Citation
Adaptations to Heterogenous Habitats: Life-History Characters of Trees and Shrubs

Material Information

Title:
Adaptations to Heterogenous Habitats: Life-History Characters of Trees and Shrubs
Creator:
ZANNE, AMY ELISE ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Cotyledons ( jstor )
Forest growth ( jstor )
Forest habitats ( jstor )
Forest trees ( jstor )
Forests ( jstor )
Grasslands ( jstor )
Seedlings ( jstor )
Shrubs ( jstor )
Species ( jstor )
Trees ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Amy Elise Zanne. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
5/1/2005
Resource Identifier:
80818452 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

ADAPTATIONS TO HETEROGENOUS HABITATS: LIFE-HISTORY CHARACTERS OF TREES AND SHRUBS By AMY ELISE ZANNE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2003

PAGE 2

To my mother, Linda Stephenson, who has always supported and encouraged me from near and afar and to the rest of my fam ily members, especially my brother, Ben Stephenson, who wanted me to keep this short.

PAGE 3

ACKNOWLEDGMENTS I would like to thank my advisor, Colin Chapman, for his continued support and enthusiasm throughout my years as a graduate student. He was willing to follow me along the many permutations of potential research projects that quickly became more and more botanical in nature. His generosity has helped me to finish my project and keep my sanity. I would also like to thank my committee members, Walter Judd, Kaoru Kitajima, Jack Putz, and Colette St. Mary. Each has contributed greatly to my project development, research design, and dissertation write-up, both in and outside of their areas of expertise. I would especially like to thank Kaoru Kitajima for choosing to come to University of Florida precisely as I was developing my dissertation ideas. Without her presence and support, this dissertation would be a very different one. I would like to thank Ugandan field assistants and friends, Tinkasiimire Astone, Kaija Chris, Irumba Peter, and Florence Akiiki. Their friendship and knowledge carried me through many a day. Patrick Chiyo, Scot Duncan, John Paul, and Sarah Schaack greatly assisted me in species identifications and project setup. A number of willing undergraduates agreed to toil through my many seed and seedling lab measurements including Carolyn Cross, Robyn Gartner, Jessi Patti, Laura Stockman, and Gabe Wagman. I thank them for their patience and diligence. I received help with soil analyses from Jacob Aniku, Nick Comerford, Hugh Popenoe, Larry Schwandes, and Ted Schuur. Both Nick Comerford and Larry Schwandes graciously opened their labs for me to conduct my analyses. Ben Bolker, Kavita Isvaran, Michael Palmer, Suhel Quader, and Nat Seavy provided assistance with statistical analyses. I would especially like to thank Ben Bolker, as I would have been unable to conduct many of the analyses without his iii

PAGE 4

guidance. I also thank Nat Seavy who was always on the lookout for new ways to analyze my data. Kavita Isvaran, Rebecca Kimball, Suhel Quader, and Doug Soltis assisted me with phylogenetic analyses. I would like to thank the staff of the Department of Zoology and Botany for their assistance throughout my graduate career. I would like to thank my family, especially my mother. They were supportive and encouraging whether or not they knew what I spent all those years doing. I would also like to thank Patrick Chiyo for being such a wonderful companion whether it was exploring the remote wilds of his home in Uganda or the remote wilds of my home in New Hampshire, especially in the winter with no heat. Many thanks are extended to friends in Gainesville. They have been both warm friends and academically-stimulating colleagues. I would like to thank the members of AAPIGS as our meetings led to many important discussions and insights. I extend special thanks to my officemates, Kavita Isvaran and Suhel Quader, who have appeared in many of the categories throughout this list. They were kind enough to open both their home and their office to me (especially toward the end of my dissertation, when their help was most needed). I would also like to thank Sarah Bouchard and Laura Sirot, who entered the graduate program with me and Scot Duncan and Denise Trunk, who were already here when I arrived. They have been strong supports for many years. This research was supported in part by College of Liberal Arts and Sciences Grinter and Dissertation Fellowships and a Ford Foundation Grant. I would also like to thank the Uganda Wildlife Authority, Makerere University, and Government of Uganda for allowing me to conduct my research in Kibale National Park. iv

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS....................................................................................................iii LIST OF TABLES.............................................................................................................vii LIST OF FIGURES...........................................................................................................ix ABSTRACT......................................................................................................................xi CHAPTER 1 GENERAL INTRODUCTION.....................................................................................1 Life-History Strategies................................................................................................1 Life-History Strategies: Study Objectives...................................................................2 2 TROPICAL FOREST DIVERSITY AND HABITAT ASSOCIATIONS: COMMUNITY VERSUS SPECIES PATTERNS........................................................5 Introduction................................................................................................................5 Materials and Methods...............................................................................................9 Study Site............................................................................................................9 Vegetation.........................................................................................................10 Soils and Light...................................................................................................11 Analyses............................................................................................................12 Results.....................................................................................................................14 Environment......................................................................................................14 Vegetation: Community Response....................................................................16 Vegetation: Species Associations.....................................................................17 Discussion................................................................................................................18 3 DO LIFE-HISTORY CHARACTERS OF JUVENILE TREES AND SHRUBS REFLECT ADAPTATIONS TO CONTRASTING HABITATS?.................................37 Introduction..............................................................................................................37 Materials and Methods.............................................................................................40 Study Site..........................................................................................................40 Methods............................................................................................................40 Analyses............................................................................................................43 Results.....................................................................................................................45 Life-History Character Variation........................................................................45 Cotyledon Types and Other Categorical Variables...........................................46 Mean Character Comparisons among Discrete Categories..............................47 v

PAGE 6

Seed Mass and Continuous Variables..............................................................48 Multivariate Analyses........................................................................................48 Phylogenetic Effects..........................................................................................49 Discussion................................................................................................................51 Cotyledon Type.................................................................................................51 Seed Size..........................................................................................................55 Suites of Characters..........................................................................................57 Juvenile Characters and Environmental Associations......................................57 Phylogeny.........................................................................................................58 Conclusions.......................................................................................................59 4 ADAPTATION AND VARIATION OF TREE AND SHRUB SEEDLINGS: SURVIVAL AND GROWTH ACROSS HABITATS...................................................69 Introduction..............................................................................................................69 Materials and Methods.............................................................................................72 Study Site..........................................................................................................72 Methods............................................................................................................73 Analyses............................................................................................................75 Results.....................................................................................................................80 Growth Analyses...............................................................................................80 Survival Analyses..............................................................................................81 Leaf and Cotyledon Turnover Rates.................................................................82 Discussion................................................................................................................82 Plasticity among Habitats..................................................................................82 Adaptations to Habitats.....................................................................................84 Other Determinants of Growth and Survival.....................................................85 Other Determinants of Leaf and Cotyledon Turnover Rates and Herbivory.....86 Conclusions.......................................................................................................87 5 GENERAL CONCLUSIONS...................................................................................100 APPENDIX A JUVENILE AND ADULT CHARACTERS FOR 80 SPECIES.................................102 B TEST RESULTS FOR 80 SPECIES......................................................................108 C SPECIES SCORES FOR 48 SPECIES.................................................................111 D ECOLOGICAL CHARACTERS AND GROWTH FOR 24 SPECIES......................113 LIST OF REFERENCES...............................................................................................117 BIOGRAPHICAL SKETCH............................................................................................128 vi

PAGE 7

LIST OF TABLES Table page 2-1. Eigenvalues, cumulative % variance, and factor loadings for environmental variables along axes with strong eigenvalues.......................................................23 2-2. Means for stem density and species richness using rarefaction in each habitat by growth form for 24 small plots..........................................................................24 2-3. Species, growth forms, and results from randomization tests with 10,000 iterations for species abundances in 24 small plots..............................................25 2-4. Species, growth forms, and results from randomization tests with 10,000 iterations for species abundances in 16 large plots...............................................26 3-1. Summary of the results for univariate ANOVAs and t tests comparing means of continuous characters among categories of discrete characters.......................60 3-2. Pearson product moment correlations between total seed mass and other continuous characters for data when phylogenetic relationships are not incorporated and incorporated using Independent Contrasts................................61 3-3. Eigenvalues and factor loadings for seed, seedling, and adult characters along the first four axes using principal components analyses for 47 species in Kibale National Park, Uganda................................................................................62 3-4. Distribution of cotyledon type categories and seed size categories among the 15 families with more than one species for 70 tree and shrub species from Kibale National Park, Uganda................................................................................63 3-5. Significant correlated changes incorporating phylogenetic relationships between discrete characters for 70 tree and shrub species in Kibale National Park, Uganda.........................................................................................................64 3-6. Eigenvalues and factor loadings for continuous seed, seedling, and adult characters with phylogenetic relationships considered along the first two axes using principal components analyses....................................................................65 4-1. Seed and seedling characters for 24 species in Kibale National Park, Uganda....89 4-2. Pearson product moment correlations between growth and mortality for seedlings planted in forest and that same character measured for seedlings planted in the three other habitats.........................................................................90 vii

PAGE 8

4-3. Results for the analysis with generalized linear mixed models to test whether relative growth rate depends on other species characters....................................91 4-4. Results for Forward Stepwise Cox Regression Analyses for seedling mortality....92 4-5. Results for the analysis with generalized linear mixed models to test whether % leaf and cotyledon herbivory, leaf production, and leaf and cotyledon mortality depend on other species characters.......................................................93 A-1. Juvenile and adult characters for 80 species.......................................................103 B-1. Test results for 80 species...................................................................................109 C-1. Species scores for 48 species.............................................................................111 D-1. Ecological characters and growth rates for 24 species.......................................114 viii

PAGE 9

LIST OF FIGURES Figure page 2-1. Means for environmental variables along 10 transects in four habitats in Kibale National Park, Uganda................................................................................27 2-2. Ordination along Axes 1 and 2 using principal components analyses in 10 transects each in four habitats in Kibale National Park, Uganda...........................29 2-3. Ordination along Axes 1 and 2 using principal components analyses in 10 transects each in three forested habitats in Kibale National Park, Uganda...........31 2-4. Number of species versus number of stems using rarefaction for different growth forms in all size classes in small plots in four habitats..............................33 2-5. Plot ordination based on species along Axes 1 and 2 using detrended correspondence analyses in 24 small plots each in four habitats in Kibale National Park, Uganda...........................................................................................34 2-6. Plot ordination based on species along Axes 1 and 2 using detrended correspondence analyses in 16 large plots each in two habitats...........................35 2-7. Plot ordination based on species along Axes 1 and 2 using principal components analyses in 24 small plots each in three habitats..............................36 3-1. Frequency of cotyledon types among seed, seedling, and adult characters for species in Kibale National Park, Uganda.........................................................66 3-2. Relationship between total seed mass and inverse cotyledon thickness using Independent Contrasts for 54 species in Kibale National Park, Uganda......67 3-3. Biplot of seed, seedling, and adult characters and species along Axes 1 and 2 using principal components analyses for 47 species in Kibale National Park, Uganda..................................................................................................................68 4-1. Relationship between growth and its components in forest and that same character in the other three habitats for 24 species in Kibale National Park, Uganda..................................................................................................................94 4-2. Distribution of species mean relative growth rate for species with photosynthetic versus storage cotyledons calculated in three ways of biomass estimation................................................................................................95 ix

PAGE 10

4-3. Proportion of seedlings surviving as a function of time since planting for different planting habitats for 24 species in Kibale National Park, Uganda...........96 4-4. Proportion of seedlings surviving as a function of time since planting for species with different habitat associations for 24 species in Kibale National Park, Uganda.........................................................................................................97 4-5. Means of species measures of leaf and cotyledon characters for forestand open-associated species for 12 forestand 12 open-associated species in Kibale National Park, Uganda................................................................................98 4-6. Means of species measures of leaf and cotyledon characters for seedlings growing in forest, gap, edge, and grassland habitats for 24 species in Kibale National Park, Uganda...........................................................................................99 x

PAGE 11

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ADAPTATIONS TO HETEROGENOUS HABITATS: LIFE-HISTORY CHARACTERS OF TREES AND SHRUBS By Amy Elise Zanne May 2003 Chair: Dr. Colin A. Chapman Major Department: Zoology Species distributions are thought to be influenced by life-history characters that allow individuals to establish and persist in particular environments. Understanding which characters represent adaptations to particular habitats is especially important in tropical environments where tree species richness is extremely high. In this study conducted in Kibale National Park, Uganda, I first examined tree and shrub species distributions in four of the major habitats: closed canopy forest, treefall gap, grass/forest edge, and grassland. I also measured environmental factors: light, root length density, bulk density, organic matter, particle size, pH, field moisture capacity, and water content. Forest and gaps were similar environmentally, except that gaps had higher light. Edge and grassland were fairly similar with edge being drier. Forested habitats had similar species composition, but grassland was distinct. Despite many similarities, common species had strong associations with one or two habitats. From this study, species were assigned habitat associations. Seeds and seedlings from 80 species were then described for seed size, dispersal type, days to germination, % germination, cotyledon morphology, initial seedling size, and cotyledon thickness. Large seeds with thick xi

PAGE 12

storage cotyledons, slow germination, large stature adults, and dispersal by large bodied animals were common in forest and gap species. Seedlings for 24 of these species varying in habitat association, seed size, and cotyledon morphology were transplanted in the four habitats. Seedlings were measured for growth, survival, and leaf and cotyledon herbivory and turnover rates. Seedling growth was greatest for seedlings planted along the edge. Survival was greatest for seedlings planted in gaps. Leaf herbivory and turnover rates were highest for seedlings growing in gaps. No differences were found between forest(forest and gap) and open(edge and grassland) associated species for growth. However, survival was higher and leaf herbivory and mortality were lower in forestthan open-associated species. These results suggest that light, and perhaps soil moisture, may exert strong selective pressures, with species associated with high light, where competition may be high, allocating more to construction of less well defended tissue, whereas species associated with low light allocating more to longer lived tissue. Such understanding of juvenile life-history character associations provides insight into adult distributions and perhaps tropical species richness. xii

PAGE 13

CHAPTER 1 GENERAL INTRODUCTION Life-History Strategies Organisms exhibit variation in life-history characters that maximize reproduction and survival (fitness) in particular environments (Stearns 1992). Ideally an organism should allocate its resources to optimize all possible characters (e.g., early reproductive maturity, large size at reproductive maturity, and large and numerous offspring; Partridge & Harvey 1988) to maximize fitness. In actuality, organisms are kept from realizing these optimizations by a lack of available resources. Linked characters may compete for limited resources leading to trade-offs between a set of characters (Southwood 1988, Stearns 1992). Such a trade-off has been suggested, for example, between offspring size and number (Smith & Fretwell 1974, Southwood 1988, Stearns 1992). A parent is constrained from producing many large offspring by limited resources; as resources are increased per offspring, the number of offspring must decrease. But, as energy per offspring is increased, the fitness of that offspring is increased (Smith & Fretwell 1974). Solutions to the choice of allocation patterns between competing characters may differ between phylogenetically distinct organisms. Also constraints in possible expression of characters are due to the limitations in genetic variability (Partridge & Harvey 1988, Southwood 1988). By determining patterns of trade-offs within species and comparing these patterns between species, one can relate patterns caused by recent selection within the environment and patterns attributable to shared lineages (Stearns 1992). While some characters are negatively related (trade-offs), others are positively associated (Stearns 1980, Southwood 1988). These suites of related characters are often correlated with variation in habitats (Franco & Silvertown 1997). An early proposal 1

PAGE 14

2 of such a framework was the idea of r-and K-selected species in which r-selected species were early colonizers experiencing density-independent growth (short life, fast development, early reproduction, and many small offspring), and K-selected species were good competitors experiencing density-dependent growth (long life, slow development, late reproduction, and few large offspring; MacArthur & Wilson 1967, Loehle 1988). Support for this classification has been equivocal (Stearns 1977, Partridge & Harvey 1988). Other researchers (Grime 1977) extended this framework, contrasting life-history patterns along environmental gradients with changes in severity of conditions and disturbance levels (Franco & Silvertown 1997). Grime (1977) recognized three suites of characters including the ruderal strategy that is typical of species adapted to high disturbance but low severity conditions, the competitive strategy that is typical of species adapted to low disturbance and low severity conditions, and the tolerant strategy that is typical of low disturbance and high severity conditions. Some studies have supported these schemes (Leps et al. 1982, Southwood 1988), and while simplistic, these models suggest some general trends. A more recent approach has been developed by Charnov (1991) for animals and expanded by Franco and Silvertown (1997) to plants in which organisms can be separated between those living a ‘fast pace’ versus ‘slow pace,’ which are similar to the r-and K-selected species, respectively. Empirical work supports associations between some characters and habitats, but the generality of particular suites of characters is still under debate. Life-History Strategies: Study Objectives In this study, I explore the associations among tree and shrub juvenile life-history characters and associations between characters and habitats: closed canopy forest, treefall gaps, grass/forest edge, and grassland, in Kibale National Park, Uganda. Understanding how life-history characters influence tree and shrub species distributions

PAGE 15

3 in a tropical forest and embedded grassland is particularly interesting as they can provide insight into the maintenance of high diversity of species found in the tropics. Chapter 2 examines tree and shrub distributions among the four habitats by placing small plots (5 x 5 m) in all four habitats, measuring all size classes of shrubs and trees, placing large plots (10 x 50 m) in forest and grassland habitats, measuring just larger size classes of shrubs and trees. Community composition was compared among the four habitats using Principal Components and Detrended Correspondence Analyses to determine the extent to which communities overlap. Next, common species distributions were examined across the four habitats to determine if species are strongly associating with a single habitat. Both light and soil moisture have been proposed as particularly limiting factors to seedlings (Burslem 1996, Agyeman et al. 1999). Environmental variables related to light and soil moisture (photosynthetically active radiation; root length density; bulk density; % organic matter; particle size distribution; pH; field capacity; and mean, minimum and maximum % water content) were measured along transects in the four habitats. By investigating differences among habitats and how species are associated with those habitats, the role of habitat specialization in influencing high tropical tree diversity is explored. In this study both (within forested habitats) and (between forested habitats and grassland habitat) diversity are examined. In Chapter 3, seeds were collected from all four habitats and seedlings germinated for 70 tree and shrub species to determine juvenile characters. I was especially interested in characters indicating resource availability to the developing seedling (seed size and cotyledon function: photosynthetic versus storage), as these characters should be under strong selective pressure. Seeds were also measured for dispersal type, amount of seed made up of embryo and endosperm, number of days to germination, and % germination. Seedlings were also measured for initial seedling size and cotyledon

PAGE 16

4 thickness. Using principal components analyses, I was able to explore the extent to which these juvenile characters are interrelated. Based on results from Chapter 2 and from the literature, species were assigned to particular habitats; species characters were then associated directly with habitats. Because many species were being compared, I used phylogenetic analyses to examine if associations among characters and between characters and habitats were being driven by the phylogenetic relatedness of species Finally, in the Chapter 4, seedlings for 24 species (a subset of the species measured in Chapter 3) were planted in each of the four habitats at the first photosynthetic organ stage. These seedlings varied in their habitat association and their seed size and cotyledon type. Seedlings were followed until the second photosynthetic organ stage for growth. Seedlings were also measured every 3 months for 1 year to determine differential survival and leaf and cotyledon production, mortality, and % herbivory. By comparing species typically associated with different habitats, especially species from more forested habitats (forest and gap) versus species from more open (edge and grassland) habitats, I was able to investigate whether such species show fundamental differences in growth and survival across habitats. Furthermore, I was able to examine whether seedlings vary overall in their responses to different habitat (e.g., if some habitats have seedlings with high growth or survival). The influence of other characters (such as seed size and cotyledon function) was also examined with relation to growth and survival. Put together, Chapters 2 through 4 allowed me to investigate species habitat associations, associations among juvenile life-history characters and these habitats, and finally how these characters influence growth and survival of seedlings when planted in the habitats. Results from Chapters 2 to 4 are compared and summarized in Chapter 5.

PAGE 17

CHAPTER 2 TROPICAL FOREST DIVERSITY AND HABITAT ASSOCIATIONS: COMMUNITY VERSUS SPECIES PATTERNS Introduction Tropical rain forests are noted for their extraordinary species diversity (Connell 1978, Phillips et al. 1994, Wright 2002) with many such forests containing well over 100 tree species per ha, and records as high as 283 species per ha (Phillips et al. 1994). This diversity has spawned numerous studies of what permits and maintains coexistence (Grubb 1977, 1996, Connell 1978, Hubbell 1997, Brokaw & Busing 2000, Harms et al. 2001), and various processes have been proposed to explain this maintenance, including both deterministic and random processes (reviewed by Wright 2002). Species specialization to particular habitats or niches (e.g., resource levels within habitats; Whittaker et al. 1973) in landscapes that are heterogeneous over space and time is one deterministic model of coexistence that has been suggested (Grubb 1977, Chesson & Huntly 1997, Clark et al. 1999, Wright 2002). Grubb (1977) argued that environmental heterogeneity might be particularly important at the regeneration stage. As a community varies temporally, species with different regeneration strategies (e.g., ability to germinate in low light) may establish at different times but come to share contiguous canopy positions as adults, thus increasing diversity (Wright 2002). I explore the degree to which habitat specialization may be occurring among trees and shrubs in different habitats in Kibale National Park, Uganda. In the literature, habitat specialization has found some measure of support (Grubb 1977, Wright 2002), but others have failed to find strong evidence of specialization (Hubbell & Foster 1986, 5

PAGE 18

6 Webb & Peart 2000, Harms et al. 2001). In a study in Panama, Harms et al. (2001) found some species to be differentially distributed across habitats (less than one-third), but most showed little affinity to a particular habitat, suggesting specialization is unlikely to be a mechanism maintaining high tropical tree diversity. To evaluate the potential role of specialization in influencing species distributions across habitats, I use several lines of evidence that have been used in other studies. First, I examine if environmental variation occurs among habitats. Without understanding underlying habitat differences, it is difficult to understand potential mechanisms influencing species distributions. A second line is to examine if community differences occur across habitats. Communities that differ greatly from one another in species composition suggest greater specialization of species to particular habitats. This is a coarse-grain examination of differences among habitats; however, even if large-scale community differences do not occur, some individual species may still be specialized to that habitat. A third line is a more fine-grain approach that involves determining if individual species show habitat associations. If no associations are found, then specialization of species to habitats is unlikely to be determining species occurrences. It must be remembered that habitat association does not necessarily mean habitat specialization. Species presence in a habitat may be due to historical patterns, anthropogenic causes, biological interactions (fundamental habitat does not equal realized habitat), current transient patterns (invasion into new territory), or habitat specialization (Harms et al. 2001). Many studies examining habitat associations have investigated environmental gradients and species associations within habitats (niche specialization), but have failed to find strong associations for most of the species in their studies (Hubbell & Foster 1986, Clark et al. 1999, Svenning 1999, Webb & Peart 2000, Harms et al. 2001). Furthermore, tree species often occur in more than one habitat type (de Carvalho et al.

PAGE 19

7 2000). These results suggest that investigating environmental gradients and species associations may be best if first explored among habitats (Terborgh et al. 1996, Williams-Linera et al. 1998, de Carvalho et al. 2000, Oosterhoorn & Kappelle 2000). I examine differences at two levels: among several forested habitats examining ( diversity); and between these forested habitats and grassland habitat ( diversity). The first line of evidence, to identify environmental differences among habitats, involves examining influential environmental variables. Burslem (1996) found that light is an important factor affecting seedling growth in tropical forests. Similarly, Agyeman et al. (1999) showed that light was the strongest abiotic determinant of seedling growth rates in shade-house experiments. Other studies found that water availability is an important environmental factor influencing forest composition (Swaine 1996, Bongers et al. 1999, Clark et al. 1999). In a 17-year study in Ghana, Veenendaal and Swaine (1998) determined that moist semi-deciduous forest soils dried to below wilting point each year, while wet evergreen forest soils dried to below wilting point only every few years. Multiple factors should interact to influence species composition, and this was indeed found in a study in Ghana in which seedling characters were associated with light and drought gradients (Agyeman et al. 1999). Both light and drought significantly contributed to species growth and survival as determined through principal components analyses, although light was more important. Light relationships contributed more to the stronger Axis 1; drought relationships contributed more to the weaker Axis 2. Several studies focused on the second coarse-grain line of evidence, determining differences in overall community composition (Kappelle et al. 1995, Lieberman et al. 1995, Terborgh et al. 1996, Sheil 1999, de Carvalho et al. 2000). In South Africa, Kirkwood and Midgley (1999) were able to distinguish tropical dry forest from evergreen forest communities. While in a Bornean forest, Webb and Peart (2000) found that

PAGE 20

8 multivariate analyses of species were able to distinguish the three physiographic forest types. Various researchers have investigated the third fine-grain line of evidence, determining individual species habitat specializations; these studies vary in spatial scale, although all are conducted within forests. Two studies found that most species were generalists with only 1 in 5 occurring in a specific habitat in the Amazon and 1 in 2.4 occurring in a specific habitat in Panama: A large-scale study in the Amazon in which species were examined over a 400 km 2 area in different forest types (Pitman et al. 1999). A small-scale study in Panama in which species were examined over 0.5 km 2 and classified in terms of their habitat associations and regeneration strategies (Hubbell & Foster 1986). Despite large variation in the spatial scales of studies and types of environmental gradients considered, a number of studies found that generalists appeared to dominate communities [Swaine (1996) with 43% of species indifferent to habitat type, Webb and Peart (2000) with 57% of species indifferent, and Harms et al. (2001) with 25 to 49% of species indifferent (depending on the type of analysis)]. In this study, I investigated the importance of habitat association in structuring species composition in four habitats in Kibale National Park, Uganda: closed canopy forest, forest treefall gap, forest/grassland edge, and grassland. In this investigation, I wanted to characterize associations between habitats and: Environmental variables (i.e., soil bulk density, water content, root length density, field moisture capacity, organic matter, particle size distribution, and pH; and light). Tree and shrub community composition. Individual tree and shrub species distributions.

