 Rauh - 0 10 8 5 12 - 20 20 0 Rauh 5 EZM - Rauh 92 Rauh 0 1 EZ 20 0 - 0 Rauh EZ 15 15 01.01.13 EZ EZ Rauh - 0 15 120 0.7 Rauh EZ - 4 5 AUT 1 1 2 - 0.7 5 2 - Champagnat EZ 0.94 4 Champagnat 25 Koriba 18 30 3 - - 0.68 25 8 Koriba EZ EZ EZ EZ - EZ EZM EZ - - - 0.71 - 0.86 - - Roux - - - - - Lists of tree species 199 Family... EuphorbiaceaeEuphorbiaceae GenusEuphorbiaceae HeveaEuphorbiaceae HuraEuphorbiaceae HyeronimaEuphorbiaceae MabeaEuphorbiaceae MabeaEuphorbiaceae Species ManihotEuphorbiaceae guianensis Maprounea alchorneoidesEuphorbiaceae MargaritariaEuphorbiaceae crepitans MicrandraEuphorbiaceae piriri Pausandra 14Euphorbiaceae speciosa guianensis Pera brachylobaEuphorbiaceae 11 nobilis PFT Pera 1 SGEuphorbiaceae Phyllanthus 2 HG 17 elataEuphorbiaceae 4 Code Phyllanthus martiniiEuphorbiaceae 3 Pogonophora 1 2Euphorbiaceae HYER 10 NNOU Sagotia 8 8Euphorbiaceae HEVE 16 attenuatus 5 Sandwithia NPSE 1 7Euphorbiaceae 1 bicolor juglandifolius 0 Sapium schomburgkiana 3Fabaceae HURA 3 NPAR glabrata 3 Sapium 1Fabaceae 8 2 0 0 NTOT 11 2 Senefeldera 2 11Fabaceae MAPR MANB 5 5 guianensis VRI 2 12Fabaceae racemosa MABS 0 2 3 MABE 0 2Fabaceae MARG 0 0 0 0 19 Alexa 0 Height 3 9Fabaceae 2 glandulosum 5 3 0 macrophylla Andira 47Fabaceae paucinervium Disp. 2 2 MIDA PAUW 0 Andira 0Fabaceae POGO 8 0 0 0 35 Andira 0 7 0Fabaceae 0 PHYB Density PHYA 0 0 11 Bowdichia 3 59 Architecture 0Fabaceae 7 8 2 3 0 0 ClathrotropisFabaceae 0 2 0 2 PERB 5 Diplotropis washenheimii PERC 2Fabaceae 0 0 1 2 coriacea 29 3 0 3 Dipteryx 2 0 5Fabaceae 3 0 0 inermis SANG Dipteryx 0 0 brachypetalaFabaceae 2 2 virgilioides surinamensis SAGR 0 SAPG Dussia 0 0Fabaceae 11 30 29 0 SAPP SENE Erythrina purpurea 0 0 0 4Fabaceae 0 0 0 25 0 Erythrina 0 2Fabaceae 0 0 5 11 0 odorata 15 Hymenolobium 15Fabaceae 5 EZBM 1 5 0 5 3 punctuata Hymenolobium 11Fabaceae 0 2 4 1 110 SZ 40 0.7 2 0 3 2 Lonchocarpus 0Fabaceae ALEX 0 flavum 5 amazonica 0 discolor 19 0 0 0 3 0 Lonchocarpus 2Fabaceae 4 petraeum 4 fusca 25 0 5 Muellera 0 170 15Fabaceae 0.6 - 2 CLAB 3 3 EZ 3 7 2 hedyosmus Ormosia ANDC ANDS 0 4Fabaceae 16 15 0 0 14 0 heptaphyllus Ormosia BOWD... 0 ANDI 5 1 1 0 0 0 280 EZ 7 Ormosia 0 3 EZ 2 Rauh 0 EZ - 0 0 DIPL Ormosia 12 17 18 19 EZ 22 0 20 4 4 4 HY frutescens Ormosia 2 7 4 3 11 0 - 13 8 10 10 5 0 0.7 1 0 7 2 3 Ormosia - cinerea DIPO DIPP 0 0 2 1 - coarctata - EZ 2 0 2 EZ - 5 5 5 EZ 4 4 0 coccinea 2 25 ERYA Nozeran 15 0 15 EZ EZ 3 0 2 - coutinhoi 23 4 HYMP 1 HYMF DUSS 0 5 - 0 2 - 0 flava 0 - 0 12 - 0 LONC LONB - 0 5 - melanocarpa EZ 1 - 23 - ERYB 0 EZ EZ 0 19 8 16 0 14 31 0 12 10 0 0 0 8 0 EZ 2 - 3 5 3 - 0 18 0 - - 0 0 0.7 - EZ 12 6 0 MUEL - - 27 5 5 0 4 0 3 2 EZ EZ 5 - 0 0 5 3 0 11 1 ORMA ORMB 25 - 2 - 0 5 1 5 - 0 0 3 0 - 5 11 0 - 2 - 30 0 ORMC 0 ORMD 20 ORMM 0 - 0 35 0 2 EZ 0 3 10 0 0 1 - 20 5 5 0 0 40 SZR ORME 3 - - 8 EZ 0 - SZR 0 0 0 EZ 0 1 0 0.89 0 SZR 1 4 AN 2 30 0 - 0 0.89 35 1 3 - Champagnat 0.89 0 - 0 4 0 14 10 0.92 Champagnat 0 45 50 SZRbat 40 0 SZRbat 0 - Champagnat 01.01.00 20 - 15 Attims - - 0 17 EZ 5 12 0 AN AN 0 SZ 1 1 AN AN - - 3 EZ 0 0.75 0.75 6 5 - 2 - - 40 30 - 40 3 HY 15 - 25 EZ EZ EZ - 20 EZ EZ - - 0.8 - EZ - - 200 Appendix B Family... FabaceaeFabaceae GenusFabaceaeFabaceaeFabaceae OrmosiaFabaceae OrmosiaFabaceae ParamachaeriumFabaceae Species PlatymisciumFabaceae ormosioides PlatymisciumFabaceae Poecilanthe paraensisFabaceae Poecilanthe trinitatis stipularisFabaceae Pterocarpus uleiFabaceae 12 PterocarpusFabaceae PFT effusa Pterocarpus SGFabaceae 3 hostmanni 12 Spirotropsis HGFlacourtiaceae officinalis Code 11 16 Taralea 3Flacourtiaceae rohrii 3 VataireaFlacourtiaceae 2 santalinoides PARB NNOU 3 Banara VataireaFlacourtiaceae 3 longifolia NPSE Carpotroche Vataireopsis 16 0 8Flacourtiaceae 3 4 Carpotroche ORMP 8Flacourtiaceae NPAR 15 3 Casearia ORMS 15 PLAT 0Flacourtiaceae 3 oppositifolia NTOT Casearia guianensis 0Flacourtiaceae 3 2 4 0 0 longifolia surinamensis 2 15 VRI Casearia 2 paraensisFlacourtiaceae surinamensis guianensis 7 2 4 Casearia PLAUFlacourtiaceae 4 0 POEC 2 Casearia 15 HeightFlacourtiaceae 0 POEB PTEO 0 2 0 0 Casearia 0 18 acuminata PTESFlacourtiaceae 4 18 Disp. 2 32 0 8 Casearia arboreaFlacourtiaceae 8 2 4 2 18 6 Casearia PTER 0 combaymensis 2Flacourtiaceae Density 4 0 Casearia commersoniana 0 SPIL 3 Architecture Flacourtiaceae 0 0 0 1 5 3 2 4 5 Casearia decandra TARO 1Flacourtiaceae 0 1 9 2 Casearia grandiflora 0 VATS 0Flacourtiaceae 2 5 0 6 0 VATG 2 Casearia guianensisFlacourtiaceae 0 CARH 12 1 0 0 0 0 0 0 19 CARG VATP Casearia javitensis 0Flacourtiaceae 38 BANG 1 0 0 Casearia mariquitensis 0 1 0Flacourtiaceae 3 0 0 0 3 0 4 12 Homalium negrensis 8FlacourtiaceaeRyaniapyrifera832RYAP000038EZ-- 2 0 20 CASA 3 1 Homalium 4 pitumba 1 0Flacourtiaceae 25 1 74 CASC 6 3 CASD Laetia rusbyana 0 7 0Hernandiaceae 0 3 0 CASB 0 11 0 0 4 sylvestris 0 0Hippocrataceae 25 0 3 1 1 Ryania 0 ulmifolia 2 ANHugoniaceae 4 5 guianense 2 25 2 2 0 0 35 8 Hernandia CASFHumiriaceae racemosum 2 0 Cheiloclinium 1 0 CASG 0 2 0 6 3 1Humiriaceae EZ 0 5 0 0 0 0 - 11 CASH 0 3Humiriaceae 15 EZ 0 AN 30 Hebepetalum CASK 32 6 4 CASJ procera 1 5... 