|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 THE EVOLUTION OF LEAF PHYSICAL DEFENSE IN THE SHADE OF A NEOTROPICAL FOREST By JARED W. WESTBROOK A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DE GREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2009
2 2009 Jared W. Westbrook
3 To all those who paved the way to tree thinking
4 ACKNOWLEDGMENTS I thank Dr. Kaoru Kitajima for guidance at all stages of this project. Dr. Gordon Burleigh provided technical assistance with phylogenetic analyses and helpful comments in the preparation of this thesis. David Brasfield collected leaves, Eric Oriel measured leaf fracture toughness, and Mirna Sameniego ground the leaf samples. Dr. Karen Bjorndal generously allowed us to use her ANKOM fiber analyzer, and Alex Boulos and Kimberly Williams assisted me with the fiber analysis. I owe a debt of gratitude to Drs. S. Joseph Wright and Helene Muller Landau for hosting me during my stay in Panama. My paren ts and my partner, Maribeth Latvis have supported me through this entire process. This project has been made possible in part by a grant from the Frank Levinson Family Foundation, a supporting organization of the Silicon Valley Community Foundation. The National Science and MacArthur Foundations have supported the BCI 50ha plot censuses.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ...................................................................................................... 4 LIST OF TABLES ................................................................................................................ 6 LIST OF FIGURES .............................................................................................................. 7 LIST OF ABBREVIATIONS ................................................................................................ 8 ABSTRACT .......................................................................................................................... 9 CHAPTER 1 INTRODUCTION ........................................................................................................ 11 2 METHODS .................................................................................................................. 16 Study Site and Leaf Sampling .................................................................................... 16 Leaf Biomechanical Measurements ........................................................................... 17 Lamina Density and Thickness .................................................................................. 17 Foliar Fiber Analysis ................................................................................................... 18 Relative Growth Rate and Mortality ........................................................................... 19 Phylogeny Reconstruction .......................................................................................... 20 Phylogenetic Signal .................................................................................................... 20 Phylogenetically Independent Contrasts and Trait Correlations ............................... 21 Path Analysis .............................................................................................................. 22 Associations between Leaf Physical Defense and Demography .............................. 23 3 RESULTS .................................................................................................................... 24 Ahistorical Trait Distributions ...................................................................................... 24 Phylogenetic Signal in Leaf Physical Defense and Demography ............................. 24 Evolutionary Path Analysis of Leaf Physical Defense ............................................... 25 Relationships between Leaf Physical Defense and Demography ............................ 26 4 DISCUSSION & CONCLUSION ................................................................................ 37 APPENDI X: SUPPLEMENTARY TABLES, FIGURES, AND METHODS ...................... 42 LIST OF REFERENCES ................................................................................................... 62 BIOGRAPHICAL SKETCH ................................................................................................ 67
6 LIST OF TABLES Table page 3 -1 Phylogenetic signal leaf of physical defense traits and demography ................... 42 3 -2 Evolutionary correlations between leaf physical defense traits ............................ 42 3 -3 Associations between leaf physical defense traits and demographic rates ......... 42 A-1 List of species included in the analysis ................................................................. 42 A-2 Descriptive statistics for ahistorical traits ............................................................... 47 A-3 The effect of outlier PICs on the phylogenetic signal of RGR .............................. 54 A-4 Testing the assumptions of PIC standardization and normality with equal branch lengths. ....................................................................................................... 56 A-5 A comparison between PIC (top values) and ahistorical correlations (in parentheses) ........................................................................................................... 57 A-6 Multiple regression to te st for additive effects of the hypothesized explanatory variables on lamina fracture toughness ................................................................. 58 A-7 Multiple regression to test for additive effects of the hypothesized explanatory variables on vein fracture toughness ..................................................................... 58 A-8 Multiple regression to test for additive effects of the hypothesized explanatory variables on lamina work -to -shear ......................................................................... 59 A-9 Predicted correlation matrix (top values) and residuals (in parentheses) from path analysis. .......................................................................................................... 59 A-10 Summary of PIC loadings on principal component axes 1 5 ............................... 60 A-11 Associations between principal component axes 1 5 and demographic rates ... 61
7 LIST OF FIGURES Figure page 3 -1 Phylogenetic map of lamina fracture toughness. .................................................. 42 3 -2 Phylogenetic map of median relative growth rate. ................................................ 42 3 -3 Path model of leaf physical defense. ..................................................................... 42 3 -4 Two independent evolutionary paths to tough leaves. .......................................... 42 3 -5 Evolutionary principal component analysis of leaf physical defense. ................... 42 3 -6 Associations between leaf physical defense PC axes and demographic rates ... 42 A-1 Ahistorical trait distributions ................................................................................... 48 A-2 Phylogenetic trait maps .......................................................................................... 50
8 LIST OF ABBREVIATION S BCI Barro Colorado Island, Panama PIC (plural PICs) Phylogenetically independent contrast J Joules m meters lamWS Lamina work -to -shear (J m1) lamFT Lamina fracture toughness (J m2) veinFT Vein fracture toughness (J m2) lamD La mina density (g cm3) lamT Lamina thickness (mm) DBH Diameter and breast height of the main stem RGR Relative growth rate of the main stem (% year1) PCA Principal component analysis GRH Growth rate hypothesis of Coley, Bryant, & Chapin (1985)
9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science THE EVOLUTION OF LEAF PHYSICAL DEFENSE IN THE SHADE OF A NEOTROPICAL FOREST By Jared W Westbrook December 2009 Chair: Kaoru Kitajima Major: Botany Leaf physical defenses effectively deter a broad suite of herbivores, but species with highly fortified leaves have slower growth rates than more poorly defended competitors in canopy gaps of t ropical moist forests. The effect of these defense s on growth and survival in the deep shade, and their evolutionary history has received less attention; therefore, w e characterized the le af physical defense traits of 197 tree and shrub species, which co occur within the shaded understory of the 50 -hectare forest dynamics plot on Barro Colorado Island (BCI), Panama. Levels of leaf physical defense displayed significant phylogenetic signal; that is, the traits of closely related species were more similar th an expected by chance. Path analysis with phylogenetically independent contrasts (PICs) revealed that tough leaves have evolved independently through increases in the foliar cellulose mass fraction, increases in lamina density, and increases in leaf thick ness; however, only size -independent, material traits (lamina density, lamina fracture toughness, and vein fracture toughness) had significant negative correlations with mortality rates in the 1 10 cm diameter at breast height (DBH) class. The combination of eight leaf physical defense PIC distributions explained 18.5% of the phylogenetic variability in mortality rates, but only 4.9% of the variation in
10 median relative growth rates (RGR) in the 1 5 cm DBH class. Thus, physically fortified leaves were more strongly associated with enhanced survival than with reduced growth rates among shade tolerant species examined. These results are consistent with the Growth Rate Hypothesis, which posits that high levels of anti herbivore defense are favored by natural selection in low resource environments.
11 CHAPTER 1 INTRODUCTION For plants within the understory of a moist tropical forest, survival requires a careful balance between growth and herbivory defense (Herms & Mattson 1992, Kitajima 1994, Valladares & Niinemets 2008). Rates of herbivory are 2 to 7 fold higher in the humid tropics than in the temperate zone (Coley & Barone 1996), and leaf toughness, the amount of energy required to fracture a leaf, is an important herbivor y defense. Coley (1983) observed th at for 41 species cooccuring within canopy gaps on Barro Colorado Island (BCI), Panama, punch strength (i.e. the maximum force to puncture a leaf) and fiber content were the strongest negative correlates of rates of herbivory among 16 physical, chemical, nutritional leaf traits. Furthermore, soft, expanding leaves suffer significantly higher rates of herbivory, despite being better defended chemically than older, tougher leaves from the same species (Coley & Barone 1996). Leaf toughness may th erefore increase survival by deterring herbivory; however, the allocation of resources to produce tougher leaves may reduce the resources available for growth and reproduction (Herms & Mattson 1992) The Growth Rate Hypothesis (GRH) states that the optimal allocation to anti herbivore defense is inversely related to the availability of resources essential for plant growth (Coley et al. 1985). Where resources are abundant and rapid growth rates can be sustained, the GRH predicts that relatively low levels of defense are favored by natural selection, because competitive interactions are intense, and tissues lost to herbivores can be replaced. In resource-poor environments, higher levels of defense are selectively advantageous because tissues lost to herbivores cannot be rapidly replaced. Consequently, herbivory is more likely to cause mortality. In support of the