PAGE 21

9 Materials and Methods Study Site Kibale National Park (766 km 2 ; 0’-0’N and 30’-30’E) is located in western Uganda, 24 km east of the Rwenzori Mountains at an elevation of 1500 m. Between 1998 and 2000, a mean of 1760 mm of rain fell per year, annual mean daily maximum temperature was 23.1C, and minimum temperature was 15.1C (C.A. Chapman & L. J. Chapman, unpublished data). Kibale consists of mature forest (57%), colonizing forest (19%), grassland (15%), woodland (4%), swamp (4%), and plantations of exotic trees (1.0%; Chapman & Lambert 2000). Near the field station Kanyawara where this study was conducted, the forest is mid-altitude, moist evergreen (Howard 1991). Andisol soils are frequent in this area (Jacob Aniku, personal communication). Kibale was originally designated a Forest Reserve in 1932 and received the protected status of National Park in 1993 (Struhsaker 1997). Forestry compartment K30 near Kanyawara is a relatively undisturbed forest of 300 ha. Annual rate of natural treefalls is 1.4% (Skorupa & Kasenene 1984), and mean gap size for K30 was estimated as 256 m 2 (range: 100 to 663 m 2 ; Kasenene 1987). Grasslands are found on hilltops within the northern section of the park, and archaeological evidence suggests they had formerly been agricultural settlements (Kingston 1967). Human settlement declined in these areas during the early-1900s due to an increase in rinderpest that killed cattle (Kingston 1967). Forest has reestablished in some grasslands while the majority remained as grassland due to fire, elephant activity, and/or competitive dominance of grasses over trees (Lang Brown & Harrop 1962, Kingston 1967, Wing & Buss 1970). Queen Elizabeth National Park, contiguous with Kibale, is dominated by grasslands; many species of shrubs and trees, especially in the

PAGE 22

10 Fabaceae, are found in grasslands in both parks (Lock 1993, Lenzi-Grillini et al. 1996). Notably missing from Kibale, however, are species in the genus Euphorbia. Vegetation From May 1998 to July 2000, I quantified tree and shrub species richness (species/stem using rarefaction; Brzustowski 2002) and tree stem density (stems/m 2 ) in 24 small plots (5 x 5 m) in each of four habitats (closed canopy forest, treefall gaps, forest/grassland edge, and grassland). All size classes of shrubs and trees (newly germinated seedlings to trees) were counted. These plots were randomly located within each habitat. For gap plots, I located treefall gaps throughout the forest in which fallen trees had no fine branches and leaves remaining yet little decay had occurred on the main bole. I randomly chose a subset of these gaps. Gap size was determined by measuring the longest axis and the second longest axis perpendicular to the first that were cleared of vegetation to within 2 m of the forest floor (Brokaw 1982); gaps averaged 300 m 2 (range: 39 to 636 m 2 ). Since few large trees and shrubs (> 2 m) occurred in small plots (5 x 5 m), I sampled this larger size class of trees and shrubs using large plots (10 x 50 m) randomly located in forest and grassland habitats (not gap and edge habitats). I used the species names found in recent literature (Polhill 1952, Hamilton 1991, Katende et al. 1995, Lwanga 1996). All shrubs and trees were categorized by growth form as shrub, treelet (sometimes growing as a shrub and sometimes growing as a tree), or tree according to Lwanga (1996) and Polhill (1952). Shrubs are self-supporting woody plants branching near the ground and either having several stems from the base or a single stem but short in stature (< 2 m). Trees are woody plants with a single main stem and tall in stature (> 2 m; Henk Beentje, personal communication). When counting shrub stems in a plot, if shrubs were multi-stemmed, all stems originating from the same plant were counted as a single stem.

PAGE 23

11 Soils and Light I characterized each habitat with respect to soil bulk density (soil dry mass per volume), soil gravimetric water content (% soil water by dry mass), estimated soil field moisture capacity (% water remaining in soil 1 day after saturation), soil particle size distribution (% sand, silt, and clay by dry mass after removal of organic matter), soil organic matter (% organic matter by dry mass), soil pH, root length density (sum of the lengths of roots per volume of soil) separated for roots with diameters < 2 mm and > 2 mm, and photosynthetically active radiation (the portion of the light spectrum plants use). Ten randomly-placed, permanent, 20 m transects were sampled in each habitat. I took soil samples for soil and root measures by first clearing away litter and removing a 2 cm diameter core to 15 cm depth. In January 2000, 10 soil cores were taken approximately every 2 m along each transect; cores were combined within transects. Adjacent to the first five cores, I took five more cores of the same size for root sampling; root cores were combined within transects. From February to June 2000, five soil cores per transect were taken each month for water content measures. For bulk density, larger diameter cores (6.4 cm) were taken in June 2002 and treated similarly to the smaller diameter cores. Soil wet and dry masses were measured; dry mass was determined after drying to constant mass at 105C. I calculated bulk density (BD = soil dry mass/volume; g/cm 3 ) for June 2002 samples and gravimetric water content (WC = [wet mass-dry mass]/dry mass*100 (%)) monthly for January to June 2000 samples (Klute 1986). An estimate of field moisture capacity was taken by saturating soils and draining them for 24 hours at room temperature. Soils were weighed and then dried to constant mass at 105C and reweighed. This methodology is only a relative estimate of field moisture capacity as it was not measured in situ and much of the gross soil structure is lost.

PAGE 24

12 Particle size distribution (% sand, silt, and clay) was determined by pipet method in which dry mass of sand and clay was determined for a soil sample after organic matter was removed (Klute 1986). Particle size distribution was measured for all forest, gap, and edge plots but only three grassland plots, as other grassland plots had high organic matter values that would have lead to inaccurate measures of particle size distribution. Soil organic matter was determined by measuring loss on ignition in a muffle furnace at 550C for 24 H, and pH was measured on 3 g of soil in 3 mL distilled water (Page et al. 1982). For root cores, roots were separated from soils by sieving samples through 2 m and 250 mm mesh and extracting remaining roots; roots were preserved in ethanol. Root length densities (Lv, cm/cm 3 ) for small (< 2 mm) and large (> 2 mm) roots were determined by determining total root length in a given volume of soil using GSRoot 4.0 Automated Root Length Measurement (PP Systems, Haverhill, Maine). Photosynethically active radiation (PAR;mol/m 2 s), the amount of light in the portion of the spectrum useed by plants hitting a given area over a given time period, was measured during dry (June) and wet (October) seasons with an LI-250 light meter and an LI-190SA quantum sensor (Licor, Lincoln, NE) along 20 randomly placed transects in the four habitats. All measurements were taken between 1000 and 1400 h. Two light readings were taken at 5 cm above the ground at each of four locations approximately 5 m apart along each transect. Measurements were averaged per transect. Analyses Principal component analyses (PCA) were used for two analyses to describe 1) relationships among environmental variables in all four habitats and 2) ecological relationships among tree and shrub species in forested habitats. Detrended correspondence analyses (DCA) were used to describe ecological relationships among

PAGE 25

13 tree and shrub species in all four habitats (forested and grassland). I chose to use PCA for comparisons of my three forested habitats, because limited species compositional differences occurred among them. Because of the substantial difference in species composition between the forested habitats and the grassland, I used DCA for these comparisons. Both analyses were conducted with Canoco 4.5 (ter Braak & Smilauer 1999). For PCA, eigenvalues greater than 0.1 were considered strong axes explaining a large percentage of the variation in the data, and when scores or factor loadings were greater than 0.5 for a given axis, those variables were considered strongly related to that axis (Norusis 1994). Organic matter was extremely high in grassland habitat so in seven of 10 grassland plots I was unable to measure particle size distribution (PSD). As a result, PCA were run without particle size distribution. In DCA, because species scores are standardized to 0 mean and 1 standard deviation, axes have constant diversity and are in standard deviation (SD) of species turnover units with 4 units representing complete species turnover. Eigenvalues represent the maximized dispersion of species along an axis so typically an eigenvalue greater than 0.5 denotes that species are well separated along an axis; only eigenvalues greater than 0.5 are considered (Jongman et al. 1987). Since my species abundance data are highly skewed with many small values and a few large values per species, I transformed species data using ln (x+1) transformations (Jongman et al. 1987). To determine species associations with habitats, I ran randomization tests using confidence intervals for common species using programs written in R 1.6.2 (R Development Core Team 2002). Thirty of 105 species in small plots and 35 of 103 species in large plots were common enough to examine their habitat associations. I kept the natural structure of the data within the plots by randomly reassigning plots to habitats without replacement for 10,000 iterations. Significant species associations with a habitat are reported if observed stem counts were more extreme than 95% of randomizations.

PAGE 26

14 With a significance level of 0.05, I would expect that six of the species (30 species with four habitats) in small plots and three to four of the species (35 species in two habitats) in large plots would be falsely found associated (either positively or negatively) by chance alone. Randomization tests with confidence intervals are more appropriate than Contingency table goodness-of-fit (GOF) tests, as GOF tests assume that all stems and plots are independent (Webb & Peart 2000, Harms et al. 2001). This assumption is violated, for example, if several stems establish within a plot after the seeds were deposited in a single dispersal event. For my data, randomization tests with confidence intervals are also more appropriate than randomization tests with deviations statistics ((Randomized – Expected) 2 /Expected; from Webb & Peart 2000) using expected values (null model) determined by multiplying total number of a given species in all habitats by proportion of stems of all species for each habitat (Webb & Peart 2000, Harms et al. 2001). From my analyses, I found these deviation statistics to be sensitive to stem number in each habitat and determined they are more appropriate when stem number is similar among habitats. Results Environment Based on environmental characteristics, forest and gap sites were similar to one another, but differed from edge and grassland sites (Figure 2-1). Factors influencing soil water availability (% organic matter; % clay; field moisture capacity; and mean, maximum, and minimum water contents) were significantly greater in grassland than in other habitats (Figure 2-1). Percent sand and bulk density were significantly greater in forest and gap than grassland and edge; root length density of small roots was significantly greater in edge than in other habitats. Roots less than 2 mm in diameter are the roots absorbing water and nutrients in the soil, so their density is an important

PAGE 27

15 indicator of the amount of belowground competition for resources. Light was significantly greater in gap, edge, and grassland sites than in forest. No significant differences were found with respect to pH and root length density of large roots. The higher root length density of small roots in edge plots and a seedling experiment in which many seedlings appear to die from wilting in edge than any other habitat (Zanne unpublished data) suggest that edge is a more water-limited habitat. In terms of soil color, dry soils are reddish brown in forest and gap, dusky red in edge, and reddish black in grasslands (Munsell Colour Company 1975). Wet soils are dusky red in forest and gaps, very dusky red in edge, and reddish black in grasslands. In the PCA of environmental data using all four habitats (both forested and grassland), most variation (74.1%) was explained by the first three axes, which had eigenvalues > 0.1; thus, only these three axes were considered further (Table 2-1). Most environmental factors were strongly related to Axis 1 with seven of 10 variables having loadings greater than 0.5 (Table 2-1, Figure 2-2a), root length densities of small and large roots were strongly related to Axis 2, and pH was strongly related to Axis 3. When habitat transects are plotted along the first two axes, grassland transects have positive scores and forest and gap transects have negative scores on Axis 1 (Figure 2-2b). Grassland and edge habitats do not overlap with any other habitat, but forest and gap habitats completely overlap with one another. In the PCA of environmental data with only forest habitats (forest, gap, edge), most variation (71.4%) was explained by the first three axes, which all had eigenvalues > 0.1 (Table 2-1); only the first two axes are plotted (Figure 2-3). Most factors were strongly related to Axis 1 with eight of 10 variables having factor loadings greater than 0.5 (Table 2-1, Figure 2-3a), root length densities of large roots and field moisture capacity were strongly related to Axis 2, and pH was strongly related to Axis 3 (although pH was more strongly related to Axis 1). When habitat transects were plotted along the first two axes,

PAGE 28

16 edge transects had positive scores and forest and gap transects had negative scores on Axis 1 (Figure 2-3b). Edge habitats do not overlapped with any other habitat, but forest and gap habitats completely overlapped with one another. Vegetation: Community Response In small plots, stem density of shrubs was highest in grassland and species richness of shrubs was highest in grassland and edge habitats (Table 2-2, Figure 2-4). Stem density of treelets was highest in edge and species richness of treelets was highest in forest and gaps. Forest and gaps had similarly high tree species richness. Tree stem density was greatest in forest plots but was also highly variable. High stem density for forest trees was due to one plot that contained 1149 stems of Diospyros abyssinica. When stem density for D. abyssinica in this plot was removed, no differences were found in stem density between forest, gap, and edge, but these habitats had higher tree stem densities than grassland. In large plots, forest and grassland had similar shrub stem densities per plot (forest: 23.8 + 18.4; grassland: 24.5 + 8.0) but forest had higher shrub species richness per stem (forest: 12.0 + 0.0; grassland: 8.0 + 0.2). Forest had higher stem density per plot and species richness per stem of treelets (forest stems: 50.0 + 22.3, forest species: 14.6 + 1.6; grassland stems: 3.2 + 4.3, grassland species: 6.0 + 0.0) and trees (forest stems: 208.3 + 51.4, forest species: 28.3 + 2.3; grassland stems: 6.9 + 5.3, grassland species: 5.0 + 0.0; Figure 2-4). Detrended correspondence analyses of tree and shrub species abundances for small and large plots revealed complete species turnover among plots along Axis 1 (gradient lengths > 4; Figures 2-5, 2-6), and only this axis showed strong separation of species (eigenvalue > 0.5). Relative abundance of species in small plots separated grassland habitats from edge, forest, and gap habitats (Figure 2-6), but among edge, forest, and gap habitats a complete species turnover did not occur. In large plots,

PAGE 29

17 species data showed complete species turnover between grassland and forest plots (Figure 2-6). Principal components analyses of tree and shrub species abundances for small plots in forested habitats (forest, gap, edge) separated edge plots from forest and gap plots along Axis 1, but forest and gap plots were completely overlapping (Figure 2-7). Only the Axis 1 explained a significant percentage of variance (eigenvalue > 0.1). Principal components analyses of tree and shrub species abundances for small plots using just forest and gap plots was unable to detect any strong axes (eigenvalue > 0.1) suggesting that the forest and gap communities are indeed indistinguishable. Vegetation: Species Associations In small plots using randomization tests, 27 out of 30 species were significantly positively associated with a habitat (Table 2-3). All but one positively forest-associated species were trees. Nine positively gap-associated species were trees, one was a shrub (Coffea eugenioides), and one was a treelet (Dasylepis sp.). Five positively edge-associated species were treelets and five were trees. Three positively grassland-associated species were shrubs and one was a tree. Only three species (Diospyros abyssinica, Dovyalis macrocalyx, and Acalypha spp.) were not significantly positively associated with a habitat. Diospyros abyssinica and D. macrocalyx were both negatively associated with grassland and commonly found in other habitats. Acalypha is a genus with species found in all habitats; of those species with habitat associations with two habitats, only forest and gap or edge and grass shared species. In randomizations tests for species in large plots in grassland and forest, only one species was not significantly positively or negatively associated with any habitat (Table 2-4). Of species positively associated with forest 19 were trees, eight were treelets, and three were shrubs. Of four species positively associated with grass three were shrubs and one was a tree.

PAGE 30

18 The majority of positively forestand grassland-associated species in small plots were also positively associated with forest and grassland in large plots (Tables 2-3, 2-4). Symphonia globulifera was a common positively forest-associated species and Lantana trifolia was a common positively grassland-associated species in small plots but not common in large plots; in the latter case, the short stature of the species (< 2 m) meant it was undetected in large plots. Many common positively forest-associated species in large plots were not common in small plots, and Hoslundia opposita was a common positively grassland-associated species in large plots, but was not a common species in small plots. Furthermore, a number of positively gapand edge-associated species in small plots were positively forest-associated in large plots; however, no large plots of gap and edge were measured to compare since large plot size dimensions were best fit to continuous forest and grassland habitats. Interestingly, Celtis durandii, a tree, was negatively associated with forest in small plots but positively associated in large plots. Most species found by randomization tests to be positively associated with a habitat loaded close to the same habitat in DCA (graphs not shown). Discussion I examined the importance of habitat associations in influencing plant distributions using several lines of evidence. In my first line, I examined how habitats vary in terms of their environment and found the four habitats differed in a number of ways. Forest and gap were most similar with only light distinguishing them. This difference may be critical since light has been shown to be one of the most influential factors determining forest tree growth (Burslem 1996, Agyeman et al. 1999) and researchers have differentiated among tropical trees depending on abilities to establish in high or low light (Swaine & Whitmore 1988). Forest and gap habitats differed in most environmental variables from edge and grassland habitats, while edge and grassland habitats differed from one another in many environmental factors affecting water availability. Edge appeared to

PAGE 31

19 have the lowest availability of water of all four habitats. Water availability is also an important environmental factor influencing forest composition (Swaine 1996, Bongers et al. 1999, Clark et al. 1999). In Ghana, light and drought gradients were both important determinants of species growth (Agyeman et al. 1999). My results suggest that strong environmental differences exist among habitats, thus species could specialize on specific sets of environmental differences. In my second line of evidence, I examined habitat-level vegetative community differences corresponding to potential environmental differences I had identified. The grassland tree and shrub community was very different from the other three habitats’ communities. Thus this coarse-grain approach clarifies the distinction between the forested and grassland communities. These results were supported by data from large and small plots that showed extreme separation between forest and grassland plots, although less separation occurred between grassland and edge plots. Studies by Williams-Linera et al. (1998) in Mexico and Oosterhoorn and Kappelle (2000) in Costa Rica, also found that forest edge and interior communities varied greatly in their species composition. I also found differences in species composition between edge plots and the other two forested plots (gap and forest). But, I did not find a complete species turnover suggesting that overall species composition among the three types of forest plots is similar. One of the most interesting findings of the study was that forest and gap plots are almost completely overlapping in species DCA ordination plots, suggesting that their community composition does not differ. Several other studies found that forest and gap species composition is similar suggesting perhaps that gaps are mainly comprised of advanced regeneration of species typically found also in the forest (Uhl et al. 1988, Raich & Christensen 1989, de Carvalho et al. 2000, Webb & Peart 2000, Schnitzer & Carson 2001). I expected edge habitat to share equal affinity with both forest and

PAGE 32

20 grassland habitats. Instead, edge community had much greater affinity to forest and gap communities. These results may be due to greater woody stem density occurring in the forest part of edge plots versus grassland part of edge plots. In my third fine-grain line of evidence, I explored individual species associations with particular habitats. I found many common species in small plots had associations with one or two habitats. Almost all common species in large plots also preferred a single habitat. Furthermore, many species in small plots and most in large plots avoided at least one habitat suggesting the habitat is unsuitable for establishment and/or growth of that species. These results are uncharacteristic as researchers at other tropical sites, in both Paleo and Neotropics at a variety of spatial scales but all within forested habitats, found most species to be generalists (Hubbell & Foster 1986, Swaine 1996, Pitman et al. 1999, Webb & Peart 2000, Harms et al. 2001, but see Svenning 1999). It is curious that more species are associated with habitats in my study than have been previously found. But, these studies differ in their scale of analysis. As stated previously, many studies make comparisons of different tree strategies in surveys of mature forests, a single habitat, that vary in some characteristic, such as, topography and excluding gaps (Duivenvoorden 1995, Swaine 1996, Clark et al. 1999, Pitman et al. 1999, Webb & Peart 2000), while a few, including the current, focus on varying habitat types (Terborgh et al. 1996, Williams-Linera et al. 1998, de Carvalho et al. 2000). Furthermore, a number of studies define habitats based on differences in soil type (Duivenvoorden 1995, Swaine 1996, Terborgh et al. 1996, Clark et al. 1999), topography (Clark et al. 1999, Svenning 1999, Webb & Peart 2000, Harms et al. 2001, Pyke et al. 2002), or rainfall (Swaine 1996), while others define habitat based on gross vegetative characteristics (Terborgh et al. 1996, Williams-Linera et al. 1998, de Carvalho et al. 2000), such as, the present study (i.e., vegetation dominated by trees and presence of a

PAGE 33

21 treefall, etc.). Even still it is surprising that the similar forest and gap communities still had many species with strong associations. A seemingly surprising finding from this study is complete species overlap between forest and gap and high species overlap between edge and forest and gap at the community scale, yet individual species associations with particular habitats for many common species. Such differences can be attributed to differences between analyses. The community analysis represents all interactions among species both common and rare; it is the sum of many individual species parts, including both their presence and abundance. The individual species analysis, on the other hand, is a representation of presence and abundance of a single species with higher and lower densities indicating association with or avoidance of that habitat. Thus, when present, that species is contributing to the overall community, but may be found there less than expected in individual-species habitat associations. Results from the second and third lines of evidence using communityand species-level analyses allow me to examine the importance of habitat associations in influencing species distributions. Strong associations of individual species with particular habitats suggest that habitat specialization may be occurring in this community leading to both high diversity found between the forested habitats and grassland and high diversity found among the forested habitats. As strong environmental differences were found among the habitats with forest being most light-limited and edge most water-limited. These environmental differences are strong candidates for determinants of species distributions. Evidence for specializations can be better supported by conducting field and/or greenhouse studies of species abilities to grow and persist in these different habitats or environmental conditions. Furthermore, the strong similarities I found, both vegetative and environmental, between forest and gap habitats suggest that strong

PAGE 34

22 distinctions between these two communities are unmerited. These communities appear to be comprised of the same suite of shade-tolerant species with individual species perhaps differing in light tolerance and thus specializing to one of the two habitats (Webb & Peart 2000) but still able to grow in the other habitat.

PAGE 35

23 Table 2-1. Eigenvalues, cumulative % variance, and factor loadings for environmental variables along axes with strong eigenvalues using PCA for 10 transects each in all habitats (forest, gap, edge, and grassland) and in just forested habitats (forest, gap, and edge) in Kibale National Park, Uganda. Numbers in bold denote variables having strong correlations (>0.50) with a given axis. All habitats Forested habitats Axis 1 Axis 2 Axis 3 Axis 1 Axis 2 Axis 3 Eigenvalue 0.50 0.13 0.11 0.44 0.17 0.11 Cumulative % variance 49.4 63.1 74.1 44.1 60.7 71.4 Mean % water content 0.96 0.00 -0.11 0.89 -0.06 0.01 Maximum % water content 0.95 0.06 -0.13 0.88 0.20 0.03 % Organic matter 0.88 0.03 0.05 0.84 0.00 0.22 Bulk density (g/cm 3 ) -0.84 0.07 -0.10 -0.67 0.23 0.27 Minimum % water content 0.83 -0.06 -0.03 0.71 -0.31 -0.48 Field moisture capacity (%) 0.70 0.35 -0.27 0.44 0.70 0.34 PAR (mol/m 2 s) 0.63 0.01 0.21 0.54 -0.10 0.32 Root length density ( > 2 mm;cm/cm -0.27 3 ) 0.81 0.02 -0.06 0.78 0.00 Root length density (< 2 mm;cm/cm 0.09 3 ) 0.62 0.59 0.65 0.41 -0.46 pH 0.17 -0.36 0.77 0.53 -0.44 0.53

PAGE 36

24 Table 2-2. Means ( + 1 SD) for stem density and species richness using rarefaction in each habitat by growth form (shrub, treelet, tree) for 24 small plots (5 x 5 m) in each of four habitats (forest, gap, grassland, and edge). Differences found among means for characteristics analyzed using analysis of variance for stem density were further analyzed using Tukey Multiple Comparison procedure. Growth forms with different superscripts have different means at P < 0.05. Stem # Forest Gap Edge Grass Stems/plot Shrub 8.4 a (7.1) 6.9 a (3.9) 15.2 a (19.4) 28.5 b (26.6) Treelet 12.9 a (7.6) 11.8 a (7.8) 55.0 b (41.2) 2.6 a (3.9) Tree 139.8 a (229.2) 102.3 a (77.8) 82.1 a,b (48.7) 7.9 b (7.5) Species/stem Shrub 163 7.0 a (0.1) 7.0 a (0.1) 11.1 b (0.7) 11.1 b (1.0) Treelet 62 13.9 a (1.4) 14.9 a (1.5) 8.8 b (1.6) 6.0 c (0.0) Tree 190 21.2 a,b (2.3) 25.2 a (2.4) 19.9 b (1.7) 13.0 c (0.0)

PAGE 37

25 Table 2-3. Species, growth forms (shrub, treelet, and tree), and results from randomization tests with 10,000 iterations for species abundances in 24 small plots (5 x 5 m) each in four habitats (forest, gap, grassland, and edge). (-) = associated significantly less than expected, (+) = associated significantly more than expected. Species are listed alphabetically within their habitat of positive association. Family Species Form + Moraceae Ficus asperifolia shrub grass forest Apocynaceae Funtumia africana tree forest Guttiferae Symphonia globulifera tree gap, edge, grass forest Rutaceae Teclea nobilis tree grass forest, gap Annonaceae Uvariopsis congensis tree grass forest, gap Sapindaceae Lepisanthes senegalensis tree grass gap Sapindaceae Blighia spp. tree grass gap Celtidaceae Celtis durandii tree forest gap Rubiaceae Coffea eugenioides shrub grass gap Flacourtiaceae Dasylepis sp. treelet edge, grass gap Sapotaceae Mimusops bagshawei tree edge, grass gap Annonaceae Monodora myristica tree edge, grass gap Sapindaceae Pancovia turbinata tree edge, grass gap Apocynaceae Tabernaemontana spp. tree grass gap Sapindaceae Allophylus dummeri tree grass edge Celtidaceae Celtis africana tree grass edge Celtidaceae Chaetacme aristata treelet grass edge Rutaceae Clausena anisata treelet forest, gap, grass edge Rutaceae Fagaropsis angolensis tree grass edge Thymelaeaceae Peddiea fischeri treelet forest, grass edge Rubiaceae Psychotria sp. treelet grass edge Loganiaceae Strychnos mitis tree grass edge Rubiaceae Tarenna pavettoides treelet grass edge Fabaceae Albizia grandibracteata tree forest, gap edge, grass Acanthaceae Acanthus arborescens shrub forest, gap grass Verbenaceae Lantana trifolia shrub forest, gap, edge grass Compositae Vernonia spp. shrub forest, gap, edge grass Ebenaceae Diospyros abyssinica tree grass Salicaceae Dovyalis macrocalyx shrub grass Euphorbiaceae Acalypha spp. shrub

PAGE 38

26 Table 2-4. Species, growth forms (shrub, treelet, and tree), and results from randomization tests with 10,000 iterations for species abundances in 16 large plots (10 x 50 m) each in two habitats (forest and grassland). (-) = associated significantly less than expected, (+) = associated significantly more than expected. Species are listed alphabetically within their habitat of positive association. Family Species Form + Moraceae Antiaris toxicaria tree grass forest Sapindaceae Lepisanthes senegalensis tree grass forest Sapindaceae Blighia spp. tree grass forest Celtidaceae Celtis durandii tree grass forest Celtidaceae Chaetacme aristata treelet grass forest Rubiaceae Coffea eugenioides shrub grass forest Achariaceae Dasylepis sp. treelet grass forest Ebenaceae Diospyros abyssinica tree grass forest Salicaceae Dovyalis macrocalyx shrub grass forest Rubiaceae Euclinia longiflora treelet grass forest Moraceae Ficus asperifolia shrub grass forest Apocynaceae Funtumia africana tree grass forest Malvaceae Leptonychia mildbraedii tree grass forest Bignoniaceae Markhamia platycalyx tree grass forest Sapotaceae Mimusops bagshawei tree grass forest Fabaceae Newtonia buchananii tree grass forest Rubiaceae Oxyanthus speciosus treelet grass forest Sapindaceae Pancovia turbinata tree grass forest Apocynaceae Pleiocarpa pycnantha tree grass forest Rubiaceae Psychotria sp. treelet grass forest Rubiaceae Rothmannia sp. treelet grass forest Salicaceae Scolopia rhamniphylla treelet grass forest Olacaceae Strombosia scheffleri tree grass forest Loganiaceae Strychnos mitis tree grass forest Rubiaceae Tarenna pavettoides treelet grass forest Rutaceae Teclea nobilis tree grass forest Meliaceae Trichilia spp. tree grass forest Moraceae Trilepisium madagascariense tree grass forest Meliaceae Turraeanthus africanus tree grass forest Annonaceae Uvariopsis congensis tree grass forest Acanthaceae Acanthus arborescens shrub forest grass Fabaceae Albizia grandibracteata tree forest grass Labiatae Hoslundia opposita shrub forest grass Compositae Vernonia spp. shrub forest grass Violaceae Rinorea brachypetala treelet

PAGE 39

Figure 2-1. Means ( + 1 SD) for environmental variables along 10 transects in four habitats (forest, gap, grassland, edge) in Kibale National Park, Uganda. Differences found among means for characteristics analyzed using analysis of variance were further analyzed using Tukey Multiple Comparison procedure. Bars with different letters are statistically different at P < 0.05. A) % Organic matter (F = 20.1, P = < 0.001). B) % Sand (F = 41.3, P = < 0.001). C) % Clay (F = 4.3, P = 0.010). D) Field moisture capacity (% water content; F = 5.7, P = 0.003). E) pH (F = 1.1, P = 0.353). F) Bulk density, g cm -3 (F = 26.6, P = < 0.001). G) Mean water content (% dry mass; F = 38.2, P = < 0.001). H) Minimum water content (% dry mass; F = 9.7, P = < 0.001). I) Maximum water content (% dry mass; F = 39.6, P = < 0.001). J) Root length density (Lv) for small roots, cm cm -3 (diameter < 2 mm; F = 4.5, P = 0.009). K) Root length density (Lv) for large roots, cm cm -3 (diameter > 2 mm; F = 1.7, P = 0.177). L) PAR, mol m -2 s -1 (F = 7.0, P = 0.001).

