Citation
Biomechanical Properties of Tropical Tree Seedlings as a Functional Correlate of Shade Tolerance

Material Information

Title:
Biomechanical Properties of Tropical Tree Seedlings as a Functional Correlate of Shade Tolerance
Creator:
ALVAREZ-CLARE, SILVIA ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Ecology ( jstor )
Forests ( jstor )
Leaves ( jstor )
Mortality ( jstor )
Seedlings ( jstor )
Species ( jstor )
Toughness ( jstor )
Trees ( jstor )
Understory ( jstor )
Vertebrates ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Silvia Alvarez-Clare. Permission granted to University of Florida to digitize and display 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:
436098658 ( OCLC )

Downloads

This item has the following downloads:

alvarez_s ( .pdf )

alvarez_s_Page_09.txt

alvarez_s_Page_26.txt

alvarez_s_Page_29.txt

alvarez_s_Page_67.txt

alvarez_s_Page_75.txt

alvarez_s_Page_80.txt

alvarez_s_Page_55.txt

alvarez_s_Page_87.txt

alvarez_s_Page_12.txt

alvarez_s_Page_61.txt

alvarez_s_Page_60.txt

alvarez_s_Page_47.txt

alvarez_s_Page_62.txt

alvarez_s_Page_56.txt

alvarez_s_Page_40.txt

alvarez_s_Page_43.txt

alvarez_s_Page_07.txt

alvarez_s_Page_05.txt

alvarez_s_Page_38.txt

alvarez_s_Page_42.txt

alvarez_s_Page_01.txt

alvarez_s_Page_57.txt

alvarez_s_Page_52.txt

alvarez_s_Page_68.txt

alvarez_s_Page_20.txt

alvarez_s_Page_17.txt

alvarez_s_Page_86.txt

alvarez_s_Page_65.txt

alvarez_s_Page_85.txt

alvarez_s_Page_78.txt

alvarez_s_Page_02.txt

alvarez_s_Page_10.txt

alvarez_s_Page_06.txt

alvarez_s_Page_33.txt

alvarez_s_Page_32.txt

alvarez_s_Page_25.txt

alvarez_s_Page_50.txt

alvarez_s_Page_23.txt

alvarez_s_Page_03.txt

alvarez_s_Page_74.txt

alvarez_s_Page_22.txt

alvarez_s_Page_24.txt

alvarez_s_Page_64.txt

alvarez_s_Page_63.txt

alvarez_s_Page_83.txt

alvarez_s_Page_82.txt

alvarez_s_Page_39.txt

alvarez_s_Page_36.txt

alvarez_s_Page_04.txt

alvarez_s_Page_21.txt

alvarez_s_Page_44.txt

alvarez_s_Page_37.txt

alvarez_s_Page_49.txt

alvarez_s_Page_91.txt

alvarez_s_Page_53.txt

alvarez_s_pdf.txt

alvarez_s_Page_14.txt

alvarez_s_Page_59.txt

alvarez_s_Page_28.txt

alvarez_s_Page_27.txt

alvarez_s_Page_30.txt

alvarez_s_Page_18.txt

alvarez_s_Page_16.txt

alvarez_s_Page_13.txt

alvarez_s_Page_95.txt

alvarez_s_Page_48.txt

alvarez_s_Page_77.txt

alvarez_s_Page_66.txt

alvarez_s_Page_69.txt

alvarez_s_Page_84.txt

alvarez_s_Page_93.txt

alvarez_s_Page_76.txt

alvarez_s_Page_88.txt

alvarez_s_Page_71.txt

alvarez_s_Page_31.txt

alvarez_s_Page_11.txt

alvarez_s_Page_90.txt

alvarez_s_Page_58.txt

alvarez_s_Page_94.txt

alvarez_s_Page_51.txt

alvarez_s_Page_41.txt

alvarez_s_Page_34.txt

alvarez_s_Page_45.txt

alvarez_s_Page_46.txt

alvarez_s_Page_73.txt

alvarez_s_Page_15.txt

alvarez_s_Page_35.txt

alvarez_s_Page_72.txt

alvarez_s_Page_92.txt

alvarez_s_Page_79.txt

alvarez_s_Page_54.txt

alvarez_s_Page_08.txt

alvarez_s_Page_81.txt

alvarez_s_Page_19.txt

alvarez_s_Page_70.txt

alvarez_s_Page_89.txt


Full Text












BIOMECHANICAL PROPERTIES OF TROPICAL TREE SEEDLINGS AS A
FUNCTIONAL CORRELATE OF SHADE TOLERANCE
















By

SILVIA ALVAREZ-CLARE


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2005

































Copyright 2005

by

Silvia Alvarez-Clare




























To my parents, who let me fly;
and to Abuelita Betty, who gave me the wings.















ACKNOWLEDGMENTS

I would like to thank my advisor Kaoru Kitajima for academic and financial

support but also for her kind, yet strict guidance while I took my initial steps as a

scientist. Kaoru and her daughter, Sachi, treated me as family when I first arrived in

Gainesville. I thank my other committee members (Emilio Bruna, Michael Daniels, and

Jack Putz) for providing support and valuable comments that improved my research and

finally my thesis. I also thank Gerardo Avalos, who was the first to believe in me as a

scientist. I extend thanks to the Department of Botany staff, who offered logistic support.

Also, the graduate students and professors in the Botany Plant Ecology Group helped

develop my ideas and scientific thinking, through stimulating discussion sessions.

I thank the Smithsonian Tropical Research Institute for financial and institutional

support. Staff and researchers at Barro Colorado Island assisted in numerous ways during

my fieldwork. Roberto Cordero shared his biomechanical knowledge and was always

willing to help. Sarah Tarrant, Liza Coward, Jeffrey Hubbard, Sebastian Bemal, and

Marta Vargas provided invaluable field assistance. I especially thank Jeff for sharing

many "eco-challenges" with me through the forest, and Marta for all the late-night leaf-

cutting sessions that also resulted in a great friendship. Momoka Yao provided great help

with fiber analysis and data entry.

I thank my "Tico" friends, who have been my family away from home for the past

3 yrs. I also thank Jenny Schaffer, Carla Stefanescu, Cat Cardelus, Eddie Watkins, Sarah

Bray, and all the rest of my Gainesville friends, who always help me put life in









perspective. Jonathan Myers has enhanced my passion for science. His philosophical

questions have taught me that we do not really understand something until we are able to

explain it.

I thank my family for supporting me in every enterprise I take. Their love and

advice have been precious tools in helping me achieve my goals. I would have not

completed this thesis and managed to maintain my sanity without the support and

patience of Chuck Knapp. He has survived my frustrations and deadlines, and has been

my best friend and companion. Finally, I thank God in his universal, nondenominational

form, for allowing us to seek the answers for the miracles of nature through science.
















TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ................................................................................................. iv

LIST O F TA B LE S ...... .. .. .. ........................................ .. .. .... .............. viii

LIST OF FIGURES ............................... ... ...... ... ................. .x

ABSTRACT ........ .............. ............. ...... .......... .......... xi

CHAPTER

1 BIOMECHANICAL PROPERTIES OF TROPICAL TREE SEEDLINGS AS A
FUNCTIONAL CORRELATE OF SHADE TOLERANCE.................. ..........1

Intro du action .................................................................................................... ..... .
M materials and M methods ................................................................. ....................... 5
Biomechanical M easurements............................................................ .........7
Young's modulus of elasticity........................................7
F racture toughness.......... .............................................. .......... ........ 7
D en sity ............................................... .......................... 8
C hem ical analy sis............ ... ................................................ ......... .... .8
P percent critical height.......................................................... ............. 9
F lexural stiffness ....................................................... 9
W ork-to-bend ................................... .... .. ..... .. ............10
W hole stem flexibility .............................................. ............................ ....... 10
Force of fracture ........................................... .. ........ ................ .. 11
Specific leaf area .......................................................... ...... .......... 11
Statistical A analyses .................. ............................... .... ... ............... 11
R e su lts ...................................... .......................................................12
Stem B iom echanics ......................... ........................ .. .. ...... .......... 12
L eaf B iom echanics .............. ... ...... ................... .. ... .. .............. ............. 14
Relationship between Biomechanical Traits of Stems and Leaves....................15
Relationship between Seedling Biomechanics and Survival.............................16
D iscu ssio n ...................................... ................................................. 16
Stem B iom echanics ......................... ........................ .. .. ...... .......... 16
L eaf B iom echanics .............. ......... ........... .... ..... ................ ........ ......... .... 20
Relationship between Biomechanical Traits of Stems and Leaves.....................22
Relationship between Seedling Biomechanics and Survival..............................22
C o n clu sio n s ..................................................... ................ 2 4









2 SPECIES DIFFERENCES IN SEEDLING SUSCEPTIBILITY TO BIOTIC AND
ABIOTIC HAZARDS IN THE FOREST UNDERSTORY ....................................41

In tro d u ctio n ...................................... ................................................ 4 1
M materials and M methods ....................................................................... ..................43
Study Site and Species................ ............. .... ........ ............ .............. 43
Experim mental D design ........................................... ... ............................ 44
Survival, Damage Agents, and Types of Mechanical Damage...........................45
A artificial Seedlings .................. ........................... .... .... ............... ... 47
Statistical A naly ses........... ........................................................ .. ... .... .... 48
R e su lts ...............................................................................................4 9
Seedling Survival .................. .............................. .... .... .. ........ .... 49
D am age A gents .............................. .... ....................... ... ...... .... ...........50
Types of M echanical D am age .................................... .......................... ......... 51
A artificial Seedlings .................. ........................... .... .. .. .............. ... 52
D discussion .................. .... ............................. ........................ 53
Survival, Damage Agents, and Types of Mechanical Damage...........................53
A artificial Seedlings .................. ........................... .... .... ............... ... 56
C o n c lu sio n s ..................................................................... 5 7

APPENDIX SPECIES MEANS AND STANDARD DEVIATIONS FOR
BIOMECHANICAL MEASUREMENTS, FIBER ANALYSIS, AND
B IO M A S S ................................................... ...................... ..... 67

L IST O F R E FE R E N C E S ............................................................................. ............. 75

B IO G R A PH IC A L SK E TCH ..................................................................... ..................83
















LIST OF TABLES


Table pge

1-1 Ecological characteristics of eight tropical tree species used in my study,
listed by increasing shade tolerance. .............................................. ............... 26

1-2 Percent seedling survival for the eight study species over specified periods from
four independent studies in BCNM ....................................................................... 27

1-3 Effect of species and harvest time on material and structural properties of
seedling stem s. ..................................................... ................. 2 8

1-4 Relationships among stem biomechanical traits for seedlings of eight tree
sp e cie s. ........................................................... ................ 2 9

1-5 Effect of species and harvest time on material and structural traits of leaves.. .......31

1-6 Relationships among leaf biomechanical traits for seedlings of eight tree species..32

1-7 Relationships among biomechanical traits of stems and leaves for seedlings
of eight tree species. .......................... ...................... ... .. ...... .... ........... 33

1-8 Relationships among % survival in shade and various seedling biomechanical
traits of stems and leaves for seedlings of eight tree species. ................................34

2-1 Relationships among species rankings of survival probability during the
specified interval for seedlings of eight tree species............................................. 59

2-2 Percent damage fatality of four types of mechanical damage on eight tree
species during 1 yr in the forest understory. ................................. .................59

2-3 Relationships among stem biomechanical traits and % damage fatality for
seedlings of eight tree species. ............................................................................ 60

2-4 Percentage of artificial seedlings affected by specified damage agents
in this and other published studies in different forest communities.......................61

A-i Biomechanical measurements of seedling stems from eight tree species ..............68

A-2 Biomechanical measurements of seedling leaves from eight tree species ..............70

A-3 Fiber fractions of seedling stems from eight tree species. .....................................72









A-4 Fiber fractions of seedling leaves from eight tree species. .....................................73

A-5 Biomass measurements of seedling stems and leaves from eight tree species. .......74















LIST OF FIGURES


Figure page

1-1 Means (+ 1 SD) material biomechanical traits of stems for seedlings of eight
tre e sp e c ie s ...............................................................................................................3 5

1-2 Mean (+ 1 SD) % fiber content (% NDF) for seedlings of eight tree species..........36

1-3 Log-log relationships between some material properties of stems for seedlings
of eight tree species .................. .................................. .. ...... .. ........ .... 37

1-4 Means (+ 1 SD) structural biomechanical traits of stems for seedlings of
eight tree species. ................................................... ................. 38

1-5 Means (+ 1 SD) biomechanical traits of leaves for seedlings of eight tree species. 39

1-6 Log-log relationships between some biomechanical properties measured at
6 mos after first leaf expansion (T2), and % mean survival in shade for
seedlings of eight tree species ............................................................................ 40

2-1 Kaplan-Meier survivorship curves for seedlings of eight tree species
transplanted to the forest understory. ............................................ ............... 63

2-2 Kaplan-Meier survivorship curves (proportion of seedlings yet to be hit by
specified damage agents plotted against time) for seedlings of eight tree species
transplanted to the forest understory ............................................. ............... 64

2-3 Percent of real and artificial seedlings (AS) damaged during 1 yr in the forest
understory by specific dam age agent ............................................ ............... 65

2-4 Kaplan-Meier survivorship curves for mechanical damage experienced by
artificial (AS) and real seedlings during 1 yr in the forest understory ...................66















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

BIOMECHANICAL PROPERTIES OF TROPICAL TREE SEEDLINGS AS A
FUNCTIONAL CORRELATE OF SHADE TOLERANCE

By

Silvia Alvarez-Clare

May 2005

Chair: Kaoru Kitajima
Major Department: Botany

Physical disturbances by vertebrates and litterfall are important causes of seedling

mortality in the understory of tropical forests. Thus, the capacity to resist or recover from

mechanical damage should enhance seedling survival in shade. I explored interspecific

variation in seedling biomechanical properties across a shade tolerance gradient, using

eight tropical tree species from Barro Colorado Island (BCI), Panama. The stems and

leaves of shade-tolerant species were constructed of stronger materials than were those of

light-demanding species, as measured by a higher Young's modulus of elasticity, fracture

toughness, and tissue density. These traits were highly correlated with tissue fiber content

(especially % cellulose, but not % lignin) and with seedling survival during the first

6 mo. There were no correlations between seedling survival and structural measurements

that integrated material and morphological traits, such as flexural stiffness, work-to-bend,

and whole stem flexibility. The lack of correlations suggests that investment in strong









material, rather than in large plant size is more beneficial for seedlings at early

developmental stages.

Next, I described first-year temporal patterns of seedling mortality, susceptibility to

damage agents, and types of damage suffered by seedlings in the forest understory.

Seedling mortality was highest during the first 2 mos (due to vertebrate activity) and

gradually decreased over the remaining 8 mo. Species differed significantly in their

temporal patterns of mortality and in the proportions of seedling surviving at the end of

the study. The three main causes of damage were (in order of severity) vertebrate activity,

disease, and litterfall. The four main types of mechanical damage (in order of severity)

were leaf damage, bent stems, broken stems, and uprooted seedlings. All species suffered

similar levels of mechanical damage but shade-tolerant species (which often had stems

constructed of strong materials) were less likely to die when damaged than light-

demanding species.

My study provides evidence that, in Barro Colorado Island, physical disturbance is

a major cause of seedling mortality during the first year, and that shade-tolerant species

survive better than light-demanding species after suffering mechanical damage. Higher

survival is potentially influenced by higher carbon investment of shade-tolerant tree

species into structural support of stems at very early developmental stages. However,

greater carbon allocation to structural defense must be accompanied by slower relative

growth rates. Thus, functional diversity in biomechanical properties is an important

aspect of multiple trait associations that lead to the growth-survival trade-offs observed

among coexisting tropical tree species.














CHAPTER 1
BIOMECHANICAL PROPERTIES OF TROPICAL TREE SEEDLINGS AS A
FUNCTIONAL CORRELATE OF SHADE TOLERANCE

Introduction

Mechanical damage is a major cause of mortality in understory plants, including

tree seedlings (Clark and Clark 1989, Gillman, Wright & Ogden 2002), saplings

(Hartshorn 1972, Aide 1987), and understory herbs (Gartner 1991, Sharpe 1993). Clark

and Clark (1991) found that litterfall caused 11% of the annual mortality of seedlings

< 1 cm in diameter in a lowland tropical rain forest. In a study in seasonal tropical forest,

Alvarez-Clare (Chapter 2) found that 77% of 755 seedlings, from eight species of tropical

trees transplanted to the forest understory, suffered some type of mechanical damage

after 1 yr. Mechanical damage can be caused by falling debris (Aide 1987, Putz et al.

1983), vertebrate activity (Roldan & Simonetti 2001, G6mez, Garcia & Zamora 2004),

water or ice flow (Mou & Warrillow 2000), and herbivory (Coley 1983). In the tropical

rain forest (where there is high frequency of such disturbances) survival of seedlings

depends on their ability to avoid or recuperate from mechanical damage.

An increase in carbon allocation to structural tissues can increase seedling

performance in the forest understory by increasing biomechanical toughness and stiffness

(Sibly & Vincent 1997) and thereby decreasing susceptibility to damage. For example,

Augspurger (1984a) found that from nine species of tree seedlings, species less affected

by pathogen attack were those that became woody more rapidly. Additionally,

mechanical defenses in leaves play a substantial role in deterring loss to herbivores









(Coley 1983, Choong 1996) and are correlated with leaf life span (Wright & Cannon

2001).

Biomechanical strength results from carbon investment in tissues of stems, leaves,

and roots, and from the organization and structure of those tissues within the plant. For

example, fiber is an important contributor of mechanical strength in leaves. Choong

(1996) found that high fracture toughness was correlated with fiber content in leaves of

Castanopsisfissa (Fagaceae). Resistance to mechanical damage is also influenced by the

type and organization of fiber components (e.g., cellulose, hemicellulose, and lignin).

Within a tissue, high cellulose content results in increased toughness, while high lignin

content increases hardness (Niklas 1992). Additionally, the orientation of the cellulose

microfibrils in the S2 layer of the secondary cell wall affects the ability of the material to

resist cracking under plastic tension (Lucas et al. 2000). In stems of adult trees, high

tissue density and cell-wall volume fraction in the xylem increase toughness and stiffness

(Barnett & Jeronimidis 2003). Similarly, higher resistance to mechanical stress in root

tissues can improve anchoring capacity, and reduce risk of uprooting (Campbell &

Hawkins 2004).

Considering the limited carbon budgets of seedlings, increased investments in

structural materials must be accompanied by decreases in allocation to growth, to

reserves, and/or to chemical defenses, such as tannins and alkaloids (Kitajima 1994,

Kobe 1997, Shure & Wilson 1993). Thus, resource limitation leads to trade-offs

involving biomechanical attributes, such as light acquisition vs. structural safety, growth

vs. tissue density, and photosynthetic capacity vs. leaf toughness (Loehle 1988, Niklas

1992, Givnish 1995, Bazzaz & Grace 1997).









Natural selection should favor stems and leaves with forms, biomechanical

properties, and growth dynamics that maximize carbon gain, competitive ability, and

safety, but that minimize costs of construction and maintenance (Givnish 1995).

Obviously, conflicts among these aspects make it impossible to optimize all factors

simultaneously. Thus diverse ecological strategies have evolved, affected by the

evolutionary forces dominating particular ecological niches. For example, plants can cope

with mechanical damage through investment in resistant structures or by allocating

resources to reserves that enable them to replace damaged tissues (Harms & Dalling

1997, Pauw et al. 2004). In both cases, a strategy will only be selected for, if it confers a

benefit relative to the cost, such as increased survival (Sibly & Vincent 1997). My study

focused on the defensive strategy of investing in damage resistance, by exploring the

influence of biomechanical traits on seedling survival in shade.

Although biomechanical properties clearly influence plant survival and competitive

ability, and potentially influence their ecological distribution (Coley 1983, Niklas 1992,

Lucas et al. 2000), investigations evaluating plant biomechanical properties in an

ecological context are few. Exploring the functional diversity of biomechanical properties

in tropical tree seedlings should help in describing multiple trait associations that lead to

growth-survival trade-offs observed among coexisting tropical tree species.

My study explored interspecific variation in seedling biomechanical properties

across a shade-tolerant gradient, among eight tropical tree species from Barro Colorado

Island (BCI), Panama. Because my goal was to understand the ecological role of

biomechanical traits in tropical tree seedlings, I evaluated a variety of biomechanical

attributes at the material and structural level. Plant stem and leaf material traits consist of









the composite, anisotropic material of which they are composed. At this level,

mechanical (toughness, stiffness, and density) and chemical (fiber fraction) traits were

measured, without considering size or anatomical organization. At the structural level,

material properties combined with morphological traits (e.g., size and shape) were

measured for individual plant organs (stems or leaves). Measurements at the structural

level for stems included flexural stiffness, percent critical height, work-to-bend, and

whole stem flexibility. To describe the structure of leaves, I measured specific leaf area

and force of fracture. Specifically, I addressed the following three questions:

* Do material properties of stems and leaves of tropical tree seedlings differ
among species in relation to their shade tolerance? Because strong material
confers an advantage against mechanical damage and presumably increases
survival probabilities, I predicted that shade-tolerant species, which survive better
in shade (Wright et al. 2003, Chapter 2), should have stronger stem and leaf
materials than light-demanding species. More specifically, stems of shade-tolerant
species should have higher Young's modulus of elasticity, fracture toughness, and
density. In addition, shade-tolerant species should have higher fiber content, which
reflects the chemical composition of the material. Likewise, leaves of shade-
tolerant species should have higher lamina and midvein fracture toughness, density,
and fiber content.

* What is the relationship between material and structural biomechanical
properties, and between material properties of stems and leaves? Carbon
allocation to stronger tissues should also contribute to overall structural strength.
Therefore, unless there are important morphological differences between species,
material traits should be reflected at the structural level. Stems of shade-tolerant
species should have lower percent critical height, higher flexural stiffness, and
higher resistance to bending in the field. Leaves of shade-tolerant species should
have a lower specific leaf area (SLA) and a higher overall resistance to fracture
(force of fracture) than leaves of shade intolerant species. I also expected a
concordance of biomechanical attributes between structures. Thus, biomechanical
properties of stems and leaves should be correlated. Because it was my ultimate
objective to evaluate the implications of biomechanics for seedling performance in
the forest, it was key to examine biomechanical traits at the material level but also
at the structural level, integrating morphological attributes that can influence
overall plant response to mechanical stress.

* How do biomechanical properties of seedling stems and leaves change over the
first 6 mos after initial development? It has been shown that free-standing plants,
as opposed to lianas, increase their stem resistance to bending and breaking during









growth and maturity (Rowe & Speck 1996). Similarly, leaves become tougher with
aging (Wright & Cannon 2001). Therefore, I predicted an increase in mechanical
strength of stems and leaves, both at the tissue and at the structural level.

Materials and Methods

The study was conducted in Barro Colorado Natural Monument (BCNM), Panama

(90 10' N, 790 51' W). Data were collected during the rainy season (May-December),

when 78% of the average annual precipitation of 2600 mm falls. Climate, flora, and

ecological characteristics of the seasonally moist tropical forest in BCNM are well

described by Croat (1978) and by Leigh, Windsor & Rand (1982).

I collected seeds from eight common species in BCNM that differ in ecological

characteristics such as dispersal mode, cotyledon type, and seedling establishment

probability (Table 1-1). A seedling-recruitment index was calculated as # seeds falling

m2year1 / # seedlings established m2year 1 obtained from a long-term experiment on

Barro Colorado Island (BCI). Seed rain density (# seeds m-2yea-1) was measured in

weekly censuses from 1995-1999 in two hundred 1 m2 seed traps. Recruitment of new

-2
seedlings (# seedlings m2year 1) was measured once a year from 1995-1998 during the

dry season in six hundred, lm2 recruitment plots,2 each located 2 m from three sides of

the seed traps (Wright et al. 2003).

Seedling shade tolerance was ranked according to measurements of seedling

survival in the shaded understory from four independent studies conducted in BCNM

(Table 1-2). Alvarez-Clare (Chapter 2) determined first-year survival of seedlings

transplanted to the forest understory at first leaf expansion, and censused weekly for the

first 3 mos, then biweekly for the rest of the year. Kitajima (unpublished data) and Myers

(2005), both quantified first year survival of seedlings transplanted to fenced enclosures

from which vertebrate predators were excluded. Wright et al. (2003) estimated survival









probability after 1 yr, using naturally recruited seedlings in the forest understory. Only

studies by Alvarez-Clare and by Kitajima included all the species assessed here, and

therefore survival per species from these two sources was averaged to determine mean

% survival (Table 1-2). Mean % survival was obtained from 2-6 mo survival from

Alvarez-Clare, which excludes initial transplant shock and most vertebrate predation

(Chapter 2), and survival from 0-4 mo from Kitajima, which was the interval before most

plants were harvested. Mean percent survival was correlated with biomechanical traits at

1 and 6 after mos the expansion of the first leaf.

Seeds were germinated in trays in a shaded house where daily total photosynthetic

photon flux density was adjusted with shade cloth to approximately 2% of full sun. I

transplanted 45 seedlings of each species to each of three 6 x 6 m common gardens

located on 70-year-old secondary forest on Buena Vista Peninsula. To standardize by

ontogenetic stage based on development of photosynthetic organs, I transplanted

seedlings at expansion of the first leaf for all species with reserve cotyledons and at

expansion of cotyledons for Tabebuia rosea, a species with photosynthetic cotyledons.

Time from germination until leaf expansion varied across species from one week for

Anacardium excelsum, to four weeks for Eugenia nesiotica and Tetragastris panamensis.

Each garden was situated under closed canopy and surrounded by a 1 m tall wire mesh

fence to exclude large, ground-dwelling herbivores.

In each garden, seedlings were transplanted 50 cm with each species randomly

located within planting positions in each plot. I replaced those that died within the first

week after transplanting. Half of the seedlings from each species were harvested after

1 mo (T1), and the remaining plants were harvested approximately 6 mo later (T2).









Forty-five plants (15 per garden) of each species were randomly chosen at T and used to

perform biomechanical tests in situ before being harvested, and then used in the

laboratory to test biomechanical properties in different organs (e.g., if stem flexibility

was measured in the field, leaves were measured in the laboratory). Because of mortality,

at T2 only 30 seedlings per species were measured. After being harvested, all plants were

refrigerated for less than 12 h until laboratory biomechanical tests were performed. After

testing, plants were separated into stems, roots, and leaves, weighed and then dried at

100 C for 1 h and then at 60 C for 48 h to determine dry weight. Samples were saved

for fiber analysis.

Biomechanical Measurements

Young's modulus of elasticity

Young's modulus of elasticity (E) of stems was measured in a three point bending

test with a Portable Universal Tester (Darvell et al. 1996), as described in Lucas et al.

(2001). More specifically, Young's modulus was calculated from the slope of the linear

regression of the applied bending force vs. deflection. Span distance varied with stem

size and bending resistance. Span ratios of>10 were always used, as suggested by Niklas

(1992). Young's modulus of elasticity is defined as the ratio between forces of stress and

strain, measured within the plastic range of a homogeneous material (i.e., stiffness) In the

case of stems, because they are constructed of heterogeneous and composite materials, I

measured an apparent Young's modulus, which describes the overall bending properties

of a stem independent of size and shape (Niklas 1992).

Fracture toughness

I measured fracture toughness for stems and leaves by performing cutting tests with

a sharp pair of scissors mounted on a Portable Universal Tester as described by Lucas &









Pereira (1990). Toughness obtained through cutting tests is the work required to

propagate a crack over a unit area (Lucas et al. 2000) and has been used in leaves as an

indicator of resistance to herbivory, pathogens, and other physical damage (Lucas &

Pereira 1990, Choong 1996). For leaves, toughness was measured for lamina and midrib

separately. When measuring stems, I cut the stem at half of the total length, or just above

the cotyledons in A. excelsum and T. panamensiss, which have epigeal cotyledons.

Density

Tissue density was calculated for leaves and stems as the ratio of dry mass to

volume. For leaves, volume was calculated as total leaf area (measured in the leaf area

meter) multiplied by the lamina thickness, and dry mass was obtained for the total leaf

including midrib and veins. For stems, volume was obtained from the formula:

V =(r 2)h (1-1)

where r is the radius measured at the middle of the stem and h is stem length, both

measured in mm. For measuring density and the other biomechanical properties, stems

were considered perfect cylinders, ignoring taper.

Chemical analysis

To evaluate fiber content and relate it with biomechanical measures, fiber fractions

were determined for stem and leaf tissues separately, using a series of increasingly

aggressive extractants (Ryan, Melillo & Ricca 1989) with a fiber analyzer system

(ANKOM Technology, NY, USA). Dried plants of each species from the same common

garden and same harvest were combined and ground as one sample to have a minimal of

0.5 g required for analysis. Because of the small size of T rosea, all harvested plants

were combined and ground as one sample. In the first step, each ground sample was

weighted and sealed in a chemical resistant filter bag. The bagged samples were









submerged, heated, and agitated in neutral detergent fiber solution removing soluble cell

contents and leaving non-detergent fiber (% NDF). In the second step, the bagged

samples were treated with acid-detergent solution, which removed hemicellulose and left

acid-detergent fiber (% ADF) consisting of cellulose, lignin, cutin and insoluble ash. In

the third step, samples were treated with 70% sulfuric acid, which removed cellulose and

left lignin, cutin and insoluble ash inside the bags. Between steps, sample bags were dried

at 1000C overnight to determine the dry mass, and each fiber fraction was calculated by

subtraction. Afterwards, the remaining sample was combusted at 5000C to determine

percent insoluble ash. Mass of labile cell contents + hemicellulose + cellulose + lignin +

insoluble ash add up to 100% of the original dry mass.

Percent critical height

Percent critical height (% Her) measures the relationship between stem height and

how tall it could be before it buckles under its own weight (Holbrook & Putz 1989).

Percent critical height was calculated for each seedling stem according to the formula

given by Greenhill (1881):

Her = 1.26(E/w)" (db) 23 (1-2)

where E = Young's modulus of elasticity (Pa), w = fresh weight/unit volume (Nm-3), and

db = diameter at base (m). The ratio of Her to the actual stem height multiplied by 100 is

% Her, which is an indication of mechanical risk-taking. In other words, the higher the

% Her the lower the margin of safety for the stem to remain free-standing.

Flexural stiffness

Flexural Stiffness (El) describes the ability of a structure to withstand mechanical

loads, taking into account the size and shape of the structure as well as the material

properties of its tissues (Gartner 1991). It is the product of E, which describes the









flexibility of the material, and the second moment of area (1), which reflects size and the

geometry of the structure to which a force is being applied. I estimated flexural stiffness

(El) for cylindrical stems using the formula:

I= 0.25nr4 (1-3)

where r (mm) is the radius measured in the middle of the stem and the Young's modulus

of elasticity (E) obtained with three point bending tests, as described above (Niklas

1992).

Work-to-bend

Resistance of stems to bending, here referred to as "work-to-bend", was obtained

empirically in the field by applying a force vertically from above a seedling until the stem

was deflected to 70-60% of its original height. To estimate work-to-bend a 2 L plastic

container was mounted on a 30 cm2 Styrofoam platform and hung from a tripod with a

spring balance just above the seedling. The Styrofoam platform was in contact with the

uppermost part of the seedling, without bending it. Then, water was poured slowly into

the container, until the weighted platform bent the stem to the specified extent. Assuming

that acceleration was nil, water weight (force) times vertical displacement, was calculated

as work to bend the seedling.

Whole stem flexibility

To further describe the behavior of intact seedlings rooted in the ground in response

to mechanical stress, I measured whole stem flexibility (Holbrook & Putz 1989) in the

field. A stem was pulled horizontally in four directions with spring balances until bent

200 from vertical. This procedure was repeated in the four canonical directions and the

forces averaged. Whole stem flexibility (WSF) was expressed as angular deflection

divided by applied force (radians/N). In the case ofE. nesiotica, I bent the stem 40,









because the force required to bend the stem 200 was too small to be detected in its small

seedlings. Whole stem flexibility is a measure of elasticity whereas flexural stiffness is a

measure of rigidity; therefore I expected them to be inversely correlated. Because WSF

applies a lateral tensile force (the stem is pulled laterally), and work-to-bend applies a

vertical compressive force (the plant is pushed down), slightly different stem properties

are being measured, and thus I performed both tests.

Force of fracture

For leaves, force of fracture was calculated as the product of fracture toughness by

lamina thickness. This structural measurement indicates total force necessary to

propagate a crack considering leaf thickness (Wright & Cannon 2001).

Specific leaf area

Specific leaf area (SLA) was calculated as the ratio of leaf area, measured with a

leaf area meter (LICOR-3100), and leaf total dry mass. Because species with low SLA

are usually thick and/or dense (Wright & Cannon 2001), I expected SLA to be inversely

correlated with leaf fracture toughness and force of fracture.

Statistical Analyses

Every biomechanical measurement was averaged for each species, and species

means were log-transformed to meet normality assumptions for ANOVA tests

(Shapiro-Wilk, a = 0.05). For each measurement, the effect of species (N= 8) and

harvest time (N = 2) was evaluated using two-way ANOVAs. When the species*time

interaction was significant, the data for each harvest were analyzed separately. To test if

means differed between T1 and T2 within each species, t-tests with subsequent

Bonferroni corrections were applied. For across-species comparisons between two

biomechanical measurements or between a biomechanical measurement and survival,









linear regressions on log-log plots were calculated. For multiple across-species

comparisons between non-normal variables, Spearman rank correlations were applied.

Two means per species (one per harvest) were obtained to evaluate the correlation

between biomechanical traits (N= 16). Work-to-bend was only measured at T1 (N= 8)

for logistic reasons. For Spearman correlations between biomechanical properties and

% survival in shade, species means were evaluated at each harvest separately (N= 8). All

analyses were performed using JMP IN 4.0 (SAS Institute Inc., Cary, NC, USA) with a

significance level of a = 0.05.