12 Growth Rate Hypothesis, Coley (1998) observed a strong negative relationship between relative growth rates and a linear combination of 16 anti herbivore defense traits among 41 species occurring in the resource rich environment of canopy gaps on BCI. Here, we complement this study by focusing how leaf physical defense is related to growth and survival among 197 species occurring within the shaded understory of the 50ha long -term forest dynamics plot on BCI. Multiple traits may influence leaf toughness, and these traits may have different effects on growth or mortality rates. Thus, in order to address t he relationship between physical defense and demographic rates, it is first necessary to address the more basic question of what makes a leaf tough. First, the leaves of different species vary in the relative mass fractions of cellulose, hemicellulose, and lignin, and it uncertain as to how this chemical variation in the cell wall influences leaf toughness. W e predict that the dry mass fraction of cellulose (cellulosemass) is positively related to leaf toughness, because a large amount of energy is consumed in breaking the covalent bonds between the glucan monomers that makeup cellulose microfibrils (Carpita & Gibeaut 1993). Some authors have assumed that lign in also contributes to lamina and vein toughness (e.g. Coley 1983, Choong 1996); however, it could also be argued that lignin depos ition, by imparting rigidity to cell walls, makes tissues more brittle (Lucas et al. 2000). H emicellulose affects the organization and space between cellulose mi crofibrils (Cosgrove 2000), but its effect on leaf toughness is unknown The leaves of different species also vary in density, or dry mass per unit volume of fresh tissue, and the effect of tissue density on fracture toughness is not straightforward. Lucas et al. (2000) hypothesized that tissue density is negatively related to fracture
13 toughness because dense tissue has little capacity for plastic intracellular collapse. They reasoned that plastic deformation consumes energy as a crack propagates, and this increases toughness. On the other hand, tissue density may be positively related to toughness, because denser tissue is associated with thicker cell walls that are more resistant to fracture (Choong et al. 1992). Lamina thickness is also an important d eterminant of leaf fracture mechanics (Hanley et al. 2007). Materials scientists describe the fracture mechanics of solids in terms of material properties (size independent) and structural properties (size dependent). Fracture toughness, a material prop erty, is defined as the work (force*displacement) to fracture a tissue normalized to fracture area. Work -to -shear, a structural property, is work to fracture normalized to fracture length (Lucas et al. 2000). Lamina work -to -shear (lamWS) is considered a s tructural property because it depends, in part, on the thickness of leaf, whereas lamina fracture toughness (lamFT) is a material property because it does not depend directly on thickness (although thickness may be correlated with fracture toughness). Fi nally, leaves are a composite of different tissues (i.e. epidermis, palisade layer, mesophyll, and vascular bundles). The properties of these individual tissues synergistically influence the fracture mechanics of whole leaves (Choong 1996). Here, we spe cifically focus on the effect of vein fracture toughness on lamina fracture toughness Choong (1996) observed that veins of Castanopsis fissa (Fagaceae) were at least one order of magnitude tougher than the estimated toughness of vein free lamina. Furthermore, the fracture toughness of primary, secondary, and higher order
14 veins were of similar magnitude. Hence, we expect lam ina fracture toughness to depend strongly on vein fracture toughness In summary, we hypothesize the following functional relationships among leaf physical defense variables: 1 Lamina fracture toughness (lamFT) is a function of the chemical makeup of cell walls (hemicellulose, cellulose, and/or lignin per unit leaf mass), lamina density (lamD), and vein fracture t oughness (veinFT). 2 Vein fracture toughness is a function of cell well chemistry and lamD. 3 Lamina work -to -shear (lamWS) is the product of lamFT and lamina thickness (lamT). We evaluate these hypotheses with phylogenetically independent contrasts (PICs) in a path analytical framework. The use of PICs allows us to determine if these traits have repeatedly evolved in a correlated fashion along all branches of a phylogeny (Felsenstein 1985). In the context of leaf physical defense, incorporating PICs allows us to determine whether tough leaves, in both the material and structural sense, evolved through correlated changes in the traits that influence leaf toughness, or whether these traits evolved independently of one another. Returning now to the question of how leaf physical defense influences growth and survival in the forest understory, individual leaf physical defense traits may have differential effects on demographic rates. The strength of the relationships between leaf biomechanical traits (veinFT, l amFT, and lamW S) and their hypothesized predictor variables may vary, which could influence how effectively these traits deter herbivory and enhance survival. In addition, leaf physical defense traits may affect demography independently of their effects on leaf biomechanical properties. For instance, very few insects and no vertebrates can digest the (1 4) glycosidic bonds of cellulose without
15 microbial gut symbionts (Sanson 2006). Cellulosemass may therefore deter herbivory through its effects on dig estibility independently of its effect on leaf toughness. Finally, few studies have examined whether leaf structural traits versus material traits differ in their relationships to growth and survival. Kitajima & Poorter (2009) found that leaf material properties, but not leaf thickness were correlated with leaf lifespan, growth, and survival of 19 tree species from Bolivia. We address the following questions: 1 Do closely related species tend to resemble each other with respect to leaf physical defense traits? 2 How does the chemical makeup of cell walls, tissue density, and leaf thickness affect leaf toughness at both the material and structural level? 3 Do the predictors of leaf toughness evolve in concert or independently of one another? 4 How is leaf ph ysical defense related to growth and mortality rates of coexisting shade -tolerant tree and shrub species? 5 Do the traits that influence leaf fracture mechanics have differential effects on demographic rates?
16 CHAPTER 2 METHODS Study Site and Leaf Sampl ing The leaves of 197 tree and shrub species were sampled from the shaded understory of the 50 -ha long -term forest dynamics plot on Barro Colorado Island (BCI), Panama (9 10N, 79 51 W). The seasonally moist lowland tropical forest of BCI receives ~2600 mm of precipitation per year with a distinct dry season between January and May (Croat 1978). The majority of the plot area (48 of 50 hectares) has not been cleared by humans for at least 2500 years (Piperno 1990) and more than 95% of the ground area remains under canopy cover at any given time. The 50 ha plot has been censused every five years since 1985, and the first census occurred in 1982. Abundance, stem growth, and mortality rates of individuals > 1 cm diameter at br east height (DBH) have been recorded for ~300 species through 5 census intervals (Condit et al. 2006). The present analysis includes 197 of these species because they occur within the shaded understory, they are represented in a phylogeny of BCI (Kress et al. 2009), and demographic data are available for at least 5 individuals of each species (Condit et al. 2006). Two mature leaves from the most exposed portion of the canopy were collected from the 6 smallest individuals of each species growing in a forest understory environment that receives no direct light or direct light from only one side (canopy classes 1 2 according to Dawkins & Field 1978). Freshly collected leaves were placed in ziplock bags with moist filter paper and stored in a cooler. Leaf biom echanical measurements were made within 2 24 hours of collection. We limited our interspecies leaf trait comparisons to shaded individuals to maximize the detection of inherent
17 variation between species, and minimize variation in leaf traits du e to differences in light environment (Dominy et al. 2003). Leaf Biomechanical Measurements Work to fracture lamina and veins was measured with the shearing method developed by Lucas (1990), using a portable universal tester that is described in Darvell et al. (1996). Work to fracture (units: Joules) was measured by integrating force to fracture the lamina by displacement of the scissor blades, and then subtracting the friction from the scissor blades passing one another. Measurements were made on smal l rectangular pieces of lamina (8x16mm) that included midvein on one of the longedges of the rectangle. First, the work to fracture the lamina was measured along a cut path that minimized the fracture of secondary and tertiary veins. After the lamina was cut, the central vein was cut in a perpendicular direction. Lamina fracture toughness (units: J m2) was calculated by dividing work to fracture by the cross sectional area of the cut surface (lamina thickness*fracture length, assuming a rectangular cros s section). Vein fracture toughness was calculated similarly, except veins were assumed to be circular in cross section. Lamina thickness (excluding secondary veins) and midvein diameter were measured with a dial thickness gauge (SM112, Teclock, Japan). Lamina work -to -shear was calculated by multiplying lamina fracture toughness by thickness (units: J m1). Duplicate biomechanical measurements were made for each species on leaves from different individuals, and the values were averaged. Lamina Densit y and Thickness Lamina density was measured as average dry mass per unit fresh volume (g cm3). Circular disks 12.9 mm in diameter were cut with a cork borer from sections of lamina that excluded the primary vein. Lamina thickness was measure d on fresh
18 tissue as described above. The leaf disks were then dried at 65 C for at least three days before the dry mass was determined. Lamina density measurements were made on 2 12 disks per species (one disk per leaf), and the values were averaged. W e were unable to obtain reliable measurements of vein density because of difficulties in measuring vein volume. Foliar Fiber Analysis The leaves that were used for biomechanical and density measurements were also used for fiber analysis. Prior to fiber analysis, petioles were removed and leaves from conspecifics were pooled, ground, and dried at 105 C. Fiber mass fractions were determined through sequential extractions with the ANKOM 200 fiber analyzer (Macedon, NY) and the method of Vansoest et al. (19 91). Briefly, 500 mg of dried and ground leaf sample from each species was sealed in a chemical resistant bag. Nonpolar extracts (pectins, lipids, sugars, starch, and soluble proteins) were washed away with neutral detergent to obtain an estimate of the crude fiber (hemicellulose + cellulose + lignin) mass fraction. Hemicellulose and structural proteins were extracted in acid detergent, and cellulose w as removed in 70% H2SO4. Finally, samples were incinerated in a muffle furnace at 500 C to determ ine mineral content. Hemicellulose, cellulose, and lignin mass fractions (mg g1) were calculated by subtraction after correcting for mineral ash mass. Duplicate fiber measurements were performed on 50 randomly selected species. All values were within 20% of their original values, with an average measurement error of 7%.