PAGE 40

28 B01020304050forestgapedgegrass% Sandaabb C0102030405060forestgapedgegrass% Claybaaa D020406080forestgapedgegrassField Capacity (%)baaa F0.00.40.81.21.6forestgapedgegrassBulk Densityabcc G010203040506070forestgapedgegrassAvg % Water Contentbaaa H010203040forestgapedgegrassMin % Water Contentaabc I020406080100forestgapedgegrassMax % Water Contentaabc A010203040forestgapedgegrass% Organic Matte r acba E4.85.25.666.46.8forestgapedgegrasspHaaaa L0102030405060forestgapedgegrassPARbbba J00.40.81.21.6forestgapedgegrassLv small rootsaaba K0.000.010.020.030.040.050.06forestgapedgegrassLv large rootsaaaa

PAGE 41

Figure 2-2. Ordination along Axes 1 and 2 using principal components analyses in 10 transects each in four habitats (forest, gap, grassland, and edge) in Kibale National Park, Uganda. Environmental variable abbreviations are as follows: Lv (> 2 mm) = root length density for roots > 2 mm diameter, Lv (< 2 mm) = root length density for roots < 2 mm diameter, % OM = % organic matter, WCmax = maximum % water content, WCavg = mean % water content, WCmin = minimum % water content. A) Environmental variables. B) Habitat transects with circles represent edge, triangles represent forest, x’s represent gap, and squares represent grass.

PAGE 42

30

PAGE 43

Figure 2-3. Ordination along Axes 1 and 2 using principal components analyses in 10 transects each in three forested habitats (forest, gap, and edge) in Kibale National Park, Uganda. Environmental variable abbreviations are as follows: Lv (> 2 mm) = root length density for roots > 2 mm diameter, Lv (< 2 mm) = root length density for roots < 2 mm diameter, % OM = % organic matter, WCmax = maximum % water content, WCavg = mean % water content, WCmin = minimum % water content. A) Environmental variables. B) Habitat transects with circles represent edge, triangles represent forest, and x’s represent gap.

PAGE 44

32

PAGE 45

33 A0246810120306090120150Number of StemsNumber of Species forest gap grass edge D02468101214060120180240300360Number of StemsNumber of Species B0246810121416020406Number of StemsNumber of Species 0 E024681012141602040Number of StemsNumber of Species C05101520253004080120160Number of StemsNumber of Species F051015202530020406080100Number of StemsNumber of Species Figure 2-4. Number of species versus number of stems using rarefaction for different growth forms (shrub, treelet, and tree) in all size classes in small (5 x 5 m) plots in four habitats (forest, gap, grassland, and edge) and for stems > 2 m in large (10 x 50 m) plots in forest and grassland habitats in Kibale National Park, Uganda. A) Shrubs in small plots. B) Treelets in small plots. C) Trees in small plots. D) Shrubs in big plots. E) Treelets in big plots. F) Trees in big plots.

PAGE 46

34 Figure 2-5. Plot ordination based on species along Axes 1 and 2 using detrended correspondence analyses in 24 small plots (5 x 5 m) each in four habitats (forest, gap, grassland, and edge) in Kibale National Park, Uganda (Axis 1: Eigenvalue = 0.68, Gradient Length = 4.7, Cumulative % Variance = 15.0%) with circles representing edge, triangles representing forest, x’s representing gap, and squares representing grass. Axes are in standard deviation units (see Methods).

PAGE 47

35 Figure 2-6. Plot ordination based on species along Axes 1 and 2 using detrended correspondence analyses in 16 large plots (10 x 50 m) each in two habitats (forest and grassland) in Kibale National Park, Uganda (Axis 1: Eigenvalue = 0.95, Gradient Length = 9.1, Cumulative % Variance = 35.5%) with triangles representing forest and squares representing grass. Axes are in standard units (see Methods).

PAGE 48

36 Figure 2-7. Plot ordination based on species along Axes 1 and 2 using principal components analyses in 24 small plots (5 x 5 m) each in three habitats (forest, gap, and edge) in Kibale National Park, Uganda (Axis 1: Eigenvalue = 0.10, Cumulative % Variance = 10.0%) with circles representing edge, triangles representing forest, and x’s representing gap.

PAGE 49

CHAPTER 3 DO LIFE-HISTORY CHARACTERS OF JUVENILE TREES AND SHRUBS REFLECT ADAPTATIONS TO CONTRASTING HABITATS? Introduction Seeds and seedlings represent important stages in plant establishment as this is when the highest risk of mortality typically occurs (Hammond & Brown 1995, Kitajima 1996a). Thus, characters that increase the ability to establish and survive should be strongly selected for (Hammond & Brown 1995, Leishman et al. 2000) and influential in determining an area’s species composition. Indeed, seeds and seedlings exhibit interspecific variation in life-history characters (e.g., seed size, cotyledon morphology), and for species found in a particular habitat, it is thought that they often use similar solutions to selective pressures imposed by the habitat (Osunkoya 1996). Seed size is one life-history character that has received great scrutiny, because it is a bridge between parental reproduction, dispersal biology, and offspring establishment (Leishman et al. 1995, 2000). Small seed size increases probability of dispersal and increases a parent’s potential progeny number (Foster 1986), whereas large seed size has been proposed as an adaptation for growing in limiting conditions (e.g., low light; Kitajima 1996b, Osunkoya 1996, Hewitt 1998, but see Putz & Appanah 1987, Grubb 1998, Leishman et al. 2000). Because seedling growth is typically strongly light-limited in forest habitats (Burslem 1996, Agyeman et al. 1999), associations between high resource stores and low light seem logical. In an attempt to understand the functional significance of seed and seedling characters, studies have largely focused on seed size and ignored other seed and seedling characters, such as cotyledon type. Just as seeds represent available 37

PAGE 50

38 resources for future seedlings, cotyledons represent the form (photosynthetic or storage) and amount of resources currently available to seedlings and should be under strong selective pressure. Photosynthetic cotyledons are generally thought to be an adaptation for high light and species with this cotyledon type are more frequently dispersed by wind (Hladik & Miquel 1990, Garwood 1996, Hladik & Mitja 1996). Storage cotyledons may be an adaptation to growing in the low light of closed canopy forest, and these species are more frequently dispersed autochorously or by animals (Hladik & Miquel 1990, Ibarra-Manrquez et al. 2001). But, the frequency of different cotyledon types and their associations with habitats differ from region to region with storage cotyledons typically making up a greater percentage of the species in the Paleotropics than the Neotropics (Garwood 1996, Ibarra-Manrquez et al. 2001). Furthermore, species in a given habitat typically exhibit more than one cotyledon type (e.g., foliaceous and storage cotyledons in closed canopy forest; Garwood 1996). The associations between characters and environments are likely to be difficult to ascertain due to phylogenetic relationships, with closely related species resembling one another due to shared history and not recent environmental pressures. For example, Hewitt (1998) found that trees adapted to growing in shady environments have larger seeds than those adapted to sunny environments for angiosperms but not gymnosperms. This study found seed size was most strongly correlated with shade tolerance at the generic level, suggesting that differentiation in shade tolerance has been recent or that shade tolerance is labile. Cotyledon type appears to be a more conserved character phylogenetically than seed size with only one or two cotyledon types typically found in a family (Garwood 1996, Kitajima & Fenner 2000). Furthermore, the presence of particular characters is likely not just a result of phylogenetic relationships and selective pressures on that character, but also selective pressures on other correlated

PAGE 51

39 characters. Characters are likely acting in concert, such that it is a suite of characters conferring adaptations to particular habitats (Osunkoya 1996). In this study, I was especially interested in seed size and cotyledon type as they represent the type and amount of resources available to the germinating seeds and developing seedlings. I examined seed size and cotyledon type and their relationships with other seed and seedling characters to determine if suites of juvenile characters exist and if these suites are associated with particular habitats. I was also interested in the influence of phylogeny in determining characters occurring in species. Finally, as mentioned above, several studies have compared cotyledon type distributions across the tropics (Hladik & Miquel 1990, Garwood 1996, Ibarra-Manrquez et al. 2001), and their results suggest that storage cotyledons held above the ground comprise a larger proportion of the species in the Paleotropics than the Neotropics. I was interested in determining if an East African site (previously only West African sites were included) would have a distribution of cotyledon types similar to those of other Paleotropical sites. I quantify associations between cotyledon type and seed size and other seed and seedling characters (dispersal type, % of seed comprised of seed reserve mass i.e. embryo plus endosperm, days to germination, % germination, cotyledon thickness, initial seedling mass just after germination) and maximum adult height, growth form, and habitat association (closed canopy forest, forest treefall gap, forest/grassland edge, and grassland) in tree and shrub species in Kibale National Park, Uganda. Based on findings from other studies, I predicted that species with large seeds, large seedlings, animal dispersal, and thick storage cotyledons would be most common among large trees growing in closed canopy forest. Conversely, I expected that species with small seeds, small initial seedlings, wind and autochorous dispersal, and thin foliaceous cotyledons would typify early successional shrubs and small trees growing in high-light conditions found in gaps, forest/grassland edges, and grasslands.

PAGE 52

40 Materials and Methods Study Site Kibale National Park (766 km 2 ; 0’-0’N and 30’-30’E) is located in western Uganda, 24 km east of the Rwenzori Mountains at an elevation of approximately 1500 m. Between 1998 and 2000, a mean of 1760 mm of rain fell per year, annual mean daily maximum temperature was 23.1C, and minimum temperature was 15.1C (C.A. Chapman & L.J. Chapman, unpublished data). Kibale consists of mature forest (57%), colonizing forest (19%), grassland (15%), woodland (4%), swamp (4%), and plantations of exotic trees (1.0%; Chapman & Lambert 2000). Near the field station at Kanyawara, where this study was conducted, the forest is mid-altitude, moist evergreen (Howard 1991). I considered species character associations with four habitat types (closed canopy forest, treefall gap, grassland, grass/forest edge). The forest has a canopy height of 25 to 30 m (Howard 1991), and natural disturbances leading to treefall gaps are common (Skorupa & Kasenene 1984). Grasslands are thought to be anthropogenic in origin, but have been abandoned for at least 50 to 100 years (Kingston 1967). Forest has reestablished in some grasslands while the majority are still dominated by grasses, which are maintained by fire, elephant damage, and/or competitive dominance of grasses over trees (Lang Brown & Harrop 1962, Kingston 1967, Wing & Buss 1970). Queen Elizabeth National Park, contiguous with Kibale, is dominated by grasslands; many species of shrubs and trees, especially in the Fabaceae, are found in grasslands in both parks (Lock 1993, Lenzi-Grillini et al. 1996). Notably missing from Kibale, however, are species in the genus Euphorbia. Methods From May 1999 to May 2000, fruits and seeds were collected from any fruiting shrub or tree species that provided a large enough number of viable seeds so that

PAGE 53

41 seedling characters could be measured. An emphasis was placed on including species that differed in various characters, such as seed size, cotyledon type, and habitat association (habitat in which it is typically found). A sub-sample of seeds (2 to 30) were separated from fruit pulp and any dispersal appendages (e.g., wings or hairs as per Westoby 1998), sun dried, and transported to the University of Florida. A mean of 24.6 seeds per species was oven dried at 60C to a constant dry mass to determine the following characters: seed size (total seed mass (TSM) and seed reserve mass, i.e., embryo plus endosperm; SRM), and % of total seed mass comprised of seed reserve mass (% SRM), and seed dispersal syndrome (wind, small-animal, large-animal, autochorous; from Zanne 1998). Animal dispersal was determined based on seed area from its two longest axes (length x width), presence of a fleshy fruit or aril, and feeding records of largeand small-bodied animals. Large-animal dispersed fruits were those eaten only by large-bodied dispersers capable of swallowing seeds > 1 cm 2 and small-animal dispersed fruits were eaten both by large-and small-bodied dispersers (Zanne 1998). Other seeds of each species were germinated in raised nursery beds (1 x 13 m) in Kibale. Three light treatments (as measured at 1200 H in June 2000, heavy shade: 27.3 mol m -2 s -1 with range = 6.8 to 13.1 mol m -2 s -1 Photosynthetically Active Radiation PAR; medium shade: 182.3 mol m -2 s -1 with range = 23.7 to 617.8 mol m -2 s -1 ; light shade: 366.8 mol m -2 s -1 with range = 183.0 to 740.4 mol m -2 s -1 ) were used, and seeds were matched to their typical habitat’s relative light conditions for germination. The number of days until germination began and % germination (measured when seeds were all germinated or obviously rotten) were recorded. During the dry season, seedbeds were watered every 2 to 3 days by saturating the surface soil (~ 2.1 L/m 2 ). Seed collection, storage, and germination guidelines follow Katende et al. (1995).

PAGE 54

42 Seedlings were described for cotyledon type as in Garwood (1996) with exposure first [cryptocotylar (C): covered by the seed coat or phanerocotylar (P): not covered by the seed coat], position second [epigeal (E): cotyledons raised above soil surface or hypogeal (H): cotyledons at or below soil surface], and function third [photosynthetic (F): green cotyledons < 0.9 mm or storage (R): white or green cotyledons > 0.9 mm thick; modified from Kitajima 1992a]. Only five combinations (PEF, PER, PHR, CHR, and CER) exist among dicotyledonous species (Garwood 1996). Seedlings were harvested for determination of initial seedling mass (ISM) when the first photosynthetic organ was fully expanded. Cotyledon thickness was measured at this time to calculate inverse of cotyledon thickness (InvCotTh), which is linearly correlated with photosynthetic capacity per cotyledon mass (Kitajima 1992a). To increase the sample size for cotyledon thickness, cotyledons for older seedlings were included if their mean did not differ from seedlings used for initial seedling mass. Seedlings were dried at 60C to constant mass and weighed. Species names, maximum adult height, growth form (shrubs, treelets, trees), and primary habitat (forest, gap, grass, edge) were determined from Eggling and Dale (1952), Polhill (1952), Hamilton (1991), Katende et al. (1995), Lwanga (1996), and Chapter 2. Shrubs are self-supporting woody plants branching near the ground and either having several stems from the base or a single stem but then are short in stature (< 2 m; H. Beentje, personal communication). Trees are woody plants with a single main stem and tall in stature (> 2m). Treelets include species that sometimes grow in shrub form and sometimes grow in tree form. Although some species are positively associated with more than one habitat (Chapter 2), each species was assigned to the single habitat category in which it was most frequent as supported in the following literature (Eggling & Dale 1952, Polhill 1952, Hamilton 1991, Katende et al. 1995).

PAGE 55

43 Analyses For several discrete variables, as sample size was small for some categories, categories were combined (dispersal type: non-animal = wind + autochorous; growth form: shrubs and treelets; habitat type: open = edge + grassland; cotyledon type: CHR + PHR = *HR with * representing both phanerocotylar and cryptocotylar). Mean seed size was 1.02 g ( + 4.06 g). Seed size was divided into size classes based on 10-fold differences in seed size (very small: < 0.01 g, small: 0.01 to 0.1 g; medium: 0.1 to 1 g; large: > 1 g); however, because only one species was < 0.001 g and 2 species were > 10 g, both very small and large categories have ranges of 10 2 g (Hughes et al. 1994). For inverse of cotyledon thickness, 0 was assigned to all cryptocotylar species following Kitajima (1992b). To increase normality of distributions, all percent data were arcsine square root transformed, and all mass and height data were log transformed. Distributions for days to germination and inverse cotyledon thickness were normally distributed. Seed reserve mass and total seed mass were highly correlated (r = 0.95, P <0.001), so only total seed mass is considered in analyses. Since I was unable to collect data on all variables for every species (Appendix A), sample size varies in tests. Categorical data (cotyledon type versus growth form, dispersal type, habitat association, and seed size) were compared using contingency table goodness-of-fit (GOF) tests using programs written in R 1.6.2 (R Development Core Team 2002). Only comparisons in which differences were significant in an overall comparison including all three cotyledon types using GOF tests were analyzed further with pair-wise GOF tests with appropriate Bonferroni corrections such that, experiment-wise, P = 0.05. As expected values were small for some cells, all GOF tests were run with P values computed by Monte Carlo randomizations with 10,000 iterations in which the observed test statistic is compared to test statistics generated from random sampling from the set of all contingency tables with given marginals (R Development Core Team 2002).

PAGE 56

44 Principal Components Analyses (PCA) were done with Canoco 4.5 (ter Braak & Smilauer 1999). I did not have data for all 11 variables for all 70 species, thus a compromise between species number and variable inclusion was made. An analysis containing eight variables (see below) and 48 species was the best compromise allowing the most species and variables. Results for analyses with more variables (all 11 variables) were similar to the analysis presented here. In PCA, characters with categorical variables were assigned as follows, dispersal type: small-animal = 1, large-animal = 2, non-animal = 3; growth form: shrubs and treelets = 1, trees = 2; habitat type: forest = 1, gap = 2, open = 3; cotyledon type: PEF = 1, PER = 2, *HR = 3. PCA were run without categorical variables and results were similar to analyses with categorical variables included. Species were also analyzed incorporating phylogenetic relationships. A phylogeny was constructed from the literature with familial relationships based upon Soltis et al. (2000) and intrafamilial relationships based upon numerous family-level studies (Apocynaceae: Potgieter & Albert 2001, Sennblad & Bremer 2002; Bignoniaceae: Bigazzi 1995, Spangler & Olmstead 1999; Celtidaceae: Sytsma et al. 2002; Euphorbiaceae: Webster 1994, K. Wurdack pers. comm.; Fabaceae: Hu et al. 2000, Kajita et al. 2001; Flacourtiaceae: Chase et al. 2002; Rubiaceae: Andreason & Bremer 1996, Andersson & Rova 1999, Lantz et al. 2002; Rutaceae: Chase et al. 1999, Scott et al. 2000, Sapindaceae: no information, Sapotaceae: Pennington 1991, Morton et al. 1996). It should be noted that all these families are circumscribed so that they are monophyletic. When insufficient data were available to define phylogenetic relationships, relationships were left unresolved. Continuous characters were analyzed with Independent Contrasts (Felsenstein 1985) using Compare 4.4 (Martins 2001). Ancestral states were reconstructed by taking the mean in the states of the descendants. Contrast values are generated by taking the difference between the sister taxa across the

PAGE 57

45 phylogeny with taxa being both species at the tree tips and ancestors at nodes. The resulting contrast values were analyzed in the same way as uncorrected values using Pearson product moment correlations and PCA. Discrete characters were examined for correlated change between cotyledon type and other characters (dispersal type, total seed mass, habitat type, growth form) using the Omnibus Test in Discrete 4.0 (Pagel 1994). The Omnibus Test examines whether a change in states for cotyledon type is related to changes in states for other characters across the phylogeny. As Discrete 4.0 only allows for binary states in a character, characters with more than two states were broken up into a series of binary characters. This method allowed one state to be contrasted with all other states. Significance was determined using Randomizations tests with 100 iterations. Tests were considered significant if their Likelihood Ratio between independent and dependent models of character evolution was greater than 95% of the iterations. Significant correlations were analyzed further to determine contingent change (e.g., whether the change of character 1’s state depends on the state of character 2). These tests were considered significant if their Likelihood Ratio was significant at P < 0.05 assuming a 2 distribution with 1 degree of freedom. As branch lengths were unknown, a punctuated model of evolution was used setting all branch lengths, except those differentiating polytomies, to 1. As neither program allows for polytomies, branch lengths for polytomies were set to 0.0000001. Results Life-History Character Variation Total seed mass ranged five orders of magnitude from 0.0003 g for Maesa lanceolata to 28.5 g for Balanites wilsoniana (Appendix A); the mean total seed mass was 1.2 g (SD = 4.1 g, n = 65 species). Within habitats seed size varied less (forest: 0.02 to 28.53 g, gap: 0.03 to 0.92 g, open: 0.0003 to 0.5334 g). Ranges were overlapping for all three categories, but forest was the only habitat with seeds >1 g (n =

PAGE 58

46 6) and open was the only habitat with seeds < 0.01 g (n = 6). The percentage of total seed mass made up of seed reserve (% SRM) varied from 5.6 to 100%, but averaged 62.3% (SD = 26.0%, n = 63). Seeds germinated on average in 44.7 days, but time to onset of germination ranged from 5.7 to 125.0 days (SD = 35.1 days, n = 53). The majority of species were fast germinators (60.4%), germinating in less than 6 weeks (as defined by Garwood 1996). Percent germination ranged from 0.8% to 100.0%, and on average 31.0% (SD = 29.9%, n = 36). Initial seedling mass ranged from 0.003 to 2.269 g and averaged 0.169 g (SD = 0.36 g, n = 44), and inverse of cotyledon thickness ranged from 0.2 to 7.7 mm -1 and averaged 3.5 mm -1 (SD = 1.9 mm, n = 54) excluding CHR cotyledons as these were fixed at 0 (see methods). Only four of five cotyledon types reported in the literature (Garwood 1996) were in this sample of species from Kibale (Appendix A). PEF was the most common (74.3%), followed by CHR (15.7%), PER (8.6%), and PHR (1.4%). Only one species, Erythrina abyssinica, had PHR cotyledons. Most of the sample of 70 species were trees (61.4%), but the sample also included treelets (34.3%) and three shrubs (4.3%; Appendix A). Maximum adult height ranged from 4 to 60 m and averaged 23 m (SD = 13.2 m, n = 70). Most species were animal dispersed (55.7% small-animal dispersed and 31.4% large-animal dispersed). Wind-dispersed species are not common in Kibale and in this sample only 7.1% of species were wind-dispersed; 5.7% of species were legumes with autochorous-dispersal. Most of the species were typically found in forest (35.7%) or edge (45.7%) habitats, but some species also were associated with gap (14.3%) and grassland (4.3%) habitats (Appendix A). Cotyledon Types and Other Categorical Variables Of the three cotyledon types (PEF: aboveground exposed photosynthetic cotyledons; PER: aboveground exposed storage cotyledons; *HR: belowground storage

PAGE 59

47 cotyledons), both shrub/treelets and trees had significantly more PEF cotyledons than PER and *HR cotyledons (Figure 3-1a). Large-animal dispersed species had similar abundances of the three cotyledon types. Both small-animal dispersed and non-animal dispersed species had more PEF cotyledons than PER and *HR cotyledons (Figure 3-1b); wind-dispersed species only had PEF cotyledons (n = 4). All three habitat types (forest, gap, open) had species dominated by PEF cotyledons (Figure 3-1c). *HR cotyledons were found in species from all three habitats, but PER cotyledons only occurred in species from forest and open habitats. Small and medium seeds were mostly comprised of PEF cotyledons, and very small seeds only had PEF cotyledons. *HR cotyledons were also abundant in medium sized seeds (Figure 3-1d). No significant differences in abundances were found among cotyledon types for large seeds. PEF cotyledons were found in all seed sizes, but PER cotyledons were only found in smallto medium-sized seeds, and *HR cotyledons were only found in mediumto large-sized seeds. Mean Character Comparisons among Discrete Categories Significant differences in total seed mass, initial seedling mass, and cotyledon thickness occurred among cotyledon types (Table 3-1; see also Appendix B for more detailed results of analyses). *HR cotyledons had the largest seed and initial seedling masses (Table 3-1). Differences in adult height, days to germination, total seed mass, initial seedling mass, and cotyledon thickness were found among dispersal categories (Table 3-1). Large-animal and non-animal dispersed species had taller adults than small-animal dispersed species. Small-animal dispersed species had thinner cotyledons than non-animal dispersed species. Large-animal dispersed species took more days to germinate than non-animal dispersed species, and had larger seed and initial seedling masses (with and without cotyledons) than small-animal dispersed and non-animal dispersed species. Differences in adult height, total seed mass, and cotyledon thickness

PAGE 60

48 were found between growth form categories with trees having taller adults, heavier seeds, and thicker cotyledon than shrub/treelets. Differences in adult height, days to germination, total seed mass, and initial seedling mass were found among habitat types. Forest species had taller adults than gap species, and forest and gap species germinated more slowly, and had greater seed and initial seed masses (with and without cotyledons) than open species. Seed Mass and Continuous Variables Total seed mass was significantly positively associated with seed reserve mass, days to germination, initial seedling mass (with and without cotyledons), and maximum adult height and negatively associated with cotyledon thickness (Table 3-2, Figure 3-2a). Total seed mass was not correlated with % germination and % SRM. Multivariate Analyses In a PCA of seed, seedling, and adult characters, most variation (81.5%) was explained by the first four axes, which all had eigenvalues > 0.1; only these four axes are considered (Table 3-3). Most factors were strongly related to Axis 1 with six (adult height, days to germination, dispersal type, habitat type, cotyledon thickness, total seed mass) of eight characters having factor loadings greater than 0.5 (Table 3-3, Figure 3-3). This axis expresses relationships among important seed and seedling characters and other associated characters. (When the analysis is run with initial seedling mass included and the 39 species for which data were available, initial seedling mass is also strongly positively related to Axis 1.) Days to germination and growth form were strongly related to Axis 2. Only % SRM was strongly related to Axis 3, and only cotyledon thickness was strongly related to Axis 4. A biplot of species and characters along Axes 1 and 2 shows that large-animal dispersed species with thick cotyledons, many days to germination, and tall adult stature associated with forest and gaps load positively on Axis 1, while small-animal or