Results

Stem Biomechanics

Mean Young's modulus of elasticity (E) of the seedling stems varied 20-fold

among species (Figure 1-1A). Most species increased their resistance to bending (E)

during the six-month period between T1 and T2 (Table 1-3) resulting in significant time

effect without a species*time interaction. Mean stem fracture toughness also varied

among species and between harvests (Figure 1-1B), but the amount of increase in fracture

toughness varied among species (Table 1-3). While E. nesiotica increased its mean

fracture toughness threefold from 1-6 mos after leaf expansion, A. exelsum and T.

panamensis showed no increase (Figure 1-1B). Mean stem tissue density also varied

among species and between harvests, increasing from T1 to T2 for all species except

A. cruenta, which decreased its mean stem tissue density over time (Table 1-3, Table A-

1). Total fiber (% NDF) was generally higher for more shade-tolerant species (Table 1-3),

but A. cruenta, the species with highest survival in shade, had a mean % NDF similar to

the three least shade-tolerant species (Figure 1-2A). Mean % NDF did not differ









significantly between harvests (Table 1-3 and Table A-3). All individual fiber fractions

varied between species, but only % hemicellulose increased between harvests (Table 1 3).

Most material properties of stems were inter-correlated (Figure 1-3 and Table 1-4).

Material mechanical traits, such as toughness and modulus of elasticity, were positively

correlated (Figure 1-3A). Additionally, material properties describing mechanical

strength (e.g., modulus of elasticity and density) were correlated with chemical indicators

of tissue strength (e.g., fiber content; Figure 1-3B-D). Among chemical properties,

% cellulose was the best predictor of mechanical strength, as measured by modulus of

elasticity, fracture toughness, and density (Table 1-4). Percent lignin was not correlated

with toughness or density but was a good predictor of modulus of elasticity (i.e., stem

stiffness).

Mean percent critical height (% Her) varied among species and significantly

decreased in three out of eight species from T1 to T2 (Figure 1-4A and Table 1-3). All

species had low % Her (their actual height was 14-28% of their critical height), indicating

that seedlings were overbuilt relative to their potential maximum height before buckling

under their own weight. Mean flexural stiffness (El) varied among species and between

harvests, with an interaction between factors (Figure 1-4B and Table 1-3). The significant

interaction was apparently influenced by A. exelsum and G. superba, the species with the

largest seedlings (i.e., largest I), which disproportionately increased El from T1 to T2.

Mean work-to-bend (i.e., work necessary to bend a stem to 70% of its original height)

varied four-fold among species (Figure 1-4C and Table 1-3). Stem diameter was a good

predictor of work-to-bend (r = 0.53, F = 73.0, d.f. = 1,66, P < 0.001), and consequently

there was a positive correlation between El and work-to-bend (Table 1-4). Mean whole









stem flexibility, measured as angular deflection, varied among species decreasing over

time as stems became more lignified (Figure 1-4D and Table 1-3). Plants with large stem

diameters were less flexible than plants with small stem diameter (r = 0.64, F = 289.70,

d.f. = 1,160, P < 0.001).

Although mechanical and chemical traits of stem tissues were intercorrelated, they

were never correlated with structural measurements that integrated material and

morphological traits (Table 1-4). The only exception was % Her, which was negatively

correlated with modulus of elasticity (E), fracture toughness, % NDF, and % cellulose.

This observation indicates that species with stronger material had a lower % Her and

hence a greater safety margin.

Second moment of area (1), did not correlate with any of the material properties. In

contrast, both I and flexural stiffness (El) correlated positively with structural traits, such

as work-to-bend and whole stem flexibility measured on intact seedlings in the field

(Table 1-4).

Leaf Biomechanics

Material biomechanical traits of leaves differed among species and between

harvests, although not all species varied consistently between TI and T2. Lamina fracture

toughness differed among species with a significant interaction between species and time

(Table 1-5). Two species increased their lamina toughness, two decreased, and four

species did not vary between T and T2. Fracture toughness of midveins varied between

species and between harvests, with a significant interaction between these two factors

(Figure 1-5B, Table 1-5). In general, for each species midvein fracture toughness was

lower or similar than stem toughness, but much higher (ca. x10) than lamina toughness.

Leaf density also varied among species and between harvests (Table 1- 5). A significant









interaction between species and time was probably because leaf density increased from

Tl to T2 in B. pendula much more than in other species (Table A-2). Percent NDF

differed among species but not between harvest times (Figure 1-2B and Table 1-5). All

individual fiber fractions varied among species, but only % lignin changed between

harvests (Table 1-5).

Structural properties of leaves, integrating material properties and morphology

varied among species, but only SLA differed between harvests (Table 1-5). Tabebuia

rosea had the highest SLA, while A. cruenta had the lowest SLA. Force of fracture was

different among species, but not between harvest times (Figure 1-5 and Table 1-5).

Biomechanical attributes of midveins highly influenced mechanical traits of the

whole leaf. Across species, there was a positive correlation between lamina and midvein

toughness (Table 1-6). Total leaf density was best correlated with midvein than with

lamina toughness, suggesting that biomechanical attributes of the midvein significantly

influence overall leaf density. Percent cellulose was the chemical trait that most

correlated with the rest of the material traits. Force of fracture (toughness*thickness) was

more correlated with toughness than with thickness, indicating a stronger effect of leaf

material properties than of leaf dimensions.

Relationship between Biomechanical Traits of Stems and Leaves

Across species, there was a positive correlation between stem toughness and

midvein toughness, but not between stem toughness and lamina toughness (Table 1-7).

Tissue density and % NDF were also correlated between stems and leaves, but the other

fiber fractions were not (data not shown).









Relationship between Seedling Biomechanics and Survival

Several stem material biomechanical properties were positively correlated with

% mean survival in shade (Table 1-8). Fracture toughness and tissue density measured at

T2 showed the highest correlations with survival in both stems and leaves (Figure 1-6,

Table 1-8). Furthermore, if A. cruenta (the species with high survival but with low E and

% fiber content) was removed from the analyses, all correlations between material

biomechanical traits and survival increased. Although individual fiber fractions exhibited

no significant correlation with survival, % NDF (i.e., total fiber) was positively correlated

with survival in both stems and leaves. Stem and leaf structural properties, at 1 and 6 mos

after expansion of the first leaf, were not correlated with survival in shade.

Discussion

Stem Biomechanics

Mechanical traits and chemical composition of seedling stems varied widely among

eight species of tropical trees but as predicted, stems of shade-tolerant species were

generally stiffer, tougher, and denser, and with higher total fiber content (% NDF) than

stems of shade intolerant species (Figures 1-1 and 1-2A). Among the biomechanical

properties tested there were positive correlations between Young's modulus of elasticity,

fracture toughness, and stem density suggesting a greater overall investment in strong

material properties in shade-tolerant species. Similar results were obtained by Cooley,

Reich & Rundel (2004) for understory herbs. Although in my study there were positive

correlations between mechanical and chemical material traits, the fiber components

contributing to these correlations differed, with the mechanical property considered. For

example, fracture toughness was correlated with % cellulose and % hemicellulose, but

not with % lignin (Table 1-4). As a complex, heterogeneous polymer with strong









covalent bonds, lignin acts as an adhesive agent in the cell wall, and therefore is expected

to increase stiffness rather than toughness (Lucas at al. 2000). In fact, the only

mechanical property correlated with % lignin was E, a measure of stem stiffness.

Modulus of elasticity, however, can also be affected by other tissue properties such as

volume fraction of cell wall materials (Lucas et al. 2000, Niklas et al. 2000),

hemicellulose and cellulose contents, and microfibril angles in the cell wall of fiber cells

(Hoffman 2003, Savidge 2003).

Differences in material properties at the time of first leaf expansion (T ) suggest

that shade-tolerant species invested earlier in stem mechanical construction than shade

intolerant species. Thus, shade-tolerant species potentially had a more developed vascular

cambium and greater secondary cell wall deposits than shade intolerant species. Mean

moduli of elasticity (E) for shade intolerant species at T1 were similar to those reported

for stems ofunderstory herbs (Cooley, Reich & Rundel 2004, Niklas 1995). This

suggests that 1 mo after leaf expansion, vascular cambium development (and thus

secondary growth) was still limited, and seedlings were relying on primary tissues for

mechanical support (Niklas 1992, Isnard, Speck & Rowe 2003). In contrast, shade-

tolerant species (e.g., T. panamensis and E. nesiotica) had moduli of elasticity at T1 of

the same order of magnitude as wood from 15 of 33 adult temperate trees evaluated by

Niklas (1992). Species with stronger material properties had higher fiber contents as well.

Specifically, they had higher % lignin and % cellulose fractions, which are correlated

with vascular cambium maturation, high cell wall volume fraction, and secondary cell

wall development (Niklas et al. 2000, Lucas et al. 2000). Because shade-tolerant species

are usually slow growers (Kitajima 1994), it is not likely that further stem maturity at the









time of first leaf expansion in shade-tolerant species was a product of accelerated stem

development. On the contrary it reveals an ecological strategy, characterized by

substantial investment in material starting very early in ontogeny.

Variation in stem development at T1 could be influenced by leaf emergence times.

Kitajima (2002) demonstrated that T rosea, a light-demanding species with

photosynthetic cotyledons, became dependent on photosynthetic carbon gain earlier in

development than shade-tolerant species with storage cotyledons. Rapid photosynthetic

cotyledon expansion after radicle emergence (22.5 + 1.9 d), allows little time for stem

structural development and toughening. In contrast, T. panamensis a shade-tolerant

species with reserve cotyledons, expands its first leaves relatively quickly (23.6 + 2.4 d),

but has a high modulus of stem elasticity. Although age (time after radicle emergence)

may potentially affect stem stiffness and toughness, this is evidently not the sole cause of

variation. Among species variation in biomechanical properties of stems at first leaf

expansion is a function of differences in material composition and structural arrangement,

which suggest the existence of different ecological strategies among species of tropical

tree seedlings.

I predicted that material traits of seedling stems would be reflected at the structural

level. Thus, I expected stems with stronger material properties per unit area (or mass) to

be more resistant to bending and breaking. Results confirmed this prediction, but only

when stems of similar size were compared. When different sized seedlings were

compared, species with larger seedlings (at comparable developmental stages) were more

resistant to bending, both for tests performed in the laboratory and on intact seedlings in

the field. A plant can obtain a high flexural stiffness by increasing E (material stiffness),









or by increasing I, a measure of size and shape (Niklas 1992). Given that seedlings of all

eight species included in my study had circular stems, the observed differences in I

reflect differences in size only. Likewise, differences in flexural stiffness among species

were mostly influenced by size of the stem (1), as opposed to flexibility of the material

(E). Similar results have been reported for neotropical understory herbs (Cooley, Reich &

Rundel 2004), vines (Rowe & Speck 1996), shrubs (Gartner 1991), and trees (Holbrook

and Putz 1989). In contrast, other studies have found an influence of both E and I when

comparing flexural stiffness of stems growing in environments differing in wind intensity

and shade conditions (Cordero 1999, Henry and Thomas 2002), and when comparing

stems from congeneric species differing in growth form (Isnard, Speck & Rowe 2003,).

When intact, live stems were tested in the field, work-to-bend and whole stem

flexibility correlated with other structural traits, but not with material properties

(Table 1-4). The results of these field tests correlated well with flexural stiffness, which

was measured using harvested stems in the laboratory. Whole stem flexibility and work-

to-bend proved good field indicators of stem rigidity for tropical tree seedlings, and

should be taken into account in future research regarding seedling biomechanics.

The structural property that best correlated with material properties was % critical

height. Seedlings from shade-tolerant species had higher safety factors (i.e., lower % Her),

than seedlings from shade intolerant species. As suggested by Givnish (1995), my results

indicate that there is a trade-off between light acquisition and mechanical safety. While

some trees maximize their height to reach light and overtop competitors, this increases

vulnerability to toppling (Holbrook & Putz 1989, Bruchert, Becker & Speck 2000).

Although all species in my study were overbuilt (Figure 1-2A), light-demanding species









had higher % Her and weaker material traits than shade-tolerant species, suggesting that

they were maximizing height growth at the expense of safety and structure.

As predicted, all species increased their mean E between 1 and 6 mos after leaf

expansion, although not always significantly (Figure 1-1). In contrast, there was no

pattern to the proportional increase in fracture toughness between T and T2 among

species, revealing that species do not necessarily increase toughness and stiffness

proportionally during ontogeny. Thus, for seven out of eight species in which stem fiber

content did not increase from T1 to T2, increases in stiffness and toughness over time

must have been caused by changes in stem anatomy, such as fiber distribution and

packaging, as opposed to increased fiber content (Hoffman et al. 2003), but further

anatomical and histological analyses are necessary.

Leaf Biomechanics

Mean lamina and midvein toughness varied 30-fold among species, with values

from 71 to 395 J m -2 for laminas and 984 to 3475 J m-2 for midveins. In a study

performed on BCI with leaves from adult trees and understory saplings, Dominy, Lucas

& Wright (2003) reported considerably higher values for lamina and midvein toughness

than reported here. Nevertheless, for the three species used in both studies (A. excelsum,

C. elastic, and A. cruenta), the same ranking prevails: A. excelsum had the lowest

lamina and midvein toughness while A. cruenta had the highest. Although the

relationship was weaker than in stems, leaves of shade-tolerant species had higher

mechanical strength than leaves of shade intolerant species. Potentially, evolutionary

forces favoring selection of other leaf traits, such as photosynthetic capacity, vein

distribution, presence of secondary compounds, and water-use efficiency also influence

differences in leaf toughness among species (Choong et al. 1992, Wright et al. 2004).









Fiber content is an indicator of biomechanical strength in leaves (Choong 1996). In

the present study, mean % cellulose was the fiber fraction that best correlated with leaf

fracture toughness, suggesting that cell wall material was the predominant cellular

component influencing fracture toughness (Esau 1977), but it is not clear which tissues

make a leaf tough. Both the cuticles (Taylor 1971) and the epidermis (Grubb 1986) have

been proposed as toughening tissues. Additionally, Wright & Illius (1995) reported that

the proportion of sclerenchyma in leaves was correlated with fracture toughness of

grasses, and Choong (1996) found that the non-venous lamina contributed little to overall

leaf toughness. In the present study, the positive correlation between midvein and lamina

toughness suggests that vascular bundles (and probably fibers associated) were the major

determinants of fracture toughness in leaves.

Structural measurements integrating leaf dimensions and size were correlated with

material traits but not with morphological traits (Table 1-6). For example, force of

fracture, calculated as the product of lamina toughness and leaf thickness, was better

correlated with lamina toughness than with leaf thickness. Thus, unlike stems, overall

leaf biomechanical properties were influenced more by material traits than by leaf

dimensions. Similar results were reported by Wright & Cannon (2001) in a study with

17 sclerophyllous species from low-nutrient woodland in eastern Australia.

I expected that biomechanical strength of leaves would increase over time;

however, most species did not change, and some even decreased in their mechanical

strength between T1 and T2. In fact, for G. superva and T panamensis mean lamina

fracture toughness decreased significantly after 6 mos. Although there is no evident

explanation for this observation, Lucas & Pereira (1990) found the same trend (where









leaves decreased their fracture toughness over time). They suggested that an increase in

parenchymatous tissue and air species in older leaves could result in low fracture

toughness per unit volume.

Relationship between Biomechanical Traits of Stems and Leaves

Measurements of the material traits of stems and leaves were positively correlated

for the eight species combined. Species with tough, dense stems also had tough, dense

leaves. An exception was A. cruenta, which had tough, thick leaves, but stems

constructed of weak and flexible material. Aspidosperma cruenta also stores substantial

amounts of nonstructural carbohydrates in its stems, which may augment its ability to

recuperate from damage, rather than avoid it (Myers 2005). Across species, the

correlation between stem and midvein toughness was stronger than the correlation

between stem and lamina toughness. The strong relationship between stems and midveins

could be driving the relationship between stem and leaf density or fiber content,

suggesting consistent investments in vascular structure throughout the plant. Collectively,

these results suggest that there is a whole-plant pattern of carbon investment in

mechanical defenses, as opposed to a trade-off between investment in stem and leaves.

Further investigations might evaluate whether this pattern remains consistent in roots.

Relationship between Seedling Biomechanics and Survival

Material properties of stems correlated with 0-6 mo survival in shade (Table 1-8,

Figure 1-6). Species stems constructed of tougher, stiffer, denser, and more fibrous

material showed higher percent survival than species composed of weaker material. This

is direct evidence that biomechanical strength of stem tissues increases seedling

performance in the tropical forest understory. As suggested in previous studies, strong

material is likely to confer an advantage against mechanical damage caused by litterfall,









vertebrate trampling, and herbivory (Augspurger 1984a, Clark & Clark 1991, Moles &

Westoby 2004a). The only species that deviated from the trend was Aspidosperma

cruenta. Seedlings from this species had the highest survival in shade, but its stems were

constructed of weak material. Most likely, high survival in A. cruenta was due to the

presence of chemical defenses and large reserve pools of carbohydrates in stems and

roots. Aspidosperma cruenta is well known for its poisonous alkaloids

(e.g., obscurinervine and obscurinervidine, Harper et al. 1993), and well-developed

chemical defense that may compensate for its low structural defense, revealing a unique

ecological strategy among the eight species tested. It should be noted that chemical

defenses confer herbivore resistance (Coley 1983), but do not protect seedlings from

mechanical damage due to litterfall or vertebrate trampling. The high survival of A.

cruenta on BCI, albeit its lack of mechanical defenses, suggests that for this species

defense against herbivory and pathogens (through secondary compounds) was more

important as a selective factor, than defense against mechanical damage, at least during

the first 6 mos.

Surprisingly, structural traits that integrate material properties with seedling size

and shape were not correlated with six-month survival in shade. Larger seedlings had

higher overall resistance to bending (Figure 1-4), but with no apparent consequence for

seedling survival. Although previous studies have emphasized the advantages of large

size for seedlings (reviewed in Moles and Westoby 2004a), my results suggest that

evolutionary pressures selecting for large seedlings are probably related to stress-

tolerance (Green & Juniper 2004) and light acquisition (Turner 1990), not to

biomechanical strength.









For leaves, there was a positive correlation between some of the biomechanical

traits and survival in shade but the trends were not as strong as for stems. Most likely,

leaf biomechanical traits are directly correlated with leaf performance (e.g., leaf lifespan

or risk of herbivory), but not with whole plant performance (e.g., survival). For example,

Wright and Cannon (2001) found that mean leaf toughness, force of fracture, leaf

thickness, and leaf area explained between 30 and 40% of variation in leaf life span of

17 species of sclerophyllous plants. In a study with 2,548 species, Wright et al. (2004)

found that leaf mass per area (LMA), explained 42% of the variation in leaf life span,

indicating that thicker, denser leaves, usually live longer. Additionally, the weaker

correlations I observed between survival and mechanical traits of leaves suggest that

invertebrate herbivores that cause leaf damage are not crucial determinants of whole-

plant survival during the first 6 mo, for the eight species considered in my study (Chapter

2).

Conclusions

Interspecific variation in material flexibility and fracture toughness of seedling

stems as early as one month after leaf expansion, revealed different ecological strategies

to cope with mechanical damage in the forest understory. Shade-tolerant species had

stems constructed of strong materials, which may promote their survival in shade.

However, stronger material properties of stems did not always reflect strength at the

structural or whole-plant level. Size and several morphological traits contributed to

overall resistance to bending and breaking stress, but they apparently were not crucial for

seedling survival from 0-6 mo. As opposed to stems, leaf biomechanical properties were

influenced more by material traits than by leaf dimensions, and biomechanical attributes

of leaves were not always correlated with whole-plant survival. In tropical tree seedlings,






25


differential survival in shade is the product of a suit of traits of which biomechanics is an

important component.












Table 1-1. Ecological characteristics of eight tropical tree species used in my study, listed by increasing shade tolerance.
Sp. Species Family Cot. Dispersal %Rec. Seed mass (g)
code type index
TABR Tabebuia rosea Bignoniaceae PEF Wind 0.6 0.035 + 0.007 (12)
ANAE Anacardium excelsum Anacardiaceae PER Animal 0.1* 1.811 + 0.316 (9)
CASE Castilla elastica Moraceae CHR Animal 0.315 + 0.005 (8)
BEIP Beilschmiediapendula Lauraceae CHR Animal 13.7 2.360 + 0.090 (10)
GUSS Gustavia superba Lecythidaceae CHR Animal 3.7* 5.566 + 1.746 (7)
TETP Tetragastris panamensis Burseraceae PER Animal 3.5 0.179 + 0.026 (10)
EUGN Eugenia nesiotica Myrtaceae CHR Animal 27.8* 0.474 + 0.067 (10)
ASPC Aspidosperma cruenta Apocynaceae PHRt Wind 2.9* 0.492 + 0.002 (6)
Cotyledon types are according to Garwood (1996): PEF = phanerocotylar epigeal foliaceous, PER = phanerocotylar epigeal reserve,
CHR = cryptocotylar hypogeal reserve, and PHR = phanerocotylar hypogeal reserve. Percent recruitment (% Rec. Index) refers to
percent recruits per seeds per area (Wright et al. 2003). Mean + 1 SD (N) seed mass without seed coat. Data obtained with between 5
and 10 recruits.t Cotyledons are partially cryptocotylar.











Table 1-2. Percent seedling survival for the eight study species over specified periods from four independent studies in BCNM.
Sp. code Mean % Alvarez-Clare a Kitajima b Myers b Wright
survival 0-2 mo 2-6 mo 6-12 mo 0-4 mo 4-12 mo 0-6 mo 6-12 mo 0-12 mo
TABR 45.5 33 (55) 44 (18) 29 (7) 47 (48) 30 (23) 33 (71) 46(14) 31(58)
ANAE 53.0 20(100) 40(20) 11(9) 66(51) 26(34) -
CASE 65.0 40(100) 73(40) 72(25) 57 (28) 67 (18) 65(101) 86(44) -
BEIP 82.5 8(100) 88(8) 60(5) 77(61) 19(47) -- 52(826)
GUSS 76.0 54 (99) 83 (54) 79 (43) 69 (42) 86 (32) 57 (213)
TETP 82.0 62(100) 79(62) 82(71) 85(20) 90(10) 64(361)
EUGN 87.5 43 (100) 100 (42) 75 (32) 75 (63) 96 (47) -- 81(22)
ASPC 87.0 78(100) 93(78) 82(71) 81(27) 99(21) 98(111) 97(104) -
Numbers in parentheses indicate sample size, (i.e., the total number of individuals at the beginning of the measurement period).
Values shown in bold were averaged for each species and used to calculate mean % survival. Refer to Table 1-1 for species codes.
aThis study. Seedlings transplanted to the forest and monitored for 1 yr (Chapter 2). Time is divided into different stages because
initial mortality during 0-2 mo was due mainly to vertebrate activity, and thus is not a good indicator of shade tolerance. bSeedlings
transplanted at the time of germination (K. Kitajima, unpublished data) or at time of first leaf full expansion (Myers 2005) to
exclosures in the forest understory and monitored weekly for 1 yr. These seedlings were protected from vertebrate herbivores.
"Percent of seedlings that survived at least 1 yr after germinating naturally in the forest understory(Wright et al. 2003).









Table 1-3. Effect of species and harvest time on material and structural properties of
seedling stems. Shown are F values from two way ANOVAs performed on
log-transformed values; d.f. = 7,1; ** P < 0.001
Biomechanical measurement Effect
Species Time Species*Time
Modulus of elasticity (MN m2) 151.5** 90.0** 1.5
Fracture toughness (J m-2) 70.5** 114.4** 9.7**
Stem tissue density (g cm-3) 219.8** 85.6** 19.6**
% NDF 57.7** 0.8 1.9
% Hemicellulose 25.3** 17.8** 4.0**
% Cellulose 30.7** 0.1 1.3
% Lignin 38.4** 2.0 1.4
% Critical height 60.8** 133.8** 4.8**
Flexural stiffness (N cm2) 111.3** 163.9** 4.7**


Work-to-bend (J)
Whole stem flexibility (radians/ N)


18.0**
100.9**


102.8**


5.81**










Table 1-4. Relationships among stem biomechanical traits for seedlings of eight tree
species.


E
(MN m2)
0.80
Toughness 0.80
(< 0.001)


Tough
(Jm-2)


Density % NDF
(g cm-3)


%Hemicell %Cellulose %Lignin


Density


0.77 0.73
(<0.001) (0.002)


0.90
% NDF 0
(<0.001)


% Hemicell


% Cellulose


0.75
(<0.001)


0.83
(<0.001)


0.51 0.58 0.46 0.40
(0.041) (0.024) (0.070) (0.128)


0.77
(<0.001)


0.70
(0.004)


0.60
(0.014)


0.86
(<0.001)


-0.11
(0.704)


S0.63 0.36 0.33 0.76 0.03 0.62
Ligni (0.009) (0.191) (0.213) (<0.001) (0.905) (0.011)


0.27
(0.316)
-0.68
(0.004)


-0.19
(0.491)
-0.71
(0.003)


-0.45
(0.083)
-0.40
(0.122)


-0.36
(0.165)
-0.59
(0.014)


-0.08
(0.837)
-0.44
(0.085)


-0.29
(0.284)
-0.67
(0.005)


0.14 0.18 -0.17 -0.05 0.14 -0.01
(0.612) (0.526) (0.519) (0.841) (0.593) (0.97)


Work-to-bend


-0.29
(0.535)


0.14
(0.760)


-0.57
(0.180)


-0.39
(0.383)


-0.25
(0.589)


-0.25
(0.589)


-0.28
(0.300)
0.43
(0.094)
-0.01
(0.970)
0.00
(1.000)


0.14 0.06 0.40 -0.34 -0.11 0.11 0.00
(0.612) (0.829) (0.140) (0.221) (0.704) (0.819) (1.00)


Seed mass (g)


0.21
(0.610)


0.33
(0.420)


0.24
(0.570)


0.14
(0.736)


0.29
(0.493)


0.21
(0.610)


0.10
(0.823)


% Hcr


WSF









Table 1-4. Continued
I (mm4)


% H El (N iM2) Work-to- WSF
bend (J)* (radians/N)
bend (J)* (radians/N)


Toughness

Density

% Total
fiber

% Hemicell

% Cellulose

% Lignin

I


% Hcr


0.17
(0.528)


0.87 -0.12
El
(<0.001) (0.667)
Work-to- 0.86 0.00 0.99
bend (0.014) (1.000) (<0.001)
-0.81 0.08 0.87 -0.857
WSF
(<0.001) (0.790) (<0.001) (0.014)
Seed mass 0.74 0.10 0.79 0.64 -0.82
(g) (0.037) (0.823) (0.021) (0.119) (0.023)
Shown are Spearman correlation coefficients for tests performed on species means
obtained at each harvest (N= 16) with P values in parentheses and significant
correlations in bold. For correlations with work-to-bend and seed mass, only values of Ti
were used (N= 8). Material properties included modulus of elasticity (E), fracture
toughness (tough), density, % NDF (non-detergent fiber), % cellulose, % hemicellulose,
and % lignin. Structural measurements integrating material traits and morphology
included % critical height (% Her), flexural stiffness (El), work-to-bend, and whole stem
flexibility (WSF). Second moment of area (1) considered only size and shape.






31


Table 1-5. Effect of species and harvest time on material and structural traits of leaves.
Shown are F values from two way ANOVAs performed on log-transformed
values; d.f.= 7,1; 0.001

Biomechanical measurement Effect
Species Time Species*Time
Lamina fracture toughness (J m-2) 32.9** 1.5 10.9**
Midvein fracture toughness (J m-2) 47.7** 49.9** 6.4**
Leaf density (g cm3) 44.0** 113.6** 15.1**
% NDF 5.5** 1.7 1.9
% Hemicellulose 36.2** 0.1 0.6
% Cellulose 6.9** 0.1 1.6
% Lignin 24.4** 10.4* 2.2
Specific leaf area (cm2g 1) 119.3** 46.6** 2.0
Force of fracture (N) 46.7** 2.4 5.0**












Table 1-6. Relationships among leaf biomechanical traits for seedlings of eight tree species.
Lamina Midvein Leaf % NDF % % % Lignin Leaf SLA
toughness toughness density Hemicell Cellulose thickness (cm2g-1)
(J m-2) (J m-2) (g cm-3) (mm)


Midvein
toughness
Leaf density

% NDF

% Hemicell

% Cellulose


0.68
(0.004)
0.52
(0.039)
0.21
(0.438)
-0.17
(0.520)
0.74
(< 0.001)


0.67
(0.005)
0.04
(0.897)
-0.11
(0.664)
0.67
(0.005)


0.31
(0.249)
0.15
(0.579)
0.33
(0.213)


0.30
(0.264)
0.29
(0.279)


-0.07
(0.787)


0.03 -0.33 -0.05 0.59 -0.25 -0.12
Lignin(0.914) (0.217) (0.846) (0.017) (0.350) (0.664)
Leaf 0.15 -0.03 -0.44 0.04 (-0.39) 0.29 0.17
thickness (0.580) (0.910) (0.087) (0.871) (0.131) (0.274) (0.535)
-0.58 -0.67 -0.56 -0.33 0.24 -0.61 -0.01 -0.34
SLA
(0.019) (0.005) (0.020) (0.209) (0.380) (0.012) (0.983) (0.200)
0.84 0.51 0.18 0.24 -0.41 0.79 0.17 0.61 -0.65
rce ) (< 0.001) (0.043) (0.513) (0.374) (0.119) (< 0.001) (0.535) (0.012) (0.006)
Shown are Spearman correlation coefficients for tests performed on species means obtained at each harvest (N= 16) with P values in
parentheses, and significant correlations in bold. Material properties included lamina toughness, midvein toughness, % NDF
(nondetergent fiber), % hemicellulose, % cellulose, % lignin, and whole-leaf tissue density. Structural variables integrating material traits
and morphology were specific leaf area (SLA) and force of fracture (Force), which was the product of leaf toughness and thickness.









Table 1-7. Relationships among biomechanical traits of stems and leaves for seedlings of
eight tree species.
Biomechanical trait r, P

Toughness (J m-2)
Lamina 0.45 0.092
Midvein 0.58 0.002
Density (g cm-3) 0.80 < 0.001
% NDF 0.58 0.019
Shown are Spearman correlation coefficients (rs) for tests performed on species means
obtained at each harvest (N=16) and corresponding P values.









Table 1-8. Relationships among % survival in shade and various seedling biomechanical
traits of stems and leaves for seedlings of eight tree species.


STEM
Material properties
E
Toughness
Density
% NDF
% Hemicellulose
% Cellulose
% Lignin
Structural properties
% Her
El
Work-to-bend
WSF
LEAF
Material properties
Lamina toughness
Midvein toughness
Density
% NDF
% Hemicellulose
% Cellulose
% Lignin
Structural properties
SLA
Force of fracture


Mean % survival refers
coefficients (rs) from te
T2 = time when biomec
expansion, respectively


0.79
0.60
0.93
0.74
0.50
0.48
0.38

-0.17
-0.17
-0.68
0.54


0.33
0.21
0.64
0.81
0.33
0.36
0.43


0.021
0.120
<0.001
0.037
0.120
0.233
0.352

0.693
0.693
0.094
0.215


0.420
0.610
0.086
0.015
0.420
0.385
0.289


0.69
0.89
0.90
0.71
0.52
0.45
0.60

-0.38
-0.12

0.19


0.76
0.31
0.93
0.45
0.14
-0.05
0.24


0.058
0.007
0.002
0.047
0.183
0.260
0.120

0.352
0.779

0.651


0.028
0.456
<0.001
0.260
0.736
0.912
0.570


-0.52 0.183 -0.60 0.120
0.29 0.493 0.048 0.911
to the first column in Table 1-2. Shown are Spearman correlation
sts performed on species means (N= 8) and their P values. T1 and
;hanical measurements were taken (1 and 6 mo after leaf
).











7000

S6000 A
Z*
r 5000

S4000

3000
4--
u2000

1000
0
0


-20000
E" B

-16000

(D
C
r 12000
0)

p 8000 -


LL 00


0
TABR ANAE CASE BEIP GUSS TETP EUGN ASPC
Species

Figure 1-1. Means (+ 1 SD) material biomechanical traits of stems for seedlings of eight
tree species ordered from left to right by increasing shade tolerance at 1 mo
(T1, filled bars), and 6 mos (T2, open bars) after leaf expansion. A) Young's
modulus of elasticity (E). B) Fracture toughness. Asterisks indicate significant
difference between T1 and T2 (P value < 0.006 with Bonferroni correction).
Refer to Table 1-1 for species codes. No data available for stem toughness of
GUSS at T2 because the size of the stems exceed size capacity of the tester
(3 mm diameter).










A
100%


80%


60%


40%


20%


0%


100% r--


80%


60%


40% -


20%


0%


m


Afr rL


% Non fiber


% Hemicell.


% Cellulose


% Lignin


TABR ANAE CASE BEIP GUSS TETP EUGN ASPC
Species
Figure 1-2. Mean (+ 1 SD) % fiber content (% NDF) for seedlings of eight tree species
ordered from left to right by increasing shade tolerance at 6 mos after first leaf
expansion (T2). A) Stems. B) Leaves. ANOVA results are shown in Table 1-3
for stems and Table 1-5 for leaves. For species codes refer to Table 1-1.


E-


F-P


% Non fiber


% Hemicell.


% Cellulose


-% Lignin


-E rFE-
































3.0 3.2 3.4 3.6


3.8 4.0 4.2 4.4


B r 2 =0.72**
0.2 Slope = 0.57


-0.4 -
A


A 0
- ANAE
* ASPC
A BEIP P
v CASE n
* EUGN
_ GUSS
O TABR v A
- TETP




0 r2= 0.72**
0 Slope =1.15


3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4


log Toughness (Jm -2)


-0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2

log Density (g cm 3)


log Toughness (Jm -2)


D r2 =0.80**
.85 Slope = 0.14

.85A


2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8

log Modulus of elasticity (MN m -2)


Figure 1-3. Log-log relationships between some material properties of stems for seedlings
of eight tree species. Each point is a species mean at 1 mo (T1, filled
symbols), and 6 mos (T2, open symbols) after leaf expansion; ** P < 0.001.
Refer to Table 1-1 for species codes.


-0.6 -


-0.8 -


-1.0
2.8


3.0

2.8

2.6

2.4

2.2

2.0
2.