0 1 cognatum 2 2 Humiria 12 6 0 - 12 0 0 guianensis 8 0 0 0 speciosa 0 Humiriastrum 1 32 0 0 2 1 3 - 0 - 3 EZ CASL 0 3 AN HY Humiriastrum 0 humiriifolium 1 3 3 3 30 CASM 2 0 0 CASN EZ 3 5 0 0 exelsum HY 2 2 4 0 1 0.65 1 8 2 15 - 0.6 0 1 CASO 17 5 30 0 10 subcrenatum - balsamifera 0 HOMG 12 0 CASP HY 40 0 - 8 HOMR 40 0 1 0 - 0 3 0.6 1 3 0 8 0 0 Troll 4 EZ 12 11 0 40 4 HY 0 1 - 5 12 16 2 3 4 3 - 0 3 AN 5 17 16 HY 0 Troll - 0 LAET CHEI 0 2 HERG 0 3 - - HEBE EZB 0 15 EZB AN 0 0 4 1 3 1 EZ 2 20 0 0 1 0.69 0 RYAS 4 0.8 1 6 - 1 16 5 4 - - 0 0 0 - HUMS 1 0 6 - - - - 0 5 0 EZ HUMB EZM HUME - 1 0 7 0 0 18 4 13 EZ 0 0 EZ 15 0.9 - 0 - 0 0 - 15 0.9 1 18 - 0 10 0 EZ 5 0.9 7 13 0 0.9 25 1 0 EZ 0 1 Cook Cook 5 EZ 0 5 EZ 8 EZ 0.9 Cook 4 Cook 1 22 39 0.9 12 0 0 20 4 0 0 0.9 0.9 15 0.9 0 Cook 20 EZ Cook - 4 12 EZ 12 EZ 7 3 Cook 5 Cook 0 1 Cook EZ HY 0.9 1 0.9 HY EZ 0.9 40 20 - 20 12 0.9 - 5 - Cook Cook - 0.9 10 Cook EZ EZ EZ Cook EZM 35 40 - 35 Cook EZ 0.75 1 - 0.75 - EZ EZBM - EZ - - 0.9 ------Lists of tree species 201 Family... HumiriaceaeHumiriaceae GenusHumiriaceaeHumiriaceae SacoglottisHumiriaceae SacoglottisHumiriaceae SchistostemonIcacinaceae SchistostemonIcacinaceae Species Vantanea cydonioidesIcacinaceae dichotomum Vantanea guianensisIcacinaceae sylvaticum DendrobangiaLascistemataceae DiscophoraLascistemataceae 16 Lascistema 12 EmmotumLauraceae guianensis Lascistema boliviana PoraqueibaLauraceae 14 PFT parviflora 3 12 3 SGLauraceae HG 1Lauraceae guianensis 4 3 3 Code grandifoliumLauraceae fagifolium Aiouea 4 16 SACC polystachyum 3 SCID guianensisLauraceae Aniba NNOU 11 16Lauraceae 0 SACG 3 Aniba SCIS NPSE 0Lauraceae 10 8 Aniba 2 0 3 8Lauraceae 4 NPAR Aniba 0 5 1Lauraceae 3 4 12 NTOT Aniba 0 3 VANG 3Lauraceae guianensis 0 Aniba 3 VRI 0 DEND VANP 0 2Lauraceae 3 2 Aniba citrifolia 2 9Lauraceae 13 0 LASC Aniba guianensis 2 3 Height DISC 21Lauraceae LASP Aniba hostmanniana 0 1Lauraceae Disp. 0 EMMO 0 Aniba PORA jenmanii 15 0 0Lauraceae 24 0 0 Aniba kappleri 0 58 21Lauraceae Density 2 12 Chlorocardium megaphylla 16 8 Architecture Lauraceae 0 0 Cinnamomum panurensis 0 2 4Lauraceae 55 1 3 0 3 Cryptocarya parviflora 1 27 9 rodiei 3Lauraceae 0 3 Endlicheria rosaeodora 7 AIOU 3Lauraceae 4 cinnamomifolia 11 Endlicheria 1 taubertiana 0 7 2Lauraceae 92 0 0 5 Endlicheria ANIH williamsii 0 0 ANIC 3 14 2 guianensisLauraceae 0 2 - ANIG 30 Endlicheria 0 8 2Lauraceae 0 12 0 2 bracteolata Endlicheria 1 15 3 25 0Lauraceae 5 1 7 endlicheriopsis Kubizkia 0 2 1 30 3 3Lauraceae 1 ANIJ 0 19 88 4 multiflora EZM Licaria 2 ANIM 1 25 1Lauraceae 16 ANIK 16 punctulata Licaria 0 3 2 2 0 0 ANIP EZM 0 3Lauraceae 12 4 0.9 0 pyriformis 16 Licaria 25 4 0 EZ 3Lauraceae 0 35 3 4 ANIQ 2 CINN 0 Licaria 4 0 5 ANIR - EZ 3 1Lauraceae 3 mezii 35 Licaria 4 0 4 ANIT 0 9 12Lauraceae - 0 CHLO 0 0 0 Licaria 0.9 EZ 3 4 0 4 18... EZ cannella 0 ANIW CRYP 0 Licaria 0 0 23 - 3 11 0 13 8 chrysophylla ENDB EZ - 15 Licaria ENDA 0 0 3 debilis 0 0.75 - 0 9 3 Licaria 0 2 0 0 0 0 - 0 EZ globifera 0 1 Licaria EZ 0 0 EZ 4 0 ENDC 0.9 guianensis - 3 0 EZ EZ - 4 7 0 ENDD martiniana 0 0 15 0 - 0 5 ENDE 0 1 0 polyphylla 0 0 0 0.9 0 - - 1 3 - 2 0 rigida 2 - - 0 0 35 0 0 8 subbullata 12 0 1 15 2 0 0 4 0 - vernicosa 0 20 12 - 0 27 1 0 0 3 0 3 LIRD 15 KUBI - 3 2 EZBM LIRC 16 3 0 - 0 - 0 3 1 3 - 0 11 0 2 4 0 0 0 EZBM 3 8 EZBM 1 0 3 0 10 0 3 25 EZBM 0 LIRG 16 0.6 LIRE LIRH 0.6 4 8 13 0 3 3 LIRM 0.6 35 3 0 0 3 0 - 0 0 1 1 LIRP 0 0 0 20 EZBM 7 EZBM 3 2 Rauh 2 Rauh 1 30 0 4 EZBM Rauh 0 0.6 0.6 0 EZBM 15 40 2 LIRS 0 0.6 0 LIRR 0 0 0 30 0.6 EZBM 35 7 5 0 LIRV EZBM EZBM 25 0 30 Rauh 0 1 Rauh 0.6 EZBM 0 EZBM - 0.6 4 Rauh 0 Rauh 11 EZBM 7 EZBM 0.6 - 0 0 0 7 18 EZBM 0 Rauh EZBM 0.6 - 0 0 5 Rauh - 0 - - 25 0 - Rauh 7 EZBM 3 - 0 0 Rauh 4 0 0 - - 1 EZBM EZBM - 0 - 0 - - - - 7 5 4 30 4 0 1 - 4 0 EZBM - 25 - EZBM EZBM 30 15 20 - 3 1 - 4 30 1 EZBM EZBM EZBM EZBM 0.7 - - - 12 - 30 - - EZBM 12 - - EZBM EZBM - - - - EZBM 0.7 - - - - - 202 Appendix B grandiskunthiana722RHOK0000315EZBM-- 16 3 4 RHOD 0 2 0 2 5 30 EZBM - - Family... LauraceaeLauraceae GenusLauraceaeLauraceaeLauraceae MezilaurusLauraceae NectandraLauraceae NectandraLauraceae Nectandra SpeciesLauraceae Nectandra itaubaLauraceae Nectandra cissifloraLauraceae Ocotea cuspidataLauraceae Ocotea globosaLauraceae Ocotea pearceiLauraceae PFT Ocotea reticulata SGLauraceae Ocotea 18 12 HGLauraceae 16 Ocotea CodeLauraceae aciphylla 2 Ocotea 3Lauraceae argyrophylla 3 15 Ocotea NNOU 5Lauraceae canaliculata 3 12 12 Ocotea NPSE 4Lauraceae 2 cernua Ocotea MEZI NECCLauraceae cymbarum NPAR 3 3 Ocotea NECD 8 4 12Lauraceae discrepens 0 0 NTOT Ocotea 0 16 3 3Lauraceae fasciculata NECG Ocotea VRI 3 3Lauraceae floribunda Ocotea NECP NECR 3 0Lauraceae glomerata 16 3 2 Ocotea 0 0 0 0 12 HeightLauraceae guianensis 0 4 Ocotea 12Lecythidaceae OCOA OCOB indirectinervia 3 Disp. 8 Ocotea 3 OCOCLecythidaceae 0 oblonga 0 3 7 Persea 16 4Lecythidaceae 0 Density petalanthera 0 0 0 0 3 Rhodostemonodaphne 16 3 0 12 Bertholletia 3 Architecture Lecythidaceae puberula 3 OCOE Rhodostemonodaphne 10 Couratari OCODLecythidaceae rubra 2 3 0 OCOF 3 0 0 2 Couratari 4Lecythidaceae schomburgkiana 0 1 0 0 0 12 0 0 0 Couratari OCOGLecythidaceae 4 splendens 0 3 16 OCOH excelsa Couratari 3Lecythidaceae 0 10 nivea 3 OCOI 0 Couratari OCOK 0Lecythidaceae 16 0 0 0 3 gloriosa OCOJ 0 Couroupita 1 1 4 2 3Lecythidaceae 0 0 0 0 1 guianensis 2 3 Eschweilera 4 1Lecythidaceae 15 multiflora 0 3 OCOP 12 EschweileraLecythidaceae 0 5 oblongifolia 4 OCOO 0 0 9 Eschweilera 0Lecythidaceae 0 2 OCOS 19 40 4 25 stellata 0 3 1 1 0 0 0 guianensis EschweileraLecythidaceae 30 OCOQ 0 0 10 apiculata 4 14 Eschweilera 19 0Lecythidaceae 3 3 0 0 collina 30 Eschweilera 0Lecythidaceae 1 18 1 0 EZBM 19 1 OCOR EZBM 0 congestiflora 1 3 5 OCOT 3 Eschweilera 0Lecythidaceae EZBM 25 20 9 0 1 35 - coriacea - 2 13 Eschweilera... 