PAGE 61

49 non-animal dispersed species with thin cotyledons, few days to germination, and short adult stature associated with open habitats load negatively on Axis 1 (Figure 3-3). Species were continuously distributed along these axes suggesting that species range in the degree to which they are differentiated by these axes with no discrete differences. Species with faster germination and taller adult stature load positively on Axis 2. Axis 1 differentiates species based on the occurrence of a suite of characters. Six species were related positively and six species were related negatively to Axis 1 more strongly than any other axis (Appendix C). Eleven other species were also strongly positively and 12 other species were also strongly negatively related to Axis 1. Thus, 73% of species included in PCA express these suites (+ or – scores on Axis 1) of characters. Phylogenetic Effects Cotyledon type is a fairly conserved character across genera and families. All species in the same genera had the same cotyledon type. Of the 15 families with more than one species, only five families had more than one cotyledon type (Table 3-4). Meliaceae and Rutaceae, which each had two species measured, had PEF and CHR, and PEF and PER cotyledons, respectively, and Sapotaceae with three species had PEF and PER cotyledons. Sapindaceae, which had four species measured, had predominately CHR cotyledons, but one species had PER, and Fabaceae, which had five species measured, had PEF, PER, and PHR, but not CHR cotyledons. Both of the most speciose families (Euphorbiaceae and Rubiaceae) in the sample had only one cotyledon type. Total seed mass is less conserved than cotyledon type (Table 3-4). Of the 15 families with more than one species, ten families had more than one seed size type meaning seed size ranged more than an order of magnitude within a family. But, most families with more than one species measured varied only by two orders of magnitude. Three families had seed sizes ranging more than two orders of magnitude including

PAGE 62

50 Meliaceae, which had two species measured, had small and large seeds, Euphorbiaceae, which had four species measured, had all four seed sizes, and Fabaceae, which had five species measured, had very small, small, and medium seeds. Continuous characters (total seed mass, % germination, % seed reserve mass, days to germination, cotyledon thickness, initial seedling mass, seed reserve mass, maximum adult height) showed equivalent or stronger correlations with total seed mass when phylogenetic relationships were incorporated (Table 3-2), except for % SRM, which showed a slightly weaker correlation after correction. Cotyledon thickness was the character that showed the greatest increase in the strength of the correlation with seed size (Figures 3-2a, b). These results suggest that relationships between total seed mass and other seed and seedling characters are not dependent on phylogeny except perhaps for cotyledon thickness. Discrete characters showed correlated change between cotyledon type and dispersal type, habitat type, and total seed mass but not growth form, and between seed size and habitat type (Table 3-5). Results from these tests are similar to findings when phylogeny is not considered. Correlated change was found between characters for large dispersers, large seeds, and gap habitats with *HR cotyledons. In fact, results from the Contingent Change Tests suggest that the transition from photosynthetic to storage cotyledons is more likely when dispersers are large animals and seeds are large, and the transition from PE* to *HR cotyledons is more likely when seed size is large. Correlated changes directly between habitat type and total seed mass are less clear as no Contingent Change Tests were significant. Principal components analyses of continuous characters with phylogenetic relationships included also showed similar patterns as analyses without phylogenetic relationships (Table 3-6) with total seed mass and cotyledon thickness strongly negatively related to Axis 1 and adult height strongly positively related to Axis 2. Both

PAGE 63

51 % SRM and days to germination were each strongly negatively related to their own axes, As in the PCA using uncorrected data, when the analysis is run with initial seedling mass included and only 27 species for which data were available, initial seedling mass is also strongly positively related to Axis 1. Discussion In this exploration of relationships among ecological and morphological life-history characters of tree and shrub species to determine if these characters were associated with species presence in particular environments. Both seed size and cotyledon morphology, two characters that indicate energetic resources available to the germinating seed and developing seedling, showed strong associations with other characters and with particular habitats. Species with larger seeds and storage cotyledons were found associated with the more forested habitats (closed canopy forest and treefall gaps) suggesting that seedling success in low-light environments is greater when seedlings have access to stored reserves. Surprisingly though, closed canopy forest, where light intensities were lowest (Chapter 2) was dominated by species with photosynthetic cotyledons. Such results imply that the importance of storage cotyledons in low-light environments may not be as important as was predicted. Further implications of my results are explored below. Cotyledon Type A comparison of these results for cotyledon type occurrence and distribution with other tropical sites is intriguing. Other studies from West Africa found a lower percentage of PEF (aboveground photosynthetic) cotyledons in their floras (Hladik & Miquel 1990, Hladik & Mitja 1996 in Gabon; Okali & Onyeachusim 1991 in Nigeria) than in Neotropical studies (Garwood 1996, Ibarra-Manrquez et al. 2001), but in this study PEF cotyledons dominated the 70 species examined. In fact, only one species had PHR (belowground exposed storage) cotyledons and no CER (aboveground hidden storage) cotyledons

PAGE 64

52 occurred among these species. The results for PHR and CER cotyledons may not be surprising as these two cotyledon types are usually poorly represented in other floras (Garwood 1996). Unfortunately, no other data on cotyledon type from eastern Africa are available for comparison. These trends suggest distribution of cotyledon types in Kibale to be different from other forests in Africa. While the reason for such differences is unclear (see discussion below), it does suggest that some of the large-scale patterns suggested by previous studies may prove incorrect as more sites are examined. When cotyledon type is associated with seed size, very small seeds have only PEF cotyledons, while larger seeds have all cotyledon types. The incidence of *HR (belowground storage) cotyledons increases with seed size. The occurrence of PEF cotyledons in small seeds is likely due to space constraints within seeds, such that large storage cotyledons do not fit in small seeds. Photosynthetic cotyledons that expand after germination then should be a typical solution for small seeds. It is also not surprising to find species with the largest seeds and greatest amount of stored reserves having *HR cotyledons. The findings for small seeds were similar to those found by Hladik and Miquel (1990) and Hladik and Mitja (1996) in Gabon. On the other hand, larger seeds in other studies also had species with PHR (belowground exposed storage) and CER (aboveground hidden storage) cotyledons; as noted above, these two cotyledon types are extremely rare or absent in the Kibale flora. Other studies found only weak or no differences between growth form and cotyledon type (Garwood 1996, Ibarra-Manrquez et al. 2001), thus my results comparing cotyledon type with growth form are not unexpected. Relationship between cotyledon type and dispersal type were similar to other studies (Hladik & Miquel 1990) except for non-animal dispersed species. Small animals typically disperse species with photosynthetic cotyledons but large animals disperse species in all cotyledon categories. These relationships are likely driven by relationships between cotyledon type and seed size as species with very small seeds all

PAGE 65

53 have PEF cotyledons. The results for the non-animal dispersed species can be better understood when this category is examined more closely, wind dispersed species all had PEF cotyledons and autochorous species had all three cotyledon types. In the review by Garwood (1996), she synthesizes results from tropical studies examining cotyledon function and makes predictions about relationships between cotyledon morphology and various other characters. She reports that studies found associations between seed size and cotyledon type, although a lot of variation in seed size exists within a given cotyledon type and associations between seedling size and cotyledon type. The latter relationship may be due to associations among seed size, cotyledon type, and regeneration strategy as pioneers typically have small seeds with photosynthetic cotyledons and small fast-growing seedlings, and shade-tolerant species typically have large seeds with storage cotyledons and large slow-growing seedlings. She reports studies have also found differences in germination rates between cotyledon types; however, these studies are not consistent in the relationships they find. Finally, she reports some evidence suggests that differences in cotyledon type occur between shrubs of shorter and trees of taller adult stature, but these relationships are not consistently found. I did find that seed size and initial seedling mass differed, but days to germination and adult height did not. The lack of relationship between adult height and cotyledon type is not that surprising as other studies have not consistently found a pattern; however, the lack of relationship between days to germination and cotyledon type is surprising. Some of the relationship between cotyledon type and germination rate may be driven by the relationship of those characters with regeneration strategy (shade tolerant versus shade intolerant), and in my study, I found different associations between cotyledon type and species associated with the low-light forest habitat than was found in other studies. Thus my results may be driven by differences in relationships between habitat association and cotyledon type in this study as compared to other studies.

PAGE 66

54 Distributions of cotyledon types among habitats were especially surprising. From other studies, I expected that species associated with forest would mainly have large storage cotyledons and that species associated with gap, edge, and grassland would mainly have PEF cotyledons. Instead, I found PEF cotyledons were the most common cotyledon type in forest and open species. No differences were found among frequencies of the three cotyledon types for gap species; however, it should be kept in mind that no PER (aboveground exposed storage) cotyledons occurred in gap species. Results for gap species though must be taken with some caution as only 10 species were examined. My results differ from those of Ibarra-Manrquez et al. (2001) who found that PEF seedlings dominated pioneer species and CHR cotyledons dominated persistent species. Hladik and Miquel (1990), though, note that PEF cotyledons occurred among large-seeded shade-tolerant species, and these seedlings develop very large photosynthetic cotyledons. In contrast to my study, Hladik and Miquel (1990) found that a large proportion of their forest species were PER. I found few PER species in the forest and very few overall. Some of these differences may be due to differences in delimiting PEF versus PER cotyledons. My study used a cut off point modified from Kitajima (1992a) of seedlings < 0.9 mm thick being PEF. This cut off led me to describe one species, Piptadeniastrum africanum, as having PEF cotyledons, and Hladik and Miquel (1990) and Hladik and Mitja (1996) described this species as having PER cotyledons. Differences in delimiting between PEF and PER cotyledons, though, can only account for a few species, so the dominance of PEF cotyledons in my study overall and in forest species is surprising. If , however, the risk of tissue loss (e.g., from herbivores) is especially high in closed canopy forest in Kibale, species may have been selected to translocate stored reserves directly into seedling stems or roots rather than leaving them exposed in cotyledons. Such a relationship is supported by findings that both seeds and initial seedlings (measured with and without cotyledons) are larger in

PAGE 67

55 both forest and gap habitats suggesting that in the forest, seed reserves are being moved into the seedling body. Seed Size In my study, seed size ranged five orders of magnitude, which is comparable to values reported from other floras (Leishman et al. 2000). Seed size was also related to a number of characters, many of which are well supported in the literature. As expected from Leishman et al. (2000), total seed mass and seed reserve mass were strongly correlated and percent total seed mass comprised of seed reserve was not correlated with either measure of seed mass. Differences in total seed mass then are mainly being driven by differences in embryo and endosperm (seed reserve mass) and not differences in seed coat. In my study, seed size also differed among growth forms, with trees having heavier seeds (Hammond & Brown 1995, Leishman et al. 1995, 2000). Dispersal distance has been suggested as potentially driving the relationship between seed size and tree height such that large seeds can be dispersed far only if they are dropped from a greater height; oddly this relationship is found for species in all dispersal categories but would be expected more in windor autochorously-dispersed species if driven by dispersal distance (Leishman et al. 2000) I found seed size was associated with dispersal types with large-animal dispersed species, not surprisingly, having the largest seeds of all disperser types. The largest seeds were all large-animal dispersed, but small-animal and non-animal dispersed were intermixed in sizes. As seed size is used, in part, to define largeversus small-animal dispersed species, the important comparison is the relationship between non-animal and the two animal-dispersed categories. The seed size relationship with dispersal type is similar to that reported by Hughes et al. (1994) in which seeds > 0.1 g were animal dispersed, < 0.0001 g had unassisted dispersal, and between 0.0001 and 0.1 g were dispersed by many different modes. My study did not include seeds < 0.0003 g.

PAGE 68

56 I found total seed mass was also correlated with days to germination, initial seedling mass, cotyledon thickness, and maximum adult height with larger seeds taking longer to germinate and having larger initial seedlings, thicker cotyledons, and taller adults. Other studies have also documented a positive relationship between seed size and seedling size (Stock et al. 1990, Leishman & Westoby 1994, Kitajima 1996b, Leishman et al. 2000) and seed size and adult height (Hammond & Brown 1995, Leishman et al. 1995, 2000). A relationship between seedling size and seed size is expected as larger seeds have more reserves to form a larger initial seedling. A relationship between seed size and adult height is also not surprising since growth form and seed size were also related and the definition of growth form is based, in part, on adult height; however, explanations for the relationship are unclear (see discussion above with adult height; Leishman et al. 2000). The relationship between growth form and seed size may also be due to a correlation with another character (e.g., habitat association). In relation to habitat types, forest and gap species in my study had larger seeds than species from other habitats. Seed size ranges overlapped across habitats suggesting a single seed size solution for a habitat does not occur, but forest habitats had the seven heaviest seeds (top 10.8%), and open habitats had the seven lightest seeds (bottom 10.8%). Interestingly, when seed size was compared among species of a given habitat type in my flora seed size range was smaller (one to three orders of magnitude) than across the entire flora. If other studies comparing across floras (Leishman et al. 1995, 2000) reported not only seed size values across floras with large geographical separation, but also species associations with specific habitats within their floras, then the high variation within floras and low variation between floras may be better understood.

PAGE 69

57 Suites of Characters In a series of univariate examinations of cotyledon type and seed size, associations among characters begin to emerge. Multivariate analyses support these univariate associations. I found larger seeds were associated with thicker cotyledons (i.e., hypogeal storage cotyledons), more days to germination, and taller stature adults, in large-animal dispersed species of forest and gap habitats, whereas smaller seeds were associated with thinner cotyledons (i.e., PEF cotyledons), fewer days to germination, and shorter stature adults, in non-animal dispersed species of open habitats. A majority of the species examined was strongly related to the Axis 1 and thus expressed the major suite of characters along the range of Axis 1. Interestingly, species were continuously distributed along the Axis 1 suggesting a single solution for each habitat did not occur. Juvenile Characters and Environmental Associations I previously measured a number of environmental variables (light and soil variables affecting water availability, i.e., soil bulk density, water content, root length density, field capacity, organic matter, particle size distribution, and pH) in Kibale and found forest and gap habitats to be similar environmentally only differing with forest having lower light than all other habitats (Chapter 2). Edge and grassland sites (open sites in the current study) were more similar to one another than to forest and gaps, but edge had higher root length density than any other habitat (and thus perhaps lower water availability), and grassland had the highest values for variables associated with water availability than all other habitats (Chapter 2). Thus forest is the most light-limited and edge is likely the most water-limited of the habitats. In the present study, I found forest and gap sites had larger seeds than open species. This result partially supports associations between low light and seed size. Furthermore the top 10.8% of seeds by mass were all found in forest habitat where light

PAGE 70

58 was lowest. Ranges of seed size within habitats was less and segregation of seed size by habitats was greater in my study than noted by Leishman et al. (2000), but my sample size within habitats (e.g., n = 10 for gap species) are smaller than those they considered. That large seeds should also be associated with gaps that have higher light is odd. But in the previous study (Chapter 2), light was highly variable from gap to gap perhaps due to differences in gap size. Large-seeded species may be more closely associated with small gaps where light availability is also low. Another possibility is that within the PEF category, cotyledons differ based on habitat association between allocations to cotyledon thickness and reserve storage in stem and roots in low light and cotyledon area in high light (Hladik & Miquel 1990). If this is the case, it should not be surprising to find PEF cotyledons in low-light conditions, and all species with PEF cotyledons may not be functionally equivalent. But, findings from Kitajima (1994) and Veneklaas and Poorter (1998) show that although seedlings of a given species, especially pioneers, are able to increase light capturing surfaces when growing in low light versus high light, seedlings of shade-adapted species actually allocate less to light capturing surfaces than seedlings of sun-adapted species. Low-light associated species with PEF cotyledons may also allocate more to cotyledon longevity than open-associated species. For instance, in a transplant experiment into the four habitats with seedlings from 24 species, many seedlings from five forest-associated species and one gap-associated species retained their cotyledons for at least a year (Zanne unpublished data). Only one seedling from one edge-associated species and no grassland-associated species retained their cotyledons for this long. Phylogeny As reported in the literature (Lord et al. 1995, Garwood 1996), cotyledon type was fairly conserved, with most families having only one type, whereas seed size varied in most families by at least two orders of magnitude. These results were supported by

PAGE 71

59 analyses incorporating phylogenetic relationships. Only the strength of the correlation between the cotyledon thickness and seed mass changed after phylogenetic relationships were considered; all other character correlations were unaffected by the correction. By definition, cotyledon thickness is directly related to cotyledon type as PEF cotyledons are > 1.11 mm -1 thick, PER cotyledons are 0 to 1.11 mm -1 , and *HR cotyledons are 0 mm -1 thick. With phylogeny incorporated, cotyledon type did show correlated change between *HR cotyledons and gap habitats, large seeds, and large-animal dispersers. Conclusions This study, which explores relationships among 70 tree and shrub species, is the first comparison of seed size and cotyledon type from eastern Africa and explores relationships of these characters across different habitats. A suite of characters including large seed size, large initial seedling size, thick hypogeal storage cotyledons, many days to germination, and tall adults in large-animal dispersed species were found associated with forest and gap habitats. These results suggest that these suites of characters are important for establishment and growth in different environments. The next step should be greenhouse and field experiments with seedlings ranging in seed size and cotyledon type to determine if such character combinations translate into differences in species abilities to establish and grow in different habitats. Interestingly, cotyledon distribution is very different than found in other studies. Further exploration of cotyledon type in different areas within the tropics will prove important in understanding the role of different cotyledon morphologies with different geographical and environmental regions.

PAGE 72

60 Table 3-1. Summary of the results for univariate ANOVAs and t tests comparing means of continuous characters (total seed mass, % total seed mass comprised of seed reserve, days to germination, % germination, initial seedling mass, inverse of cotyledon thickness, maximum adult height) among categories of discrete characters (cotyledon type, dispersal type, growth form, habitat type). Several categories had low sample size and were combined with other categories (wind + autochorous = non-animal; shrub + treelet; edge + grass = open). Only relationships where differences were found among means are shown and were analyzed further using Tukey’s b Multiple Comparison procedure. Categories with different superscripts have different means at P < 0.05. Categories with means denoted by a are smaller than means denoted by b. For more complete results, see Appendix B. Independent Dependent 1 2 3 Cotyledon type Total seed mass PEF a PER a *HR b Initial seedling mass with cotyledons PEF a PER a,b *HR b Dispersal type Adult height large animal b small animal a non animal b Days to germination large animal b small animal b non animal a Total seed mass large animal b small animal a non animal a Initial seedling mass with cotyledons large animal b small animal a non animal a Initial seedling mass without cotyledons large animal b small animal a non animal a Inverse cotyledon thickness large animal a small animal b non animal a Growth form Adult height shrub/treelet a tree b Total seed mass shrub/treelet a tree b Inverse cotyledon thickness shrub/treelet b tree a Habitat type Adult height forest b gap a open a,b Days to germination forest b gap b open a Total seed mass forest b gap b open a Initial seedling mass with cotyledons forest b gap b open a Initial seedling mass without cotyledons forest b gap b open a

PAGE 73

61 Table 3-2. Pearson product moment correlations between total seed mass and other continuous characters (seed reserve mass, % germination, % total seed mass comprised of seed reserve (% SRM), days to germination, initial seedling mass, inverse cotyledon thickness, maximum adult height) for data when phylogenetic relationships are not incorporated and incorporated using Independent Contrasts. Significant relationships are in bold. Total seed mass Phylogenetically not incorporated Phylogenetically incorporated R P R P Seed reserve mass 0.95 <0.001 0.95 <0.001 % Germination 0.09 0.602 0.22 0.207 % Seed reserve mass 0.08 0.534 0.07 0.600 Days to germination 0.39 0.005 0.40 0.004 Initial seedling mass with cotyledons 0.88 <0.001 0.89 <0.001 Initial seedling mass without cotyledons 0.84 <0.001 0.84 <0.001 Inverse cotyledon thickness with cryptocotylar cotyledons -0.59 <0.001 -0.70 <0.001 Inverse cotyledon thickness without cryptocotylar cotyledons -0.42 0.005 -0.55 <0.001 Maximum adult height 0.37 0.002 0.42 0.001

PAGE 74

62 Table 3-3. Eigenvalues and factor loadings for seed, seedling, and adult characters along the first four axes using principal components analyses for 47 species in Kibale National Park, Uganda. Numbers in bold denote the strongest factor loadings for a character with an axis and numbers in italic denote strong factor loadings (>0.50) for a character with other axes. Inverse cotyledon thickness includes species with cryptocotylar cotyledons. Axis 1 Axis 2 Axis 3 Axis 4 Eigenvalue 0.37 0.19 0.13 0.12 Cumulative % variance 36.8 56.2 69.2 81.5 Total seed mass 0.90 -0.04 -0.10 -0.23 Dispersal type -0.76 0.43 0.21 -0.03 Habitat type -0.68 0.19 -0.23 -0.42 Adult height 0.61 0.49 -0.04 0.39 Growth form 0.39 0.76 -0.04 0.21 Days to germination 0.52 -0.61 0.19 -0.03 % Seed mass comprised of seed reserve 0.15 0.20 0.92 -0.24 Inverse of cotyledon thickness -0.52 -0.34 0.20 0.71

PAGE 75

63 Table 3-4. Distribution of cotyledon type categories (PEF, PER, CHR, PHR) and seed size categories (total seed mass: very small = < 0.01 g, small = 0.1 to 0.01 g, medium = 1 to 0.1 g, large = > 1 g) among the 15 families with more than one species for 70 tree and shrub species from Kibale National Park, Uganda. # Cotyledon types/family # Seed sizes/family # Species/family 1 2 3 1 2 3 4 2 3 2 3 3 3 4 1 2 2 4 1 1 2 1 5 1 1 1 7 1 1

PAGE 76

64 Table 3-5. Significant correlated changes (Omnibus Test) incorporating phylogenetic relationships between discrete characters (Character 1: cotyledon type or total seed mass and Character 2: habitat type, dispersal type, growth form, or total seed mass) for 70 tree and shrub species in Kibale National Park, Uganda. The Omnibus Test examines whether changes in state for character 1 are related to changes in state for character 2 across the phylogeny (changes in state can either be from state 1 to state 2 or state 2 to state 1). Categories are as follows cotyledon type: F = photosynthetic versus R = storage or PE* = phanerocotylar epigeal versus *HR = hypogeal storage or PER = PER versus other (PEF and *HR); dispersal type: large-animal versus other (small-animal and non-animal) or small-animal versus other (large-animal and non-animal); habitat: forest and gap versus open or gap versus other (forest and open) or forest versus other (forest and gap); total seed mass: small (< 0.1 g) versus large (> 0.1 g) or very large (> 1 g) versus other (< 1 g). Significant results were further tested for contingent change (if the change of one character’s state is dependent on the state of the second character). For significance in Omnibus Tests, * = P < 0.05 and NS = P > 0.05. Significant contingent changes are denoted by letters with the direction of the change noted at the bottom of the table. Character 1 Cotyledon type Seed size Character 2 F vs. R PE* vs. *HR PER vs. other Small vs. large Very large vs. other Habitat type Forest/gap vs. open NS * NS * * Gap vs. other NS * * NS NS Forest vs. other NS NS NS NS * Dispersal type Large-animal vs. other *a * NS Small-animal vs. other * * NS Total seed mass Small vs. large * b * NS Very large vs. other NS *c NS Direction a F to R change is more likely if dispersers are large b F to R change is more likely if seeds are large c PE to *HR change is more likely if seeds are large

PAGE 77

65 Table 3-6. Eigenvalues and factor loadings for continuous seed, seedling, and adult characters with phylogenetic relationships considered along the first two axes using principal components analyses for 47 species in Kibale National Park, Uganda. Numbers in bold denote the strongest factor loading for a character with an axis and numbers in italic denote strong factor loadings (>0.50) for a character with other axes. Axis 1 Axis 2 Axis 3 Axis 4 Eigenvalue 0.44 0.20 0.19 0.13 Cumulative % variance 43.7 63.5 82.2 95.2 Total seed mass -0.89 0.02 0.16 0.23 Inverse of cotyledon thickness 0.86 0.16 -0.09 -0.35 Adult height -0.44 0.75 0.33 -0.36 % Seed mass comprised of seed reserve -0.33 0.34 -0.88 0.02 Days to germination -0.59 -0.54 -0.12 -0.59

PAGE 78

66 0.00.20.40.60.81.0PEFPER*HRFrequency large animal small animal non animalBacdabdabd 0.00.20.40.60.81.0PEFPER*HRFrequency forest gap openCbceacdacd 0.00.20.40.60.81.0PEFPER*HRFrequency very small small medium largeDbdfgacece,fgga 0.00.20.40.60.81.0PEFPER*HRFrequency shrub/treelet treeAbaadcc Figure 3-1. Frequency of cotyledon types (PEF, PER, and *HR) among seed, seedling, and adult characters for species in Kibale National Park, Uganda. Only comparisons in which differences among counts of all three cotyledon types using GOF tests were analyzed further with pair-wise GOF tests with appropriate Bonferroni corrections with being divided by the number of categories being compared such that, experiment-wise, P = 0.05. All GOF tests were run with P values computed by Monte Carlo randomizations with 10,000 iterations. Within a category, cotyledon types with different superscripts differ at P < 0.05. A) Growth form (shrub + treelet: n = 27, tree: n = 43). B) Dispersal type (small-bodied animal: n = 39, large-bodied animal: n = 22, non-animal: n = 9). C) Habitat type (forest: n = 25, gap: n = 10, open: n = 35). D) Seed size (for total seed mass, very small = < 0.01 g, n = 6; small = 0.01 to 0.1 g, n = 25; medium = 0.1 to 1 g, n = 28; large = > 1 g, n = 6).

PAGE 79

67 -4-20246-2-1012Total Seed Mass (g)BInverse Cotyledon Thickness (mm-1) 02468-3-2-1012Total Seed Mass (g)AInverse Cotyledon Thickness (mm-1) Figure 3-2. Relationship between total seed mass and inverse cotyledon thickness using Independent Contrasts for 54 species in Kibale National Park, Uganda. A). Incorporating phylogenetic relationships. B) Without incorporating phylogenetic relationships.

PAGE 80

68 Figure 3-3. Biplot of seed, seedling, and adult characters and species along Axes 1 and 2 using principal components analyses for 47 species in Kibale National Park, Uganda. Species abbreviations are the first two letters of the Genus and the first two letters of the species (see Appendix C for names). Character abbreviations are as follows Habitat = habitat type [forest = 1, gap = 2, open (edge + grassland) = 3], Dispersal = dispersal type [large-animal = 1, small-animal = 2, non-animal (wind + autochorous) = 3], InvCotTh = inverse cotyledon thickness, % SRM = percent of total seed mass comprised of seed reserve, GrFrm = growth form [shrub and treelets = 1, trees = 2], Height = maximum adult height, TSM = total seed mass, DaytoGerm = days to germination.