19 Relative Growth Rate and Mortality Mortality rates and stem relative growth rates (RGR) were obtained from Condit et al. (2006). Leaf traits were correlated with mortality rates from th e 10 99mm diameter at breast height (DBH) size class, and relative growth rates from the 10 49mm DBH size class. These size classes corresponded well with the size class of individuals sampled for leaf traits, as 97.5% percent of the sampled individuals were 10 99mm DBH, and 69% were 10 49mm DBH. Condit et al. (2006) calculated mortality rates (m) and relative growth rates (RGR) as follows: m ln( N ) ln( S ) t 100 (2 -1) RGR ln( dbhfinal) ln( dbhinitial) t 100 (2 -2) Where N = number of individuals initia lly present, S= number of survivors, dbh = diameter at breast height, and t = time between censuses in years. Median relative growth rates and mortality rates (both in units of % yr1) were averaged from the three census intervals that spanned 1990 to 200 5. Data from the 1982 and 1985 censuses were excluded because DBH was measured only to the nearest 5mm, which introduced major error in the calculation of RGR. In later censuses diameter was measured to the nearest 0.1mm. The number of individuals per species (N) from which to estimate median RGR varied from 5 individuals to 84,689 individuals with a median sample size of 164 individuals per species ( Table A -1 ). Sample sizes from which to estimate mortality rates varied from 9 to 108,385 with a median sample size of 362 ( Table A 1 ). Species with less than 5 individuals were excluded. Mortality rates of zero were observed for three species, thus to make mortality rates amenable to log transformation,
20 1/N was added to the average mortality rates of all spe cies. Demographic data were compiled without respect to the light environment of the individuals (Condit et al. 2006). Phylogeny Reconstruction A phylogeny of the tree and shrub species from the 50hectare plot on BCI was obtained from Kress et al. ( 2009). The phylogeny was based on the plastid markers rbcL, matK and trnH -psbA and was reconstructed with maximum parsimony using PAUP v.4.0 (Swofford 2002). A single optimal topology was selected to perform further analyses. The phylogeny was pruned, using the APE package in R (Paradis et al. 2004), to include only the species for which trait data were available. Phylogenetic Signal Both u ntransformed and log10-transformed trait values were randomly permuted across the tips of the phylogeny 10,000 times using Phylocom AOTF (Webb et al. 2008). If closely related species tend to resemble each other with respect to a certain trait (i.e. there is phylogenetic signal), then the observed variance of the PIC distribution for that trait is expected to be less than variance of the PICs calculated after randomization (Blomberg et al. 2003). The one tailed pvalue of the phylogenetic signal for each trait was calculated as the number of randomizations where the observed variance in the PICs was less than the randomized PIC variance divided by 10,000. As an additional measure of phylogenetic signal, Blombergs K -statistic was calculated from untransformed trait values with the picante package in R (Kembell et al. 2009). The K -statistic measures how closely the observed phylogenetic pattern for each trait corresponds to the trait patterns expected from a Brownian motion model of trait diversification. K values less than 1 indicate that species trait values are less similar
21 than expected by Brownian motion, whereas K values greater than 1 indicate that traits are more similar than expected from Brownian motion (Blomberg et al. 2003). Phylogenetically Independent Contrasts and Trait Correlations Phylogenetically independent contrasts (PICs) were calc ulated for each trait according to the method of Felsenstein (1985) with the Phylocom Analysis of Traits Module (Webb et al. 2008). With two polytomies in the phylogenetic tree, 193 independent contrasts were obtained from a phylogeny of 197 species. Ide ally, PICs for each trait should be standardized by expected rates of evolutionary change so that the PICs share a common variance, and standard probability tables can be used to determine the significance of correlated evolution between traits (Felsenstei n 1985, Diaz -Uria rte et al. 1996). Five different branch length distributions (DNA base pair substitutions, log10-DNA substitutions, time, log10-time, and equal branch lengths ) in combination with both untrans formed and log10-transformed trai t values were tested with the method of Garland et al. (1992) to determine which combination best standardized the PICs Log10-transformation of the trait values with equal branch lengths resulted in the best standardiz ation ; however, the PIC distributions of 7/11 traits remained i nadequately standardized (Table A 3). Therefore, the sign -test was used to determine the significance (p < 0.05) of pair wise correlations because this non -parametric method is robust to outliers and inadequate PIC standardization. Statistical simulations show that the power of the signtest approaches the power of parametric methods with sample sizes greater than ~150 (Ackerly 2000). We also report Pearson correlation coefficients to convey the relative magnitude of pairwise trait correlations. Correlations were calculated through the origin because the direction of subtraction is arbitrary in the calculation of PICs (Garland et al. 1992).
22 Path Analysis Predictor variables of leaf biomechanical traits were included in the path model if they had an ad ditive effect on their response variable(s) in the presence of other predictor variables. Because the PICs were inadequately standardized, a non parametric multiple regression procedure was performed in JMP v. 7 to assess the significance of additive effec ts. First, predictor variables were regressed against their hypothesized response variables with both positive and negative values of each PIC included in the analysis. Second, the leverage residuals ( i.e., the effect of individual predictor variables on their response variable, while controlling for other predictor variables) were extracted. Third, the data were filtered to include only positive PICs for each predictor variable so that one set of contrasts (NPIC=193) was subsequently evaluated. Finally, the number of positive PICs of the response variable w as tallied, and the additive effects of individual predictor variables were assessed with the sign test against a null expectation that 50% of the response variable PICs would be positive or negativ e. To assess the significance of pairwise correlations, one variable was designated as the predictor variable, the other variable was designated as the response variable (the designations make no difference), and significance was assessed with the sig n test as described above. SAS PROC CALIS was used to perform path analysis on the Pearson correlations between the PICs of the relevant log10-transformed traits (Hatcher 1996), and a 2 goodness of -fit test was used to assess the significance of the path model. The null hypothesis in this goodness of -fit test is that the model accurately predicts the observed trait correlations. Thus, pvalues greater than 0.1 fail to reject the model (Shipley 2000).
23 A ssociations b etween Leaf Physical Defense a nd Demogr aphy First, pairwise correlations between individual leaf physical defense traits and species demographic rates (RGR and mortality) were assessed with the sign test. Next, a principal component analysis (PCA) was performed on eight leaf physical defense PIC distributions (hemicellulose, cellulose, lignin, lamW S, lamFT, veinFT, lamD, and lamT), and the factor scores fr om the PC axes were plotted against RGR and mortality rates respectively. The PC axes were computed through the origin by incl uding both positive and negative values of each PIC in the principal component analysis (Ackerly & Donoghue 1998). The significance of the correlations between the PC axes and demographic rates w as assessed with the sign-test as described above. The adv antages of using PCA in this context are 1) it can be used to assess patterns of multiple -correlation among leaf physical defense traits, and 2) the power of the PC axes to predict demographic rates (i.e. R2values) are independent and additive.
24 CHAPTER 3 RESULTS Ahistorical Trait Distributions Exploration of the untransformed, ahistorical data revealed that demographic rates were more variable than leaf physical defense traits among the 197 tree and shrub species examined (Table A 2). Leaf physical d efense traits varied between 1.8 fold (Cellulosemass) and 12.1 fold (lamWS). Median RGR varied 26-fold (1.950%), and average mortality rates varied 7177 -fold (0.00427%). High variability in demographic rates is statistically attributable to large posi tive skew in both the median RGR and mortality rate distributions (Figure A 1 ). Phylogenetic Signal i n Leaf Physical Defense and Demography Phylogenetic signal was highly significant for all untransformed leaf physical defense traits and mortality rates but was not significant for median RGR (Table 3 -1 ) due to three pairs of sister species with widely divergent growth rates (Fig. 3 -2 ). Removal of the faster growing species of these three pairs ( Apeiba tibourbou Turpinia occidentalis, and Zanthoxylum ekmanii) would result in significant phylogenetic signal in RGR (p = 0.018; Table A -3) as t he rest of the species included in the analysis had similar growth rates (Figure 3 -2 ). Among species, both the median and upper q uartile of RGR were within the same order of magnitude (5.6 and 7.4 % yr1 respectively), while the maximum RGR (51.2% yr1) was one order of magnitude greater than the median. Log10 transformation of the trait values resulted in significant phylogenetic signal for all traits under consideration (Table 3-1). The K -statistic was less than one for all untransformed traits, indicating that the species were less similar than expected by a Brownian motion model of trait
25 diversification (Table 3 -1 ). The largest K-statistic was observed for mortality rates (K = 0.376), while smallest K -statistic was observed for RGR (K = 0.2058). The low K statistic for RGR is consistent with the relatively weak phylogenetic signal detected by the randomization test. Evolutiona ry Path Analysis of Leaf Physical Defense In multiple regression analyses, cellulosemass, veinFT, and lamD had significant positive additive effects on lamFT, whereas hemicellulosemass ,had a negative effect on lamFT, and the effect on ligninmass on lamF T was not significant. Together, hemicellulosemass, cellulosemass, veinFT, and lamD explained 56.7% of the variation in lamFT (model F4,189 = 61.98, P < 0.0001; Table A 6 ). Cellulosemass and lamD had significant positive effec ts on veinFT, while the effects of hemicellulosemass and ligninmass were non-significant. Cellulosemass together with lamD explained 37.8% of the variation in veinFT (F2,191= 58.12, P < 0.0001; Table A -7). Lamina fracture toughness and lamT had approximately equal positive effects on lamWS, and both variables explained 99.7% of the variation in lamWS (F2,191 = 28518, P < 0.0001; Table A -8). None of exogenous predictor variables of leaf fracture mechanics (hemicellulosemass ,, cellulosemass, lamD, and lamT) were significantly correlated with any other predictor variable (Table 3 2 Figure 3 -4 ), which indicates that these traits evolved independently of one another. In addition, the lack of correlation between lamina thickness and lamina fracture toughness (Table 3-2) indicates that thick leaves are not necessarily tough at the material level. The multiple regressions described above were combined into a single path model (Figure 3 -3 ), which was supported by the observed correlation matrix (2 7df = 6.05, p = 0.53, p values greater than 0.1 fail to reject the model).
26 Relationships b etween Leaf Physical Defense and Demography Leaf physical defense was more strongly associated with reduced mortality rates than with reduced median RGR among the shade tolerant species examined (Table 3 -3 Fig. 3 6 ). A multiple regression with all eight leaf physical defense trait PIC distributions (hemicellulosemass cellulosemass, ligninmass, lamWS, lamFT, veinFT, lamD, and lamT) explained 18.5% of the variation in mortali ty (F8,185 = 5.26, P < 0.0001), and 4.9% of the variation in median RGR (F8,185 = 1.897, P = 0.3037). Lamina fracture toughness, veinFT, and lamD were the only leaf physical defense traits that were negatively correlated with mortality rates (Table 3 -3 ). No leaf physical defense trait was significantly correlated with RGR. The fact that leaf physical defense was more strongly associated with mortality than with RGR was not unique to the PIC distributions. The log10transformed, ahistorical distributions of all eight leaf physical defenses traits explained 25.6% of the variation in mortality rates (model F8,185 = 8.12, P < 0.0001), and 7.7% of the variation in median RGR (F8,185 = 1.97, P = 0. 0527 ). In the principle component analysis (PCA) on the leaf physical defense traits, the first two principal components captured 57% of the variation in eight leaf physical defense trait PIC distributions Seven of 8 leaf physical defense traits (with the exception of hemicellulosemass) loaded positively on the f irst PC axis, which accounted for 38.8% of the total variation in leaf physical defense. Lamina density, veinFT, lamFT, and hemicellulosemass loaded positively on PC2, while cellulose, lignin, lamT, and lamWS were negatively associated with the second PC which accounted for 18.2% of the total variation (Fig. 3 -5 ). A more complete breakdown the PC loadings can be found in Table A 9.