4 0 3 3 19 - 4 5 0 BERE decolorans 0 EZBM Eschweilera 19 0 0 18 4 5 15 grandiflora PERN 0 0 5 Eschweilera 4 COUF COUG 3 EZBM EZBM - 15 28 0 9 2 3 3 micrantha 3 14 0 COUH 5 - 0 - - - COUI 0 0 16 parviflora 5 3 1 - 2 0 3 EZBM 5 EZBM 4 pedicellata 30 16 16 0 0 0 EZBM 3 COUK 20 1 4 sagotiana - 0.6 0.6 4 4 1 5 2 45 25 COUJ 0 0 3 0.6 0 3 3 4 - 0 - 4 ESCA 16 ESCE 12 0 0 9 EZBM 28 3 4 EZBM 4 4 0 0 Massart Massart ESCD 6 0 0 3 16 83 EZBM - 25 25 Massart 32 3 - 1 0 0 ESCF 0 ESCG 0 0 EZBM 0.6 5 0 16 4 3 3 EZBM 0 4 137 ESCH 0 - 721 0 14 5 25 0 ESCM 3 1 0 0.6 EZBM EZBM 4 EZBM - 2 30 Massart - 0 0 8 2 0 - 0.6 0 0.6 149 25 3 4 ESCP ESCQ 13 0 30 0 19 Massart EZBM - 0 5 0 ESCS 30 3 49 EZBM 1 Massart 5 - Massart EZBM 3 0 5 166 0 3 1 0 - 0.6 22 EZBM 0 0 0.6 14 EZBM 49 - 4 62 10 45 0 0 40 0.6 Massart - EZBM 20 5 55 0 35 117 734 Massart 4 - 19 0.6 2 0 0 SZR 45 - Massart AN 168 937 5 3 EZBM AN AN 40 1 - Massart 1 - 45 - 28 AN 67 10 0.65 2 1055 - - 30 - EZ 30 Massart AN 3 - - 35 - SZR - 5 SZR 35 - Massart SZR Massart - SZR - 1 1 SZR 35 Massart 1 30 - SZR 1 1 - Troll SZR Massart 1 Troll SZR SZR Troll SZR Troll 1 1 Troll 1 Troll 1 Troll Troll Troll Troll Lists of tree species 203 Family... LecythidaceaeLecythidaceae GenusLecythidaceae EschweileraLecythidaceae EschweileraLecythidaceae EschweileraLecythidaceae GustaviaLecythidaceae simiorum GustaviaLecythidaceae Species subglandosa LecythisLecythidaceae wachenheimii LecythisLecythidaceae LecythisLecythidaceae augusta LecythisLecythidaceae 18 12 hexapetala 12 LecythisLecythidaceae Lecythis chartaceaLecythidaceae 2 PFT 3 3 Lecythis SG confertifloraLoganiaceae 5 3 HG Lecythis corrugataMalpighiaceae 3 Code Lecythis 8 holcogyneMalpighiaceae 7 ESCU ESCT ESCV idatimonMalpighiaceae NNOU 16 Antonia 16 Byrsonima 0 persistensMalpighiaceae 0 3 2 0 NPSE Byrsonima poiteauiMalpighiaceae 3 3 12 2 Byrsonima 2 simiorumMalpighiaceae NPAR 16 Byrsonima zabucajo18Malpighiaceae 4 4 0 3 NTOT 0 GUSH GUSA 0 Byrsonima 12Malpighiaceae crassifolia 3 12 VRI LECB LECC 5 2 Spachea ovata 0Malpighiaceae eugeniifolia 3 3 Byrsonima 4Melastomataceae spicata 0 13 19 3 5 LECD 12 0 Byrsonima 2 HeightMelastomataceae stipulacea 0 3 LECH Bellucia ByrsonimaMelastomataceae 3 verbascifolia 3 0 LECZ Disp. 19 0 3 5 Henriettea LECIMelastomataceae 0 9 1 0 11 LECP 5 HenrietteaMelastomataceae aerugo elegans Density 3 0 2 0 Henriettea 32 0 Architecture Melastomataceae densa 14 58 136 0 LECQ 14 26 1 13 Henriettea LECS 0Melastomataceae laevigata 0 2 0 1 Henriettella 0 1 3 grossularioides maroniensisMelastomataceae 0 1 0 0 5 Loreya 321 3 160 0 multifloraMelastomataceae 1 198 BYRC 4 BYRE 737 Miconia ramiflora 4Melastomataceae 4 0 0 1 11 14 Miconia 14 0 0 succosaMelastomataceae 14 13 17 ANTO 1 5 caudata 0 BYRT 1 Miconia BYRS 6 1 14Melastomataceae 40 763 BYRV 25 14 0 2 1 25 1 Miconia 1Melastomataceae 0 0 6 Miconia 1 6 0Melastomataceae 1 1 1 3 25 4 52 4 0 0 mespiloides 6 Miconia 354 12 6 5 argyrophyllaMelastomataceae SZR 257 SZR 0 4 SZR 4 2 Miconia SPAE BELL 1 0 BYRA chrysophyllaMelastomataceae 0 6 0 Miconia 7 0 holosericeaMelastomataceae BYRL 1 5 69 0 10 0 4 - 2 0 BYRD HENA 1 2 35 - Miconia - 0 0 ZMR kappleriMelastomataceae 25 6 0 35 7 0 0 Miconia 1 2 1 lepidotaMelastomataceae 2 HENC 13 7 0.85 Miconia 0 longifolia... 0 3 ZMR 0 2 0 HENE SZR 45 1 0 2 2 0 Miconia 20 0 0 Troll - mirabilis SZR 6 0 35 25 - SZR 2 0 0 - myriantha 0 HENG 0 0.85 - 2 13 HENK 2 1 plukenetii 10 1 0 1 2 1 SZR 0 0 - 40 SZR 0 0 LORE MICA poeppigii 0 0 0 6 SZR 10 SZR - 1 6 MICC prasina 1 2 0 0 0 4 1 0 0 1 20 1 11 pubipetala 1 Troll 5 1 0 1 SZR 3 MICH - 0 Troll 5 0 1 4 - 0 8 0 0 2 10 3 2 20 0 MICK 0 0 2 0 0 30 SZR 1 15 0 0 0 Troll MICL Troll 3 0 3 Troll 0 Troll 0 2 7 30 MICM 35 0 11 5 2 EZ 0 0 - 1 5 - MICN EZ 2 0 0 AN MICO 0 4 1 Troll 4 0 2 - 3 2 0 MICP 0 0 EZ 0 0.75 EZ 0 0.75 0 MICQ 3 EZ 2 4 0.54 0 Troll 25 0 35 30 0 0 0 3 30 1 MICR Aubreville 1 MICS 11 35 0 0.75 Aubreville 0.75 - 0 0 0 0.75 6 0 3 0 AUT EZ 15 EZ 0 0 Aubreville 0 Aubreville 0 EZ 0 EZB 11 EZ 6 Aubreville 3 - 0 0 0 4 0.75 - 10 0 EZB 6 0 - 0 0.75 0 0.75 5 EZB 0 0 15 Aubreville 9 - 0 - EZB 3 0 Aubreville 12 0 Aubreville Leeuwenberg - EZB 0 - 4 0 9 5 EZB - 0 EZB 5 - - EZB 20 5 0 0.7 - 0 20 0.7 4 - EZB 13 4 - 20 EZB - Leeuwenberg 10 5 0.7 Mangenot EZB 4 9 EZB - Mangenot 27 EZB - EZB Mangenot - 18 7 0.7 EZB EZB - Mangenot Mangenot EZB 0.7 Mangenot Mangenot - EZB Mangenot 0.7 Mangenot - Mangenot Mangenot Mangenot 204 Appendix B Family... MelastomataceaeMelastomataceaeMiconiaserrulata321MICU000044EZB-Mangenot Genus MiconiaMelastomataceaeMelastomataceae MiconiaMelastomataceae MiconiaMelastomataceae MiconiaMelastomataceae Mouriri ruficalyxMelastomataceae Species MouririMelastomataceae Mouriri tomentosaMelastomataceae Mouriri trimeraMelastomataceae Mouriri tschudyoidesMelastomataceae Mouriri 7Melastomataceae acutiflora MouririMelastomataceae collocarpa PFT 6 Mouriri SGMelastomataceae crassifolia 2 4 Mouriri HGMelastomataceae dumetosa 6 1 Code Mouriri 2Melastomataceae francavillana 3 Mouriri 15Melastomataceae grandiflora 2 NNOU MICT 15 Mouriri 1Melastomataceae guianensis 1 2 NPSE 16 Mouriri MICV 1Melastomataceae huberi 2 2 8 Mouriri 8 MICUMelastomataceae nervosa NPAR 4 0 3 MouririMelastomataceae MICT 4 nigra 11 0 NTOT Myriaspora 3 MOUAMelastomataceae 3 oligantha 4 0 0 7 MOUC Tococa VRI 0Meliaceae 2 sagotiana 2 2 Votomita 0 0 MOUDMeliaceae sideroxylon 3 2 2 1 HeightMeliaceae 15 MOUF subumbellata MOUE 7 egensis 0 0MeliaceaeGuareacostata1233GUAC0000318EZ-- vernicosa 0 Disp. MOUG 0 0 2 2 12Meliaceae 0 0 0 11 Carapa 2 7Meliaceae MOUH Density 15 0 guianensis 5 12 guianensis 4 Carapa 3 Architecture Meliaceae 0 1 2 34 2 0 Cedrela 0 0 2Meliaceae 2 3 MOUI 3 80 0 0 10 3Meliaceae MOUM 7 0 4 Guarea 2 17 2 3 0 MOUOMeliaceae 34 0 Guarea MOUN 6 1 3 12 0Meliaceae 0 guianensis MOUT 0 0 MOUS MOUU 2 Guarea 80 0Meliaceae 6 procera 0 0 3 0 0 3 Guarea 23 1 0 odorataMeliaceae 1 2 - 1 Guarea 0 MOUVMeliaceae 10 3 0 0 2 Guarea 1 3 0 MYRXMeliaceae 0 0 gomma 13 15 Guarea 0 - 0 0 VOTG 0Meliaceae 1 grandifolia TOCG 0 5 Guarea 0 10Meliaceae 0 guidonia EZB 2 0 0 Trichilia 15 0 3Meliaceae 30 kunthiana 30 1 18 0 EZB 0 Trichilia 4Meliaceae 0 macrophylla 5 29 0.7 0 0 2 0 Trichilia 0 EZB EZBMeliaceae pubescens 2 1 19 0 Trichilia - CARG 12 0 3... EZBM 4 scabra EZBM 0 14 Trichilia 15 5 0 0 0 Mangenot 0.7 0.7 silvatica 3 12 EZBM 01.01.00 - 3 0 16 25 Trichilia cipo12 CARF 12 0 0 1 Mangenot - CEDR 01.01.00 Trichilia euneura - 5 0 3 226 3 Mangenot 0 Mangenot 15 EZBM 3 3 3 Trichilia EZBM 8 martiana 4 1 Mangenot Mangenot 0 0 4 GUAE EZBM 3 - micrantha GUAD - 1 4 Mangenot 3 3 1 0 - 16 pallida 0 30 - 3 0 12 GUAG EZBM 12 12 0 quadrijuga GUAK 0 TRIC GUAM 0 20 1 2 - schomburgkii 0 - 0 3 0 25 12 3 Mangenot 