PAGE 81

CHAPTER 4 ADAPTATION AND VARIATION OF TREE AND SHRUB SEEDLINGS: SURVIVAL AND GROWTH ACROSS HABITATS Introduction Plant species are often considered to be adapted to grow and survive best in particular habitats (e.g., closed canopy forest or treefall gaps). But, species are known to be plastic in their growth and survival responses to different habitats (Veenendaal et al. 1996a, Kobe 1999, Poorter 1999, 2001, Montgomery & Chazdon 2002). For instance, seedlings may grow slowly in the shade under closed canopy, but when a treefall gap is created near that seedling, growth rates typically increase in response to the greater light (Whitmore 1989). How plastic different types of species are to varying conditions and the mechanisms they use to respond to growing in different habitats are poorly understood. From work measuring morphological and physiological characters in Panama on 13 species of seedlings and Australia on 12 species of seedlings, it has been suggested that tree seedlings from species adapted to low-light conditions are both less plastic and differ in resource allocation in changing light levels than seedlings from species adapted to high-light conditions (Kitajima 1994, Osunkoya et al. 1994). In this study, I was interested in the degree to which species with different habitat associations (habitat in which they were most typically found) vary in their ability to grow and persist both within and among habitats in Kibale National Park, Uganda. Species adapted to higher light are expected to grow faster than those adapted to shade (Veneklaas & Poorter 1998). Furthermore, several studies have documented a trade-off between high survivorship (typical of species adapted to growing in shaded habitats), and high growth rate and light-capturing ability (typical of species adapted to growing in 69

PAGE 82

70 high-light habitats; Kitajima 1994, 1996a, Westoby et al. 1997, Poorter 1999, but see Walters & Reich 2000, Sack & Grubb 2001). Species with slow growth and high survivorship are thought to allocate resources to defensive characters, such as, tough well-protected leaves (Kitajima 1994). Shade-intolerant species have also been shown to have greater plasticity in growth than shade-tolerant species in their responses to increasing light, such that shade-intolerant species had a greater increase in growth as light increased (Kitajima 1994, 2002, Osunkoya et al. 1994, Veenendaal et al. 1996b). Seedling growth and survival then is thought to be strongly influenced by the environment in which seedlings are growing. I was also interested in the variation in growth and survival when seedlings of the same species are planted in different habitats: closed canopy forest, treefall gap, grass/forest edge, grassland. These habitats differ in light levels with closed canopy forest having lower light than the other three habitats, but the other three habitats do not differ from one another (Chapter 2). In general, species have been found to grow and persist better as light increases (Kitajima 1994, Walters & Reich 2000). Although, in a previous study in Kibale, one species, Uvariopsis congensis, was found to grow best in forest understory and small canopy gaps rather than large canopy gaps (Chapman et al. 1999), and Poorter (1999) found that growth was best at intermediate light at approximately 25 to 50% of full sun. These habitats also differ in the amount of soil moisture with edge likely being the driest and grassland the wettest (Chapter 2). Biotic factors, such as herbivores, are also likely to differ among these habitats. A variety of characters, such as seed size and cotyledon type, are thought to be associated with a species’ habitat associations (Kitajima 1994, Leishman & Westoby 1994, Osunkoya et al. 1994). Characters that are indicators of the amount and type of energy supplies for the developing seedling should be particularly important because they assist seedlings in their initial establishment and growth. Seed size has been

PAGE 83

71 correlated with plant attributes that may confer tolerance to poor growing conditions including positively correlated with seedling survivorship (Leishman & Westoby 1994, Kitajima 1996b) and negatively correlated with relative growth rate (RGR; Kitajima 1994, 1996b, Osunkoya et al. 1994, Westoby et al. 1997, but see Stock et al. 1990 for comparisons within a family). Photosynthetic cotyledons are generally thought to be an adaptation for high light (Hladik & Miquel 1990, Garwood 1996, Hladik & Mitja 1996). The inverse of cotyledon thickness has be found to be negatively related with photosynthetic capacity per unit mass of cotyledons, meaning that as cotyledons became thicker photosynthetic rates decreased per unit mass (Kitajima 1996b). Large seeds and thick storage cotyledons then are predicted to be associated with species from low-light conditions, because the larger seeds and cotyledons provide energetic resources for the developing seedling where light is low. In a previous study, I found that large seed size and thick storage cotyledons were indeed associated with species from the more forested of the four habitats (closed canopy forest and treefall gap, Chapter 3). But, surprisingly, thick storage cotyledons comprised a greater component of species from gap habitats and photosynthetic cotyledons comprised a greater component of species from forest habitats (Chapter 3). Since energy resources are typically limiting for species adapted to growing in low-light habitats, these species have a difficult time replacing lost tissue. Thus, these species are thought to invest more reserves in storage and protection than in growth (Kitajima 1994). Thus, reduced herbivory on leaves and cotyledons, low leaf and cotyledon turnover rates, and tough, well-protected cotyledons and leaves are expected in species growing in low-light conditions. In this study, I had three main goals. For 24 tree and shrub species, I wanted to compare: Growth and survival of seedlings of each species across different habitats: closed canopy forest, treefall gap, grass/forest edge, and grassland (thus characterizing plasticity).

PAGE 84

72 Species contrasting in habitat associations in terms of their growth and survival both within and among habitats (to investigate adaptation). How characters typically related to particular habitat associations (e.g., seed size and cotyledon type) are related to growth and survival. I expected that forest-associated species would have higher survival, lower growth, larger seeds, and lower turnover rates of cotyledons and leaves than species associated with gaps, grassland, and grass/forest edges that have higher light levels. I also expected that, within species, seedlings growing in high-light habitats would have higher growth and survival than seedlings growing in closed canopy forest where light is less plentiful. Furthermore, seedlings within a species were expected to show differences in growth and survival with high-light associated species showing greater plasticity in their responses. Materials and Methods Study Site Kibale National Park (766 km 2 ; 0’-0’N and 30’-30’E) is located in western Uganda, 24 km east of the Rwenzori Mountains at an elevation of approximately 1500 m. Between 1998 and 2000, a mean of 1760 mm of rain fell per year, annual mean daily maximum temperature was 23.1C, and minimum temperature was 15.1C (C.A. Chapman & L.J. Chapman, unpublished data). Kibale consists of mature forest (57%), colonizing forest (19%), grassland (15%), woodland (4%), swamp (4%), and plantations of exotic trees (1.0%; Chapman & Lambert 2000). Seedling growth and survival were measured in four habitat types: closed canopy forest, treefall gap, grassland, and grass/forest edge. The forest is mid-altitude moist tropical forest with a canopy height of 25 to 30 m (Howard 1991), and natural disturbances leading to treefall gaps are common (Skorupa & Kasenene 1984). Grasslands are thought to be anthropogenic in origin, but have been abandoned for at least 50 to 100 years (Kingston 1967). Forest has reestablished in some grasslands,

PAGE 85

73 while the majority are still dominated by grasses that are maintained by fire, elephants, and/or competitive dominance of grasses over trees (Lang Brown & Harrop 1962, Kingston 1967, Wing & Buss 1970). Queen Elizabeth National Park, contiguous with Kibale, is dominated by grasslands; many species of shrubs and trees, especially in the Fabaceae, are found in grasslands in both parks (Lock 1993, Lenzi-Grillini et al. 1996). Notably missing from Kibale, however, are species in the genus Euphorbia. Methods From May 1999 to May 2000, fruits and seeds were collected from fruiting shrub and tree species. Seeds were germinated in raised nursery beds (1 x 13 m). Three light treatments (as measured at 12 pm in June 2000, heavy shade: 27.3 mol m -2 s -1 PAR Photosynthetically Active Radiation; medium shade: 182.3 mol m -2 s -1 ; light shade: 366.8 mol m -2 s -1 ) were used, and seeds were matched to their typical habitat’s relative light conditions for germination. During the dry season, seedbeds were watered every 2 to 3 days by saturating the surface soil (~ 2.1 L/m 2 ). Seedling germination was high enough (~160 seedlings) for 24 species and these were included in the experiment (Table 4-1, Appendix D); however, four species had lower germination and were still used (see below). Once germinated and at the first photosynthetic organ stage (either cotyledon or first leaf depending on the species), 80 seedlings were haphazardly selected for bare root transplanting along 20 randomly located transects in each habitat. Photosynthetic cotyledons were defined as those cotyledons that were green and < 0.9 mm thick (definition adapted from Kitajima 1992a). For gap transects, I located recent treefall gaps and randomly chose a subset of these gaps. In four species, fewer seedlings germinated and so for Blighia unijugata, Celtis africana, and Myrianthus arboreus, 15 seedlings were planted in forest and 15 seedlings were planted in grassland, and for Drypetes gerrardii, 14 to 15 seedlings were planted in each of the four

PAGE 86

74 habitats. One seedling of each species was planted along a transect at least 1 m from other transplants for a total of 1748 seedlings. As seeds became available at different times during the year and species varied in the germination rates, species were planted throughout the year (June 1999 to May 2000). Each seedling transplanted during the dry season received four handfuls of water on the day of planting. Two days after planting, seedlings were measured for survival, aboveground tissue lengths (stem, leaf, and photosynthetic cotyledon), % leaf and cotyledon herbivory, and leaf and cotyledon presence. Seedlings were monitored every 1 to 2 weeks for transition to the second photosynthetic organ stage. Upon reaching this stage, measurements were repeated. Seedlings were then measured every 3 months for 1 year for survival, % leaf and cotyledon herbivory, and leaf and cotyledon presence (to determine turnover rates). Leaves and cotyledons were monitored for turnover rates by assigning each leaf and cotyledon a number and examining the seedling for presence of each leaf and cotyledon or leaf and cotyledon scar. At 1 year, seedlings were harvested to determine aboveground biomass. Leaves, cotyledons, and stems were separated, dried, and transported to University of Florida. As I wanted to monitor changes in aboveground biomass but was only able to measure aboveground lengths for field seedlings, a population of seedlings for each species was maintained in the nursery to establish allometric relationships between lengths and biomass. A maximum of 20 seedlings was haphazardly harvested at the first and second photosynthetic organ stages. Seedlings were measured for stem, leaf, and aboveground cotyledon lengths (excluding storage cotyledons). Leaves and photosynthetic cotyledons were traced onto paper and transported to University of Florida to determine area using Scion Image (Scion Corporation, Frederick, Maryland, USA). Cotyledon thickness was also measured for all species to calculate inverse cotyledon thickness, which is linearly correlated with photosynthetic capacity per

PAGE 87

75 cotyledon mass (Kitajima 1992a). To increase the sample size for cotyledon thickness, cotyledons for older seedlings were included if their mean did not differ from seedlings used for initial seedling mass. Leaves, cotyledons, and stems were separated, dried, and transported to University of Florida. Several other characters were measured for each species including leaf and cotyledon force of penetration, total seed mass, cotyledon type, and habitat association. Force of penetration was determined at the third leaf stage (or fourth leaf stage for plants with opposite leaves) on photosynthetic cotyledons and the oldest leaves by measuring the amount of force needed to push a rod (Pesola Medio-Line Spring Scale with a pressure set; Pesola AG, Baar, Switzerland) through the lamina avoiding major veins. Total seed mass was measured for each species after seeds were separated from fruit pulp and dispersal appendages and dried for transport to University of Florida (Chapter 3). Seeds and harvested seedlings (nursery and 1 year) were dried at 60C to a constant dry mass and weighed. Species names and habitat associations (forest, gap, grass, edge; Table 4-1, Appendix D) were determined from Eggling and Dale (1952), Polhill (1952), Hamilton (1991), Katende et al. (1995), Lwanga (1996), and Chapter 2. While some species were typically found in more than one habitat (positive habitat associations; Chapter 2); each species was assigned to the single habitat association category in which it was most frequent and also supported in the following literature (Eggling & Dale 1952, Polhill 1952, Hamilton 1991, Katende et al. 1995). Analyses As only two of the 24 species were gap-associated and three species were grassland-associated, habitat association categories were combined (forest = forest + gap, open = grass + edge). These combinations are justified since forest and gap species were more similar to one another and grass and edge species were more similar

PAGE 88

76 to one another with respect to various seed and seedling characters and environmental variables relating to light and water availability (Chapters 2, 3). Seedlings growing in different habitats (planting habitats; forest, gap, edge, grassland) are all considered separately. For inverse of cotyledon thickness, 0 was assigned to all cryptocotylar species following Kitajima (1992b). To increase normality of distributions, percent and proportion data (% herbivory, proportion dead at 3 months and 1 year) were arcsine square root transformed, and total seed mass and 1-year seedling mass were log transformed. Inverse cotyledon thickness was normally distributed. For leaf turnover rates and herbivory, many seedlings had missing values (e.g., due to periods of leaf absence or to no leaf production until 9 months for some species) at different time periods, so individuals were averaged across the four time periods (0 to 3 months, 3 to 6 months, 6 to 9 months, 9 to 12 months). For the first and second photosynthetic organ stages, area and biomass values were estimated for field seedlings by producing allometric relationships between stem, leaf, and aboveground cotyledon lengths versus total aboveground seedling mass and leaf and cotyledon mass and area from nursery seedlings. Nursery seedling lengths were ln transformed. Stepwise multiple regressions were used to determine equations to estimate total aboveground seedling mass for each species using leaf, aboveground cotyledon, and stem lengths. Simple linear regressions were used to determine equations to estimate leaf area and mass for each species using leaf length and photosynthetic cotyledon area and mass using cotyledon length. All regressions were significant at P < 0.05 with mean R 2 values of 0.80 (range: 0.20 to 0.99). Biomass and area estimates for field seedlings were used to determine relative growth rate between the first and second photosynthetic organ stages [RGR = (ln M2 – ln M1)/(t2 – t1) g g -1 d -1 ] and its components for seedlings where M = aboveground biomass, t = time in days, 1 = first photosynthetic organ stage, and 2 = second

PAGE 89

77 photosynthetic organ stage (Hunt 1982). RGR is comprised of net assimilation rate (NAR) and leaf area ratio (LAR) such that RGR = NAR * LAR. NAR is estimated using the equation [(M2 – M1) * (ln LA2 – ln LA1)]/[(t2 t1) * (LA2 – LA1)] (g cm -2 d -1 ), and LAR is estimated using the equation LA2/LM2 (cm 2 g -1 ) where LM = leaf mass and LA = leaf area. LAR is comprised of specific leaf area (SLA) and leaf mass ratio (LMR) such that LAR = SLA * LMR. SLA is estimated using the equation LA2/LM2 (cm 2 g -1 ), and LMR is estimated using the equation LM2/M2 (g g -1 ). RGR was calculated in three ways using planting mass (see Appendix D for RGR data): Excluding storage cotyledons (and treating photosynthetic cotyledons with leaves). Excluding all cotyledons. Including all cotyledons. Storage cotyledons transfer resources to the rest of the seedlings. Thus, RGR calculated with the first option likely biases RGR of storage-cotyledon species upwards, while the second and third options are likely to bias RGR of storage-cotyledons species downwards. Data is mainly discussed with respect to RGR measured excluding storage cotyledons unless specified. Pearson product moment correlations were used to examine relationships between RGR excluding storage cotyledons (or its components: NAR, LAR, SLA, LMR) for seedlings planted in forest relative to the higher light habitats (e.g., RGR in forest versus RGR in gap, edge, and grassland). For these analyses, growth data were averaged for all seedlings planted in the same habitat for each species. I was interested in examining whether growth and leaf and cotyledon herbivory and turnover rates (response variables) depend on a number of species and environmental variables including planting habitat, habitat association, total seed mass, inverse cotyledon thickness, 1-year seedling mass, cotyledon and leaf force of penetration, and rain in the month of planting. One difficulty with this data set is that

PAGE 90

78 seedlings within a species are not independent (Krackow & Tkadlec 2001). If the clumping of data within species is not taken into consideration then results could be driven by such non-independence (e.g., a species better represented in the surviving seedlings). A solution to this problem is to use generalized linear mixed models (GLMMs, see Pinheiro & Bates 2000, Venables & Ripley 2002). GLMMs are a type of generalized linear models (GLMs). GLMs can be compared to linear models (e.g., regressions, ANOVAs) except that they allow for non-linear relationships between independent and response variables using other response distributions besides normal (Gaussian), such as Poisson and binomial; these response distributions are achieved through transformations of the response variables via one of several link functions (Venables & Ripley 2002). For my data, GLMMs are able to incorporate clustering of data by treating species as random effects (Jovani & Serrano 2001). This is accomplished by removing variation due to differences among species from the error term and allowing species to vary randomly around the overall mean (Pinheiro & Bates 2000, Krackow & Tkadlec 2001). The other independent variables (fixed effects) can then be examined. A useful aspect of this mixed effects model is that results for significant fixed effects can be generalized to other species not included in the study (Pinheiro & Bates 2000, Jovani & Serrano 2001). Growth and leaf and cotyledon turnover rates were considered as having Gaussian distributions of errors and examined using an identity link function. Percent leaf and cotyledon herbivory were considered to have binomial distributions of errors (damage = 0, no damage = 1) because many leaves and cotyledons show no damage; these data were examined using a logit link function. GLMMs using penalized quasi-likelihood methods (Venables & Ripley 2002) were run with programs written in R 1.6.2 (R Development Core Team 2002). An analysis was run for each of the response variables with the fixed effects of habitat association, planting habitat, rain in the month

PAGE 91

79 of planting, total seed mass, and inverse cotyledon thickness all included. These variables were measured for all species. Leaf and cotyledon force of penetration and 1-year seedling mass were not measured for all species so these variables were each run in separate analyses as fixed effects along with planting habitat; species was considered a random effect in each analysis. Planting habitat must be entered in every analysis as seedlings within a species are assigned to different habitats so all seedlings within a species are also affected by the habitat in which they grew. Planting habitat though was only examined in the more comprehensive model containing all species as the inclusion of more species should lead to a more accurate estimation of the relationship between response variables and planting habitat. For discrete fixed effects (planting habitat, habitat association), categories of effects are compared to the forest category as this allows forest habitat association and planting habitat to be compared to higher light habitats. Results reported for these analyses include estimations of the partial coefficients (B) for each of the fixed effects and their P values as determined from the Wald’s chi squared statistic (derived from t values for individual contrasts at different levels of discrete fixed effects). This Wald statistic examines whether the coefficients differ significantly from 0 (Pinheiro & Bates 2000). Survival data were evaluated with Forward Stepwise Cox Regression Analyses using SPSS 10.0.5 (SPSS Inc., Chicago, Illinois). This procedure analyzes time to event data (i.e., time to death) allowing for censored cases (e.g., cases still alive at the end of the study; Norusis 1994, Allison 1995). The analysis is based upon the hazard function that estimates the instantaneous probability of an event (i.e., death) given that an individual has survived to that time (Norusis 1994). Results for these analyses include the partial coefficients (B) and standard errors of the covariates, the Wald statistics, and the relative hazards (Exp(B)). For the B coefficients, negative coefficients indicate that as that covariate increases risk of mortality decreases and positive coefficients indicate

PAGE 92

80 that as that covariate increases so does the risk of mortality. The Wald statistic (described above) tests whether B coefficients are different from 0. The relative hazards for discrete covariates are the ratios of the estimated hazards of each of the categories to the first category (i.e., the increase or decrease in risk by being in condition 2 versus condition 1), and the relative risks for continuous covariates when subtracted from 1 and multiplied by 100 are the estimated percent change in hazard for each one unit change in the covariate (Norusis 1994, Allison 1995). My data included right-censored individuals as some seedlings survived for the entire study period, and time was discrete steps of 3-month intervals. Stepwise analyses using forward selection were run with habitat of planting, relative growth rate excluding storage cotyledons, and species as covariates. As a number of species-specific characters have been measured including habitat association, rain in month of planting, total seed mass, and inverse cotyledon thickness, once species was determined to be a significant component of the first analysis, the analysis was rerun to include these species-specific characters and habitat association as covariates. Simple contrasts were used to compare categories of discrete covariates to the forest category as this allowed forest habitat association and planting habitat to be compared to higher light habitats for habitat association and planting habitat. Plots for these data are survival functions with time increments as means of the sampling times (i.e., 1.5, 4.5, 7.5, or 10.5 months) as seedlings that were found dead at each sampling time (3, 6, 9, or 12 months) could have died any time between the two sampling points. Results Growth Analyses Growth (excluding storage cotyledons) and its components in forest were positively related to growth and its components in all other habitats (Table 4-2, Figure 4-1). RGR did not differ between species with storage and photosynthetic cotyledons for

PAGE 93

81 any of the three measures of growth (Figure 4-2); although growth measures were less variable for both types of species when cotyledon mass was included in analyses. RGR was also negatively related to inverse cotyledon thickness for RGR measured without storage cotyledons but not when all cotyledons were included or excluded (Table 4-3). RGR was negatively related to force of cotyledon penetration when measured without all cotyledons but not when measured with all cotyledons or just without storage cotyledons. For all three measures of RGR, growth differs by planting habitat with seedlings planted along edges having higher growth than seedlings planted in forest. Survival Analyses At 3 months, 43.8% of seedlings were still alive, while at 12 months, 17.5% of seedlings remained alive. The proportion of individuals dying in the forest was positively correlated with proportion dying in all other habitats both at 3 months and 12 months (Table 4-2), except for the proportion dying in grassland at 12 months which was only marginally significantly correlated with the proportion dying in the forest (P = 0.052). In Forward Stepwise Cox Regression Analyses for the first model, seedling survival was related to the habitat in which it was planted and species. In the second model, seedling survival was related to the habitat in which it was planted, its habitat association, seed size, inverse cotyledon thickness, and rain in the month of planting (Table 4-4). Seedlings showed lower mortality when planted in gaps than forest, but mortality was equally likely in forest versus edge and grassland (Figure 4-3). Forest-associated species had higher survival than open-associated species across all habitats (Figure 4-4), and seedling survival decreased by 22.9% for each ten-fold decrease in seed mass between species. Seedlings also declined in survival by 4% for every 10 mm decrease in rainfall in the month of planting and by 6.3% for every 1 mm -1 decrease in inverse cotyledon thickness (i.e., as cotyledons became thinner survival increased). RGR in

PAGE 94

82 model 1 and the interaction term between habitat of planting and planting habitat in model 2 were not significant predictors of survival. Leaf and Cotyledon Turnover Rates Percent leaf herbivory was positively related to total seed mass and 1-year seedling mass and negatively related to cotyledon force of penetration (Table 4-5). Forest-associated species had lower leaf herbivory than open-associated species in the same habitat (Figure 4-5), and seedlings growing in forest had lower leaf herbivory than seedlings growing in gaps but higher leaf herbivory than seedlings growing in grassland and edge (Figure 4-6). Percent cotyledon herbivory was also positively related to 1-year seedling mass. Leaf production was positively related to 1-year seedling mass and negatively related to leaf and cotyledon force of penetration. Leaf production was greater in seedlings growing in gaps and grassland than in seedlings growing in forest. Leaf mortality was negatively related to both leaf and cotyledon force of penetration, and open-associated species had higher leaf mortality than forest-associated species. Seedlings growing in gaps had higher leaf mortality than seedlings growing in forest. Cotyledon mortality was negatively related to inverse cotyledon thickness, leaf force of penetration, and cotyledon force of penetration. Seedlings growing in grassland had lower cotyledon mortality than seedlings growing in forest. Discussion Plasticity among Habitats Seedlings differed in a number of ways based upon the habitat into which they were planted (Kitajima 1994, Kobe 1999, Montgomery & Chazdon 2002). Seedlings planted in the higher light habitat of the edges had higher growth than seedlings planted in forest. Similar to findings by Veenendaal et al. (1996a) and Augspurger (1984a), gap-planted seedlings suffered lower mortality than forest-planted seedlings. In the former study, it was thought that forest seedlings suffering greater drought stress in the

PAGE 95

83 dry season than gap seedlings, while in the latter study seedlings growing in gaps were less likely to suffer damage from pathogens. In Costa Rica, Kobe (1999) also found seedlings in higher light habitats suffered lower mortality. In the present study, forest and gap habitats did not differ in soil moisture but did differ in light. Thus, gap habitats may be functionally drier habitats than forest perhaps leading to less pathogen damage than forest as found by Augspurger (1984a). Interestingly, in this study, the mortality of forest-planted seedlings did not differ from edge and grassland-planted seedlings. It would be useful to determine the source of mortality in these habitats as one would expect that survival in the low light of closed canopy forest may differ from the two other habitats where light is higher. Anecdotally, elephants periodically visited the grassland transects killing many seedlings, but never visited forest transects. Also many seedlings wilted in the edge habitat, and in a previous study (Chapter 2), edge habitat appeared to have the lowest water availability of the four habitats. Leaf turnover rates and leaf and cotyledon herbivory were higher in gap-planted seedlings than forest-planted seedlings and these results are similar to findings from other studies (Augspurger 1984b, Poorter 1998). It is surprising that seedlings grown along the edge did not show the greatest leaf turnover rates as they had the greatest growth; however, gap-grown seedlings showed a trend for greater growth than forest-grown seedlings. Also, the lower survival in edge relative to forest and lower soil water in edge versus all other habitats (Chapter 2) suggests that low water availability may be limiting seedlings from utilizing the higher light for increased leaf production (but see Wright et al. 2002 for species adapted to low rainfall sites). In general the forest grown seedlings had lower leaf turnover rates than the higher light habitats, suggesting that seedlings growing in an environment where light is limiting are maintaining their leaves longer than are seedlings growing in a higher light habitat with higher herbivory

PAGE 96

84 (gap). These results corroborate other results suggesting that in low-light environments the importance of allocating resources to minimize tissue loss is more important than growth (Poorter 1999). Adaptations to Habitats Growth rates did not differ among species with different habitat associations (forest versus open). This result is surprising as species associated with low-light habitats typically have been documented to have slower growth than species adapted to high-light habitats (Veneklaas & Poorter 1998, Agyeman et al. 1999). It should be kept in mind though that seedlings in this experiment were growing in natural settings where more factors than just light are varying. For instance, both % herbivory and leaf turnover rates for seedlings were lower for forest-associated than open-associated species, similar to findings by Coley (1996) and Agyeman et al. (1999). The faster leaf turnover and greater herbivory of open-associated species suggests that inherent growth may be faster and leaf tissue defense may be lower than in forest-associated species. Forest-associated species may be investing more in tissue protection. Furthermore, forest-associated species had a lower risk of mortality than open-associated species across all habitats. Coley (1987), in fact, found a negative relationship between height growth and degree of defense, and a positive relationship between height growth and degree of herbivory for saplings growing in shade. In combination, the leaf herbivory, leaf turnover rates, and survival data suggest forest-associated species are better defended than open-associated species. Open-associated species may in fact have higher inherent growth rates, but if forest-associated species allocate more to defense, the net result is that growth rates do not differ between the two categories. To tease apart these differences, it would be useful to either grow these seedlings in a greenhouse experiment in which only light is manipulated or in a field experiment in which some seedlings are caged thus decrease herbivory and stem snaps. As no

PAGE 97

85 interaction between habitat association and planting habitat was found, forest-associated species had higher survival across all habitats of planting, and gap-planted seedlings had higher survival across both forest and open habitat association categories. Both growth and survival in forest showed positive correlations with growth and survival in the other three habitats. Thus, species with high growth or mortality in one habitat had high growth or mortality in all other habitats, (Augspurger 1984b, Kitajima 1994, Poorter 1999). So, while plasticity of response does occur for a species across habitats (see above), the direction of the response is similar. Species adapted to high-light habitats are not showing greater plasticity in growth than species adapted to growing in low-light habitats as was found in other studies (Kitajima 1994, 2002, Osunkoya et al. 1994, Veenendaal et al. 1996b). Other Determinants of Growth and Survival The negative relationship between both growth and mortality and inverse cotyledon thickness is particularly surprising (as growth and mortality increased with increasing cotyledon thickness). Typically one would expect that species with thin cotyledons would be adapted to higher light conditions with higher growth (Garwood 1996) and higher mortality. These results can be better understood by examining the list of species included. Seven species had storage cotyledons; of these species five species were open-adapted, and two species, Erythrina abyssinica and Milletia dura, had the highest growth rates of any species included in the study. As many seedlings with storage cotyledons still had cotyledons at the first photosynthetic organ stage, the high growth rates, especially in these two species, is likely driven by translocation of cotyledon reserves between the two sampling points rather than gains due to photosynthesis. With regard to survival, the two species with the highest rates of mortality, Cordia africana and Sesbania sesban, had fairly thick photosynthetic

