biological processes influencing nutrient...

1

BIOLOGICAL PROCESSES INFLUENCING NUTRIENT LIMITATION IN A LOWLAND TROPICAL WET FOREST IN COSTA RICA

By

SILVIA ALVAREZ CLARE

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT

OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2012

2

© 2012 Silvia Alvarez Clare

3

To my parents, my rudder to Chuck, my anchor

and to Lucia Jane, my shining star in dark seas

4

ACKNOWLEDGMENTS

I would like to thank my advisor Michelle C. Mack. Her great ideas, unconditional

support of my research, and patience have made this dissertation possible. Her passion

for science and mechanistic approach to ecological questions, have greatly shaped my

views as a scientist. I would also like to thank my committee members Emilio Bruna,

Nick Comerford and Ted Schuur for their support and mentorship during this long

process. All my professors at UCR and at UF have influenced me and helped me to

achieve this goal, and I thank them.

I acknowledge the National Science Foundation, the Tropical Conservation and

Development Program at the University of Florida, the South Eastern Alliance for

Graduate Education and Professoriate (SEAGEP), Elizabeth Clare-Rhoades, and

Patricia Clare for financial support. The Department of Biology and the School of

Natural Resources and the Environment at the University of Florida, as well as the

Escuela de Agricultura de la Región del Trópico Húmedo (EARTH), provided valuable

institutional support.

I would like to extend my deepest appreciation to all the people that came to the

field with me during these six years and that endured mud, rain, mosquitoes, spines,

snakes, and endless hours of fieldwork: Agustin Alvarez, Adolfo Artavia, Adrián

Villalobos, Andy Retzler, Balbina García, Carlos (Pelón), Catherine Cardelús, Charles

Knapp, Danielle Pallow, Eduardo Chacón, Elida Madrid, Enrique Salicetti, Faeleen Tais,

Fernanda Arhernas, Hanna Lee, Ismael Herrera, Jenny Bermudez, Jonathan Artavia

(La Selva), José Antonio Dominguez, José Zuniga, Josué Beltetón, Karla Ayala, Katy

Evans, Laura Morales, Laura Schreeg, Ma Alexandra Chicaciza, Maga Gei, María de

los Angeles, Melania Fernandez, Michelle Mack, Ondřej Vybíral, Rady Ho, Rigoberto

5

Gonzalez, Rolo, Sara Kovachich, Shantelle Bartra, Tony Arévalo, Violeta Rodriguez,

Virginia Guillén, and Yariela Ugalde. I also thank the people who helped me during what

seemed like endless hours in the laboratory; especially Grace Crummer and Julia

Reiskind, who with their advice and friendship made my life better during these years.

All members of the Mack and Schuur labs were great colleagues and friends,

contributing to exciting science discussions and providing support during difficult times.

My office mates Jennie DeMarco, Jenny Shafer, and Caitlin Hicks were great friends,

who I will always remember.

During my field work at EARTH University, many people provided logistical and

technical support. Professors Bert Kolhlmann and Ricardo Russo served as my local

mentors. Warner Vargas, Checo, and all the staff at the “Unidad de Ingeniería Agrícola”

allowed me to use the workshop and the tractor to fertilize my plots. Melissa Arce and

Alejandra Carvajal helped with lodging. Carlos Sandí and the staff at the “Vivero

Forestal” provided help in the field and with processing seedlings. Herbert Arrieta and

the staff at the “Laboratorio de Suelos” provided invaluable support, which included

work and storage space, assistance with laboratory analysis, and their friendship.

My friends, both the “ticos” and “the gang” were critical to keep me sane during my

PhD. I would like to thank them for making these years some of the most special of my

life. I do not have enough words to thank my family (my parents, my brothers, my

abuelitas) for their unconditional love and support. My mother is my example to follow; I

would like to thank her and tell her: you are next! This dissertation should probably be

co-authored with Chuck Knapp. He has read and edited every page, provided feedback,

and supported me every step of the way. His love, patience, and encouragement have

6

been infinite. My baby Lucia Jane, who was conceived, born, and turned one during my

PhD, reminds me every day of what really matters in life. Finally, I thank God, in his

universal non-denominational form, for giving me health, strength and perseverance to

fulfill this goal.

7

TABLE OF CONTENTS page

ACKNOWLEDGMENTS .................................................................................................. 4

LIST OF TABLES .......................................................................................................... 10

LIST OF FIGURES ........................................................................................................ 12

LIST OF ABBREVIATIONS ........................................................................................... 14

ABSTRACT ................................................................................................................... 15

CHAPTER

1 BACKGROUND AND MOTIVATION ...................................................................... 17

2 INFLUENCE OF PRECIPITATION ON SOIL AND FOLIAR NUTRIENTS ACROSS NINE COSTA RICAN FORESTS ............................................................ 27

Introduction ............................................................................................................. 27 Methods .................................................................................................................. 31

Study Site ......................................................................................................... 31 Soil Sampling and Analysis .............................................................................. 31

Foliage Sampling and Analysis ........................................................................ 33 Data Analysis ................................................................................................... 34

Results .................................................................................................................... 34 Soil Characteristics ........................................................................................... 34 Foliar Measurements ........................................................................................ 35

Relationship Between Soil and Foliar Measurements ...................................... 36 Discussion .............................................................................................................. 36

Patterns of Soil Nutrient Availability .................................................................. 36 Pattern of Foliar Nutrients ................................................................................. 39 Conclusions ...................................................................................................... 41

3 DIRECT TEST OF NUTRIENT LIMITATION TO NET PRIMARY PRODUCTIVITY IN A LOWLAND TROPICAL WET FOREST ............................... 53

Introduction ............................................................................................................. 53

Methods .................................................................................................................. 58

Site Description ................................................................................................ 58 Experimental Design ........................................................................................ 61 Soil Measurements ........................................................................................... 61 Tree Diameter Measurements .......................................................................... 63 Seedling Measurements ................................................................................... 65 Litterfall Collection ............................................................................................ 66 Root Measurements ......................................................................................... 67

8

Statistical Analysis ............................................................................................ 68

Results .................................................................................................................... 70 Soil Measurements ........................................................................................... 70

Tree Growth ..................................................................................................... 71 Seedling Measurements ................................................................................... 73 Litterfall and Litter Pool Measurements ............................................................ 73 Root Biomass and Productivity ......................................................................... 74

Discussion .............................................................................................................. 75

Soil Measurements ........................................................................................... 75 Nutrient Limitation to NPP ................................................................................ 76 Nutrient Limitation and Tree Size ..................................................................... 79 Effect of Taxa on Nutrient Limitation ................................................................ 81 Conclusions ...................................................................................................... 82

4 EFFECT OF NUTRIENT ADDITIONS ON FOLIAR, LITTER AND ROOT CHEMISTRY ......................................................................................................... 109

Introduction ........................................................................................................... 109

Methods ................................................................................................................ 113 Experimental Design ...................................................................................... 113 Chemical Analysis .......................................................................................... 115

Statistical Analysis .......................................................................................... 116 Results .................................................................................................................. 117

Foliar Nutrients ............................................................................................... 117 Litterfall Nutrients ........................................................................................... 119 Root Nutrients ................................................................................................ 120

Discussion ............................................................................................................ 121 Effects of Fertilization on Foliar Nutrients ....................................................... 121

Influence of Tree Size on Foliar Nutrients ...................................................... 123 Influence of Taxa on Foliar nutrients .............................................................. 124

Effects of Fertilization on Litterfall and Root Nutrients .................................... 125 Total Soil P as a Driver of Tissue Nutrient Concentrations ............................. 126 Conclusions .................................................................................................... 127

5 CONCLUSIONS AND LESSONS LEARNED ....................................................... 146

Conclusions .......................................................................................................... 146 The Myth of P Limitation in the Tropics .......................................................... 147 Heterogeneous Nutrient Limitation ................................................................. 147

Environmental and Biological Processes Influence Nutrient Limitation and Carbon Cycling ............................................................................................ 148

Lessons Learned .................................................................................................. 149 APPENDIX

A SPATIAL AND TEMPORAL VARIATION OF LIGHT REACHING THE UNDERSTORY IN A WET TROPICAL FOREST IN COSTA RICA ...................... 156

9

B BIOTIC AND ABIOTIC FACTORS INFLUENCING TREE GROWTH IN A LOWLAND TROPICAL WET FOREST: A MIXED MODEL APPROACH ............. 160

C ADDITIONAL TABLES AND FIGURES ................................................................ 167

LIST OF REFERENCES ............................................................................................. 174

BIOGRAPHICAL SKETCH .......................................................................................... 190

10

LIST OF TABLES

Table page 1-1 Fertilization experiments conducted in tropical forests where components of net

primary production were measured ........................................................................ 24

2-1 Characteristics of study sites in Costa Rica ............................................................ 43

2-2 Species and number of samples collected per site in nine sites in Costa Rica ....... 44

2-3 Soil nutrients and isotopic signatures (mean + SE) for the nine study sites in Costa Rica .............................................................................................................. 46

2-4 Foliar carbon, phosphorus, nitrogen and stable isotopic signatures (mean + SE) for the nine study sites in Costa Rica ..................................................................... 48

3-1 Particle sizes for soils from the top 10cm of the study plots .................................... 84

3-2 Means (with standard errors) for various soil parameters ....................................... 84

3-3 Means (with standard errors) for various soil parameters. ...................................... 85

3-4 Taxa selected to study the effect of fertilization on different functional groups ....... 86

3-5 Results from repeated measures MANOVAs for several soil variables measured at three depths ........................................................................................................ 87

3-6 Results from two-way ANOVA analyses for the percent difference between 2 yrs and pre-fertilization values for several soil parameters ........................................... 88

3-7 Results from repeated measures MANOVAs for “total basal area increase” by tree size class ......................................................................................................... 89

3-8 Results from repeated measures MANOVAs for “proportion of tree growth” by tree size class ......................................................................................................... 90

3-9 Contingency table describing the proportion of trees that grew or did not grow between 2 and 2.7 yrs after initial fertilization ......................................................... 91

3-10 Results from Pearson chi-square tests for seedling variables measured 1yr and 2 yrs after fertilization ..................................................................................... 93

3-11 Results from repeated measures MANOVAs for foliar (leaves and sticks <2mm diameter), reproductive (flowers and fruits), and coarse litterfall ........................... 93

3-12 Results from two-way ANOVA analyses for litterpool fractions ............................. 94

11

3-13 Results from repeated measures MANOVAs for fine (<2mm diameter) and large (>2mm diameter) root biomass collected at 0-15 cm depth.......................... 94

4-1 Results from repeated measures MANOVAs for foliar chemistry by tree size class .................................................................................................................... 130

4-2 Results from repeated measures MANOVAs comparing foliar %N, P (mg g-1) and N:P ratios among the six most common taxa ............................................... 131

4-3 Results from repeated measures MANOVAs for foliar %N by species. ................ 132

4-4 Results from repeated measures MANOVAs for foliar P by species. .................... 133

4-5 Results from repeated measures MANOVAs for litterfall chemistry ...................... 134

4-6 Results from repeated measures MANOVAs for root chemistry ........................... 135

5-1 Summary of f responses ratios (RR) of treatments relative to the control in the fertilization experiment .......................................................................................... 154

B-1 Parameters used in the mixed models .................................................................. 163

B-2 Models used to test the hypotheses ..................................................................... 164

B-3 Models used to test the hypotheses organized by increasing AIC values ............ 164

B-4 Maximum likelihood estimates, their standard error, and T-value for parameters included in model M1 ............................................................................................ 165

C-1 Species of trees found in the study plots .............................................................. 167

C-2 Results from repeated measures MANOVAs for foliar N:P ratios by tree size class ..................................................................................................................... 170

C-3 Results from repeated measures MANOVAs for foliar N:P ratios by species ....... 171

12

LIST OF FIGURES

Figure page 1-1 Diagram showing the evolution of nitrogen (N) and phosphorus(P) availability

through soil development. ...................................................................................... 26

1-2 Plant-soil- microbial feedbacks after addition of the limiting nutrient ....................... 26

2-1 Map of sites where soil and foliar samples were collected in Costa Rica ................ 49

2-2 Relationship between mean annual precipitation (+ SE) and soil parameters ....... 50

2-3 Mean N:P ratios (+ SE) .......................................................................................... 51

2-4 Mean soil phosphorus (+ SE) for the top 10 cm mineral layer of nine study sites in Costa Rica ......................................................................................................... 52

3-1 Monthly average maximum and minimum air temperatures (dashed lines) and precipitation (solid lines) at the study site .............................................................. 95

3-2 Floristic description of the study site ...................................................................... 96

3-3 Distribution of plots and blocks within the EARTH forest reserve ........................... 97

3-4 Mean + SE soil parameters ..................................................................................... 98

3-5 Mean + SE percent change in various soil parameters two years after fertilization .............................................................................................................. 99

3-6 Mean (+ SE) total basal area increase per treatment ............................................ 100

3-7 Mean (+ SE) percentage of trees that grew per plot ............................................ 101

3-8 Box plots of relative growth rates (RGR) measured between 0.4 and 2.7 yrs after initial fertilization, in the four nutrient addition treatments ............................. 102

3-9 Box plots showing relative growth rates for six common tree species, measured between 0.4 and 2.7 yrs. after initial fertilization, in the four nutrient addition treatments............................................................................................................. 103

3-10 Mean (+ SE) percent of seedlings ....................................................................... 104

3-11 Mean (+ SE) foliar litterfall production (foliage + sticks <2 mm in diameter) ........ 105

3-12 Mean (+ SE) coarse litterfall production .............................................................. 106

3-13 Mean (+ SE) fine root biomass (roots <2mm diameter) ...................................... 107

13

3-14 Box plot (25th percentile, median and 75th percentile) for root production from ingrowth cores ..................................................................................................... 108

4-1 Mean (+ SE) foliar N and P for trees with DBH between 5-10 cm ........................ 136

4-2 Box Plot comparing variability in N:P ratios at EARTH forest before fertilization and 2yrs after fertilization .................................................................................... 137

4-3 Mean (+ SE) foliar %N for six common tree species for the four nutrient addition treatments ............................................................................................................ 138

4-4 Mean (+ SE) foliar P for six common tree species for the four nutrient addition treatments ............................................................................................................ 139

4-5 Mean (+ SE) foliar N:P ratios for six common tree species for the four nutrient addition treatments .............................................................................................. 140

4-6 Relationship between several soil variables and plot-averaged foliar %N ............ 141

4-7 Relationship between several soil variables and plot-averaged foliar P ............... 142

4-8 Mean (+ SE) litterfall chemistry ............................................................................. 143

4-9 Mean litterfall nutrient concentrations by taxa ....................................................... 144

4-10 Mean (+ SE) root chemistry. ............................................................................... 145

5-1 Diagram representing how environmental and biological factors could interact to influence nutrient cycling in a diverse tropical forest ............................................. 155

A-1 Mean (+ SE) transmitted diffuse light for the four nutrient addition treatments ..... 158

A-2 Hemispherical canopy photograph taken in the same position at three successional dates ............................................................................................... 159

B-1 Box plots showing stem diameter increase (calculated as relative growth rate, see methods) for the four fertilizer treatments ...................................................... 165

B-2 Box plots showing stem diameter increase (calculated as relative growth rate, see methods) for the most common canopy palm, Socratea exorrhiza ................ 166

B-3 Relationship between initial diameter at breast height (DBH) and stem diameter increase ................................................................................................................ 166

C-1 Relationship between several soil variables and plot-averaged foliar N:P ratios .. 172

C-2 Relationship between soil Total P and Melich P ................................................... 173

14

LIST OF ABBREVIATIONS

BA Basal area (m2)

DBH Diameter at breast height (1.3 m above ground)

DIN Dissolved Inorganic Nitrogen (ug N g-1)

IVI Importance value index. Measured from the relative frequency, the relative density and the relative basal area of a species in a plot

K Potassium

N Nitrogen

NPP Net primary productivity (Mg C ha-1yr-1)

P Phosphorus

RGR Relative growth rate (ln(mm yr-1)). Calculated as the slope from a line of all the diameter measurements for each stem

15

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

BIOLOGICAL PROCESSES INFLUENCING NUTRIENT LIMITATION IN A LOWLAND

TROPICAL WET FOREST IN COSTA RICA

By

Silvia Alvarez Clare

May 2012 Chair: Michelle Caitlin Mack Major: Interdisciplinary Ecology

Nutrient limitation by nitrogen (N) and phosphorus (P) is pervasive in most

ecosystems on Earth. Thus, increases of available nutrients due to agricultural practices

and pollution may alter key ecosystem processes, such as net primary productivity

(NPP), and carbon sequestration in soils. However, our understanding of how nutrients

influence C dynamics in tropical forests remains far from complete. During my

dissertation work, I conducted two studies (one observational and one experimental) to

explore how environmental and biological processes influence nutrient limitation in

lowland tropical forests.

In my first study, I explored patterns of soil and foliar nutrients (specifically N and

P) across nine, relatively wet, mature lowland forests in Costa Rica. My objective was to

investigate the relationship between rainfall and plant or soil nutrients to better

understand the potential long-term effects that alterations in MAP could have on the

nutrient dynamics of wet forest plant communities. Across the gradient, soil N was

relatively more abundant than P but was also more sensitive to changes in MAP. I

concluded that complex feedbacks and interactions among environmental and biological

factors make it difficult to predict, by conducting observations on current patterns, how

16

changes in climate (e.g., MAP) or plant resources (e.g., nutrient availability) will

influence nutrient limitation in these forests.