27 Significant negative correlations were observed between log10 mortality PICs and the factor scores from the first two princi pal components of leaf physical defense (Fig. 3 6 ). This result is consistent with the pairwise correlations between individual leaf physical defense traits and mortality as lamD, veinFT, and lamFT were the only variables that loaded positively on PC axes 1 and 2, and were negatively correlated with mortality rates (Fig. 3 5 Table 3 4 ). Together, PC axes 1 and 2 explained 16% of the variation in mortality (Fig. 3 6 ), while these axes explained only 2% of the variation in RGR (Fig. 3 -6 ).
28 Table 3 1 Phylogenetic signal leaf of physical defense traits and demography Trait Phylogenetic signal untransformed trait s Phylogenetic signa l log 10 trait s K statistic untransformed traits Mortality < 0.0001 < 0.0001 0.3760 RGR 0.14 0.009 0.2058 Hemicellulose 0.001 < 0.0001 0.2167 Cellulose 0.004 0.005 0.2274 Lignin < 0.0001 < 0.0001 0.3053 LamWS < 0.0001 < 0.0001 0.2452 L am FT < 0.0001 < 0.0001 0.2630 VeinFT < 0.0001 < 0.0001 0.3123 LamD 0.0005 < 0.0001 0.2743 LamT < 0.0001 < 0.0001 0.2700 Phylogenetic signal was assessed by comparing the variance of PICs calculated after randomizing trait values across the tips of the phylogeny to the observed PIC variance. Pvalues less than 0.05 (column s 2 and 3) indicate that closely related species have more similar trait values than expected by chance. K -statistic values less than 1 (column 4 ) indicate that closely related species are less similar than expected under a Brownian motion model of trait diversification.
29 Figure 3 1 Phylogenetic map of lamina fracture toughness. Closely related species tend to have similar lamina fracture toughness values, and clusters of distantly related species have evolved similar fracture toughness values. The inset displays the ahistorical distribution of lam ina fracture toughness.
30 Figure 3 2 Phylogenetic map of median relative growth rate. The weak phylogenetic signal for median RGR arises from three closely related sister species pairs with widely divergent growth rates (rapidly growing outliers are labeled). The inset displays the ahistorical distribution of median RGR. Note the large positive skew.
31 Table 3 2 Evolutionary correlations between leaf physical defense traits Crude fiber Hemi cellulose Cellulose Lignin LamWS LamFT VeinFT LamD Hemicellu lose 0.379 (0.0015) Cellulose 0.687 (<0.0001) 0.121 (0.8856) Lignin 0.583 (<0.0001) 0.311 (0.0015) 0.178 (0.0305) LamWS 0.284 (0.0015) 0.272 (0.0024) 0.396 (<0.0001) 0.275 (0.0024) LamFT 0.249 (0.0039) 0 .184 (0.1131) 0.371 (<0.0001) 0.174 (0.1131) 0.788 (<0.0001) VeinFT 0.182 (0.0610) 0.012 (0.5648) 0.355 (<0.0001) 0.039 (0.6659) 0.497 (<0.0001) 0.594 (<0.0001) LamD 0.032 (0.7735) 0.096 (1.0000) 0.130 (0.7735) 0.156 (0.0305) 0.329 (0.061) 0.541 (<0.0001) 0.452 (0.0005) LamT 0.176 (0.3878) 0.228 (0.0024) 0.219 (0.0838) 0.239 (0.1131) 0.721 (<0.0001) 0.145 (0.1498) 0.135 (0.0305) 0.069 (1.0000) Pearsons correlations between the PICs of log10 trait values are displayed in the top of each cell, and two-tailed pvalues calculated from the non -parametric signtest are displayed below in parentheses.
32 Figure 3 3 Path model of leaf physical defense. Path coefficients (standardized partial regressio n coefficients) next to the arrows were derived from evolutionary correlations (Table 3 -2 ). All paths depicted by singled headed arrows were significantly different than zero (p < 0.05, Tables A 6, A 7, and A -8), while correlations between independent var iables (hemicellulosemass, cellulosemass, lamina density, and lamina thickness) were non-significant (Table 3 2 ). Ligninmass was left out of the model because it did not have significant additive effects on either lamFT or veinFT. The model explained 99.7% of the variation of lamina work -to -shear, 56.8% of the variation of lamina fracture toughness, and 37.8% of the variation of vein fracture toughness. The path model provided a good fit to the observed correlation matrix ( 2 7df = 6.05, p = 0.53, pvalu es > 0.1 fail to reject the model).
33 Figure 3 4 Two independent evolutionary paths to tough leaves. Both lamina density and cellulosemass were positively correlated with lamina fracture toughness, but lamina density and cellulosemass were uncorrelated with each other. The significance of trait correlation was determined by the sign test using one set of contrasts. B oth positive and negative values of each PIC are displayed
34 Table 3 3 Associations between leaf physical defense traits and demographic rates Median RGR Mortality Direction of correlation p value Direction of correlation p value Crude fiber Not significant 0.06 05 N ot significant 0.1950 Hemicellulose Not signif i cant 0.2494 Not significant 0.3136 Cellulose Not signific ant 0.6659 Not significant 0.5648 Lignin Not significant 0.0838 Not sig nificant 0.7735 LamWS Not significant 0.8856 Not sig nificant 0.4717 LamF T Not significant 0.5648 Negative 0.0024 VeinFT Not significant 0.7735 Negative 0.0015 LamD Not significant 0.2494 Negative < 0.0001 LamT Not significant 0.3878 Not significant 0.3878 Significance was assessed by the sign test.
35 Figure 3 5 Evolu tionary principal component analysis of leaf physical defense. The first two principal components captured 57% of the variation in eight leaf physical defense PIC distributions (PC1 = 38.8%, PC2 = 18.2%). Abbreviations: lamD lamina density, lamFT lami na fracture toughness, lamT lamina thickness, Ws lamina work -to -shear, and veinFT vein fracture toughness.
3 6 Figure 3 6 Associations between leaf physical defense PC axes and demographic rates. A) log10 mortality PICs v. leaf physical defense PC1. B) log10 mortality PICs v. leaf physical defense PC2. C) log10 RGR PICs v. leaf physical defense PC1. D) log10 RGR PICs v. leaf physical defense PC2. Both positive and negative values of each PIC are displayed; however, only one set of contrasts was used to determine the significance of the relationships between leaf physical defense PC axes and demographic rates.
37 CHAPTER 4 DISC USSION & CONCLUSION Significant phylogenetic signal in all leaf physical defense traits examined demonstrates t hat these traits are relatively slowly evolving (Table 31) Phylogenetic signal may contribute to niche conservatism among closely related spec ies to the extent that leaf physical defense influences ecological performance (e.g. growth, survival, and reproduction) along resource gradients ( Webb et al. 2002). In this regard, leaf physical defenses may confer a fitness advantage to trees and shrubs persisting in the shade, as the defense traits examined here were more strongly correlated with reduced mortality rates than with reduced growth rates (Table 3-3, Fig. 36) Leaf physical defense improves survival in the deep shade of moist tropical for ests by increasing resistance to herbivory and prolonging leaf lifespan, thus contributing to the maintenance of a positive balance (Coley 1983, Coley 1988, Kitajima 1994, Alvarez Clare & Kitajima 2007). Furthermore, t he result s presented here and those o f Kitajima & Poorter (2009) support the notion that leaf material traits (LamD, lamFT, veinFT) are better predictors of rates of survival than structural traits (lamT and lamWS). Material level leaf physical defense traits may effectively deter chewing ins ects, the dominant herbivore guild in tropical forests (Coley & Barone 1996), because insect mouthparts are small relative to the size of most leaves (Lucas et al. 2004, Sanson 2006). Lamina density the trait with the strongest negative association with m ortality rates (Table 3 3), may be an especially effect ive deterrent against chewing insects because it augments resistance to crack propagation (i.e. fracture toughness of lamina and veins ; Fig. 3 -3), and may also increase resistance to crack formation b y increasing lamina hardness (i.e. resistance to deformation; Lucas et al. 2000). Tough v eins were also correlated
38 with increased surv ival in the forest understory (Table 3 3) and may contribute to a positive carbon balance by supplying water and nutrien ts to regions of intact lamina remaining after an herbivore attack (Delaney & Higley 2006, Delaney 2008). Enhanced resistance to herbivory may not be only mechanism by which material level leaf physical defense s enhance survival in the shade. Both LamD and cellulosemass had comparable positive effects on lamFT and veinFT (Fig. 3 3 Fig. 3 -4 ); however only lamD was significantly associated with reduced mortality rates (Table 3 3) The lack of correlation between cellulosemass and mortality was surprising given that Coley (1983) observed a strong negative correlation between foliar fiber content and rates of herbivory among many of the species examined here, and cellulose reduces digestibility of leaves (Sanson 2006). In addition to increasing resistance to herbivory, lamina density may confer greater resistance to pathogens as has been observed for wood densi ty (Ausperger 1984). Furthermore, dense tissue has greater bending stiffness (Lucas et al. 2000), which maintains horizontal leaf alignment to maxim ize light capture in the shade (Read & Stokes 2006). The lack of correlation between leaf physical defense and growth observed here, contrasts with the results of Coley (1988) who observed a strong negative relationship between herbivore defense traits and RGR among 41 tree and shrub species from BCI, most of which were also included in this study. One explanation for these contrasting findings is Coley (1988) measured defense characteristics and RGR among species occurring within canopy gaps while t h e species median RGR analyzed here more likely reflect growth rates in the shaded understory. Although, species growth rate distributions were compiled without respect to light environment (Condit et al. 2006), the
39 median RGR for stems 1 5 cm DBH analyzed here are likely to reflect growth rates in the shade as more than 95% of the ground area within the 50 -ha plot is under forest cover at any given time point (Smith et al. 1992, Hubbell et al. 1999), and all species in the present analy sis persist in the shade Differences between species growth rates are greater in treefall gaps than in the shaded understory (Kobe 1999). Species with the inherent potential for fast growth approach their maximum growth rates in the high light of can opy gaps, whereas shade tolerant species may show much smaller changes in RGR in high light (Kobe 1999, Valladares & Niinemets 2008). Although, inherently faster growing species allocate less to defense, and consequently suffer higher rates of herbivory than slower growing, well -defended species (Coley & Barone, 1996), individuals growing in high light have more resources to replace tissue loss to herbivores (Coley et al. 1985). Hence, allocation of photosynthetic resources to defense is expected to be n egatively correlated with RGR in canopy gaps, as was observed by Coley (1988). In the shaded understory, on the other hand, differences in RGR between species are small because growth rates of all species are constrained by low light availability (P oorter 1999) Competitive interactions are also less intense in the understory, because stem density is lower than in canopy gaps (Denslow 1987) Although a single well -defended leaf requires more resources to produce than a poorly defended leaf, each leaf is retained for longer periods of time (Coley 1983, Coley 1988, Reich et al. 1992). Species with well -defended leaves may, therefore, produce less leaf mass over their lifecycle (Westoby et al. 2002). Increased leaf lifepan however does not completely compensate for the allocation cost s of producing highly defended leaves.