0 28 Mangenot 15 147 25 septentrionalis 3 EZBM 15 10 0 GUAP EZBM 0 Mangenot 3 3 3 1 5 01.01.00 EZBM 2 0 01.01.00 0 12 Mangenot 1 16 EZBM 0 GUAQ 389 0 Mangenot 0 0 Mangenot 16 3 25 11 GUAR EZBM - EZBM 0 EZBM 4 - 7 0 3 0 0 3 3 - 0 - TRIE 3 2 01.01.00 20 6 TRIL 6 0 Mangenot 4 TRIM 3 4 0 EZBM 6 0 0 0 4 3 Mangenot 0 0 0 EZB 01.01.00 0 Mangenot TRIS TRIQ 0 0 EZ 0 Mangenot Mangenot TRIT TRIP Mangenot EZB 35 0 4 0 0 - 1 35 0 0 0 2 0 0 0 - 0 - 0 3 27 3 SZRM 0 40 0 SZ 0 5 - 0.7 0 0 5 0 0 1 0 27 0 - 0 - 50 25 0.7 AN 3 Cook 0 0 20 0 0 35 0 25 0 20 0 EZ 0 1 EZ 0.46 Cook 15 EZ 0 0 EZ EZ 0 1 2 EZ - Chamberlain - 20 3 5 20 0.7 EZ 0.7 0.7 4 - 4 3 20 EZ 5 - - EZ 35 20 - 0.7 - - 0.7 30 18 EZ Rauh - 30 18 EZ EZ - - 0.8 EZ EZ - EZ EZ 0.8 - Rauh 0.8 0.8 0.8 0.8 Rauh Rauh Rauh Rauh Rauh Rauh Lists of tree species 205 Family... MeliaceaeMimosaceae GenusMimosaceaeMimosaceaeMimosaceae Trichilia AbaremaMimosaceae AbaremaMimosaceae AnadenantheraMimosaceae Species BaliziaMimosaceae CalliandraMimosaceae peregrina jupunba surinamensis CedrelingaMimosaceae laeta EnterolobiumMimosaceae EnterolobiumMimosaceae Hydrochorea surinamensisMimosaceae 16 PFT pedicellaris Hydrochorea oldemanii 16 cateniformis SGMimosaceae Inga 18 schomburgkii 3 HGMimosaceae Code 3 IngaMimosaceae corymbosa 3 4 2 IngaMimosaceae 8 gonggrijpii 4 12 NNOU 18 Inga 18Mimosaceae TRIU 5 3 2 NPSE Inga ANBPMimosaceae 3 3 2 Inga 0 2 ABARMimosaceae NPAR 0 1 10 2 Inga 3 2 5Mimosaceae 4 10 acreana 5 NTOT Inga CALAMimosaceae 1 1 acrocephala BALI ABAL CEDP VRI IngaMimosaceae ENTS 1 0 0 alata 0 Inga 3 ENTO 0 0Mimosaceae 4 alba 0 1 Inga 3 HeightMimosaceae 2 auristellae HYDC IngaMimosaceae 12 Disp. bourgoni HYDG 0 0 Inga 0 12Mimosaceae 0 0 brachystachys 0 0 Inga 0 1 1 3Mimosaceae Density bracteosa 3 0 Inga Architecture Mimosaceae capitata 3 Inga 8 3Mimosaceae 0 0 0 3 8 cayennensis 0 Inga 0 0Mimosaceae 0 15 12 INAB 15 crassiflora 23 6 Inga INAA 3Mimosaceae 0 2 disticha 3 Inga 20 2Mimosaceae 7 2 edulis 15 5 22 0 2 Inga 1 1Mimosaceae 2 16 7 0 0 fanchoniana 24 0 4 12 Inga 4 6Mimosaceae INAE 2 INAC fastuosa 2 Inga INBB 0 3Mimosaceae INAD 3 flagelliformis 0 INBA 2 22 Inga 14 26 35 2Mimosaceae 4 5 5 gracilifolia 3 30 1 Inga 12 3 52 0 53 3Mimosaceae 3 2 2 37 heterophylla INBC Inga -Mimosaceae 0 huberi 15 3 - INCA Inga INCB 1 3 0 1 0 EZMimosaceae 0 2 ingoides EZ 5 25 3 Inga 1 40 0 0Mimosaceae 6 3 60 18 EZM 2 4 4 jenmanii INCC Inga 7 5 1Mimosaceae 2 lateriflora - 0.8 Inga 20 INDA 2 4 2... 25 0.66 - 15 0 0 laurina EZ INFA 0 1 Inga EZ AUT 2 0 SZMB 33 0 0 0 leiocalycina 0 1 EZ INEA Inga 0 20 13 4 INFC lomatophylla Rauh - - 2 6 Inga 3 2 - EZ EZ INFB 23 3 0.6 melinonis 3 0 0.75 - 15 26 10 9 0 86 INGA - nobilis INHA 3 0 3 0 18 EZ 1 53 7 38 11 nouragensis 2 3 8 1 - 0.55 0.75 0 12 - 1 - 25 0 nuda - 2 25 3 - 0 2 5 1 paraensis 3 0 - 0 0 INHB 3 5 - 2 3 18 - - INIA 1 42 0 pezizifera 0 3 INJA INLA EZMB 0 3 pilosula 105 0 3 1 0 EZMB 0 11 15 2 2 INLC - 0.65 35 23 1 0 0 0.65 INLB 5 - INLD 4 0 30 2 8 0 0 1 2 56 20 3 0 Troll 0 12 2 0 12 23 1 EZMB Troll 0 INMA EZMB 3 0 - EZMB 0 EZMB 9 2 15 0.65 20 3 15 18 INNA 4 0.65 3 22 0 EZMB - 3 0.65 INMB 26 0.65 1 8 5 0 3 0 4 0 0.65 3 0 Troll 0 3 0 Troll - EZMB EZMB Troll 0 EZMB - INPA - INNB Troll 0 INPB 131 0 20 Troll 1 - - 0.65 0.65 22 0 1 30 0 0 0 6 15 0 INPC 0 23 3 EZMB 1 Troll Troll - - 9 EZMB 56 0 0.65 EZMB - 0 EZMB 0 - 0 13 0 2 0 12 - 0.65 0 0.65 5 - 5 Troll EZMB 31 3 EZMB 1 0 Troll EZMB Troll 0.65 EZMB 0.65 0 0 - 2 - 4 0 - 0.65 10 - - 25 EZMB 22 Troll Troll 0 22 0.65 Troll 15 0 20 - - EZMB 6 EZMB EZMB - EZMB - Troll - 0.65 - 0.65 1 0.65 5 EZMB 1 - EZMB 25 - Troll Troll Troll EZMB 0.65 - 4 - 25 0.65 20 EZMB EZMB Troll - - 0.65 Troll EZMB - EZMB EZMB 0.65 0.65 Troll - - Troll EZMB Troll 0.65 - Troll 206 Appendix B ioaeeakali1935PAUU1001545ZMR0.45- Family... MimosaceaeMimosaceae GenusMimosaceaeMimosaceae IngaMimosaceae IngaMimosaceae IngaMimosaceae IngaMimosaceae Species IngaMimosaceae IngaMimosaceae IngaMimosaceae poeppigiana IngaMimosaceae retinocarpa IngaMimosaceae rubiginosa IngaMimosaceae PFT semialata Inga SGMimosaceaeParkiaulei 3 splendens Macrosamanea HGMimosaceae 3 sterulifera Code ParkiaMimosaceae 2 stipularis Parkia 12Mimosaceae kegelii thibaudiana NNOU 2 1Mimosaceae 12 tubaeformis 3 NPSE Parkia 16Mimosaceae 1 umbellifera INPD Penthaclethra 12 3Mimosaceae 3 NPAR virgultosa Pseudopiptadenia 3 INRAMimosaceae 10 11 3 NTOT nitida 3 3 Pseudopiptadenia INRBMimosaceae 11 psilostachya 1 4 pendula macroloba VRI Stryphnodendron 1 2Mimosaceae suaveolens INSB 8 3 7 31 Stryphnodendron 2Mimosaceae INSC guianense velutina 3 3 0 Stryphnodendron 6 8 Height INSDMimosaceae moricolor 3 3 2 Zygia 16 4 0Mimosaceae INTA Disp. INSE polystachyum 0 0 Zygia 11Monimiaceae 2 1 INTB 2 17 3 18 Zygia 20 19Monimiaceae 6 Density 0 0 Zygia MACK 2 2 0 6 INUA Architecture Monimiaceae 4 1 19 0 0 2 3 Zygia 18 0 0 4Monimiaceae 0 Siparuna 3 INVA 0 2 PSEC 5Moraceae 5 3 1 5 Siparuna 3 2 3 5 cataractae 0Moraceae PENT 3 0 Siparuna 2 PSED PAUN 5 0 1 2 PAUP inaequalisMoraceaeBrosimumacutifolium1634BROA0156335EZMR1Troll 5 0 31 Siparuna 0 0 1 latifolia 0 6 131Moraceae 0 2 STRP STRG PAUV morii 0 -MoraceaeBrosimumlactescens1825BROL0000345EZMR1Troll cuspidata 13 0 STRM Bagassa 7 1 1 0 racemosaMoraceae 2 4 - cristata 8 5 1 0 Batocarpus 1 0 0Moraceae decipiens 0 0 11 22 0MoraceaeBrosimumutile1935BROU0000150EZMR1Troll guianensis 1 9 3 - Brosimum 0Moraceae 7 2 0 0 0 11 0 5 25 - 6 0Moraceae 0 2 0 0 5 6 amazonicus Brosimum 16 0 0 3 guianensisMoraceae 5 2 16 4 20 Brosimum ZYGC 5Moraceae 7 1 EZMB 0 3 ZYGI 7 - 0 0 2 EZMB 3 36... guianense 3 30 0 EZMB 132 0.65 1 0 Clarisia 0 2 20 3 0.65 25 4 0 3 6 2 ZYGL 7 4 0.65 EZMB Ficus 4 2 parinarioides 25 18 SIPC 1 Ficus Troll 1 ZYGR 0 2 0.65 38 5 rubescens EZMB 1 1 1 25 Troll ZYGM 2 0 Ficus 3 EZMB Troll 12 EZMB 8 1 2 5 SIPB 0 0.65 1 SIPD 0 19 10 5 5 - SIPG EZMB Troll 0.65 19 1 5 - 3 ilicifolia 15 EZMB 0 0 