In my second study, I conducted a fertilization experiment to directly test if N or P

availability limit NPP in a lowland tropical forest in Costa Rica and to explore how

biological factors, such as tree species composition and size, responded to added

nutrients. There was no significant effect of either N or P fertilization on tree diameter

increase, litterfall production, or root biomass two years after initial fertilization.

However, there were interesting and contrasting responses among tree species and

size classes. Results stemming from this experiment suggest that although soils have

high N relative to P, NPP is not necessarily limited by P, or at least does not respond to

either N or P fertilization in the short term, and highlights the importance of considering

biological factors, such as species composition and life history traits, before making

generalizations regarding nutrient limitation in tropical forests.

.

17

CHAPTER 1 BACKGROUND AND MOTIVATION

Tropical forest ecosystems provide numerous vital services for people and society

such as, biodiversity preservation, biomedical exploration, water resources, food, and

fiber (Zarin et al. 2004). Additionally, tropical forests contribute 59% of global forest

vegetation, 27% of soil carbon (C) pools (Dixon et al. 1994). They also account for at

least one-third of the annual biosphere-atmosphere carbon dioxide (CO2) exchange

(Field et al. 1992, Grace et al. 2001, 2006), thus making these systems an important

component of the global C cycle.

Contemporary human activities worldwide, however, are altering tropical forests at

an alarming rate. Perturbations include not only changes in climatic variables, such as

temperature and precipitation, but also an increase in limiting resources used by plants

and microbes. Nutrient availability, for example, has been largely increased by

atmospheric nutrient deposition that results from changing global biogeochemical cycles

(Smil 2000, Tilman et al. 2001, Galloway and Cowling 2002, Galloway et al. 2004). In

recent years global mobilization of phosphorus (P) has roughly tripled compared to its

natural flows (Smil 2000). In addition, fire, land use change, and indiscriminate use of

fertilizers have significantly increased atmospheric nitrogen (N) and P inputs in the

tropics (Galloway et al. 2004, Okin et al. 2004, Mahowald et al. 2008).

Nutrient limitation, especially of N and P, is pervasive in many ecosystems on

earth (Vitousek and Howarth 1991, Elser et al. 2010, Vitousek et al. 2010), and has

been shown to be an important control on C storage and cycling in multiple ecosystems

(Elser et al. 2007). Therefore, human-caused increases in nutrient availability could

have profound effects on tropical forest ecosystems by altering how nutrients influence

18

C accumulation through net primary productivity (NPP). However, we have an

incomplete understanding of the mechanisms that link nutrient limitation and availability

with C storage and cycling in tropical forests.

Traditionally, it has been assumed that in tropical forests, NPP is limited by P

(Figure 1-1, Walker and Syers 1976, Vitousek 1984, Vitousek and Farrington 1997).

Tropical forests usually occur in old soils, where the original P-rich parent material has

been weathered and most of the remaining P is occluded on iron and aluminum oxides

(Sanchez 1976, Miller et al. 2001). Nitrogen, by contrast, accumulates over time through

biological fixation, and is therefore relatively more available than P in old soils.

However, multiple abiotic and biotic factors can influence nutrient availability, so that not

all tropical forests occurring in old soils are limited only by P. In fact, direct evidence

from the few existing fertilization experiments conducted in the tropics (Table 1-1) and

indirect evidence from other biochemical and biological parameters (e.g., Reich and

Oleksyn 2004, McGroddy et al. 2004, Cleveland et al. 2011) suggest that N, P, or other

nutrients (e.g., potassium; Wright et al. 2011) can limit NPP in tropical forests. In this

dissertation, I explored how environmental and biological processes influence nutrient

limitation in moist-to-wet lowland tropical forests in Costa Rica. I used two

complementary approaches, consisting of an observational gradient study and an

experimental nutrient manipulation.

In Chapter 2, I present results from a study concerning patterns of soil and foliar

nutrients (specifically N and P) across nine, relatively wet (mean annual precipitation

(MAP) > 3500mm), mature lowland forest sites in Costa Rica. My objective was to

investigate the relationship between rainfall, and plant or soil nutrient characteristics to

19

better understand the potential long-term effects that alterations in mean annual

precipitation (MAP) could have on the nutrient dynamics of wet forest plant

communities. A gradient approach was a good alternative for this study because it

provided an integrated view of the long term effects that precipitation regimes can have

on ecosystem processes. I focused on the high-end of the precipitation spectrum (MAP

from 3500 to 5500 mm) because limited information exists concerning the

biogeochemical effects of abundant precipitation in wet tropical forests (Wieder et al.

2009). In areas with abundant precipitation, high water inputs can cause removal of

mobile nutrients in the soil solution via leaching (Radulovich and Sollins 1991), reduced

mineralization of nitrogen (N) and phosphorus (P) in poorly drained soils or anaerobic

microsites (Schuur and Matson 2001), slower decomposition of organic matter (Schuur

2001), and more intense weathering over time (Walker and Syers 1976, Crews et al.

1995). As a result, areas with high precipitation, where nutrients have been either

leached or occluded, can have soils with low nutrient supply rates. In this study, I tested

if patterns of soil and foliar nutrient concentrations followed patterns of MAP regardless

of co-varying factors such as altitude and species composition.

To complement the gradient study, and to learn more about the mechanisms

driving the patterns of soil and foliar nutrients observed across the nine Costa Rican

forests, I conducted one of the few fertilization experiments existing in the lowland

tropics to date. This experiment was established at the EARTH (Escuela de Agricultura

de la Región del Trópico Húmedo) Forest Reserve, in the lowlands of the Caribbean

slope of Costa Rica. As part of this experiment, I tested if N or P availability limits NPP

in this forest and explored how biological factors, such as tree species composition and

20

size, responded to added nutrients. Although this study included a significantly smaller

area than the gradient study and explored the effect of nutrient additions in a short time

scale, it provided drastically more detailed information than the gradient study on the

effects that changes in nutrient availability can have on tropical forest processes.

As a theoretical background to formulate hypotheses in the fertilization

experiment, I utilized a mechanistic model proposed by Vitousek (2004), which stems

from long-term experiments in the Hawaiian archipelago (Figure 1-1). According to this

model, in a system limited by P (as is traditionally expected of tropical forests in old

soils), increased P availability results in a positive feedback that reinforces high nutrient

availability and causes the P-limited forest to resemble a naturally P-rich system (Figure

1-2A). Specifically, increased P availability after fertilization leads to decreased P

residence times in the forest canopy, decreased overall P-use efficiency, increased P

concentrations in plant tissues and in leaf litter, increased rates of litter decomposition,

and more rapid regeneration of P from decomposing litter. The Hawaii experiments

provide a remarkable starting point for understanding the mechanisms reinforcing

nutrient limitation in the tropics; however, these experiments occurred in montane wet

forests dominated by a single tree species, Metrosideros polymorpha. By contrast, more

diverse continental tropical forests, such as the one at EARTH, are likely to contain a

wider array of plant functional traits related to nutrient use. Consequently, plant-soil-

microbial feedbacks in diverse tropical forests may respond differently to an increase in

nutrient availability than the monodominant forests in Hawaii.

One key aspect in which monodominant and diverse forests may differ is that in

nutrient-limited, diverse continental forests, species composition and identity could

21

influence the ecosystem response to nutrient addition. For example, in a diverse P-

limited forest, the more successful and thus more abundant species are expected to

have low P metabolic demand or high P-use efficiency (Chapin, Vitousek and Van

Cleve 1986). At the same time, these successful (i.e., abundant) species are expected

to contribute to the majority of the litterfall production of the system. If, for example,

these species respond to increased P availability by increasing their leaf (and litterfall)

production, but not their litterfall nutrient concentrations, then an alternative pathway

than that observed in the Hawaii experiments, may occur (Figure 1-2B). Moreover, if the

dominant species due to their low P metabolic demand do not respond to P fertilization,

no positive feedback to nutrient addition would be observed. In consequence, even if

the rest of the species in the community are strongly P limited and respond to P

additions, their response to fertilization will have no effect on community dynamics in

the short term (i.e., no feedback could be observed) because their litterfall mass is not

enough to dominate the plant community average. In the long-term, however, if P soil

availability remains high there could be a shift in species composition stemming from

differential regeneration of species with higher P demand.

In synthesis, nutrient limitation in diverse tropical forests is framed by individual

species and their interactions. Therefore, it is critical that species composition and life

history traits are considered in studies investigating nutrient limitation in diverse forests.

One innovative aspect of this dissertation is the exploration of taxa-specific responses

of some common tree species, with the objective of elucidating the role that individual

species have on the “community wide” signal of nutrient limitation.

22

Another understudied aspect of nutrient limitation in tropical forests is the

relationship between different demographic groups and nutrient cycling. In

monodominant forests (e.g., Hawaii), population growth rates (λ) will only be determined

by the vital rates of a single tree species and thus changes in nutrient availability will not

have consequences for community composition and biodiversity. By contrast in diverse

tropical forests, alterations in vital rates of different species due to changes in nutrient

availability can lead to shifts in community composition (Ceccon et al. 2004), which in

turn affect plant-soil-microbial feedbacks. Additionally, there may be resource

partitioning in space and time among age groups and different age groups may be

limited by different nutrients. For example adult trees, which intercept the majority of

light reaching the canopy and possess greater root area, may assimilate more nutrients

than light-limited saplings and seedlings (Lambers et al. 1998). Seedlings in the

understory may access nutrients added in fertilizer, but may be unable to incorporate

them in their tissues due to light co-limitation (Burslem et al. 1995). In this dissertation I

tested the effect of fertilization on stem diameter increase for trees from different size

classes, including measurements on tree seedlings.

In Chapter 3, I report the effects of a two-year N and P fertilization experiment on

different components of NPP, including tree diameter increase, litterfall productivity, root

productivity, and seedling growth and survival. In Chapter 4, I report the effects of

fertilization on nutrient (N and P) concentrations of leaves from the most common tree

species, of litterfall, and of roots collected during the experiment. Finally, in Chapter 5 I

provide some overall conclusions and discuss lessons learned and potential further

directions.

23

Overall, my dissertation work encompasses a comprehensive investigation of

nutrient limitation in the humid-to-wet lowland tropical forests of Costa Rica. Soil and

foliar nutrient data from the sites included in the gradient study provide a useful baseline

for future investigations in these sites. In addition, by conducting one of the few factorial

fertilization experiments in a lowland tropical wet forest, I hope to advance the general

understanding of plant-soil-microbial feedbacks in these systems and the role that these

feedbacks play for carbon sequestration and biodiversity preservation.

24

Table 1-1. Fertilization experiments conducted in tropical forests where components of net primary production were measured. Trt. = treatment, n.s = non-significant results, TSP = Triple Super Phosphate, ky = 1,000 yrs.

Source Duration (yrs)

Plot size and replicates

Trt. Amount (Kg Ha

-1

yr-1

)

Component of primary production Location Comments

Wood Fine roots Fine litter Coarse debris

Tanner et al. 1992

4.5yr 12x12 m x 5 reps

+N +P +NP

225/150 (urea) 75/50 (TSP)

No data No data Increased

No data n.s. n.s. Increased

No data Montane forest, Venezuela

Changed fertilizer quantities after 2 yrs

Mirmanto et al. 1999

Litterfall =1 yr Girth = 5 yrs

50 x 50 m x 5 reps

+N +P +NP

225 (urea) 75 (TSP)

n.s. n.s. n.s.

No data Increased Increased Increased

No data Lowland primary forest, Kalimantan, Borneo

Dipterocarp dominated forest

Newbery et al. 2002

2 yrs +P n.s. No data No data Cameroon

Hawaii (Summarized in Vitousek 2004)

variable Tree-centered 10 x 10 m or 15 x 15 m reps 4-6

+N +P +T* +NT +NP +PT +NPT

100 (NH4NO3 + urea) 100 (TSP)

Data exist for all these fractions but responses varied among sites and experiments.

Montane forests, Hawaii

Parent material ranging in age from 0.3 ky to 4100 Ky

Davidson et al. 2004

3 yrs (prefert. + 2 post-fert.)

20 x 20 m x 3 reps

+N +P +NP

100 (urea) 50 (TSP)

Increased

n.s

Increased

No data No data No data 6 yr-old secondary forest, Amazon basin, Brazil

25

Table 1-1. Continued.

Source Duration (yrs)

Plot size and replicates

Trt. Amount (Kg Ha

-1

yr-1

)

Component of primary production

Location Comments Wood Fine roots Fine litter

Coarse debris

*T = Other elements essential to plant growth, except N and P. These are Ca, Mg, K, S, Fe, Mn, Zn, Cu, B, Mo. For quantities and forms of elements added, refer to Vitousek 2004, Table 5.1)

Wright et al. 2011

11 yrs 40 x 40 m

x 4 reps +N +P +K +NK +NP +PK +NPK

125 (urea) 50 (TSP) 50 (KCl)

+P: Less decrease (large trees) +NK: Less decrease (small trees)

+K: Decrease

+P: small Increase

No data >200 yr secondary seasonal forest, Panama

There was a general decrease in growth over time but the decrease was less in several treatments

26

Figure 1-1. Diagram showing the evolution of nitrogen (N) and phosphorus(P) availability through soil development. Modified from Walker and Syers (1976).

Addition

of limiting

Nutrient

+ Nutrient

Availability -MRT

- NUE

+ Foliar

nutrient

concentration

+ Decomposition

rate

+ Litter nutrient

concentration

+ Regeneration

of nutrients

A + Nutrient

Availability

= MRT

= NUE

= Foliar

nutrient

concentration

= Decomposition

rate

= Litter nutrient

concentration

+ Regeneration

of nutrients

Addition

of limiting

Nutrient

+ Litterfall pool

+ Foliage

production

B

Figure 1-2. Plant-soil- microbial feedbacks after addition of the limiting nutrient. (A)

Evidence from Hawaii (Vitousek 2004). (B) Potential alternative feedback for a diverse tropical forest (this study) if the dominant species increases productivity, but not foliar nutrient concentrations, after P addition of the limiting nutrient. “MRT” refers to mean residence time and “NUE” to nutrient use efficiency. A plus sign indicates an increase, a minus sign a decrease, and an equal sign no change in the specific process.

N

27

CHAPTER 2

INFLUENCE OF PRECIPITATION ON SOIL AND FOLIAR NUTRIENTS ACROSS

NINE COSTA RICAN FORESTS

Introduction

Within the tropics, precipitation is expected to be one of the most important

variables affected by global climate change (Enquist 2002, Douville et al. 2006, Neelin

et al. 2006, IPCC 2007). Indeed, alterations in frequency, intensity and seasonality of

rainfall have the potential to affect multiple aspects of tropical ecosystems. Alterations in

precipitation can affect some ecosystem processes, such as forest regeneration (Sack

and Grubb 2002, Bunker and Carson 2005), net primary productivity (NPP; Clark et al.

2001), and litter decomposition (Powers et al. 2009, Wieder et al. 2009) within a short

time scale. Other consequences of altered precipitation regimes, however, are likely to

take decades to millennia to become evident. For example, changes in tree species

composition (Engelbrecht et al. 2007), biodiversity (Bazzaz 1998), soil development

(Walker and Syers 1976) and multiple aspects of nutrient cycling (Miller et al. 2001,

Schuur and Matson 2001, Vitousek 2004) usually occur over long periods.

Two approaches to investigate how changes in precipitation affect ecosystem

processes include water-manipulation experiments and observations across natural

precipitation gradients. There are benefits and limitations to both approaches.

Manipulation experiments, such as water addition (Yavitt and Wright 2008) or

precipitation exclusion (Nepstad et al. 2002), have the advantage of changing only one

or few factors at a time (e.g., water and nutrients), and are appropriate to investigate

short-term effects of precipitation. These experiments, however, are expensive, usually

This chapter has been published in Biotropica, 43: 433–441.

28

performed at a small scale, and are not appropriate to investigate long-term effects of

changes in mean annual precipitation (MAP). Natural precipitation gradients present an

interesting alternative to explore the long term effects of MAP on ecosystems. Ideally, in

a gradient study all “state factors” (climate, parent material, topography, time and biota;

sensu Jenny 1941), are kept constant, except for the factor being studied. This is mostly

the case in the Hawaiian archipelago, where multiple studies using MAP gradients have

been conducted (Austin and Vitousek 1998, Schuur and Matson 2001, Idol et al. 2007,

Houlton et al. 2007). One limitation of gradient studies, however, is that in most non-

island systems, multiple state factors vary systematically, making it challenging to

separate direct effects of MAP, for example, from other environmental factors. Despite

these limitations, gradient studies are useful because they provide insight on the long

term effects that precipitation regimes can have on ecosystem processes and can help

predict the effects that changes in MAP will have on ecosystems.

In this study, I used a precipitation gradient in Costa Rica to investigate the

potential long term influence that alterations in precipitation could have on nutrient

cycling and nutritional status of diverse- forest communities. I attempted to keep other

environmental factors constant by selecting nine, predominantly lowland, mature- forest

sites where altitude ranged from 200 to 1200m elevation. I focused on the high-end of

the precipitation spectrum (MAP from 3500 to 5500 mm) because limited information

exists concerning the biogeochemical effects of the abundant precipitation characteristic

of wet tropical forests (Wieder et al. 2009). In areas with abundant precipitation, high

water inputs can cause removal of mobile nutrients in the soil solution via leaching

(Radulovich and Sollins 1991), reduced mineralization of nitrogen (N) and phosphorus

29

(P) in poorly drained soils or anaerobic microsites (Schuur and Matson 2001), slower

decomposition of organic matter (Schuur 2001), and more intense weathering over time

(Walker and Syers 1976, Crews et al. 1995). As a result, areas with high precipitation,

where nutrients have been either leached or occluded, can have soils with low nutrient

supply rates.