40 In a common garden experiment conducted in nutrient -poor sandy soils o f the Amazon basin specialists of these soils had higher levels of foliar herbivory defense s (including physical defenses) and grew more slowly than less -defended specialists of more nutrient -rich clay soil s when herbivores were excluded Without herbivo re exclusion, however sandy soil specialist s outperformed clay soil specialists in both growth and survival (Fine et al. 2004, Fine et al 2006). H ence, in resource limited environments, the allocation cost of defense is balanced by the cumulative resource savings of both increased leaf lifespan and decreased rates of herbivory which may explain the lack of correlation between leaf physical defense and growth observed here. Together t he results presented here, and those of Coley (1988) b oth may be explained by the Growth Rate Hypothesis, which predicts that the tradeoff between defense and growth is inversely proportional to resource availability (Coley et al. 1985). Although material level leaf physical defense traits were significantly correlated with reduced mortality rates, 81.5% of the variance in mortality and 95% percent of the variance in median RGR was unexplained by the leaf traits examined here. Undoubtedly, more variation in demographic rates will be ex plained as additional functional traits are analyzed together with leaf physical defense. For example, wood density and percentage of recruits in gaps explained 54% of the variation in sapling RGR of the 73 most abundant tree species from the 50-hectare pl ot on BCI (Wright et al. 2003). Although we did not detect significant relationships between leaf physical defense and growth, it is possible that these traits tradeoff with allocation to reproduction and carbohydrate storage (Stamp 2003, Agrawal et al. 2008). In addition,
41 some of this unexplained variation may be attributable to stochastic factors that affect species demographic rates independently of functional traits (Hubbell 20 01). Conclusions. Leaves of the woody tropical understory plants on BCI attain leaf toughness through increases in cellulose mass fraction, lamina density, or leaf thickness. The size -independent leaf material properties (lamFT, veinFT, and la mD) were negatively correlated with mortality rates, suggesting the importance of leaf physical defenses for survival in the shade ; h owever, no leaf properties were correlated with median RGR. These results are consistent with Growth Rate Hypothesis which states that optimal levels of anti herbivore defense are inversely related to resource availability.
42 APPENDIX SUPPLEMENTARY TABLES, FIGURES, AND METHODS Table A -1 List of species included in the analysis Family Species Lifeform N mort N RGR Euphorbiaceae Acalypha diversifolia SHRUB 1827 776 Euphorbiaceae Acalypha macrostachya SHRUB 123 44 Euphorbiaceae Adelia triloba SHRUB 383 124 Verbenaceae Aegiphila panamensis SHRUB 164 46 Rubiaceae Alibertia edulis SHRUB 1107 83 2 Sapindaceae Allophylus psilospermus UNDERSTORY 310 148 Rubiaceae Alseis blackiana UNDERSTORY 21523 15890 Rubiaceae Amaioua corymbosa SHRUB 74 49 Annonaceae Anaxagorea panamensis SHRUB 2032 1863 Fabaceae Andira inermis UNDERSTORY 803 498 Annonaceae Annona acuminata UNDERSTORY 1606 1197 Annonaceae Annona spraguei MIDSTORY 380 138 Malvaceae Apeiba membranacea TREE 194 31 Malvaceae Apeiba tibourbou TREE 29 8 Acanthaceae Aphelandra sinclairiana SHRUB 20 11 Myrsinaceae Ardisia bartlettii SHRUB 10 6 Myrsinaceae Ardisia guianensis SHRUB 57 40 Myrsinaceae Ardisia standleyana SHRUB 268 199 Apocynaceae Aspidosperma spruceanum TREE 1278 969 Anacardiaceae Astronium graveolens TREE 110 61 Lauraceae Beilschmiedia tovarensis TREE 6780 4557 Moraceae Brosim um alicastrum TREE 2161 1284 Moraceae Brosimum guianense TREE 15 8 Clusiaceae Calophyllum longifolium TREE 2856 2126 Brassicaceae Capparis frondosa SHRUB 9823 7800 Salicaceae Casearia aculeata UNDERSTORY 1284 809 Salicaceae Casearia arborea MIDSTORY 1 71 63 Salicaceae Casearia commersoniana UNDERSTORY 54 36 Salicaceae Casearia guianensis SHRUB 49 22 Salicaceae Casearia sylvestris SHRUB 376 177 Ulmaceae Celtis schippii TREE 294 126 Myrtaceae Chamguava schippii UNDERSTORY 983 784 Clusiaceae Chrysoch lamys eclipes UNDERSTORY 1215 849 Sapotaceae Chrysophyllum argenteum TREE 1815 1358 Sapotaceae Chrysophyllum cainito TREE 298 223 Lauraceae Cinnamomum triplinerve TREE 170 65 Polygonaceae Coccoloba coronata UNDERSTORY 380 194 Polygonaceae Coccoloba ma nzinellensis MIDSTORY 1288 843 Melastomataceae Conostegia cinnamomea SHRUB 487 209 Boraginaceae Cordia bicolor MIDSTORY 1835 785 Boraginaceae Cordia lasiocalyx UNDERSTORY 3415 1595 Rubiaceae Coussarea curvigemmia UNDERSTORY 5911 4258 Sapindaceae Cupan ia cinerea MIDSTORY 28 11
43 Table A 1. Continued Family Species Lifeform N mort N RGR Sapindaceae Cupania latifolia UNDERSTORY 104 59 Sapindaceae Cupania rufescens UNDERSTORY 290 196 Sapindaceae Cupania seemannii UNDERSTORY 3392 2335 Araliaceae Dend ropanax arboreus TREE 95 26 Annonaceae Desmopsis panamensis UNDERSTORY 35146 27700 Ebenaceae Diospyros artanthifolia UNDERSTORY 192 142 Fabaceae Dipteryx oleifera TREE 48 28 Erythroxylaceae Erythroxylum macrophyllum SHRUB 768 508 Erythroxylaceae Eryth roxylum panamense SHRUB 325 263 Myrtaceae Eugenia coloradoensis UNDERSTORY 2012 1337 Myrtaceae Eugenia galalonensis MIDSTORY 4427 3722 Myrtaceae Eugenia oerstediana UNDERSTORY 5965 3844 Rubiaceae Faramea occidentalis UNDERSTORY 75564 50160 Moraceae Fi cus insipida TREE 24 13 Moraceae Ficus maxima MIDSTORY 14 5 Moraceae Ficus tonduzii UNDERSTORY 33 5 Clusiaceae Garcinia intermedia TREE 12702 10459 Clusiaceae Garcinia madruno TREE 1233 915 Rubiaceae Genipa americana TREE 161 100 Nyctaginaceae Guapir a standleyana TREE 243 92 Meliaceae Guarea grandifolia TREE 160 94 Meliaceae Guarea guidonia MIDSTORY 4595 3078 Annonaceae Guatteria dumetorum TREE 3021 1442 Malvaceae Guazuma ulmifolia MIDSTORY 53 27 Rubiaceae Guettarda foliacea UNDERSTORY 774 439 L ecythidaceae Gustavia superba UNDERSTORY 434 71 Rubiaceae Hamelia axillaris SHRUB 313 157 Malvaceae Hampea appendiculata UNDERSTORY 42 13 Olacaceae Heisteria acuminata TREE 303 159 Olacaceae Heisteria concinna MIDSTORY 2058 1405 Malvaceae Herrania pur purea SHRUB 1584 1215 Euphorbiaceae Hieronyma alchorneoides TREE 143 67 Chrysobalanaceae Hirtella americana MIDSTORY 104 57 Chrysobalanaceae Hirtella triandra MIDSTORY 12808 9133 Euphorbiaceae Hura crepitans TREE 35 9 Violaceae Hybanthus prunifolius S HRUB 108385 84689 Fabaceae Inga acuminata UNDERSTORY 928 651 Fabaceae Inga goldmanii MIDSTORY 1056 625 Fabaceae Inga marginata UNDERSTORY 1365 589 Fabaceae Inga nobilis UNDERSTORY 1965 1157 Fabaceae Inga pezizifera TREE 394 211 Fabaceae Inga punctata UNDERSTORY 39 24