50 3 0.65 1 4 30 Troll 0 EZ 0 BATO 3 BAGS 0.65 3 - 25 Troll 0 EZMB 19 0 broadwayi 0 5 50 1 40 - 18 Troll 5 0.65 catappifolia EZMB AN 0 3 45 AN 0 BROG - Troll 6 gomelleira 12 45 0 0 0.65 BROP HY 0 0 0 5 - ZMR Troll 0 10 ZMR 0 0 0 0 0.79 0.79 11 ZMR Troll BROR 6 0 EZ EZ 0.45 - 0 1 1 0.45 - 0 3 0 2 - 3 17 0.45 11 - 0 EZ 1 3 4 1 0 0 0 - - - - 3 1 3 CLAA 4 7 - 2 - 0 10 FICB 4 5 - 10 5 0 4 - FICC 25 1 0 18 0 - - FICG - 0 20 5 EZ 6 13 30 29 - 0 0 EZ 10 - 30 - 0 0 EZ - - 0 15 6 EZ 34 12 - SZC 0 0 EZ 45 3 - - 0 SZ - 01.03.00 - 5 0 SZ SZ 45 - 01.03.00 - 0 0 EZMR - 40 EZ - 0.8 0 0.75 0.8 0.8 - EZMR 40 0 0 1 5 Rauh EZMR - Roux Roux Roux 2 1 EZMR 1 1 1 Troll 20 - 20 Troll 15 Troll EZM 40 EZ - EZ EZ 0.4 0.4 - 0.4 Rauh Rauh Rauh Lists of tree species 207 Family... MoraceaeMoraceae GenusMoraceaeMoraceaeMoraceae FicusMoraceae FicusMoraceae FicusMoraceae Species FicusMoraceae FicusMoraceae FicusMoraceae Ficus guianensisMoraceae Ficus insipidaMoraceae Ficus malacocarpaMoraceae PFT Ficus SG maximaMoraceae Helicostylis HG nymphaeifoliaMoraceae 6 Code Helicostylis paludicaMoraceae 10 Maquira pertusaMoraceae 17 NNOU 1 Maquira 14 pulchellaMoraceae 1 pedunculata NPSE Naucleopsis schumacheriMoraceae 1 18 2 tomentosa Naucleopsis 1 3 trigonaMoraceae NPAR Perebea 5 FICHMyristicaceae 2 10 4 NTOT FICM Perebea guianensisMyristicaceae 11 FICI 14 glabra 15 0 VRI Pseudolmedia 5 sclerophylla 1 16 FICOMyristicaceae 0 guianensis Trymatococcus 15Myristicaceae 2 Iryanthera 1 2 FICN 0 3 0 Trymatococcus 3 HeightMyristicaceae Virola 14 laevis 3 2 4 guianensis 4Myristicaceae 0 0 Virola oligandrus FICP 10 Disp. 4 0 14 mollisMyristicaceae paraensis Iryanthera HELP 1 4 FICQ FICS 0 0 11 FICR 1Myristicaceae Density Iryanthera 0 paraensis 1 10 0 Architecture Myrsinaceae 4 Virola HELT 0 0 0 0 3 0 2 0Myrsinaceae 4 Virola 1 12 11 3 FICTMyrsinaceae melinonii Virola MAQG hostmannii 0 0 3 0 MAQS 15 17 3Myrsinaceae sebifera 3 0 2 0 sagotiana 0 Ardisia 0 0 3 0 0 NAUGMyrsinaceae 15 0 12 0 NAUF Cybianthus 2 1 3 3 2Myrsinaceae 1 Cybianthus 0 0 0 2 3 0Myrtaceae michelii 4 5 12 PERG Cybianthus TRYO 0 0 0 11Myrtaceae multicostata 0 1 3 0 4 4 0 3 Myrsine 0 0 1 TRYP PSEBMyrtaceae 16 surinamensis 3 19 fulvopuverulentus 11 Stylogyne 0 2Myrtaceae guianensis PERH IRYP 1 0 0 0 2 microbotrys 0 3 3 7Myrtaceae 0 1 0 2 3 Calycolpus surinamensis 0 0 15 0 5 0 12 0 0Myrtaceae 1 1 Calycolpus 20 24 15 15 IRYH 4 3 VIRLMyrtaceae 1 2 Calycorectes 2 40 4 0 0 guianensis 35 0Myrtaceae 0 IRYS surinamensis 7 2 2 Calycorectes 3 VIRS 2 1 0 0 2 EZ 6 0 5 4... 1 40 1 0 Calyptranthes EZ 4 1 goetheanus 6 33 4 21 4 0 Calyptranthes CYBF 2 revolutus 2 VIRN EZ 2 bergii EZ 1 20 0 0 Calyptranthes 0 22 VIRM 8 VIRT 0.4 0 grandifolius 25 4 0.4 0 2 5 amshoffae 31 6 EZ Calyptranthes 0 1 30 1 34 30 0 2 30 36 30 30 7 3 fasciculata 0.4 8 0.4 CYBM 0 3 3 ARDG EZ lucida Rauh CYBS 1 121 0 1 Rauh 0 0 0.4 EZM speciosa 0 0 6 0 2 EZ 25 0 30 8 4 EZ 3 EZ 0 0 109 EZ 13 30 2 Rauh 0 Rauh 1 1 10 0.4 - 11 143 STYL 0 2 20 8 1 Rauh MYRY 3 5 20 0.4 0 5 166 0.4 1 EZ 0.6 0 EZ 0.4 0 0 0 2 0 46 0 CALC EZ 20 2 0 4 2 4 Rauh 5 3 20 0 3 8 EZ 6 0 3 - 0 CALD Rauh 5 EZM CALG Rauh 0.93 - 0.4 2 Rauh 30 0 40 CALK 0.93 89 0 0 CALE 0 31 EZ 0 0 5 3 0 1 0 30 CALL - EZ 20 23 0 0.8 4 0 - 0 Rauh 4 2 - 2 30 0 EZ EZBM 0.75 - - 0 0 5 0 0 0 EZM 0.68 CALP 25 - 0 CALS EZBM - EZBM 0 0 - 6 0 Troll 0 0.75 0 0.75 - EZBM 0 20 - 0 Roux 35 1 0.75 35 EZBM 0 15 0 2 0 0 Troll 5 Massart - 0 - 35 0 EZBM 0 0 0 - Massart - 0 EZBM EZBM - EZB 6 4 0 4 5 0.48 0 0.48 12 EZBM 0 - 0 0 - 2 0 0.46 0 Massart Massart - EZB 8 6 EZB EZB 11 4 1 Massart 4 0 - 12 - 0 - - 4 EZB EZB 10 7 17 EZBM 18 1 - - 5 - - 12 - - EZBM EZBM EZBM - EZBM 15 - - EZBM 10 ------EZBM EZBM ------208 Appendix B Family... MyrtaceaeMyrtaceae GenusMyrtaceaeMyrtaceaeMyrtaceae CampomanesiaMyrtaceae CampomanesiaMyrtaceae Eugenia aromaticaMyrtaceae Eugenia Species grandifloraMyrtaceae EugeniaMyrtaceae EugeniaMyrtaceae EugeniaMyrtaceae 6 Eugenia albicans 7Myrtaceae Eugenia anastomosansMyrtaceae PFT Eugenia 1 citrifolia SGMyrtaceae 2 Eugenia coffeifolia HGMyrtaceae 2 Eugenia cucullata 12 Code 2Myrtaceae Eugenia excelsa 8 CAMAMyrtaceaeEugeniaramiflora321EUGS000023EZBM0.9- Eugenia 3 CAMG NNOU feijoi 0Myrtaceae Eugenia 0 florida NPSE 8 3Myrtaceae 3 8 Eugenia lambertianaMyrtaceae NPAR Eugenia 8 latifolia EUGA 2 3Myrtaceae 0 NTOT 3 macrocalyx 0 0Myrtaceae 8 Eugenia EUGA omissa 2 3 VRIMyrtaceae 2 8 Marlierea patrisii 0Myrtaceae 8 EUGB 2 Marlierea 11 3 pseudopsidium 0 EUGC HeightMyrtaceae Myrcia 0 0 8 0 3 punicifolia EUGD 7Myrtaceae 0 2 Disp. 2 3 MyrciaMyrtaceae 0 2 5 0 Myrcia tapacumensis 3 EUGE 3 2 montanaMyrtaceae Density 2 0 Myrcia 6 0 EUGK 0 0 Architecture schomburgkianaMyrtaceae 0 8 Myrcia 2 0 2 EUGG EUGF 2 1Myrtaceae 0 Myrcia 0 1 0 2 0 0 amazonicaMyrtaceae EUGM 6 2 Myrcia 3 EUGL 3 bracteataMyrtaceae 0 0 0 2 3 Myrcia 0 0 EUGQ 0 29 9 1 citrifoliaMyrtaceae 2 1 Myrcia 0 1 EUGO 0 0 decorticansMyrtaceae 0 0 Myrcia 0 1 EUGP 2 deflexaMyrtaceae 0 10 0 9 Myrcia EUGR 0 1 10 0 31 0 fallaxMyrtaceae 0 EUGT 0 Myrcia MARS 5 0 3 0 gracilifloraMyrtaceae Myrcia 3 0 0 0 0 0 0 0 guianensis 14Myrtaceae 5 EZBM Myrcia MARM 0 EZBM 0 5 inaequiloba 0Myrtaceae 23 3 1 Myrcia 0 - 0 1 0 1 magnoliifoliaMyrtaceae - 0 Myrcia 0 0 2 7 10 0 MYRA - 0 multifloraMyrtaceae 0 5 0 0 Myrcia 4 0 pistrinalis 0 2MyrtaceaePsidiumguineense502PSIG000047EZBM-- EZBM MYRB 10 Myrcia 5 15 0 14 1 0 2 5 0 pyrifolia - ... MYRD 0 0 Myrcia 11 0.9 0 0 - MYRC EZBM 3 rufipila 1 0 0 Myrciaria 0 4 2 10 0 1 0 0 saxatilis 0.9 2 0 2 Myrciaria 0 EZBM 0 5 3 EZBM 4 servata - MRYE 2 8 0 1 5 1 0.9 0 MYRG 2 3 spendens 0.9 0 12 MYRF 0 EZBM 0 4 0 0 - 0 MYRH 16 sylvatica 6 15 0 0 MYRI MYRJ 0 0 0 0.9 1 floribunda tomentosa 0 10 - 2 8 EZBM 0 - 0 5 0 vismiifolia 8 1 0 EZBM 1 2 5 0 EZBM 0.9 0 MYRK EZBM 10 0 4 0.9 14 - 1 0 3 7 0 2 MYRM 0 0 0.9 12 EZBM 0.9 0 0 5 0 0 5 15 0 6 8 1 2 EZBM 0.9 0 MYRN - 5 1 2 EZBM 0 - 6 0 0 4 0 0.9 4 - 15 0 MYRO - 0 EZBM 0 0.9 0 1 3 5 MYRP 2 0 0 EZBM - 0 1 0.9 0 17 MYRR 0 2 0 4 2 2 EZBM MYRS 0.9 - 0 0 1 EZBM 0 - EZBM 0 2 MYRT 1 0.9 0 0 MYRU MYRV 0 17 0 - - 0.9 0 0 EZBM 0 MYRW 0 0 0 0 - 0 6 0 - 2 0 30 - 3 0 13 0 EZBM - 5 0 6 - 0 0 0 0 - 0 3 3 0 EZBM 0 EZBM - 15 0 EZBM 6 0 - 0 - 16 1 15 - 0 0 1 0 0 22 - 14 0 EZBM 0 0 EZBM 5 EZBM - - 0 EZBM - 10 - - 0 1 EZBM - EZBM 15 0 0 - 1 0 - - 1 10 EZBM - 4 EZBM 5 - 15 - - 1 6 - 5 5 - - 30 EZBM - 8 - EZBM 6 13 10 - EZBM - 10 EZBM - - - EZBM - EZBM EZBM EZBM - EZBM ------Lists of tree species 209 -Roux RhizophoraceaeRosaceaeRubiaceae RhizophoraRubiaceaeRubiaceaeRubiaceae PrunusRubiaceae racemosa AlibertiaRubiaceae AlibertiaRubiaceae AlibertiaRubiaceae AlseisRubiaceae Amaioua 7Rubiaceae edulis Amaioua myrtifoliaRubiaceae myrciifolia Bertiera... 