Foliar nutrient concentrations (especially N and P) can track soil fertility and

therefore can be good indicators of the nutritional status of an ecosystem. For example,

in a study across a precipitation gradient in Hawaii, Schuur and Matson (2001) reported

a decrease in soil and foliar N concentrations (but not P) with increased precipitation.

Increased precipitation along a gradient in Panama was associated with decreased

foliar P, calcium and magnesium (Santiago et al. 2004). Relative concentrations of foliar

nutrients can also provide insight regarding nutrient limitation in a system.

Stoichiometric analyses of foliar nutrient concentrations (especially N and P) have

been used as an index of nutritional status of plants (e.g., Aerts and Chapin 2000,

Vitousek 2004) and can provide insight into processes such as net primary productivity

(NPP), decomposition, nutrient mineralization, trace gas emissions, and leaching

losses. Several studies have shown that N:P ratios above 16 (mass based) are typical

of sites limited by P and values below 14 are typical of sites limited by N, although

there is a substantial amount of variation around these threshold values (Güsewell

2004). At a global scale, foliar P increases and N:P ratios decline with increasing

latitude, supporting the basic hypothesis that the tropics are relatively more P limited

than temperate zones (Reich and Oleksyn 2004). However multiple factors, in addition

to soil fertility, can influence foliar nutrient concentrations and their ratios. Two factors

30

that can strongly influence nutrient concentrations are life history strategy and

physiology. For example, potentially N-fixing leguminous species in several tropical

forests have higher foliar N concentrations than species from other families (Martinelli et

al. 2000, Townsend et al. 2007).

Nitrogen stable isotopic composition of leaves and soil is another measurement

used to indicate N sources, sinks, and losses in a system (reviewed by Handley and

Raven 1992, Högberg 1997). For example, if N is found in excess in a system, losses

as nitrate and trace gas fluxes can result in 15N enriched soils (Högberg 1997).

Because plants take up N from the soil and incorporate it into their tissues, foliage 15N

signatures can broadly trace soil signatures. Thus, forests with more open N cycles can

have soils and foliage with higher 15N signatures (Garten 1993, Martinelli et al. 1999).

At a global scale, as MAP increases 15N signatures for both soils and plants decrease

(Amundson et al. 2003). In the present study, I tested if patterns of soil and foliar

nutrient concentrations followed patterns of MAP regardless of co-varying factors such

as altitude and species composition. Specifically, I addressed the following questions:

(1) Do sites with higher precipitation have relatively lower soil nutrient concentrations?

(2) Do patterns of foliar nutrients track patterns of soil nutrients? (3) Do foliar nutrients

differ among genera or functional groups? and finally (4) Do other environmental factors

(e.g., altitude) interact with MAP to influence patterns of soil and foliar nutrients? I

predicted that sites with higher MAP would have lower soil nutrients (specifically N and

P). I expected greater differences in foliar nutrients (specifically N and P) among genera

within a site than in community-averaged foliar nutrients among sites. Within genera,

31

however, I expected that foliar nutrients would track measurements of soil nutrient

availability.

Methods

Study Site

In June and July 2006 I sampled nine forest sites in Costa Rica (Figure 2-1 and

Table 2-1) ranging from 3500 – 5500 mm MAP and from 100-1200 m in altitude.

Vegetation at all sites was mature tropical forest but physical and historical factors

varied, such as geology, topography, and degree of human intervention (Table 2-1). Our

study sites were: Universidad EARTH Forest Reserve (Earth), Instituto Tecnológico de

Costa Rica forest plots at Mogos (Mogos), forest fragment near La Palma (La Palma),

La Selva Biological Station (La Selva), Rancho Mastatal (Mastatal), private forest

reserve in Dos Brazos de Río Tigre de Osa (Dos Brazos), Alberto Jimenez Forest

Reserve in San Ramón (San Ramón), Parque Nacional Tapantí (Tapantí), and La

Gamba Reserve in Golfito (Golfito). In each site I walked trails until I found at least three

mature individuals from five common tree species that were likely to be found in all (or

most) study sites and that included at least one leguminous species. I chose replicate

trees from the same species that were separated from each other by at least 20 m. I

then collected foliage from these trees and soil samples from the surrounding area.

Soil Sampling and Analysis

At each site, I randomly collected five to eight 10 cm deep soil cores within areas

where foliage samples were collected. All soil cores were separated by at least 30 m.

When a visible organic horizon deeper than 5 cm was present, I analyzed it separately

and collected a 10cm -deep mineral core starting at the depth that the organic horizon

ended. I stored soils at 4°C for ≤ 48 hours before being extracted at the La Selva

32

Biological Station. Because of the remoteness of the Mogos and Golfito sites, these

soils were stored for five days at 4°C before we could conduct extractions, which could

potentially have decreased mineralization and nitrification values (Arnold et al. 2008).

However, from the two sites for which we had to store soils for more than 48hrs, Golfito

had not only low N mineralization and nitrification but also low foliar nutrients (which

were analyzed on dried foliage), suggesting that low N mineralization and nitrification

values are real and not an analytical artifact. Mogos had substantially higher

mineralization and nitrification values, and high foliar N. Once in the laboratory, I

homogenized the bulk soil samples by separating roots, rocks, and other coarse debris.

To measure NO3- and NH4

+ I extracted 10 g of fresh soil in 50 ml 2M KCl and the

solution was measured using an Astoria Pacific colorimetric autoanalyzer (Clackamas,

Oregon, USA). I calculated dissolved inorganic nitrogen (DIN) as the sum of NO3- and

NH4+

at this initial extraction. To obtain a measurement of potential mineralization, I

incubated 10 g of soil for eight days at field moisture and at room temperature

(approximately 25 °C); then I extracted and measured NO3- and NH4

+ as above. I

calculated net N mineralization from changes in the NO3- and NH4

+ concentrations

(Riley and Vitousek 1995, Robertson et al. 1999). I calculated net nitrification as the

change in NO3- concentration per gram dry soil mass divided by the time of incubation. I

measured soil pH and available P on air-dried soils that were forced with a rubber

stopper through a 2 mm sieve. Soil pH was measured using 2:1 water: soil ratio on an

electronic pH meter (Thermo Orion 250A+, Orion Research, Inc., Boston,

Massachusetts, USA).I extracted Soil P with a Melich I solution (Kuo 1996). The

resulting P in solution was read by colorimetric determination of ortho-phosphate

33

(Murphy and Riley’s (1962) method with modified volumes for analysis in a microplate

reader) using a spectrophotometer microplate reader (PowerWave XS Microplate

Reader, Bio-Tek Instruments, Inc., Winooski, Vermont, USA). I measured total percent

N, percent C, 15N and 13C on ground soils dried at 60 °C with an elemental analyzer

(ECS 4010, Costech Analytical, Valencia, California, USA) coupled with an isotope ratio

mass spectrometer (Delta Plus XL, ThermoFinnigan, Bremen, Germany).

Foliage Sampling and Analysis

I collected fully-expanded, sun-leaf samples, using a crossbow with a bolt affixed

with monofilament line, from five common tree genera occurring at most of our sites. I

collected samples from at least nine trees from the five most common evergreen

species at each site. I tried to maximize species– or at least genus– overlap among

sites, and included at least one abundant legume in each location (Table 2-2). I dried

the collected foliage at 60°C and ground resulting samples using a Wiley Mill (Thomas

Scientific, Swedesboro, NJ, USA) passed through a #40 screen for chemistry

measurements of foliage from each tree (samples were not bulked). I measured total

percent N, percent C, 15N and 13C of foliage with an elemental analyzer (ECS 4010,

Costech Analytical, Valencia, California, USA) coupled with an isotope ratio mass

spectrometer (Delta Plus XL, ThermoFinnigan, Bremen, Germany). I measured P in

foliage samples using an ash digestion (Jones and Case 1996) followed by colorimetric

determination of ortho-phosphate using a spectrophotometer microplate reader

(PowerWave XS Microplate Reader, Bio-Tek Instruments, Inc., Winooski, Vermont,

USA).

34

Data Analysis

I used linear regressions to describe relationships among variables and one-way

analysis of variance (ANOVA) to compare variables among sites. When ANOVAs were

significant I conducted Tukey tests to compare pairs of means. I tested distributions

obtained from the nine site averages using Shapiro-Wilk tests. In most cases

distributions were normal but when this was not the case, I ln-transformed data to

improve fit. If data still deviated from normality, I used non-parametric statistics. In all

figures, site means with respective standard errors are shown. For soil variables, each

site mean results from averaging 5-8 soil cores; for foliar variables, each site mean

results from averaging foliage from 9-23 trees (Table 2-2). I calculated regressions

using means from each site (Sample size = 9). Analyses were performed using JMP IN

5.0 (SAS Institute Inc., Cary, NC, U.S.A).

Results

Soil Characteristics

I observed differences among site means for several soil parameters. For some

parameters, however, within site variability was higher than among site variability (Table

2-3). Mean N concentrations for mineral soils, for example, ranged from 0.32 percent in

Dos Brazos to 0.87 percent in Tapantí but individual core measurements ranged from

0.26 percent to 1.42 percent in Mastatal alone. Average percent C in the mineral soil

ranged from 4.13 percent at La Palma to 11.8 percent at San Ramón. However, soils in

Tapantí (the highest altitude site), presented a distinct organic (O) horizon containing up

to 33.2 percent C (Table 2-3). This organic layer had higher nutrient levels than the

other mineral soils, and was thus not included in among-site comparisons. From the N

measurements, only DIN differed significantly among sites. Overall, net nitrification rates

35

were 100 percent or more of net mineralization rates, suggesting that most mineralized

N was converted rapidly to nitrate. All sites had very low (negative) net ammonification

rates and all sites had at least one soil core where net immobilization of N occurred

after incubation (data not shown). There was no relationship between mean soil N

concentrations and net mineralization or nitrification (r2 = 0.18, P = 0.26 for

mineralization and r2 = 0.01, P = 0.99 for nitrification), suggesting that differences in

these processes were not driven by the initial N pool size. In addition, there was a

positive relationship between mean soil C and N (r2 = 0.95, P < 0.01) but not between

soil C and P (r2 = 0.15, P = 0.30) or N and P (r2 = 0.06, P < 0.52).

Precipitation was a better predictor of soil N than soil P. There was a tendency

for net N mineralization to decrease with increasing MAP but only net nitrification

decreased significantly with increased MAP (Figure 2-2). In higher altitude sites, soils

had higher C (r2 = 0.49, P = 0.04) and N (r2 = 0.59, P = 0.02) concentrations but not

higher P (r2 = 0.01, P = 0.86) concentration.

Foliar Measurements

All foliar parameters measured varied significantly among sites (Table 2-4) but

were not correlated with MAP or altitude (data not shown). There was only a marginal

correlation between mean foliar percent C and N across sites (r2 = 0.42, P = 0.058).

When observing foliar nutrients by plant functional type, legumes had higher N:P ratios

than non-legumes because legumes had higher N concentrations than non-legumes (T

= 10.48, d.f. = 137, P < 0.001; Figure 2-3); legumes and non-legumes did not differ in

percent P concentrations (T = 1.08, d.f. = 135, P = 0.28). Foliar N:P ratios varied greatly

among genera. For the four most common genera present in the majority of the study

sites (Hyeronima, Inga, Protium, and Virola), there were significant N:P differences

36

among genera and across sites (two-way ANOVA for genera and site: Model F11, 75 =

11.99, P < 0.001; effect tests: genera F3 = 34.01, P < 0.001, site F8 = 3.04, P = 0.006;

Figure 2-3B), with Inga usually presenting the highest N:P ratios and Protium the

lowest.

Relationship Between Soil and Foliar Measurements

Soil Melich-extractable P was a better predictor of foliar N and P than soil N. Soil

Melich-extractable P was positively correlated with foliar P concentrations (Figure 2-4A)

and marginally correlated with foliar N concentrations; r2 = 0.34, P= 0.059), as well as

negatively correlated with foliar N:P (Figure 2-4B). Soil total N and DIN concentrations

were not correlated with foliar N concentrations (r2 = 0.14, P= 0.321 and r2 = 0.06, P =

0.543). Nitrogen fluxes (e.g., net mineralization and nitrification) were also not good

predictors of foliar N concentrations. Isotope δ15 N values, however, were well

correlated between soils and plants (r2 = 0.84, P < 0.001). Interestingly, both study sites

located in the Caribbean slope of the country (Earth and La Selva) had soils and plants

significantly more enriched in δ15N than the rest of the sites (Wilcoxon rank test: X2 =

4.2, d.f. = 1, P = 0.040; Tables 2-3 and 2-4). These two sites also had the smallest

relative offset between soil and foliar signatures (Earth = -0.57 and La Selva = -0.29).

Soil C concentration was not correlated with any foliar measurements (data not shown).

Discussion

Patterns of Soil Nutrient Availability

Across nine forest sites in Costa Rica, there was a decrease in net N

mineralization and nitrification rates with increased MAP, although the pattern was

much stronger for nitrification than for mineralization. Reductions in net nitrification,

specifically, could be due to the lysis of aerobic microorganisms under anaerobic

37

conditions (high rainfall) or by the activity of facultative anaerobes adapting to

fluctuating oxic and anoxic conditions (Davelaar 1993, Wright et al. 2001, Rinklebe and

Langer 2006). Decreased net N mineralization with high MAP has been observed in

other tropical forests (Chandler 1985, Schuur and Matson 2001) and has also been

attributed to anaerobic conditions. However, Pandey et al. (2009) in three land-use

systems in India found higher net N mineralization (but not net nitrification) during dry

spells of the wet season compared to net N mineralization during high precipitation

periods, highlighting different controls on these two biogeochemical processes.

Although I found a decrease in both net N mineralization and nitrification with increased

MAP, the higher variability in the net N mineralization patterns suggests that net N

mineralization rates could also be influenced by other factors such as litter quality

(Austin and Vitousek 2000, Schuur 2001, Santiago et al. 2005). In addition, soil total N

and DIN concentrations did not decrease with increasing MAP, suggesting that in my

study sites, excess water from high MAP may result in a decrease in the rate of N

cycling but not in N pool sizes.

In contrast to N availability, measurements of soil P did not decrease with

increasing MAP. Potentially in the wettest sites, anoxic conditions resulting from high

precipitation can cause low redox potentials in soils, which can release mineral-sorbed

P (Miller et al. 2001, Schuur and Matson 2001, Schuur et al. 2001). Phosphorus

released by this mechanism could offset slow organic matter decomposition and

mineralization, resulting in reduced available N but not P at the wettest sites.

Alternatively, we may have failed to detect a correlation between P and MAP in this

study because we did not measure the P fractions that are most susceptible to changes

38

in MAP. For example, in a study in a precipitation gradient in Hawaii only the most

soluble P fraction, “resin-extractable P”, decreased with increasing MAP (Idol et al.

2007). In another Hawaii study, however, Miller et al. (2001) reported an increase of

labile P (defined as resin extractable P + bicarbonate extractable P) and a decline of

recalcitrant inorganic P with increasing MAP. These examples demonstrate that

different P forms may be influenced differently by MAP, making it challenging to detect

patterns. In my study, although I did not measure multiple P fractions, the strong

positive correlation between Melich-extractable P and foliar P suggests that we

captured a significant part of the “plant available” P with my analysis. Thus, I conclude

that at least this P fraction is not correlated with MAP in my gradient, although the

mechanisms for this result are unclear.

Percent C in the soil was also not correlated with MAP either. Most likely, the

lack of correlation is caused by the interaction of multiple environmental factors, such as

variation in altitude (Tanner et al. 1998), parent material (Jenny 1941), and topography

(Porder et al. 2006). For example, there was an interaction between elevation and

precipitation that caused a dramatic effect on soils in Tapantí. In this site, located at

1200 m elevation and with a MAP of 5000 mm, soils exhibited a definite organic horizon

with significantly higher C (and N and P) than the rest of the sites (Table 2-3). Thus,

Tapantí had the highest soil percent carbon and nutrient concentrations due to this

organic horizon. The distinct organic horizon in Tapantí soils is probably the result of

slow decomposition of organic matter caused by a combination of lower temperatures,

higher elevation, and lower litter quality (Tanner et al. 1998, Schuur 2001, McGroddy

and Silver 2004, Cusack et al. 2009). Surprisingly the other two high elevation sites –

39

San Ramón (elevation 1000 m; MAP 4500mm) and Mastatal (elevation 900 m; MAP

4000mm) – did not show a distinct organic horizon and thus had significantly lower soil

nutrients and C than Tapantí. From the data presented here, it is difficult to determine

the cause of this difference. I suspect, however, that a combination of higher

temperature (due to slightly lower elevation), lower MAP, and potentially higher litter

quality due to different species composition could impede the formation of an organic

horizon in San Ramón and Mastatal.

Pattern of Foliar Nutrients

Consistent with other studies, foliar P was a good predictor of soil P (e.g., Schuur

and Matson 2001, Santiago et al. 2005, Wu et al. 2007). This implies that plants are

incorporating in their tissues as much P as is available at each site, and supports the

idea that foliar concentrations are good predictors of soil nutrient availability (Vitousek

and Farrington 1997, Aerts and Chapin 2000, Vitousek 2004); moreover, it suggests

that across the study sites P may be more limiting to plant growth than N. Tapantí,

however, did not have higher foliar nutrients than the rest of the sites, even though it

had the highest concentrations of soil extractable P and N, when considering the

organic horizon. There are several possible explanations for this result. One possibility

is that high soil nutrient concentrations in Tapantí do not result in higher nutrient pools

than the rest of the sites due to relatively lower bulk density. However, we did not

measure bulk density and therefore we cannot directly test this hypothesis.