44 Table A -1. Continued Family Species Lifeform N mort N RGR Fabaceae Inga spectabilis TREE 20 8 Fabaceae Inga umbellifera UNDERSTORY 2786 1766 Bignoniaceae Jacaranda copaia TREE 132 29 Salicaceae Lacistema aggregatum SHRUB 4634 3381 Apocynaceae Lacmella panamensis TREE 154 72 Salicaceae Laetia procera TREE 46 13 Salicaceae Laetia thamnia UNDERSTORY 1435 872 Chrysobalanaceae Licania hypoleuca MIDSTORY 359 256 Chrysobalanaceae Licania platypus TREE 877 658 Salicaceae Lindackeria l aurina UNDERSTORY 56 12 Fabaceae Lonchocarpus heptaphyllus TREE 2108 1291 Malvaceae Luehea seemannii TREE 375 130 Rubiaceae Macrocnemum roseum UNDERSTORY 233 145 Malpighiaceae Malpighia romeroana SHRUB 155 111 Moraceae Maquira guianensis MIDSTORY 3950 2611 Clusiaceae Marila laxiflora MIDSTORY 36 27 Melastomataceae Miconia affinis SHRUB 1179 660 Melastomataceae Miconia argentea UNDERSTORY 2096 698 Melastomataceae Miconia elata UNDERSTORY 64 32 Melastomataceae Miconia hondurensis UNDERSTORY 120 80 Melastomataceae Miconia impetiolaris SHRUB 41 27 Melastomataceae Miconia nervosa SHRUB 863 370 Rubiaceae Morinda seibertii SHRUB 9 8 Annonaceae Mosannona garwoodii MIDSTORY 1160 858 Melastomataceae Mouriri myrtilloides SHRUB 21210 17134 Myrtaceae Myrc ia gatunensis UNDERSTORY 140 101 Fabaceae Myrospermum frutescens LIANA 28 14 Lauraceae Nectandra cissiflora TREE 676 320 Lauraceae Nectandra lineata MIDSTORY 286 147 Lauraceae Nectandra purpurea MIDSTORY 236 129 Nyctaginaceae Neea amplifolia SHRUB 195 143 Lauraceae Ocotea cernua MIDSTORY 777 396 Lauraceae Ocotea oblonga TREE 410 169 Lauraceae Ocotea puberula TREE 468 200 Fabaceae Ormosia coccinea TREE 223 157 Fabaceae Ormosia macrocalyx TREE 315 250 Ochnaceae Ouratea lucens SHRUB 3657 3075 Rubia ceae Palicourea guianensis SHRUB 3392 783 Rubiaceae Pentagonia macrophylla SHRUB 1131 781 Moraceae Perebea xanthochyma UNDERSTORY 709 480 Simaroubaceae Picramnia latifolia UNDERSTORY 3211 2178 Piperaceae Piper aequale SHRUB 162 74 Piperaceae Piper arb oreum SHRUB 120 52
45 Table A -1. Continued Family Species Lifeform N mort N RGR Piperaceae Piper carrilloanum SHRUB 9 5 Piperaceae Piper colonense SHRUB 103 32 Piperaceae Piper cordulatum SHRUB 2257 407 Piperaceae Piper perlasense SHRUB 132 52 Piper aceae Piper reticulatum SHRUB 378 156 Fabaceae Platymiscium pinnatum TREE 305 160 Fabaceae Platypodium elegans TREE 297 135 Rubiaceae Posoqueria latifolia SHRUB 186 124 Moraceae Poulsenia armata TREE 3070 802 Urticaceae Pourouma bicolor TREE 106 69 S apotaceae Pouteria reticulata TREE 4185 2809 Sapotaceae Pouteria stipitata TREE 115 78 Burseraceae Protium costaricense TREE 2134 1262 Burseraceae Protium panamense TREE 8810 6663 Malvaceae Pseudobombax septenatum TREE 27 12 Myrtaceae Psidium friedric hsthalianum UNDERSTORY 142 111 Rubiaceae Psychotria acuminata SHRUB 41 12 Rubiaceae Psychotria chagrensis SHRUB 30 11 Rubiaceae Psychotria deflexa SHRUB 105 33 Rubiaceae Psychotria grandis SHRUB 186 104 Rubiaceae Psychotria horizontalis SHRUB 14690 97 82 Rubiaceae Psychotria limonensis SHRUB 64 27 Rubiaceae Psychotria marginata SHRUB 1871 1154 Fabaceae Pterocarpus rohrii TREE 4538 3123 Simaroubaceae Quassia amara UNDERSTORY 409 205 Rubiaceae Randia armata UNDERSTORY 2520 1509 Violaceae Rinorea syl vatica SHRUB 7341 6077 Euphorbiaceae Sapium glandulosum MIDSTORY 64 17 Fabaceae Schizolobium parahyba TREE 44 18 Fabaceae Senna dariensis SHRUB 249 105 Monimiaceae Siparuna guianensis SHRUB 53 9 Monimiaceae Siparuna pauciflora SHRUB 838 377 Solanacea e Solanum hayesii SHRUB 129 24 Moraceae Sorocea affinis SHRUB 9172 6259 Anacardiaceae Spondias mombin TREE 186 69 Anacardiaceae Spondias radlkoferi TREE 501 226 Malvaceae Sterculia apetala TREE 101 46 Myrsinaceae Stylogyne turbacensis UNDERSTORY 2147 1536 Fabaceae Swartzia simplex_var.grandiflora UNDERSTORY 7197 5938 Fabaceae Swartzia simplex_var.ochnacea UNDERSTORY 8254 6304 Clusiaceae Symphonia globulifera TREE 422 296 Bignoniaceae Tabebuia guayacan TREE 130 86
46 Table A -1. Continued Fa mily Species Lifeform N mort N RGR Bignoniaceae Tabebuia rosea TREE 620 365 Apocynaceae Tabernaemontana arborea TREE 3472 2512 Fabaceae Tachigali versicolor TREE 8236 5723 Sapindaceae Talisia croatii UNDERSTORY 1935 1454 Combretaceae Terminalia amazon ia TREE 74 30 Combretaceae Terminalia oblonga UNDERSTORY 149 81 Apocynaceae Thevetia ahouai SHRUB 248 122 Burseraceae Trattinnickia aspera TREE 87 12 Acanthaceae Trichanthera gigantea UNDERSTORY 20 6 Meliaceae Trichilia pallida MIDSTORY 1424 899 Meli aceae Trichilia tuberculata TREE 33056 23695 Polygonaceae Triplaris cumingiana MIDSTORY 437 133 Moraceae Trophis caucana UNDERSTORY 564 161 Moraceae Trophis racemosa MIDSTORY 825 570 Staphyleaceae Turpinia occidentalis UNDERSTORY 67 8 Annonaceae Unono psis pittieri MIDSTORY 1667 895 Myristicaceae Virola multiflora MIDSTORY 78 28 Myristicaceae Virola nobilis TREE 189 71 Myristicaceae Virola sebifera TREE 3745 2206 Clusiaceae Vismia baccifera SHRUB 200 95 Vochysiaceae Vochysia ferruginea TREE 42 20 Salicaceae Xylosma oligandra UNDERSTORY 362 181 Rutaceae Zanthoxylum ekmanii TREE 241 64 Rutaceae Zanthoxylum juniperinum TREE 306 164 Rutaceae Zanthoxylum panamense MIDSTORY 361 137 Salicaceae Zuelania guidonia TREE 89 50 Demographic data from the fo ur censuses spanning 1990-2005 were obtained from Condit et al. (2006). Lifeforms were classified according to expert opinion (J. Wright, personal communication)
47 Table A -2. Descriptive statistics for ahistorical traits Trait Units Mean Median Min Max Fo ld variation RGR % yr1 7.51 5.57 1.90 51.18 26.9 Mortality % yr1 4.38 2.88 0.004 28.71 7177.5 Crude fibermass mg g1 538.03 545.10 297.17 789.06 2.7 Hemicellulosemass mg g1 160.25 152.43 52.57 388. 96 7.4 Cellulosemass mg g1 198.83 194.90 59.84 348.39 1.8 Ligninmass mg g1 166.07 160.50 49.35 337.59 6.8 LamWS J m1 0.060 0.056 0.014 0.170 12.1 Lam FT J m2 398.16 391.03 143.15 761.54 5. 3 VeinFT J m2 2958.74 2809.20 951.6 0 8546.40 9 .0 LamD g cm3 0.281 0.275 0.133 0.490 3.7 LamT mm 0.149 0.143 0.065 0.334 5.1
48 Figure A1 Ahistorical trait distributions
49 Figure A 1. Continued.
50 Figure A 2. Phylogenetic trait maps
51 Figure A 2. Continued
52 Figure A 2. Continued
53 Figure A 2. Continued
54 Table A -3. The effect of outlier PICs on the phylogenetic signal of RGR A comparison of the relative growth rates and abundances of species that share a common ancestor, yet have widely divergent growth rates. The column furthest to th e right indicates the p value from the randomization test of phylogenetic signal after stepwise removal of the rapidly growing species from each outlying node. PIC Outlier rank Fast RGR species Slow RGR species or node Fast RGR Slow RGR N fast RGR N slow RGR P hylo signal p value 1 Apeiba tibourbou Apeiba membracea 51.1 10.3 8 31 0.1261 2 Turpinia occidentalis Picramnia latifolia 40.3 4.2 8 2178 0.0590 3 Xanthoxylum ekmanii Xanthoxylum panamense 38.2 6.0 146 351 0.0173 4 Vochysia ferruginea M yrtaceae 34.0 5.0 20 13669 0.0093 5 Solanum hayesii Lamiales 35.5 8.6 24 1086 0.0007
55 Phylogenetically independent contrast standardization A contrast arising from a single node i is standardized by dividing the difference in trait values of the direct descendents of node i by the square root of the sum the branch lengths that connect no de i to its descendents (Felsenstein 1985). A test to determine if a branch length distribution ( e.g., molecular branch lengths, branch lengths proportional to time, etc.) adequately standardizes the contrasts is to plot the absolute values of the sta ndardized PICs against their standardization terms, and determine if the least squares regression slope differs significantly from zero (Garland et al. 1992). A regression slope indistinguishable from zero indicates that branch lengths are proportional to evolutionary change, and thus the PICs are adequately standardized. Branch lengths in units of time (millions of years) were obtained with the Penalized Likelihood method (Sanderson 2002 ) as implemented in the program R8S (Sanderson 2003) using six different fossil constraints (minimum ages of clades) and a two fixed fossil calibration point (Angiosperms = 170 MYA, Eudicots = 125 MYA).