8


C r2 =0.56**
Slope = 0.23


A a


v


0 0


1.85 F


LLi
01.80
z
Z

'1.75


1.70


1.65
-1.0


0 0


I I I I I I I


3











40
A

30
20












0) *
u 20


10









S40


20


0.005
S0.004
U) l



U-








0.003
o
0.002
0.001


0.000
0





D
0.15


0.10


0.00


0.00
TABR ANAE CASE BEIP GUSS TETP EUGN ASPC

Species
Figure 1-4. Means (+ 1 SD) structural biomechanical traits of stems for seedlings of eight
tree species ordered from left to right by increasing shade tolerance at 1 mo
(T1, filled bars), and 6 mos (T2, open bars) after leaf expansion. A) Percent
critical height. B) Flexural stiffness. C) Work-to-bend. D) Whole stem
flexibility (WSF). Asterisks indicate significant difference between T1 and T2
(P value < 0.006 with Bonferroni correction). Refer to Table 1-1 for species
codes. Work-to-bend was only measured at T1. Work-to-bend and WSF could
not be measured for TABR at T1 because of small size of seedlings.










-600
E A
E500

400 -
'- *
o 300

200

5 100
L- 0

5000

E B
4000

o 3000
'-c
z 2000
O

n 1000

LL 0

100

z C
Z 80


t 60
4--
0 40

o 20
i-

0
TABR ANAE CASE BEIP GUSS TETP EUGN ASPC
Species

Figure 1-5. Means (+ 1 SD) biomechanical traits of leaves for seedlings of eight tree
species ordered from left to right by increasing shade tolerance at 1 mo (T1,
filled bars), and 6 mos (T2, open bars) after leaf expansion. A) Lamina
fracture toughness. B) Midvein fracture toughness. C) Leaf force of fracture.
Asterisks indicate significant difference between T1 and T2 (P value <0.006
with Bonferroni correction). Notice the different scales between A and B.
Refer to Table 1-1 for species codes.











2.00

1.95

1.90

1.85

1.80

1.75

1.70

1.65


1.60 L
3.2


2.00


1.95

1.90

1.85

1.80

1.75

1.70

1.65


3.4 3.6 3.8 4.0 4.2

log Stem toughness (Jm 2)



C r2 =0.15P= 0.33
Slope = 0.31
A











0


2.0 2.1 2.2 2.3 2.4 2.5

log Lamina toughness (Jm-2)


-0.8 -0.7 -0.6 -0.5 -0.4 -0.3

log Stem density (g cm -3)


-0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -

log Leaf density (g cm -3)


Figure 1-6. Log-log relationships between some biomechanical properties measured at
6 mos after first leaf expansion (T2), and % mean survival in shade for
seedlings of eight tree species. A and B refer to stem traits, and C and D to
leaf traits. Each point represents a species mean; symbols as in Figure 1-3.


A r 2= 0.56P= 0.54
Slope = 0.28





V
rc


B r2 = 0.78 P= 0.004
Slope = 0.58



A



V


*


D r2 = 0.90 P< 0.001
Slope = 0.67


* /














CHAPTER 2
SPECIES DIFFERENCES IN SEEDLING SUSCEPTIBILITY TO BIOTIC AND
ABIOTIC HAZARDS IN THE FOREST UNDERSTORY

Introduction

For most plant species, mortality rates is highest during the seed and seedling

phases, thus early developmental stages are critical determinants of adult abundance and

distribution (Clark & Clark 1985, Condit, Hubbell & Foster 1995, Kitajima & Fenner

2000). Multiple mechanisms have been proposed to explain existing differences in

recruitment and survival across microsites in the forest understory, which lead to species-

specific distribution patterns and niche partitioning. For example, in the tropics

differential seedling survival in contrasting light environments has been related to

trade-offs involving growth, resource allocation, and defense (Kitajima 1994, Kobe 1999,

Montgomery & Chazdon 2002). Additionally, differences in seed size (Moles & Westoby

2004a,b), dispersal mechanisms (Howe & Schupp 1984, Chapman & Chapman 1996),

cotyledon function (Kitajima 1996, Ibarra-Manriquez, Martinez Ramos & Oyama 2001),

carbon allocation to storage (Canham et al. 1999, Myers 2005), and biomechanical

properties (Chapter 1) can all contribute to differential seedling survival across a light

gradient by allowing seedlings to avoid or respond to stresses in different ways.

A wide range of biotic and abiotic hazards resulting in physical damage can

seedling cause mortality. Numerous studies have addressed the importance of

invertebrate herbivory (Coley 1983, Aide & Zimmerman 1990, Benitez-Malvido,

Garcia-Guzman, & Kossmann-Ferraz 1999), pathogens (Auspurger 1983, 1984a),









vertebrate consumption and trampling (Osunkoya et al. 1992, Weltzin, Archer, &

Heitschmidt 1997, Gillman, Wright & Ogden 2002), and litterfall (Guariguata 1998,

Scariot 2000, Gillman et al. 2004) as mortality agents for seedlings. Mortality agents

such as disease, invertebrate herbivory, and vertebrate consumption are species-specific,

while mortality agents such as litterfall or vertebrate trampling affect seedlings from

different species indiscriminately. In other words, species vary in their susceptibility to

diseases or herbivores, but the chance of being trampled by animals or impacted by

falling litter is random. Additionally, all else being equal, species differences in mortality

due to severe mechanical hazards such as vertebrate trampling or large falling debris arise

mainly from their differential ability to respond and recuperate from damage, rather than

from the ability to evade it. In contrast, differential mortality among species due to

invertebrate herbivory, disease, or minor mechanical hazards (i.e., impact by small

litterfall) is also influenced by species differential ability to resist physical disturbances

by investing in strong materials and structures (Chapter 1). Species with high mechanical

strength in stems and leaves should 1) be affected less frequently by species-specific

damage agents such as herbivory and disease; 2) Suffer less and less intense mechanical

damage when affected by any mechanical damage agent; and, 3) die less frequently when

damaged by herbivores, disease, litterfall, and vertebrate trampling (Niklas 1992,

Chapter 1).

There are also differences in the types of damage inflicted on seedlings by damage

agents. Pathogens cause damping-off or tissue necrosis, while invertebrate herbivores,

vertebrate activity, and litterfall result in mechanical damage. Vertebrate activity often

results in bent, broken, uprooted or chewed stems, or in missing seedlings. Falling leaves









and branches can bend or break seedlings. Because seedlings face multiple stresses

simultaneously or over short periods, it is difficult to determine the damage type resulting

in the ultimate cause of mortality (Kitajima & Fenner 2000).

The main objective of my study was to describe first-year temporal patterns of

seedling mortality, susceptibility to damage agents, and types of damage in eight species

of tropical trees differing in their shade tolerance. Seedlings were transplanted to the

forest understory and temporal patterns of damage susceptibility analyzed by constructing

survivorship curves. Additionally, artificial seedlings constructed of plastic straws and

wire, were "planted" with the real seedlings and monitored to assess susceptibility to

damage agents from a community perspective. The artificial seedling approach has been

frequently used to estimate potential damage by litterfall and vertebrate trampling (Clark

& Clark 1989, McCarthy & Facelli 1990, Mack 1998, Scariot 2000, Roldan & Simonetti

2001, Drake & Pratt 2001, Gillman, Wright & Ogden 2002). Here, I further assess the

efficacy of the artificial seedling method for predicting real seedling damage and

mortality by comparing artificial and adjacent real seedlings that were transplanted to the

forest understory.

Materials and Methods

Study Site and Species

This study was conducted in a seasonally moist tropical forest, on Barro Colorado

Island (BCI), Panama (90 10' N, 790 51' W). Average annual precipitation is 2600 mm,

90% of which falls primarily between May and December (Windsor 1990). Ecological

characteristics of BCI are well described in Croat (1978) and Leigh, Windsor & Rand

(1982). The experiment took place in young forest (100-300 yrs, Foster & Brokaw 1982)









with abundant palms (mainly Astrocarium standleyanum and Oenocarpuspanamanus)

and evidence of frequent physical disturbances such as tree and branch falls.

Seeds of eight common tree species, that differ in ecological characteristics such as

dispersal mode, cotyledon type, and seedling establishment probability (Table 1-1) were

collected during the dry season of 2003 and germinated in a shade house. Seedlings from

the eight species differed in time of germination (from approximately 5 days for

Aspidosperma cruenta to 36 days for Eugenia nesiotica), number of leaves, and size

(Table A-3). Species with the largest seedlings at germination were Gustavia superba and

Anacardium excelsum, while the smallest was Tabebuia rosea. Species were selected

based on seed availability during 2003 on BCI from a range of shade tolerance (K.

Kitajima personal communication). Species ranged from shade-tolerant for an

intermediate period (T rose) to shade-tolerant with slow growth in shade (A. cruenta)

according to Augspurger (1984b). Additionally, according to a long-term study in BCI

the study species differ dramatically in their recruitment index, which is a measure of a

species ability to recruit seedlings that survive at least 1 yr relative to the number of seeds

dispersed to the same microsite throughout that year (Wright et al. 2003, Table 1-1). The

eight study species also varied substantially in the biomechanical properties of their stems

and leaves (Chapter 1), which I expected to lead to differential susceptibilities and

responses to mechanical damage.

Experimental Design

In June 2003, 755 seedlings from the eight study species were transplanted, at first

leaf expansion, to 100 stations located randomly along a 9 km network of trails. The

stations were at least 5 m away from the trail and separated by a minimum of 50 m. At

each station, one seedling of each species was planted within a 1 m2 area, and examined









every 1-2 wks for 8 mos and every 4 wks for four additional months, for a total of

22 censuses between June 2003 and May 2004. Because of limited seed availability,

T. rosea seedlings were transplanted to only 55 of 100 stations. During transplantation,

microsites were minimally altered (no litterfall or debris were removed). No evident

transplant shock (wilted or dried plants) was observed, presumably because abundant

rainfall during the transplanting period diminished the risk of desiccation.

In addition to the eight natural seedlings, two artificial seedlings made of plastic

and wire were "planted" at each station. The design of artificial seedlings followed Clark

& Clark (1989), but I used two sizes of artificial seedlings to evaluate the effect of

physical disturbance and mechanical damage on seedlings of different size classes. Each

large artificial seedling was made of two 200 mm-long transparent plastic straws oriented

in a cross and attached together with staples. A stiff, 3 mm-diameter x 100 mm-long wire

was inserted 20 mm into the vertical straw and the remaining 80 mm into the ground to

simulate a root (Figure 1 in Clark & Clark 1989). Small artificial seedlings were

constructed in the same way, except that they were made from 100 mm-long straws and

50 mm-long wire "root".

Survival, Damage Agents, and Types of Mechanical Damage

At each census, mortality, apparent damage agent, and damage types were

recorded. For real seedlings four main types of damage agents were recorded: vertebrate

activity, disease, litterfall, and missing without obvious indication of damage agent.

"Vertebrate activity" was characterized by plants that were uprooted, flattened, or having

damaged, broken, or chewed stems with no evidence of falling leaves or branches.

Although it was impossible to differentiate between vertebrate consumption and

trampling, chewed stems, missing cotyledons, and uprooted seedlings suggest that









vertebrate activity was mostly consumption-related. Seedlings were diagnosed as

"diseased" when they exhibited necrotic tissue, or when they were severely wilted with at

least one dry leaf. "Litterfall" was recorded when a seedling had a bent, damaged, or

broken stem and there was direct evidence of litterfall or debris above it. "Missing"

seedlings were those that could not be located and were presumed dead. This

classification scheme underestimates litterfall damage and overestimates vertebrate

activity, since all damaged stems without evidence of litterfall were considered as

damaged by vertebrates. Likewise, missing seedlings within a relatively short census

interval of 1-2 wks was likely to be caused by consumption of vertebrate browsers, also

leading to an underestimation of vertebrate activity. A seedling was considered dead

when it was completely dried, when the stem was cut in two and the lower portion

uprooted, or when the whole seedling was missing. The four damage agents were not

mutually exclusive. In fact, seedlings often died after being affected by two or more

agents. It was not the intent of my study to determine the ultimate cause of mortality, but

to describe the temporal patterns of these damage agents that may synergistically kill

seedlings.

I also recorded the first occurrence of the four main types of mechanical damage

that could be fatal: leaf damage, stem bent, stem broken, and seedling uprooted. A leaf

was considered "damaged" if it was fractured, incomplete or had missing sections larger

than 10% of the leaf area. A seedling with at least one damaged leaf was classified with

"leaf damage". A stem was considered "bent" if it was curved or tilted at least 450, and

"broken" if it was fractured in two or more sections. A seedling was uprooted when it

was completely pulled from the ground. Although the four types of mechanical damage









were not mutually exclusive, only the first type recorded on each plant was used for

calculation of "% damage fatality" at the end of the study. Furthermore, to be certain that

death (if it occurred) was caused by a particular damage agent, only plants affected by a

single damage type were considered. For each type of mechanical damage (Mx), I

calculated the likelihood of dying after receiving a given type of damage P(D Mx)

expressed as the following formula (Gothelli & Ellison 2004):

P(D M,) = P (D M) (2-1)
P (Mx)

where P(M,) is the probability of receiving the damage type M, and P(D) is the

probability of death. The conditional probability P(D M), multiplied by 100 and

expressed in %, was called "% damage fatality". I also evaluated the relationship between

material and structural stem properties and seedling susceptibility to mechanical damage,

by comparing species mean biomechanical traits measured in chapter 1 (Table 1-4) vs.

% damage fatality.

Artificial Seedlings

The artificial seedling method has been used in other studies to quantify damage

due to random disturbance agents such as litterfall and vertebrate trampling because

artificial seedlings are not susceptible to biotic species-specific agents of mortality such

as pathogens and herbivory. Here, artificial seedlings were censused simultaneously with

real seedlings to provide a comparison between real and artificial seedling damage and

mortality. An artificial seedling was considered damaged when it was bent such that at

least one of its arms was touching the ground, when it was flattened, cut, chewed, or

missing (Clark & Clark 1989). For comparison with previous studies, three standardized

categories of damage were recorded for artificial seedlings: vertebrate activity, litterfall,









and unknown. An artificial seedling was classified as damaged by "vertebrate activity" if

it was flattened, chewed, cut, or missing with no evidence of litter or branch fall that

could have caused the damage. If an artificial seedling was bent or flattened, with

evidence of litterfall or debris above it, it was considered damaged by "litterfall". If a

seedling was bent but there was no obvious cause, it was classified as "unknown".

Therefore, the estimate of damage caused by litterfall is conservative, since it only

reports artificial seedlings that were damaged by conspicuous litterfall and debris.

I compared the first occurrence of damage agents affecting real vs. artificial

seedlings. Because missing artificial seedlings were included in the vertebrate activity

category in previous studies (Gillman, Wright & Ogden 2002), I included missing real

seedlings in the vertebrate activity category, such that when comparing artificial

seedlings and real seedlings vertebrate activity refers to seedlings flattened, chewed, cut,

or missing. In any case, it is likely that seedlings that suddenly disappeared (in an interval

of one week) were eaten or uprooted by vertebrates. Additionally, when comparing real

and artificial seedlings the "unknown" category was obtained from seedlings that were

recorded as bent with no further evidence of damage; the "litterfall" category remained

the same.

Statistical Analyses

Temporal patterns of the occurrence of seedling death and damage agents were

analyzed using non-parametric Kaplan-Meier survival distribution functions (Collett

2003). Survival functions (also called hazard functions) describe the probability that an

individual survives longer than a specified period, considering individuals at risk at the

beginning of each interval and excluding censored values. An interval is defined as the

lapse between two mortality events. Censors are individuals that "left" the study before









its conclusion (e.g., removed from the study) or individuals that were not dead at the time

of finalization of the experiment. In my study, censored plants were only those that were

alive at the end of the observation period. From the probability of hazard occurrence

during each census interval, survivorship curves throughout the entire period were drawn,

plotting the proportion of seedlings unaffected by the respective damage agent against

time. To observe each species behavior throughout different periods, additional

survivorship curves were drawn considering shorter intervals, of 0-2 mo, 2-6 mo, and

6-12 mo. The log-rank test and the Wilcoxon test (Pyke & Thompson 1986, Collett 2003)

were used to compare survival distribution functions for different species. The log-rank

test is more sensitive to differences in late survival times, while the Wilcoxon test is more

sensitive to differences in early survival times. However, here both tests had similar

outcomes, and therefore only results from log-rank tests are reported. For across-species

comparisons between survival proportions at the end of each period, Spearman rank

correlation coefficients were used. For interspecific comparison within types of damage,

likelihood chi-squared tests were used. Lastly, nonparametric Spearman correlation tests

were used to compare species mean biomechanical and ecological traits vs. % damage

fatality. For all analyses c = 0.05 and all were performed JMP IN 4.0 (SAS Institute Inc.,

Cary, NC, USA).

Results

Seedling Survival

Within the first 2 mos 59% of the transplanted seedlings died, and by the end of

1 yr 76% of transplanted seedlings were dead. The temporal pattern of mortality and

overall % mortality at the end of the first year differed among species (Figure 2-1).









Survival distribution functions varied among species for the entire 1 yr period (log-rank

x2 = 220.2, d.f. = 7, P < 0.001; Figure 2-1D), and within each of shorter intervals

(0-2 mo: log-rank X2 = 220.2, d.f. = 7, P < 0.001; 2-6 mo: log-rank X2 = 55.4, d.f. = 7, P <

0.001; 6-12 mo: log-rank 2 = 49.6, d.f. = 7, P < 0.001; Figures 2-1A to 2-1C). In

addition, species rankings of survival probability switched between intervals. Survival for

0-1 yr was determined mainly by survivorship during the first 2 mos, which differed from

survival in the following intervals. This is demonstrated by the high correlation between

the 0-2 mo period and the overall lyr survival, and the lack of correlation between the

0-2 mo period with both the 2-6 mo and the 6-12 mo periods (Table 2-1).

Damage Agents

Vertebrate activity was the most common damage agent (Figure 2-2). Survival

functions differed significantly among damage agents calculated for all species combined

(log-rank X2 = 1496.9, d.f. = 4, P < 0.001; Figure 2-2A). In addition, damage agents

affected species differentially. The percentage of seedlings affected by vertebrate activity

after 1 yr ranged from 31% for Beilschmiedia pendula, to 65% for E. nesiotica (log-rank

72= 56.0, d.f. = 7, P <0.001; Figure 2-2B). However, there were more missing seedlings

of B. pendula than of the other species (log-rank 2 = 209.3, d.f. = 7, P <0.001; Figure

2-2C) and most of these events happened in the first four weeks after transplant. In B.

pendula, 87% of the seedlings would have been affected by vertebrate activity if all

missing seedlings were included in the vertebrate activity category. Tetragastris

panamensis, A. excelsum and, T rosea were the species most affected by disease.

Although, T. rosea, and A. excelsum were affected by disease mostly in the first 2 mos

after transplant, T panamensis had a constant intensity of infection (log-rank X2 =98.7,









d.f. = 7, P <0.001; Figure 2-2D). Only 4.1% of the seedlings were affected by litterfall,

with no interspecific differences in damage agent distribution over time (log-rank X2

13.7, d.f. = 7, P >0.05).

Types of Mechanical Damage

Four types of mechanical damage were recorded during 1 yr: damaged leaves, bent

stems, broken stems, and uprooted seedlings, caused by vertebrates, litterfall, or

invertebrate herbivores (in the case of leaf damage). At the end of the study, 77% of

seedlings showed some form of damage, of which leaf damage was the most frequent.

After 1 yr, 30.6% exhibited leaf damage, 28.7% of the seedlings had broken stems,

23.9% had bent stems, and 25.6% had been uprooted (Table 2-2). These categories were

not mutually exclusive. In fact, 45% of the damaged seedlings had two or more types of

damage.

Species differed in their likelihood to die after suffering leaf damage or bent stems,

but not after being uprooted or having their stem broken. Percent damage fatality (as

defined by Formula 2-1) differed among species for leaf damage or bent stem (leaf

damage: 2 = 78.6, d.f. = 7, P <0.001; stem bent: 2 = 57.3, d.f. = 7, P <0.001). In

contrast, there was no interspecific difference in damage fatality for uprooted seedlings or

those with broken stems (uprooted: 2 = 14.0, d.f. = 7, P = 0.052; stem broken: 2 = 9.4,

d.f. = 7, P = 0.226). It should be pointed out that because of the low number of uprooted

and broken seedlings in some species (cell N< 5), results should be interpreted with

caution. From all the biomechanical and ecological measurements tested (Table 1-4 in

Chapter 1), only stem toughness, stem tissue density, and second moment of area were

correlated with damage fatality (Table 2-3). Tougher, denser stems died less when their









stem was broken or bent than stems with weaker, less dense material. In contrast, species

with larger stems (i.e., with large second moment of area) were more likely to die when

their stem was broken.

Artificial Seedlings

After 1 yr, 9.5% of artificial seedlings were damaged by litterfall, 15.5% were

damaged by vertebrate activity, and 22.5% were bent by unknown causes. Thus, damage

levels on artificial seedlings were within the range of damage reported for other sites

(Table 2-4). Overall, real seedlings were damaged more than artificial seedlings

(X2= 64.3, d.f. = 1, P <0.001; Figure 2-3) Artificial seedlings were damaged more by

litterfall and by unknown causes, and less by vertebrate activity than each of the species

of real seedlings. Large and small artificial seedlings did not differ in their damage

frequencies (X2 = 6.0, d.f. = 3, P = 0.111), and therefore they were averaged for

comparisons with real seedlings and with previous studies.

Temporal patterns for each type of damage differed between artificial and real

seedlings (Figure 2-4). Vertebrate activity damaged a much higher proportion of real than

artificial seedlings, especially during the first 50 days. If missing seedlings were also

considered to be affected by vertebrate activity, the difference became even stronger

(log-rank X2= 118.1, d.f. =1, P <0.001; Figure 2-4A). Real seedlings were particularly

vulnerable to vertebrate activity during the first 2 mos after transplant, but artificial

seedlings received a more constant rate of vertebrate damage. Litterfall damage was less

than 10% for both artificial seedlings and real seedlings. Artificial seedlings were less

affected during the first 6 mo, but the trend reversed during the subsequent period (log-

rank X2= 6.5, d.f. =1, P = 0.011; Figure 2-4B). In addition, survival functions describing









the proportion of seedlings damaged by unknown causes, differed between artificial and

real seedlings. Although real seedlings were more affected during the first 3 mo, the trend

inverted and at the end of 1 yr, a larger proportion of artificial seedlings were affected by

unknown causes (log-rank 2 = 9.79, d.f. =1, P = 0.002; Figure 2-4C).

Discussion

Survival, Damage Agents, and Types of Mechanical Damage

The combination of species-specific and indiscriminate damage agents, their

temporal patterns, and the differential susceptibility to damage among species influences

seedling performance in the forest understory. Consistent with other studies (Augspurger

1984a, Kitajima & Augspurger 1989, De Steven 1994), the proportion of seedlings dead

after 2 mos was higher than in the 2-6 or 6-12 mo periods. Seedling mortality was highest

during the first 2 mos after transplant, decreasing gradually and then becoming more

constant over the remaining 8 mos (Figure 2-1). Although this was the trend for the eight

species individually, the species mortality ranks during the 0-2 mo interval differed from

the ranks during the 2-6 mo interval, or the 6-12 mo interval (Table 2-1). This

observation suggests that mortality agents that had a greater effect during the first 2 mos

became less important in the following 8 mo.

Vertebrates were the most common cause of damage overall, especially during the

0-2 mo period (Figure 2-2). Vertebrate activities included non-trophic interactions such

as trampling, and trophic interactions including leaf herbivory and cotyledon

consumption. During the initial 2 mo, low percentages of leaf herbivory (i.e., leaf

damage) and high percentages of seedlings uprooted but left partially uneaten, with stems

cut in half and cotyledons missing, suggest that cotyledon predation by vertebrates was

the primary cause of mortality. Predation of large storage cotyledons has been recorded









as an important cause of mortality in previous studies (Sork 1987, Molofsky & Fisher

1993). Here, six of the eight study species have large-seeds (Table 1-1) and abundant

cotyledon reserves that can attract vertebrate consumers, even months after germination

(Smythe 1978). For example, B. pendula, an animal-dispersed species with large reserve

cotyledons, suffered extremely high mortality during the first month after transplanting.

Within the first week after transplanting 42% of the seedlings were missing, and by the

end of the first month, 56% of the seedlings were missing and presumed dead. Cotyledon

predation was potentially enhanced by soil disturbance during transplant, which could

have attracted agoutis (Dasyproctapunctata). However, it has been shown that agoutis

find buried seeds using predominantly olfactory cues (Smythe 1978). Thus, further

investigations comparing cotyledon consumption in naturally germinated vs. transplanted

seedlings are required to reach definitive conclusions.

Contrary to vertebrate damage, which affected almost 40% of seedlings during the

first 2 mo, disease affected less than 10% of seedlings throughout the study. This

observation differs from disease prevalence reported by Augspurger (1984a), who studied

naturally germinated seedlings of nine wind-dispersed species on BCI. She found that the

largest fraction of early seedling mortality under shaded conditions was due to pathogens.

However, Augspurger (1984a) studied naturally germinated seedlings that were already

established, diminishing the probability of recording early cotyledon predation, similar to

that reported for B. Pendula here. Nevertheless, the two species (A. cruenta and T. rose)

overlapping between studies exhibited similar levels of mortality at the end of 2 mos.

Mortality recorded for A. cruenta in Augspurger (1984a) and in my study was 20% and

27%, respectively. For T. rosea mortality was 65% in both studies.









Differences in survivorship between species reflected the interaction between

likelihood of damage and the ability to tolerate damage-induced stress. Contrary to my

expectation shade-tolerant species with stronger material properties (i.e., tougher stems

and higher stem tissue density), did not differ from species with weaker material

properties in the probability of suffering mechanical damage, but once damaged they

were less likely to die than species with weaker stems (Table 2-2). Contrary to this

pattern, A.cruenta, a shade-tolerant species with weak stems (Chapter 1), was the least

likely to die after suffering mechanical damage, suggesting that other factors, such as

carbohydrate storage reserves in stems and roots play an important role in the ability of

seedlings to tolerate mechanical damage (Myers 2005). Additionally, the probability of

survival after being damaged (measured as damage fatality) was different depending on

the type of damage received. For example, seedlings that were uprooted or had broken

stems usually died, while seedlings that suffered leaf damage or bent stems were more

likely to survive. My results (and previous studies, Marquis & Braker 1994) suggest that

leaf damage (caused by leaf herbivory) constitutes a less severe stress than stem bending

and breakage.

Numerous studies have emphasized the benefits of having a large seed, resulting in

a large seedling and increased survival (Paz & Martinez- Ramos 2003, Green & Juniper

2004, Moles & Westoby 2004b). In contrast, in my study there was no correlation

between seed size and seedling survival, susceptibility to damage agents, or incidence of

different types of damage (Table 1-4). One possible reason is that on BCI large seeded

species face strong pressures from vertebrate cotyledon predators. Barro Colorado Island,

due to the absence of top predators, supports high densities of medium sized mammals









(Glanz 1982) that can have a large effect on tree regeneration patterns through high

predation rates on large seeded species before and after germination (Asquith, Wright &

Clauss 1997). Additionally, the likelihood of death after stem breakage increased with

stem size (Table 2-3), suggesting that large size is not always beneficial.

Artificial Seedlings

The comparison between artificial vs. real seedlings revealed that on BCI damage

agents affecting artificial seedlings were not good predictors of damage agents affecting

the real seedling community (Figure 2-3). Moreover, artificial seedling damage was not a

good predictor of mortality for transplanted seedlings because real seedlings were more

affected by vertebrate consumption than artificial seedlings, especially during the first 2

mos. However, artificial seedlings were more affected by litterfall, in terms of cumulative

damage, after 1 yr. Consistent with my results, Gillman, Wright & Ogden (2002) found

that artificial seedling damage was not a good predictor of mortality of naturally

germinated seedlings in five evergreen temperate forests in New Zealand. Because

artificial seedlings are not significantly consumed by vertebrates, they would be accurate

predictors of mechanical damage, only in environments where indiscriminate, non-

trophic damage agents (e.g., litterfall, vertebrate trampling) are more frequent than

trophic interactions.

Temporal patterns of mechanical damage also differed between real and artificial

seedlings. Real seedlings were severely affected by vertebrate activity during the first

2 mos with a rapid decline afterwards, while artificial seedlings experienced a relatively

constant rate of damage (Figure 2-4). This difference suggests that damage agents that

affect seedlings indiscriminately (e.g., litterfall) became important after the early

establishment period when seedlings suffer heavily from vertebrate consumers.









The total percentage of artificial seedlings damaged on BCI after 1 yr, was within

the range reported in other studies (Table 2-4). However, the percentage (9.5 + 2.1% yr-)

of artificial seedlings damaged by litterfall on BCI was lower than the percentage

reported in most other tropical forests, possibly due to differences in rainfall, canopy

composition, and topography (Van Der Meer & Bongers 1996,Gillman et al. 2004).

Seedlings in the old secondary forest on BCI suffered an intermediate frequency of

mechanical damage. La Selva, Costa Rica (Clark & Clark 1989) exhibited the highest

percentage of damaged artificial seedlings (82.4% yr-1), and the intensive hunted forest in

Beni, Bolivia (Roldan & Simonetti 2000) the lowest (25% yr-1). However, in the Bolivian

study, litterfall damage was not considered. The large variation between studies suggests

that either the probability of being affected by mechanical damage differs widely across

forest communities, or researchers used different criteria to classify agents of mechanical

damage. Interestingly, the most frequent agent of mechanical damage reported in each

study, was usually the focal agent of interest for each author.

Conclusions

Survivorship analyses revealed that diverse ecological pressures, such as

vertebrate predation and disease affect seedlings differentially through time and among

species. For example, seedlings from T. panamensis suffered little seedling mortality due

to vertebrate predation during the initial 2 mo, but became more susceptible to disease in

the following period. In contrast, seedlings from E. nesiotica and B. pendula were

severely damaged by vertebrate predation in the initial 2 mo, but attained high survival

rates if they escaped predation. Furthermore, species such as A. cruenta that suffered little

mechanical damage overall can be limited by other factors, such as seed production and

dispersal (Augspurger 1984a). Additionally, differences among damage agents affecting









real and artificial seedlings indicate that litterfall can become an important cause of

damage after the early establishment period. Although there was no relationship between

shade tolerance and the types of mechanical damage affecting each species,

shade-tolerant species with stronger stems survived more often after damage, suggesting

that investment in strong stems is beneficial for seedling performance. The combined

effects of species-specific and indiscriminate damage agents, and species differences in

the responses to damage determine seedling performance. The resulting differences in

survival allow different species to succeed under different ecological conditions,

ultimately contributing to plant diversity in tropical forests.









Table 2-1. Relationships among species rankings of survival probability during the
specified interval for seedlings of eight tree species.
0-2 mo 2-6 mo 6-12 mo
0.50
2-6 mo
(0.207)
0.55 0.90
6-12 mo
(0.160) (0.002)
0.90 0.79 0.83
0-12 mo
(0.002) (0.021) (0.010)
Order of species codes listed in Fig 2-1 A-C. Shown are Spearman correlation
coefficients with their corresponding P values in parentheses and significant correlations
in bold.




Table 2-2. Percent damage fatality of four types of mechanical damage on eight tree
species during 1 yr in the forest understory.


Leaf damage


Stem bent


N % Fatality N % Fatality


20 100.0
37 97.3
45 64.4
7 100.0
47 44.7
35 48.6


20 100.0
39 100.0
17 76.5
9 88.9
25 48.0
19 57.9


22.2
32.3


Stem broken


Uprooted


N % Fatality N % Fatality


4 100.0
36 100.0
31 100.0
22 100.0
36 94.4
19 94.7


57.1
47.8


93.2
92.0


22 95.5
36 100.0
16 100.0
16 100.0
35 85.7
9 100.0


97.6
88.9


Sp.
code
TABR
ANAE
CASE
BEIP
GUSS
TETP
EUGN
ASPC


Total 231 61.5 180 72.2 217 96.3 193 95.3
N= total number of seedlings affected by each damage type. Damage types are not
mutually exclusive. For species codes refer to Table 1-1.









Table 2-3. Relationships among stem biomechanical traits and % damage fatality for
seedlings of eight tree species.
Types of mechanical
damage
Biomechanical properties Stem broken Stem bent
-0.36 0.61
Modulus of elasticity (0.3 0.1
(0.388) (0.108)
-0.77 0.82
Stem toughness -
(0.044) (0.024)
-0.79 -0.71
tem density (0.019) (0.048)
-0.406 -0.69
% NDF
(0.318) (0.056)
Second moment of 0.79 0.61
area of stem (0.019) (0.108)
0.61 0.54
Flexural stiffness0.10.
(0.109) (0.169)
Shown are Spearman correlation coefficients for tests performed on species means
(N= 8) with P values in parentheses and significant correlations in bold;
% NDF = percent non-detergent fiber.










Table 2-4. Percentage of artificial seedlings affected by specified damage agents in this
and other published studies in different forest communities.
Means (yr1) + 1 SD when applicable.


Forest type


Tropical wet
forest


Tropical wet
forest


La Selva,
Costa
Rica

Crater
Mountain,
Papua New
Guinea


N

y.


C)


Study


1 yr 500 19.2 21.0 42.2 17.6 Clark & Clark
(1989)


1 yr 418 13.8 7.0 11.0 65.3 Mack (1998)a


Seasonal
tropical forest


Seasonal
tropical forest


Tropical
terra fire
forest


Montane
tropical forest








Evergreen
temperate
forest


Temperate
forest


BDFFP,
Manaus,
Brazil

BCI,
Panama

OHF
Beni,
Bolivia
IHF
Beni,
Bolivia

(+ Pigs)
Mauna
Loa,
Hawaii
(- Pigs)
Mauna
Loa,
Hawaii

North
Island,
New
Zealand

New
Jersey,
USA


lyr 100 21.7 9.7 8.6 60 Scariot
(2000) b


1 yr 100 9.5
+2.1


6 mo 500


500


15.5
+4.9


22.5
+4.9


52.5
+6.4


This study c


80 5 15 Roldan &
Simonetti
(2000) d
13 12 75


1 yr 150 15.3 4.7 11.3 68.7 Drake & Pratt
(2001)e


150 20 0 0 80


2 yr 200 6.1 2.8
+5.8 +2.2


10 mo 200


2 39


Gillman,
Wright &
Ogden
(2002)f


13 44 McCarthy &
Facelli
(1990)g






62


a Total shown adds to 99.1% seedlings, and the remaining 02.9% artificial seedlings were
damaged by water erosion. b data for continuous forest averaged for the three study sites.
BDFFP = Biological Dynamics of Forest Fragments Project. C Average from large and
small artificial seedlings damaged ( 1 SD). BCI = Barro Colorado Island. See text for
methodological details. d OHF = Occasionally Hunted Forest, IHF = Intensively Hunted
Forest; approximate damages were taken from Figure 2, since exact numbers were not
shown in the study. Authors did not record % litterfall damage. e Half of the artificial
seedlings were fenced to exclude pigs (- pigs). f Shown are averages (+ 1 SD) from five
sites included in the study. Only non-trophic vertebrate activity was recorded.
Undamaged or unknown fractions were not reported. g Only forest habitat data are
shown. Total adds to 98% seedlings, the remaining 2% AS were damaged by "frost
heaving".