Alternatively, Tapantí may in fact have the largest soil nutrient pools but other factors,

such as floristic composition in Tapantí, may be characterized by species with inherent

low foliar nutrient concentrations (area based) as is typical in other pre-montane and

40

montane forests (Tanner et al. 1998; Table 2-2), thus not reflecting soil nutrient

availability.

In contrast to foliar P, foliar N was not a good predictor of soil N (or soil P),

suggesting that factors other than soil N availability are influencing foliar N

concentrations across the plant community. Indeed, I found higher foliar N (but not P) in

leguminous vs. non-leguminous species, with legume abundance varying among sites

(personal observation). Thus in a given ecosystem, a higher proportion of legumes can

result in a higher community-wide foliar N (regardless of MAP), which highlights the

importance that species (or functional group) composition can have for a site’s

biogeochemistry (Hooper and Vitousek 1998, Townsend et al. 2007). High foliar

concentrations of N in leguminous species have been observed in other studies (e.g.,

Martinelli et al. 2000, Townsend et al. 2007) and are typical for nodulating and non-

nodulating legumes (McKey 1994). In contrast to foliar N, foliar δ 15N was well

correlated with soil δ 15N (and with foliar %N). The two Caribbean sites –La Selva and

Earth– had soils and foliage most enriched in δ 15N. One possible explanation for this

pattern is that these two sites could have a more open N cycle, with more N losses and

thus a more positive δ 15N signature (Martinelli et al. 1999). In addition, more enriched

δ 15N signatures could reflect differences in parent material or N deposition rates

(Eklund et al. 1997). Further investigations are required to obtain a conclusive

mechanism for the observed pattern.

Across sites, foliar N:P ratios above 16 (Figure 2-3) support the general view that

P limits plant productivity in tropical forests (Vitousek 1984, Hedin et al. 2003, Reich and

Oleksyn 2004). However, when observing individual genera within sites, sizable

41

differences in N:P ratios suggest that the community averaged, foliar N:P-ratio

thresholds used in temperate systems to infer nutrient limitation may not be applicable

to tropical forests. In my study, variability among genera and functional type (legumes

vs. non-legumes) reflects the high diversity of life history traits and physiological

strategies in tropical forests. Moreover, it stresses that even though individual-taxa foliar

nutrient concentrations can provide useful information regarding multiple ecosystem

processes, community averaged N:P ratios do not always reflect nutrient limitation. In

fact, physiological properties and life history strategies may be more important than

environmental limitations in determining foliar nutrient concentrations (Ågren 2004,

Niklas et al. 2005). To confidently draw conclusions concerning the nutritional status of

a plant community based on foliar nutrient ratios, these values should be calibrated by

conducting a direct test of nutrient limitation, such as a nutrient addition experiment

(Chapin et al. 1986, Koerselman and Meuleman 1996).

Conclusions

The results from this study suggest that soil N dynamics may be more sensitive to

increasing precipitation than soil P dynamics in these tropical wet forests in Costa Rica.

However, the lack of relationship between soil N dynamics and foliar N makes the

consequences of this sensitivity uncertain. On the contrary, although factors other than

MAP may control soil P in these forests, the positive relationship between soil and foliar

P suggests that plant foliar nutrients are more responsive to changes in soil P than in N.

Large differences in foliar nutrient concentrations and N:P ratios within sites stress the

key role that plant functional group composition plays in tropical ecosystem

biogeochemistry. For example within a site, a decrease in N availability could be

irrelevant (or even beneficial) for N-fixing legumes, but detrimental for non-legume

42

species with high N metabolic requirements, such as Virola sp. that has high alkaloid

concentrations (Dominy et al. 2003). Thus, my study highlights the importance of

considering species composition and identity before making community-wide

generalizations regarding nutrient limitation in tropical forests. Furthermore, interactions

among multiple environmental and biological factors, such as altitude, temperature,

redox potential, and decomposability of soil organic matter, can influence the sensitivity

of N and P dynamics to changes in MAP in each site, making predictions extremely

difficult and inferences regarding nutrient limitation uncertain.

With climate change, several models predict a decrease in MAP for certain areas

in Costa Rica (Neelin et al. 2006). My results suggest that this decrease could lead to

an increase in available N from mineralization in the wetter sites of my gradient. More

investigation is required to reach conclusive results regarding the effect of MAP on soil

nutrient availability and the relationship between soil and foliar nutrients in these forests.

Key measurements conducted at the specific sites should include (1) meteorological

measurements, including temperature, (2) detailed soil characterization including bulk

density, cation exchange capacity, and texture, (3) investigation of parent material, and

finally (4) direct tests of nutrient limitation. Nevertheless, my study provides important

baseline biogeochemical data that can be used to improve understanding of nutrient

cycling and limitation in tropical forests, and to encourage future studies that identify the

possible consequences that alterations in MAP resulting from climate change will have

in tropical systems.

43

Table 2-1. Characteristics of study sites in Costa Rica. Altitude = meters above sea level, MAP = mean annual precipitation, Dry months = average number of months with less than 100 mm precipitation. Data from Proyecto Atlas Digital de Costa Rica 2008.

a Soil order from Proyecto Atlas Digital de Costa Rica 2008 and Sub-group from Pérez et al. (1978)

b Recently reclassified as Oxisols (Kleber et al. 2007)

Site Location Altitude

(m) MAP ( mm)

Dry months

Parent material Soil order and

Sub-groupa

Earth 10° 10’ 21.1” N 83° 36’ 9.8” W

100 3500 1 Alluvial and colluvial deposits and distal facles of modern volcanic rocks

Inceptisols and Ultisols (Oxic palehumult)

Mogos 8° 45’ 8.6” N 83° 23’ 7.1”W

100 3500 1 Sea floor basalts Ultisols (Typic tropohumult)

La Palma 8° 39’ 5.7” N 83° 23’ 36.9” W

100 4000 1 Alluvial and colluvial deposits Entisols or Mollisols (Fluvaquentic hapludoll)

La Selva 10° 26’ 24.8” N 84° 0’ 56.0” W

100 4000 1 Alluvial and colluvial deposits and distal facles of modern volcanic rocks

Inceptisols and Ultisols b

(Oxic dystropept)

Mastatal 9° 40’ 52.1” N 84° 22’ 34.0” W

900 4000 3 Deep water sedimentary rocks Inceptisols and Ultisols (Typic Dystropept)

Dos Brazos

8° 31’ 59.3” N 83° 23’ 30.3” W

200 4500 1 Deep water sedimentary rocks Ultisols (Typic tropohumult)

San Ramón

10° 9’ 4.8” N 84° 29’ 15.2” W

1000 4500 3 Volcanic intrusive rocks from the Tertiary Inceptisols (Typic dystrandept)

Tapantí 9° 45’ 59.1” N 83° 48’ 9.9” W

1200 5000 2 Volcanic intrusive rocks from the Tertiary and alluvial and colluvial deposits

Inceptisols (Andic humitropept)

Golfito 8° 38’ 39.8” N 83° 9’ 54.8” W

200 5500 1 Sea floor basalts and deep water sedimentary rocks

Ultisols (Typic tropohumult)

44

Table 2-2. Species and number of samples collected per site in nine sites in Costa Rica. See methods section for site description. DB=Dos Brazos, EAR=Earth, GOL= Golfito, LP=La Palma, LS=La Selva, MOG=Mogos, MAS= Mastatal, SR= San Ramón, TAP=Tapantí

Species DB EAR GOL LP LS MOG MAS SR TAP Total

Alchornea latifolia

2 3 5

Ardisia sp.

1

1

Billia hippocastanum

1 1

Billia rosea

1

1

Brunellia standleyana

3 3 6

Citharexylum caudatum

3 3

Dendropanax arboreus 3 3

2

3

11

Elaegia auriculata

3

3

Hasseltia quinquenervia

1

1

Hyeronima alchorneoides 3 2

3

3

11

Hyeronima oblonga

1

2

3

Inga acrocephala 1

1

Inga barbourii

3

3

Inga nobilis

1

1

Inga oerstediana 2

1 3

Inga pezizifera

1

1

Inga polita

1

1

Inga sp.

1 1

Inga sp. Leonis

1

1

Inga sp. tonduzii

1 1

Inga thibaudiana

3

1

4

Inga venusta

2

2

Inga vera

3

3

Laetia procera

4

4

45

Table 2-1. Continued

Species DB EAR GOL LP LS MOG MAS SR TAP Total

Marila pluricostata

2

2

Micropholis meloniana

2

2

Pentaclethra macroloba

1

1

2

Protium aracouchini

3

3

Protium confusum

1

1

Protium glabrum 3

1

1

5

Protium panamense

4

4

Protium pittieri

3

1

4

Protium ravenii

2

2

Rollinia pittieri

1

1

Saurauria montana

2 2

Saurauria rubiformis

1 1

Virola guatemalensis 3

3

2

8

Virola koschnyi

3 1 1 1

4

10

Virola multiflora

1

1

Virola sebifera

1

1

2

Virola surimamensis

1

1

Vochysia allenii

2

2

Vochysia ferruginea

2 2

4

8

Warzewiczia coccinea

1

3

4

Total number of samples 15 16 12 16 9 11 23 19 16 137

46

Table 2-3. Soil nutrients and isotopic signatures (mean + SE) for the nine study sites in Costa Rica shown in order of increasing mean annual precipitation (Table 2-1). Soils were sampled 10 cm deep in the mineral layer, except for Tapantí, where a distinct organic layer was present and thus organic (O) and mineral (M) horizons were separated. Soil pH was measured in water; DIN = dissolved inorganic nitrogen (nitrate + ammonium), Net min = net mineralization, and Net nit = net nitrification. Shown are also F tests from one-way ANOVA comparing sites (N = 9) with significant values bolded (* = P < 0.05, ** = P< 0.001).For significant ANOVAs, means that do not share a common letter superscript are significantly different. Organic soils were excluded from the analysis.

Site pH %C

δ13C Melich P (µg g-1)

%N

EARTH 4.77 + 0.20bc 5.17 + 0.98ab -28.36 + 0.56 1.24 + 0.34a 0.48 + 0.08

Mogos 4.86 + 0.08bc 4.92 + 0.42ab -28.01. + 0.21 0.28 + 0.04a 0.36 + 0.04

La Palma 5.32 + 0.26b 4.13 + 0.66b -27.56 + 0.13 0.42 + 0.14a 0.38 + 0.07

La Selva 4.16 + 0.04c 5.42 + 0.63ab -27.72 + 0.28 0.89 + 0.19a 0.46 + 0.04

Mastatal 5.12 + 0.17b 5.57 + 1.45ab -28.11 + 0.72 0.80 + 0.21a 0.52 + 0.13

Dos Brazos 6.25 + 0.06a 3.42 + 0.48b -28.28 + 0.29 1.51+ 0.80a 0.32 + 0.05

San Ramón 5.15 + 0.06b 11.88 + 1.74a -28.15 + 0.08 0.19 + 0.03a 0.87 + 0.14

Tapantí M 4.70 + 0.08bc 7.96 + 2.34ab -28.64 + 0.81 0.90 + 0.32a 0.63 + 0.17

Tapantí O 3.93 + 0.05 33.23 + 4.94 -29.32 + 0.42 25.67 + 11.17 2.21 + 0.34

Golfito 5.06 + 0.11b 4.65 + 0.49ab -28.44 + 0.19 0.28 + 0.03a 0.37 + 0.03

F values 12.33** 2.48* 0.29 2.44* 1.58

47


Site δ15N DIN (µg g-1)

Net N min (µg N g-1 d-1)

Net N nit (µg N g-1 d-1)

EARTH 5.01 + 0.48ab 11.92 + 0.57 3.25 + 1.06 3.77 + 1.09

Mogos 2.77 + 0.49c 14.15 + 0.69 3.74 + 1.13 4.48 + 1.68

La Palma 2.99 + 0.23bc 11.49 + 0.33 0.57 + 0.41 0.78 + 0.58

La Selva 5.53 + 0.32a 12.68 + 0.47 2.50 + 0.81 2.72 + .77

Mastatal 2.23. + 0.38c 11.42 + 0.70 3.15 + 1.90 3.12 + 1.39

Dos Brazos 2.27 + 0.56c 11.94 + 1.30 0.87 + 0.3 1.94 + 0.62

San Ramón 3.19 + 0.30abc 14.42 + 0.70 1.53 + 0.65 3.64 + 1.16

Tapantí M 2.73 + 0.64bc 13.03 + 0.64 3.00 + 1.86 1.18 + 1.67

Tapantí O 0.88 + 0.45 40.77 + 11.51 15.50 + 7.24 10.03 + 1.93

Golfito 2.12 + 0.17c 12.92 + 0.40 0.73 + 0.55 1.05 + 0.16

F values 7.72** 2.09 0.71 0.92

48

Table 2-4. Foliar carbon, phosphorus, nitrogen and stable isotopic signatures (mean + SE) for the nine study sites in Costa Rica. For complete list of collected specimens refer to Table 2-2. Shown are also F tests from one-way ANOVA comparing sites (N = 9); significant values are bolded (* = P < 0.05, ** = P < 0.001).Within a column, means that do not share a common letter superscript are significantly different

Site %C δ13C %P %N δ15N

Earth 48.65 + 0.49a -31.92 + 0.25bc 0.13 + 0.01a 2.27 + 0.12ab 2.16 + 0.19b

Mogos 45.37 + 0.79bc -32.72 + 0.33c 0.09 + 0.01bc 1.88 + 0.14ab -0.75 + 0.26cd

La Palma 46.72 + 0.59abc -31.17 + 0.4bc 0.09 + 0.01bc 2.08 + 0.13ab -1.51 + 0.34de

La Selva 49.44 + 0.32a -30.41 + 0.39ab 0.12 + 0.01ab 2.5 + 0.15a 3.91 + 0.37a

Mastatal 46.81 + 0.59abc -31.66 + 0.45bc 0.11 + 0.01ab 2.27 + 0.09ab -1.42 + 0.29de

Dos Brazos 45.75 + 0.58bc -31.81 + 0.34bc 0.14 + 0.01a 2.41 + 0.13a 0.16 + 0.27c

San Ramon 45.39 + 0.4bc -31.36 + 0.33bc 0.12 + 0.01ab 2.3 + 0.18ab -0.82 + 0.38cd

Tapantí 47.63 + 0.46ab -29.57 + 0.24a 0.14 + 0.01a 2.32 + 0.16ab -0.46 + 0.40cd

Golfito 44.83 + 0.90c -31.60 + 0.32bc 0.07 + 0.01c 1.67 + 0.11b -2.60 + 0.17e

F values 6.01** 6.05** 9.78** 2.85* 28.34**

49

Figure 2-1. Map of sites where soil and foliar samples were collected in Costa Rica.

Mean annual precipitation is shown in shades of gray. Precipitation data obtained from Proyecto Atlas Digital Costa Rica (2008).

50

Ne

t N

Min

era

liza

tio

n (

g N

/g/d

)

0

1

2

3

4

5

6

MAP (m)

3.0 3.5 4.0 4.5 5.0 5.5 6.0

So

il P

(m

g/k

g)

0.0

0.5

1.0

1.5

2.0

2.5

3.0Earth

Mogos

La Palma

La Selva

Mastatal

Dos Brazos

San Ramon

Tapanti

Golfito

R2 = 0.07 P = 0.452

A

B

R2 = 0.31 P = 0.095

C

R2 = 0.63 P = 0.010

+ P

recip

ita

tio

n

Net

Nitrification (g N

/g/d

)

0

2

4

6

8

B R2 = 0.63 P = 0.010

Figure 2-2. Relationship between mean annual precipitation (+ SE) and soil

parameters, including (A) average net nitrogen mineralization rates, (B) net nitrification rates, and (C) soil Melich I phosphorus for the top 10 cm mineral layer of nine study sites in Costa Rica. Circles represent lowland sites (< 900 m) and triangles are mid-elevation sites (900-1200 m).

51

N:P

0

10

20

30

40

50

60

Legumes

Non-legumes

F1,135 = 84.9, P < 0.001 A

Earth

Mog

os

La P

alm

a

La S

elva

Mas

tata

l

Dos

Bra

zos

San R

amon

Tapan

ti

Golfito

N:P

0

10

20

30

40

50

60

Hyeronima (2 sp.)

Inga (13 sp.)

Protium (6 sp.)

Virola (5 sp.)

B Genera: F3 = 34.01, P < 0.001

Site: F8 = 3.04, P= 0.006

- Precipitation +

Figure 2-3. Mean N:P ratios (+ SE) for (A) legumes and non-legumes and (B) for the

most common genera (including the leguminous genus Inga) in nine study sites in Costa Rica. The number of species collected for each genus is shown in the legend. Dashed line represents the suggested threshold for N vs. P limitation suggested in the literature (Aerts and Chapin 2000, Güsewell 2004, Reich and Oleskyn 2004). For details of which species were collected at each site refer to Table 2-4. Sites are arranged by increasing MAP.

52

Folia

r %

P

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

Soil P (ug/g)

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Folia

r N

:P

14

16

18

20

22

24

26

28

A

B

R2 = 0.58 P = 0.017

R2 = 0.44 P = 0.051

Pre

cip

itatio

n +

Earth

Mogos

La Palma

La Selva

Mastatal

Dos Brazos

San Ramon

Tapanti

Golfito

Figure 2-4. Mean soil phosphorus (+ SE) for the top 10 cm mineral layer of nine study sites in Costa Rica plotted against (A) mean (+ SE) foliar P and (B) mean (+ SE) foliar N:P ratios. Circles represent lowland sites (< 900 m) and triangles are mid-elevation sites (900-1200 m).