PAGE 98

86 cotyledons suggesting the relationship between survival and cotyledon thickness is driven by variation among photosynthetic cotyledon species. Seedling survival was also related to a number of other variables to which growth was unrelated. Seedling survival increased as seed size increased (Walters & Reich 2000). Large seeds are likely supplying their developing seedling with greater energetic reserves and thus increasing their probability of survival. Surprisingly, a relationship was not found among species between survival and growth as expected from the literature (Kitajima 1994, Dalling et al. 1998, Poorter 1999). As seedling survival was fairly low across the study (compare to 60% survival of seedlings planted in closed canopy forest in Costa Rica at 14 months, Montgomery & Chazdon 2002; 95% survival of seedlings planted in grasslands in Kibale at 6 to 8 months, Duncan & Duncan 2000) a smaller sample of seedlings was measured for growth than survival. Nevertheless, growth data were available for 606 of the 1748 seedlings suggesting I should have a large enough sample to detect relationships with mortality if they exist. But, if as suggested above, differential tissue loss in forest-associated versus open-associated seedlings is leading to similar net growth rates, then this lack of relationship is not as surprising. Other Determinants of Leaf and Cotyledon Turnover Rates and Herbivory Herbivory can differ greatly within similar habitats, and these differences have been related to leaf force of penetration and chemical defenses (Coley 1987). In my study, leaf and cotyledon force of penetration were both negatively related to leaf production (number of new leaves per time period) and leaf and cotyledon mortality (number of leaves and cotyledons dead per time period) suggesting that leaves requiring greater force for penetration, perhaps with greater investment in defenses, are maintained for longer periods of time (Wright & Cannon 2001). A negative relationship between leaf force of penetration and leaf turnover rates has been found previously (Reich et al. 1991). Surprisingly, though, % herbivory on leaves was only related to

PAGE 99

87 cotyledon force of penetration and % herbivory on cotyledons was not related to either measure of force of penetration. One would expect that leaf and cotyledon tissue more resistant to force of penetration should be incurring less herbivory (Coley 1987). Leaf and cotyledon force of penetration though were measured on seedlings at the threeor four-leaf stage on the oldest leaves and photosynthetic cotyledons. It may be that leaf force of penetration varies with ontogeny and thus the measure of leaf force of penetration is not adequate to describe leaves and cotyledons across their lifetimes. One year seedling mass was positively related to % herbivory of leaves and cotyledon and leaf production, suggesting that plants with less well-defended leaves that are produced frequently are found on seedlings that reach a larger mass at 1 year. Interestingly, growth rate was not related to 1-year seedling mass. Total seed mass was only related positively to % herbivory of leaves. From a previous study (Chapter 3), large seeds produced larger initial seedlings. Thus seedlings with larger seeds, larger initial seedlings, and larger seedlings at 1 year are suffering the greatest leaf herbivory. This result is surprising as small seeds have been found to have faster growing seedlings (Osunkoya et al. 1994), and faster growing seedlings typically have less investment in leaf defense and thus greater herbivory (Coley 1987). Conclusions Both plastic responses by seedlings within the same species planted in different habitats and potentially adaptive responses by different species planted in the same habitats were detected. Seedling growth, mortality, and leaf and cotyledon herbivory and turnover rates increased in higher light environments; however, these responses were not consistent across all higher light environments. Such differences may be due to differences in abiotic (moisture) and biotic (herbivores) factors among these environments. Only seedling mortality and leaf and cotyledon herbivory and turnover rates differed between seedlings of forest and open habitat associations, growth did not.

PAGE 100

88 But, the suite of characters associated with forest-associated species including lower leaf herbivory and mortality and higher survival than open-associated species suggests that through greater investment in defensive characters shade adapted species have attained similar realized growth to shade-intolerant species at least for the time period between development of the first and second photosynthetic organs. By understanding how species respond to the different hazards they encounter we gain insight into important limitations for seedlings establishment and growth. These results suggest that important limitations include both competition and low soil moisture in higher light habitats and potentially herbivores or other causes of tissue loss in the lower light habitat.

PAGE 101

89 Table 4-1. Seed and seedling characters (totals seed mass, cotyledon thickness, cotyledon type, 1-year seedling mass, habitat association) for 24 species in Kibale National Park, Uganda. Cotyledon types are as follows P = phanerocotylar, C = cryptocotylar, E = epigeal, H = hypogeal, F = foliaceous, and R = reserve/storage. Cotyledon thickness was not measured for cryptocotylar storage cotyledons. Family Species Total seed mass (g) Cotyledon thickness (mm) Cotyledon type Habitat association 1-Year seedling mass (g) Fabaceae Albizia grandibracteata 0.049 0.92 PER grass 0.11 Sapindaceae Allophylus macrobotrys 0.049 1.84 PER edge 0.08 Sapindaceae Blighia unijugata 0.371 0.00 CHR gap 0.20 Phyllanthaceae Bridelia micrantha 0.068 0.20 PEF grass 0.64 Celtidaceae Celtis africana 0.037 0.19 PEF edge 0.01 Boraginaceae Cordia africana 0.166 0.41 PEF edge 0.34 Boraginaceae Cordia millenii 3.002 0.56 PEF forest 2.84 Euphorbiaceae Croton megalocarpus 2.160 0.33 PEF forest 7.77 Ebenaceae Diospyrus abyssinica 0.115 0.18 PEF forest 0.08 Salicaceae Dovyalis macrocalyx 0.027 0.16 PEF forest 0.35 Putranjivaceae Drypetes gerrardii 0.331 0.26 PEF forest 0.28 Fabaceae Erythrina abyssinica 0.200 3.80 PHR grass 1.59 Apocynaceae Funtumia africana 0.029 0.19 PEF forest 0.17 Bignoniaceae Markhamia lutea 0.037 0.60 PEF edge 0.78 Fabaceae Milletia dura 0.190 1.72 PER edge 2.98 Sapotaceae Mimusops bagshawei 0.261 0.22 PEF forest 0.17 Annonaceae Monodora myristica 0.919 0.26 PEF gap 0.40 Moraceae Myrianthus arboreus 0.744 0.00 CHR forest 0.12 Fabaceae Piptadeniastrum africanum 0.141 0.62 PEF forest 0.12 Pittosporaceae Pittosporum manni 0.026 0.16 PEF edge 0.53 Rosaceae Prunus africana 0.196 0.00 CHR edge 0.15 Fabaceae Sesbania sesban 0.009 0.55 PEF edge NA Bignoniaceae Spathodea campanulata 0.006 0.21 PEF edge 0.36 Annonaceae Uvariopsis congensis 0.323 0.19 PEF forest 0.12

PAGE 102

90 Table 4-2. Pearson product moment correlations between growth (relative growth rate, net assimilation rate, leaf area ratio, specific leaf area, leaf mass ratio) and mortality (proportion dead) for seedlings planted in forest and that same character measured for seedlings planted in the three other habitats: treefall gap, grass/forest edge, and grassland. Correlations are for species averaged by habitat. Significant relationships are in bold; ns P > 0.05, * P < 0.05, ** P < 0.005, *** P < 0.0005. RGR in forest NAR in forest LAR in forest SLA in forest LMR in forest Mortality 3 months forest Mortality 12 months forest Gap 0.770*** 0.835*** 0.922*** 0.925*** 0.818*** 0.836*** 0.705*** Edge 0.862*** 0.948*** 0.919*** 0.985*** 0.556* 0.686*** 0.545* Grass 0.782*** 0.718*** 0.844*** 0.877*** 0.619* 0.545** 0.401 ns

PAGE 103

91 Table 4-3. Results for the analysis with generalized linear mixed models to test whether relative growth rate (calculated in three ways: using seedling mass without storage cotyledons, with all cotyledons, without all cotyledons) depends on other species characters and environmental characteristics for 24 tree and shrub species in Kibale National Park, Uganda. The effect of planting habitat was analyzed as contrasts of gap, edge, or grassland relative to forest, while the effect of habitat association was analyzed as the contrast of open relative to forest. Lines between rows separate different analyses. In all analyses, “species” was included as a random effect (B: partial coefficients; P: significance from Wald’s chi squared statistic; see methods). Significant relationships (P < 0.05) are in bold. Relative growth rate Without storage cotyledons With all cotyledons Without all cotyledons B P B P B P Habitat association (open-forest) 0.0021 0.875 -0.0006 0.958 -0.0006 0.975 Rain in month of planting 0.0001 0.111 0.0001 0.227 0.0001 0.336 Total seed mass -0.012 0.251 -0.0083 0.355 -0.0115 0.479 Inverse cotyledon thickness -0.0079 0.013 -0.0025 0.317 -0.0077 0.0926 Planting habitat (gap-forest) 0.0068 0.074 0.0046 0.152 0.0058 0.243 Planting habitat (edge-forest) 0.0100 0.014 0.0079 0.029 0.0131 0.011 Planting habitat (grass-forest) 0.0033 0.424 0.0045 0.194 0.0045 0.393 1-Year seedling mass 0.0143 0.129 0.0118 0.097 0.0237 0.054 Cotyledon force of penetration -0.0001 0.293 -0.0001 0.293 -0.0003 0.032 Leaf force of penetration -0.0003 0.077 -0.0001 0.638 -0.0003 0.276

PAGE 104

92 Table 4-4. Results for Forward Stepwise Cox Regression Analyses for seedling mortality (B: partial coefficients; P: significance from Wald’s chi squared statistic; Exp(B): relative hazards; see methods). The analysis was first run (Model 1) with species, planting habitat, and relative growth rate calculated for the seedling mass excluding storage cotyledons as covariates. Once species was determined to be a significant component of the model, the analysis was run (Model 2) with species specific characters (habitat association, total seed mass, inverse cotyledon thickness, and rain in month of planting) and planting habitat as covariates. The effects of contrasts in planting habitat and habitat association are included, but the effects of contrasts for species are not shown as I was more interested in results for species specific characters than individual species. Contrasts for planting habitat are of gap, edge, or grassland relative to forest and for habitat associations are of open relative to forest. Several variables were not significant components of the models (model 1: relative growth rate excluding storage cotyledons, model 2: planting habitat by habitat association interaction). Significant relationships are in bold. Model 1 B SE Wald D.F. P Exp(B) Planting habitats 18.11 3 0.0004 Gap relative to forest -0.4725 0.1560 9.17 1 0.0025 0.623 Grass relative to forest -0.1074 0.1577 0.46 1 0.4960 0.898 Edge relative to forest 0.1033 0.1504 0.47 1 0.4921 1.109 Species 145.33 23 <0.0001 Model 2 Planting habitats 33.96 3 <0.0001 Gap relative to forest -0.3189 0.0784 16.55 1 <0.0001 0.727 Grass relative to forest 0.0506 0.0718 0.50 1 0.4809 1.052 Edge relative to forest 0.1086 0.0736 2.17 1 0.1403 1.115 Open relative to forest-association 0.1977 0.0797 6.15 1 0.0131 1.219 Total seed mass -0.2604 0.0584 19.86 1 <0.0001 0.771 Inverse cotyledon thickness -0.0655 0.0175 13.96 1 0.0002 0.937 Rain in month of planting -0.0033 0.0004 78.84 1 <0.0001 0.997

PAGE 105

Table 4-5. Results for the analysis with generalized linear mixed models to test whether % leaf and cotyledon herbivory, leaf production, and leaf and cotyledon mortality depend on other species characters and environmental characteristics for 24 tree and shrub species in Kibale National Park, Uganda. The effect of planting habitat was analyzed as contrasts of gap, edge, or grassland relative to forest, while the effect of habitat association was analyzed as the contrast of open relative to forest. Lines between rows separate different analyses. In all analyses, “species” was included as a random effect (B: partial coefficients; P: significance from Wald’s chi squared statistic; see methods). Significant relationships (P < 0.05) are in bold. % Leaf herbivory % Cotyledon herbivory Leaf production Leaf mortality Cotyledon mortality B P B P B P B P B P Habitat association (open-forest) 2.04282 0.0030 1.364 0.1459 1.409 0.0538 0.7716 0.0409 0.3250 0.0942 Rain in month of planting 0.00001 0.9964 -0.004 0.2955 0.002 0.6306 0.0004 0.8232 -0.0005 0.6046 Total seed mass 1.23642 0.0197 1.248 0.0971 0.346 0.5489 0.2873 0.3328 -0.1944 0.1725 Inverse cotyledon thickness 0.09635 0.4692 0.247 0.4403 0.156 0.3145 -0.0563 0.4775 -0.1940 0.0006 Planting habitat (gap-forest) 0.75784 0.0054 0.393 0.1680 0.883 <0.0001 0.5483 <0.0001 0.0208 0.6743 Planting habitat (edge-forest) -0.79010 0.0101 -0.317 0.3194 0.030 0.8446 -0.0138 0.8952 0.0065 0.9032 Planting habitat (grass-forest) -0.71365 0.0126 -0.068 0.8298 0.384 0.0077 0.1958 0.0509 -0.1203 0.0230 1-Year seedling mass 1.00460 0.0146 1.326 0.0158 0.983 0.0204 0.4555 0.0584 -0.0665 0.7430 Cotyledon force of penetration -0.02207 0.0302 -0.009 0.3711 -0.038 0.0002 -0.0174 0.0033 -0.0073 0.0006 Leaf force of penetration -0.01132 0.1751 0.208 0.7831 -0.024 0.0036 -0.0170 0.0001 -0.0099 0.0003 93

PAGE 106

94 502504506500200400600LAR in forest (cm2 g-1) LAR (cm2 g-1)B 50250450650100300500700SLA in forest (cm2 g-1) SLA (cm2 g-1)C -0.040.000.040.080.12-0.020.030.08RGR in forest (g g-1 d-1)RGR (g g-1 d-1) gap edge grassA -0.00040.00020.0008-0.00010.00020.0005NAR in forest (g cm-2 d-1) NAR (g cm-2 d-1)D 0.30.81.30.51.0LMR in forest (g g-1) LMR (g g-1)E Figure 4-1. Relationship between growth and its components in forest and that same character in the other three habitats (gap, edge, grassland) for 24 species in Kibale National Park, Uganda. A) Relative growth rate (excluding storage cotyledons). B) Leaf area ratio. C) Specific leaf area. D) Net assimilation rate. E) Leaf mass ratio.

PAGE 107

95 StoragePhotosyntheticRGR (g g-1 d-1).2.10.0-.1 0.2 0.1 0.0 -0.1 Photosynthetic Storage StoragePhotosyntheticRGR (g g-1 d-1).2.10.0-.1 0.2 0.1 0.0 -0.1 StoragePhotosyntheticRGR (g g-1 d-1).2.10.0-.1 0.2 0.1 0.0 -0.1 StoragePhotosyntheticRGR (g g-1 d-1).2.10.0-.1 0.2 0.1 0.0 -0.1 0.2 0.1 0.0 -0.1 Photosynthetic Storage Figure 4-2. Distribution of species mean relative growth rate for species with photosynthetic (n = 16) versus storage (n = 7) cotyledons calculated in three ways of biomass estimation: excluding storage organs (black hatched plots), including all cotyledons (white plots), excluding all cotyledons (white hatched plots). Circles represent outliers that are 1.5 to 3 box lengths from the box edge and stars represent extreme cases that are more than 3 box lengths from the box edge with boxes representing interquartile ranges. Means of RGR do not differ between photosynthetic and storage cotyledons for any measure of RGR.

PAGE 108

96 Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Planting Habitat edge grass gap forest Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Planting Habitat edge grass gap forest 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0 1.5 4.5 7.5 10.5 12.0 Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Planting Habitat edge grass gap forest Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Planting Habitat edge grass gap forest Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Planting Habitat edge grass gap forest Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Planting Habitat edge grass gap forest 1.2 1.0 0.8 0.6 0.4 0.2 0.0 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0 1.5 4.5 7.5 10.5 12.0 Figure 4-3. Proportion of seedlings surviving as a function of time since planting for different planting habitats (forest, gap, edge, grassland) for 24 species in Kibale National Park, Uganda. Gap-planted seedlings had significantly higher survival than seedlings planted in the other three habitats.

PAGE 109

97 Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Habitat Association open forest Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Habitat Association open forest 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0 1.5 4.5 7.5 10.5 12.0 Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Habitat Association open forest Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Habitat Association open forest Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Habitat Association open forest Time (months)6543210-1Cumulative Survival1.21.0.8.6.4.20.0 Habitat Association open forest 1.2 1.0 0.8 0.6 0.4 0.2 0.0 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0 1.5 4.5 7.5 10.5 12.0 Figure 4-4. Proportion of seedlings surviving as a function of time since planting for species with different habitat associations (forest, open) for 24 species in Kibale National Park, Uganda. Forest-associated species had significantly higher survival than open-associated species.

PAGE 110

98 0510152025forestopen% Cotyledon herbivor y B 01234forestopenLeaf productionC 0.00.51.01.52.02.5forestopenLeaf mortalityD 0.00.51.01.52.0forestopenCotyledon mortalityE 051015forestopen% Leaf herbivoryA Figure 4-5. Means (+ 1 SD) of species measures of leaf and cotyledon characters for forestand open-associated species for 12 forestand 12 open-associated species in Kibale National Park, Uganda. First, species means were calculated across four planting habitats, and then group means were calculated within each habitat association. A) % Leaf herbivory. B) % Cotyledon herbivory. C) Leaf production (number of leaves produced per 3 month time interval). D) Leaf mortality (number of leaves dead per 3 month time interval). E) Cotyledon mortality (number of cotyledons dead per 3 month time interval).

PAGE 111

99 0510152025forestgapedgegrass% Cotyledon herbivor y B 012345forestgapedgegrassLeaf productionC 0.00.51.01.52.02.53.0forestgapedgegrassLeaf mortalityD 0.00.51.01.5forestgapedgegrassCotyledon mortalityE 048121620forestgapedgegrass% Leaf herbivoryA Figure 4-6. Means (+ 1 SD) of species measures of leaf and cotyledon characters for seedlings growing in forest, gap, edge, and grassland habitats for 24 species in Kibale National Park, Uganda. First, species means were calculated, and then group means were calculated within each planting habitat A) % Leaf herbivory. B) % Cotyledon herbivory. C) Leaf production (number of leaves produced per 3 month time interval). D) Leaf mortality (number of leaves dead per 3 month time interval). E) Cotyledon mortality (number of cotyledons dead per 3 month time interval).

PAGE 112

CHAPTER 5 GENERAL CONCLUSIONS Species distributions are thought to be due, at least in part, to specific life-history characters allowing individuals of that species to establish and persist in particular environments. Understanding which characters represent adaptations to particular habitats is especially important in tropical environments where tree species richness is extremely high. In this study, I examined associations between tree and shrub species for four habitats (closed canopy forest, treefall gap, grass/forest edge, grassland) in Kibale National Park, Uganda. In 80 species, I then examined how ecological and morphological characters in juvenile trees and shrubs differ based upon species habitat associations. Finally, for a subset of these species, I examined growth and survival of seedlings contrasting in habitat associations when planted in different habitats. The following are my major findings. 1. Habitat differences. a. Environmental. Forest and gap environments had similar soils, but forest had lower light levels than gap. Grassland and edge were more similar to one another than to forest and gap, but differed in a number of important biotic and abiotic factors that determine soil water availability, including root length density of small roots; bulk density; % organic matter; % clay; field moisture capacity; and mean, minimum and maximum % water content. These results suggest forest is the most light-limited habitat and edge is the most water-limited habitat. b. Vegetative. For the three forested habitats, gap and forest had indistinguishable communities with edge being similar, but distinguishable, from both communities. A complete species turnover was only found between grassland and the other three forested habitats. Even though overall community composition was similar in the three forested habitats, in analyses of individual species, many common species were more frequently associated with only one or two of the four habitats. 2. Ecological and morphological characteristics. For 70 tree and shrub species, cotyledon type distributions differed from expectations with phanerocotylar, 100

PAGE 113

101 epigeal, foliaceous cotyledons being the most frequent cotyledon type overall and in forest-associated species. Associations were found among seed size, cotyledon type, and other characters and associations between these suites of characters and habitat associations. Large seeds with thick storage cotyledons, slow germination, large-stature adults, and dispersal by large-bodied animals were common in species from forest and gap habitats, and small seeds with thin cotyledons, rapid germination, small-stature adults, and dispersal by small-bodied animals were common in species from open habitats. Phylogenetic analyses support these suites and indicate that seed size may be less influenced than cotyledon type by phylogeny. 3. Growth and survival differences among and within species. For 24 tree and shrub species, seedlings of the same species varied in their responses to different habitats with seedling growth highest along edges and survival highest in gaps. Leaf herbivory and turnover rates were highest for seedlings growing in gaps. No differences were found between forestand open-associated species for growth, but forest-associated species had higher survival and lower leaf herbivory and mortality than open-associated species. Leaf and cotyledon turnover rates were negatively related to leaf and cotyledon force of penetration. Total seed mass was positively related to seedling survival and % leaf herbivory, and cotyledon thickness was positively related to growth (when storage cotyledons are excluded from growth calculations) and negatively related to seedling survival and cotyledon mortality. Growth did not differ between storage and photosynthetic cotyledons. This study demonstrates that seedlings show variation in responses among habitats with forest grown seedlings allocating less resources to growth and more resources to leaf maintenance than species associated with the higher light habitats. Forest-associated species also allocated more resources to leaf maintenance which may translate into the higher survival. In summary, these results suggest that light, and perhaps soil moisture, differences among habitats may exert strong selective pressures with species associated with high light, where competition may be high, allocating more to construction of less well defended tissue (and perhaps faster growth), and species associated with low light allocating more to longer lived (and likely better defended) tissue. Seed size, and perhaps its associated characters, was related to seedling survival at least through the first year of growth. Relationships with cotyledon morphology were less conclusive. An increased understanding of the importance of juvenile life-history characters for growth and survival provides insight into adult distributions across a range of habitats and perhaps insight into factors promoting high tropical species richness.

PAGE 114

APPENDIX A JUVENILE AND ADULT CHARACTERS FOR 80 SPECIES Phylogenetic relations and seed, seedling, and adult characters (growth form (Gr frm), maximum adult height (Ad ht), days to germination (Day to germ), % germination (% Germ), seed reserve mass (SRM), totals seed mass (TSM), % total seed mass comprised of seed reserve (%SRM), dispersal type (Disp), initial seedling mass (ISM), inverse of cotyledon thickness (Inv CtTh), cotyledon type (Ct typ), habitat type (Hab)) for 70 species in Kibale National Park, Uganda. % Phylogenetic relations include the family, order, and higher taxonomic levels. Higher taxonomic levels are not mutually exclusive, but instead, internested clades. Thus, eurosids and euasterids are subgroups within eudicots. SRM is percent of total seed mass comprised of seed reserve. Growth form abbreviations are as follows s = shrub, tl = treelet, t = tree. Dispersal type abbreviations are as follows lg anim = large-animal dispersed, sm anim = small-animal dispersed, wind = wind-dispersed, auto = autochorously-dispersed. Cotyledon types are as follows P = phanerocotylar, C = cryptocotylar, E = epigeal, H = hypogeal, F = foliaceous, and R = storage.