1


'E
S0.8


c
0.6
0
o 0.4
L..

a 0.2

0


ANAE
BEIP


0 10 20 30 40 50 60


.- EUGN
\ -- .. -- ASPC
GUSS

-- -.--~- -TETP
BEIP
CASE

ANAE
TABR
B


60 80 100 120 140 160 180


S---- ------------
-
N^. .


240 270 300 330 360
Time (days)


0 50 100 150 200 250 300 350
Time (days)


Figure 2-1. Kaplan-Meier survivorship curves for seedlings of eight tree species transplanted to the forest understory. Survivorship is
relative to the number of seedlings alive at the beginning of each period; A) 0-2 mo, B) 2-6 mo, C) 6-12 mo, and D) total
study period (0-12 mo). For species codes refer to Table 1-1.


--A
--N-,____


1


._0.8
>

~0.6


E 0.4
L..
o-
0-"0.2


EUGN
ASPC
TETP
GUSS
CASE
BEIP


TABR
ANAE


180 210













B. Vertebrate




---- -------- -----------


-- --. --- -
,,,


C. Missing


------------


Others
TETP
CASE



BEIP


0 50 100 150 200 250 300 350 400
Time (days)


D. Disease

--


Others

TETP
ANAE
TABR


0 50 100 150 200 250 300 350 400
Time (days)


Figure 2-2. Kaplan-Meier survivorship curves (proportion of seedlings yet to be hit by specified damage agents plotted against
time) for seedlings of eight tree species transplanted to the forest understory. A) Comparison of agents of
mechanical damage for the pooled data of the eight study species. B) Vertebrate activity on each species (uprooted,
flattened, chewed or broken stems; see methods). C) Seedlings missing most likely due to consumption by
vertebrates and presumed dead. D) Disease. For species codes refer to Table 1-1.


1


-0
S0.8

C
0.6
o
* 0.4
o
Q..
0
Q 0.2


.Death by
all causes


BEIP
TETP
ASPC
CASE
TABR
ANAE
GUSS
EUGN


0 1
(D
( 0.8

S0.6
c

0
* 0.4

n_ 0.2










100%

80%
80% Intact

60%

40% Unknown

20% Litterfall
I Vertebrate
0% activity



I II I
Real seedlings AS


Figure 2-3. Percent of real and artificial seedlings (AS) damaged during 1 yr in the forest
understory by specific damage agent. Small and large artificial seedlings were
pooled as "both AS" (N= 200), and real seedlings were pooled as "all real"
(N= 755). Categories are mutually exclusive, as each seedling was assigned
only to the first damage agent it experienced. For species codes refer to
Table 1-1.














-0 1
(D

S0.8
CU
S0.6

r-0.4
0
C..
0
_0.2

0






1
0

0.9
C
o

00.8
2
0

0.7






S1
-o


C

o
&0.s

0

0_


Real


AS


Real


'- .------. -AS


0 50 100 150 200 250 300
Time (days)


350 400


Figure 2-4. Kaplan-Meier survivorship curves for mechanical damage experienced by
artificial (AS) and real seedlings during 1 yr in the forest understory.
A) Vertebrate activity (estimated excluding missing seedlings and assuming
those missing seedlings were all due to vertebrate consumption). B) Litterfall
damage. C) Bent by unknown causes. For real seedlings, the average from
eight species is shown. Notice the different scales in y-axes.


B. Litterfall















APPENDIX
SPECIES MEANS AND STANDARD DEVIATIONS FOR BIOMECHANICAL
MEASUREMENTS, FIBER ANALYSIS, AND BIOMASS










Table A-1. Biomechanical measurements of seedling stems from eight tree species at T1
and T2 (1 and 6 mos after first leaf expansion, respectively). Shown are means
( 1 SD) and N= number of individuals. For species codes refer to Table 1-1.


Modulus of elasticity
(MN m2)


Fracture toughness
(J m2)


Density (g cm 3)











Flexural stiffness
(N cm2)


Species
ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


Tl
Mean
255.06
356.16
507.93
336.08
3590.92
916.04
171.72
1674.51


1551.82
2476.14
3430.16
1325.15
5433.04
5463.60
1135.16
5692.55


0.15
0.52
0.35
0.14
0.47
0.29
0.13
0.44


23.85
3.38
7.76
4.22
1.43
36.15
0.24
3.96


SD
90.66
126.64
102.44
110.51
768.37
178.94
107.84
402.98


414.41
709.15
636.75
353.23
1064.57
953.86
672.99
869.12


0.03
0.14
0.09
0.04
0.13
0.04
0.03
0.06


12.96
1.30
3.25
1.69
0.96
11.75
0.23
1.21


T2
Mean
451.10
768.01
944.71
779.27
5265.72
1244.89
331.35
2269.71


1883.80
7471.83
6015.59


SD
108.15
289.30
488.98
255.60
1215.35
354.93
193.40
538.67


560.45
2466.70
1760.58


0 2657.60 842.52
9 16004.96 2914.35


3831.43
5778.97


0.19
0.39
0.40
0.25
0.58
0.38
0.24
0.43


52.11
12.19
16.38
13.79
7.69
84.46
2.80
10.16


1774.03
2245.26


0.04
0.07
0.10
0.05
0.14
0.07
0.10
0.07


29.23
9.25
7.73
6.73
2.38
41.12
2.17
4.85










Table A-1. Continued


Whole stem
flexibility
(Radians/N)


Work to bend (J)


% Critical height


Species
ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP

ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


Tl
N Mean SD
10 0.034 0.015
15 0.062 0.017
14 0.024 0.008
15 0.061 0.041
10 0.15 0.035
15 0.009 0.003

15 0.069 0.036

13 0.0034 0.0016
10 0.0008 0.0005
9 0.0019 0.0008
7 0.0010 0.0006
9 0.0007 0.0004
11 0.0041 0.0012

9 0.0016 0.0006


14 28.36 6.03
15 22.31 2.71
13 21.05 3.35
15 20.69 4.35
15 14.49 2.60
14 13.68 1.90
14 18.46 2.84
15 17.87 2.21


T2
N Mean SD
10 0.014 0.004
12 0.018 0.008
10 0.016 0.005
11 0.03 0.021
12 0.092 0.052
11 0.007 0.002
4 0.087 0.076
8 0.055 0.023


14 23.35 4.53
10 16.93 2.43
14 16.62 3.80
14 19.78 4.82
14 14.99 1.97
9 14.31 3.21
15 16.66 6.18
7 14.44 1.95










Table A-2. Biomechanical measurements of seedling leaves from eight tree species at T
and T2 (1 and 6 mos after first leaf expansion, respectively). Shown are means
( 1 SD) and N= number of individuals. For species codes refer to Table 1-1.


Lamina fracture
toughness (J m 2)










Midvein fracture
toughness (J m 2)


Force of fracture (N)


Density (g cm-3)


Species
ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


Tl
Mean
274.12
371.28
130.34
71.29
191.55
395.43
117.78
339.27


2198.47
2370.67
1405.83
984.15
1609.67
3124.26
1521.65
2217.99


43.27
74.52
38.45
12.41
20.71
56.81
15.93
41.53


0.25
0.29
0.20
0.20
0.33
0.30
0.21
0.36


SD
75.12
92.76
60.45
39.63
59.37
150.29
90.17
135.27


419.80
455.09
494.77
183.35
461.49
426.99
387.46
348.91


15.85
19.97
18.34
8.00
8.36
20.53
12.08
12.72


0.04
0.02
0.03
0.03
0.04
0.02
0.06
0.06


T2
Mean
225.73
324.28
229.19
107.40
245.71
218.39
189.49
198.51


2532.61
2545.19
2271.53
1001.03
2417.33
2906.78
2194.12
3475.15


40.58
63.89
32.23
11.84
20.04
30.54
28.67
25.49


0.30
0.34
0.48
0.28
0.45
0.35
0.23
0.42


SD
71.25
59.60
51.23
35.21
32.90
58.59
46.37
71.58


360.55
403.04
477.84
185.94
501.64
582.72
427.60
666.77


14.92
13.64
6.84
4.73
3.72
8.93
10.54
8.80


0.04
0.02
0.07
0.04
0.04
0.03
0.05
0.04







71


Table A-2. Continued
Tl T2
Specific leaf area Species N Mean SD N Mean SD
(cm2 g-) ANAE 15 317.02 36.46 13 257.89 28.66
ASPC 15 193.95 14.78 14 172.03 8.88
BEIP 13 248.38 47.50 15 196.52 29.83
CASE 15 486.00 83.97 6 427.39 102.53
EUGN 15 360.38 24.02 8 305.14 36.79
GUSS 16 281.02 19.03 9 248.32 28.84
TABR 4 478.64 130.97 18 501.46 90.48
TETP 13 287.33 26.60 10 231.80 27.27







72


Table A-3. Fiber fractions of seedling stems from eight tree species at T1 and T2 (1 and
6 mos after first leaf expansion, respectively). Shown are means (+ 1 SD) and
N= number of samples (each sample composed of 5-15 seedlings); percent
NDF = Non detergent fiber. For species codes refer to Table 1-1.


% NDF


% Hemicellulose


% Cellulose


% Lignin


Species
ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


Tl
Mean SD
48.60 2.49
51.33 1.60
66.83 1.40
52.95 3.20
76.36 1.04
62.29 0.88
48.38 -
67.90 1.24

8.54 1.00
13.95 0.82
11.98 0.43
13.71 0.82
14.34 0.09
14.89 0.60
10.12 -
13.09 0.22


27.94 1.13
26.52 0.70
29.22 1.76
27.01 0.54
37.32 0.32
34.19 1.23
25.33 -
40.22 1.90


12.12 0.44
10.85 0.60
25.63 1.59
12.23 3.04
24.71 0.73
13.21 0.94
12.93 -
14.59 3.06


T2
Mean SD
48.43 1.31
55.28 3.20
63.57 1.68
57.13 3.03
74.41 3.46
62.43 0.42
53.01 8.57
64.46 2.19

10.28 0.66
15.93 0.54
14.71 1.54
18.82 1.57
14.39 2.65
13.47 0.60
11.58 2.13
13.20 0.47


26.18 1.01
27.80 2.78
28.54 0.98
26.66 0.40
37.59 5.54
34.11 0.54
30.58 5.17
37.21 2.51


11.97 0.36
11.55 1.13
20.33 2.31
11.65 2.24
22.43 1.03
14.85 1.08
10.85 1.27
14.06 0.44










Table A-4. Fiber fractions of seedling leaves from eight tree species at T1 and T2 (1 and
6 mos after first leaf expansion, respectively). Shown are means (+ 1 SD) and
N= number of samples (each sample composed of 5-15 seedlings); percent
NDF = Non detergent fiber. For species codes refer to Table 1-1.


% NDF


Tl
Species N Mean SD
ANAE 3 42.08 1.16
ASPC 3 52.32 1.87
BEIP 3 54.87 1.95
CASE 3 51.84 0.81
EUGN 3 56.93 4.59
GUSS 3 54.77 0.89
TABR 1 43.83 -
TETP 3 54.73 1.85


ANAE 3 7.77 1.15
ASPC 3 14.21 0.58
BEIP 3 9.68 0.55
CASE 3 20.43 0.47
EUGN 3 19.05 2.12
GUSS 3 16.37 0.95
TABR 1 10.78 -
TETP 3 12.63 0.18


ANAE 3 18.43 1.16
ASPC 3 20.62 1.03
BEIP 3 16.96 0.48
CASE 3 16.73 1.34
EUGN 3 17.45 0.30
GUSS 3 22.65 0.30
TABR 1 14.51 -
TETP 3 24.18 1.14


ANAE 3 15.87 0.82
ASPC 3 17.48 0.56
BEIP 3 28.23 1.21
CASE 3 14.67 1.76
EUGN 3 20.43 2.24
GUSS 3 15.75 0.20
TABR 1 18.54 -
TETP 3 17.91 1.38


T2
N Mean SD
3 43.11 1.58
3 54.23 4.12
3 55.87 0.31
3 44.83 10.05
3 50.28 6.60
3 52.67 0.67
2 50.82 0.16
3 45.86 6.90


3 8.68 2.35
3 14.29 2.10
3 10.34 1.51
3 18.27 2.54
3 16.99 2.31
3 17.58 1.22
2 12.24 0.50
3 11.84 1.58


3 19.85 0.88
3 22.11 2.45
3 17.98 0.97
3 16.51 5.29
3 16.18 0.91
3 20.48 1.01
2 18.98 0.31
3 20.23 3.38


3 14.58 1.20
3 17.83 1.02
3 27.55 1.50
3 10.05 2.84
3 17.11 3.69
3 14.61 0.86
2 19.60 0.35
3 13.80 2.17


% Hemicellulose


% Cellulose


% Lignin










Table A-5. Biomass measurements of seedling stems and leaves from eight tree species at
Tl and T2 (1 and 6 mos after first leaf expansion, respectively). Shown are


means ( 1 SD) and N
Table 1-1.


Stem height (mm)











Leaf area (cm2)











Leaf thickness (mm)











Plant biomass (g)


Species
ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


ANAE
ASPC
BEIP
CASE
EUGN
GUSS
TABR
TETP


number of individuals. For species codes refer to


Tl
Mean
238.2
140.7
165.3
140.1
140.3
176.7
64.1
154.5


112.78
35.40
67.70
66.24
13.89
138.05
6.54
28.22


0.16
0.20
0.29
0.17
0.11
0.15
0.14
0.13


0.8066
0.5446
1.8675
0.3022
0.3990
3.6938
0.0459
0.2467


SD
35.4
19.6
36.9
34.5
27.7
29.8
12.6
19.0


T2
Mean
253.4
154.8
162.8
173.3
182.5
196.3
115.9
162.7


125.42
55.98
67.94
122.44
34.32
171.96
52.09
41.82


41.03
8.26
34.61
42.23
5.97
41.97
6.52
6.52


0.03
0.01
0.04
0.04
0.02
0.02
0.03
0.03


0.18
0.20
0.14
0.11
0.08
0.14
0.15
0.13


0.2689
0.1265
0.5233
0.2076
0.1743
1.3287
0.0240
0.0543


1.1550
0.7385
0.9564
0.6178
0.3508
2.5911
0.2540
0.3735


SD
40.7
26.9
35.0
32.7
25.0
50.2
39.4
21.8


57.70
20.29
25.32
80.94
9.23
55.02
16.84
16.84


0.05
0.01
0.01
0.02
0.01
0.01
0.05
0.02


0.4724
0.2422
0.4232
0.3901
0.2056
0.8554
0.2180
0.1564















LIST OF REFERENCES


Aide, T.M. (1987) Limbfalls-a major cause of sapling mortality for tropical forest
plants. Biotropica 19, 284-285.

Aide, T.M. & Zimmerman J.K. (1990) Patterns of insect herbivory, growth, and
survivorship in juveniles of a neotropical liana. Ecology 70,1412-1421.

Asquith, N.M., Wright, S.J. & Clauss, M.J. (1997) Does mammal community
composition control recruitment in neotropical forests? Evidence from Panama.
Ecology 78, 941-946.

Augspurger, C.K. (1983) Seed dispersal of the tropical tree, Platypodium elegans, and the
escape of its seedlings from fungal pathogens. Journal of Ecology 71,759-771.

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.

Barnett, J.R. & Jeronimidis, G., eds. (2003) Wood Quality and its Biological Basis.
Blackwell Publishing, Oxford.

Bazzaz, S.A. & Grace, J., eds. (1997) Plant Resource Allocation. Academic Press, San
Diego, California.

Benitez-Malvido, J., Garcia-Guzman, G. & Kossmann-Ferraz, I.D. (1999) Leaf-fungal
incidence and herbivory on tree seedlings in tropical rainforest fragments: an
experimental study. Biological Conservation 91,143-150.

Bruchert, F., Becker, G. & Speck, T. (2000) The mechanics of Norway spruce (Picea
abies (L.) Karst) mechanical properties of standing trees from different thinning
regimes. Forest Ecology andManagement 135, 45-62.

Campbell, K.A. & Hawkins, C.D.B. (2004) Effect of seed source and nursery culture on
paper birch (Betulapapyrifera) uprooting resistance and field performance. Forest
Ecology and Management 196, 425-433.

Canham, C.D., Kobe, R.K., Latty, E.F. & Chazdon, R.L. (1999) Interspecific and
intraspecific variation in tree seedling survival: effects of allocation to roots versus
carbohydrate reserves. Oecologia 121,1-11.









Chapman, C.A. & Chapman L.J. (1996) Frugivory and the fate of dispersed and non-
dispersed seeds of six African tree species. Journal of Tropical Ecology 12, 491-
504.

Choong, M.F. (1996) What makes a leaf tough and how this affects the pattern of
Castanopsisfissa leaf consumption by caterpillars. Functional Ecology 10, 668-
674.

Choong, M.F., Lucas, P.W., Ong, J.S.Y., Pereira, B., Tan, H.T.W. & Turner, I.M. (1992)
Leaf fracture-toughness and sclerophylly-their correlations and ecological
implications. New Phytologist 121, 597-610.

Clark, D.B. & Clark, D.A. (1985) Seedling dynamics of a tropical tree: impacts of
herbivory and meristem damage. Ecology 66, 1884-1892.

Clark, D.B. & Clark, D.A. (1989) The role of physical damage in the seedling mortality
regime of a neotropical rain forest. Oikos 55, 225-230.

Clark, D.B. & Clark, D.A. (1991) The impact of physical damage on canopy tree
regeneration in tropical rain-forest. Journal of Ecology 79, 447-457.

Coley, P.D. (1983) Herbivory and defensive characteristics of tree species in a lowland
tropical forest. Ecological Monographs 53, 209-233.

Collet, D. (2003) Modeling Survival Data in Medical Research, 2nd edn. Chapman &
Hall/CRC, Washington D.C.

Condit, R., Hubbell, S.P., & Foster, R.B. (1995) Mortality rates of 205 neotropical tree
and shrub species and the impact of a severe drought. Ecological Monographs 65,
419-439.

Cooley, A.M., Reich, A. & Rundel P. (2004). Leaf support biomechanics of neotropical
understory herbs. American Journal ofBotany 91, 573-581.

Cordero, R.A. (1999) Ecophysiology of Cecropia schreberiana saplings in two wind
regimes in an elfin cloud forest: growth, gas exchange, architecture and stem
biomechanics. Tree Physiology 19, 153-163.

Croat, T.B. (1978) Flora ofBarro Colorado Island. Stanford University Press, Stanford.

Darvell, B.W., Lee, P.K.D., Yuen, T.D.B. & Lucas. P.W. (1996) A portable fracture
toughness tester for biological materials. Measurement Science & Technology 7,
954-962.

De Steven, D. (1994) Tropical tree seedling dynamics: recruitment patterns and their
population consequences for three canopy species in Panama. Journal of Tropical
Ecology 10, 369-383.









Dominy, N.J., Lucas, P.W. & Wright, S.J. (2003) Mechanics and chemistry of rain forest
leaves: canopy and understorey compared. Journal ofExperimental Botany 54,
2007-2014.

Drake, D.R. & Pratt, L.W. (2001) Seedling mortality in Hawaiian rain forest: The role of
small-scale physical disturbance. Biotropica 33, 319-323.

Esau, K. (1977) Anatomy of Seed Plants, 2nd edn. John Wiley & Sons, New York.

Foster, R.B. & Brokaw, N.V.L. (1982) Structure and history of the vegetation of Barro
Colorado Island. The Ecology of a Tropical Forest: Seasonal Rhythms and Long-
Term Changes. (eds E.G. Leigh, Jr., A.S. Rand & D.M. Windsor), pp. 133-150.
Smithsonian Institution Press, Washington, D. C.

Gartner, B.L. (1991) Structural stability and architecture of vines vs. shrubs of poison
oak, Toxicodendron diversilobum. Ecology 72, 2005-2015.

Gillman, L.N., Ogden, J., Wright, S.D., Stewart, K.L. & Walsh, D.P. (2004) The
influence of macro-litterfall and forest structure on litterfall damage to seedlings.
AustralEcology 29, 305-312.

Gillman, L.N., Wright, S.D. & Ogden, J. (2002) Use of artificial seedlings to estimate
damage of forest seedlings due to litterfall and animals. Journal of Vegetation
Science, 13, 635-640.

Givnish, T.J. (1995). Plant stems: biomechanical adaptation for energy capture and
influence on species distributions. Plant Stems Physiology and Functional
Morphology (ed. B.L. Gartner), pp. 3-49. Academic Press, San Diego, California.

Glanz, W.E. (1982) The terrrestrial mammal fauna of Barro Colorado Island: censuses
and long-term changes. The Ecology of a Tropical Forest: Seasonal Rhythms and
Long-Term Changes (eds E.G. Leigh, A.S. Rand & D.M. Windsor), pp 455-468,
Smithsonian Institution Press, Washington D.C.

G6mez, J.M., Garcia, D. & Zamora, R. (2003) Impact of vertebrate acorn- and seedling-
predators on a Mediterranean Quercus pyrenaica forest. Forest Ecology and
Management 180, 125-134.

Gotelli, N.J. & Ellison, A.M. (2004) A Primer of Ecological Statistics. Sinauer
Associates Inc., Massachusetts.

Green, P.T. & Juniper, P.A. (2004) Seed mass, seedling herbivory and the reserve effect
in tropical rainforest seedlings. Functional Ecology 18, 539-547.

Greenhill, G. (1881) Determination of the greatest height consistent with stability that a
vertical pole or mast can be made, and the greatest height to which a tree of given
proportions can grow. Procedures of the Cambridge Philosophical Society 4, 65-
73.









Grubb, P.J. (1986) Sclerophylls, pachyphylls, and pycnophylls: the nature and
significance of hard leaf surfaces. Insect andPlant Surfaces (eds. B. Juniper &
T.R.E. Southwood), pp.137-150. Edward Arnold, London.

Guariguata, M.R. (1998) Response of forest tree saplings to experimental mechanical
damage in lowland Panama. Forest Ecology and Management 102, 103-111.

Harms, K.E. & Dalling, J.W. (1997) Damage and herbivory tolerance through
resprouting as an advantage of large seed size in tropical trees and lianas. Journal
of Tropical Ecology 13, 617-621.

Harper, J.K., Dalley, N.K., Owen, N.L., Wood, S.G. & Cates, R.G. (1993) X-ray
structure and C-13 NMR assignments of indole alkaloids from Aspidosperma
cruenta. Journal of Crystallographic and Spectroscopic Research 23, 1005-1011.

Hartshorn, G. S. (1972) The ecological life history and population dynamics of
Penthaclethra macroloba, a tropical wet forest dominant and Stryphnodendron
excelsum, an occasional associate. Ph.D. Thesis, University of Washington, Seattle.

Henry, H.A.L. & Thomas, S.C. (2002) Interactive effects of lateral shade and wind on
stem allometry, biomass allocation, and mechanical stability in Abutilon
theophrasti (Malvaceae). American Journal ofBotany 89, 1609-1615.

Hoffmann, B., Chabbert, B., Monties, B., & Speck, T. (2003) Mechanical, chemical and
x-ray analysis of wood in the two tropical lianas Bauhinia guianensis and
Condylocarpon guianense: variations during ontogeny. Planta 217, 32-40.

Holbrook, N.M. & Putz, F.E. (1989) Influence of neighbors on tree form: effects of
lateral shade and prevention of sway on the allometry ofLiquidambar styraciflua
(sweet gum). American Journal of Botany 76,1740-1749.

Howe, H.F. & Schupp, E.W. (1984) Early Consequences of seed dispersal for neotropical
tree (Virola surinamensis). Ecology 66, 781-791.

Ibarra-Manriquez G., Martinez Ramos, M. & Oyama, K. (2001) Seedling functional
types in a lowland rain forest in Mexico. American Journal ofBotany 88, 1801-
1812.

Isnard, S., Speck, T. & Rowe, N.P. (2003) Mechanical architecture and development in
Clematis: implications for canalized evolution of growth forms. New Phytologist
158, 543-559.

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. (1996) Cotyledon functional morphology, patterns of seed reserve utilization
and regeneration niches of tropical tree seedlings. The Ecology of Tropical Forest
Tree Seedlings (ed. M.D. Swaine), pp. 193-208,UNESCO, Paris.

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. & Augspurger C.K. (1989) Seed and seedling ecology of a monocarpic
tropical tree, Tachigalia versicolor. Ecology 70, 1102-1114.

Kitajima, K. & Fenner, M. (2000) Seedling regeneration ecology. Seeds: Ecology of
Regeneration in Plant Communities, 2nd edn (ed. M. Fenner), pp. 331-360. CAB
International, Wallingford, UK.

Kobe, R.K. (1997) Carbohydrate allocation to storage as a basis of interspecific variation
in sapling survivorship and growth. Oikos 80, 226-233.

Kobe, R. K. (1999) Light gradient partitioning among tropical tree species through
differential seedling mortality and growth. Ecology 80, 187-201.

Leigh, E.G., Jr., Windsor, D.M. & Rand, S.A., eds. (1982) The Ecology of a Tropical
Forest: Seasonal Rhythms and Long-Term Changes. Smithsonian Institution Press,
Washington D.C.

Loehle, C. (1988) Tree life history strategies: the role of defenses. Canadian Journal of
Forest Research 18, 209-222.

Lucas, P.W., Beta, T., Darvell, B.W., Dominy, N.J., Essackjee, H.C., Lee, P.K.D.,
Osorio, D., Ramsden, L., Yamashita, N. & Yuen, T.D.B. (2001) Field kit to
characterize physical, chemical and spatial aspects of potential primate foods. Folia
Primatologica 72, 11-25.

Lucas, P.W. & Pereira, B. (1990) Estimation of the fracture toughness of leaves.
Functional Ecology 4, 819-822.

Lucas, P.W., Turner, I.M., Dominy, N.J. & Yamashita. N. (2000) Mechanical defenses to
herbivory. Annals ofBotany 86, 913-920.

Mack, A.L. (1998) The potential impact of small-scale physical disturbance on seedlings
in a Papuan rainforest. Biotropica 30 547-552.

Marquis J. & Braker H.E. (1994) Plant-herbivore interactions: diversity, specificity, and
impact. La Selva Ecology and Natural History of a Neotropical Rain Forest (eds.
L.A. McDade, K.S. Bawa, H.A. Hespenheide & G.S. Hartshorn), pp. 261-
281.University of Chicago Press, Chicago.









McCarthy, B.C. & Facelli, J.M. (1990) Microdisturbances in old fields and forests:
implications for woody seedling establishment. Oikos 58 55-60.

Moles, A.T. & Westoby, M. (2004a) What do seedlings die from and what are the
implications for evolution of seed size? Oikos 106, 193-199.

Moles, A.T. & Westoby, M. (2004b) seedling survival and seed size: a synthesis of the
literature. Journal of Ecology 92, 372-383.

Molofsky, J. & Fisher B.L. (1993) Habitat and predation effects on seedling survival and
growth in shade-tolerant tropical trees. Ecology 74, 261-265.

Montgomery, R.A. & Chazdon, R.L. (2002) Light gradient partitioning by tropical tree
seedlings in the absence of canopy gaps. Oecologia 131, 165-174.

Mou, P. & Warrillow, M.P. (2000) Ice storm damage to a mixed hardwood forest and its
impacts on forest regeneration in the ridge and valley region of southwestern
Virginia. Journal of the Torrey Botanical Society 127, 66-82.

Myers, J.A. (2005) Seedling carbohydrate storage, survival, and stress tolerance in a
neotropical forest. Master of Science Thesis, University of Florida, Gainesville.

Niklas, K.J. (1992) Plant Biomechanics. An Engineering Approach to Plant Form and
Function. University of Chicago Press, Chicago.

Niklas, K.J. (1995) Plant height and the properties of some herbaceous stems. Annals of
Botany 75, 133-142.

Niklas, K.J., Molina-Freaner, F., Tinoco-Ojanguren, C. & Paolillo, D.J. (2000) Wood
biomechanics and anatomy of Pachycereus pringlei. American Journal ofBotany
87, 469-481.

Osunkoya, O., Ash, J.E., Hopkins, M.S. & Graham, A.W (1992) Factors affecting
survival of tree seedlings in North Queensland rainforests. Oecologia 91, 569-578.

Pauw, A., Van Bael, S.A., Peters, H.A., Allison S.D., Camargo, J.L.C., Cifuentes-Jara,
M., Conserva, A., Garcia Restom, T., Heartsill-Scalley, T., Mangan, S.A., Nunez-
Iturri, G., Rivera-Ocasio, E., Rountree, M., Vetter, S. & Volkmer de Casilho, C.
(2004) Physical damage in relation to carbon allocation strategies to tropical forest
tree saplings. Biotropica 36, 410-413.

Paz, H. & Martinez-Ramos, M. (2003) Seed mass and seedling performance within eight
species ofPsychotria (Rubiaceae). Ecology 84, 439-450.

Putz, F.E., Coley, P.D., Lu, K., Montalvo, A. & Aiello, A. (1983) Uprooting and
snapping of trees structural determinants and ecological consequences. Canadian
Journal of Forest Research 13, 1011-1020.






81


Pyke, D.A. & Thompson, J.N. (1986) Statistical analysis of survival and removal rate
experiments. Ecology 67, 240-245.

Roldan, A.I. & Simonetti, J.A. (2001) Plant-mammal interactions in tropical Bolivian
forests with different hunting pressures. Conservation Biology 15, 617-623.

Rowe, N.P. & Speck, T. (1996) Biomechanical characteristics of the ontogeny and
growth habit of the tropical liana Condylocarpon guianense (Apocynaceae).
International Journal ofPlant Sciences 157, 406-417.

Ryan M.G., Melillo, J.M. & Ricca, A. (1989) A comparison of methods for determining
proximate carbon fractions of forest litter. Canadian Journal of Forest Research
20, 166-171.

Savidge, R.A. (2003) Tree growth and wood quality. Wood Quality and its Biological
Basis. (eds. J.R. Bamett, & G. Jeronimidis), pp. 1-29. Blackwell Publishing,
Oxford.

Scariot, A. (2000) Seedling mortality by litterfall in Amazonian forest fragments.
Biotropica 32, 662-669.

Sharpe, J.M. (1993) Plant growth and demography of a neotropical herbaceous fern
Danea wendlandii (Marattiaceae) in a Costa Rican rain forest. Biotropica 25, 85-
94.

Shure, D. J. & Wilson, L. A. (1993) Patch-size effects on plant phenolics in successional
openings of the Southern Appalachians. Ecology 74, 55-67.

Sibly, R.M. & Vincent, J.F.V (1997) Optimality approaches to resource allocation in
woody tissues. Plant Resource Allocation (eds. F.A. Bazzaz & J. Grace), pp. 143-
160. Academic Press, California.

Smythe, N. (1978) The natural history of the Central American agouti (Dasyprocta
punctata). S.,nmith/,uian Contributions to Zoology 157, 1-52.

Sork, V.L. (1987) Effects of predation and light on seedling establishment in Gustavia
superba. Ecology 68, 1341-1350.

Taylor, F.J. (1971) Some aspects of the growth of mango (Manguifera indica L.) leaves.
III. A mechanical analysis. New Phytologist 70, 911-922.

Turner, I.M. (1990) Tree seedling growth and survival in a Malaysian rain forest.
Biotropica 22, 146-154.

Van Der Meer, P.J. & Bongers, F. (1996) Patterns of tree-fall and branch-fall in a tropical
rain forest in French Guiana. Journal of Ecology 84, 19-29









Weltzin, J.F., Archer, S. & Heitschmidt, R.K. (1997) Small-mammal regulation of
vegetation structure in a temperate savanna. Ecology 78, 751-763.

Windsor D.M. (1990) Climate and moisture variability in a tropical forest: long- term
records from Barro Colorado Island, Panama. S.un,ii/huian Contributions to the
Earth Sciences 29,145.

Wright, I.J. & Cannon, K. (2001) Relationships between leaf lifespan and structural
defences in a low-nutrient, sclerophyll flora. Functional Ecology 15, 351-359.

Wright, I.J., Reich, P.B., Westoby, M., Ackerly, D.D., Baruch, Z., Bongers, F.,
Cavender-Bares, J., Chapin, T., Cornelissen, J.H.C., Diemer, M., Flexas, J.,
Gamier, E., Groom, P.K., Gulias, J., Hikosaka, K., Lamont, B.B., Lee, T., Lee, W.,
Lusk, C., Midgley, J.J., Navas, M.L., Niinemets, U., Oleksyn, J., Osada, N.,
Poorter, H., Poot, P., Prior, L., Pyankov, V.I., Roumet, C., Thomas, S.C., Tjoelker,
M.G., Veneklaas, E.J., & Villar, R. (2004) The worldwide leaf economics
spectrum. Nature 428, 821-827.

Wright, S.J., Muller-Landau, H.C., Condit, R., & Hubbell, S.P. (2003) Gap-dependent
recruitment, realized vital rates, and size distributions of tropical trees. Ecology 84,
3174-3185.

Wright, W. & Illius, W. (1995) A comparative study of the fracture properties of five
grasses. Functional Ecology 9, 269-278.















BIOGRAPHICAL SKETCH

Silvia Alvarez-Clare was born in San Jose, Costa Rica in September 1977. She

attended Saint Francis High School where she was inspired to become a biologist. Silvia

earned her bachelor's degree in biology at the Universidad de Costa Rica (UCR) in July

2001. During her undergraduate studies at UCR, she became interested in plant

ecophyisiology and conservation partly because she had many opportunities to work in

diverse tropical ecosystems throughout Costa Rica. She also worked as a research

assistant in the Centro de Investigaci6n en Biologia Celular y Molecular at Universidad

de Costa Rica (CIBCM) and at the university herbarium. Silvia was also a naturalist

guide in the Monteverde cloud forest, an experience that reinforced her love of nature and

her desire to continue her studies. When she finished her undergraduate studies, Silvia

taught biology for a year in the International Baccalaureate program at Lincoln High

School in San Jose. She then came to the University of Florida where she obtained her

master's degree in Botany with a minor in Statistics in May 2005. Silvia will continue her

education toward her doctoral degree in interdisciplinary ecology under the guidance of

Dr. Michelle Mack at the University of Florida. For her dissertation research she plans to

study tropical forest regeneration, restoration, and nutrient cycling in Costa Rica.