53

CHAPTER 3 DIRECT TEST OF NUTRIENT LIMITATION TO NET PRIMARY PRODUCTIVITY IN A

LOWLAND TROPICAL WET FOREST

Introduction

Nutrient availability controls key processes in all ecosystems on earth. Net

primary productivity (NPP), nutrient use efficiency (NUE) by plants, and nutrient

turnover through decomposition, are all processes affected by nutrient availability.

Nitrogen (N) and phosphorus (P), either individually or in combination, limit primary

productivity in most terrestrial ecosystems (Vitousek and Howarth 1991, Elser et al.

2007, Vitousek et al. 2010). In turn, plant adaptations to limitation by these nutrients

feed back strongly to control ecosystem rates of nutrient cycling (Chapin 1980, Hobbie

1992, Vitousek 2004). Most tropical forests occur on old soils, where P-rich parent

material has been weathered and most of the remaining P is occluded on iron and

aluminum oxides (Sanchez 1976, Miller et al. 2001). Nitrogen, by contrast, accumulates

over time through biological fixation, and is therefore expected to be relatively more

available than P in old, weathered soils. Thus, it is generally believed that NPP is limited

by P in these systems (Walker and Syers 1976, Vitousek 1984, Vitousek and Farrington

1997).

However, multiple abiotic and biotic factors can influence nutrient availability, so

that not all tropical forests occurring in old soils are limited by P. In montane tropical

forests, for example, erosion rates caused by topographic variation are sufficient to

provide renewal of solum P from weathering parent material. Thus, NPP in these forests

is generally limited by N, instead of P (reviewed by Tanner et al. 1998, Porder et al.

2006). Other factors, such as precipitation (Schuur 2003, Santiago et al. 2005, Alvarez-

Clare and Mack 2011), disturbance (Davidson et al. 2004), and life-form diversity

54

(Hiremath and Ewel 2001) can also influence nutrient availability in tropical forests,

resulting in limitation by nutrients other than P. In fact, data from the few fertilization

experiments conducted in the tropics, revealed that N limitation is pervasive among

tropical forests (Le Bauer and Treseder 2008), although this sample was dominated by

tropical montane forests.

Historically, our conceptual understanding of nutrient limitation was one derived

from Liebig’s law of the minimum, where the single scarcest nutrient in relation to plant

demand, usually N or P, was the most limiting (Liebig 1842). However, recent

investigations showing synergistic interactions between limited supplies of N and P are

widespread across aquatic and terrestrial systems (Elser et al. 2007, Harpole et al.

2011), and evidence of limitation by other nutrients, such as potassium, is emerging

(e.g., Kaspari et al. 2008, Wright et al. 2011). Consistent with these findings, nutrient

limitation in tropical forests is probably not an N versus P question, but instead one that

includes complex interactions among nutrient cycles, and their linkages with biological

processes. For example, N inputs have been shown to accelerate phosphorus cycling

rates through enhancement of soil and root phosphatase activity (Marklein and Houlton

2011), and tree species composition is known to influence rates of nutrient turnover

through decomposition (Cornwell et al. 2008, Wieder et al. 2009).

The influence of biological processes on nutrient cycling is expected to be of

particular importance in lowland tropical forests, where there is a high diversity of flora

(Losos et al. 2004). Here, a wide variety of tree functional traits related to nutrient

acquisition and use is likely to influence biogeochemical processes, creating complex

linkages with nutrient cycles (Townsend et al. 2008). For example, large interspecific

55

differences in foliar nutrient concentrations and resulting litter quality can influence

plant-soil microbial feedbacks related to nutrient use (Vitousek 2004, Wood et al. 2011).

In a lowland tropical forest in the south of Costa Rica, tree species variation in foliar P

and carbon (C) chemistry were associated with tree-specific differences in both free-

living N fixation and soil respiration (Reed et al. 2007, 2008, Wieder et al. 2008) In

addition, differences among species in the functional use of resources (e.g., nitrogen

fixers vs. non-fixers, shade tolerant vs. light demanding, species with superficial roots

vs. species with deep roots) can enable them to mitigate nutrient limitation by accessing

different nutrient pools (Vance 2003).Therefore, to understand nutrient limitation in

lowland tropical forests it is imperative to consider the influence of species-specific

effects on nutrient cycling.

Another understudied aspect of nutrient limitation in tropical forests is the

relationship between different demographic groups and nutrient cycling. In

monodominant forests (e.g., Hawaii, summarized by Vitousek 2004)), population growth

rates (λ) will only be determined by the vital rates of a single tree species and thus

changes in nutrient availability will not have consequences for community composition

and biodiversity. By contrast in diverse tropical forests, alterations in vital rates of

different species due to changes in nutrient availability can lead to shifts in community

composition (Ceccon et al. 2004), which in turn affect plant-soil-microbial feedbacks.

Additionally, there may be resource partitioning in space and time among age groups

and different age groups may be limited by different nutrients. For example adult trees,

which intercept the majority of light reaching the canopy and possess greater root area,

may assimilate more nutrients than light-limited saplings and seedlings (Lambers et al.

56

1998). Seedlings in the understory may access nutrients added in fertilizer, but may be

unable to incorporate them in their tissues due to light co-limitation (Burslem et al.

1995).

To date, only two fertilization experiments have been conducted in lowland tropical

forests to directly test nutrient limitation of NPP. One found evidence for N and P co-

limitation (Mirmanto et al. 1999), and one found evidence of limitation by N, P, and K

(Wright et al. 2011). These results differ from the traditional view that lowland tropical

forests are P limited. Moreover, the study by Wright et al. (2011) reported that in the

lowland tropical forest in Panama where their study was conducted, different tree age

groups and different fractions of NPP were limited by different nutrients. None of these

studies explored the effects of tree species composition and identity on the observed

responses to nutrient additions. The scarcity of fertilization experiments in lowland

diverse tropical forests, the conflicting results obtained from these few existing studies,

and the critical role that these forests have on the global C cycle motivated my

research. I conducted a factorial fertilization experiment with N and P additions, to

directly test nutrient limitation in a lowland tropical wet forest in Costa Rica.

My first objective was to test if N and P limited various components of NPP.

Because theory (Walker and Syers 1976) and indirect evidence from stoichiometry

(Hedin 2004, McGroddy et al. 2004, Reich and Oleksyn 2004) and soil microbial

dynamics (Cleveland and Townsend 2006) suggest that lowland tropical forests in

nutrient poor, clayey soils are P-limited, I hypothesized that P fertilization would

increase productivity by enhancing stem diameter growth and fine litterfall production.

Considering that fine litterfall has responded faster to fertilization than wood increment

57

in other studies (Mirmanto et al. 1999, Wood et al. 2009 but see Tanner et al. 1992),

and that litterfall constitutes the major component of productivity in tropical lowland

forests (Clark et al. 2001a), I expected NPP to increase mainly due to litterfall

production.

My second objective was to investigate the effect of fertilization on stem diameter

increase for trees from different size classes. Consistent with Wright et al. 2011, I

expected different age groups to respond differently to nutrient additions. Large trees (>

10 cm diameter at breast height (DBH)) usually have slow growth rates (Lieberman and

Lieberman 1994) or may utilize extra nutrients for reproduction instead of growth, and

small trees (5-10 cm DBH) are usually light limited (Lambers et al. 1998). Thus, I

predicted that intermediate sized trees (10-30cm DBH)) would have the largest

response to fertilization (mainly with P) because these trees have moderate access to

light but must reach the canopy to obtain full sun. These trees, therefore, would benefit

from investing extra nutrients to stem growth. Small trees and seedlings in the

understory may access nutrients added in fertilizer, but may be unable to incorporate

them in their tissues as biomass due to light co-limitation (Burslem et al. 1995). I

hypothesized no difference in stem diameter increase for small trees and no difference

in stem length or number of leaves in seedlings with increased nutrient availability.

My third objective was to explore the effect of fertilization on stem diameter

increase for trees from different taxa. In my experiment, I used four species and three

genera that were present in all treatments and that encompassed a variety of life history

traits to explore how fertilization affected diameter increase in different taxa (Table 3-1).

Because fast-growing, light-demanding species acclimate faster to increases in light or

58

nutrient regimes (Lambers et al. 1998), I predicted that fast-growing canopy species

(e.g., Pentaclethra macroloba and Goethalsia meiantha) would demonstrate greater

diameter increases after fertilization (especially with P) than slow-growing, shade-

tolerant trees (e.g., trees from the genus Virola). In addition, I predicted that subcanopy

trees with no access to appreciable light (e.g. Dendropanax arboreus or trees from the

genus Protium) would not demonstrate differences in diameter increase.

Methods

Site Description

The study was conducted at the Forest Reserve of the EARTH University (Escuela

de Agricultura de la Región del Trópico Húmedo), in Guácimo, Limón, Costa Rica (10°

11’ N and 84° 40’ W). This private reserve is located approximately 30 m above sea

level and consists of 900 ha of mature and regenerating wet forest and wetlands. Mean

annual temperature is 25.1 °C and mean annual precipitation (MAP) is 3,464 mm,

distributed in a bimodal pattern with peaks traditionally occurring in July-August and

November-December; March is the driest month with a MAP of 124 mm. Relative

humidity is above 80% during the year. An early ecological map (Tosi 1969) classifies

the EARTH forest within two life zones (sensu Holdridge 1971): tropical wet forest and

tropical moist forest with transition to wet. Throughout the duration of my study, MAP

averaged 3,844 mm ranging from 3,580 mm in 2007 to 4,097 in 2009. In 2009,

precipitation was unusually high during February and March (Figure 3-1). Mean

temperature between 2007 and 2009 was 24.8 °C, ranging from 23.2 °C in 2009 to 25.1

°C in 2008.

According to Sancho et al. (1990), the study area is located in the distal section

of a coalescence of alluvial fans. Parent material is from volcanic origin, mainly lava

59

flows and mud flows (lahars), boulders and pyroclastic materials. The undulating

topography is characterized with few eroded slopes and depressions where the eroded

material has been deposited. Soils in the area are mainly described in two units: 1)

Typic Dystrandepts, which are deep soils from volcanic origin, with low base saturation

and that remain humid throughout the year, or 2) Oxic Palehumult and Aeric

Tropaquept, described as deep, reddish soils that occur on old alluvial terraces and

suffer poor drainage. Detailed soil information exists for the rest of the EARTH property

but not for the forest reserve (Sancho et al. 1990).

Soil texture was analyzed in two ways. Texture analyzed in the traditional way with

a dispersant agent (Bouyoucous 1950) revealed that soils at the ERATH Forest

Reserve are clayey, with approximately 50% clay, 20% silt and 30% sand (Alvarez-

Clare, unpublished data). Results obtained with a Masterizer Particle Size Analyzer

revealed that most of that “clay” is actually very fine or fine silt with particle sizes

between 2-15.6 µm (Table 3-1). Overall, both total nitrogen (N) and phosphorus (P)

measurements are relatively high (Table 3-2), as well as most total micronutrients

(Table 3-3), although I did not conduct analyses of extractable ions for elements other

than P, which are probably low (Sancho et al. 1990).

The EARTH forest reserve is comprised of mostly secondary forest and disturbed

primary forest, which has not been altered in the past 25 yrs, since EARTH bought the

property and it became a reserve. Mean tree density is 390 trees with >10 cm diameter

at breast height (1.37 m) per hectare, with a basal area range of 15-25 m2 ha-1 (R.

Russo, unpublished data). In my study plots, which did not include swampy or riparian

areas, I found a higher density of trees (542 trees ha-1) and basal area (34.43 m2 ha-1).

60

Tree density is similar to that found at La Selva Biological Station, a well-studied wet

forest in Costa Rica (446 trees ha-1; Lieberman and Lieberman 1994). Although there

was variation in basal area among plots in my study site (Figure 3-2A), there were no

differences among blocks (F5, 23 = 0.64, P = 0.67) or among assigned treatments (F3, 23 =

1.72, P = 0.20), when the experiment was established.

Limited information exists regarding the floristic composition of the EARTH forest

reserve but within my study plots (total of 0.96 ha) I identified 104 tree species

comprising 82 genera and 46 families. Similar than other forests in the area (Hartshorn

and Hammel 1994), the forest at EARTH is dominated by the legume tree Pentaclethra

macroloba. This species contributed almost 30% of the total basal area within my study

plots (Figure 3-2B). High palm density is also characteristic of forests in the Caribbean

lowlands of Costa Rica (Hartshorn 1983, Hartshorn and Hammel 1994). At EARTH

forest, the second most important species was Socratea exohrriza, a canopy or

subcanopy palm with large leaves and distinctive stilt roots. In contrast with

Pentaclethra, however, this palm was important because of the high frequency and

density in which it occurs at EARTH, and not because of a large basal area (Figure 3-

2B).

Another inherent trait of these forests is their high dynamism. Canopy gaps are a

major source of environmental heterogeneity both in the canopy and in the forest floor

(Denslow and Hartshorn 1994). I characterized the variation in canopy openness among

plots using hemispherical photographs (Appendix A).

61

Experimental Design

The study plots were located in two flat, relatively high, non-flooding, mature-

forest areas (termed “Rancho” and “Rio”) within the forest reserve and were separated

by less than 1 km (Figure 3-3).In May 2007, I established 24 30 x 30 m plots and

assigned them randomly to three fertilizer treatments or a control in a complete block

design (n = 6). Three blocks were located at the Rancho site and three at the Rio site

(Figure 3-3). Though fertilization treatments are ongoing, data included here were

collected from August 2007 to March 2010. Besides the control plots, the three

treatments included +P (47 kg ha-1yr-1 of P as super triple phosphate), +N (100 kg ha-

1yr-1 of N applied as ammonium nitrate and urea), and +NP (N and P added together in

quantities as in +N and +P plots). Fertilizer was broadcast by hand twice a year on the

surface of the 900 m2 plots. All measurements were restricted only to the central 400 m2

of each plot (20 x 20 m) to reduce edge effects. The amount of fertilizer added is

consistent with similar studies conducted in montane (Harrington et al. 2001, Tanner

and Kapos 1992), secondary (Davidson et al. 2004), and lowland seasonal (Wright et al.

2011) tropical forests. In my experiment, all plots were separated by at least 50 m and

were selected to avoid leaf-cutter ant (Atta) nests, and extremely large trees that would

dominate an entire plot. To minimize the variability due to high tree species diversity,

each plot included at least one Pentaclethra macroloba tree and one Socratea exhorriza

palm, two of the most common species in this forest (Figure 3-2).

Soil Measurements

I collected soils before applying treatments and 1 and 2 years after the onset of

fertilization. At each collection time, soils were cored at depths of 0-10, 10-30 and 30-50

cm at randomly preselected, varying sites in each of four quadrants in a plot and then

62

combined into one composite sample per depth for each plot. I collected soil samples

using an open soil corer and kept them in a cooler for less than 6 hrs until they were

brought to the lab and stored in a 4°C cold room. All samples were homogenized by

separating roots, rocks, and other coarse debris, within 72 hrs of collection.

Homogenized soils were sub-sampled and extracted for chemical analysis.

Nitrate (NO3-), ammonium (NH4

+), dissolved inorganic nitrogen (DIN), net

mineralization, and net nitrification were measured on fresh soils by extracting 10 g of

soil in 50 ml 2M KCl at EARTH, freezing the extracts , and then conducting colorimetric

readings in an Astoria Pacific colorimetric autoanalyzer (Clackamas, Oregon, USA) at

the University of Florida. I calculated dissolved inorganic nitrogen (DIN) as the sum of

NO3- and NH4

+ at this initial extraction. To obtain a measurement of potential

mineralization, I incubated 10 g of soil for eight days at field moisture and at room

temperature (approximately 25 °C), then extracted and measured NO3- and NH4

+ as

above. I calculated net N mineralization from changes in the NO3- and NH4

+

concentrations (Riley and Vitousek 1995, Robertson et al. 1999) and net nitrification as

the change in NO3- concentration per gram dry soil mass divided by the time of

incubation.

Soil pH and extractable P were measured on air-dried soils that were forced with

a rubber stopper through a 2 mm sieve. Soil pH was measured using a 2:1 water: soil

ratio on a electronic pH meter (Thermo Orion 250A+, Orion Research, Inc., Boston,

Massachusetts, USA). I extracted soil P with a Melich I solution (Kuo 1996). The

resulting phosphate (PO4-3) in solution was read by colorimetric determination of ortho-

phosphate with modified volumes as described by Murphy and Riley (1962), using a

63

spectrophotometer microplate reader (PowerWave XS Microplate Reader, Bio-Tek

Instruments, Inc., Winooski, Vermont, USA). Total percent C and N were measured on

ground soils dried at 60 °C using an elemental analyzer (ECS 4010, Costech Analytical,

Valencia, California, USA). For soils collected in 2007 (pre-fertilization) and in 2009 (2

yrs after fertilization) at 0-10cm depth, a detailed “48-element 4-acid ICP-MS analysis”

(code ME-MS61) was conducted at ALS Chemex laboratories. A summary of Pre-

fertilization values is presented in Table 3-1.

Tree Diameter Measurements

In May 2007, all trees larger than 9 cm diameter at breast height (DBH) in each

20 x 20 m plot were identified to species or genus and labeled with a metal tag nailed to

the bole. A complete list of species of trees is included in Table C-1. Lianas with stems

larger than 10cm with roots in the plot were considered trees. I also included felled trees

with live leaves. For larger trees (approximately 10 per plot), I installed dendrometer

bands constructed with aluminum packing tape and springs as described by Keeland

and Joy Young (2004). I removed lianas and loose bark at point of attachment and

installed dendrometers ≥10 cm above identification tags or higher when buttresses were

present. Dendrometers were allowed to settle at least one month prior to initial marking.