56 Table A -4. Testing the assumptions of PIC standardization and normality with equal branch lengths. Log 10 trait PICs standardized? PICs normally distributed? Median RGR N (p=0.02) N, (W=0.92, p<10 4 ) Mortality Y (p=0.21) N, (W=0.96, p<10 4 ) Crude fiber Y (p=0.055) Y, (W=0.99, p=0.74) Hemicellulose N (p=0.0005) N (W=0.98, p<10 4 ) Cellulose N (p=0.0003) N (W=0.99, p=0.01) Lignin N (p=0.02) Y (W=0.994, p=0.11) Work to shear Y (p=0.086) Y (W=0.994, p=0.10) Lamina toughness N (p=0.002) Y ( W=0.995, p=0.28) Vein toughness N (p=0.028) Y (W=0.993, p=0.07) Lamina density N (p=0.005) Y (W=0.996, p=0.34) Lamina thickness Y (p=0.10) Y (W=0.998, p=0.95) Traits were log10 transformed before the PICs were calculated. The pvalues in the second c olumn indicate whether the regression slope s between the absolute value of the PICs and their associated standardization terms were significantly different from zero. The ShapiroWilk test of normality was performed on PIC distributions that included bot h positive and negative values of each contrast
57 Table A -5. A comparison between PIC (top values) and ahistorical correlations (in parentheses) Crude fiber Hemicell Cellu lose Lignin LamWS LamFT VeinFT LamD Crude fiber 1 Hemicell 0.382 (0.431) 1 Cellulose 0.688 (0.725) 0.119 (0.186) 1 Lignin 0.581 (0.659) 0.308 ( 0.189) 0.177 (0.286) 1 LamWS 0.283 (0.356) 0.271 ( 0.224) 0.396 (0.4 40) 0.274 (0.376) 1 LamFT 0.248 (0.372) 0.183 ( 0.128) 0.371 (0.497) 0.172 (0.278) 0.788 (0.795) 1 VeinFT 0.183 (0.298) 0.013 ( 0.010) 0.354 (0.479) 0.04 (0.136) 0.499 (0.580) 0.595 (0.686) 1 LamD 0.029 (0.108) 0.094 ( 0.173) 0. 131 (0.017) 0.153 (0.242) 0.329 (0.422) 0.540 (0.618) 0.452 (0.528) 1 LamT 0.176 (0.152) 0.228 ( 0.220) 0.218 (0.146) 0.239 (0.291) 0.720 (0.772) 0.145 (0.158) 0.138 (0.167) 0.070 ( 0.015) All traits were log10-transformed before the calcula tion of PICs and ahistorical correlations.
58 Table A -6.Multiple regression to test for additive effects of the hypothesized explanatory variables on lamina fracture toughness Explanatory variable Standardized partial regression coefficient St andard error Direction of correlation Number of positive lamFT residuals out of 193 2 tailed p value Hemicellulosemass 0.1913 0.0513 Negative 77 0.0061 Cellulosemass 0.3671 0.0574 Positive 127 < 0.0001 Ligninmass 0.0331 0.0531 Not sig. 99 0.7735 Vein fracture toughness 0.2582 0.0616 Positive 119 0.0015 Lamina density 0.4596 0.0588 Positive 118 0.0024 The full model explained 56.8% of the variation in lamFT, and was significant (F=49 .51, P < 0.0001). Removing lignin from the model marginally changed the model R2 value to 0.567 and increased the model significance (F = 61.98, P < 0.0001). Zeros indicate non-significant correlations. Table A -7. Multiple regression to test for ad ditive effects of the hypothesized explanatory variables on vein fracture toughness Explanatory variable Standardized partial regression coefficient Standard error Direction of correlation Number of positive veinFT residuals out of 193 2 tailed p value Hemicellulosemass 0.0592 0.0606 Not sig. 91 0.4717 Cellulosemass 0.4562 0.0592 Positive 132 < 0.0001 Ligninmass 0.1379 0.0619 Not sig. 88 0.2494 Lamina density 0.5256 0.0581 Positive 132 < 0.0001 The full model explained 39.4% of the variation in veinFT, and was significant (F = 30.78, P < 0.0001). Removing hemicellulosemass and ligninmass reduced the model R2value to 0.378, but increased the model significance (F = 58.12, P < 0.000 1).
59 Table A -8. Multiple regression to test for additive effects of the hypothesized explanatory variables on lamina work to -shear Explanatory variable Standardized partial regression coefficient Standard error Direction of correla tion Number of positive lamWS residuals out of 193 2 tailed p value Lamina thickness 0.6189 0.0042 Positive 187 < 0.0001 Lamina fracture toughness 0.6988 0.0042 Positive 187 < 0.0001 The model explained 99.7% of the variation in lamWs, and was highly significant (F = 28518, P < 0.0001). Table A -9. Predicted correlation matrix (top values) and residuals (in parentheses) from path analysis. Hemi cellulose Cellulose LamWS LamFT VeinFT LamD LamT Hemi cellu lose 1 (0) Cellulose 0.1210 ( 0.0048) 1 (0) LamWS 0.2678 ( 0.0037) 0.3947 (0.0013) 0.9635 (0.0365) LamFT 0.1803 ( 0.0143) 0.3710 (0) 0.7614 (0.0266) 0.9987 (0.0013) VeinFT 0.0023 ( 0 ) 0.3550 (0) 0.4483 (0.0487) 0.5914 (0.0026) 1 (0) LamD 0.0960 ( 0 ) 0.1300 (0) 0.3349 ( 0.0059) 0.5410 (0) 0.4520 (0) 1 (0) LamT 0.228 ( 0 ) 0.2190 (0) 0.6921 (0.0289) 0.1036 (0.0414) 0.0572 (0.0778) 0.069 (0) 1 (0) Residuals are the differences between observed PIC correlations and co rrelations predicted by the path model.
60 Table A -10. Summary of PIC loadings on principal component axes 1 5 PC1 (38.8%) PC2 (18.2%) PC3 (15.9%) PC4 (10.6%) PC5 (7.9% ) Hemicellulose 0.3086 0.4556 0.6148 0.045 0.5488 Cellulose 0.4817 0.077 0.6983 0.3922 0.2237 Lignin 0.3877 0.4408 0.3006 0.6779 0.2937 LamWS 0.9293 0.1847 0.0802 0.2388 0.0736 LamFT 0.8562 0.3093 0.0386 0.0182 0.1493 VeinFT 0.6969 0.4485 0 .1376 0.0208 0.1032 LamD 0.5093 0.556 0.5098 0.007 0.2754 LamT 0.5332 0.6384 0.1727 0.4138 0.2895 The percentages in the top row are the proportion of variance captured by each principal component axis. The first five principal components explain 91.4% of the variation in leaf physical defense PIC distributions.
61 Table A -11. Associations between principal component axes 1 5 and demographic rates Median RGR Mortality Direction of correlation nPos p value Direction of correlat ion nPos p value PC1 Not sig. 98 0.8856 Negative 77 0.0061 PC2 Not sig. 94 0.7735 Negative 81 0.0305 PC3 Not sig, 98 0.8856 Not sig. 103 0.3878 PC4 Positive 111 0.0436 Not sig. 86 0.1498 PC5 Negative 79 0.0142 Not sig. 84 0.0838 PC6 Not sig. 90 0. 3878 Not sig. 99 0.7735 PC7 Not sig. 92 0.5648 Not sig. 93 0.6659 PC8 Not sig. 98 0.8856 Not sig. 99 0.7735 Significance of the relationships between PC axes and demographic rates was assessed by the sign test. nPos stands for the number of positive mortality or growth PICs predicted by the PC axes out of a maximum of 193. Although the sign test detected significant relationships between median RGR versus PC axes 4 and 5 together these axes explained only 1.5% of the variance in median RGR.