Full Text

PAGE 1

BIOMECHANICAL PROPERTIES OF TR OPICAL TREE SEEDLINGS AS A FUNCTIONAL CORRELATE OF SHADE TOLERANCE By SILVIA ALVAREZ-CLARE A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Silvia Alvarez-Clare

PAGE 3

To my parents, who let me fly; and to Abuelita Betty, who gave me the wings.

PAGE 4

iv ACKNOWLEDGMENTS I would like to thank my advisor Kaor u Kitajima for academic and financial support but also for her kind, yet strict gui dance while I took my initial steps as a scientist. Kaoru and her daughter, Sachi, treat ed me as family when I first arrived in Gainesville. I thank my other committee members (Emilio Bruna, Michael Daniels, and Jack Putz) for providing support and valuable comments that improved my research and finally my thesis. I also thank Gerardo Avalos who was the first to believe in me as a scientist. I extend thanks to the Department of Botany staff, who offe red logistic support. Also, the graduate students and professors in the Botany Plant Ecology Group helped develop my ideas and scientific thinki ng, through stimulating discussion sessions. I thank the Smithsonian Tropical Research Institute for financial and institutional support. Staff and researchers at Barro Colo rado Island assisted in numerous ways during my fieldwork. Roberto Cordero shared his biomechanical knowledge and was always willing to help. Sarah Tarrant, Liza Cowar d, Jeffrey Hubbard, Sebastian Bernal, and Marta Vargas provided invaluab le field assistance. I especi ally thank Jeff for sharing many eco-challenges with me through the forest and Marta for all the late-night leafcutting sessions that also re sulted in a great friendship. Momoka Yao provided great help with fiber analysis and data entry. I thank my Tico friends, who have been my family away from home for the past 3 yrs. I also thank Jenny Schaffer, Carla Stef anescu, Cat Cardelus, Eddie Watkins, Sarah Bray, and all the rest of my Gainesville friends, who always help me put life in

PAGE 5

v perspective. Jonathan Myers has enhanced my passion for science. His philosophical questions have taught me that we do not rea lly understand something until we are able to explain it. I thank my family for supporting me in ev ery enterprise I take. Their love and advice have been precious tools in helpi ng me achieve my goals. I would have not completed this thesis and managed to ma intain my sanity without the support and patience of Chuck Knapp. He has survived my frustrations and d eadlines, and has been my best friend and companion. Finally, I tha nk God in his universal, nondenominational form, for allowing us to seek the answers fo r the miracles of natu re through science.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES.............................................................................................................x ABSTRACT....................................................................................................................... xi CHAPTER 1 BIOMECHANICAL PROPERTIES OF TROPICAL TREE SEEDLINGS AS A FUNCTIONAL CORRELATE OF SHADE TOLERANCE.......................................1 Introduction................................................................................................................... 1 Materials and Methods.................................................................................................5 Biomechanical Measurements...............................................................................7 Youngs modulus of elasticity........................................................................7 Fracture toughness..........................................................................................7 Density...........................................................................................................8 Chemical analysis...........................................................................................8 Percent critical height.....................................................................................9 Flexural stiffness............................................................................................9 Work-to-bend...............................................................................................10 Whole stem flexibility..................................................................................10 Force of fracture...........................................................................................11 Specific leaf area..........................................................................................11 Statistical Analyses..............................................................................................11 Results........................................................................................................................ .12 Stem Biomechanics.............................................................................................12 Leaf Biomechanics..............................................................................................14 Relationship between Biomechanical Traits of Stems and Leaves.....................15 Relationship between Seedling Biomechanics and Survival...............................16 Discussion...................................................................................................................16 Stem Biomechanics.............................................................................................16 Leaf Biomechanics..............................................................................................20 Relationship between Biomechanical Traits of Stems and Leaves.....................22 Relationship between Seedling Biomechanics and Survival...............................22 Conclusions.........................................................................................................24

PAGE 7

vii 2 SPECIES DIFFERENCES IN SEEDLING SUSCEPTIBILITY TO BIOTIC AND ABIOTIC HAZARDS IN THE FOREST UNDERSTORY......................................41 Introduction.................................................................................................................41 Materials and Methods...............................................................................................43 Study Site and Species.........................................................................................43 Experimental Design...........................................................................................44 Survival, Damage Agents, and Types of Mechanical Damage...........................45 Artificial Seedlings..............................................................................................47 Statistical Analyses..............................................................................................48 Results........................................................................................................................ .49 Seedling Survival.................................................................................................49 Damage Agents...................................................................................................50 Types of Mechanical Damage.............................................................................51 Artificial Seedlings..............................................................................................52 Discussion...................................................................................................................53 Survival, Damage Agents, and Types of Mechanical Damage...........................53 Artificial Seedlings..............................................................................................56 Conclusions.........................................................................................................57 APPENDIX SPECIES MEANS AND STANDARD DEVIATIONS FOR BIOMECHANICAL MEASUREMENTS, FIBER ANALYSIS, AND BIOMASS.....................................................................................................67 LIST OF REFERENCES...................................................................................................75 BIOGRAPHICAL SKETCH.............................................................................................83

PAGE 8

viii LIST OF TABLES Table page 1-1 Ecological characteristics of eight tr opical tree species used in my study, listed by increasing shade tolerance.........................................................................26 1-2 Percent seedling survival for the eigh t study species over specified periods from four independent studies in BCNM..........................................................................27 1-3 Effect of species and harvest time on material and structural properties of seedling stems..........................................................................................................28 1-4 Relationships among stem biomechanical traits for seedlings of eight tree species......................................................................................................................29 1-5 Effect of species and harvest time on ma terial and structural traits of leaves.........31 1-6 Relationships among leaf biomechanical tr aits for seedlings of eight tree species..32 1-7 Relationships among biomechanical trai ts of stems and leaves for seedlings of eight tree species.................................................................................................33 1-8 Relationships among % survival in shade and various seedling biomechanical traits of stems and leaves for seedlings of eight tree species..................................34 2-1 Relationships among species rankings of survival probability during the specified interval for seedlings of eight tree species................................................59 2-2 Percent damage fatality of four type s of mechanical damage on eight tree species during 1 yr in the forest understory.............................................................59 2-3 Relationships among stem biomechanical traits and % damage fatality for seedlings of eight tree species..................................................................................60 2-4 Percentage of artificial seedlings affected by specified damage agents in this and other published studies in different forest communities.........................61 A-1 Biomechanical measurements of se edling stems from eight tree species................68 A-2 Biomechanical measurements of seed ling leaves from eight tree species...............70 A-3 Fiber fractions of seedling stems from eight tree species........................................72

PAGE 9

ix A-4 Fiber fractions of seedling l eaves from eight tree species.......................................73 A-5 Biomass measurements of seedling stems and leaves from eight tree species........74

PAGE 10

x LIST OF FIGURES Figure page 1-1 Means (+ 1 SD) material biomechanical traits of stems for seedlings of eight tree species...............................................................................................................35 1-2 Mean (+ 1 SD) % fiber content (% NDF ) for seedlings of eight tree species..........36 1-3 Log-log relationships between some mate rial properties of stems for seedlings of eight tree species..................................................................................................37 1-4 Means (+ 1 SD) structural biomechani cal traits of stems for seedlings of eight tree species......................................................................................................38 1-5 Means (+ 1 SD) biomechanical traits of leaves for seedlings of eight tree species.39 1-6 Log-log relationships between some biomechanical properties measured at 6 mos after first leaf expansion (T2), and % mean survival in shade for seedlings of eight tree species..................................................................................40 2-1 Kaplan-Meier survivorship curves for seedlings of eight tree species transplanted to the forest understory........................................................................63 2-2 Kaplan-Meier survivorship curves (p roportion of seedlings yet to be hit by specified damage agents plotted against time) for seedlings of eight tree species transplanted to th e forest understory........................................................................64 2-3 Percent of real and artificial seedlings (AS) damaged during 1 yr in the forest understory by specific damage agent.......................................................................65 2-4 Kaplan-Meier survivorship curves for mechanical damage experienced by artificial (AS) and real seedlings dur ing 1 yr in the forest understory.....................66

PAGE 11

xi Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science BIOMECHANICAL PROPERTIES OF TR OPICAL TREE SEEDLINGS AS A FUNCTIONAL CORRELATE OF SHADE TOLERANCE By Silvia Alvarez-Clare May 2005 Chair: Kaoru Kitajima Major Department: Botany Physical disturbances by vertebrates and lit terfall are important causes of seedling mortality in the understory of tr opical forests. Thus, the capacity to resist or recover from mechanical damage should enhance seedling su rvival in shade. I explored interspecific variation in seedling biomechan ical properties across a shad e tolerance gradient, using eight tropical tree species from Barro Colo rado Island (BCI), Panama. The stems and leaves of shade-tolerant species were constructed of stronger materials than were those of light-demanding species, as measured by a hi gher Youngs modulus of elasticity, fracture toughness, and tissue density. Thes e traits were highl y correlated with tis sue fiber content (especially % cellulose, but not % lignin) a nd with seedling survival during the first 6 mo. There were no correlations between seed ling survival and structural measurements that integrated material and morphological traits, such as fl exural stiffness, work-to-bend, and whole stem flexibility. The lack of corre lations suggests that investment in strong

PAGE 12

xii material, rather than in large plant size is more beneficial for seedlings at early developmental stages. Next, I described first-year temporal patte rns of seedling mortality, susceptibility to damage agents, and types of damage suffere d by seedlings in the forest understory. Seedling mortality was highest during the fi rst 2 mos (due to vertebrate activity) and gradually decreased over the remaining 8 m o. Species differed si gnificantly in their temporal patterns of mortality and in the pr oportions of seedling surviving at the end of the study. The three main causes of damage were (in order of severity) vertebrate activity, disease, and litterfall. The f our main types of mechanical damage (in order of severity) were leaf damage, bent stems, broken stems, and uprooted seedlings. All species suffered similar levels of mechanical damage but sh ade-tolerant species (w hich often had stems constructed of strong materials) were less likely to die when damaged than lightdemanding species. My study provides evidence that, in Barro Colorado Island, physical disturbance is a major cause of seedling mortality during the first year, and that shade-tolerant species survive better than light-demanding species after suffering mechan ical damage. Higher survival is potentially influenced by higher carbon investment of shade-tolerant tree species into structural suppor t of stems at very early developmental stages. However, greater carbon allocation to st ructural defense must be ac companied by slower relative growth rates. Thus, functional diversity in biomechanical properties is an important aspect of multiple trait associations that le ad to the growth-survival trade-offs observed among coexisting tropical tree species.

PAGE 13

1 CHAPTER 1 BIOMECHANICAL PROPERTIES OF TR OPICAL TREE SEEDLINGS AS A FUNCTIONAL CORRELATE OF SHADE TOLERANCE Introduction Mechanical damage is a major cause of mortality in understory plants, including tree seedlings (Clark and Clark 1989, G illman, Wright & Ogden 2002), saplings (Hartshorn 1972, Aide 1987), and understory herbs (Gar tner 1991, Sharpe 1993). Clark and Clark (1991) found that litterfall caused 11% of the annual mortality of seedlings 1 cm in diameter in a lowland tropical rain forest. In a study in seasonal tropical forest, Alvarez-Clare (Chapter 2) found that 77% of 755 seedlings, from eight species of tropical trees transplanted to the forest understory, suffered some type of mechanical damage after 1 yr. Mechanical damage can be caused by falling debris (Aide 1987, Putz et al 1983), vertebrate activity (Roldan & Sim onetti 2001, Gmez, Garca & Zamora 2004), water or ice flow (Mou & Warrillow 2000), and herbivory (Coley 1983). In the tropical rain forest (where there is high frequency of such disturbances) survival of seedlings depends on their ability to avoid or recuperate from mechanical damage. An increase in carbon allocation to stru ctural tissues can increase seedling performance in the forest understory by incr easing biomechanical toughness and stiffness (Sibly & Vincent 1997) and thereby decreasing susceptibility to damage. For example, Augspurger (1984a) found that from nine specie s of tree seedlings, species less affected by pathogen attack were those that be came woody more rapidly. Additionally, mechanical defenses in leaves play a substa ntial role in deterring loss to herbivores

PAGE 14

2 (Coley 1983, Choong 1996) and ar e correlated with leaf life span (Wright & Cannon 2001). Biomechanical strength results from carbon in vestment in tissues of stems, leaves, and roots, and from the organization and struct ure of those tissues within the plant. For example, fiber is an important contributor of mechanical strength in leaves. Choong (1996) found that high fracture toughness was corre lated with fiber content in leaves of Castanopsis fissa (Fagaceae). Resistance to mechanical damage is also influenced by the type and organization of fiber components (e.g., cellulose, hemicellulose, and lignin). Within a tissue, high cellulose content resu lts in increased toughness, while high lignin content increases hardness (Nik las 1992). Additionally, the or ientation of the cellulose microfibrils in the S2 layer of the secondary cell wall affects the abilit y of the material to resist cracking under plastic tension (Lucas et al 2000). In stems of adult trees, high tissue density and cell-wall volume fraction in the xylem increase toughness and stiffness (Barnett & Jeronimidis 2003). Similarly, higher resistance to mechanical stress in root tissues can improve anchoring capacity, a nd reduce risk of uprooting (Campbell & Hawkins 2004). Considering the limited carbon budgets of seedlings, increased investments in structural materials must be accompanied by decreases in allocation to growth, to reserves, and/or to chemical defenses, su ch as tannins and alkaloids (Kitajima 1994, Kobe 1997, Shure & Wilson 1993). Thus, resource limitation leads to trade-offs involving biomechanical attributes, such as li ght acquisition vs. struct ural safety, growth vs. tissue density, and photos ynthetic capacity vs. leaf toughness (Loehle 1988, Niklas 1992, Givnish 1995, Bazzaz & Grace 1997).

PAGE 15

3 Natural selection should favor stems a nd leaves with forms, biomechanical properties, and growth dynamics that ma ximize carbon gain, competitive ability, and safety, but that minimize costs of cons truction and maintenance (Givnish 1995). Obviously, conflicts among these aspects ma ke it impossible to optimize all factors simultaneously. Thus diverse ecological st rategies have evolved, affected by the evolutionary forces dominating particular ecol ogical niches. For example, plants can cope with mechanical damage through investment in resistant structures or by allocating resources to reserves that enable them to replace damaged tissues (Harms & Dalling 1997, Pauw et al 2004). In both cases, a strategy will onl y be selected for, if it confers a benefit relative to the cost, su ch as increased survival (S ibly & Vincent 1997). My study focused on the defensive strategy of investi ng in damage resistance, by exploring the influence of biomechanical traits on seedling survival in shade. Although biomechanical propert ies clearly influence plant survival and competitive ability, and potentially influence their ecol ogical distribution (Col ey 1983, Niklas 1992, Lucas et al 2000), investigations evaluating plan t biomechanical properties in an ecological context are few. E xploring the functional diversit y of biomechanical properties in tropical tree seedlings should help in descri bing multiple trait associ ations that lead to growth-survival trade-offs observed am ong coexisting tropical tree species. My study explored interspecific variati on in seedling biomechanical properties across a shade-tolerant gradient, among eight tropical tree species from Barro Colorado Island (BCI), Panama. Because my goal was to understand the ecological role of biomechanical traits in tropical tree seedlin gs, I evaluated a variety of biomechanical attributes at the material and st ructural level. Plant stem and leaf material traits consist of

PAGE 16

4 the composite, anisotropic material of wh ich they are composed. At this level, mechanical (toughness, stiffness, and density) a nd chemical (fiber fraction) traits were measured, without considering size or anatomical organizatio n. At the structural level, material properties combined with morphol ogical traits (e.g., size and shape) were measured for individual plant organs (stems or leaves). Measurements at the structural level for stems included flexural stiffness, percent critical height, work-to-bend, and whole stem flexibility. To describe the structur e of leaves, I measured specific leaf area and force of fracture. Specifically, I addressed the following three questions: Do material properties of stems and le aves of tropical tree seedlings differ among species in relation to their shade tolerance? Because strong material confers an advantage against mechanical damage and presumably increases survival probabilities, I predicted that shad e-tolerant species, which survive better in shade (Wright et al. 2003, Chapter 2), should have stronger stem and leaf materials than light-demanding species. More specifically, stems of shade-tolerant species should have higher Youngs modulus of elastic ity, fracture toughness, and density. In addition, shade-tolerant species should have higher fiber content, which reflects the chemical composition of the material. Likewise, leaves of shadetolerant species should have higher lami na and midvein fracture toughness, density, and fiber content. What is the relationship between mate rial and structural biomechanical properties, and between material properties of stems and leaves? Carbon allocation to stronger tissues should also contribute to overall structural strength. Therefore, unless there are important mo rphological differences between species, material traits should be refl ected at the structural leve l. Stems of shade-tolerant species should have lower pe rcent critical height, highe r flexural stiffness, and higher resistance to bending in the field. Leaves of shade-tolerant species should have a lower specific leaf area (SLA) and a higher overall resistance to fracture (force of fracture) than leaves of shad e intolerant species. I also expected a concordance of biomechanical attributes between structures. Thus, biomechanical properties of stems and leaves should be correlated. Because it was my ultimate objective to evaluate the implications of biomechanics for seedling performance in the forest, it was key to examine biomechanical traits at the material level but also at the structural level, integrating morphological attr ibutes that can influence overall plant response to mechanical stress. How do biomechanical properties of seed ling stems and leaves change over the first 6 mos after initial development? It has been shown that free-standing plants, as opposed to lianas, increase their stem resistance to bending and breaking during

PAGE 17

5 growth and maturity (Rowe & Speck 1996). Similarly, leaves become tougher with aging (Wright & Cannon 2001). Therefore, I predicted an increase in mechanical strength of stems and leaves, both at th e tissue and at the structural level. Materials and Methods The study was conducted in Barro Colorado Natural Monument (BCNM), Panama (9 10 N, 79 51 W). Data were collect ed during the rainy season (May-December), when 78% of the average annual precipitat ion of 2600 mm falls. Climate, flora, and ecological characteristics of the seasonally moist tropical forest in BCNM are well described by Croat (1978) and by Leigh, Windsor & Rand (1982). I collected seeds from eight common speci es in BCNM that differ in ecological characteristics such as dispersal mode, cotyledon type, and seedling establishment probability (Table 1-1). A seedling-recruitmen t index was calculated as # seeds falling m2year / # seedlings established m-2year obtained from a long-term experiment on Barro Colorado Island (BCI). Seed rain density (# seeds m-2year-1) was measured in weekly censuses from 1995-1999 in two hundred 1 m2 seed traps. Recruitment of new seedlings (# seedlings m-2year) was measured once a year from 1995-1998 during the dry season in six hundred, 1m2 recruitment plots,2 each located 2 m from three sides of the seed traps (Wright et al. 2003). Seedling shade tolerance was ranked acco rding to measurements of seedling survival in the shaded understory from four independent studies conducted in BCNM (Table 1-2). Alvarez-Clare (C hapter 2) determined firstyear survival of seedlings transplanted to the forest unde rstory at first leaf expansion, and censused weekly for the first 3 mos, then biweekly for the rest of the year. Kitajima (unpublished data) and Myers (2005), both quantified first year survival of s eedlings transplanted to fenced enclosures from which vertebrate pred ators were excluded. Wright et al. (2003) estimated survival

PAGE 18

6 probability after 1 yr, using na turally recruited seedlings in the forest understory. Only studies by Alvarez-Clare and by Kitajima in cluded all the species assessed here, and therefore survival per species from these tw o sources was averaged to determine mean % survival (Table 1-2). Mean % survival was obtained from 2-6 mo survival from Alvarez-Clare, which excludes initial transp lant shock and most vertebrate predation (Chapter 2), and survival from 0-4 mo from Kitajima, which was the interval before most plants were harvested. Mean percent survival was correlated with biomechanical traits at 1 and 6 after mos the expa nsion of the first leaf. Seeds were germinated in trays in a sh aded house where daily total photosynthetic photon flux density was adjusted with shade cloth to approximately 2% of full sun. I transplanted 45 seedlings of each species to each of three 6 x 6 m common gardens located on 70-year-old secondary forest on Buena Vista Peninsula. To standardize by ontogenetic stage based on development of photosynthetic organs, I transplanted seedlings at expansion of the first leaf for all species with reserve cotyledons and at expansion of cotyledons for Tabebuia rosea a species with phot osynthetic cotyledons. Time from germination until leaf expansion varied across species from one week for Anacardium excelsum, to four weeks for Eugenia nesiotica and Tetragastris panamensis. Each garden was situated under closed ca nopy and surrounded by a 1 m tall wire mesh fence to exclude large, ground-dwelling herbivores. In each garden, seedlings were transpla nted 50 cm with each species randomly located within planting positions in each plot I replaced those that died within the first week after transplanting. Half of the seedlings from each species were harvested after 1 mo (T1), and the remaining plants were harvested approximately 6 mo later (T2).

PAGE 19

7 Forty-five plants (15 per garden) of each spec ies were randomly chosen at T1 and used to perform biomechanical tests in situ before being harvested, and then used in the laboratory to test biomechanica l properties in different orga ns (e.g., if stem flexibility was measured in the field, leaves were measur ed in the laboratory). Because of mortality, at T2 only 30 seedlings per species were meas ured. After being harvested, all plants were refrigerated for less than 12 h until laboratory biomechanical tests were performed. After testing, plants were separated into stems, r oots, and leaves, weighed and then dried at 100 C for 1 h and then at 60 C for 48 h to determine dry weight. Samples were saved for fiber analysis. Biomechanical Measurements Youngs modulus of elasticity Youngs modulus of elasticity ( E ) of stems was measured in a three point bending test with a Portable Universal Tester (Darvell et al. 1996), as described in Lucas et al. (2001). More specifically, Youngs modulus was calculated from the slope of the linear regression of the applied bending force vs. de flection. Span distance varied with stem size and bending resistance. Span ratios of > 10 were always used, as suggested by Niklas (1992). Youngs modulus of elasticity is defined as the ratio between forces of stress and strain, measured within the plas tic range of a homogeneous materi al (i.e., stiffness) In the case of stems, because they are constructed of heterogeneous and composite materials, I measured an apparent Youngs modulus, wh ich describes the ove rall bending properties of a stem independent of size and shape (Niklas 1992). Fracture toughness I measured fracture toughness for stems and leaves by performing cutting tests with a sharp pair of scissors mounted on a Portab le Universal Tester as described by Lucas &

PAGE 20

8 Pereira (1990). Toughness obtained through cu tting tests is the work required to propagate a crack over a unit area (Lucas et al 2000) and has been used in leaves as an indicator of resistance to herbivory, pathogens, and othe r physical damage (Lucas & Pereira 1990, Choong 1996). For leaves, toughness was measured for lamina and midrib separately. When measuring stems, I cut the stem at half of th e total length, or just above the cotyledons in A. excelsum and T. panamensiss which have epigeal cotyledons. Density Tissue density was calculated for leaves and stems as the ratio of dry mass to volume. For leaves, volume was calculated as total leaf area (measured in the leaf area meter) multiplied by the lamina thickness, a nd dry mass was obtained for the total leaf including midrib and veins. For stems, volume was obtained from the formula: V = ( r 2) h (1-1) where r is the radius measured at the middle of the stem and h is stem length, both measured in mm. For measuring density and the other biomechanical properties, stems were considered perfect cy linders, ignoring taper. Chemical analysis To evaluate fiber content a nd relate it with biomechanic al measures, fiber fractions were determined for stem and leaf tissues separately, using a series of increasingly aggressive extractants (Ryan, Melillo & Ri cca 1989) with a fiber analyzer system (ANKOM Technology, NY, USA). Dr ied plants of each species from the same common garden and same harvest were combined and ground as one sample to have a minimal of 0.5 g required for analysis. Because of the small size of T. rosea all harvested plants were combined and ground as one sample. In the first step, each ground sample was weighted and sealed in a chemical resi stant filter bag. The bagged samples were

PAGE 21

9 submerged, heated, and agitated in neutral de tergent fiber solution removing soluble cell contents and leaving non-detergent fibe r (% NDF). In the second step, the bagged samples were treated with acid-detergent solution, which removed hemicellulose and left acid-detergent fiber (% ADF) consisting of cel lulose, lignin, cutin and insoluble ash. In the third step, samples were treated with 70% sulfuric acid, which removed cellulose and left lignin, cutin and insoluble ash inside the bags. Between steps, sample bags were dried at 100 C overnight to determine the dry mass, and each fiber fraction was calculated by subtraction. Afterwards, the remaining sample was combusted at 500C to determine percent insoluble ash. Mass of labile cell contents + hemicellulose + cellulose + lignin + insoluble ash add up to 100% of the original dry mass. Percent critical height Percent critical height (% Hcr) measures the relationship between stem height and how tall it could be before it buckles under its own weight (Holbrook & Putz 1989). Percent critical height was calculated for each seedling st em according to the formula given by Greenhill (1881): Hcr = 1.26(E/w)1/3 (db)2/3 (1-2) where E = Youngs modulus of elasticity (Pa), w = fresh weight/unit volume (Nm-3), and db = diameter at base (m). The ratio of Hcr to the actual stem height multiplied by 100 is % Hcr, which is an indication of mechanical ri sk-taking. In other words, the higher the % Hcr the lower the margin of safety for the stem to remain free-standing. Flexural stiffness Flexural Stiffness ( EI ) describes the ability of a structure to withstand mechanical loads, taking into account the size and shap e of the structure as well as the material properties of its tissues (Gartn er 1991). It is the product of E which describes the

PAGE 22

10 flexibility of the material, and the second moment of area ( I ), which reflects size and the geometry of the structure to which a force is being applied. I estimated flexural stiffness ( EI ) for cylindrical stems using the formula: I = 0.25 r4 (1-3) where r (mm) is the radius measured in the middle of the stem and the Youngs modulus of elasticity ( E ) obtained with three point bending tests, as described above (Niklas 1992). Work-to-bend Resistance of stems to bending, here refe rred to as work-to-bend, was obtained empirically in the field by applying a force ve rtically from above a seedling until the stem was deflected to 70-60% of its original he ight. To estimate work-to-bend a 2 L plastic container was mounted on a 30 cm2 Styrofoam platform and hung from a tripod with a spring balance just above the seedling. The Styrofoam platform was in contact with the uppermost part of the seedling, without bendi ng it. Then, water was poured slowly into the container, until the weighted platform be nt the stem to the specified extent. Assuming that acceleration was nil, water weight (force ) times vertical displacement, was calculated as work to bend the seedling. Whole stem flexibility To further describe the behavior of intact seedlings rooted in the ground in response to mechanical stress, I measured whole st em flexibility (Holbr ook & Putz 1989) in the field. A stem was pulled horizontally in four directions with spri ng balances until bent 20 from vertical. This proce dure was repeated in the four canonical directions and the forces averaged. Whole stem flexibility (WSF) was expressed as angular deflection divided by applied force (radi ans/N). In the case of E. nesiotica I bent the stem 40,

PAGE 23

11 because the force required to bend the stem 20 was too small to be detected in its small seedlings. Whole stem flexibility is a measure of elasticity whereas flexural stiffness is a measure of rigidity; therefor e I expected them to be inversely correlated. Because WSF applies a lateral tensile force (the stem is pulled laterally), and work-to-bend applies a vertical compressive force (the plant is pushed down), slightly different stem properties are being measured, and thus I performed both tests. Force of fracture For leaves, force of fracture was calculat ed as the product of fracture toughness by lamina thickness. This structural measur ement indicates total force necessary to propagate a crack considering leaf thickness (Wright & Cannon 2001). Specific leaf area Specific leaf area (SLA) was calculated as th e ratio of leaf area, measured with a leaf area meter (LICOR-3100), and leaf tota l dry mass. Because species with low SLA are usually thick and/or dense (Wright & Ca nnon 2001), I expected SLA to be inversely correlated with leaf fracture t oughness and force of fracture. Statistical Analyses Every biomechanical measurement was av eraged for each species, and species means were log-transformed to meet normality assumptions for ANOVA tests (Shapiro-Wilk, = 0.05). For each measurement, the effect of species ( N = 8) and harvest time ( N = 2) was evaluated using twoway ANOVAs. When the species*time interaction was significant, the data for each ha rvest were analyzed separately. To test if means differed between T1 and T2 within each species, t-test s with subsequent Bonferroni corrections were applied. Fo r across-species comparisons between two biomechanical measurements or between a biomechanical measurement and survival,

PAGE 24

12 linear regressions on log-log plots were calculated. For multiple across-species comparisons between non-normal variables, Spearman rank correlations were applied. Two means per species (one pe r harvest) were obtained to evaluate the correlation between biomechanical traits ( N = 16). Work-to-bend was only measured at T1 ( N = 8) for logistic reasons. For Spearman correlati ons between biomechanical properties and % survival in shade, species means were evaluated at each harvest separately ( N = 8). All analyses were performed using JMP IN 4.0 (S AS Institute Inc., Cary, NC, USA) with a significance level of = 0.05. Results Stem Biomechanics Mean Youngs modulus of elasticity ( E ) of the seedling st ems varied 20-fold among species (Figure 1-1A). Most species increased their resistance to bending ( E ) during the six-month period between T1 and T2 (Table 1-3) resulting in significant time effect without a species*tim e interaction. Mean stem fr acture toughness also varied among species and between harvests (Figure 1-1B ), but the amount of increase in fracture toughness varied among speci es (Table 1-3). While E. nesiotica increased its mean fracture toughness threefold from 1-6 mos after leaf expansion, A. exelsum and T. panamensis showed no increase (Figur e 1-1B). Mean stem tissu e density also varied among species and between harvests, increasing from T1 to T2 for all species except A. cruenta which decreased its mean stem tissue de nsity over time (Table 1-3, Table A1). Total fiber (% NDF) was generally higher for more shade-tolerant species (Table 1-3), but A. cruenta the species with highest survival in shade, had a mean % NDF similar to the three least shade-tolerant species (F igure 1-2A). Mean % NDF did not differ

PAGE 25

13 significantly between harvests (Table 1-3 and Table A-3). A ll individual fiber fractions varied between species, but only % hemicellulo se increased between harvests (Table 1 3). Most material properties of stems were in ter-correlated (Figure 1-3 and Table 1-4). Material mechanical traits, such as toughne ss and modulus of elas ticity, were positively correlated (Figure 1-3A). Additionally, material properties describing mechanical strength (e.g., modulus of elasti city and density) were correlat ed with chemical indicators of tissue strength (e.g., fiber content; Figure 1-3B-D). Am ong chemical properties, % cellulose was the best predictor of mechan ical strength, as m easured by modulus of elasticity, fracture toughness, a nd density (Table 1-4). Per cent lignin was not correlated with toughness or density but was a good predictor of modulus of elasticity (i.e., stem stiffness). Mean percent critical height (% Hcr) varied among species and significantly decreased in three out of eight species from T1 to T2 (Fi gure 1-4A and Table 1-3). All species had low % Hcr (their actual height was 14-28% of their critical height), indicating that seedlings were overbuilt relative to th eir potential maximum height before buckling under their own weight. Mean flexural stiffness ( EI ) varied among species and between harvests, with an interaction between factor s (Figure 1-4B and Table 1-3). The significant interaction was apparently influenced by A. exelsum and G. superba the species with the largest seedlings (i.e., largest I ), which disproportionately increased EI from T1 to T2. Mean work-to-bend (i.e., work necessary to bend a stem to 70% of its original height) varied four-fold among species (Figure 1-4C and Table 1-3) Stem diameter was a good predictor of work-to-bend ( r2 = 0.53, F = 73.0, d.f. = 1,66, P < 0.001), and consequently there was a positive correlation between EI and work-to-bend (Table 1-4). Mean whole

PAGE 26

14 stem flexibility, measured as angular defl ection, varied among species decreasing over time as stems became more lignified (Figure 14D and Table 1-3). Plants with large stem diameters were less flexible than plants with small stem diameter ( r2 = 0.64, F = 289.70, d.f. = 1,160, P < 0.001). Although mechanical and chemical traits of stem tissues were intercorrelated, they were never correlated with st ructural measurements that integrated material and morphological traits (Table 14). The only exception was % Hcr, which was negatively correlated with modulus of elasticity ( E ), fracture toughness, % NDF, and % cellulose. This observation indicates that species w ith stronger material had a lower % Hcr and hence a greater safety margin. Second moment of area ( I ), did not correlate with any of the material properties. In contrast, both I and flexural stiffness ( EI ) correlated positively with structural traits, such as work-to-bend and whole stem flexibility measured on intact seedlings in the field (Table 1-4). Leaf Biomechanics Material biomechanical traits of l eaves differed among species and between harvests, although not all species varied consis tently between T1 and T2. Lamina fracture toughness differed among species with a signific ant interaction between species and time (Table 1-5). Two species increased their lamina toughness, two decreased, and four species did not vary between T1 and T2. Fracture toughness of midveins varied between species and between harvests, with a significant interaction between these two factors (Figure 1-5B, Table 1-5). In general, for each species midvein fracture toughness was lower or similar than stem toughness, but mu ch higher (ca. x10) than lamina toughness. Leaf density also varied among species and be tween harvests (Table 15). A significant

PAGE 27

15 interaction between species a nd time was probably because le af density increased from T1 to T2 in B. pendula much more than in other sp ecies (Table A-2). Percent NDF differed among species but not between harves t times (Figure 1-2B and Table 1-5). All individual fiber fractions varied among species, but onl y % lignin changed between harvests (Table 1-5). Structural properties of leaves, integr ating material properties and morphology varied among species, but only SLA diffe red between harvests (Table 1-5). Tabebuia rosea had the highest SLA, while A. cruenta had the lowest SLA. Force of fracture was different among species, but not between harv est times (Figure 1-5 and Table 1-5). Biomechanical attributes of midveins highly influenced mechanical traits of the whole leaf. Across species, th ere was a positive correlation between lamina and midvein toughness (Table 1-6). Total leaf density was best correlated with midvein than with lamina toughness, suggesting that biomechanical attributes of the midvein significantly influence overall leaf density. Percent cellu lose was the chemical trait that most correlated with the rest of the material tr aits. Force of fracture (toughness*thickness) was more correlated with to ughness than with thickness, indica ting a stronger effect of leaf material properties than of leaf dimensions. Relationship between Biomechanic al Traits of Stems and Leaves Across species, there was a positive correlation between stem toughness and midvein toughness, but not between stem t oughness and lamina toughness (Table 1-7). Tissue density and % NDF were also correlate d between stems and leaves, but the other fiber fractions were not (data not shown).