PAGE 115

Table A-1. Juvenile and adult characters for 80 species Higher taxonomic levels A d ht (m) Day Order Family Species Gr. % SRM (g) TSM % Disp ISM (g) Inv Ct Hab to frm Germ (g) SRM CotTh typ germ (mm) eurosid I Fabales Fabaceae t 30 5.7 4.2 0.036 0.0490 73.2 auto 0.028 1.08 PER grass Albizia grandibracteata eurosid II Sapindales Sapindaceae t 103 Allophylus macrobotrys 20 17.0 22.2 0.036 0.0490 74.3 smanim 0.057 0.54 PER edge eurosid I Zygophyllales Zygophyllaceae t 35 6.285 28.528 22.0 lganim 0 CHR forest Balanites wilsoniana rosids Geraniales Melianthaceae t 15 50.0 4.6 0.175 0.2034 85.9 smanim 1.73 PEF edge Bersama abyssinica eurosid II Sapindales Sapindaceae t 15 29.0 7.9 0.315 0.3710 85.0 lg anim 0.223 0 CHR gap Blighia unijugata eurosid I Malpighiales Phyllanthaceae tl 15 11.0 8.0 0.048 0.0680 70.0 sm anim 0.026 5.02 PEF grass Bridelia micrantha eurosid II Sapindales Meliaceae t 25 36.5 8.3 10.69 16.592 64.4 lganim 0 CHR forest Carapa procera eurosid I Rosales Celtidaceae t 35 21.0 51.4 0.008 0.0370 22.5 smanim 0.007 5.19 PEF edge Celtis africana eurosid I Rosales Celtidaceae t 25 0.011 0.0289 39.4 smanim 0.010 2.7 PEF gap Celtis durandii eurosid I Rosales Celtidaceae tl 10 67.0 28.3 0.023 0.0911 Chaetacme aristata 25.4 smanim 0.016 1.71 PEF edge asterids Ericales Sapotaceae Chrysophyllum albidum t 60 59.0 10.4 0.147 0.4232 34.8 lganim 0.137 4.03 PEF forest eurosid II Sapindales Rutaceae Clausena anisata tl 8 0.067 0.0712 94.5 smanim PER edge euasterid I Gentianales Rubiaceae Coffea euginioides s 4 0.156 0.1738 87.7 smanim PEF gap euasterid I [unplaced] Boraginaceae Cordia africana t 15 20.5 0.024 0.1660 14.4 smanim 0.024 2.41 PEF edge euasterid I [unplaced] Boraginaceae Cordia millenii t 35 27.0 92.3 0.168 3.0020 5.6 lganim 0.158 1.78 PEF forest eurosid I Malpighiales Euphorbiaceae Croton macrostachyus tl 25 58.0 50.0 0.017 0.0358 47.8 smanim 0.019 4.75 PEF edge

PAGE 116

104 Table A-1. Continued Higher taxonomic levels Order Family Species Gr. frm A d ht (m) Day to germ % Germ SRM (g) TSM (g) % SRM Disp ISM (g) Inv CotTh (mm) Ct typ Hab eurosid I Malpighiales Euphorbiaceae Croton megalocarpus t 40 30.0 100.0 1.294 2.1600 59.9 lganim 0.198 3.06 PEF forest eurosid I Malpighiales Achariaceae Dasylepis eggelingii tl 10 82.0 0.054 0.0827 65.4 smanim 0.048 4.14 PEF gap asterids Ericales Ebenaceae Diospyros abyssinica t 40 44.0 22.6 0.115 0.1150 100. smanim 0.060 5.43 PEF forest eurosid I Malpighiales Salicaceae Dovyalis macrocalyx s 8 24.0 64.0 0.004 0.0270 14.0 smanim 0.021 6.28 PEF forest eurosid I Malpighiales Salicaceae Dovyalis spinosissima tl 6 0.031 0.0632 49.2 smanim PEF edge eurosid I Malpighiales Putranjivaceae Drypetes gerrardii t 30 39.0 17.5 0.259 0.3310 78.2 lganim 0.268 3.81 PEF forest euasterid I [unplaced] Boraginaceae Ehretia cymosa tl 20 26.5 smanim 4.95 PEF edge eurosid I Fabales Fabaceae Erythrina abyssinica t 15 8.7 100.0 0.169 0.2000 84.5 auto 0.158 0.26 PHR grass eurosid II Sapindales Rutaceae Fagaropsis angolensis t 30 30.0 2.7 0.014 0.0290 46.9 sm anim 0.023 3.41 PEF edge euasterid I Gentianales Apocynaceae Funtumia africana t 30 26.0 17.7 0.024 0.0290 82.4 wind 0.024 5.24 PEF forest euasterid I Lamiales Bignoniacae Kigelia africana t 18 14.0 9.9 smanim 0.084 2.02 PEF edge eurosid II Sapindales Sapindaceae Lepisanthes senegalensis t 15 24.0 64.9 0.632 0.7535 83.9 lganim 0.511 0 CHR gap eurosid I Malpighiales Achariaceae Lindackeria schweinfurthii tl 15 0.034 0.0464 73.8 smanim 5.92 PEF forest euasterid I Lamiales Oleaceae Linociera johnsonii tl 15 56.0 3.8 0.326 0.6143 53.0 lg anim 0.443 0 CHR gap eurosid I Malpighiales Euphorbiaceae Macaranga schweinfurthii t 25 smanim PEF edge asterids Ericales Myrsinaceae Maesa lanceolata tl 15 35.0 0.0003 smanim PEF edge euasterid I Lamiales Bignoniaceae Markhamia lutea t 25 14.3 0.026 0.0370 70.6 wind 0.025 1.68 PEF edge

PAGE 117

105 Table A-1. Continued Higher taxonomic levels Order Family Species Gr. frm A d ht (m) Day to germ % Germ SRM (g) TSM (g) % SRM Disp ISM (g) Inv CotTh (mm) Ct typ Hab eurosid I Celastrales Celastraceae Maytenus gracilipes s 10 35.0 68.9 0.028 0.0342 81.6 smanim 3.33 PEF edge eurosid I Fabales Fabaceae Milletia dura tl 15 10.7 63.8 0.156 0.1900 82.1 auto 0.033 0.58 PER edge asterids Ericales Sapotaceae Mimusops bagshawei t 40 63.0 27.6 0.146 0.2610 56.1 lganim 0.086 4.5 PEF forest eumagnoliids Magnoliales Annonaceae Monodora myristica tl 30 91.0 0.724 0.9190 78.8 lganim 0.405 3.78 PEF gap eurosid I Rosales Moraceae Myrianthus arboreus tl 20 70.5 12.3 0.185 0.7440 24.9 lganim 0.150 0 CHR forest eurosid I Malpighiales Euphorbiaceae Neoboutonia melleri t 15 38.0 17.3 0.128 0.2431 52.8 smanim 0.145 3.79 PEF edge eurosid I Malpighiales Salicaceae Oncoba sp. tl 8 29.0 0.011 0.0169 66.1 smanim 0.011 7.74 PEF edge eurosid II Sapindales Sapindaceae Pancovia turbinata t 20 55.0 2.0 0.210 0.2529 83.1 lganim 0.143 0 CHR gap eurosid I Malpighiales Chrysobalanaceae Parinari excelsa t 45 lganim 0 CHR forest eurosid I Fabales Fabaceae Piptadeniastrum africanum t 50 16.0 0.124 0.1410 87.4 wind 0.116 1.61 PEF forest euasterid II Apiales Pittosporaceae Pittosporum manni tl 10 38.3 0.026 0.0260 100 smanim 0.017 6.22 PEF edge euasterid I Gentianales Apocynaceae Pleiocarpa pycnantha t 10 58.0 0.109 0.1575 69.3 smanim 0.052 2.86 PEF gap euasterid II Apiales Araliaceae Polyscias fulva t 25 38.0 0.004 0.0085 43.6 smanim 0.003 7.24 PEF edge asterids Ericales Sapotaceae Pouteria altissima t 50 42.0 16.2 0.746 0.9930 75.1 lganim 0.607 0.16 PER forest eurosid I Rosales Rosaceae Prunus africana t 40 29.0 28.0 0.155 0.1960 79.0 smanim 0.110 0 CHR edge eurosid II Sapindales Anacardiaceae Pseudospondias microcarpa t 45 54.3 2.1 0.058 0.5334 10.8 lg anim 0.96 PER edge

PAGE 118

106 Table A-1. Continued Higher taxonomic levels Order Family Species Gr. frm A d ht (m) Day to germ % Germ SRM (g) TSM (g) % SRM Disp ISM (g) Inv CotTh (mm) Ct typ Hab euasterid I Gentianales Rubiaceae Psychotria capensis tl 6 0.030 0.0356 84.0 smanim PEF edge euasterid I Gentianales Rubiaceae Psychotria laurace tl 7.5 0.015 0.0213 69.3 smanim PEF forest eurosid II Malvales Malvaceae Pterygota mildbraedii t 50 wind PEF edge euasterid I Gentianales Rubiaceae Rothmania urcelliformis tl 10 32.0 2.8 0.040 0.0531 75.4 sm anim 5.26 PEF forest euasterid I Gentianales Rubiaceae Rytigynia sp. tl 15 0.090 0.1049 86.2 smanim PEF edge eurosid I Malpighiales Euphorbiaceae Sapium ellipticum t 25 0.001 0.0030 8.8 smanim 0.042 4.32 PEF edge eurosid I Fabales Fabaceae Sesbania sesban t 8 13.0 0.003 0.0090 34.2 auto 0.006 1.83 PEF edge euasterid I Lamiales Bignoniaceae Spathodea campanulata t 20 13.5 0.002 0.0060 34.2 wind 0.007 4.78 PEF edge basal eudicot Santalales Olacaceae Strombosia scheffleri t 30 125 0.8 1.310 1.5913 82.3 lg anim PEF forest euasterid I Gentianales Loganiaceae Strychnos mitis t 35 0.291 0.3068 94.7 lganim 0.165 4.19 PEF forest eurosid I Malpighiales Guttiferae Symphonia globulifera t 40 113 61.1 3.544 3.6483 97.1 lganim 2.269 0 CHR forest euasterid I Gentianales Apocynaceae Tabernaemontana pachysiphon t 10 96.0 0.089 0.1263 70.6 smanim 0.102 3.41 PEF gap euasterid I Gentianales Rubiaceae Tarenna pavettoides tl 10 0.0370 smanim PEF edge eurosid I Rosales Celtidaceae Trema orientalis tl 15 37.0 0.001 0.0035 33.0 smanim 4.85 PEF edge eurosid I Rosales Moraceae Trilepisium madagascariense t 30 74.0 58.8 0.422 0.4964 85.0 lganim 0.381 0 CHR forest eurosid II Sapindales Meliaceae Turraea vogelioides tl 10 0.022 0.0301 74.0 smanim PEF forest

PAGE 119

107 Table A-1. Continued Higher taxonomic levels Order Family Species Gr. frm A d ht (m) Day to germ % Germ SRM (g) TSM (g) % SRM Disp ISM (g) Inv CotTh (mm) Ct typ Hab eumagnoliids Magnoliales Annonaceae Uvariopsis congensis t 15 46.0 0.196 0.3230 60.8 lganim 0.114 5.39 PEF forest euasterid I Gentianales Rubiaceae Vangueria apiculata tl 10 40.0 13.8 0.086 0.2782 31.0 lganim 1.66 PEF edge eumagnoliids Canellales Canellaceae Warburgia ugandensis t 40 0.110 0.1097 100 smanim PEF edge eumagnoliids Laurales Monimiaceae Xymalos monospora tl 15 44.0 51.0 0.040 0.0860 46.7 smanim 0.031 4.06 PEF forest eurosid II Sapindales Rutaceae Zanthoxylum gilletii t 35 0.022 0.0379 56.9 smanim PEF forest

PAGE 120

APPENDIX B TEST RESULTS FOR 80 SPECIES Summary of the results for univariate ANOVA’s and t tests comparing means of continuous characters (total seed mass, % total seed mass comprised of seed reserve, days to germination, % germination, initial seedling mass with and without cotyledon mass, inverse of cotyledon thickness, maximum adult height) among categories of discrete characters (cotyledon type, dispersal type, growth form, habitat type). Several categories had low sample size and were combined with other categories (wind + autochorous = non-animal; shrub + treelet; edge + grass = open). Only relationships where differences were found among means in ANOVA’s were analyzed further using Tukey’s b Multiple Comparison procedure. Categories with different superscripts have different means at P <0.05. Categories with means denoted by a are smaller than means denoted by b . Means and standard errors are back transformations for all continuous variables except days to germination and inverse cotyledon thickness, which were untransformed in analyses.

PAGE 121

Table B-1. Test results for 80 species. Discrete Continuous Mean SE Mean SE Mean SE Cotyledon type F P PEF PER *HR Adult height (m) 1.9 0.155 17.7 a 1.1 23.3 a 1.3 25.1 a 1.1 Days to germination 1.4 0.266 41.4 a 4.2 26.0 a 9.5 49.6 a 9.7 Total seed mass (g) 11.7 <0.001 0.068 a 1.279 0.161 a 1.690 1.007 b 1.683 Initial seedling mass with cotyledons (g) 8.8 0.001 0.041 a 1.255 0.080 a,b 1.997 0.321 b 1.407 Initial seedling mass without cotyledons (g) 3.9 0.027 0.021 a 1.279 0.034 a 2.190 0.081 a 1.161 % Germination 0.3 0.718 29.3 a 0.7 17.9 a 1.9 32.9 a 2.0 % Total seed mass comprised of seed reserve 0.5 0.581 62.0 a 0.2 69.6 a 1.9 71.4 a 0.8 Dispersal type F P Large animal Small animal Non animal Adult height (m) 11.0 <0.001 28.6 b 1.1 14.7 a 1.1 23.3 b 1.2 Days to germination 10.7 <0.001 56.5 b 6.5 39.0 b 4.1 13.5 a 2.1 Total seed mass (g) 37.5 <0.001 0.911 b 1.334 0.043 a 1.247 0.044 a 1.604 Initial seedling mass with cotyledons (g) 27.8 <0.001 0.290 b 1.243 0.029 a 1.237 0.029 a 1.509 Initial seedling mass without cotyledons (g) 27.1 <0.001 0.112 b 1.186 0.014 a 1.236 0.013 a 1.523 Inverse cotyledon thickness (mm) 10.4 <0.001 1.6 a 0.4 4.0 b 0.4 2.1 a 0.7 % Germination 0.7 0.497 26.5 a 1.1 25.7 a 0.5 49.9 a 8.9 % Total seed mass comprised of seed reserve 0.2 0.790 61.5 a 0.5 64.9 a 0.3 69.9 a 0.7 Habitat type F P Forest Gap Open Adult height (m) 5.3 0.007 25.9 b 1.1 14.2 a 1.2 17.0 a,b 1.101 Days to germination 8.0 0.001 50.6 b 7.0 61.4 b 9.5 29.0 a 3.2 Total seed mass (g) 10.4 <0.001 0.322 b 1.514 0.225 b 1.406 0.043 a 1.327 Initial seedling mass with cotyledons (g) 11.1 <0.001 0.147 b 1.365 0.123 b 1.631 0.025 a 1.274 Initial seedling mass without cotyledons (g) 14.6 <0.001 0.062 b 1.290 0.062 b 1.463 0.011 a 1.275 Inverse cotyledon thickness (mm) 1.1 0.335 2.9 a 0.5 1.9 a 0.6 3.2 a 0.4 % Germination 0.6 0.575 33.8 a 1.0 14.4 a 3.4 27.4 a 0.9 % Total seed mass comprised of seed reserve 0.6 0.528 65.5 a 0.4 72.6 a 0.3 60.6 a 0.4 Growth form t P Shrub/treelet Tree Adult height (m) -7.3 <0.001 11.5 a 1.1 26.6 b 1.1 Days to germination 0.4 0.659 43.7 a 5.3 40.3 a 4.9 Total seed mass (g) -2.6 0.012 0.056 a 1.374 0.189 b 1.377 109

PAGE 122

110 Table B-1. Continued Discrete Continuous Mean SE Mean SE Mean SE Growth form t P Shrub/treelet Tree Initial seedling mass with cotyledons (g) -1.0 0.304 0.044 a 1.436 0.073 a 1.307 Initial seedling mass without cotyledons (g) -0.8 0.407 0.022 a 1.461 0.032 a 1.274 Inverse cotyledon thickness (mm) 2.6 0.013 3.9 b 0.54 2.4 a 0.3 % Germination 0.1 0.883 29.9 a 0.9 28.0 a 0.7 % Total seed mass comprised of seed reserve 0.2 0.875 65.2 a 0.3 64.0 a 0.3

PAGE 123

APPENDIX C SPECIES SCORES FOR 48 SPECIES Species scores for 48 species in Kibale National Park, Uganda along the first four axes using principal components analyses. Numbers in bold denote the strongest scores for a species with an axis, and numbers in italics denote strong species scores (> 0.50) for a species with other axes. Species abbreviations for species shown in Figure 3-3 are the first two letters of the Genus and the first two letters of the species. Table C-1. Species scores for 48 species. Species Abbreviation Axis 1 Axis 2 Axis 3 Axis 4 Albizia grandibracteata Algr -0.71 1.97 0.08 -0.50 Allophylus macrobotrys Aldu -0.37 1.22 -0.11 -0.94 Bersama abyssinica Beab -0.05 0.45 0.63 -1.11 Blighia unijugata Blun 0.85 0.71 0.14 -0.84 Bridelia micrantha Brmi -1.11 -0.34 0.17 -0.14 Carapa procera Capr 1.80 0.29 -0.49 -0.70 Celtis africana Ceaf -0.65 0.77 -1.33 1.43 Chaetacme aristata Char -0.61 -1.36 -1.04 -1.24 Chrysophyllum albidum Chal 1.26 -0.05 -0.70 1.93 Cordia africana Coaf -0.49 0.52 -1.90 -0.17 Cordia millenii Comi 1.28 0.19 -2.38 0.91 Croton macrostachyus Crma -0.71 -0.88 -0.16 0.34 Croton megalocarpus Crme 1.31 0.37 -0.30 0.93 Dasylepis eggelingii Daeg -0.39 -1.80 0.76 -0.33 Diospyrus abyssinica Diab 0.45 0.64 2.49 1.47 Dovyalis macrocalyx Doma -1.11 -1.55 -0.85 1.35 Drypetes gerrardii Drge 0.91 0.11 0.55 1.05 Erythrina abyssinica Erab -0.61 1.65 0.41 -1.56 Fagaropsis angolensis Faan -0.50 0.84 -0.62 0.57 Funtumia africana Fuaf -0.46 1.02 1.44 1.74 Lepisanthes senegalensis Lese 0.75 0.49 0.07 -1.28 Linociera johnsonii Lijo 0.61 -1.21 -0.59 -1.47 Markhamia lutea Malu -0.82 1.66 0.14 -0.38 Maytenus gracilipes Magr -1.05 -0.73 0.69 -1.03 Milletia dura Midu -0.92 0.54 0.43 -1.88 Mimusops bagshawei Miba 1.05 -0.29 0.08 1.67 Monodora myristica Momy 0.94 -1.65 0.75 -0.09 Myrianthus arboreus Myar 1.06 -1.55 -1.16 -0.61 Neoboutonia melleri Neme -0.33 0.28 -0.42 -0.15 Oncoba sp. Onsp -1.69 -1.26 0.55 0.48 111

PAGE 124

112 Table C-1. Continued Species Abbreviation Axis 1 Axis 2 Axis 3 Axis 4 Pancovia turbinate Patu 0.89 0.15 0.32 -0.97 Piptadeniastrum africanum Piaf 0.24 1.83 1.13 0.66 Pittosporum manni Pima -1.23 -0.87 2.29 -0.43 Pleiocarpa pycnantha Plpy -0.06 -0.24 0.45 -0.38 Polyscias fulva Pofu -1.04 0.23 -0.23 1.85 Pouteria altissima Poal 1.60 0.67 0.04 0.11 Prunus africana Praf 0.27 1.46 -0.04 -0.88 Pseudospondias microcarpa Psmi 0.84 0.30 -2.34 0.00 Rothmania urcelliformis Rour -0.65 -1.17 1.06 0.53 Sesbania sesban Sese -1.59 0.87 -0.84 -0.64 Spathodea campanulata Spca -1.55 1.10 -0.61 1.02 Symphonia globulifera Sygl 2.35 -0.47 1.57 -0.66 Tabernaemontana pachysiphon Tapa 0.14 -0.91 0.87 -0.23 Trema orientalis Tror -1.50 -0.86 -0.60 0.48 Trilepisium madagascariense Trma 1.56 -0.06 0.71 -0.35 Uvariopsis congensis Uvco 0.54 -0.63 0.20 1.24 Vangueria apiculata Vaap -0.23 -1.30 -1.43 -1.38 Xymalos monospora Xymo -0.28 -1.15 0.10 0.60

PAGE 125

APPENDIX D ECOLOGICAL CHARACTERS AND GROWTH FOR 24 SPECIES Appendix D. Species characters (planting habitat, rain in month of planting, force of cotyledon and leaf penetration, 1-year seedling mass, and RGR measured three ways (with all cotyledons, without all cotyledons, without storage cotyledons) averaged for each species within each planting habitat for 24 species in Kibale National Park, Uganda.

PAGE 126

Diospyrus abyssinica edge 234.0 121.6 80.3 0.00036 0.00074 0.00036 Diospyrus abyssinica grass 234.0 121.6 80.3 0.00136 0.00291 0.00136 Dovyalis macrocalyx forest 195.5 34.9 18.8 0.02377 0.03822 0.02377 114Table D-1. Ecological characters and growth rates for 24 species. Species Planting habitat Rain in month of planting (mm) Force of cotyledon penetration (g) Force of leaf penetration (g) RGR with all cotyledons (g g-1 d-1) RGR without all cotyledons (g g-1 d-1) RGR without storage cotyledons (g g-1 d-1) Albizia grandibracteata forest 79.7 0.00042 0.00128 0.00128 Albizia grandibracteata gap 79.7 0.00470 0.01397 0.01397 Albizia grandibracteata edge 79.7 0.00291 0.00636 0.00636 Albizia grandibracteata grass 79.7 0.00386 0.00940 0.00940 Allophylus macrobotrys forest 65.7 2.3 0.00392 0.05143 0.05143 Allophylus macrobotrys gap 65.7 2.3 -0.02703 -0.08647 -0.08647 Blighia unijugata forest 14.4 33.7 0.00252 0.04302 0.04302 Bridelia micrantha forest 234.0 20.5 50.1 0.00243 0.00613 0.00243 Bridelia micrantha gap 234.0 20.5 50.1 0.00081 0.00426 0.00081 Bridelia micrantha edge 234.0 20.5 50.1 0.00622 0.01635 0.00622 Bridelia micrantha grass 234.0 20.5 50.1 0.00323 0.00992 0.00323 Celtis africana forest 22.4 23.8 2.6 0.01037 0.01301 0.01037 Celtis africana grass 22.4 23.8 2.6 -0.00296 -0.01211 -0.00296 Cordia africana forest 22.4 27.6 28.6 -0.00854 0.00311 -0.00854 Cordia africana gap 22.4 27.6 28.6 -0.00853 0.02568 -0.00853 Cordia africana edge 22.4 27.6 28.6 -0.01904 0.01357 -0.01904 Cordia millenii forest 234.0 41.4 28.8 0.01005 0.01848 0.01005 Cordia millenii gap 234.0 41.4 28.8 0.01701 0.03211 0.01701 Cordia millenii edge 234.0 41.4 28.8 0.04530 0.08294 0.04530 Cordia millenii grass 234.0 41.4 28.8 0.01994 0.03523 0.01994 Croton megalocarpus forest 61.4 54.1 57.4 0.00312 0.03085 0.00312 Croton megalocarpus gap 61.4 54.1 57.4 0.02764 0.04569 0.02764 Croton megalocarpus edge 61.4 54.1 57.4 0.00799 0.02211 0.00799 Diospyrus abyssinica forest 234.0 121.6 80.3 -0.00105 -0.00143 -0.00105 Diospyrus abyssinica gap 234.0 121.6 80.3 0.00053 0.00124 0.00053

PAGE 127

Mimusops bagshawei grass 195.5 107.1 126.0 0.00328 0.00409 0.00328 Monodora myristica forest 98.2 97.3 89.6 0.00723 0.00723 Monodora myristica gap 98.2 97.3 89.6 0.00732 0.00732 115Appendix D. Continued Species Planting habitat Rain in month of planting (mm) Force of cotyledon penetration (g) Force of leaf penetration (g) RGR with all cotyledons (g g-1 d-1) RGR without all cotyledons (g g-1 d-1) RGR without storage cotyledons (g g-1 d-1) Dovyalis macrocalyx gap 195.5 34.9 18.8 0.02775 0.04410 0.02775 Dovyalis macrocalyx edge 195.5 34.9 18.8 0.01937 0.02990 0.01937 Dovyalis macrocalyx grass 195.5 34.9 18.8 0.03307 0.05236 0.03307 Drypetes gerrardii forest 234.0 82.0 101.4 0.00242 0.00445 0.00242 Drypetes gerrardii gap 234.0 82.0 101.4 0.00306 0.00494 0.00306 Drypetes gerrardii edge 234.0 82.0 101.4 0.00335 0.00541 0.00335 Drypetes gerrardii grass 234.0 82.0 101.4 0.00453 0.00805 0.00453 Erythrina abyssinica forest 195.5 19.0 0.01818 0.05305 0.05305 Erythrina abyssinica gap 195.5 19.0 0.03954 0.10790 0.10790 Erythrina abyssinica edge 195.5 19.0 0.03590 0.10188 0.10188 Erythrina abyssinica grass 195.5 19.0 0.03696 0.09543 0.09543 Funtumia africana forest 79.7 55.1 50.9 0.01818 0.02695 0.01818 Funtumia africana gap 79.7 55.1 50.9 0.02349 0.03516 0.02349 Funtumia africana edge 79.7 55.1 50.9 0.00081 0.00274 0.00081 Funtumia africana grass 79.7 55.1 50.9 0.00048 0.00090 0.00048 Markhamia lutea forest 114.0 54.5 0.07557 0.14041 0.07557 Markhamia lutea gap 114.0 54.5 0.05626 0.10402 0.05626 Markhamia lutea edge 114.0 54.5 0.11350 0.18782 0.11350 Markhamia lutea grass 114.0 54.5 0.09254 0.16160 0.09254 Milletia dura forest 98.2 4.7 0.02056 0.04383 0.04383 Milletia dura gap 98.2 4.7 0.01340 0.03697 0.03697 Milletia dura edge 98.2 4.7 -0.02479 0.01286 0.01286 Milletia dura grass 98.2 4.7 0.01335 0.03246 0.03246 Mimusops bagshawei gap 195.5 107.1 126.0 -0.00374 -0.00044 -0.00374 Mimusops bagshawei edge 195.5 107.1 126.0 0.00281 0.00303 0.00281

PAGE 128

116Appendix D. Continued Species Planting habitat Rain in month of planting (mm) Force of cotyledon penetration (g) Force of leaf penetration (g) RGR with all cotyledons (g g-1 d-1) RGR without all cotyledons (g g-1 d-1) RGR without storage cotyledons (g g-1 d-1) Monodora myristica grass 98.2 97.3 89.6 0.00776 0.00776 Myrianthus arboreus forest 195.5 31.8 0.00279 0.02880 0.02880 Myrianthus arboreus grass 195.5 31.8 0.00179 -0.02973 -0.02973 Piptadeniastrum africanum forest 79.7 0.00233 0.06043 0.00233 Piptadeniastrum africanum gap 79.7 0.00797 0.05076 0.00797 Piptadeniastrum africanum edge 79.7 0.00064 0.04687 0.00064 Piptadeniastrum africanum grass 79.7 0.00046 0.05610 0.00046 Pittosporum manni forest 234.0 45.5 65.8 0.00361 0.00555 0.00361 Pittosporum manni gap 234.0 45.5 65.8 0.01597 0.02802 0.01597 Pittosporum manni edge 234.0 45.5 65.8 0.01354 0.02273 0.01354 Pittosporum manni grass 234.0 45.5 65.8 0.01050 0.02442 0.01050 Prunus africana forest 14.4 45.5 0.00179 0.00806 0.00806 Prunus africana gap 14.4 45.5 0.00048 0.00492 0.00492 Prunus africana edge 14.4 45.5 0.00053 0.01301 0.01301 Sesbania sesban grass 21.7 -0.00544 -0.02259 -0.00544 Spathodea campanulata forest 22.4 28.0 28.0 0.00000 0.00233 0.00000 Spathodea campanulata gap 22.4 28.0 28.0 0.04199 0.05133 0.04199 Spathodea campanulata grass 22.4 28.0 28.0 0.00949 0.03288 0.00949 Uvariopsis congensis forest 61.4 80.6 64.6 0.00308 0.00464 0.00308 Uvariopsis congensis gap 61.4 80.6 64.6 0.00264 0.00445 0.00264 Uvariopsis congensis edge 61.4 80.6 64.6 0.00282 0.00433 0.00282 Uvariopsis congensis grass 61.4 80.6 64.6 0.00668 0.01326 0.00668

PAGE 129

LIST OF REFERENCES Agyeman, V. K., M. D. Swaine, and J. Thompson. 1999. Responses of tropical forest tree seedlings to irradiance and the derivation of a light response index. Journal of Ecology 87:815-827. Allison, P. D. 1995. Survival Analysis Using the SAS System: A Practical Guide. SAS Publishing, Cary, North Carolina, USA. Andersson, L. and J. H. E. Rova. 1999. The rps16 intron and the phylogeny of the Rubioideae (Rubiaceae). Plant Systematics and Evolution 214:161-186. Andreason, K. and B. Bremer. 1996. Phylogeny of the subfamily Ixoroideae (Rubiaceae). Opera Botanica Biologique 7:119-138. Augspurger, C. K. 1984a. Seedling survival of tropical tree species: Interactions of dispersal distance, light-gaps, and pathogens. Ecology 65:1705-1712. Augspurger, C. K. 1984b. Light requirements of Neotropical tree seedlings: A comparative study of growth and survival. Journal of Ecology 72:777-795. Bigazzi, M. 1995. Investigation on occurrence and ultrastructure of the proteinaceous nuclear inclusions (PNIs) in the Bignoniaceae, with special reference to the geographic distribution patterns. Caryologia 48:211-223. Bongers, F., L. Poorter, R. S. A. R. Van Rompaey, and M. P. E. Parren. 1999. Distribution of twelve moist forest canopy tree species in Liberia and Cote d'Ivoire: Response curves to a climatic gradient. Journal of Vegetation Science 10:371-382. Brokaw, N. 1982. The definition of treefall gap and its effect on measures of forest dynamics. Biotropica 14:158-160. Brokaw, N. and R. T. Busing. 2000. Niche versus chance and tree diversity in forest gaps. Trends in Ecology and Evolution 15:183-188. Brzustowski, J. 2003. Rarefaction Calculator. http://www.biology.ualberta.ca/jbrzusto/rarefact.php. University of Alberta, Alberta, Canada. Accessed 30 January 2003. Burslem, D. F. R. P. 1996. Differential responses to nutrients, shade and drought among tree seedlings of lowland tropical forest in Singapore. Pages 211-244 in M. D. Swaine, editor. The Ecology of Tropical Forest Tree Seedlings. UNESCO, Paris, France. 117

PAGE 130

118 Chapman, C. A., L. J. Chapman, L. Kaufman, and A. E. Zanne. 1999. Potential causes of arrested succession in Kibale National Park, Uganda: Growth and mortality of seedlings. African Journal of Ecology 37:81-92. Chapman, C. A. and J. E. Lambert. 2000. Habitat alteration and the conservation of African primates: Case study of Kibale National Park, Uganda. American Journal of Primatology 50:169-185. Charnov, E. L. 1991. Evolution of life history variation among female mammals. Proceedings of the National Academy of Sciences of the U. S. A. 88:1134-1137. Chase, M. W., C. M. Morton, and J. A. Kallunki. 1999. Phylogenetic relationships of Rutaceae: A cladistic analysis of the subfamilies using evidence from rbcL and atpB sequence variation. American Journal of Botany 86:1191-1199. Chase, M. W., S. Zmarzty, M. D. Lledo, K. J. Wurdack, S. M. Swensen, and M. F. Fay. 2002. When in doubt, put it in Flacourtiaceae: a molecular phylogenetic analysis based on plastid rbcL DNA sequences. Kew Bulletin 57:141-181. Chesson, P. and N. Huntly. 1997. The roles of harsh and fluctuating conditions in the dynamics of ecological communities. American Naturalist 150:519-553. Clark, D. B., M. W. Palmer, and D. A. Clark. 1999. Edaphic factors and the landscape-scale distributions of tropical rain forest trees. Ecology 80:2622-2675. Coley, P. D. 1987. Interspecific variation in plant anti-herbivore properties: The role of habitat quality and rate of disturbance. New Phytologist 106:251-263. Coley, P. D. 1996. Rates of herbivory on different tropical trees. Pages 123-132 in E. G. Leigh Jr., A. S. Rand, and D. M. Windsor, editors. The Ecology of a Tropical Forest: Seasonal Rhythms and Long-term Changes. Smithsonian Institution Press, Washington, D.C., USA. Connell, J. H. 1978. Diversity in tropical rain forests and coral reefs: High diversity of trees and corals is maintained only in a non-equilibrium state. Science 199:1302-1310. Cornelissen, J. H. C., P. C. Diez, and R. Hunt. 1996. Seedling growth, allocation and leaf attributes in a wide range of woody plant species and types. Journal of Ecology 84:755-765. Dalling, J. W., S. P. Hubbell, and K. Silvera. 1998. Seed dispersal, seedling establishment and gap partitioning among tropical pioneer trees. Journal of Ecology 86:674-689. de Carvalho, L. M. T., M. A. L. Fontes, and A. T. de Oliveira-Filho. 2000. Tree species distribution in canopy gaps and mature forest in an area of cloud forest of the Ibitipoca Range, south-eastern Brazil. Plant Ecology 149:9-22.