62 LIST O F REFERENCES Ackerly DD. 2000. Taxon sampling, correlated evolution, and independent contrasts. Evolution 54: 1480 92 Ackerly DD, Donoghue MJ. 1998. Leaf size, sapling allometry, and Corner's rules: Phylogeny and correlated evolution in maples (Acer). Amer ican Naturalist 152: 767-91 Agrawal AA, Fishbein M. 2008. Phylogenetic escalation and decline of plant defense strategies. Proceedings of the National Academy of Sciences of the United States of America 105: 1005760 Alvarez -Clare S, Kitajima K. 2007. Phys ical defence traits enhance seedling survival of neotropical tree species. Functional Ecology 21: 1044 54 Augspurger CK, Kelly CK. 1984. Pathogen mortality of tropical tree seedlings experimental studies of the effects of dispersal distance, seedling density, and light conditions. Oecologia 61: 211 -7 Blomberg SP, Garland T, Ives AR. 2003. Testing for phylogenetic signal in comparative data: Behavioral traits are more labile. Evolution 57: 71745 Carpita NC, Gibeaut DM. 1993. Structural models of primary -c ell walls in flowering plants consistency of molecular -structure with the physical properties of the walls during growth. Plant Journal 3: 1-30 Choong MF. 1996. What makes a leaf tough and how this affects the pattern of Castanopsis fissa leaf consumpti on by caterpillars. Functional Ecology 10: 66874 Choong MF, Lucas PW, Ong JSY, Pereira B, Tan HTW, Turner IM. 1992. Leaf fracture toughness and sclerophylly their correlations and ecological implications. New Phytologist 121: 597610 Coley PD. 1983. Her bivory and defensive characteristics of tree species in a lowland tropical forest. Ecological Monographs 53: 209 -33 Coley PD. 1988. Effects of plant -growth rate and leaf lifetime on the amount and type of anti -herbivore defense. Oecologia 74: 531-6 Coley P D, Barone JA. 1996. Herbivory and plant defenses in tropical forests. Annual Review of Ecology and Systematics 27: 30535 Coley PD, Bryant JP, Chapin FS. 1985. Resource availability and plant antiherbivore defense. Science 230: 895 -9 Condit R, Ashton P, Bunyavejchewin S, Dattaraja HS, Davies S, et al. 2006. The importance of demographic niches to tree diversity. Science 313: 98101
63 Cosgrove DJ. 2000. Expansive growth of plant cell walls. Plant Physiology and Biochemistry 38: 10924 Croat TB. 1978. Flora of Barro Colorado Island. Stanford University Press. Stanford, CA. 956 pp. Darvell BW, Lee PKD, Yuen TDB, Lucas PW. 1996. A portable fracture toughness tester for biological materials. Measurement Science & Technology 7: 954 62 Dawkins HC, Field, D.R. 1978. A long -term surveillance system for British woodland vegetation. In Occasional Paper Number 1. Oxford, UK: Oxford University Press Delaney KJ. 2008. Injured and uninjured leaf photosynthetic responses after mechanical injury on Nerium oleander leaves, and D anaus plexippus herbivory on Asclepias curassavica leaves. Plant Ecology 199: 187200 Delaney KJ, Higley LG. 2006. An insect countermeasure impacts plant physiology: midrib vein cutting, defoliation and leaf photosynthesis. Plant Cell and Environment 29: 1 24558 Denslow JS. 1987. Tropical rain-forest gaps and tree species diversity. Annual Review of Ecology and Systematics 18: 431-51 Diaz -Uriarte R, Garland T. 1996. Testing hypotheses of correlated evolution using phylogenetically independent contrasts: Sensitivity to deviations from Brownian motion. Systematic Biology 45: 27 47 Dominy NJ, Lucas PW, Wright SJ. 2003. Mechanics and chemistry of rain forest leaves: canopy and understorey compared. Journal of Experimental Botany 390: 200714 Felsenstein J. 1985. Phylogenies and the Comparative Method. American Naturalist 125: 1-15 Fine PVA, Mesones, I, Coley PD. 2004. Herbivores promote habitat specialization by trees in Amazonian forests. Science 305: 663 -65 Fine PVA, Miller ZJ, Mesones I, Irazuzta S, Appel HH, Stevens HH, Saaksjarvi I, Schultz JC, Coley PD. 2006. The growth -defense tradeoff and habitat specialization by plants in Amazonian forests. Ecology 87 : S15062 Garland T, Harvey PH, Ives AR. 1992. Procedures for the analysis of comparative data using phy logenetically independent contrasts. Systematic Biology 41: 1832 Hanley ME, Lamont BB, Fairbanks MM, Rafferty CM. 2007. Plant structural traits and their role in anti -herbivore defence. Perspectives in Plant Ecology Evolution and Systematics 8: 157 -78
64 Hat cher L. 1996. Using SAS PROC CALIS for path analysis: an introduction. Structural Equation Modeling 3: 17692 Herms DA, Mattson WJ. 1992. The dilemma of plants to grow or defend. Quarterly Review of Biology 67: 283335 Hubbell SP. 2001. The Unified Neutr al Theory of Biodiversity and Biogeography Princeton University Press. Princeton, NJ. 448 pp. Hubbell SP, Foster RB, O'Brien ST, Harms KE, Condit R, et al. 1999. Light -gap disturbances, recruitment limitation, and tree diversity in a neotropical forest. Science 283: 554 -7 Kembel SW, Ackerly DD, Blomberg SP, Cowan MR, Helmus H, et al. 2009. picante: R tools for integrating phylogenies and ecology. Kitajima K. 1994. Relative importance of photosynthetic traits and allocation patterns as correlates of seedli ng shade tolerance of 13 tropical trees. Oecologia 98: 419 -28 Kitajima K, Poorter, L. 2009. Tissue level leaf toughness, but not lamina thickness predict leaf lifespan, shade tolerance, and ontogenetic shifts of tropical tree species. New Phytologist Accep ted Kobe RK. 1999. Light gradient partitioning among tropical tree species through differential seedling mortality and growth. Ecology 80: 187201 Koricheva J. 2002. Meta analysis of sources of variation in fitness costs of plant antiherbivore defenses. Ec ology 83: 176-90 Kress WJ, Erickson DL, F.A. J, Swenson NG, Perez R, et al. 2009. Plant DNA barcodes and community phylogeny of a tropical forest dynamics plot in Panama. Proceedings of the National Academy of Sciences of the United States of America Luca s PW, Pereira B. 1990. Estimation of the Fracture -Toughness of Leaves. Functional Ecology 4: 819 -22 Lucas PW, Turner IM, Dominy NJ, Yamashita N. 2000. Mechanical defences to herbivory. Annals of Botany 86: 91320 Paradis E, Claude J, Strimmer K. 2004. APE: Analyses of Phylogenetics and Evolution in R language. Bioinformatics 20: 28990 Piperno DR. 1990. Fitolitos, arquelogia y cambios prehistoricos de la vegetacion en un lote de cincuenta hectareas de la isla de Barro Colorado. In Ecologia de un bosque trop ical, ed. EG Leigh, AS Rand, DM Windsor, pp. 153-6. Smithsonian Institution Press. Washington, D.C.
65 Poorter L. 1999. Growth responses of 15 rain-forest tree species to a light gradient: the relative importance of morphological and physiological traits. Fu nctional Ecology 13: 396410 Read J, Stokes A. 2006. Plant biomechanics in an ecological context. American Journal of Botany 93: 1546-65 Reich PB, Walters MB, Ellsworth DS. 1992. Leaf life -span in relation to leaf, plant, and stand characteristics among di verse ecosystems. Ecological Monographs 62: 36592 Sanderson MJ. 2002. Estimating absolute rates of molecular evolution and divergence times: A penalized likelihood approach. Molecular Biology and Evolution 19: 1019 Sanderson MJ. 2003. r8s: inferring absolute rates of molecular evolution and divergence times in the absence of a molecular clock. Bioinformatics 19: 3012 Sanson G. 2006. The biomechanics of browsing and grazing. American Journal of Botany 93: 1531-45 Shipley B. 2000. Cause and Correlation in Biolo gy Cambridge University Press. Cambridge, UK. 317 pp. Smith AP, Hogan KP, Idol JR. 1992. Spatial and temporal patterns of light and canopy structure in a lowland tropical moist forest. Biotropica 24: 50311 Stamp N. 2003. Out of the quagmire of plant defe nse hypotheses. Quarterly Review of Biology 78: 2355 Swofford DL. 2002. PAUP*: Phylogenetic Analysis Using Parsimony (and Other Methods). Sinauer, Sunderland Valladares F, Niinemets U. 2008. Shade tolerance, a key plant feature of complex nature and consequences. Annual Review of Ecology Evolution and Systematics 39: 23757 Vansoest PJ, Robertson JB, Lewis BA. 1991. Methods for dietary fiber, neutral detergent fiber, and nonstarch polysaccharides i n relation to animal nutrition. Journal of Dairy Science 74: 358397 Webb CO, Ackerly DD, McPeek MA, Donoghue MJ. 2002. Phylogenies and community ecology. Annual Review of Ecology and Systematics 33: 475-505 Webb CO, Ackerly DD, Kembel SW. 2008. Phylocom: software for the analysis of phylogenetic community structure and trait evolution. Bioinformatics 24: 2098-100
66 Westoby M, Falster DS, Moles AT, Vesk PA, Wright IJ. 2002. Plant ecological strategies: Some leading dimensions of variation between species. Ann ual Review of Ecology and Systematics 33: 125-59 Wright SJ, Muller -Landau HC, Condit R, Hubbell SP. 2003. Gapdependent recruitment, realized vital rates, and size distributions of tropical trees. Ecology 84: 3174 -85
67 BIOGRAPHICAL SKETCH Jared Westbrook s pent his format ive years Grand Rapids, MI. He graduated from East K entwood High School in 2000, and went on to study at the University of Michigan (U of M Ann Arbor). A summer at the U of M Biological Station sparked his interest in plants. He later worked for Dr. David Ellsworth studying the physiological responses of trees to elevated CO2, and spent one summer monitoring the spread of Beech Bark Disease throughout Michigan under Dr. John Witter. He earned his Bachelor of Science (cum laude) in environmen tal s cience in 2004. Upon graduation, he joined Americorps to teach vocational forestry to high school students in rural Washington for one year. After completing his Americorps tenure, he moved to Boston, MA where he worked as a research assistant at H arvard University studying the biomechanics of fern spore dispersal under Jacques Dumais. In 2007, he walked the Appalachian Trail from Georgia to Maine with his partner, Maribeth Latvis. He started graduate school in the Botany Department at University of Florida in f all 2007 under Dr. Kaoru Kitajima, and received his Master of Science in plant e cology and evolution in the fall of 2009. He continues to work toward his Ph.D. in the Plant Molecular and Cellular Biology program at UF. He hopes to use his diverse training obtained at UF to work at nexus between ecology, evolutionary biology, and genomics.