In addition in each plot, I labeled 10 trees of common species with DBH of 4 to 9 cm

with plastic tags attached to a plastic string. I measured diameter increase with either

calipers (for trees with dendrometers) or with a DBH tape (for remaining trees) every six

months until approximately 2.5 yrs after initial fertilization, for a total of six censuses.

To test the community response to fertilization, I used the plot as the

experimental unit (n = 24). I calculated the basal area (BA) occupied by the trees with

the following formula: BA (m2) = (DBH/2) 2 * π. Total basal area increase for a plot (m2

64

ha-1 yr-1) is extrapolated from the total increase in tree diameter for a plot during a time

interval for each size class. This metric was calculated as the plot sum of all the

differences in BA between two consecutive measurements divided by the time interval,

and was dependent on the number and size of the trees in a plot. This metric only

considers trees that increased in diameter. Trees that died, broke, shrunk or did not

change in BA were excluded from the calculations. Overall, when a tree “shrunk”, this

was caused by (1) real shrinkage of the bole, probably due to a change in water

availability or time of day (Sheil 2003) (2) physical damage (one of the trunks was

broken or trunk split), (3) a measurement discrepancy caused by a new dendrometer. In

some cases, the original dendrometer was damaged and a second had to be installed.

Because it was hard to discern among causes of shrinkage, I eliminated these trees

from analyses. I also evaluated tree growth by calculating proportion of tree growth (%

trees that grew yr-1plot-1), which refers to the number of trees that increased in diameter

relative to the total number of trees in a plot for a given time interval. In this case I

considered all trees in a plot. I calculated a total of seven metrics of tree growth,

including relative growth rate (RGR), average of DBH per plot and maximum diameter

increase per plot. However, most metrics provided similar statistical results and

therefore I only present results from total BA increase and proportion of tree growth.

Because of high species diversity in tropical forests, it was challenging to find

trees from the same species in all plots or all treatments. Thus, I selected four common

species (Dendropanax arboreus, Goethalsia meiantha, Pentaclethra macroloba, and

Socratea exohrriza) and three common genera (Inga, Protium and Virola) of trees

ranging from 4 to 89 cm DBH to test how different taxa responded to fertilization

65

additions (Table 3-4). To analyze species responses to fertilization I used individual

trees as the experimental unit and RGR as the metric of tree growth. In this case I

considered all trees for analysis, including those that did not grow because the results of

the analysis do not change if trees that did not grow were included.

Seedling Measurements

In July 2007, I selected, marked, and photographed 20 seedlings or saplings

from the most common species in each plot that had expanded leaves and were less

than 1m high. I documented evidence of herbivory, and recorded stem length, number

of leaves, and canopy cover above each seedling using a spherical denisometer

(Forestry Supplies, Lincoln, NE, USA). In August 2008 and 2009, I conducted a census

of the marked seedlings and recorded survival, growth, number of leaves, and level of

herbivory. A seedling was considered dead if the stem was broken or dry and there was

no resprouting. Seedlings that could not be found were considered missing and were

not included in the analysis. Growth was defined as increase in stem length measured

with a ruler from the point where the root starts to the longest branch. However, some

seedlings showed no change, or a decrease in stem length, caused by a broken or

damaged stem. Other seedlings had multiple stems that were longer than the stem

measured in the previous census. This physical damage and resprouting of multiple

stems made it difficult to quantify plot-level growth. Therefore, I recorded the percent of

seedlings that grew, which I calculated as the proportion of seedlings with an increase

in stem length or number of stems relative to those with no change, or a decrease in

stem length relative to the previous census (1 yr – baseline; 2 yr – 1 yr). The percent of

seedlings that increased the number of leaves also refers to change between censuses

66

for live seedlings (1 yr – baseline; 2 yr – 1 yr). Finally, I recorded presence/absence of

herbivory at each census.

Litterfall Collection

I collected litterfall every two weeks for three years from the onset of the

experiment, from two traps installed in each plot. Traps measured 50 x 50 cm and were

constructed of mesh and pvc mounted on 1m-high metal rods. Samples from both traps

were combined, dried at 60 °C, separated into foliar, reproductive and woody

subsamples, and weighed. The foliar fraction included all leaves, petioles and fronds;

the reproductive fraction included flowers, peduncles, fruits and seeds; the woody

fraction included all sticks smaller than 2 mm in diameter. I discarded sticks larger than

2 mm in diameter, mosses, insects, and all unidentified material.

Not all litterfall-collection intervals were regular because at times inclement

weather prevented collection or traps were stolen. I therefore calculated a daily litterfall

rate (g biomass/day) for each collection by dividing dry litterfall mass by number of days

for that specific interval. If there was more than one collection per month I averaged the

rates for that month. For months where there were no collections, I used the average

from the previous and following month for that year. After obtaining all monthly rates, I

multiplied them by the number of days in that month to obtain a monthly production

value. The sum of all monthly productions 12 months post initial fertilization, was

labeled as “year one” (which corresponded to August 2007-July 2008). Litterfall

production for “year two” refers to August 2008-July 2009.

To measure coarse litter production, I marked two 1 x 1 m quadrants on the

forest floor in each plot (Clark and Clark 2001b). Here, I collected palm fronds and wood

pieces >2 mm in diameter, approximately every three months between February 2008

67

and July 2009, resulting in four collection periods. Samples from both quadrants were

combined, dried at 60 °C, and weighed. When samples were too large to transport to

the laboratory, fresh weight was obtained with a Pesola scale in the field and a

subsample collected. With the fresh and dry weight from the subsample, the %water

content of the large sample was estimated and the dry weight of the whole sample

calculated. In August 2009, I collected all standing litter in two 50 x 50 cm quadrats

placed randomly in the buffer zone of each plot in order to measure the pool of standing

litter on the forest floor. Samples were dried at 60 °C, separated into leaves,

reproductive parts (flowers and fruits), sticks <2 mm diameter, sticks >2 mm diameter or

wood pieces, and roots, and then weighed. Leaves and sticks <2 mm diameter were

combined as “foliar litter fraction” and the sum of all fractions is the total litterpool.

Surface roots in the litter were only found in seven plots, thus the root fraction was not

analyzed separately.

Root Measurements

To estimate the effect of fertilization on root biomass, I collected 0-15 cm deep

root cores using a pounding corer with a volume of 4.75 x15 cm. Prior to treatments,

followed by 1-yr and 2-yr post treatments, I collected root cores at four points per plot

adjacent to where soil samples were collected, and combined them into one composite

sample for each plot. Intact root cores were refrigerated at 4 °C for up to three months

and then separated into large (>2 mm diameter), small (<2 mm diameter), live and dead

roots. For samples collected in 2007 and 2008, I separated roots using variable-speed

electric drills as described by Espeleta and Clark (2007). For samples collected in 2009,

I dissolved the samples in two-gallon (7.6 L) pails filled with water but stirred them

manually and hand-sorted the roots using soil sieves. Each year, immediately after

68

cleaning and separating the roots, I dried them at 60 °C for at least one week and

obtained dry weight.

To estimate the effect of fertilization on fine root production I used ingrowth cores

(Cuevas and Medina 1988). In August 2007, I randomly chose two locations within each

plot and installed two cylindrical, closed-bottom root ingrowth cores (2 mm mesh, 10 cm

deep and 7.7cm diameter).To install the ingrowth cores, I collected soil cores, removed

all visible roots by hand (using latex gloves), placed the mesh ingrowth cores in the hole

where the core was removed, and then filled the cores with the root-free soil. In July

2009, I removed the cores and brought them to the laboratory, where I separated the

roots that had grown inside by washing the cores in a bucket with water and hand

sorting the roots using soil sieves. I separated roots by size (>2 mm or < 2 mm

diameter) and status (live/dead). I dried roots at 60 °C to a constant mass and recorded

dry weight.

Statistical Analysis

To test the effect of fertilization treatments on total basal area increase,

proportion of tree growth, litterfall and coarse debris production, and root biomass, I

used plot-averaged values (n=24) in repeated measures MANOVAs with measurements

at different times as dependent variables and treatment and block as independent

variables. I selected this approach over univariate repeated measures ANOVA because

in some cases the sphericity assumption was not met. I also calculated relative growth

rates (RGR) from individual trees and conducted one-way ANOVAs with RGR (mm yr-1)

as the dependent variable and treatment as the independent variable. I calculated RGR

for each tree as the slope from a line from the log-transformed DBH at each census (n =

6). I chose this approach to test fertilizer effects in addition to plot-averaged analyses

69

(MANOVAS) because this approach incorporates differences in tree size and can still

be calculated if one time measurement is missing. Because RGR is calculated as a

slope, it incorporates differences between measurements and provides an integrated

index of growth throughout the timescale of the experiment. I used Dunnett’s method to

test if treatment means were different from the control. I also used these one-way

ANOVAs with RGR as the dependent variable as a complementary analysis when there

was a treatment *time interaction in the repeated measures MANOVA and it was not

evident where the differences were. Finally, I used a contingency table to explore the

likelihood that a tree would grow in a control plot versus in a treatment plot. I calculated

odds ratios from this contingency table as (# trees that grew in treatment x/ # trees that

did not grow in treatment x)/ (# trees that grew in control/ # trees that did not grow in

control). I calculated these odds ratios for each tree size class and each treatment (x=

+N, +P, or +NP).

To explore which of the measured variables were important to predict tree RGR at

EARTH forest, I also utilized a mixed model approach. I fitted different models and

compared them using the AIC criteria. Results were similar than those obtained using

the repeated MANOVA approach and therefore were not included here. However, a

detailed description of the methodology and results from this analysis is included in

Appendix B.

To test for treatment and block effects in root production (results from ingrowth

cores) and litterpool samples I used a two-way ANOVA. Here, when necessary, data

were log-transformed to meet normality assumptions. To test for the effect of fertilization

on categorical variables deriving from seedling measurements, I used Pearson chi-

70

squared tests. For these analyses, I used individual seedlings (not plot-averaged

values) because the number of seedlings surviving in each plot varied significantly. For

tests of growth, number of leaves and herbivory, I only considered live seedlings.

In August 2008, just after my field season ended, a strong storm with severe

winds struck the EARTH Forest Reserve destroying plots 17 and 22 (Appendix A). Both

plots were on the Rio side and belonged to the +N treatment. This event resulted in

missing data for tree censuses four, five and six and for seedling census three (2 yrs

after fertilization), as well as for soil and root collections for 2009 (2 yrs after

fertilization). To avoid an unbalanced design in my statistical analyses, I replaced

missing data with averages from other plots for that treatment for that year. In the case

of litterfall collection, however, because of the significant amount of missing points I

eliminated these two plots from the analysis (n = 22). For the seedling analyses I

conducted two different tests for 1 yr and 2 yrs after fertilization and excluded seedlings

from plots 17 and 22 from the 2 yr analyses.

In all figures and tables as well as in the text, means (+ standard errors) are

shown. All figures were constructed using Sigmaplot 11.0 and analyses were conducted

in JMP 8.0(SAS Institute Inc., Cary, NC, USA).

Results

Soil Measurements

Repeated measures MANOVAs revealed that spatial and temporal variation in

soil parameters was larger than variation stemming from fertilization treatments, which

resulted in no treatment effect but significant block and time effects in most soil

parameters (Figure 3-4 and Table 3-5). Spatial variation was not only large among

blocks but also within blocks. For example at the 0-10 cm depth, DIN ranged between

71

3.2-26.6 µg g-1 within one block (a coefficient of variation (CV) of 40%) and Melich P

ranged between 2.1-6.7 µg g-1 (a CV of 58%).

This large variation in soil parameters within blocks made it difficult to detect

treatment effects. Therefore, to distinguish the effect of fertilization additions from

background heterogeneity, I compared the percent difference in soil parameters

between initial values (pre-fertilization) and those obtained two years after fertilization

(Table 3-6). Two years after the initiation of the experiment, there was no significant

percent change in pH, percent C, DIN, or percent N in any treatment (Figure 3-5A-D).

However, after two years of fertilization there was a significant increase in Melich P and

total P at the 0-10 cm depth in the +P and +NP treatments (Figure 3-5E, F). For Melich

P, there was a mean increase of 59 + 9.8 % and 87 + 25 % for the +P and +NP

treatments respectively. By contrast, there was a 9.3 + 23% and 20 + 5.6 % decrease in

Melich P in the control and +N treatments respectively. This increase in Melich P was

not observable in the 10-30 or 30-50cm depth increments (Figure 3-5E). Total P

increased 7.2 + 2.2 % and 8.6 + 2.5 % in the +P and +NP treatments respectively but

decreased by 1.8 + 1.0 % and 4.6 + 3.4 % in the control and +N treatments

respectively. The observed total P increase represents a recuperation of 97% and 93%

of the P added as fertilizer in the +N and +NP treatments, respectively. Melich P and

Total P were positively correlated before and 2yrs after initial fertilization (Figure C-2). I

did not analyze samples from deeper profiles for total P.

Tree Growth

At the community level there was no detectable difference in tree growth

measured as total increase in basal area (Figure 3-6 and Table 3-7). However, there

was a time*treatment interaction, as well as a significant block and time effect, in the

72

percentage of trees that grew per plot (Figure 3-7 and Table 3-8). This indicates that for

some blocks at different times there were significant treatment effects, although from

plotting the data and conducting ANOVAS at each time point, it was not possible to

discern where the differences occurred (non-significant ANOVAS not shown). To further

explore this result I compared the RGRs of all trees across treatments (experimental

unit here was individual tree as opposed to plot total). Tree RGRs were higher in the

+NP treatment relative to the control (F3,768 = 2.17, P = 0.08; Dunnett’s test = 2.35, P =

0.04 ). This difference was probably due to small trees (5-10cm DBH) growing more in

the +NP treatment (Figure 3-6a) and is consistent with the number of trees that grew in

each plot across time (Figure 3-7 and Table 3-8). This result was consistent if all trees

were included in the analysis or if only trees that grew were included in the analysis.

Overall, a tree in the +P treatment was 1.65 times more likely to grow than in a control

plot and a tree in the +NP treatment was 1.68 times more likely to grow than in a control

treatment, 2 yrs after initial fertilization (Table 3-9).

In addition to the community level response, I explored the effect of fertilizer

additions on trees of different size classes. For the 5-10 cm size class there was a time

effect and a time* treatment interaction in the number of trees that grew in each plot

across time (Figure 3-7 and Table 3-8). When considering tree RGRs among treatments

for this size class, there were significantly higher RGRs for trees in the +NP treatment

(F3,266 = 3.94, P = 0.01; Dunnett’s test = 2.36, P = 0.01; Figure 3-8). In addition for this

size class, trees in the +P and +NP treatment were 2.16 and 2.68 times more likely to

grow than trees in the control plots 2 yrs after initial fertilization (Table 2-9). For this size

class there was a positive correlation between percent change in total P in surface soils

73

(0-10 cm) after two years and the number of trees that grew between 2 and 2.5 yrs after

fertilization (r2 = 0.22, d.f. = 20, P = 0.03). For all the size classes combined there was

also a positive correlation between percent change in total P in surface soils (0-10 cm)

after two years and the number of trees that grew between 2 and 2.5 yrs after

fertilization (r2 = 0.33, d.f. = 20, P = 0.01).

I also studied the effect of fertilizer additions on trees from different taxa within

the community. From the four species and three genera where I analyzed RGR

individually (here again individual tree was the experimental unit), only Goethalsia

meiantha and Socratea exohrriza showed evidence of a response to fertilization (Figure

3-9). Goethalsia had the highest growth in the control treatment (Figure 3-9B), although

this pattern was highly influenced by a single tree and should be interpreted with

caution. Socratea exohrriza, a fast growing canopy palm, had higher growth with P

additions (Figure 3-9F). Protium had higher RGR when both N and P were added

together, although this difference was not statistically significant (P = 0.51).

Seedling Measurements

There was a significant treatment effect on the proportion of seedlings that

survived, grew, and showed an increase in leaf number two years after initial

fertilization. Presence of herbivory was common (more than 50% in every plot) and did

not differ among treatments (Figure 3-10 and Table 3-10).

Litterfall and Litter Pool Measurements

Mean foliar litterfall productivity was 5.0 + 0.27 Mg C ha-1 yr-1, mean reproductive

litterfall productivity was 0.72 + 0.19 Mg C ha-1 yr-1, and mean coarse litterfall

productivity was 1.98 + 0.34 Mg C ha-1 yr-1. There were no differences among

treatments in foliar, reproductive, or coarse litterfall productivities. However, foliar and

74

reproductive litterfall productivities decreased over time (Figures 3-11 and 3-12 and

Table 3-11) and the decrease was consistent across blocks (no time*block interaction).

The standing pool of litter measured in 2009 in 50 x 50 cm plots established in the

buffer zones of each plot averaged 2.77 + 0.18 Mg C ha-1 foliar litter, 0.56 + 0.17 Mg C

ha-1 reproductive litter, and 0.64 + 0.08 Mg C ha-1 sticks with diameters >2 mm. These

values summed up to a total litterpool of 6.76 + 0.36 Mg C ha-1. There was no difference

among treatments or blocks in foliar, reproductive or total standing litter or in the mass

of sticks with diameters >2 mm collected from these litterpool plots (Table 3-12).