2 surinamensis Capirona Chimarrhis 2 corymbosa Chomelia longifolia 11 guianensis Coussarea RHIZ 5 2 4 guianensis 2 0 decorticans turbinata 0 1 3 6 3 tenuiflora 7 2 1 7 mapourioides PRUN 1 0 1 4 ALIM 16 ALIS 0 2 2 ALIE 19 2 0 3 4 0 2 3 2 0 3 3 AMAC 0 4 0 ALSE 1 0 AMAG 3 5 2 0 0 CAPI 0 BERG 0 1 0 CHIM 0 1 0 8 0 0 COUL 25 CHOM 0 0 0 0 0 0 0 4 0 0 2 0 0 0 0 0 0 0 0 0 10 0 0 18 3 0 4 0 0 0 1 0 AUT 3 45 0 8 20 3 - 11 5 1 0 5 0 5 4 EZ 3 4 10 EZM - 15 15 3 EZ 3 - EZ 37 - 5 30 EZ EZ - AN 5 - EZ 3 - 0.7 AN - EZ 0.7 - - 0.85 EZ - - - EZ - - Petit ------Family... NyctaginaceaeNyctaginaceae GenusOchnaceae GuapiraOchnaceae NeeaOchnaceaeOlacaceaeOlacaceae OurateaOlacaceae Ouratea Species olfersianaOlacaceae OurateaOlacaceae ChaunochitonOlacaceae floribunda DulaciaPiperaceaePiperarboreum722PIPA000057EZBRC-- decagyna HeisteriaPolygonaceae kappleri guianensis Heisteria 14Polygonaceae PFT melinonii Minquartia SGPolygonaceae 1 Ptychopetalum 12 HGProteaceae Coccoloba Code guianensisProteaceae 4 Coccoloba 10 cauliflora 3 2 olacoidesProteaceaeRoupalamontana612ROUM000047AN-- Triplaris densifrons NNOU guianensis 12 GUAA 1Quiinaceae 3 2 NPSE 0Quiinaceae 1 Euplassa 3 latifolia 3 NEEAQuiinaceaeLacunariaumbonata612LACU000006EZM-- Panopsis NPAR 16 1 mollis 1Quiinaceae 3 827 8 OURD NTOT 11 8 15Quiinaceae 1 Lacunaria 3 surinamensis OURG 0 0 VRI CHAUQuiinaceae Lacunaria 3 0 0 2 2 OURM 4 0Quiinaceae pinnata 3 0 Height 6Quiinaceae sessilifolia Quiina 2 3 4 DULA 10 0 2 0Quiinaceae Quiina Disp. crenata 14 0 HEIC PTYO 2Rhabdodendraceae MING 5 Quiina 1 1 0 HEID jenmani RhabdodedronRhamnaceae Density Quiina 2 17 0 0 1 2 Architecture 3Rhizophoraceae 7 Quiina 0 0 12 12 4 Touroulia 0 0 832 COCL amazonicum TRIV 18 guianensis 0 Cassipourea 3 3 Ziziphus 0 COCM 5 0 0 macrophylla 8 0 1 5 0 0 obovata 7 3 3 4 1 18 2 oiapocensis 7 3 EUPL guianensis 17 PANS sessilis guianensis 0 2 7 7 0 0 12 0 10 2 0 30 0 4 0 2 20 cinamomum 2 5 1 3 4 2 LACC - 14 2 15 LACJ 2 12 0 0 3 18 0 7 0 EZBM 2 14 1 EZ 0 RHAB 25 0 1 2 0 4 25 - 1 QUIM 6 5 QUIG 0 1 - 2 4 4 5 0 4 5 0 6 0.75 0 3 4 0 EZB 4 0 1 2 QUIJ 0 0 QUII EZBM 0 30 - 1 EZ 2 CASU 0.58 - 2 TOUG 0 - EZ 0 30 0 15 20 5 7 ZIZI 0 1 0 23 4 QUIS 1 0 5 - 0 - EZBM - 0 3 0 EZBM 0 0.8 EZBM - EZBM 0 0 27 12 EZBM 0 0 19 25 0.9 0.8 - 1 6 30 0 0.8 Roux - Roux 2 0 0 0 5 0 EZBM Nozeran 0 0 - AN 0 0 EZBM - 18 - - 20 5 - 4 0 0 - 0 13 0 0 19 3 1 AN HY - 15 - 15 3 0 EZM 0 - 5 - 5 - 18 7 3 EZM 0.98 EZ - 0 30 - EZ 30 30 - - - EZ 15 - - EZ 10 EZ(myrmec) - EZ - - EZ - - - EZ - - EZ - - Chamberlain - Chamberlain - - - 210 Appendix B Family... RubiaceaeRubiaceae GenusRubiaceaeRubiaceaeRubiaceae CoussareaRubiaceae CoussareaRubiaceae CoussareaRubiaceae Coutarea SpeciesRubiaceae Duroia micrococcaRubiaceae Duroia paniculataRubiaceae Duroia racemosaRubiaceae DuroiaRubiaceae hexandra Duroia 4Rubiaceae PFT Faramea SGRubiaceae amapana 8 Faramea HGRubiaceae aquatica 3 Faramea Code 8Rubiaceae eriopila 3 Faramea 1 2Rubiaceae longiflora Faramea NNOU 3Rubiaceae 2 micrantha corymbosa Faramea COUM NPSE 7Rubiaceae 1 costata Ferdinandusa 0 2 COUNRubiaceae NPAR 8 guianensis Genipa 1 0Rubiaceae COUO 2 multiflora Genipa NTOT 7 16Rubiaceae 0 occidentalis Gonzalagunia 3 paraensis COUP 4 8 VRI 2 0Rubiaceae quadricostata Guettarda 3 0 2Rubiaceae 2 0 Guettarda DURA 3 4 3 Height 4Rubiaceae 4 Isertia dicocca 2 0 DURB 0 8 americanaRubiaceae 8 Disp. 0 Isertia 1 4 2 DURL 2 spruceanaRubiaceae 3 DURE 3 0 11 Isertia acreana 0Rubiaceae DURM Density FARA 3 0 Isertia 0 3 1 3 macrantha 1 Architecture Rubiaceae 2 0 0 2 Ixora 0 0 2Rubiaceae 2 FARG 1 12 Pagamea 2 FARC 0 3Rubiaceae 4 0 coccinea Palicourea FARM 12 0 0 FARO 0 0 FARQRubiaceae 3 hypoleuca FERD Palicourea 3 0 0 2 0 7Rubiaceae 0 10 1 3 0 3 parviflora Palicourea 0 3 11RubiaceaePsychotriaracemosa431PSYR000044EZ-- 0 spiciformis Palicourea 4 1 3 1Rubiaceae 0 0 2 guianensis 0 20 Palicourea GENA calophylla gracilifolia 0Rubiaceae 0 0 3 0 Posoqueria 6 0 GONZ GENS 3 5 crocea 0 0 7 2 15Rubiaceae 1 Psychotria 3 guianensis 0 5 0Rutaceae 12 GUEA 20 0 GUAM 6 0 1 longiflora 0RutaceaeEsenbeckiacowanii832ESEC000006SZ(myrm)-- 2 Rudgea 3 103 3 0 0 12 5 rigida 0 4 0Rutaceae 4 2 EZ 0 0 0 2 Rudgea latifolia 0 2 1 5Rutaceae EZ 0 5 Tocoyena 0 erecta 2 0 0 ISECSapindaceae 0 3 6 0 3 ISEH 1 2 EZ 8 Erythrochiton - - 15 0 0... 