PAGE 28

16 Relationship between Seedlin g Biomechanics and Survival Several stem material biomechanical prope rties were positively correlated with % mean survival in shade (Table 1-8). Fr acture toughness and tissue density measured at T2 showed the highest correlations with surv ival in both stems and leaves (Figure 1-6, Table 1-8). Furthermore, if A. cruenta (the species with high survival but with low E and % fiber content) was removed from the an alyses, all correlations between material biomechanical traits and survival increased. Although individual fibe r fractions exhibited no significant correlation with survival, % NDF (i.e., total fiber) was positively correlated with survival in both stems and leaves. Stem a nd leaf structural prope rties, at 1 and 6 mos after expansion of the first leaf, were not correlated with survival in shade. Discussion Stem Biomechanics Mechanical traits and chem ical composition of seedling stems varied widely among eight species of tropical trees but as predicted, stems of shade-tolerant species were generally stiffer, tougher, and denser, and w ith higher total fiber content (% NDF) than stems of shade intolerant species (Figures 1-1 and 12A). Among the biomechanical properties tested there were positive correlations between Y oungs modulus of elasticity, fracture toughness, and stem density suggesti ng a greater overall i nvestment in strong material properties in shade-tolerant speci es. Similar results we re obtained by Cooley, Reich & Rundel (2004) for understory herbs. Although in my study there were positive correlations between mechanical and chemi cal material traits, the fiber components contributing to these correlations differed, w ith the mechanical property considered. For example, fracture toughness was correlated w ith % cellulose and % hemicellulose, but not with % lignin (Table 1-4). As a co mplex, heterogeneous polymer with strong

PAGE 29

17 covalent bonds, lignin acts as an adhesive agent in the cell wa ll, and therefore is expected to increase stiffness rather than toughness (Lucas at al 2000). In fact, the only mechanical property correlated with % lignin was E a measure of stem stiffness. Modulus of elasticity, however, can also be affected by othe r tissue properties such as volume fraction of cell wall materials (Lucas et al 2000, Niklas et al. 2000), hemicellulose and cellulose contents, and microf ibril angles in the cell wall of fiber cells (Hoffman 2003, Savidge 2003). Differences in material properties at the time of first leaf expansion (T1) suggest that shade-tolerant species invested earlier in stem mechanical construction than shade intolerant species. Thus, shadetolerant species potentially ha d a more developed vascular cambium and greater secondary cell wall depos its than shade intolerant species. Mean moduli of elasticity ( E ) for shade intolerant species at T1 were similar to those reported for stems of understory herbs (Cooley, Reich & Rundel 2004, Niklas 1995). This suggests that 1 mo after leaf expansion, vascular cambium development (and thus secondary growth) was still limited, and seed lings were relying on primary tissues for mechanical support (Niklas 1992, Isnard, Sp eck & Rowe 2003). In contrast, shadetolerant species (e.g., T. panamensis and E. nesiotica ) had moduli of elasticity at T1 of the same order of magnitude as wood from 15 of 33 adult temperat e trees evaluated by Niklas (1992). Species with st ronger material properties had higher fiber contents as well. Specifically, they had higher % lignin and % cellulose fractions, which are correlated with vascular cambium maturation, high cel l wall volume fraction, and secondary cell wall development (Niklas et al. 2000, Lucas et al 2000). Because shadetolerant species are usually slow growers (Kitajim a 1994), it is not likel y that further stem maturity at the

PAGE 30

18 time of first leaf expansion in shade-tolerant species was a product of accelerated stem development. On the contrary it reveal s an ecological stra tegy, characterized by substantial investment in material starting very early in ontogeny. Variation in stem development at T1 coul d be influenced by leaf emergence times. Kitajima (2002) demonstrated that T. rosea a light-demanding species with photosynthetic cotyledons, became dependent on photosynthetic carbon gain earlier in development than shade-tolera nt species with storage co tyledons. Rapid photosynthetic cotyledon expansion after radicle emergence (22.5 + 1.9 d), allows little time for stem structural development and toughening. In contrast, T. panamensis a shade-tolerant species with reserve cotyle dons, expands its first leaves relatively quickly (23.6 + 2.4 d), but has a high modulus of stem elasticity. Although age (time after radicle emergence) may potentially affect stem stiffness and toughness, this is evidently no t the sole cause of variation. Among species variation in biomechan ical properties of stems at first leaf expansion is a function of differences in ma terial composition and structural arrangement, which suggest the existence of different ecological strategi es among species of tropical tree seedlings. I predicted that material tra its of seedling stems would be reflected at the structural level. Thus, I expected stems with stronger ma terial properties per unit area (or mass) to be more resistant to bending and breaking. Results confirmed this prediction, but only when stems of similar size were compare d. When different sized seedlings were compared, species with larger seedlings (at comparable developmental stages) were more resistant to bending, both for tests performed in the laboratory and on intact seedlings in the field. A plant can obtain a hi gh flexural stiffness by increasing E (material stiffness),

PAGE 31

19 or by increasing I a measure of size and shape (Niklas 1992). Given that seedlings of all eight species included in my study had circular stem s, the observed differences in I reflect differences in size only. Likewise, di fferences in flexural stiffness among species were mostly influenced by size of the stem ( I ), as opposed to flexibility of the material ( E ). Similar results have been reported for neotropical understory herbs (Cooley, Reich & Rundel 2004), vines (Rowe & Speck 1996), shr ubs (Gartner 1991), and trees (Holbrook and Putz 1989). In contrast, other st udies have found an influence of both E and I when comparing flexural stiffness of stems growi ng in environments differing in wind intensity and shade conditions (Cordero 1999, Henry and Thomas 2002), and when comparing stems from congeneric species differing in gr owth form (Isnard, Speck & Rowe 2003,). When intact, live stems were tested in the field, work-to-bend and whole stem flexibility correlated with other structural traits, but not with material properties (Table 1-4). The results of thes e field tests correlated well wi th flexural stiffness, which was measured using harvested stems in the laboratory. Whole stem flexibility and workto-bend proved good field indicators of stem rigidity for tropical tree seedlings, and should be taken into account in future re search regarding seed ling biomechanics. The structural property that best correlated with material properties was % critical height. Seedlings from shade-tolerant specie s had higher safety factors (i.e., lower % Hcr), than seedlings from shade into lerant species. As suggested by Givnish (1995), my results indicate that there is a trad e-off between light acquisition and mechanical safety. While some trees maximize their height to reach light and overtop competitors, this increases vulnerability to toppling (Holbrook & Putz 1989, Brchert, Becker & Speck 2000). Although all species in my st udy were overbuilt (Figure 12A), light-demanding species

PAGE 32

20 had higher % Hcr and weaker material traits than sh ade-tolerant species, suggesting that they were maximizing height growth at the expense of safety and structure. As predicted, all species increased their mean E between 1 and 6 mos after leaf expansion, although not always significantly (Figure 1-1). In contrast, there was no pattern to the proportional increase in fracture toughness between T1 and T2 among species, revealing that speci es do not necessarily incr ease toughness and stiffness proportionally during ontogeny. Thus for seven out of eight sp ecies in which stem fiber content did not increase from T1 to T2, in creases in stiffness and toughness over time must have been caused by changes in stem anatomy, such as fiber distribution and packaging, as opposed to increased fiber content (Hoffman et al 2003), but further anatomical and histological analyses are necessary. Leaf Biomechanics Mean lamina and midvein toughness vari ed 30-fold among species, with values from 71 to 395 J m -2 for laminas and 984 to 3475 J m-2 for midveins. In a study performed on BCI with leaves from adult trees and understory saplings, Dominy, Lucas & Wright (2003) reported considerably highe r values for lamina and midvein toughness than reported here. Nevertheless, for th e three species used in both studies ( A. excelsum, C. elastica, and A. cruenta ), the same ranking prevails: A. excelsum had the lowest lamina and midvein toughness while A. cruenta had the highest. Although the relationship was weaker than in stems, l eaves of shade-tolerant species had higher mechanical strength than leaves of shade in tolerant species. Pote ntially, evolutionary forces favoring selection of other leaf tr aits, such as photosynt hetic capacity, vein distribution, presence of sec ondary compounds, and water-use efficiency also influence differences in leaf toughness among species (Choong et al. 1992, Wright et al 2004).

PAGE 33

21 Fiber content is an indicat or of biomechanical strengt h in leaves (Choong 1996). In the present study, mean % cellulose was the fibe r fraction that best correlated with leaf fracture toughness, suggesting that cell wall material was the predominant cellular component influencing fracture toughness (Esa u 1977), but it is not clear which tissues make a leaf tough. Both the cuticles (Tay lor 1971) and the epidermis (Grubb 1986) have been proposed as toughening tissues. Additiona lly, Wright & Illius (1995) reported that the proportion of sclerenchyma in leaves was correlated with fracture toughness of grasses, and Choong (1996) found th at the non-venous lamina c ontributed little to overall leaf toughness. In the present study, the posit ive correlation between midvein and lamina toughness suggests that vascular bundles (and probably fibers associated) were the major determinants of fracture toughness in leaves. Structural measurements integrating leaf di mensions and size were correlated with material traits but not with morphological tr aits (Table 1-6). For example, force of fracture, calculated as the product of lami na toughness and leaf thickness, was better correlated with lamina toughness than with l eaf thickness. Thus, unlike stems, overall leaf biomechanical properties were influenced more by ma terial traits than by leaf dimensions. Similar results were reported by Wright & Cannon (2001) in a study with 17 sclerophyllous species from low-nutrient woodland in eastern Australia. I expected that biomechanical strength of leaves would increase over time; however, most species did not change, and so me even decreased in their mechanical strength between T1 and T2. In fact, for G. superva and T. panamensis mean lamina fracture toughness decreased si gnificantly after 6 mos. Although there is no evident explanation for this observation, Lucas & Pe reira (1990) found the same trend (where

PAGE 34

22 leaves decreased their fractur e toughness over time). They s uggested that an increase in parenchymatous tissue and air species in ol der leaves could result in low fracture toughness per unit volume. Relationship between Biomechanic al Traits of Stems and Leaves Measurements of the material traits of stems and leaves were positively correlated for the eight species combined. Species w ith tough, dense stems also had tough, dense leaves. An exception was A. cruenta which had tough, thick leaves, but stems constructed of weak and flexible material. Aspidosperma cruenta also stores substantial amounts of nonstructural carbohydra tes in its stems, which ma y augment its ability to recuperate from damage, rather than a void it (Myers 2005). Across species, the correlation between stem and midvein t oughness was stronger than the correlation between stem and lamina toughness. The str ong relationship between stems and midveins could be driving the relationship between st em and leaf density or fiber content, suggesting consistent investment s in vascular structure thr oughout the plant. Collectively, these results suggest that there is a w hole-plant pattern of carbon investment in mechanical defenses, as opposed to a trade-o ff between investment in stem and leaves. Further investigations might evaluate whether this pattern remains consistent in roots. Relationship between Seedlin g Biomechanics and Survival Material properties of stems correlated w ith 0-6 mo survival in shade (Table 1-8, Figure 1-6). Species stems cons tructed of tougher, stiffer, denser, and more fibrous material showed higher percent survival than species composed of weaker material. This is direct evidence that biomechanical stre ngth of stem tissues increases seedling performance in the tropical forest understory. As suggested in prev ious studies, strong material is likely to confer an advantage against mechanical damage caused by litterfall,

PAGE 35

23 vertebrate trampling, and herbivory (A ugspurger 1984a, Clark & Clark 1991, Moles & Westoby 2004a). The only species that deviated from the trend was Aspidosperma cruenta Seedlings from this species had the highes t survival in shade, but its stems were constructed of weak material. Most likely, high survival in A. cruenta was due to the presence of chemical defenses and large reserve pools of carbohydrates in stems and roots. Aspidosperma cruenta is well known for its poisonous alkaloids (e.g., obscurinervine and obscurinervidine, Harper et al 1993), and well-developed chemical defense that may compensate for it s low structural defens e, revealing a unique ecological strategy among the ei ght species tested. It shoul d be noted that chemical defenses confer herbivore resistance (Col ey 1983), but do not protect seedlings from mechanical damage due to litterfall or ve rtebrate trampling. The high survival of A. cruenta on BCI, albeit its lack of mechanical defenses, suggest s that for this species defense against herbivory and pathogen s (through secondary compounds) was more important as a selective factor than defense against mechanical damage, at least during the first 6 mos. Surprisingly, structural trai ts that integrate material properties with seedling size and shape were not correlated with six-mont h survival in shade. Larger seedlings had higher overall resistance to bending (Figure 14), but with no apparent consequence for seedling survival. Although previous studies have emphasized the advantages of large size for seedlings (reviewed in Moles and Westoby 2004a), my results suggest that evolutionary pressures selecting for large seedlings are probably related to stresstolerance (Green & Juniper 2004) and light acquisition (Turner 1990), not to biomechanical strength.

PAGE 36

24 For leaves, there was a positive correlation between some of the biomechanical traits and survival in shade but the trends were not as str ong as for stems. Most likely, leaf biomechanical traits are directly correla ted with leaf performan ce (e.g., leaf lifespan or risk of herbivory), but not with whole pl ant performance (e.g., survival). For example, Wright and Cannon (2001) found that mean l eaf toughness, force of fracture, leaf thickness, and leaf area explaine d between 30 and 40% of varia tion in leaf life span of 17 species of sclerophyllous plants. In a study with 2,548 species, Wright et al. (2004) found that leaf mass per area (L MA), explained 42% of the va riation in leaf life span, indicating that thicker, denser leaves, us ually live longer. Additionally, the weaker correlations I observed between survival and m echanical traits of leaves suggest that invertebrate herbivores that cause leaf da mage are not crucial determinants of wholeplant survival during the first 6 mo, for the ei ght species considered in my study (Chapter 2). Conclusions Interspecific variation in material flexibility and fr acture toughness of seedling stems as early as one month after leaf expans ion, revealed different ecological strategies to cope with mechanical damage in the fo rest understory. Shade-tolerant species had stems constructed of strong materials, whic h may promote their survival in shade. However, stronger material properties of st ems did not always reflect strength at the structural or whole-plant le vel. Size and several morphologi cal traits contributed to overall resistance to bending a nd breaking stress, but they apparently were not crucial for seedling survival from 0-6 mo. As opposed to stems, leaf biomechan ical properties were influenced more by material traits than by l eaf dimensions, and biom echanical attributes of leaves were not always co rrelated with whole-plant surviv al. In tropical tree seedlings,

PAGE 37

25 differential survival in shade is the product of a suit of traits of which biomechanics is an important component.

PAGE 38

26Table 1-1. Ecological characterist ics of eight tropical tree sp ecies used in my study, listed by increasing shade tolerance. Sp. code Species Family Cot. type Dispersal %Rec. index Seed mass (g) TABR Tabebuia rosea Bignoniaceae PEF Wind 0.6 0.035 + 0.007 (12) ANAE Anacardium excelsum Anacardiaceae PER Animal 0.1* 1.811 + 0.316 (9) CASE Castilla elastica Moraceae CHR Animal 0.315 + 0.005 (8) BEIP Beilschmiedia pendula Lauraceae CHR Animal 13.7 2.360 + 0.090 (10) GUSS Gustavia superba Lecythidaceae CHR Animal 3.7* 5.566 + 1.746 (7) TETP Tetragastris panamensis Burseraceae PER Animal 3.5 0.179 + 0.026 (10) EUGN Eugenia nesiotica Myrtaceae CHR Animal 27.8* 0.474 + 0.067 (10) ASPC Aspidosperma cruenta Apocynaceae PHR Wind 2.9* 0.492 + 0.002 (6) Cotyledon types are according to Garwood (1996) : PEF = phanerocotylar epigeal foliaceous PER = phanerocotyla r epigeal reserve, CHR = cryptocotylar hypogeal reserve, and PHR = phanerocotylar hypog eal reserve. Percent recruitmen t (% Rec. Index) refers to percent recruits per seeds per area (Wright et al. 2003). Mean + 1 SD ( N ) seed mass without seed coat. Data obtained with between 5 and 10 recruits. Cotyledons are partially cryptocotylar.

PAGE 39

27Table 1-2. Percent seedling survival for th e eight study species over specified periods from four independent studies in BCNM. Numbers in parentheses indicate sample size, ( i.e., the total number of individuals at th e beginning of the measurement period). Values shown in bold were averaged for each species and used to calculate mean % survival. Refer to Table 1-1 for species codes aThis study. Seedlings transplanted to the fore st and monitored for 1 yr (Chapter 2). Ti me is divided into different stages beca use initial mortality during 0-2 mo was due mainly to vertebrate activity, and thus is not a good i ndicator of shade tolerance. bSeedlings transplanted at the time of germination (K. Kitajima, unpublished data) or at time of first leaf full expansion (Myers 2005) to exclosures in the forest understory and mon itored weekly for 1 yr. These seedlings were protected from vertebrate herbivores. cPercent of seedlings that survived at least 1 yr after germinating natu rally in the forest understory(Wright et al 2003). Sp. code Alvarez-Clare a Kitajima b Myers b Wright c Mean % survival 0-2 mo 2-6 mo 6-12 mo 0-4 mo 4-12 mo 0-6 mo 6-12 mo 0-12 mo TABR 45.5 33 (55) 44 (18) 29 (7) 47 (48) 30 (23) 33 (71) 46(14) 31 (58) ANAE 53.0 20 (100) 40 (20) 11 (9) 66 (51) 26 (34) CASE 65.0 40 (100) 73 (40) 72 (25) 57 (28) 67 (18) 65 (101) 86(44) BEIP 82.5 8 (100) 88 (8) 60 (5) 77 (61) 19 (47) 52 (826) GUSS 76.0 54 (99) 83 (54) 79 (43) 69 (42) 86 (32) 57 (213) TETP 82.0 62 (100) 79 (62) 82 (71) 85 (20) 90 (10) 64 (361) EUGN 87.5 43 (100) 100 (42) 75 (32) 75 (63) 96 (47) 81 (22) ASPC 87.0 78 (100) 93 (78) 82 (71) 81 (27) 99 (21) 98 (111) 97(104)

PAGE 40

28 Table 1-3. Effect of species and harvest time on material and structural properties of seedling stems. Shown are F values from two way ANOVAs performed on log-transformed values; d.f. = 7,1; ** P < 0.001 Effect Biomechanical measurement Species Time Species*Time Modulus of elasticity (MN m-2) 151.5**90.0**1.5 Fracture toughness (J m-2) 70.5**114.4** 9.7** Stem tissue density (g cm-3) 219.8**85.6** 19.6** % NDF 57.7** 0.8 1.9 % Hemicellulose 25.3**17.8** 4.0** % Cellulose 30.7** 0.1 1.3 % Lignin 38.4** 2.0 1.4 % Critical height 60.8** 133.8** 4.8** Flexural stiffness (N cm2) 111.3** 163.9** 4.7** Work-to-bend (J) 18.0** Whole stem flexibility (radians/ N)100.9**102.8** 5.81**

PAGE 41

29 Table 1-4. Relationships among stem biomechani cal traits for seedlings of eight tree species. E (MN m2) Tough (J m-2) Density (g cm-3) % NDF %Hemicell%Cellulose % Lignin Toughness 0.80 (< 0.001) Density 0.77 (<0.001) 0.73 (0.002) % NDF 0.90 (<0.001) 0.75 (<0.001) 0.83 (<0.001) % Hemicell 0.51 (0.041) 0.58 (0.024) 0.46 (0.070) 0.40 (0.128) % Cellulose 0.77 (<0.001) 0.70 (0.004) 0.60 (0.014) 0.86 (<0.001) -0.11 (0.704) % Lignin 0.63 (0.009) 0.36 (0.191) 0.33 (0.213) 0.76 (<0.001) 0.03 (0.905) 0.62 (0.011) I 0.27 (0.316) 0.19 (0.491) 0.45 (0.083) -0.36 (0.165) -0.08 (0.837) -0.29 (0.284) -0.28 (0.300) % Hcr -0.68 (0.004) 0.71 (0.003) 0.40 (0.122) -0.59 (0.014) -0.44 (0.085) -0.67 (0.005) 0.43 (0.094) EI 0.14 (0.612) 0.18 (0.526) -0.17 (0.519) -0.05 (0.841) 0.14 (0.593) -0.01 (0.97) -0.01 (0.970) Work-tob end -0.29 (0.535) 0.14 (0.760) -0.57 (0.180) -0.39 (0.383) -0.25 (0.589) -0.25 (0.589) 0.00 (1.000) WSF 0.14 (0.612) 0.06 (0.829) 0.40 (0.140) -0.34 (0.221) -0.11 (0.704) 0.11 (0.819) 0.00 (1.00) Seed mass (g) 0.21 (0.610) 0.33 (0.420) 0.24 (0.570) 0.14 (0.736) 0.29 (0.493) 0.21 (0.610) 0.10 (0.823)

PAGE 42

30 Table 1-4. Continued I (mm4) % Hcr EI (N m2) Work-tobend (J)* WSF (radians/N) Toughness Density % Total fiber % Hemicell % Cellulose % Lignin I % Hcr 0.17 (0.528) EI 0.87 (<0.001) -0.12 (0.667) Work-tobend 0.86 (0.014) 0.00 (1.000) 0.99 (<0.001) WSF -0.81 (<0.001) 0.08 (0.790 ) 0.87 (<0.001) -0.857 (0.014) Seed mass (g) 0.74 (0.037) 0.10 (0.823) 0.79 (0.021) 0.64 (0.119) -0.82 (0.023) Shown are Spearman correlation coefficients for tests performed on species means obtained at each harvest ( N = 16) with P values in parenthe ses and significant correlations in bold. For correlations with work -to-bend and seed mass, only values of T1 were used ( N = 8). Material properties in cluded modulus of elasticity ( E ), fracture toughness (tough), density, % NDF (non-detergen t fiber), % cellulose, % hemicellulose, and % lignin. Structural m easurements integrating mate rial traits and morphology included % critical height (% Hcr), flexural stiffness ( EI ), work-to-bend, and whole stem flexibility (WSF). Second moment of area ( I ) considered only size and shape.

PAGE 43

31 Table 1-5. Effect of species and harvest time on material and structur al traits of leaves. Shown are F values from two way A NOVAs performed on log-transformed values; d.f. = 7,1; 0.001 < P < 0.05, ** P < 0.001. Effect Biomechanical measurement Species Time Species*Time Lamina fracture toughness (J m-2) 32.9** 1.5 10.9** Midvein fracture toughness (J m-2) 47.7** 49.9** 6.4** Leaf density (g cm) 44.0** 113.6** 15.1** % NDF 5.5** 1.7 1.9 % Hemicellulose 36.2** 0.1 0.6 % Cellulose 6.9** 0.1 1.6 % Lignin 24.4** 10.4* 2.2 Specific leaf area (cm2g) 119.3** 46.6** 2.0 Force of fracture (N) 46.7** 2.4 5.0**

PAGE 44

32Table 1-6. Relationships among leaf biomechanical traits for seedlings of eight tree species. Lamina toughness (J m-2) Midvein toughness (J m-2) Leaf density (g cm-3) % NDF % Hemicell % Cellulose % Lignin Leaf thickness (mm) SLA (cm2g-1) Midvein toughness 0.68 (0.004) Leaf density 0.52 (0.039) 0.67 (0.005) % NDF 0.21 (0.438) 0.04 (0.897) 0.31 (0.249) % Hemicell -0.17 (0.520) -0.11 (0.664) 0.15 (0.579) 0.30 (0.264) % Cellulose 0.74 (< 0.001) 0.67 (0.005) 0.33 (0.213) 0.29 (0.279) -0.07 (0.787) % Lignin 0.03 (0.914) -0.33 (0.217) -0.05 (0.846) 0.59 (0.017) -0.25 (0.350) -0.12 (0.664) Leaf thickness 0.15 (0.580) -0.03 (0.910) -0.44 (0.087) 0.04 (0.871) (-0.39) (0.131) 0.29 (0.274) 0.17 (0.535) SLA -0.58 (0.019) -0.67 (0.005) -0.56 (0.020) -0.33 (0.209) 0.24 (0.380) -0.61 (0.012) -0.01 (0.983) -0.34 (0.200) Force (N) 0.84 (< 0.001) 0.51 (0.043) 0.18 (0.513) 0.24 (0.374) -0.41 (0.119) 0.79 (< 0.001) 0.17 (0.535) 0.61 (0.012) -0.65 (0.006) Shown are Spearman correlation coefficients for tests pe rformed on species means obtained at each harvest ( N = 16) with P values in parentheses, and significant co rrelations in bold. Material pr operties included lamina toughne ss, midvein toughness, % NDF (nondetergent fiber), % hemicellulose, % cellulose, % lignin, and whol e-leaf tissue density. Structural variables integrating mater ial traits and morphology were specific leaf area (SLA) and force of fracture (Force), which was the product of leaf toughness and thickne ss.

PAGE 45

33 Table 1-7. Relationships among bi omechanical traits of stems and leaves for seedlings of eight tree species. Biomechanical trait rs P Toughness (J m-2) Lamina 0.45 0.092 Midvein 0.58 0.002 Density (g cm-3) 0.80 < 0.001 % NDF 0.58 0.019 Shown are Spearman correlation coefficients ( rs) for tests performed on species means obtained at each harvest ( N =16) and corresponding P values.

PAGE 46

34 Table 1-8. Relationships among % survival in shade and various seedling biomechanical traits of stems and leaves for seedlings of eight tree species. T1 T2 rs P rs P STEM Material properties E 0.79 0.021 0.69 0.058 Toughness 0.60 0.120 0.89 0.007 Density 0.93 <0.001 0.90 0.002 % NDF 0.74 0.037 0.71 0.047 % Hemicellulose 0.50 0.120 0.52 0.183 % Cellulose 0.48 0.233 0.45 0.260 % Lignin 0.38 0.352 0.60 0.120 Structural properties % Hcr -0.17 0.693 -0.38 0.352 EI -0.17 0.693 -0.12 0.779 Work-to-bend -0.68 0.094 WSF 0.54 0.215 0.19 0.651 LEAF Material properties Lamina toughness 0.33 0.420 0.76 0.028 Midvein toughness 0.21 0.610 0.31 0.456 Density 0.64 0.086 0.93 <0.001 % NDF 0.81 0.015 0.45 0.260 % Hemicellulose 0.33 0.420 0.14 0.736 % Cellulose 0.36 0.385 -0.05 0.912 % Lignin 0.43 0.289 0.24 0.570 Structural properties SLA -0.52 0.183 -0.60 0.120 Force of fracture 0.29 0.493 0.048 0.911 Mean % survival refers to the first column in Table 1-2. Shown are Spearman correlation coefficients ( rs) from tests performed on species means ( N = 8) and their P values. T1 and T2 = time when biomechanical measurements were taken (1 and 6 mo after leaf expansion, respectively).

PAGE 47

35 * * 0 1000 2000 3000 4000 5000 6000 7000Modulus of elasticity (MN m -2) A * * *0 4000 8000 12000 16000 20000 TABRANAECASEBEIPGUSSTETPEUGNASPCSpeciesFracture toughness (J m -2)B Figure 1-1. Means (+ 1 SD) material biomechanical traits of stems for seedlings of eight tree species ordered from left to righ t by increasing shade tolerance at 1 mo (T1, filled bars), and 6 mos (T2, open ba rs) after leaf expansion. A) YoungÂ’s modulus of elasticity ( E ). B) Fracture toughness. Aste risks indicate significant difference between T1 and T2 ( P value < 0.006 with Bonferroni correction). Refer to Table 1-1 for species codes. No data available for stem toughness of GUSS at T2 because the size of the stem s exceed size capacity of the tester (3 mm diameter).

PAGE 48

36 0% 20% 40% 60% 80% 100%% Non fiber % Hemicell. % Cellulose % Lignin A 0% 20% 40% 60% 80% 100%TABRANAECASEBEIPGUSSTETPEUGNASPCSpecies% Non fiber % Hemicell. % Cellulose % Lignin B Figure 1-2. Mean (+ 1 SD) % fiber content (% NDF) for seedlings of ei ght tree species ordered from left to right by increasing sh ade tolerance at 6 mo s after first leaf expansion (T2). A) Stems. B) Leaves ANOVA results are shown in Table 1-3 for stems and Table 1-5 for leaves. Fo r species codes refer to Table 1-1.

PAGE 49

37 log Toughness (Jm -2 ) 2.83.03.23.43.63.84.04.24.4 log Modulus of elasticity (MN m -2 ) 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 ANAE ASPC BEIP CASE EUGN GUSS TABR TETP log Toughness (Jm -2 ) 2.83.03.23.43.63.84.04.24.4 log Density (g cm -3 ) -1.0 -0.8 -0.6 -0.4 -0.2 0.0 log Modulus of elasticity (MN m -2 ) 2.02.22.42.62.83.03.23.43.63.8 log % NDF 1.65 1.70 1.75 1.80 1.85 1.90 r 2 = 0.72** Slope =1.15B r 2 = 0.72** Slope = 0.57D r 2 = 0.80** Slope = 0.14 log Density (g cm 3 ) -1.0-0.9-0.8-0.7-0.6-0.5-0.4-0.3-0.2 log % NDF 1.65 1.70 1.75 1.80 1.85 1.90 C r 2 = 0.56** Slope = 0.23A Figure 1-3. Log-log relationships between some material properties of stems for seedlings of eight tree species. Each point is a species mean at 1 mo (T1, filled symbols), and 6 mos (T2, open symbol s) after leaf expansion; ** P < 0.001. Refer to Table 1-1 for species codes.

PAGE 50

38 A B C D % Critical height 0 10 20 30 40 Flexural stiffness (N cm2 ) 0 20 40 60 80 100 120 Species TABRANAECASEBEIPGUSSTETPEUGNASPCWSF (radians/N) 0.00 0.05 0.10 0.15 Work to bend (J) 0.000 0.001 0.002 0.003 0.004 0.005 * * * * * * * * Figure 1-4. Means (+ 1 SD) structural biomechanical tra its of stems for seedlings of eight tree species ordered from left to righ t by increasing shade tolerance at 1 mo (T1, filled bars), and 6 mos (T2, open ba rs) after leaf expa nsion. A) Percent critical height. B) Flex ural stiffness. C) Work -to-bend. D) Whole stem flexibility (WSF). Asterisks indicate si gnificant difference between T1 and T2 ( P value < 0.006 with Bonferroni correc tion). Refer to Table 1-1 for species codes. Work-to-bend was only measured at T1. Work-to-bend and WSF could not be measured for TABR at T1 because of small size of seedlings.

PAGE 51

39 * *0 100 200 300 400 500 600Fracture toughness (J m -2)A * * 0 1000 2000 3000 4000 5000Fracture toughness (J m -2)B 0 20 40 60 80 100 TABRANAECASEBEIPGUSSTETPEUGNASPCSpeciesForce of fracture (N ) C * Figure 1-5. Means (+ 1 SD) biomechanical traits of l eaves for seedlings of eight tree species ordered from left to right by increasing shade tolerance at 1 mo (T1, filled bars), and 6 mos (T2, open bars ) after leaf expansion. A) Lamina fracture toughness. B) Midvein fracture toughness. C) Leaf force of fracture. Asterisks indicate significant difference between T1 and T2 ( P value <0.006 with Bonferroni correction). Notice th e different scales between A and B. Refer to Table 1-1 for species codes.

PAGE 52

40 log Stem toughness (Jm -2) 3.23.43.63.84.04.24.4 log % Survival 1.60 1.65 1.70 1.75 1.80 1.85 1.90 1.95 2.00 A r 2 = 0.56 P = 0.54 Slope = 0.28 log Stem density (g cm -3 ) -0.8-0.7-0.6-0.5-0.4-0.3-0.2 B r 2 = 0.78 P = 0.004 Slope = 0.58 log Lamina toughness (Jm -2) 2.02.12.22.32.42.52.6 log % Survival 1.60 1.65 1.70 1.75 1.80 1.85 1.90 1.95 2.00 C r 2 = 0.15 P = 0.33 Slope = 0.31 log Leaf density (g cm -3 ) -0.9-0.8-0.7-0.6-0.5-0.4-0.3 D r 2 = 0.90 P < 0.001 Slope = 0.67 Figure 1-6. Log-log relationships between so me biomechanical properties measured at 6 mos after first leaf expansion (T2) and % mean survival in shade for seedlings of eight tree species. A and B refer to stem traits, and C and D to leaf traits. Each point represents a sp ecies mean; symbols as in Figure 1-3.