Root Biomass and Productivity

Two years after initial nutrient additions there was no difference among

treatments in biomass of fine roots (< 2 mm diameter) or large roots (> 2 mm diameter)

in the top 15 cm of the soil profile. However, across treatments, there was a larger

biomass of fine roots two years after the onset of the experiment than in the two other

sampling times (Table 3-13 and Figure 3-13). Mean fine root biomass increased from

0.89 + 0.09 Mg C ha-1 pre-fertilization to 2.02 + 0.17 Mg C ha-1 two years after

fertilization. In addition, for both root sizes there was a significant block effect; overall,

plots in the Rancho site had more roots than plots in the Rio site (MANOVA by site: fine

roots F1, 19 = 19.4 P < 0.01 and large roots F1, 19 = 8.24 P < 0.01).

Root productivity measured with ingrowth cores in the 0-10cm soil profile, did not

differ among treatments or blocks (Ftreatment = 0.52, P = 0.67, Fblock = 1.57, P = 0.67 d.f. =

3,5 ; Figure 3-14). There was no relationship between root productivity and fine or large

root biomass before fertilization or after fertilization (data not shown).

75

Discussion

Soil Measurements

Baseline nutrient concentrations in soils at the EARTH Forest were high relative

to other tropical forests. Nitrogen measurements were similar to values obtained at La

Selva, another well-studied forest in Costa Rica with similar climatic and floristic

composition (Vitousek and Denslow 1986, Sollins et al. 1994). Superficial total P (0-10

cm) at EARTH Forest was 1601 µg g-1 (or 1136.71 kg ha-1), which was particularly high

compared to other forests where fertilization experiments have been conducted: a

secondary forest in Pará, Brazil (~225 µg g-1; Davidson et al. 2004), an evergreen forest

in Kalimantan, Borneo (<250 µg g-1; Mirmanto et al. 1999), a rain forest in Korup,

Cameroon (<300 µg g-1), a seasonal forest in Gigante Peninsula, Panama (772 µg g-1

;Yavitt et al. 2011), and a wet forest in Osa Peninsula (557 µg g-1; Townsend et al.

2002). However, total P values at EARTH were similar to the alluvial soils at La Selva

(1650 µg g-1 in alluvial soils and 579 µg g-1 in residual soils; Espeleta and Clark 2007,

Wood et al. 2009). Concentrations of immobile elements, such as niobium (Nb) and

zirconium (Zr), were similar to those reported for volcanic parent material in the area

(Table A-4; Meijer and Buurman 2003). This, in conjunction with high total P, suggests

that soils at EARTH were not significantly leached during development. Alternatively,

nearby volcanoes may have provided inputs of lahar or ash to the region recently, which

would have replenished soil fertility after leaching of original parent material (Meijer and

Buurman 2003, Porder et al. 2006). High concentrations of total P could imply that P

availability at EARTH forest is higher than in other tropical forests, where most rock

derived minerals have been leached or occluded, and P limits NPP (e.g., Kauai Island in

Hawaii; Herbert and Fownes 1995, and potentially Kalimantan, Borneo; Mirmanto et al.

76

1999). However, resin-extractable P (not shown but below 1µg g-1) and Melich-

extractable P (2.14 + 0.36 µg g-1) at EARTH Forest were low, which suggests that

although P pools are large, P bioavailability could be low. In these soils high clay

content (around 50%, Table A-2), high sorption capacity (Sanchez 1976), and microbes

(Olander and Vitousek 2004) can immobilize PO4-3 making it unavailable for plant

uptake (Vitousek and Denslow 1987, Vandecaar et al. 2009).

There was a large spatial and temporal variation in soil measurements, which

was evident as significant block effects, and could have masked increases in N or P

concentrations after fertilization. To distinguish the effect of fertilization additions from

background heterogeneity, I compared the percent difference in soil parameters

between baseline and two years after fertilization. After two years of fertilization, there

was a significant increase in soil Melich P and Total P relative to baseline levels but no

increase in any N parameters measured. These results are similar to soil responses to

fertilization in a secondary forest in the Brazilian Amazon (Davidson et al. 2004) and

could be attributed to the larger N than P background pools, or to higher N ion mobility.

Because Nitrate (NO3-) is more mobile than phosphate (PO4

3-) and soil samples were

collected almost 6 mo after fertilization, it is possible that all added N was leached

(Radulovitch and Sollins 1991) or immobilized by plants and/or microbes by the time

soils were sampled (Chapter 4).

Nutrient Limitation to NPP

After 2.5 yrs, there was no clear effect of fertilization on tree diameter increase

(or basal area (BA) increase) at the community level, litterfall productivity, or root

biomass or productivity. One possible explanation is that NPP at EARTH forest is

limited or co-limited by a nutrient other than N or P, as suggested by Kaspari et al.

77

(2008) for litterfall production and decomposition in Panama. At the EARTH forest there

are relatively high concentrations of N and P in soils (Table 3-2), foliage, and litterfall

(Chapter 4). In addition, litterfall production at EARTH Forest (11.44 Mg ha-1 yr-1) is on

the high spectrum compared to other tropical forests: 7.17 Mg ha-1 yr-1 in Kalimantan

(Mirmanto et al. 1999); 10.65 Mg ha-1 yr-1 in Gigante Peninsula (Kaspari et al. 2008); 12

Mg ha-1 yr-1 in Osa Peninsula (C. Cleveland unpublished data), and ~9 Mg ha-1 yr-1 in La

Selva (D. A. Clark unpublished data), which suggests that at EARTH trees are meeting

at least some of their nutrient demands to produce biomass. Therefore, it is possible

that as N and P requirements are met, the plant community becomes limited by another

nutrient, potentially potassium or a combination of micronutrients (Herbert and Fownes

1995, Kaspari et al. 2008, Townsend et al. 2011, Wright et al. 2011).

After 2.5 yrs there was no overall NPP response to fertilization but there was a

higher proportion of trees that showed some BA increase in the +P and +NP treatments,

which suggests that NPP at EARTH forest could be limited by P to some degree but

that this experiment failed to capture statistical differences among treatments.

Treatment effects could be obscured by the high spatial and temporal variability in

individual tree growth (Clark and Clark 1999, Clark and Clark 2011), litterfall (Wood et

al. 2009), and root biomass (Espeleta and Clark 2007), which are characteristic of

forests in this area. Potentially, microclimatic variables (e.g., light, water availability, or

soil temperature) could influence NPP more than nutrient availability on the short term

and at the plot scale, confounding any experimental treatments. If so, more intense

sampling and over a longer period would be required to observe treatment effects (Clark

and Clark 2001b, Clark and Clark 2011).

78

To compensate for spatial variation in litterfall, it could be argued that more traps

were needed to capture community wide responses to fertilization. In my experiment,

however, litterfall collection effort (area of traps/ plot area) was higher (6.26 m2 ha-1)

than in experiments at La Selva (4.5 m2 ha-1 in old growth and 2.56 m2 ha-1 in

secondary forest), where significant differences in litterfall production were observed

after a litter manipulation experiment equivalent to fertilization with 5-25 kg ha-1 of

organic P (Wood et al. 2009). In the case of diameter increase, all trees >10cm DBH

and 10 trees between 4-9 cm DBH were measured per plot, twice per year for 2.7 yrs.

These recordings are similar to other fertilization experiments where differences in tree

growth were observed (e.g., Tanner and Kapos 1992, Herbert and Fownes 1995,

Davidson et al. 2004). In contrast, superficial root biomass in these forests is largely

variable and influenced by the presence of palms (Espeleta and Clark 2007).Therefore,

it is likely that random location of palms relative to where samples were collected had

an influence in the observed results. In addition, the change in methodology used to

separate roots in the last sampling event (2 yrs after fertilization, see methods) may

have caused the apparent increase in root biomass over time. In conclusion, it is

unlikely that insufficient sampling effort was the cause for a lack of difference among

treatments in litterfall production or tree growth but it is possible that more intense root

sampling with consistent methodology over time would reveal differences in root

productivity or biomass.

Temporal variation in tree growth associated with climatic variables in this type of

forest is large and heterogeneous among size classes and species (Clark et al. 2003,

Clark and Clark 2011). Thus, it is possible that more time is needed to observe

79

treatment effects in diameter increase. In the case of litterfall production, treatment

differences have been observed in less than two years in other experiments (Herbert

and Fownes 1995, Mirmanto et al. 1999, Wood et al. 2009). However, in a montane

forest in Venezuela, there was an effect of both N and P additions on litterfall only after

4 yrs (Tanner and Kapos 1992). Because of high temporal variability in the system and

high background nutrient levels, it is possible that differences among treatments at

EARTH forest will only be observable after several more years of nutrient additions. As

recently highlighted by Clark and Clark (2011), temporal variability in NPP in lowland

tropical forests make long-term observations particularly important and necessary.

Nutrient Limitation and Tree Size

After 2.5 yrs, there was no difference among treatments in community-level BA

increase (Figure 3-6) but there was a higher proportion of trees that increased in BA in

+P plots (Figure 3-7). This suggests that the largest trees, which contribute the most to

total BA increase in each plot, are not limited by P (or did not respond in a 2.5 yr time

frame), but that smaller trees (5-10 cm DBH), which are more frequent, could be P

limited. This response of small trees was consistent with traditional expectations (i.e.,

Walker and Syers 1976) and was supported by a positive relationship between percent

difference in soil total P (0-10cm depth) and the proportion of trees showing some

diameter increase.

However, when observing RGR responses of all small trees (considering individual

trees as opposed to plot averages), it is evident, that although not statistically

significant, these small trees are limited not only by P but by N as well. RGRs of small

trees showed a sub-additive response to N and P (sensu Harpole et al. 2011),

increasing RGR by 15% with +N, 26% with +P and 38% with +NP relative to the control.

80

This form of co-limitation is incompatible with Liebig’s Law of the Minimum (Liebig 1842)

and supports recent evidence that proposes synergistic interactions between N and P

availability (Elser et al. 2007, Harpole et al. 2011, Cleveland et al. 2011).

Because of light-co-limitation of small trees, I had predicted that medium (10-30

cm DBH) trees would be more likely to respond to nutrient additions. However, small

trees were the only ones that differed among treatments, either because they are the

only ones limited by N and P or because small trees have a faster growth rate (Clark

and Clark 1992, Clark 1994). Although there was no difference among treatments in the

proportion of trees >30 cm DBH that grew, there was a high proportion of large trees

that grew overall (>40 % of trees for any given census for all treatments), which

suggests that EARTH Forest is a relatively young forest and that soils are relatively

nutrient-rich.

I hypothesized that seedlings would not respond to nutrient additions because

low light levels in the understory would prevent them from incorporating extra nutrients

as biomass (Denslow et al. 1990, Burslem et al. 1995). However, although this was

probably true in some cases, a higher proportion of P-fertilized seedlings survived,

grew, and increased leaf number after two years than seedling from other treatments.

This suggests that (1) the canopy at EARTH Forest is relatively open and enough light

reaches the understory to allow seedling growth (Lawrence 2003, Baroloto et al. 2006,

Appendix B), and (2) the seedling community at EARTH may be limited by P (Palow

and Oberbauer 2009). Interestingly, there seemed to be an effect on the +P but not the

+NP treatment. Perhaps, +N cancels the +P effect by increasing herbivory if N is

incorporated into tissues, increasing their palatability (Andersen et al. 2007 but see

81

Campo and Dirzo 2003). However, I did not collect foliage from seedlings and did not

observe a significant difference in the percent of seedlings exhibiting leaves with

herbivory. More experiments are required to test this hypothesis.

Effect of Taxa on Nutrient Limitation

I had predicted that fast growing, canopy species, such as Pentaclethra and

Goethalsia would respond to nutrient additions but this was not the case. The lack of

response of Pentaclethra is interesting because as the most important species in this

forest (up to 30% of the basal area and 14% stems, Figure 3-2); it has a strong

influence over the community-averaged response to fertilization. Pentaclethra is a

nitrogen-fixing legume and therefore I did not expect a response to N fertilization

(McKey 1994). However, the lack of response to the +P treatment suggests that either

this species is well adapted to the PO43- levels found at EARTH Forest or that it has

access to enough PO43- to fulfill metabolic needs. It has been suggested that N fixing

legumes, such as Inga and Pentaclethra, could have an advantage when available P is

low. These species can invest extra N in producing phosphatase enzymes, which

enable them to access organic P pools (Olander and Vitousek 2000, Treseder and

Vitousek 2001, Houlton et al. 2008). Measurements of phosphatase activities and

available P at the base of these trees could be used to test this hypothesis.

From the studied taxa, only Socratea exohrriza (the most abundant canopy palm;

Figure 3-2), significantly responded to nutrient additions. Individuals from this species

showed very fast growth in several plots where there was a combination of P fertilization

plus available light due to a tree or branch fall (personal observation). These palms,

however, usually reach a maximum stem diameter around 30 cm, so this response was

mainly due to small individuals. Interestingly, larger individuals responded by increasing

82

foliar P instead of stem diameter with P additions (Chapter 4). Finally, as predicted,

subcanopy trees (Dendropanax and Protium) did not grow more with fertilization

probably because of light co-limitation (Fetcher et al. 1994, Holste et al. 2011).

Conclusions

This study emphasizes the complexity of nutrient limitation in lowland diverse

tropical forests (Vitousek et al. 2010, Townsend et al. 2011, Cleveland et al. 2011) and

adds to the body of experimental evidence showing that these forests may not be

limited by P (Mirmanto et al. 1999, Newbery et al. 2002, Wright et al. 2011), or at least

do not respond to either N or P fertilization in the short term as is the case in montane

forests (Vitousek 2004, Tanner 1998). Most likely, as stated by Kaspari et al. (2008),

“these systems are non-Liebig worlds of multiple nutrient limitations”. More experiments

are needed to elucidate the mechanisms driving limitation by multiple nutrients, such as

proposed by Harpole et al. (2011).

Although the plant community as a whole at EARTH Forest does not appear to

be limited by N or P, my data suggest that seedlings and a palm species (Socratea

exohrriza) are limited by P, and that on average small trees (5-10 cm DBH) are co-

limited by N and P. The lack of response of Pentaclethra to fertilization implies that this

species, the dominant at EARTH Forest, is not limited by either N or P (Chapter 4). In

contrast, this species, and potentially others at EARTH Forest, may be limited by a

different nutrient or nutrients. Alternatively other factors, such as small scale variation in

light and water availability exert more important controls on tree growth than nutrient

availability.

Recent attention has been given to the complex nature of “community nutrient

limitation” (Vitousek et al. 2010, Harpole et al. 2011) and mechanisms for multiple

83

nutrient co-limitation have been proposed as an alternative to the historical Liebig’s Law

of the Minimum (Chapin et al. 1987, Gleason and Tilman 1992, Rastetter and Shaver

1992, Danger et al. 2008, John et al. 2007). In the case of this study, I would expand to

say that at EARTH forest there is “heterogeneous nutrient limitation”, not only driven by

variability in nutrient responses among tree species but also among size classes. This

heterogeneity highlights the importance of considering different aspects of the plant

community, such as forest structure and species composition, when making predictions

concerning nutrient limitation in lowland tropical forests. Furthermore, the differential

response of size classes and species suggests that changes in nutrient availability

could lead to changes in forest structure or even diversity in the long term and could

have important implications for plant-soil-microbial feedbacks concerning nutrient

limitation.

84

Table 3-1. Particle sizes for soils from the top 10cm of the study plots (Rancho and Rio

are two areas of the forest where plots were located). Data were obtained using a Mastersizer Particle Size Analyzer. Samples were dispersed in 1% solution of Calgon soap overnight.

Class clay very fine silt

fine silt medium silt

coarse silt

very fine sand

fine sand

medium sand and

greater

Size (um) < 2 2-7.8 7.8-15.6

15.6- 31

31-62.5

62.5-125

125-250

>250

Rancho 8.62 31.38 23.78 17.23 8.96 4.50 2.27 3.27

Rio 9.44 33.87 23.27 15.36 7.68 3.90 2.12 4.37

Average 8.99 32.51 23.54 16.38 8.38 4.23 2.20 3.77

Table 3-2. Means (with standard errors) for various soil parameters measured at three

depths on each plot at the beginning of the experiment. N = 24 plots. For details about analyses see methods section.

Measurement Depth (cm)

0-10cm SE 10-30cm SE 30-50cm SE

Bulk Density (mg cm-3) 0.71 0.03 0.78 0.02 0.83 0.04

pH H2O 4.04 0.04 4.20 0.03 4.32 0.02

Melich P(µg g-1) 2.14 0.36 1.40 0.25 1.02 0.25

Total %C 4.83 0.14 2.51 0.07 1.71 0.06

Total %N 0.49 0.01 0.28 0.01 0.18 0.01

DIN (µg N g-1) 19.67 1.47 10.86 1.23 6.89 0.56

Net min. (µg N g-1 d-1) 1.68 0.28 0.50 0.17 0.36 0.09

Net nit. (µg N g-1) 1.74 0.35 0.54 0.10 0.35 0.09

85

Table 3-3. Means (with standard errors) for various soil parameters measured at EARTH Forest at the beginning of the experiment. N = 24 plots, depth = 0-10cm. For details about analyses see methods section.

Element Mean SE

Al% 14.27 0.09

Ca % 0.04 0.00

Co (ppm) 27.40 1.49

Cu (ppm) 144.56 2.18

Fe % 8.78 0.10

K % 0.12 0.00

Mg % 0.17 0.00

Mn ppm 765.88 65.73

Mo (ppm) 2.67 0.07

Na % 0.04 0.00

Ni (ppm) 63.77 1.32

P (ppm) 1601.25 104.47

S % 0.10 0.00

Zn (ppm) 110.04 2.80

86

Table 3-4. Taxa selected to study the effect of fertilization on different functional groups. Shown is the percent of the plots in which a taxa were present (n = 24 plots) and the total number of trees measured (all trees were >4 cm DBH). All taxa had at least one individual in each treatment.