1 15 0 6 ISEP 2 0 0 0 - 1 1 1 ISES 0 1 AN 0 15 Esenbeckia cornigera 0 3 PAGG 0.8 0 3 0 27 4 IXOR Cupania brasiliensis PALC 1 Galipea hostmanniana 1 EZBM 7 guianensis 2 0 EZBM 0 4 0 - - 2 6 5 - 4 0 0 0 0.9 0 2 - 3 0.9 0 PALI 3 EZBM - 0 0 PALD 2 0 EZBM 8 0 3 pilocarpoides 103 PALL 0.9 5 3 0 0 4 1 0.9 8 0 10 2 0 2 Fagerlind EZBM EZ 12 Fagerlind 7 0 4 5 0 - 0 0 diphylla 20 3 0 trifoliata 0.9 PALM 4 POSO 1 4 Fagerlind 4 3 3 8 0 EZ EZ Fagerlind 0 2 2 0 EZ - 1 0 0 PSYE EZ 0 1 1 2 AN 0 EZ 25 0 0 Fagerlind 0 2 RUDH 3 0 0 20 - 4 RUDC 20 - ERYC 0 - 0 TOCK 8 - 2 0 7 2 14 0 0 - 0 - 0 - 0 8 EZ 0 0 0 0 5 ESEP EZ 5 0 3 EZ 2 EZ - - 0 5 - 0 85 0 - - 5 5 2 0 0 0 - 2 6 EZ - 0 - 0 4 5 - 12 - 15 CUPD 0 GALI 100 0 12 0 0 - 0 6 12 - 0 6 0 0 13 0 EZ Aubreville 5 5 EZ 5 - - - 0 EZ 0 0 0 EZ 3 - 15 EZ 0 - 6 0 0 - 0 15 EZ EZ - 12 4 - 5 - 0 EZ 0 5 3 EZ 0 - 4 - - - EZ 3 EZ - 3 0.63 - 6 - - 1 0 EZ - 5 0 6 - - 0.71 7 - EZ - EZ 10 - - 2 - EZ 3 AUT EZ - - - SZ(myrm)- - - 10 - 15 - - - - EZ - AUT - - - - - Lists of tree species 211 Family... SapindaceaeSapindaceae GenusSapindaceaeSapindaceae CupaniaSapindaceae CupaniaSapindaceae DodonaeaSapindaceae MataybaSapindaceae Species MataybaSapindaceae Matayba hirsutaSapindaceae Pseudima scrobiculataSapindaceae viscosa TalisiaSapindaceae arborescens TalisiaSapindaceae guianensis TalisiaSapindaceae PFT 7 opaca Talisia SGSapindaceae frutescens 7 Talisia HGSapindaceae 7 Code Talisia 2 4Sapindaceae elephantipes 2 Talisia 8Sapindaceae 2 furfuracea NNOU Talisia 2Sapindaceae 3 2 guianensis Talisia NPSE 8 CUPSSapindaceae 3 2 longifolia Talisia 9 1 CUPH 16Sapotaceae NPAR megaphylla 0 Toulicia 2 MATA 3Sapotaceae 0 DODV micrantha NTOT Toulicia 3 16 0Sapotaceae 0 mollis MATG 0 2 Vouarana VRI 8Sapotaceae praealta 4 0 3 3 2Sapotaceae Chrysophyllum 8 PSEA 8 retusa 1 Height 3Sapotaceae TALE Chrysophyllum MATO 4 sylvatica 1 0 4 guianensis 0Sapotaceae Chrysophyllum Disp. 0 0 3 argenteum 3 2 pulvinata TALF 0Sapotaceae Chrysophyllum guianensis 1 cuneifoliumSapotaceae 3 Density Chrysophyllum 0 2 2 TALG 0 8 lucentifolium 12 3 Architecture Sapotaceae Chrysophyllum 0 0 pomiferum 0 1 0 TALL TALM 0Sapotaceae Chrysophyllum 8 8 12 3 0 prieurii 3 12 3Sapotaceae 0 Ecclinusa 0 TALN 11 sanguinolentum 8 1 8 4 0Sapotaceae 3 Ecclinusa 3 2 4 3 venezuelanense 3 3 0 0 0Sapotaceae 0 Manilkara 0 2 0 TALP 15 12 3 3 0 3 TALOSapotaceae 3 Manilkara 2 3 6 2 0 0 3 4 4Sapotaceae 0 Micropholis 0 0 2 TALR 2 2 3 1 TOUG guianensis 6 CHRB TALSSapotaceae 16 Micropholis 0 0 4 CHRC 0 0 0 ramiflora 0 0Sapotaceae TOUP Micropholis 4 3 VOUG 3 3 3 CHRL 0 bidentata 10 3 3 112Sapotaceae Micropholis 0 0 0 15 huberi 0 58 0 CHRO 0 CHRS 1 cayennensisSapotaceae Micropholis 15 4 0 5 4 egensis 0 5Sapotaceae 5 Micropholis 15 1 5 0 0 112 CHRV 1 EZ 15 0 guyanensisSapotaceae Micropholis CHRR 0 EZ 0 4 0 0 1 0 longipedicellata 0 2Sapotaceae Pouteria 0 19 0 EZ 1 4 mensalis 0 14... 6 Pouteria HY 0.85 3 3 20 4 30 6 0 EZ 0 3 porphyrocarpa 3 0 Pouteria - 4 0 16 0 venulosa 4 3 0.9 Pouteria 18 0 0 0 0 ECCG 0 1 30 5 12 0 - - 2 EZ Pouteria 3 4 - 15 3 EZ 2 4 1 EZ 13 2 2 39 ambelaniifolia 0 ECCR MANC 0 3 4 0 - - 4 4 3 bangii 1 MIDC 17 12 1 5 58 EZ 0 0 0 - 4 3 3 3 bilocularis - 74 - - EZ MANH 6 MIDL EZ 4 16 14 - 4 50 caimito MIDG 3 10 1 1 0 MIDE 1 4 cicatricata 4 5 - 2 EZ 5 19 - 3 1 0 3 1 5 MIDP 1 0.9 77 20 3 - 8 - - - 18 3 0 2 EZ 4 4 MIDN 1 1 0 20 16 - 15 20 8 15 0 25 - 4 20 Chamberlain MIDV - 4 EZ 0 4 15 - 15 3 PPAA 19 3 3 34 EZ - - 0 25 0 35 1 EZ 0 AN EZ 1 0 3 4 EZ 3 - 44 EZ 216 - 0 0 EZ AN 30 - EZ 1 5 0 PPBA 5 - PPAZ 1 EZ - - EZM - 4 6 1 - 0 0 0 243 0 1 PPCA 5 1 PPCB - - 1 5 - 1 EZ 0 0 - 0.7 - 0 30 4 EZ 6 - 1 - - - 0 2 - 1 0 - 40 4 01.05.00 ------4 EZ 1 0 1 - 3 40 5 - EZ 1 0 - 0 30 0.7 - - - 0 0 20 EZ EZ 01.01.00 4 Aubreville EZ - EZ 2 - - 0 0.7 EZ 0.78 - EZ 0 1 0.78 01.01.00 35 Aubreville - - 4 EZ - - 0.78 0.78 - 3 EZ - EZ - 0.78 - - 30 EZ 0.78 40 0.78 - - - EZ - EZ - EZ EZ - - Aubreville - - - Aubreville Aubreville Aubreville 212 Appendix B Family... SapotaceaeSapotaceae GenusSapotaceaeSapotaceaeSapotaceae PouteriaSapotaceae PouteriaSapotaceae PouteriaSapotaceae Pouteria SpeciesSapotaceae PouteriaSapotaceae Pouteria cladanthaSapotaceae Pouteria coriaceaSapotaceae Pouteria cuspidataSapotaceae Pouteria decorticansSapotaceae Pouteria egregia PFT 11 SGSapotaceae Pouteria engleri HGSapotaceae Pouteria eugeniifolia 7 2 Code 16Sapotaceae Pouteria filipes 4Sapotaceae Pouteria 3 glomerata NNOU 3 2Sapotaceae Pouteria gonggrijpii 3 NPSE PPCC 19Sapotaceae Pouteria 4 grandis 4 2 11 1Sapotaceae Pouteria guianensis 0 NPAR 3 PPCE PPCDSapotaceae Pouteria hispida 3 NTOT 4 PPDA 2Sapotaceae Pouteria jariensis 5 0 12 0 12 VRI 4 1Sapotaceae Pouteria laevigata 3 0 PPEA 3Sapotaceae 3 Pouteria melanopoda 3 16 PPEC 16 Height PPEBSapotaceae Pouteria oblanceolata 0 1 0 3 6 3Sapotaceae 7 Pouteria reticulata Disp. 3 0 0 3 11Sapotaceae 0 Pouteria retinervis PPGA 4 PPFA PPGB 16 4 4Simaroubaceae Density Pouteria 4 sagotiana 8 2 4 0 0 17 Architecture Simaroubaceae Pradosia singularis 0 0 0 3 PPGD 3 PPGC 0 2Simaroubaceae Pradosia torta 3 3 0 Quassia 4 3 29Simaroubaceae 0 Pradosia trigonosperma 1 4 Simaba 1 PPHA 4 1 0Simaroubaceae venosa 0 0 1 0 Simaba 6 7 PPJA PPMA 0 3Solanaceae virescens 0 4 0 PPLA Simaba 4 4 3 10SolanaceaeSolanumasperum612SOLA000066EZ-- cochlearia 3 4 PPOA 2 1 Simarouba 1 2SolanaceaeSolanumcrinitum211SOLC000055EZ-- 2 ptychandra 3 0 2 0 0 1 1 amaraSolanaceae 3 4 3 surinamensis PPRA 7 4 0 2 - 2Solanaceae cedron Brunfelsia 169 20 3 PPRB 0 0 1 0 16 0SolanaceaeSolanumrugosum211SOLR000055EZ-- 4 morettii PPSA 0 19 21 0 8 amara 3 0 12 PPTBSterculiaceae 1 multiflora 0 PPSB 12 15 27 3 - 1 202 0Sterculiaceae Solanum 6 3 - EZ 3 0 1 3 1 57Sterculiaceae 1 Solanum 3 0 4 1 guianensis 4 0 3 1 5Sterculiaceae PPTA 1 - Guazuma 0 5 3 EZ 40 EZ 1 7 3 PPVA... - 0 Herrania 4 - 1 0 PPVB 8 25 PRAC 0 EZ 61 PRAP 2 0 Sterculia 0 0 PRAS 15 2 1 0 0 25 2 leucocarpon - - 2 Sterculia 11 EZ 3 0 1 2 - 35 3 25 0 3 paludosum 2 EZ - - 0 2 4 2 4 ulmifolia EZ 1 0 QUAV 2 0 0 1 EZ 4 - - 2 30 kanukuensis 1 0 SIMC 8 0 - Aubreville Aubreville EZM SIMM 6 EZ 1 SIMN - excelsa - EZ 0 - SINA 1 0 1 Aubreville frondosa 0 2 0 17 30 0 0 01.01.15 1 1 0 1 EZ - Aubreville BRUN 0 - - Aubreville 0 175 1 8 12 0 - - - 1 0 2 - 0 Aubreville Aubreville EZ EZ 0 4 1 3 5 - 5 0 3 1 Aubreville 175 SOLL 0 - 15 1 - 1 Aubreville 20 15 EZ 3 0 Aubreville 2 SOLP EZ 0 - - 01.05.00 0 0 EZ 2 Aubreville 109 - GUAZ 12 2 0 - 0 HERK Aubreville 25 20 3 - - 4 0 4 1 EZ - - 0 4 0 - 109 0 Aubreville 0 EZ 4 STEE EZ STEF 40 EZ 1 20 - 0 - 35 6 Aubreville 25 0 EZ 2 2 3 0 Aubreville - 5 - Aubreville - 0 0 25 - EZ 5 - 0 EZ EZ - EZ Aubreville 0 0 5 0 10 0 4 EZ Aubreville EZ 0.9 0 Aubreville 2 - 15 - - 0.87 - 0 Aubreville EZ 0 - 0 30 0 - - 0 EZ - 6 4 EZ Aubreville Aubreville 6 0.55 EZ EZM 6 - 3 Aubreville 2 4 - 0.55 8 EZ - 0.45 0.55 5 - 6 20 - - 8 Scarrone - - EZ EZ EZM 33 35 EZ - - - - SZRM - SZRM - 0.57 0.57 - - Rauh - Rauh Lists of tree species 213 Family... SterculiaceaeSterculiaceae GenusStyracaceaeSymplocaceae SterculiaTheaceae TheobromaTheophrastaceae Styrax SymplocosTiliaceae ClavijaTiliaceae SpeciesTiliaceae subincanum pruriens GordoniaTiliaceaeTiliaceae guianensis ApeibaTiliaceae glabratus Apeiba 12Tiliaceae lancifolia ApeibaUlmaceae PFT 15 fruticosa 3 Apeiba SGVerbenaceae 5 HG LueheaVerbenaceae 2 3 Code LueheopsisVerbenaceae 11 echinata 0 4 LueheopsisVerbenaceae THES 4 NNOU Aegiphila glabra TremaVerbenaceae 2 2 3 Citharexylum 11 petoumo STEP NPSEViolaceae Citharexylum tibourbou 3 3Violaceae rosea 18 SYMP NPAR 2 Vitex speciosaViolaceae rugosa 1 STYR 0 NTOT Vitex macrophyllum 14 1 3Violaceae integrifolia spinosum VRI CLAC 0Violaceae 5 micrantha 1 14 14 GORD LeoniaViolaceae 10 0 11 Paypayrola 0 0 HeightVochysiaceae 4 1 1 62 Rinorea 7 1Vochysiaceae 2 Disp. 0 6 14 82 Rinorea APEE compressa 14 4 4Vochysiaceae 6 3 0 Rinorea 3 triflora 6 2Vochysiaceae Density 0 Erisma 0 1 0 66 APEG 1 APEP guianensis 1 Rinorea Architecture Vochysiaceae APET Qualea CITM glycycarpa 1 105 2 0 3 4Vochysiaceae 4 1 Qualea 2 4 0 0Vochysiaceae flavescens 0 10 2 Qualea LUEH 1 LUER 0 0 2 0 AEGI LUESVochysiaceae pectino-squamata 6 Ruizterania 0 CITS 0 1Vochysiaceae pubiflora 7 0 Vochysia 0 TREM 6 0 0 8 0 11 uncinatum 0Vochysiaceae riana Vochysia 0 0 0 0 3 18 caercula 29 0 Vochysia 6 2 2 34 8 1 3 dinizii albiflora 0 Vochysia VITE 0 4 0 rosea 0 2 0 0 3 Vochysia 2 2 0 0 EZM 3 0 8 5 0 29 0 densiflora 15 SZRM 4 PAYP 8 LEON VITT guyanensis RINO 2 0 0.55 2 2 0.57 0 surinamensis 18 20 0 3 3 4 0 9 4 0 0 0 6 tetraphylla RINF 0 18 1 4 0 0 4 2 17 2 25 Nozeran tomentosa EZ Rauh 14 3 0 0 2 EZ ERIU 15 18 2 0 15 RINP 5 0 1 4 - 0 9 0 6 5 1 1 35 5 5 34 0 - EZ 0 0 AN 2 2 0 18 QUAC 2 5 - RINR 4 0 RUIZ 18 0 1 4 5 - 0 4 2 QUAD 6 30 4 3 30 EZ 0 11 - 0 - VOCD 0 25 3 20 0 2 1 VOCG QUAE 6 4 - VOCS 0 5 0 Roux 0 65 0 5 0 10 0 EZ 0.38 VOCT 0 30 EZ 6 17 8 EZ 35 0 0 EZ - - 0 VOCU 0 1 0 6 8 15 0 0 Troll 1 2 0.38 AN 1 0.38 0 0 AN 0 4 0.38 5 AN - 5 EZ 4 0 34 EZ 0 23 0 20 Troll EZ - Troll 0 Troll 0 0.4 181 2 0.65 0 1 0 0 - 10 10 20 - - 4 13 0 27 - EZ 0 0 247 - - 5 Troll 0 3 1 AUT 10 5 AUT 0 EZM - EZ - 5 35 5 - 4 - 0 0.75 5 - 0.7 8 5 7 AUT 4 - 45 4 AN Fagerlind 45 - 50 4 - - 6 40 - AUT 30 6 AN 0.58 35 30 - AN AN - AUT AN - 40 - - AN AN 50 0.7 0.71 AN - - - AN - - Massart - - - AN - - - - 0.45 - - - -

Danksagung

Diese Arbeit entstand am Wissenschaftlichen Zentrum fur¨ Umweltsystemforschung (WZIII), einem interdisziplin¨aren Institut an der Universt¨at Gesamthochschule Kassel. Danken m¨ochte ich an dieser Stelle Herrn Bossel fur¨ seine Bereitschaft, selbst nach der Emeritierung dieser Arbeit als Erstgutachter zur Verfugung¨ zu stehen. Durch die Offenheit des Fachbereiches Physik fachfremden Themen gegenuber¨ konnte die Arbeit in der vorliegenden Form verwirklicht werden. Insbesondere Herrn Fricke sei an dieser Stelle fur¨ die Zweitbegutachtung sowie fur¨ die M¨oglichkeit zur Einreichung der Arbeit in seiner momentanen Form gedankt. Weiterer Dank gebuhrt¨ Andreas Huth fur¨ die Betreuung dieser Arbeit in angenehmer, freundschaftlicher Atmosph¨are. Mit ihm und Thomas Ditzer fanden etliche fruchtbare Fachdiskussionen statt. Ein dreimonatiger Forschungsaufenthalt in den W¨aldern Borneos wurde durch ihr DFG-Projekt m¨oglich gemacht. Kai Reinhard, Frank Kaspar und Thomas Noll als Raumteiler und Mitstudenten verdanke ich etliche hilfreiche Tipps und eine entspannte Arbeitsatmosph¨are. Allen anderen Mitgliedern des WZIII danke ich fur¨ die angenehme Arbeitsatmosph¨are, insbesondere Ulla Marquard fur¨ die leckeren Kuchen. Ohne unsere nationalen und internationalen Kooperationspartner h¨atte diese Arbeit in ihrem Umfang nicht durchgefuhrt¨ werden k¨onnen. Dem Malaysian-German Sustainable Forest Management Project am Forestry Department in Sandakan und allen Mitgliedern verdanke ich eine lehrreiche und wundervolle Zeit in Sabah, Malaysia. Insbesondere Ro- bert Ong und Peter Lagan m¨ochte ich fur¨ ihren freundlichen Empfang danken. Reinhold Glauner von der Bundesanstalt fur¨ Holz- und Forstwirtschaft, Hamburg, gebuhrt¨ Dank fur¨ die Einfuhrung¨ in das malayische Forestry Department, sowie fur¨ fachliche Unterstutzung¨ der Kasseler Modellierungsprojekte. Masirom Rundi danke ich fur¨ sein Wissen uber¨ Lichtbedarf einzelner Baumarten. Mit Artur Gralla verbrachte ich eine angenehme Zeit im Regenwald. Zainal Abidin Jaafar und Hubert ”We are onto somethingPerol haben den notwendigen Humor mitgebracht. Ludwig Kammesheidt (Institut fur¨ Waldbau, Universit¨at G¨ottingen) und J´erˆome Cha- ve (Service de Physique de l’Etat´ condens´e, Universit´e Paris-Sud and Department of Ecology and Evolutionary Biology, Princeton University) danke ich fur¨ die erfolgreiche Zusammenarbeit und ihr Wissen uber¨ sudamerikanische¨ W¨alder. Ihre Arbeiten haben eine Modellanwendung in Venezuela und Franz¨osisch Guayana erst m¨oglich gemacht. Freund- licherweise haben sie mir ihre Artenlisten fur¨ den Anhang dieser Arbeit zur Verfugung¨ gestellt. Fur¨ die Verwendung etlicher Felddaten aus unterschiedlichsten Projekten bin ich allen Beteiligten dankbar, insbesondere Michael Kleine, Robert Ong, Mark Schlensog (Sabah), Ludwig Kammesheidt (Venezuela), J´erˆome Chave und Bernard Ri´era (Franz¨osisch Gua- yana). 216

Finanziert wurde diese Arbeit durch ein Projekt der Deutschen Forschungsgemein- schaft (DFG) und ein Promotionsstipendium des Otto-Braun-Fonds an der Universit¨at Gesamthochschule Kassel. Mein Dank gilt auch Susa Lewit und Robin Phipps fur¨ ihre Fehlerkorrekturen. Meinen zwei Frauen, Dagi und Hannah Schneider, danke ich fur¨ all den Rest.