PAGE 53

41 CHAPTER 2 SPECIES DIFFERENCES IN SEEDLING SUSCEPTIBILITY TO BIOTIC AND ABIOTIC HAZARDS IN THE FOREST UNDERSTORY Introduction For most plant species, mortality rates is highest during the seed and seedling phases, thus early developmental stages are critical determinants of adult abundance and distribution (Clark & Clark 1985, Condit, Hubbell & Foster 1995, Kitajima & Fenner 2000). Multiple mechanisms have been propos ed to explain existing differences in recruitment and survival across microsites in the forest unders tory, which lead to speciesspecific distribution patterns and niche pa rtitioning. For example, in the tropics differential seedling survival in contrasting light environments has been related to trade-offs involving growth, resource a llocation, and defens e (Kitajima 1994, Kobe 1999, Montgomery & Chazdon 2002) Additionally, differences in seed size (Moles & Westoby 2004a,b), dispersal mechanisms (Howe & Schupp 1984, Chapman & Chapman 1996), cotyledon function (Kitajima 1996, Ibarra-Manriquez, Martinez Ramos & Oyama 2001), carbon allocation to storage (Canham et al 1999, Myers 2005), and biomechanical properties (Chapter 1) can a ll contribute to differential seedling survival across a light gradient by allowing seedlings to avoid or respond to stresses in different ways. A wide range of biotic and abiotic h azards resulting in physical damage can seedling cause mortality. Numerous studi es have addressed the importance of invertebrate herbivory (Coley 1983, Aide & Zimmerman 1990, Benitez-Malvido, Garcia-Guzman, & Kossmann-Ferraz 1999) pathogens (Auspurger 1983, 1984a),

PAGE 54

42 vertebrate consumption and trampling (Osunkoya et al 1992, Weltzin, Archer, & Heitschmidt 1997, Gillman, Wright & Ogde n 2002), and litterfall (Guariguata 1998, Scariot 2000, Gillman et al. 2004) as mortality agents for seedlings. Mortality agents such as disease, invertebrate herbivory, and vertebrate consumption are species-specific, while mortality agents such as litterfall or vertebrate trampling affect seedlings from different species indiscriminatel y. In other words, species vary in their susceptibility to diseases or herbivores, but the chance of being trampled by animals or impacted by falling litter is random. Additionally, all else be ing equal, species differences in mortality due to severe mechanical hazards such as vert ebrate trampling or large falling debris arise mainly from their differential ability to res pond and recuperate from damage, rather than from the ability to evade it. In contrast differential mortality among species due to invertebrate herbivory, diseas e, or minor mechanical hazards (i.e., impact by small litterfall) is also influenced by species differential ability to resist physical disturbances by investing in strong materials and structures (Chapter 1). Species with high mechanical strength in stems and leaves should 1) be affected less frequently by species-specific damage agents such as herbivory and diseas e; 2) Suffer less and less intense mechanical damage when affected by any mechanical damage agent; and, 3) die less frequently when damaged by herbivores, disease, litterfall, and vertebrate trampling (Niklas 1992, Chapter 1). There are also differences in the types of damage inflicted on seedlings by damage agents. Pathogens cause damping-off or tissu e necrosis, while invertebrate herbivores, vertebrate activity, and litterfall result in m echanical damage. Vertebrate activity often results in bent, broken, uprooted or chewed stems, or in missing seedlings. Falling leaves

PAGE 55

43 and branches can bend or break seedlings. Because seedlings face multiple stresses simultaneously or over short periods, it is diffi cult to determine the damage type resulting in the ultimate cause of mortality (Kitajima & Fenner 2000). The main objective of my study was to de scribe first-year temporal patterns of seedling mortality, susceptibility to damage ag ents, and types of damage in eight species of tropical trees differing in their shade tole rance. Seedlings were transplanted to the forest understory and temporal patterns of da mage susceptibility analyzed by constructing survivorship curves. Additionally, artificial seedlings constructed of plastic straws and wire, were “planted” with the real seedlings and monitored to assess susceptibility to damage agents from a community perspective. The artificial seedling approach has been frequently used to estimate potential damage by litterfall and vertebrate trampling (Clark & Clark 1989, McCarthy & Facelli 1990, Mack 1998, Scariot 2000, Roldan & Simonetti 2001, Drake & Pratt 2001, Gillman, Wright & Ogden 2002). Here, I further assess the efficacy of the artificial seedling method for predicting real seedling damage and mortality by comparing artificial and adjacent r eal seedlings that were transplanted to the forest understory. Materials and Methods Study Site and Species This study was conducted in a seasonally moist tropical forest, on Barro Colorado Island (BCI), Panama (9 10’ N, 79 51’ W). Average annual precipitation is 2600 mm, 90% of which falls primarily between Ma y and December (Windsor 1990). Ecological characteristics of BCI are well described in Croat (1978) and Leigh, Windsor & Rand (1982). The experiment took place in young fo rest (100-300 yrs, Foster & Brokaw 1982)

PAGE 56

44 with abundant palms (mainly Astrocarium standleyanum and Oenocarpus panamanus ) and evidence of frequent physical disturba nces such as tree and branch falls. Seeds of eight common tree species, that di ffer in ecological characteristics such as dispersal mode, cotyledon type, and seedling es tablishment probability (Table 1-1) were collected during the dry seas on of 2003 and germinated in a shade house. Seedlings from the eight species differed in time of germination (from approximately 5 days for Aspidosperma cruenta to 36 days for Eugenia nesiotica ), number of leaves, and size (Table A-3). Species with the largest seedlings at germination were Gustavia superba and Anacardium excelsum while the smallest was Tabebuia rosea Species were selected based on seed availability during 2003 on BCI from a range of shade tolerance (K. Kitajima personal communication). Species ranged from shade-tolerant for an intermediate period ( T. rosea ) to shade-tolerant with slow growth in shade ( A. cruenta ) according to Augspurger (1984b). Additionally, according to a long-term study in BCI the study species differ dramatically in their recruitment index, which is a measure of a species ability to recruit seedlings that survive at least 1 yr relative to the number of seeds dispersed to the same microsit e throughout that year (Wright et al 2003, Table 1-1). The eight study species also varied substantially in the biomechanical prop erties of their stems and leaves (Chapter 1), which I expected to lead to differential susceptibilities and responses to mechanical damage. Experimental Design In June 2003, 755 seedlings from the eight study species were tran splanted, at first leaf expansion, to 100 stati ons located randomly along a 9 km network of trails. The stations were at least 5 m away from the trail and separated by a minimum of 50 m. At each station, one seedling of each species was planted within a 1 m2 area, and examined

PAGE 57

45 every 1-2 wks for 8 mos and every 4 wks for four additional months for a total of 22 censuses between June 2003 and May 2004. Becau se of limited seed availability, T. rosea seedlings were transplanted to only 55 of 100 stations. During transplantation, microsites were minimally altered (no litterf all or debris were removed). No evident transplant shock (wilted or dried plants ) was observed, presumably because abundant rainfall during the transplanting period diminished the risk of desiccation. In addition to the eight natu ral seedlings, two artificial seedlings made of plastic and wire were “planted” at each station. The design of artif icial seedlings followed Clark & Clark (1989), but I used two sizes of artificial seedlings to evaluate the effect of physical disturbance and mechanical damage on seedlings of different size classes. Each large artificial seedling was made of two 200 mm-long transpar ent plastic straws oriented in a cross and attached together with staples. A sti ff, 3 mm-diameter x 100 mm-long wire was inserted 20 mm into the vertical stra w and the remaining 80 mm into the ground to simulate a root (Figure 1 in Clark & Cl ark 1989). Small artificial seedlings were constructed in the same way, except that th ey were made from 100 mm-long straws and 50 mm-long wire “root”. Survival, Damage Agents, and Types of Mechanical Damage At each census, mortality, apparent damage agent, and damage types were recorded. For real seedlings f our main types of damage agents were recorded: vertebrate activity, disease, litterfall, and missing wit hout obvious indication of damage agent. “Vertebrate activity” was characterized by plants that were uproote d, flattened, or having damaged, broken, or chewed stems with no evidence of falling leaves or branches. Although it was impossible to differentiate between vertebrate consumption and trampling, chewed stems, missing cotyledons and uprooted seedlings suggest that

PAGE 58

46 vertebrate activity was mos tly consumption-related. Seed lings were diagnosed as “diseased” when they exhibited necrotic tissue, or when they were severely wilted with at least one dry leaf. “Litterfall” was recorded when a seedling had a bent, damaged, or broken stem and there was direct evidence of litterfall or debris above it. “Missing” seedlings were those that could not be located and were presumed dead. This classification scheme underestimates litterf all damage and overestimates vertebrate activity, since all damaged stems without ev idence of litterfall we re considered as damaged by vertebrates. Like wise, missing seedlings within a relatively short census interval of 1-2 wks was likely to be caused by consumption of vert ebrate browsers, also leading to an underestimation of vertebrate activity. A seedling was considered dead when it was completely dried, when the st em was cut in two and the lower portion uprooted, or when the whole seedling was missing. The four damage agents were not mutually exclusive. In fact, seedlings ofte n died after being aff ected by two or more agents. It was not the intent of my study to determine the ultimate cause of mortality, but to describe the temporal patterns of these damage agents that ma y synergistically kill seedlings. I also recorded the first occurrence of th e four main types of mechanical damage that could be fatal: leaf damage, stem be nt, stem broken, and s eedling uprooted. A leaf was considered “damaged” if it was fractured, incomplete or had missing sections larger than 10% of the leaf area. A seedling with at least one damaged leaf was classified with “leaf damage”. A stem was considered “bent” if it was curved or tilted at least 45, and “broken” if it was fractured in two or more sections. A seedling was uprooted when it was completely pulled from the ground. Although the four types of mechanical damage

PAGE 59

47 were not mutually exclusive, only the first type recorded on each plant was used for calculation of “% damage fatali ty” at the end of the study. Furt hermore, to be certain that death (if it occurred) was caused by a particular damage agent, only plants affected by a single damage type were considered. Fo r each type of mechanical damage ( Mx), I calculated the likelihood of dying after receiving a given type of damage P ( D|Mx) expressed as the following formula (Gothelli & Ellison 2004): P ( D|Mx) = P ( D M x ) (2-1) P (Mx) where P ( Mx) is the probability of r eceiving the damage type Mx and P ( D ) is the probability of death. The conditional probability P ( D|Mx) multiplied by 100 and expressed in %, was called “% damage fatality ”. I also evaluated th e relationship between material and structural stem properties and s eedling susceptibility to mechanical damage, by comparing species mean biomechanical traits measured in chapter 1 (Table 1-4) vs. % damage fatality. Artificial Seedlings The artificial seedling method has been us ed in other studies to quantify damage due to random disturbance agents such as litterfall and vertebra te trampling because artificial seedlings are not susceptible to bio tic species-specific agents of mortality such as pathogens and herbivory. Here, artificial seedlings were censused simultaneously with real seedlings to provide a comparison betw een real and artificia l seedling damage and mortality. An artificial seedli ng was considered damaged when it was bent such that at least one of its arms was touching the gr ound, when it was flattened, cut, chewed, or missing (Clark & Clark 1989). For comparison wi th previous studies, three standardized categories of damage were recorded for artifi cial seedlings: vertebrate activity, litterfall,

PAGE 60

48 and unknown. An artificial seedling was classified as damaged by “vertebrate activity” if it was flattened, chewed, cut, or missing with no evidence of litter or branch fall that could have caused the damage. If an artifi cial seedling was bent or flattened, with evidence of litterfall or debris above it, it was considered damaged by “litterfall”. If a seedling was bent but there was no obvious cause, it was classified as “unknown”. Therefore, the estimate of damage caused by litterfall is conser vative, since it only reports artificial seedlings that were dama ged by conspicuous litterfall and debris. I compared the first occurrence of damage agents affecting real vs. artificial seedlings. Because missing artificial seedlings were included in the vertebrate activity category in previous studies (Gillman, Wr ight & Ogden 2002), I included missing real seedlings in the vertebrate activity categor y, such that when comparing artificial seedlings and real seedlings vertebrate activit y refers to seedlings flattened, chewed, cut, or missing. In any case, it is li kely that seedlings th at suddenly disappeared (in an interval of one week) were eaten or uprooted by vertebrates. Additionally, when comparing real and artificial seedlings the “unknown” categor y was obtained from seedlings that were recorded as bent with no further evidence of damage; the “litterfall” category remained the same. Statistical Analyses Temporal patterns of the o ccurrence of seedling death and damage agents were analyzed using non-parametric Kaplan-Meier survival distribution functions (Collett 2003). Survival functions (also called hazard fu nctions) describe the probability that an individual survives longer than a specified period, considerin g individuals at risk at the beginning of each interval and excluding censored values. An interval is defined as the lapse between two mortality events. Censors ar e individuals that “l eft” the study before

PAGE 61

49 its conclusion (e.g., removed from the study) or individuals that were not dead at the time of finalization of the experiment. In my st udy, censored plants were only those that were alive at the end of the obser vation period. From the probabi lity of hazard occurrence during each census interval, su rvivorship curves throughout the entire period were drawn, plotting the proportion of seed lings unaffected by the respective damage agent against time. To observe each species behavior throughout different periods, additional survivorship curves were drawn considering shorter intervals, of 0-2 mo, 2-6 mo, and 6-12 mo. The log-rank test and the Wilcoxon test (Pyke & Thompson 1986, Collett 2003) were used to compare survival distri bution functions for different species. The log-rank test is more sensitive to differences in late survival times, while th e Wilcoxon test is more sensitive to differences in early survival times. However, here both tests had similar outcomes, and therefore only results from log -rank tests are reported. For across-species comparisons between survival proportions at the end of each period, Spearman rank correlation coefficients were used. For inters pecific comparison within types of damage, likelihood chi-squared tests were used. Las tly, nonparametric Spearman correlation tests were used to compare species mean biomechanical and ecological traits vs. % damage fatality. For all analyses = 0.05 and all were performed JMP IN 4.0 (SAS Institute Inc., Cary, NC, USA) Results Seedling Survival Within the first 2 mos 59% of the transp lanted seedlings died, and by the end of 1 yr 76% of transplanted seedlings were dead. The temporal pattern of mortality and overall % mortality at the end of the first year differed among species (Figure 2-1).

PAGE 62

50 Survival distribution functions varied among species for the entire 1 yr period (log-rank 2 = 220.2, d.f. = 7, P < 0.001; Figure 2-1D), and within each of shorter intervals (0-2 mo: log-rank 2 = 220.2, d.f. = 7, P < 0.001; 2-6 mo: log-rank 2 = 55.4, d.f. = 7, P < 0.001; 6-12 mo: log-rank 2 = 49.6, d.f. = 7, P < 0.001; Figures 2-1A to 2-1C). In addition, species rankings of survival probability switched between inte rvals. Survival for 0-1 yr was determined mainly by survivorship during the first 2 mos, which differed from survival in the following intervals. This is demonstrated by the high correlation between the 0-2 mo period and the overall 1yr survival and the lack of co rrelation between the 0-2 mo period with both the 2-6 mo and the 6-12 mo periods (Table 2-1). Damage Agents Vertebrate activity was the most common damage agent (Figure 2-2). Survival functions differed significantly among damage agents calculated for all species combined (log-rank 2 = 1496.9, d.f. = 4, P < 0.001; Figure 2-2A). In addition, damage agents affected species differentially. The percentage of seedlings affected by vertebrate activity after 1 yr ranged from 31% for Beilschmiedia pendula to 65% for E. nesiotica (log-rank 2 = 56.0, d.f. = 7, P <0.001; Figure 2-2B). However, th ere were more missing seedlings of B. pendula than of the other species (log-rank 2 = 209.3, d.f. = 7, P <0.001; Figure 2-2C) and most of these events happened in the first four weeks after transplant. In B. pendula 87% of the seedlings would have been affected by vertebrate activity if all missing seedlings were included in the vertebrate activity category. Tetragastris panamensis, A. excelsum and, T. rosea were the species most affected by disease. Although, T. rosea and A. excelsum were affected by disease mostly in the first 2 mos after transplant, T. panamensis had a constant intensity of infection (log-rank 2 = 98.7,

PAGE 63

51 d.f. = 7, P <0.001; Figure 2-2D). Only 4.1% of the seedlings were affected by litterfall, with no interspecific differences in dama ge agent distribution over time (log-rank 2 = 13.7, d.f. = 7, P >0.05). Types of Mechanical Damage Four types of mechanical damage were re corded during 1 yr: damaged leaves, bent stems, broken stems, and uprooted seedlings caused by vertebrate s, litterfall, or invertebrate herbivores (in th e case of leaf damage). At the end of the study, 77% of seedlings showed some form of damage, of which leaf damage was the most frequent. After 1 yr, 30.6% exhibited leaf damage, 28.7% of the seedlings had broken stems, 23.9% had bent stems, and 25.6% had been upr ooted (Table 2-2). These categories were not mutually exclusive. In fact, 45% of the damaged seedlings had two or more types of damage. Species differed in their lik elihood to die after suffering l eaf damage or bent stems, but not after being uprooted or having thei r stem broken. Percent damage fatality (as defined by Formula 2-1) differed among species for leaf damage or bent stem (leaf damage: 2 = 78.6, d.f. = 7, P <0.001; stem bent: 2 = 57.3, d.f. = 7, P <0.001). In contrast, there was no interspecific difference in damage fatality for uprooted seedlings or those with broken stems (uprooted: 2 = 14.0, d.f. = 7, P = 0.052; stem broken: 2 = 9.4, d.f. = 7, P = 0.226). It should be pointed out that because of the low number of uprooted and broken seedlings in some species (cell N < 5), results should be interpreted with caution. From all the biomechanical and ecologi cal measurements tested (Table 1-4 in Chapter 1), only stem toughness, stem tissue density, and second moment of area were correlated with damage fatality (Table 2-3). Tougher, denser stems died less when their

PAGE 64

52 stem was broken or bent than stems with weak er, less dense material. In contrast, species with larger stems (i.e., with large second mome nt of area) were more likely to die when their stem was broken. Artificial Seedlings After 1 yr, 9.5% of artificial seedling s were damaged by litterfall, 15.5% were damaged by vertebrate activity, and 22.5% we re bent by unknown causes. Thus, damage levels on artificial seedlings were within the range of damage re ported for other sites (Table 2-4). Overall, real seedlings were damaged more than artificial seedlings ( 2 = 64.3, d.f. = 1, P <0.001; Figure 2-3) Artificial seedlings were damaged more by litterfall and by unknown causes, and less by vert ebrate activity than each of the species of real seedlings. Large and small artificia l seedlings did not differ in their damage frequencies ( 2 = 6.0, d.f. = 3, P = 0.111), and therefore th ey were averaged for comparisons with real seedlings and with previous studies. Temporal patterns for each type of damage differed between artificial and real seedlings (Figure 2-4). Vertebrate activity da maged a much higher proportion of real than artificial seedlings, especially during the fi rst 50 days. If missing seedlings were also considered to be affected by vertebrate activity, the difference became even stronger (log-rank 2 = 118.1, d.f. =1, P <0.001; Figure 2-4A). Real seedlings were particularly vulnerable to vertebrate activ ity during the first 2 mos after transplant, but artificial seedlings received a more constant rate of vertebrate damage. Litterfall damage was less than 10% for both artificial seedlings and real seedlings. Artificial seedlings were less affected during the first 6 mo, but the trend reversed during the s ubsequent period (logrank 2 = 6.5, d.f. =1, P = 0.011; Figure 2-4B). In additi on, survival functions describing

PAGE 65

53 the proportion of seedlings damaged by unknown causes, differed between artificial and real seedlings. Although real s eedlings were more affected dur ing the first 3 mo, the trend inverted and at the end of 1 yr, a larger pr oportion of artificial seed lings were affected by unknown causes (log-rank 2 = 9.79, d.f. =1, P = 0.002; Figure 2-4C). Discussion Survival, Damage Agents, and Types of Mechanical Damage The combination of species-specific and indiscriminate damage agents, their temporal patterns, and the differential suscep tibility to damage among species influences seedling performance in the forest understor y. Consistent with other studies (Augspurger 1984a, Kitajima & Augspurger 1989, De Steven 1994), the proportion of seedlings dead after 2 mos was higher than in the 2-6 or 612 mo periods. Seedling mortality was highest during the first 2 mos after transplant, d ecreasing gradually and then becoming more constant over the remaining 8 mos (Figure 21). Although this was the trend for the eight species individually, the species mortality ra nks during the 0-2 mo interval differed from the ranks during the 2-6 mo interval, or the 6-12 mo interval (Table 2-1). This observation suggests that mortality agents that had a greater effect during the first 2 mos became less important in the following 8 mo. Vertebrates were the most common cause of damage overall, especially during the 0-2 mo period (Figure 2-2). Vertebrate activit ies included non-trophic interactions such as trampling, and trophic interactions including leaf herbivory and cotyledon consumption. During the initial 2 mo, low pe rcentages of leaf herbivory (i.e., leaf damage) and high percentages of seedlings upro oted but left partiall y uneaten, with stems cut in half and cotyledons missing, suggest that cotyledon predation by vertebrates was the primary cause of mortalit y. Predation of larg e storage cotyledons has been recorded

PAGE 66

54 as an important cause of mortality in pr evious studies (Sork 1987, Molofsky & Fisher 1993). Here, six of the eight study species ha ve large-seeds (Table 1-1) and abundant cotyledon reserves that can attract vertebrate consumers, even months after germination (Smythe 1978). For example, B. pendula, an animal-dispersed species with large reserve cotyledons, suffered extremely high mortality during the first month after transplanting. Within the first week after transplanting 42% of the seed lings were missing, and by the end of the first month, 56% of the seedlings were missing and presumed dead. Cotyledon predation was potentially enha nced by soil disturbance during transplant, which could have attracted agoutis ( Dasyprocta punctata ). However, it has been shown that agoutis find buried seeds using predominantly olf actory cues (Smythe 1978). Thus, further investigations comparing cotyledon consumption in naturally germinated vs. transplanted seedlings are required to reach definitive conclusions. Contrary to vertebrate damage, which aff ected almost 40% of seedlings during the first 2 mo, disease affected less than 10% of seedlings throughout the study. This observation differs from disease prevalence reported by Augspurger (1984a), who studied naturally germinated seedlings of nine wi nd-dispersed species on BCI. She found that the largest fraction of early seed ling mortality under shaded conditions was due to pathogens. However, Augspurger (1984a) stud ied naturally germinated se edlings that were already established, diminishing the probability of r ecording early cotyledon predation, similar to that reported for B. Pendula here. Nevertheless, the two species ( A. cruenta and T. rosea ) overlapping between studies exhibited similar levels of mortality at the end of 2 mos. Mortality recorded for A. cruenta in Augspurger (1984a) and in my study was 20% and 27%, respectively. For T. rosea mortality was 65% in both studies.

PAGE 67

55 Differences in survivorship between sp ecies reflected the interaction between likelihood of damage and the ability to tolera te damage-induced stress. Contrary to my expectation shade-tolera nt species with stronger material properties (i.e., tougher stems and higher stem tissue density), did not di ffer from species with weaker material properties in the probability of suffering mechanical damage, but once damaged they were less likely to die than species with w eaker stems (Table 2-2). Contrary to this pattern, A.cruenta a shade-tolerant species with weak stems (Chapter 1), was the least likely to die after suffering m echanical damage, suggesting th at other factors, such as carbohydrate storage reserves in stems and roots play an important role in the ability of seedlings to tolerate mechanical damage (Myers 2005). Additionally, the probability of survival after being damaged (measured as damage fatality) was different depending on the type of damage received. For example, seedlings that were uprooted or had broken stems usually died, while seedlings that suffe red leaf damage or bent stems were more likely to survive. My results (and previous st udies, Marquis & Braker 1994) suggest that leaf damage (caused by leaf herbivory) cons titutes a less severe st ress than stem bending and breakage. Numerous studies have empha sized the benefits of havi ng a large seed, resulting in a large seedling and increased survival (Paz & MartinezRamos 2003, Green & Juniper 2004, Moles & Westoby 2004b). In contrast, in my study there was no correlation between seed size and seedling survival, suscepti bility to damage agents, or incidence of different types of damage (Table 1-4). One possible reason is that on BCI large seeded species face strong pressures from vertebra te cotyledon predators. Barro Colorado Island, due to the absence of top predators, supports high densities of medium sized mammals

PAGE 68

56 (Glanz 1982) that can have a large effect on tree regenerati on patterns through high predation rates on large seeded species before and after germination (Asquith, Wright & Clauss 1997). Additionally, the likelihood of d eath after stem breakage increased with stem size (Table 2-3), suggesting that large size is not always beneficial. Artificial Seedlings The comparison between artificial vs. real seedlings revealed that on BCI damage agents affecting artificial seedlings were not good predictors of damage agents affecting the real seedling community (F igure 2-3). Moreover, artificia l seedling damage was not a good predictor of mortality for transplanted s eedlings because real seedlings were more affected by vertebrate consumption than artif icial seedlings, especially during the first 2 mos. However, artificial seedlings were more affected by litterfall, in terms of cumulative damage, after 1 yr. Consistent with my re sults, Gillman, Wright & Ogden (2002) found that artificial seedling damage was not a good predictor of mort ality of naturally germinated seedlings in five evergreen temperate forests in New Zealand. Because artificial seedlings are not significantly cons umed by vertebrates, they would be accurate predictors of mechanical damage, only in environments where indiscriminate, nontrophic damage agents (e.g., litterfall, verteb rate trampling) are more frequent than trophic interactions. Temporal patterns of mechanical damage also differed between real and artificial seedlings. Real seedlings were severely aff ected by vertebrate activ ity during the first 2 mos with a rapid decline afterwards, while artificial seedlings experienced a relatively constant rate of damage (Figure 2-4). This difference suggests that damage agents that affect seedlings indiscriminately (e.g., lit terfall) became import ant after the early establishment period when seedlings suffe r heavily from vertebrate consumers.

PAGE 69

57 The total percentage of artificial seedli ngs damaged on BCI after 1 yr, was within the range reported in other studies (Table 2-4). However, the percentage (9.5 + 2.1% yr-1) of artificial seedlings damaged by litterfall on BCI was lower than the percentage reported in most other tropical forests, po ssibly due to differences in rainfall, canopy composition, and topography (Van Der Meer & Bongers 1996,Gillman et al 2004). Seedlings in the old secondary forest on BCI suffered an intermediate frequency of mechanical damage. La Selva, Costa Rica (C lark & Clark 1989) e xhibited the highest percentage of damaged ar tificial seedli ngs (82.4% yr-1), and the intensive hunted forest in Beni, Bolivia (Roldn & Simonetti 2000) the lowest (25% yr-1). However, in the Bolivian study, litterfall damage was not considered. Th e large variation between studies suggests that either the probability of being affected by mechanical damage differs widely across forest communities, or researcher s used different criteria to classify agents of mechanical damage. Interestingly, the most frequent agen t of mechanical damage reported in each study, was usually the focal agent of interest for each author. Conclusions Survivorship analyses revealed that diverse ecological pressures, such as vertebrate predation and disease affect s eedlings differentially through time and among species. For example, seedlings from T. panamensis suffered little se edling mortality due to vertebrate predation during the initial 2 mo, but became more susceptible to disease in the following period. In c ontrast, seedlings from E. nesiotica and B. pendula were severely damaged by vertebrate predation in the initial 2 mo, but at tained high survival rates if they escaped predation. Furthermore, species such as A. cruenta that suffered little mechanical damage overall can be limited by other factors, such as seed production and dispersal (Augspurger 1984a). Additionally, di fferences among damage agents affecting

PAGE 70

58 real and artificial seedlings indicate that l itterfall can become an important cause of damage after the early establishment peri od. Although there was no relationship between shade tolerance and the type s of mechanical damage affecting each species, shade-tolerant species with stronger stems survived more often after damage, suggesting that investment in strong st ems is beneficial for seedling performance. The combined effects of species-specific and indiscriminate damage agents, and species differences in the responses to damage determine seedling performance. The resulting differences in survival allow different species to su cceed under different ecological conditions, ultimately contributing to plant diversity in tropical forests.

PAGE 71

59 Table 2-1. Relationships among species ranki ngs of survival probability during the specified interval for seedlings of eight tree species. 0-2 mo 2-6 mo 6-12 mo 2-6 mo 0.50 (0.207) 6-12 mo 0.55 (0.160) 0.90 (0.002) 0-12 mo 0.90 (0.002) 0.79 (0.021) 0.83 (0.010) Order of species codes listed in Fig 2-1 A-C. Shown are Spearman correlation coefficients with their corresponding P values in parentheses a nd significant correlations in bold. Table 2-2. Percent damage fatality of four types of mechanical damage on eight tree species during 1 yr in the forest understory. Leaf damage Stem bent Stem broken Uprooted Sp. code N % Fatality N % Fatality N % Fatality N % Fatality TABR 20 100.0 20 100.0 4 100.0 22 95.5 ANAE 37 97.3 39 100.0 36 100.0 36 100.0 CASE 45 64.4 17 76.5 31 100.0 16 100.0 BEIP 7 100.0 9 88.9 22 100.0 16 100.0 GUSS 47 44.7 25 48.0 36 94.4 35 85.7 TETP 35 48.6 19 57.9 19 94.7 9 100.0 EUGN 9 22.2 28 57.1 44 93.2 41 97.6 ASPC 31 32.3 23 47.8 25 92.0 18 88.9 Total 231 61.5 180 72.2 217 96.3 193 95.3 N = total number of seedlings affected by each damage type. Damage types are not mutually exclusive. For speci es codes refer to Table 1-1.

PAGE 72

60 Table 2-3. Relationships among stem biomechan ical traits and % damage fatality for seedlings of eight tree species. Types of mechanical damage Biomechanical properties Stem broken Stem bent Modulus of elasticity -0.36 (0.388) 0.61 (0.108) Stem toughness -0.77 (0.044) 0.82 (0.024) Stem density -0.79 (0.019) -0.71 (0.048) % NDF -0.406 (0.318) -0.69 (0.056) Second moment of area of stem 0.79 (0.019) 0.61 (0.108) Flexural stiffness 0.61 (0.109) 0.54 (0.169) Shown are Spearman correlation coefficients for tests performed on species means ( N = 8) with P values in parentheses and signi ficant correlati ons in bold; % NDF = percent non-detergent fiber.

PAGE 73

61 Table 2-4. Percentage of artifi cial seedlings affected by speci fied damage agents in this and other published studies in different forest communities. Means (yr-1) + 1 SD when applicable. Forest type Site Study length N Litterfall Vertebrate activity Unknown Undamaged Study Tropical wet forest La Selva, Costa Rica 1 yr 500 19.2 21.0 42.2 17.6 Clark & Clark (1989) Tropical wet forest Crater Mountain, Papua New Guinea 1 yr 418 13.8 7.0 11.0 65.3 Mack (1998)a Seasonal tropical forest BDFFP, Manaus, Brazil 1 yr 100 21.7 9.7 8.6 60 Scariot (2000) b Seasonal tropical forest BCI, Panama 1 yr 100 9.5 + 2.1 15.5 + 4.9 22.5 + 4.9 52.5 + 6.4 This study c OHF Beni, Bolivia 500 — 80 5 15 Tropical terra firme forest IHF Beni, Bolivia 6 mo 500 — 13 12 75 Roldn & Simonetti (2000) d (+ Pigs) Mauna Loa, Hawaii 150 15.3 4.7 11.3 68.7 Montane tropical forest (Pigs) Mauna Loa, Hawaii 1 yr 150 20 0 0 80 Drake & Pratt (2001) e Evergreen temperate forest North Island, New Zealand 2 yr 200 6.1 + 5.8 2.8 + 2.2 — — Gillman, Wright & Ogden (2002) f Temperate forest New Jersey, USA 10 mo 200 2 39 13 44 McCarthy & Facelli (1990) g

PAGE 74

62 a Total shown adds to 99.1% seedlings, and th e remaining 02.9% artificial seedlings were damaged by water erosion. b data for continuous forest aver aged for the three study sites. BDFFP = Biological Dynamics of Forest Fragments Project. c Average from large and small artificial seedlings damaged (+ 1 SD). BCI = Barro Colorado Island. See text for methodological details. d OHF = Occasionally Hunted Fore st, IHF = Intensively Hunted Forest; approximate damages were taken fr om Figure 2, since exac t numbers were not shown in the study. Authors did not record % litte rfall damage. e Half of the artificial seedlings were fenced to exclude pigs (pigs). f Shown are averages (+ 1 SD) from five sites included in the study. Only non-tr ophic vertebrate activity was recorded. Undamaged or unknown fractions were not reported. g Only forest habitat data are shown. Total adds to 98% seedlings, the remaining 2% AS were damaged by “frost heaving”.

PAGE 75

63 0 0.2 0.4 0.6 0.8 1 010203040506070Proportion survivin g ASPC TETP GUSS CASE TABR EUGN A NAE BEIPA 6080100120140160180EUGN A SPC GUSS TETP BEIP CASE A NAE TABRB 0 0.2 0.4 0.6 0.8 1 180210240270300330360 Time (days)Proportion survivin g EUGN A SPC TETP GUSS CASE BEIP TABR A NAEC 050100150200250300350 Time (days) A SPC TETP GUSS EUGN CASE TABR BEIP A NAED Figure 2-1. Kaplan-Meier survivorship curves for seedlings of eight tree species transplanted to the forest understory. Survivo rship is relative to the number of seedlings alive at the beginning of each period; A) 0-2 mo, B) 2-6 mo, C) 6-12 mo, and D) total study period (0-12 mo). For speci es codes refer to Table 1-1.

PAGE 76

64 0 0.2 0.4 0.6 0.8 1Proportion unaffected Litterfall Disease Missing Vertebrate Death by all causesA. Pooled BEIP TETP A SPC CASE TABR A NAE GUSS EUGNB. Vertebrate0 0.2 0.4 0.6 0.8 1 050100150200250300350400 Time (days)Proportion unaffecte d Others TETP CASE BEIPC. Missin g 050100150200250300350400 Time (days)D. DiseaseOthers TETP A NAE TABR Figure 2-2. Kaplan-Meier survivorship curves (proportion of seedlings yet to be hit by specified damage agents plotted against time) for seedlings of eight tree sp ecies transplanted to the forest unde rstory. A) Comparison of agents of mechanical damage for the pooled data of the eight study sp ecies. B) Vertebrate activi ty on each species (uprooted, flattened, chewed or broken stems; see methods). C) Seedlings missing most likely due to consumption by vertebrates and presumed dead. D) Diseas e. For species codes refer to Table 1-1.

PAGE 77

65 0% 20% 40% 60% 80% 100%TABR A N A E C A SE B E I P GUSS TE TP EUGN A SPC Small L arge All rea l B o th A SReal seedlingsAS Intact Unknown Litterfall Vertebrate activity Figure 2-3. Percent of real and artificial seedlings (AS) damage d during 1 yr in the forest understory by specific damage agent. Sm all and large artificial seedlings were pooled as “both AS” ( N = 200), and real seedlings were pooled as “all real” ( N = 755). Categories are mutually exclus ive, as each seedling was assigned only to the first damage agent it expe rienced. For species codes refer to Table 1-1.

PAGE 78

66 0 0.2 0.4 0.6 0.8 1Proportion unaffectedA. Vertebrate AS Real Real including those missing 0.7 0.8 0.9 1Proportion unaffectedReal A SB. Litterfall 0.7 0.8 0.9 1 050100150200250300350400 Time (days)Proportion unaffected Real ASC. Unknown Figure 2-4. Kaplan-Meier survivorship curv es for mechanical damage experienced by artificial (AS) and real seedlings dur ing 1 yr in the forest understory. A) Vertebrate activity (estimated excluding missing seedlings and assuming those missing seedlings were all due to vertebrate consumption). B) Litterfall damage. C) Bent by unknown causes. For real seedlings, the average from eight species is shown. Notice th e different scales in y-axes.