PAGE 131

119 Duivenvoorden, J. F. 1995. Tree species composition and rain forest-environment relationships in the middle Caqueta area, Colombia, NW Amazonia. Vegetatio 120:91-113. Duncan, R. S. and V. E. Duncan. 2000. Forest succession and distance from forest edge in an Afrotropical grassland. Biotropica 32:33-41. Eggeling, W. J. and I. R. Dale. 1952. The Indigenous Trees of the Uganda Protectorate. Government of Uganda Printer, Entebbe, Uganda. Felsenstein, J. 1985. Phylogenies and the comparative method. American Naturalist 125:1-15. Foster, S. A. 1986. On the adaptive value of large seeds for tropical moist forest trees: A review and synthesis. The Botanical Review 52:260-299. Franco, M. and J. Silvertown. 1997. Life history variation in plants: An exploration of the fast-slow continuum hypothesis. Pages 210-227 in J. Silvertown, M. Franco, and J. L. Harper, editors. Plant Life Histories: Ecology, Phylogeny and Evolution. Cambridge University Press, Cambridge, UK. Garwood, N. C. 1996. Functional morphology of tropical tree seedlings. Pages 59-129 in M. D. Swaine, editor. The Ecology of Tropical Forest Tree Seedlings. UNESCO, Paris, France. Grime, J. P. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. American Naturalist 111:1169-1194. Grubb, P. J. 1977. The maintenance of species-richness in plant communities: The importance of the regeneration niche. Biological Reviews of the Cambridge Philosophical Society 52:107-145. Grubb, P. J. 1996. Rainforest dynamics: The need for new paradigms. Pages 215-233 in D. S. Edwards, W. E. Booth, and S. C. Choy, editors. Tropical Rainforest Research: Current Issues. Kluwer Academic, Dordrecht, The Netherlands. Grubb, P. J. 1998. Seeds and fruits of tropical rainforest plants: Interpretation of the range in seed size, degree of defense and flesh/seed quotients. Pages 1-24 in D. M. Newbery, H. H. T. Prins, and N. D. Brown, editors. Dynamics of Tropical Communities. Blackwell Science, Cambridge, UK. Hamilton, A. 1991. A Field Guide to Ugandan Forest Trees. Makerere University Press, Kampala, Uganda. Hammond, D. S. and V. K. Brown. 1995. Seed size of woody plants in relation to disturbance, dispersal, soil type in wet Neotropical forests. Ecology 76:2544-2561. Harms, K. E., R. Condit, S. P. Hubbell, and R. B. Foster. 2001. Habitat associations of trees and shrubs in a 50-ha Neotropical forest plot. Journal of Ecology 89:947-959.

PAGE 132

120 Hewitt, N. 1998. Seed size and shade-tolerance: A comparative analysis of North American temperate trees. Oecologia 114:432-440. Hladik, A. and S. Miquel. 1990. Seedling types and plant establishment in an African rain forest. Pages 261-281 in K. S. Bawa and M. Hadley, editors. Reproductive Biodiversity and Conservation 4:10-34. Ecology of Tropical Forest Plants. The Parthenon Publishing Group, Paris, France. Hladik, A. and D. Mitja. 1996. Seedlings, saplings and tree temperaments: Potential for agroforestry in the African rain forest. Pages 173-192 in M. D. Swaine, editor. The Ecology of Tropical Forest Tree Seedlings. UNESCO, Paris, France. Howard, P. C. 1991. Nature Conservation in Uganda's Tropical Forest Reserves. IUCN Publications Services Unit, Cambridge, UK. Hu, J. M., M. Lavin, M. F. Wojciechowski, and M. J. Sanderson. 2000. Phylogenetic systematics of the tribe Millettieae (Leguminosae) based on chloroplast trnK/matK sequences and its implications for evolutionary patterns in Papilionoideae. American Journal of Botany 87:418-430. Hubbell, S. P. 1997. A unified theory of biogeography and relative species abundance and its application to tropical rain forests and coral reefs. Coral Reefs 16:S9-S21. Hubbell, S. P. and R. B. Foster. 1986. Commonness and rarity in a Neotropical forest: Implications for tropical tree conservation. Pages 205-231 in M. E. Soule, editor. Conservation Biology: The Science of Scarcity and Diverstiy. Sinauer Associates, Sunderland, Massachusetts, USA. Hughes, L., M. Dunlop, K. French, M. R. Leishman, B. Rice, L. Rodgerson, and M. Westoby. 1994. Predicting dispersal spectra a minimal set of hypotheses based on plant attributes. Journal of Ecology 82:933-950. Hunt, R. 1982. Plant Growth Curves: The Functional Approach to Plant Growth Analysis. Edward Arnold, London, UK. Ibarra-Manrquez, G., M. M. Ramos, and K. Oyama. 2001. Seedling functional types in a lowland rain forest in Mexico. American Journal of Botany 88:1801-1812. Jongman, R. H., C. J. F. ter Braak, and O. F. R. van Tongeren. 1987. Data Analysis in Community and Landscape Ecology. Pudoc Wageningen, Wageningen, The Netherlands. Jovani, R. and D. Serrano. 2001. Feather mites (Astigmata) avoid moulting wing feathers of passerine birds. Animal Behaviour 62:723-727. Kajita, T., H. Ohashi, Y. Tateishi, C. D. Bailey, and J. J. Doyle. 2001. rbcL and legume phylogeny, with particular reference to Phaseoleae, Millettieae, and allies. Systematic Botany 26:515-536. Kappelle, M., P. A. F. Kennis, and R. A. J. Devries. 1995. Changes in diversity along a successional gradient in a Costa Rican upper montane Quercus forest.

PAGE 133

121 Kasenene, J. M. 1987. The influence of mechanized selective logging, felling intensity and gap-size on the regeneration of a tropical moist forest in the Kibale Forest Reserve, Uganda. PhD Thesis. Michigan State University, East Lansing, Michigan, USA. Katende, A. B., A. Birnie, and B. Tengns. 1995. Useful Trees and Shrubs for Uganda: Identification, Propagation, and Management for Agricultural and Pastoral Communities. Regional Soil Consultants Unit, Nairobi, Kenya. Kingston, B. 1967. Working Plan for the Kibale and Itwara Central Forest Reserves. Government of Uganda Printer, Entebbe, Uganda. Kirkwood, D. and J. J. Midgley. 1999. The floristics of sand forest in northern KwaZulu-Natal, South Africa. Bothalia 29:293-304. Kitajima, K. 1992a. Relationship between photosynthesis and thickness of cotyledons for tropical tree species. Functional Ecology 6:582-589. Kitajima, K. 1992b. The importance of cotyledon functional morphology and patterns of seed reserve utilization for the physiological ecology of Neotropical tree seedlings. Dissertation. University of Illinois, Illinois, USA, Urbana-Champaign. Kitajima, K. 1994. Relative importance of photosynthetic traits and allocation patterns as correlates of seedling shade tolerance of 13 tropical trees. Oecologia 98:419-428. Kitajima, K. 1996a. Cotyledon functional morphology, patterns of seed reserve utilization and regeneration niches of tropical tree seedlings. Pages 193-210 in M. D. Swaine, editor. The Ecology of Tropical Forest Tree Seedlings. The Parthenon Publishing Group, Paris, France. Kitajima, K. 1996b. Ecophysiology of tropical tree seedlings. Pages 559-596 in S. S. Mulkey, R. L. Chazdon, and A. P. Smith, editors. Tropical Forest Plant Ecophysiology. Chapman and Hall, New York, New York, USA. Kitajima, K. 2002. Do shade-tolerant tropical tree seedlings depend longer on seed reserves? Functional growth analysis of three Bignoniaceae species. Functional Ecology 16:433-444. Kitajima, K. and M. Fenner. 2000. Ecology of seedling regeneration. Pages 331-359 in M. Fenner, editor. Seeds: The Ecology of Regeneration in Plant Communities. CAB International, New York, New York, USA. Klute, A. 1986. Methods of Soil Analysis Part 1: Physical and Mineralogical Methods. Soil Science Society of America, Madison. Kobe, R. K. 1999. Light gradient partitioning among tropical tree species through differential seedling mortality and growth. Ecology 80:187-201. Krackow, S. and E. Tkadlec. 2001. Analysis of brood sex ratios: Implications of offspring clustering. Behavioral Ecology and Sociobiology 50:293-301.

PAGE 134

122 Lang Brown, J. R. and J. F. Harrop. 1962. The ecology and soils of the Kibale grasslands, Uganda. East African Agricultural and Forestry Journal 27:264-272. Lantz, H., K. Andreasen, and B. Bremer. 2002. Nuclear rDNA ITS sequence data used to construct the first phylogeny of Vanguerieae (Rubiaceae). Plant Systematics and Evolution 230:173-187. Leishman, M. R. and M. Westoby. 1994. The role of large seed size in shaded conditions: Experimental evidence. Functional Ecology 8:205-214. Leishman, M. R., M. Westoby, and E. Jurado. 1995. Correlates of seed size variation: A comparison among five temperate floras. Journal of Ecology 83:517-530. Leishman, M. R., I. Wright, A. Moles, and M. Westoby. 2000. The evolutionary ecology of seed size. Pages 31-57 in M. Fenner, editor. Seeds: The Ecology of Regeneration in Plant Communities. CAB International, New York, New York, USA. Lenzi-Grillini, C. R., P. Viskanic, and M. Mapesa. 1996. Effects of 20 years of grazing exclusion in an area of the Queen Elizabeth National Park, Uganda. African Journal of Ecology 34:333-341. Leps, J., J. Osbornov-Kosinov, and M. Rejmnek. 1982. Community stability, complexity and species life history strategies. Vegetatio 50:53-63. Lieberman, M., D. Lieberman, R. Peralta, and G. S. Hartshorn. 1995. Canopy closure and the distribution of tropical forest tree species at La Selva Costa Rica. Journal of Tropical Ecology 11:161-178. Lock, J. M. 1993. Vegetation change in Queen Elizabeth National Park, Uganda: 1970-1988. African Journal of Ecology 31:106-117. Loehle, C. 1988. Tree life history strategies: The role of defenses. Canadian Journal of Forest Research 18:209-222. Lord, J., M. Westoby, and M. Leishman. 1995. Seed size and phylogeny in 6 temperate floras: Constraints, niche conservatism, and adaptation. American Naturalist 146:349-364. Lwanga, J. 1996. Trees and shrubs. Pages 15-36 in P. Howard, T. Davenport, and R. Matthews, editors. Kibale Forest Biodiversity Report. Forest Department, Kampala, Uganda. MacArthur, R. H. and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton University Press, Princeton, New Jersey, USA. Martins, E. P. 2001. COMPARE, version 4.4. Computer programs for the statistical analysis of comparative data. Distributed by the author via the WWW at http://compare.bio.indiana.edu/. Department of Biology, Indiana University, Bloomington, Indiana, USA. Accessed 30 January 2003.

PAGE 135

123 Montgomery, R. A. and R. L. Chazdon. 2002. Light gradient partitioning by tropical tree seedlings in the absence of canopy gaps. Oecologia 131:165-174. Morton, C. M., M. W. Chase, K. A. Kron, and S. M. Swensen. 1996. A molecular evaluation of the monophyly of the order Ebenales based upon rbcL sequence data. Systematic Botany 21:567-586. Munsell Colour Company. 1975. Munsell Soil Colour Charts. Munsell Colour, Macbeth Division of Kollmorgen Corporation, Baltimore, Maryland, USA. Norusis, M. J. 1994. SPSS: SPSS Advanced Statistics 6.1. SPSS Inc., Chicago, Illinois, USA. Okali, D. U. U. and N. D. Onyeachusim. 1991. The ground flora and rain forest regeneration at Omo Forest Reserve, Nigeria. Pages 273-283 in A. Gomez-Pompa, T. C. Whitmore, and M. Hadley, editors. Rain Forest Regeneration and Management. UNESCO/Parthenon, Paris/Carnforth, France. Oosterhoorn, M. and M. Kappelle. 2000. Vegetation structure and composition along an interior-edge-exterior gradient in a Costa Rican montane cloud forest. Forest Ecology and Management 126:291-307. Osunkoya, O. O. 1996. Light requirements for regeneration in tropical forest plants: Taxon-level and ecological attribute effects. Australian Journal of Ecology 21:429-441. Osunkoya, O. O., J. E. Ash, M. S. Hopkins, and A. W. Graham. 1994. Influence of seed size and seedling ecological attributes on shade-tolerance of rain-forest tree species in northern Queensland. Journal of Ecology 82:149-163. Page, A. L., R. H. Miller, and D. R. Keeney. 1982. Methods of Soil Analysis Part 2: Chemical and Microbiological Properties. Soil Science Society of America, Madison. Pagel, M. 1994. Detecting correlated evolution on phylogenies: A general method for the comparative analysis of discrete characters. Proceedings of the Royal Society of London Series B-Biological Sciences 255:37-45. Partridge, L. and P. H. Harvey. 1988. The ecological context of life history evolution. Science 241:1449-1455. Pennington, T. D. 1991. The Genera of Sapotaceae. Royal Botanic Garden, Kew, UK. Phillips, O. L., P. Hall, A. H. Gentry, S. A. Sawyer, and R. Vasquez. 1994. Dynamics and species richness of tropical rain forests. Proceedings of the National Academy of Sciences of the United States of America 91:2805-2809. Pinheiro, J. C. and D. M. Bates. 2000. Mixed-Effects Models in S and S-PLUS. Springer-Verlag, New York, New York, USA.

PAGE 136

124 Pitman, N. C. A., J. Terborgh, M. R. Silman, and P. Nuez. 1999. Tree species distributions in an upper Amazonian forest. Ecology 80:2651-2661. Polhill, R. M. 1952. Flora of Tropical East Africa. A. A. Balkema, Rotterdam, The Netherlands. Poorter, L. 1998. Seedling growth of Bolivian rain forest tree species in relation to light and water availability. Dissertation. Utrecht University, Utrecht, The Netherlands. Poorter, L. 1999. Growth responses of 15 rain-forest tree species to a light gradient: The relative importance of morphological and physiological traits. Functional Ecology 13:396-410. Poorter, L. 2001. Light-dependent changes in biomass allocation and their importance for growth of rain forest tree species. Functional Ecology 15:113-123. Potgieter, K. and V. A. Albert. 2001. Phylogenetic relationships within Apocynaceae s.l. based on trnL intron and trnL-F spacer sequences and propagule characters. Annals of the Missouri Botanical Garden 88:523-549. Putz, F. E. and S. Appanah. 1987. Buried seeds, newly dispersed seeds, and the dynamics of a lowland forest in Malaysia. Biotropica 19:326-333. Pyke, C. R., R. Condit, S. Aguilar, and S. Lao. 2001. Floristic composition across a climatic gradient in a Neotropical lowland forest. Journal of Vegetation Science 12:553-566. R Development Core Team. 2002. R 1.6.2. www.r-project.org. Free Software Foundation, Inc., Boston, Massachusetts, USA. Accessed 30 January 2003. Raich, J. W. and N. L. Christensen. 1989. Malaysian dipterocarp forest : Tree seedling and sapling species composition and small-scale disturbance patterns. National Geographic Research 5:348-363. Reich, P. B., C. Uhl, M. B. Walters, and D. S. Ellsworth. 1991. Leaf life-span as a determinant of leaf structure and function among 23 Amazonian tree species. Oecologia 86:16-24. Sack, L. and P. J. Grubb. 2001. Why do species of woody seedlings change rank in relative growth rate between low and high irradiance? Functional Ecology 15:145-154. Schnitzer, S. A. and W. P. Carson. 2001. Treefall gaps and the maintenance of species diversity in a tropical forest. Ecology 82:913-919. Scott, K. D., C. L. McIntyre, and J. Playford. 2000. Molecular analyses suggest a need for a significant rearrangement of Rutaceae subfamilies and a minor reassessment of species relationships within Flindersia. Plant Systematics and Evolution 223:15-27.

PAGE 137

125 Sennblad, B. and B. Bremer. 2002. Classification of Apocynaceae s.l. according to a new approach combining Linnaean and phylogenetic taxonomy. Systematic Biology 51:389-409. Sheil, D. 1999. Developing tests of successional hypotheses with size-structured populations, and an assessment using long-term data from a Ugandan rain forest. Plant Ecology 140:117-127. Skorupa, J. P. and J. M. Kasenene. 1984. Tropical forest management: Can rates of natural treefalls help guide us? Oryx 18:96-101. Smith, C. C. and S. D. Fretwell. 1974. The optimal balance between size and number of offspring. American Naturalist 108:499-506. Soltis, D. E., P. S. Soltis, M. W. Chase, M. E. Mort, D. C. Albach, M. Zanis, V. Savolainen, W. H. Hahn, S. B. Hoot, M. F. Fay, M. Axtell, S. M. Swensen, L. M. Prince, W. J. Kress, K. C. Nixon, and J. S. Farris. 2000. Angiosperm phylogeny inferred from 18S rDNA, rbcL, and atpB sequences. Botanical Journal of the Linnean Society 133:381-461. Southwood, T. R. E. 1988. Tactics, strategies and templets. Oikos 52:3-18. Spangler, R. E. and R. G. Olmstead. 1999. Phylogenetic analysis of Bignoniaceae based on the cpDNA gene sequences rbcL and ndhF. Annals of the Missouri Botanical Garden 86:33-46. Stearns, S. C. 1977. The evolution of life history traits. Annual Review of Ecology and Systematics 8:145-171. Stearns, S. C. 1980. A new view of life-history evolution. Oikos 35:266-281. Stearns, S. C. 1992. The Evolution of Life Histories. Oxford University Press, Oxford, UK. Stock, W. D., J. S. Pate, and J. Delfs. 1990. Influence of seed size and quality on seedling development under low nutrient conditions in five Australian and South African members of the Proteaceae. Journal of Ecology 78:1005-1020. Struhsaker, T. T. 1997. Ecology of an African Rain Forest: Logging in Kibale and the Conflict Between Conservation and Exploitation. University Presses of Florida, Gainesville, Florida, USA. Svenning, J. C. 1999. Microhabitat specialization in a species-rich palm community in Amazonian Ecuador. Journal of Ecology 87:55-65. Swaine, M. D. 1996. Rainfall and soil fertility as factors limiting forest species distributions in Ghana. Journal of Ecology 84:419-428. Swaine, M. D. and T. C. Whitmore. 1988. On the definition of ecological species groups in tropical rain forests. Vegetatio 75:81-86.

PAGE 138

126 Sytsma, K. J., J. Morawetz, J. C. Pires, M. Nepokroeff, E. Conti, M. Zjhra, J. C. Hall, and M. W. Chase. 2002. Urticalean Rosids: Circumscription, rosid ancestry, and phylogenetics based on rbcL, trnL-F, and ndhF sequences. American Journal of Botany 89:1531-1546. ter Braak, C. J. F. and P. Smilauer. 1999. Canoco for Windows 4.02. Centre for Biometry Wageningen, Wageningen, The Netherlands. Terborgh, J., R. B. Foster, and P. Nunez. 1996. Tropical tree communities: A test of the nonequilibrium hypothesis. Ecology 77:561-567. Uhl, C., K. Clark, N. Dezzeo, and P. Maquirino. 1988. Vegetation Dynamics in Amazonian Treefall Gaps. Ecology 69:751-763. Veenendaal, E. M. and M. D. Swaine. 1998. Limits to tree species distributions in lowland tropical rainforest. Pages 163-191 in D. M. Newbery, H. T. Prins, and N. D. Brown, editors. Dynamics of Tropical Communities. Blackwell Scientific Publications, Oxford, UK. Veenendaal, E. M., M. D. Swaine, V. K. Agyeman, D. Blay, I. K. Abebrese, and C. E. Mullins. 1996a. Differences in plant and soil water relations in and around a forest gap in West Africa during the dry season may influence seedling establishment and survival. Journal of Ecology 84:83-90. Veenendaal, E. M., M. D. Swaine, R. T. Lecha, M. F. Walsh, I. K. Abebrese, and K. Owusu-Afriyie. 1996b. Responses of West African forest tree seedlings to irradiance and soil fertility. Functional Ecology 10:501-511. Venables, W. N. and B. D. Ripley. 2002. Modern Applied Statistics with S. Springer -Verlag, New York, New York, USA. Veneklaas, E. J. and L. Poorter. 1998. Growth and carbon partitioning of tropical tree seedlings in contrasting light environments. Pages 337-361 in H. Lambers, H. Poorter, and M. M. I. Van Vuuren, editors. Inherent Variation in Plant Growth: Physiological Mechanisms and Ecological Consequences. Backhuys, Leiden, The Netherlands. Walters, M. B. and P. B. Reich. 2000. Seed size, nitrogen supply, and growth rate affect tree seedling survival in deep shade. Ecology 81:1887-1901. Webb, C. O. and D. R. Peart. 2000. Habitat associations of trees and seedlings in a Bornean rain forest. Journal of Ecology 88:464-478. Webster, G. L. 1994. Synopsis of the genera and suprageneric taxa of Euphorbiaceae. Annals of the Missouri Botanical Garden 81:33-144. Westoby, M. 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant and Soil 199:213-227.

PAGE 139

127 Westoby, M., M. Leishman, and J. Lord. 1997. Comparative ecology of seed size and dispersal. Pgs 143-162 in J. Silvertown, J. Franco, M., and J. L. Harper, Eds. Plant Life Histories: Ecology, Phylogeny and Evolution. Cambridge University Press, Cambridge, UK. Whitmore, T. C. 1989. Canopy gaps and the two major groups of forest trees. Ecology 70:536-538. Whittaker, R. H., S. A. Levin, and R. B. Root. 1973. Niche, habitat, and ecotope. American Naturalist 107:321-338. Williams-Linera, G., V. Dominguez-Gastelu, and M. E. Garcia-Zurita. 1998. Microenvironment and floristics of different edges in a fragmented tropical rainforest. Conservation Biology 12:1091-1102. Wing, L. D. and I. O. Buss. 1970. Elephants and forests. Wildlife Monographs 19:1-91. Wright, I. J. and K. Cannon. 2001. Relationships between leaf lifespan and structural defences in a low-nutrient, sclerophyll flora. Functional Ecology 15:351-359. Wright, I. J., M. Westoby, and P. B. Reich. 2002. Convergence towards higher leaf mass per area in dry and nutrient-poor habitats has different consequences for leaf life span. Journal of Ecology 90:534-543. Wright, S. J. 2002. Plant diversity in tropical forests: A review of mechanisms of species coexistence. Oecologia 130:1-14. Zanne, A. E. 1998. Expediting indigenous tree regeneration in African grasslands: Plantations and the effects of distance and isolation from seed sources. Thesis. University of Florida, Gainesville, Florida, USA.

PAGE 140

BIOGRAPHICAL SKETCH Amy Elise Zanne received her B.A. degree in biology from Dartmouth College, Hanover, New Hampshire, in June 1992. In the fall of 1995, she began graduate school in the Department of Zoology, University of Florida. She completed her M.S. in August 1998 with a thesis titled “Expediting indigenous tree regeneration in African grasslands: Plantations and the effects of distance and isolations from seed sources.” Amy continued her graduate work in the Department of Zoology, University of Florida, studying tropical tree seeds and seedlings from ecological, morphological, evolutionary, and physiological perspectives. She received her Ph.D. in May 2003 and will study tree seedling life-history characters from a physiological perspective during a Postdoctoral Fellowship at Tufts University in Boston. 128