PAGE 79

APPENDIX SPECIES MEANS AND STANDARD DE VIATIONS FOR BIOMECHANICAL MEASUREMENTS, FIBER ANALYSIS, AND BIOMASS

PAGE 80

68 Table A-1. Biomechanical measurements of se edling stems from eight tree species at T1 and T2 (1 and 6 mos after first leaf expansion, respectively). Shown are means (+ 1 SD) and N = number of individuals. For species codes refer to Table 1-1. T1 T2 Species N Mean SD N Mean SD Modulus of elasticity (MN m2) ANAE 14 255.06 90.66 15 451.10 108.15 ASPC 14 356.16 126.64 14 768.01 289.30 BEIP 15 507.93 102.44 15 944.71 488.98 CASE 14 336.08 110.51 14 779.27 255.60 EUGN 15 3590.92 768. 37 9 5265.72 1215.35 GUSS 15 916.04 178.94 7 1244.89 354.93 TABR 15 171.72 107.84 14 331.35 193.40 TETP 15 1674.51 402.98 10 2269.71 538.67 Fracture toughness (J m-2) ANAE 14 1551.82 414. 41 2 1883.80 560.45 ASPC 15 2476.14 709.15 9 7471.83 2466.70 BEIP 15 3430.16 636.75 15 6015.59 1760.58 CASE 15 1325.15 353.23 0 2657.60 842.52 EUGN 15 5433.04 1064.57 9 16004.96 2914.35 GUSS 5 5463.60 953.86 — — — TABR 13 1135.16 672.99 16 3831.43 1774.03 TETP 15 5692.55 869.12 10 5778.97 2245.26 Density (g cm-3) ANAE 45 0.15 0.03 24 0.19 0.04 ASPC 45 0.52 0.14 27 0.39 0.07 BEIP 44 0.35 0.09 25 0.40 0.10 CASE 43 0.14 0.04 26 0.25 0.05 EUGN 44 0.47 0.13 20 0.58 0.14 GUSS 46 0.29 0.04 21 0.38 0.07 TABR 25 0.13 0.03 19 0.24 0.10 TETP 45 0.44 0.06 24 0.43 0.07 Flexural stiffness (N cm2) ANAE 13 23.85 12.96 15 52.11 29.23 ASPC 14 3.38 1.30 14 12.19 9.25 BEIP 15 7.76 3.25 14 16.38 7.73 CASE 14 4.22 1.69 13 13.79 6.73 EUGN 15 1.43 0.96 9 7.69 2.38 GUSS 15 36.15 11.75 7 84.46 41.12 TABR 15 0.24 0.23 13 2.80 2.17 TETP 15 3.96 1.21 10 10.16 4.85

PAGE 81

69 Table A-1. Continued T1 T2 Species N Mean SD N Mean SD ANAE 10 0.034 0.015 10 0.014 0.004 Whole stem flexibility (Radians/N) ASPC 15 0.062 0.017 12 0.018 0.008 BEIP 14 0.024 0.008 10 0.016 0.005 CASE 15 0.061 0.041 11 0.03 0.021 EUGN 10 0.15 0.035 12 0.092 0.052 GUSS 15 0.009 0.003 11 0.007 0.002 TABR — — — 4 0.087 0.076 TETP 15 0.069 0.036 8 0.055 0.023 Work to bend (J) ANAE 13 0.0034 0.0016 — — — ASPC 10 0.0008 0.0005 — — — BEIP 9 0.0019 0.0008 — — — CASE 7 0.0010 0.0006 — — — EUGN 9 0.0007 0.0004 — — — GUSS 11 0.0041 0.0012 — — — TABR — — — — — — TETP 9 0.0016 0.0006 — — — % Critical height ANAE 14 28.36 6.03 14 23.35 4.53 ASPC 15 22.31 2.71 10 16.93 2.43 BEIP 13 21.05 3.35 14 16.62 3.80 CASE 15 20.69 4.35 14 19.78 4.82 EUGN 15 14.49 2.60 14 14.99 1.97 GUSS 14 13.68 1.90 9 14.31 3.21 TABR 14 18.46 2.84 15 16.66 6.18 TETP 15 17.87 2.21 7 14.44 1.95

PAGE 82

70 Table A-2. Biomechanical measurements of seed ling leaves from eight tree species at T1 and T2 (1 and 6 mos after first leaf expansion, respectively). Shown are means (+ 1 SD) and N = number of individuals. For species codes refer to Table 1-1. T1 T2 Species N Mean SD N Mean SD Lamina fracture toughness (J m-2) ANAE 15 274.12 75.12 13 225.73 71.25 ASPC 14 371.28 92.76 14 324.28 59.60 BEIP 15 130.34 60.45 15 229.19 51.23 CASE 15 71.29 39.63 6 107.40 35.21 EUGN 15 191.55 59.37 8 245.71 32.90 GUSS 16 395.43 150.29 9 218.39 58.59 TABR 4 117.78 90.17 18 189.49 46.37 TETP 14 339.27 135.27 10 198.51 71.58 Midvein fracture toughness (J m-2) ANAE 15 2198.47 419. 80 12 2532.61 360.55 ASPC 15 2370.67 455.09 14 2545.19 403.04 BEIP 15 1405.83 494.77 15 2271.53 477.84 CASE 13 984.15 183.35 6 1001.03 185.94 EUGN 15 1609.67 461. 49 8 2417.33 501.64 GUSS 16 3124.26 426.99 9 2906.78 582.72 TABR 4 1521.65 387.46 17 2194.12 427.60 TETP 14 2217.99 348.91 10 3475.15 666.77 Force of fracture (N) ANAE 15 43.27 15.85 13 40.58 14.92 ASPC 14 74.52 19.97 14 63.89 13.64 BEIP 15 38.45 18.34 15 32.23 6.84 CASE 15 12.41 8.00 6 11.84 4.73 EUGN 15 20.71 8.36 8 20.04 3.72 GUSS 16 56.81 20.53 9 30.54 8.93 TABR 4 15.93 12.08 18 28.67 10.54 TETP 14 41.53 12.72 10 25.49 8.80 Density (g cm-3) ANAE 15 0.25 0.04 12 0.30 0.04 ASPC 14 0.29 0.02 14 0.34 0.02 BEIP 12 0.20 0.03 15 0.48 0.07 CASE 15 0.20 0.03 6 0.28 0.04 EUGN 15 0.33 0.04 8 0.45 0.04 GUSS 15 0.30 0.02 9 0.35 0.03 TABR 4 0.21 0.06 18 0.23 0.05 TETP 13 0.36 0.06 10 0.42 0.04

PAGE 83

71 Table A-2. Continued T1 T2 Species N Mean SD N Mean SD Specific leaf area (cm2 g-1) ANAE 15 317.02 36.46 13 257.89 28.66 ASPC 15 193.95 14.78 14 172.03 8.88 BEIP 13 248.38 47.50 15 196.52 29.83 CASE 15 486.00 83.97 6 427.39 102.53 EUGN 15 360.38 24.02 8 305.14 36.79 GUSS 16 281.02 19.03 9 248.32 28.84 TABR 4 478.64 130.97 18 501.46 90.48 TETP 13 287.33 26.60 10 231.80 27.27

PAGE 84

72 Table A-3. Fiber fractions of s eedling stems from eight tree species at T1 and T2 (1 and 6 mos after first leaf expansion, respectively). Sh own are means (+ 1 SD) and N = number of samples (each sample composed of 5-15 seedlings); percent NDF = Non detergent fiber. For sp ecies codes refer to Table 1-1. T1 T2 % NDF Species N Mean SD N Mean SD ANAE 3 48.60 2.49 3 48.43 1.31 ASPC 3 51.33 1.60 3 55.28 3.20 BEIP 3 66.83 1.40 3 63.57 1.68 CASE 3 52.95 3.20 3 57.13 3.03 EUGN 3 76.36 1.04 3 74.41 3.46 GUSS 3 62.29 0.88 3 62.43 0.42 TABR 1 48.38 — 2 53.01 8.57 TETP 3 67.90 1.24 3 64.46 2.19 % Hemicellulose ANAE 3 8.54 1.00 3 10.28 0.66 ASPC 3 13.95 0.82 3 15.93 0.54 BEIP 3 11.98 0.43 3 14.71 1.54 CASE 3 13.71 0.82 3 18.82 1.57 EUGN 3 14.34 0.09 3 14.39 2.65 GUSS 3 14.89 0.60 3 13.47 0.60 TABR 1 10.12 — 2 11.58 2.13 TETP 3 13.09 0.22 3 13.20 0.47 % Cellulose ANAE 3 27.94 1.13 3 26.18 1.01 ASPC 3 26.52 0.70 3 27.80 2.78 BEIP 3 29.22 1.76 3 28.54 0.98 CASE 3 27.01 0.54 3 26.66 0.40 EUGN 3 37.32 0.32 3 37.59 5.54 GUSS 3 34.19 1.23 3 34.11 0.54 TABR 1 25.33 — 2 30.58 5.17 TETP 3 40.22 1.90 3 37.21 2.51 % Lignin ANAE 3 12.12 0.44 3 11.97 0.36 ASPC 3 10.85 0.60 3 11.55 1.13 BEIP 3 25.63 1.59 3 20.33 2.31 CASE 3 12.23 3.04 3 11.65 2.24 EUGN 3 24.71 0.73 3 22.43 1.03 GUSS 3 13.21 0.94 3 14.85 1.08 TABR 1 12.93 — 2 10.85 1.27 TETP 3 14.59 3.06 3 14.06 0.44

PAGE 85

73 Table A-4. Fiber fractions of s eedling leaves from eight tree species at T1 and T2 (1 and 6 mos after first leaf expansion, respectively). Sh own are means (+ 1 SD) and N = number of samples (each sample composed of 5-15 seedlings); percent NDF = Non detergent fiber. For sp ecies codes refer to Table 1-1. T1 T2 % NDF Species N Mean SD N Mean SD ANAE 3 42.08 1.16 3 43.11 1.58 ASPC 3 52.32 1.87 3 54.23 4.12 BEIP 3 54.87 1.95 3 55.87 0.31 CASE 3 51.84 0.81 3 44.83 10.05 EUGN 3 56.93 4.59 3 50.28 6.60 GUSS 3 54.77 0.89 3 52.67 0.67 TABR 1 43.83 — 2 50.82 0.16 TETP 3 54.73 1.85 3 45.86 6.90 % Hemicellulose ANAE 3 7.77 1.15 3 8.68 2.35 ASPC 3 14.21 0.58 3 14.29 2.10 BEIP 3 9.68 0.55 3 10.34 1.51 CASE 3 20.43 0.47 3 18.27 2.54 EUGN 3 19.05 2.12 3 16.99 2.31 GUSS 3 16.37 0.95 3 17.58 1.22 TABR 1 10.78 — 2 12.24 0.50 TETP 3 12.63 0.18 3 11.84 1.58 % Cellulose ANAE 3 18.43 1.16 3 19.85 0.88 ASPC 3 20.62 1.03 3 22.11 2.45 BEIP 3 16.96 0.48 3 17.98 0.97 CASE 3 16.73 1.34 3 16.51 5.29 EUGN 3 17.45 0.30 3 16.18 0.91 GUSS 3 22.65 0.30 3 20.48 1.01 TABR 1 14.51 — 2 18.98 0.31 TETP 3 24.18 1.14 3 20.23 3.38 % Lignin ANAE 3 15.87 0.82 3 14.58 1.20 ASPC 3 17.48 0.56 3 17.83 1.02 BEIP 3 28.23 1.21 3 27.55 1.50 CASE 3 14.67 1.76 3 10.05 2.84 EUGN 3 20.43 2.24 3 17.11 3.69 GUSS 3 15.75 0.20 3 14.61 0.86 TABR 1 18.54 — 2 19.60 0.35 TETP 3 17.91 1.38 3 13.80 2.17

PAGE 86

74 Table A-5. Biomass measurements of seedling st ems and leaves from eight tree species at T1 and T2 (1 and 6 mos after first le af expansion, respectively). Shown are means (+ 1 SD) and N = number of individuals. For species codes refer to Table 1-1. T1 T2 Stem height (mm) Species N Mean SD N Mean SD ANAE 45 238.2 35.4 24 253.4 40.7 ASPC 45 140.7 19.6 27 154.8 26.9 BEIP 44 165.3 36.9 25 162.8 35.0 CASE 43 140.1 34.5 26 173.3 32.7 EUGN 44 140.3 27.7 20 182.5 25.0 GUSS 46 176.7 29.8 21 196.3 50.2 TABR 25 64.1 12.6 19 115.9 39.4 TETP 45 154.5 19.0 24 162.7 21.8 Leaf area (cm2) ANAE 45 112.78 41.03 25 125.42 57.70 ASPC 45 35.40 8.26 28 55.98 20.29 BEIP 45 67.70 34.61 25 67.94 25.32 CASE 44 66.24 42.23 26 122.44 80.94 EUGN 30 13.89 5.97 21 34.32 9.23 GUSS 45 138.05 41.97 21 171.96 55.02 TABR 23 6.54 6.52 18 52.09 16.84 TETP 45 28.22 6.52 24 41.82 16.84 Leaf thickness (mm) ANAE 15 0.16 0.03 13 0.18 0.05 ASPC 14 0.20 0.01 14 0.20 0.01 BEIP 15 0.29 0.04 15 0.14 0.01 CASE 15 0.17 0.04 6 0.11 0.02 EUGN 15 0.11 0.02 8 0.08 0.01 GUSS 16 0.15 0.02 9 0.14 0.01 TABR 4 0.14 0.03 18 0.15 0.05 TETP 14 0.13 0.03 9 0.13 0.02 Plant biomass (g) ANAE 45 0.8066 0.2689 25 1.1550 0.4724 ASPC 45 0.5446 0.1265 28 0.7385 0.2422 BEIP 43 1.8675 0.5233 25 0.9564 0.4232 CASE 44 0.3022 0.2076 26 0.6178 0.3901 EUGN 30 0.3990 0.1743 21 0.3508 0.2056 GUSS 46 3.6938 1.3287 21 2.5911 0.8554 TABR 23 0.0459 0.0240 19 0.2540 0.2180 TETP 44 0.2467 0.0543 24 0.3735 0.1564

PAGE 87

75 LIST OF REFERENCES Aide, T.M. (1987) Limbfalls—a major cause of sapling mortality for tropical forest plants. Biotropica 19 284-285. Aide, T.M. & Zimmerman J.K. (1990) Patt erns of insect herbivory, growth, and survivorship in juvenile s of a neotropical liana. Ecology 70 ,1412-1421. Asquith, N.M., Wright, S.J. & Clauss, M.J. (1997) Does mammal community composition control recruitment in neotro pical forests? Evidence from Panama. Ecology 78 941-946. Augspurger, C.K. (1983) Seed di spersal of the tropical tree, Platypodium elegans and the escape of its seedlings from fungal pathogens. Journal of Ecology 71 ,759-771. 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 requiremen ts of neotropical tree seedlings: a comparative study of growth and survival. Journal of Ecology 72 777-795. Barnett, J.R. & Jeronimidis, G., eds. (2003) Wood Quality and its Biological Basis Blackwell Publishing, Oxford. Bazzaz, S.A. & Grace, J., eds. (1997) Plant Resource Allocat ion. Academic Press, San Diego, California. Benitez-Malvido, J., Garcia-Guzman, G. & Kossmann-Ferraz, I.D. (1999) Leaf-fungal incidence and herbivory on tr ee seedlings in tropical rainforest fragments: an experimental study. Biological Conservation 91 ,143-150. Brchert, F., Becker, G. & Speck, T. ( 2000) The mechanics of Norway spruce ( Picea abies (L.) Karst) mechanical properties of standing trees from different thinning regimes. Forest Ecology and Management 135 45-62. Campbell, K.A. & Hawkins, C.D.B. (2004) Eff ect of seed source and nursery culture on paper birch ( Betula papyrifera ) uprooting resistance and field performance. Forest Ecology and Managemen t 196 425-433. Canham, C.D., Kobe, R.K., Latty, E.F. & Chazdon, R.L. (1999) Interspecific and intraspecific variation in tree seedling surv ival: effects of alloca tion to roots versus carbohydrate reserves. Oecologia 121 ,1-11.

PAGE 88

76 Chapman, C.A. & Chapman L.J. (1996) Fr ugivory and the fate of dispersed and nondispersed seeds of six African tree species. Journal of Tropical Ecology 12 491504. Choong, M.F. (1996) What makes a leaf tough and how this affects the pattern of Castanopsis fissa leaf consumption by caterpillars. Functional Ecology 10 668674. Choong, M.F., Lucas, P.W., Ong, J.S.Y., Pereira, B., Tan, H.T.W. & Turner, I.M. (1992) Leaf fracture-toughness and sclerophy lly-their correlations and ecological implications New Phytologist 121 597-610. Clark, D.B. & Clark, D.A. (1985) Seedling dynamics of a tropical tree: impacts of herbivory and meristem damage. Ecology 66 1884-1892. Clark, D.B. & Clark, D.A. (1989) The role of physical damage in the seedling mortality regime of a neotropical rain forest. Oikos 55 225-230. Clark, D.B. & Clark, D.A. (1991) The im pact of physical damage on canopy tree regeneration in trop ical rain-forest. Journal of Ecology 79 447-457. Coley, P.D. (1983) Herbivory and defensive ch aracteristics of tree sp ecies in a lowland tropical forest Ecological Monographs 53 209-233. Collet, D. (2003) Modeling Survival Data in Medical Research 2nd edn. Chapman & Hall/CRC, Washington D.C. Condit, R., Hubbell, S.P., & Foster, R.B. ( 1995) Mortality rates of 205 neotropical tree and shrub species and the im pact of a severe drought. Ecological Monographs 65 419-439. Cooley, A.M., Reich, A. & Rundel P. (2004). Leaf support biomechanics of neotropical understory herbs. American Journal of Botany 91 573-581. Cordero, R.A. (1999) Ecophysiology of Cecropia schreberiana saplings in two wind regimes in an elfin cloud forest: grow th, gas exchange, ar chitecture and stem biomechanics Tree Physiology 19 153-163. Croat, T.B. (1978) Flora of Barro Colorado Island Stanford University Press, Stanford. Darvell, B.W., Lee, P.K.D., Yuen, T.D.B. & Lucas. P.W. (1996) A portable fracture toughness tester for biological materials. Measurement Science & Technology 7 954-962. De Steven, D. (1994) Tropical tree seedli ng dynamics: recruitment patterns and their population consequences for three canopy species in Panama. Journal of Tropical Ecology 10 369-383.

PAGE 89

77 Dominy, N.J., Lucas, P.W. & Wri ght, S.J. (2003) Mechanics and chemistry of rain forest leaves: canopy and understorey compared. Journal of Experimental Botany 54 2007-2014. Drake, D.R. & Pratt, L.W. (2001) Seedling mort ality in Hawaiian rain forest: The role of small-scale physical disturbance. Biotropica 33 319-323. Esau, K. (1977) Anatomy of Seed Plants 2nd edn. John Wiley & Sons, New York. Foster, R.B. & Brokaw, N.V.L. (1982) Struct ure and history of th e vegetation of Barro Colorado Island. The Ecology of a Tropical Forest: Seasonal Rhythms and LongTerm Changes. (eds E.G. Leigh, Jr., A.S. Rand & D.M. Windsor), pp. 133-150. Smithsonian Institution Pr ess, Washington, D. C. Gartner, B.L. (1991) Structural stability a nd architecture of vines vs. shrubs of poison oak, Toxicodendron diversilobum Ecology 72 2005-2015. Gillman, L.N., Ogden, J., Wright, S.D., St ewart, K.L. & Walsh, D.P. (2004) The influence of macro-litterfall and forest structure on litterfall damage to seedlings. Austral Ecology 29 305-312. Gillman, L.N., Wright, S.D. & Ogden, J. (2002) Use of artificial seedlings to estimate damage of forest seedlings due to litterfall and animals. Journal of Vegetation Science 13, 635-640. Givnish, T.J. (1995). Plant stems: biomech anical adaptation for energy capture and influence on species distributions. Plant Stems Physiology and Functional Morphology (ed. B.L. Gartner), pp. 3-49. Academic Press, San Diego, California. Glanz, W.E. (1982) The terr restrial mammal fauna of Barro Colorado Island: censuses and long-term changes. The Ecology of a Tropical Forest: Seasonal Rhythms and Long-Term Changes (eds E.G. Leigh, A.S. Rand & D.M. Windsor), pp 455-468, Smithsonian Institution Press, Washington D.C. Gmez, J.M., Garca, D. & Zamora, R. (2003) Impact of vertebrate acornand seedlingpredators on a Mediterranean Quercus pyrenaica forest. Forest Ecology and Management 180 125-134. Gotelli, N.J. & Ellison, A.M. (2004) A Primer of Ecol ogical Statistics Sinauer Associates Inc., Massachusetts. Green, P.T. & Juniper, P.A. (2004) Seed mass, seedling herbivory and the reserve effect in tropical rainforest seedlings. Functional Ecology 18 539-547. Greenhill, G. (1881) Determination of the greates t height consistent with stability that a vertical pole or mast can be made, and th e greatest height to which a tree of given proportions can grow. Procedures of the Cambridge Philosophical Society 4 6573.

PAGE 90

78 Grubb, P.J. (1986) Sclerophylls, pachyphy lls, and pycnophylls: the nature and significance of hard leaf surfaces. Insect and Plant Surfaces (eds. B. Juniper & T.R.E. Southwood), pp.137-150. Edward Arnold, London. Guariguata, M.R. (1998) Response of forest tree saplings to experimental mechanical damage in lowland Panama. Forest Ecology and Management 102 103-111. Harms, K.E. & Dalling, J.W. (1997) Da mage and herbivory tolerance through resprouting as an advantage of large s eed size in tropical trees and lianas. Journal of Tropical Ecology 13 617-621. Harper, J.K., Dalley, N.K., Owen, N.L., W ood, S.G. & Cates, R.G. (1993) X-ray structure and C-13 NMR assignmen ts of indole alkaloids from Aspidosperma cruenta Journal of Crystallographic and Spectroscopic Research 23 1005-1011. Hartshorn, G. S. (1972) The ecological life history and population dynamics of Penthaclethra macroloba a tropical wet fo rest dominant and Stryphnodendron excelsum an occasional associate. Ph.D. Thesis, University of Washington, Seattle. Henry, H.A.L. & Thomas, S.C. (2002) Intera ctive effects of late ral shade and wind on stem allometry, biomass allocation, and mechanical stability in Abutilon theophrasti (Malvaceae). American Journal of Botany 89 1609-1615. Hoffmann, B., Chabbert, B., Monties, B., & Speck, T. (2003) Mechanical, chemical and x-ray analysis of wood in the two tropical lianas Bauhinia guianensis and Condylocarpon guianense : variations during ontogeny. Planta 217 32-40. Holbrook, N.M. & Putz, F.E. (1989) Influen ce of neighbors on tree form: effects of lateral shade and prevention of sway on the allometry of Liquidambar styraciflua (sweet gum). American Journal of Botany 76 ,1740-1749. Howe, H.F. & Schupp, E.W. (1984) Early Conseque nces of seed dispersal for neotropical tree ( Virola surinamensis ). Ecology 66 781-791. Ibarra-Manrquez G., Martinez Ramos, M. & Oyama, K. (2001) Seedling functional types in a lowland rain forest in Mexico. American Journal of Botany 88 18011812. Isnard, S., Speck, T. & Rowe, N.P. (2003) Me chanical architecture and development in Clematis : implications for canalized e volution of growth forms. New Phytologist 158 543-559. Kitajima, K. (1994) Relative importance of photosynthetic traits a nd allocation patterns as correlates of seedling shade tolerance of 13 tropical trees Oecologia 98 419428.

PAGE 91

79 Kitajima, K. (1996) Cotyledon functional morphol ogy, patterns of seed reserve utilization and regeneration niches of tropical tree seedlings. The Ecology of Tropical Forest Tree Seedlings (ed. M.D. Swaine), pp. 193-208,UNESCO, Paris. Kitajima, K. (2002) Do shade-tolerant trop ical tree seedlings depend longer on seed reserves? Functional growth analys is of three Bignoniaceae species. Functional Ecology 16 433-444 Kitajima, K. & Augspurger C.K. (1989) S eed and seedling ecology of a monocarpic tropical tree, Tachigalia versicolor. Ecology 70 1102-1114. Kitajima, K. & Fenner, M. ( 2000) Seedling regeneration ecology. Seeds: Ecology of Regeneration in Plant Communities 2nd edn (ed. M. Fenner), pp. 331-360. CAB International, Wallingford, UK. Kobe, R.K. (1997) Carbohydrate al location to storage as a basi s of interspecific variation in sapling survivorship and growth Oikos 80 226-233. Kobe, R. K. (1999) Light gradient par titioning among tropical tree species through differential seedling mortality and growth. Ecology 80 187-201. Leigh, E.G., Jr., Windsor, D.M. & Rand, S.A., eds. (1982) The Ecology of a Tropical Forest: Seasonal Rhythms and Long-Term Changes. Smithsonian Institution Press, Washington D.C. Loehle, C. (1988) Tree life history strategies: the role of defenses. Canadian Journal of Forest Research 18 209-222. Lucas, P.W., Beta, T., Darvell, B.W., Dominy, N.J., Essackjee, H.C., Lee, P.K.D., Osorio, D., Ramsden, L., Yamashita, N. & Yuen, T.D.B. (2001) Field kit to characterize physical, chemical and spa tial aspects of poten tial primate foods. Folia Primatologica 72 11-25. Lucas, P.W. & Pereira, B. (1990) Estima tion of the fracture toughness of leaves. Functional Ecology 4 819-822. Lucas, P.W., Turner, I.M., Dominy, N.J. & Yama shita. N. (2000) Mechanical defenses to herbivory. Annals of Botany 86 913-920. Mack, A.L. (1998) The potential impact of small-scale physical disturbance on seedlings in a Papuan rainforest. Biotropica 30 547-552. Marquis J. & Braker H.E. ( 1994) Plant-herbivore interactio ns: diversity, specificity, and impact La Selva Ecology and Natural History of a Neotropical Rain Forest (eds. L.A. McDade, K.S. Bawa, H.A. Hesp enheide & G.S. Hartshorn), pp. 261281.University of Chicago Press, Chicago.

PAGE 92

80 McCarthy, B.C. & Facelli, J.M. (1990) Microdi sturbances in old fields and forests: implications for woody seedling establishment. Oikos 58 55-60. Moles, A.T. & Westoby, M. (2004a) What do seedlings die from and what are the implications for evolution of seed size? Oikos 106 193-199. Moles, A.T. & Westoby, M. (2004b) seedling surv ival and seed size: a synthesis of the literature. Journal of Ecology 92 372-383. Molofsky, J. & Fisher B.L. (1993) Habitat a nd predation effects on seedling survival and growth in shade-tolerant tropical trees. Ecology 74 261-265. Montgomery, R.A. & Chazdon, R.L. (2002) Ligh t gradient partitioning by tropical tree seedlings in the absence of canopy gaps. Oecologia 131 165-174. Mou, P. & Warrillow, M.P. (2000) Ice storm da mage to a mixed hardwood forest and its impacts on forest regeneration in the ri dge and valley region of southwestern Virginia. Journal of the Torrey Botanical Society 127 66-82. Myers, J.A. (2005) Seedling carbohydrate stor age, survival, and stress tolerance in a neotropical forest. Master of Science Thes is, University of Florida, Gainesville. Niklas, K.J. (1992) Plant Biomechanics. An Engineer ing Approach to Plant Form and Function University of Chicago Press, Chicago. Niklas, K.J. (1995) Plant he ight and the properties of some herbaceous stems. Annals of Botany 75 133-142. Niklas, K.J., Molina-Freaner, F., Tinoco-O janguren, C. & Paolillo, D.J. (2000) Wood biomechanics and anatomy of Pachycereus pringlei American Journal of Botany 87 469-481. Osunkoya, O., Ash, J.E., Hopkins, M.S. & Graham, A.W (1992) Factors affecting survival of tree seedlings in North Queensland rainforests. Oecologia 91 569-578. Pauw, A., Van Bael, S.A., Peters, H.A., A llison S.D., Camargo, J.L.C., Cifuentes-Jara, M., Conserva, A., Garcia Restom, T., H eartsill-Scalley, T., Mangan, S.A., NunezIturri, G., Rivera-Ocasio, E., Rountree, M ., Vetter, S. & Volkmer de Casilho, C. (2004) Physical damage in relation to carbon allocation strategies to tropical forest tree saplings. Biotropica 36 410-413. Paz, H. & Martinez-Ramos, M. (2003) Seed mass and seedling performance within eight species of Psychotria (Rubiaceae). Ecology 84 439-450. Putz, F.E., Coley, P.D., Lu, K., Montalvo, A. & Aiello, A. (1983) Uprooting and snapping of trees structural determ inants and ecological consequences. Canadian Journal of Forest Research 13 1011-1020.

PAGE 93

81 Pyke, D.A. & Thompson, J.N. (1986) Statistica l analysis of survival and removal rate experiments. Ecology 67 240-245. Roldan, A.I. & Simonetti, J.A. (2001) Plant-mammal interactions in tropical Bolivian forests with different hunting pressures. Conservation Biology 15 617-623. Rowe, N.P. & Speck, T. (1996) Biomechani cal characteristics of the ontogeny and growth habit of the tropical liana Condylocarpon guianense (Apocynaceae). International Journal of Plant Sciences 157 406-417. Ryan M.G., Melillo, J.M. & Ricca, A. (1989) A comparison of methods for determining proximate carbon fractions of forest litter. Canadian Journal of Forest Research 20 166-171. Savidge, R.A. (2003) Tree growth and wood quality. Wood Quality and its Biological Basis (eds. J.R. Barnett, & G. Jeroni midis), pp. 1-29. Blackwell Publishing, Oxford. Scariot, A. (2000) Seedling mortality by lit terfall in Amazonian forest fragments. Biotropica 32 662-669. Sharpe, J.M. (1993) Plant growth and demogr aphy of a neotropical herbaceous fern Danea wendlandii (Marattiaceae) in a Costa Rican rain forest. Biotropica 25 8594. Shure, D. J. & Wilson, L. A. (1993) Patch-si ze effects on plant phenolics in successional openings of the Southern Appalachians. Ecology 74 55-67. Sibly, R.M. & Vincent, J.F.V (1997) Optimality approaches to resource allocation in woody tissues. Plant Resource Allocation (eds. F.A. Bazzaz & J. Grace), pp.143160. Academic Press, California. Smythe, N. (1978) The natural histor y of the Central American agouti ( Dasyprocta punctata ). Smithsonian Contribut ions to Zoology 157 1-52. Sork, V.L. (1987) Effects of predation and light on seedling establishment in Gustavia superba Ecology 68 1341-1350. Taylor, F.J. (1971) Some aspe cts of the growth of mango ( Manguifera indica L.) leaves. III. A mechanical analysis. New Phytologist 70 911-922. Turner, I.M. (1990) Tree seedling growth a nd survival in a Malaysian rain forest. Biotropica 22 146-154. Van Der Meer, P.J. & Bongers, F. (1996) Patter ns of tree-fall and branch-fall in a tropical rain forest in French Guiana. Journal of Ecology 84 19-29

PAGE 94

82 Weltzin, J.F., Archer, S. & Heitschmidt, R.K. (1997) Small-mammal regulation of vegetation structure in a temperate savanna. Ecology 78 751-763. Windsor D.M. (1990) Climate and moisture va riability in a tropica l forest: longterm records from Barro Colorado Island, Panama. Smithsonian Contributions to the Earth Sciences 29 ,145. Wright, I.J. & Cannon, K. (2001) Relationships between leaf lifespan and structural defences in a low-nutrient, sclerophyll flora. Functional Ecology 15 351-359. Wright, I.J., Reich, P.B., Westoby, M., Ac kerly, D.D., Baruch, Z., Bongers, F., Cavender-Bares, J., Chapin, T., Cornelis sen, J.H.C., Diemer, M., Flexas, J., Garnier, E., Groom, P.K., Gu lias, J., Hikosaka, K., Lamont, B.B., Lee, T., Lee, W., Lusk, C., Midgley, J.J., Navas, M.L., Niinemets, U., Oleksyn, J., Osada, N., Poorter, H., Poot, P., Prior, L., Pyankov, V.I., Roumet, C., Thomas, S.C., Tjoelker, M.G., Veneklaas, E.J., & Villar, R. (2004) The worldwide leaf economics spectrum. Nature 428 821-827. Wright, S.J., Muller-Landau, H.C., Condit, R., & Hubbell, S.P. (2003) Gap-dependent recruitment, realized vital rates, a nd size distributions of tropical trees. Ecology 84 3174-3185. Wright, W. & Illius, W. (1995) A comparativ e study of the fracture properties of five grasses. Functional Ecology 9 269-278.

PAGE 95

83 BIOGRAPHICAL SKETCH Silvia Alvarez-Clare was born in San Jos, Costa Rica in September 1977. She attended Saint Francis High School where she wa s inspired to become a biologist. Silvia earned her bachelorÂ’s degree in biology at the Universidad de Costa Rica (UCR) in July 2001. During her undergraduate studies at UCR, she became interested in plant ecophyisiology and conservation partly because she had many opportunities to work in diverse tropical ecosystems throughout Cost a Rica. She also worked as a research assistant in the Centro de Investigacin en Biologa Celular y Molecular at Universidad de Costa Rica (CIBCM) and at the university herbarium. Silvia was also a naturalist guide in the Monteverde cloud fo rest, an experience that reinfo rced her love of nature and her desire to continue her studies. When sh e finished her undergraduate studies, Silvia taught biology for a year in the Internati onal Baccalaureate program at Lincoln High School in San Jos. She then came to the University of Florida where she obtained her masterÂ’s degree in Botany with a minor in St atistics in May 2005. S ilvia will continue her education toward her doctoral degree in interdisciplinary ecol ogy under the guidance of Dr. Michelle Mack at the Univer sity of Florida. For her disse rtation research she plans to study tropical forest regene ration, restoration, and nutrien t cycling in Costa Rica.