Species Family % plots

# trees

Functional properties at EARTH Forest

Dendropanax arboreus

Araliaceae 50 21 Subcanopy tree with soft wood, fast growth, and relatively high light demand.

Goethalsia meiantha

Malvaceae 50 49 Canopy or subcanopy tree with soft wood, fast growth, and relatively high light demand. Characteristic of disturbed areas.

Inga Mimosaceae 70 25 Seven species included1. Canopy or subcanopy trees with hard or semi hard wood, and mostly shade tolerant. Can fix Nitrogen.

Pentaclethra macroloba

Mimosaceae 100 91 Canopy tree with semi hard wood, medium growth, and shade tolerant. Most abundant tree at EARTH forest. Can fix Nitrogen.

Protium Burseraceae 75 41 Four species included2. Subcanopy trees (at EARTH forest) with relatively slow growth, and shade tolerant. Resinous compounds in leaves, stems, and fruits.

Socratea exohrriza

Arecaceae 100 154 Canopy/subcanopy palm with relatively fast growth and shade tolerant. Conspicuous stilt roots.

Virola Myristicaceae 70 24 Three species included3.Canopy trees with semi hard wood, relatively slower growth, and shade tolerant. High concentrations of alkaloids and other compounds in leaves, stems, and fruits. Dioecious.

Source McDade et al. (1994) and O. Vargas (personal communication). 1 I. alba, I. leocalycina, I. pezizifera, I. sapindioides, I. thiboudiana, I. umbilifera, I. venusta

2 P. confusum, P. panamense, P. pittieri, P. Ravenii

3 V. koschnyi, V. multiflora, V. sebifera

87

Table 3-5. Results from repeated measures MANOVAs for several soil variables measured at three depths. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. Significant effects are signaled with an asterisk

Treatment Block Time Treatment* Time

Block *Time

Variable F P F P F P F P F P

pH

0-10cm 0.12 0.95 1.21 0.37 1.93 0.01* 2.07 0.11 1.17 0.37 10-30cm 0.38 0.77 0.51 0.76 9.59 <0.01* 0.76 0.61 0.99 0.48 30-50cm 1.07 0.40 4.00 0.03* 41.19 <0.01* 1.26 0.32 2.08 0.07

%C

0-10cm 0.46 0.71 4.40 0.02* 9.00 <0.01* 1.33 0.29 2.57 0.03* 10-30cm 0.51 0.69 1.26 0.34 11.96 <0.01* 1.67 0.17 0.86 0.58 30-50cm 1.10 0.38 0.53 0.75 0.57 0.58 1.21 0.34 0.98 0.49

DIN (µg g-1)

0-10cm 0.93 0.46 1.44 0.27 15.65 <0.01* 1.26 0.31 3.84 <0.01* 10-30cm 1.27 0.33 1.02 0.45 2.90 0.09 0.33 0.91 1.11 0.39 30-50cm 3.70 0.04* 4.78 0.01* 8.32 0.01* 0.69 0.66 1.96 0.09

% N

0-10cm 0.09 0.97 3.66 0.03* 26.83 <0.01* 0.75 0.62 2.10 0.07 10-30cm 0.47 0.71 0.41 0.28 5.60 0.02* 0.90 0.51 0.60 0.80 30-50cm 1.23 0.34 0.41 0.83 0.07 0.93 1.43 0.24 1.02 0.46

Melich P (µg g-1)

0-10cm 0.11 0.95 3.23 0.04* 2.59 0.12 0.79 0.59 0.95 0.51 10-30cm 0.25 0.86 3.46 0.03* 30.84 <0.01* 0.44 0.85 2.46 0.03* 30-50cm 0.49 0.70 2.93 0.06 40.25 <0.01* 1.00 0.44 3.41 0.01*

Total P (µg g-1)

0-10cm 0.59 0.63 5.31 0.01* 3.6 0.08 7.42 <0.01* 1.17 0.38 Net mineralization (µg N g-1 d-1)

0-10cm 1.66 0.23 3.06 0.05 1.15 0.35 1.90 0.13 2.59 0.03* 10-30cm 0.68 0.58 2.10 0.14 1.99 0.18 1.08 0.40 1.68 0.15 30-50cm 0.40 0.76 2.41 0.09 1.92 0.19 0.30 0.94 0.17 0.01*

Net Nitrification (µg N g-1 d-1)

0-10cm 0.64 0.60 3.38 0.04* 0.74 0.49 1.69 0.17 2.03 0.07 10-30cm 0.68 0.58 0.51 0.76 8.44 <0.01* 0.69 0.66 1.98 0.09 30-50cm 2.49 0.11 3.14 0.04* 2.42 0.13 1.88 0.12 1.26 0.30

88

Table 3-6. Results from two-way ANOVA analyses for the percent difference between 2 yrs and pre-fertilization values for several soil parameters measured at three depths. Significant effects are signaled with an asterisk.

Parameter Depth df Treatment Block

F P F P

pH 0-10cm 3,4 2.61 0.12 1.55 0.27

10-30cm 3,5 1.10 0.38 0.47 0.70

30-50cm 3,5 2.06 0.16 1.13 0.39

% C 0-10cm 3,5 2.69 0.09 2.52 0.09

10-30cm 3,5 0.07 0.97 0.13 0.98

30-50cm 3,5 0.93 0.45 0.98 0.47

DIN (µg g-1) 0-10cm 3,5 2.05 0.16 6.52 <0.01*

10-30cm 3,5 1.43 0.28 4.40 0.01*

30-50cm 3,5 0.72 0.56 2.92 0.06

Total N (%) 0-10cm 3,5 1.76 0.21 4.75 0.01*

10-30cm 3,5 0.74 0.55 0.66 0.66

30-50cm 3,5 1.03 0.41 0.77 0.59

Melich P (µg g-1)

0-10cm 3,5 9.33 <0.01* 1.61 0.23

10-30cm 3,5 0.23 0.88 1.11 0.40

30-50cm 3,5 0.78 0.53 2.23 0.12

Total P (µg g-1) 0-10cm 3,5 11.27 <0.01* 2.30 0.11

89

Table 3-7. Results from repeated measures MANOVAs for “total basal area increase” by tree size class. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

Size class dfn,d F Probability

5-10 cm

Treatment 3,15 0.72 0.55

Block 5,15 1.15 0.38

Time 4,12 2.53 0.10

Time*treatment 12,32.04 1.15 0.20

Time*block 20,40.75 0.90 0.58

10-30 cm

Treatment 3,15 1.43 0.27

Block 5,15 3.87 0.02*

Time 4,12 6.80 0.04*

Time*treatment 12,32.04 1.58 0.15

Time*block 20,40.75 1.17 0.33

>30 cm

Treatment 3,15 1.16 0.36

Block 5,15 1.02 0.44

Time 4,12 4.23 0.02*

Time*treatment 12,32.04 0.67 0.76

Time*block 20,40.75 1.20 0.30

Total

Treatment 3,15 1.01 0.42

Block 5,15 2.50 0.08

Time 4,12 7.18 <0.01*

Time*treatment 12,32.04 1.02 0.45

Time*block 20,40.75 1.43 0.16

90

Table 3-8. Results from repeated measures MANOVAs for “proportion of tree growth” by tree size class. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

Size class dfn,d F Probability

5-10 cm Treatment 3,15 0.99 0.43 Block 5,15 1.58 0.23 Time 4,12 7.27 <0.01* Time*treatment 12,32.04 2.68 0.02* Time*block 20,40.75 1.17 0.33

10-30 cm Treatment 3,15 1.00 0.42 Block 5,15 7.01 <0.01* Time 4,12 18.13 <0.01* Time*treatment 12,32.04 1.55 0.16 Time*block 20,40.75 2.15 0.02*

>30 cm Treatment 3,15 3.12 0.06 Block 5,15 1.18 0.36 Time 4,12 2.07 0.15 Time*treatment 12,32.04 0.81 0.64 Time*block 20,40.75 0.85 0.65

Total Treatment 3,15 0.69 0.57 Block 5,15 2.96 0.05* Time 4,12 34.19 <0.01* Time*treatment 12,32.04 3.76 <0.01* Time*block 20,40.75 1.78 0.06

91

Table 3-9. Contingency table describing the proportion of trees that grew or did not grow between 2 and 2.7 yrs after initial fertilization (censuses 5 and 6), by tree size class, in the four nutrient addition treatments. The odds ratio refers to the likelihood of a tree growing in a given treatment relative to the control. In each cell total number, percent of total and percent for that category are shown.

# of trees Odds Ratio Size class no yes Total

5-10 cm control 26

11.82 40.63

38 17.23 59.38

64

+N 14 6.36 32.56

29 13.18 67.44

43 1.42

+P 13 5.91 24.07

41 18.64 75.93

54 2.16

+NP 12 5.45 20.34

47 21.36 79.66

59 2.68

10-30 cm control 34

10.30 34.69

64 19.39 65.31

98

+N 25 7.58 40.32

37 11.21 59.69

62 0.79

+P 23 6.97 26.44

64 19.39 73.56

83 1.44

+NP 28 8.44 33.77

55 16.67 66.27

87 1.04

>30 cm control 5

5.88 22.73

17 20.00 77.27

22

+N 1 1.18 7.14

13 15.29 92.86

14 3.82

+P 4 4.71 15.38

22 25.88 84.62

26 1.60

+NP 2 2.35 8.71

21 24.71 91.30

23 3.09

92


# of trees Odds Ratio Size class no yes Total

Total control 67

9.96 34.54

127 18.87 65.46

194

+N 43 6.39 33.33

86 12.78 66.67

129 1.05

+P 42 6.24 28.83

132 19.91 71.17

176 1.65

+NP 42 6.24 23.86

134 19.91 76.14

174 1.68

93

Table 3-10. Results from Pearson chi-square tests for seedling variables measured 1yr

and 2 yrs after fertilization. Survival= proportion of seedlings surviving, Growth = proportion of seedlings that grew, Number of leaves = proportion of seedlings that increased the number of leaves, and Herbivory = proportion of seedlings that showed evidence of leaf herbivory. Significant effects are signaled with an asterisk.

Parameter Time df n Treatment χ2 Probability

Survival 1 yr 3 476 2.81 0.42 2 yrs 3 375 9.86 0.02* Growth 1 yr 3 396 4.36 0.23 2 yrs 3 213 9.43 0.02* Number leaves 1 yr 3 387 2.23 0.53 2 yrs 3 217 13.86 <0.01* Herbivory 1 yr 3 406 4.26 0.23 2 yrs 3 219 2.11 0.55

Table 3-11. Results from repeated measures MANOVAs for foliar (leaves and sticks

<2mm diameter), reproductive (flowers and fruits), and coarse litterfall. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk. Plots 17 and 22 excluded from analysis (n= 22 plots).

Sample dfn,d F Probability

Foliar litterfall Treatment 3,13 0.14 0.93 Block 5,13 1.21 0.36 Time 2,12 16.02 <0.01* Time*treatment 6,24 0.59 0.34 Time*block 10,24 1.12 0.39

Reproductive litterfall Treatment 3,13 0.12 0.95 Block 5,13 1.21 0.36 Time 2,12 15.72 <0.01* Time*treatment 6,24 1.23 0.33 Time*block 10,24 0.90 0.54

Coarse litterfall Treatment 3,13 3.09 0.06 Block 5,13 1.18 0.37 Time 3,11 1.94 0.18 Time*treatment 9,27 0.58 0.80 Time*block 15,30 0.88 0.59

94

Table 3-12. Results from two-way ANOVA analyses for litterpool fractions. Foliar = leaves and sticks <2mm diameter, reproductive = flowers and fruits, large sticks= wood pieces and sticks > 2mm diameter, and total= the sum of all these plus surface roots.

Parameter df Treatment Block F Probability F Probability

Foliar 3,5 0.34 0.79 1.51 0.25 Reproductive 3,5 0.69 0.58 0.44 0.82 Large sticks 3,5 2.30 0.13 1.49 0.26 Total 3,5 0.31 0.82 1.77 0.19

Table 3-13. Results from repeated measures MANOVAs for fine (<2mm diameter) and

large (>2mm diameter) root biomass collected at 0-15 cm depth. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

Sample dfn,d F Probability

Fine root biomass Treatment 3,15 1.18 0.35 Block 5,15 3.41 0.03* Time 2,14 23.18 <0.01* Time*treatment 6,28 1.35 0.27 Time*block 10,28 1.40 0.23

Large root biomass Treatment 3,15 0.05 0.98 Block 5,15 4.68 <0.01* Time 2,14 2.43 0.12 Time*treatment 6,28 0.84 0.55 Time*block 10,28 1.55 0.17

95

Te

mp

era

ture

( C

)

15

20

25

30

35

Pre

cip

ita

tio

n (

mm

)

200

400

600

800

B 2008

Tem

pera

ture

( C

)

15

20

25

30

35

Pre

cip

itation (

mm

)

200

400

600

800

A 2007

(c) 2009

Month

Jan

Feb Mar

Apr

May Ju

nJu

lAug

Sep O

ctNov

Dec

Te

mp

era

ture

( C

)

10

15

20

25

30

35

Pre

cip

ita

tio

n (

mm

)

200

400

600

800

C 2009

Figure 3-1. Monthly average maximum and minimum air temperatures (dashed lines)

and precipitation (solid lines) at the study site for the three years encompassing the study. Arrows indicate fertilizer applications. Data collected by staff at the EARTH Meteorological station in the main campus.

96

Figure 3-2. Floristic description of the study site including (A) total basal area for each of

the study plots (and blocks) for trees >10 cm DBH at the beginning of the experiment, and (B)importance value indices (relative frequency + relative density + relative basal area) for the most important 20 species of trees (>10 cm DBH) found at EARTH forest. For full species list and complete species name refer to Table C-1.

A

B

97

Figure 3-3. Distribution of plots and blocks within the EARTH forest reserve. Each block

is composed of a control plot (circles), a plot where 100 kg ha-1yr-1 of nitrogen was added (crosses), a plot where approximately 50 kg ha-1yr-1 of phosphorus were added (triangles) and a plot where both nitrogen and phosphorus were added together (squares). Plot sizes are shown enlarged for illustrative purposes.

98

pH

3.8

3.9

4.0

4.1

4.2

4.3

4.4

4.5

(a)

Tota

l C

(%

)

3.0

3.5

4.0

4.5

5.0

5.5

Control

+N

+P

+NP

(b)

DIN

(ug g

-1)

6

8

10

12

14

16

18

20

22

24

26

28

(c)T

ota

l N

(%

)

0.40

0.42

0.44

0.46

0.48

0.50

0.52

0.54

Years after fertilization

Pre Fert. 1 yr 2 yr

Melic

h P

(ug g

-1)

0

1

2

3

4

5

6

7


Pre Fert. 1yr 2 yr

Tota

l P

(ug g

-1)

1000

1200

1400

1600

1800

2000

2200

(d)

(e) (f)

Figure 3-4. Mean + SE soil parameters, including (A) pH, (B) percent total carbon,

(C)dissolved inorganic nitrogen, (D) percent total nitrogen, (E) Melich extractable phosphorus, and (F) total phosphorus for soils collected at a depth of 0-10 cm in four different fertilization treatments before, 1 yr and 2 yrs after initial fertilization.

A B

C D

E F

99

% d

iffe

rence

in

pH

0

5

10

15

20

C

+N

+P

+NP

% d

iffe

rence

in

DIN

(u

g g

-1 )

-100

-50

0

50

100

% d

iffe

rence

in

to

tal C

(%

)

-40

-20

0

20

40

% d

iffe

rence

in

to

tal N

(%

)

-40

-20

0

20

40

Depth (cm)

0-10 10-30 30-50

% d

iffe

rence

in

Me

lich

P(u

g g

-1)

-40

-20

0

20

40

60

80

100

120

Depth (cm)

0-10 10-30 30-50

% d

iffe

rence

in

to

tal P

(ug g

-1)

-10

-5

0

5

10

15

20

(a) (b)

(c) (d)

(e) (f)

aa

b

b

b

aa

b

Figure 3-5. Mean + SE percent change in various soil parameters two years after

fertilization for samples collected at three different depths. Different letters represent significant differences among treatments.

A B

C D

E F

100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9 B 10-30 cm dbh

B

asa

l are

a incre

ase

(m

2 h

a-1

yr--1

)

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35A 5-10 cm dbh


0.0 0.5 1.0 1.5 2.0 2.5 3.0

Ba

sa

l are

a incre

ase

(m

2 h

a-1

yr-1

)

0.0

0.2

0.4

0.6

0.8

1.0

1.2 C >30 cm dbh


0.0 0.5 1.0 1.5 2.0 2.5 3.00.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

D All size classes

Control

+N+P+NP

Figure 3-6. Mean (+ SE) total basal area increase per treatment for trees within (A) 5-10

cm DBH, (B) 10-30 cm DBH, (C) >30 cm DBH and (D) all size classes combined, for the four nutrient addition treatments measured between 0.4 and 2.7 yrs after initial fertilization. Values for all trees per plot were summed and plots for each treatment averaged.

101

Control

+N+P+NP

B 10-30 cm dbh

Ave

rag

e p

rop

ort

ion o

f tr

ee

gro

wth

(% tre

es tha

t g

rew

yr-

1 p

lot-

1)

0

20

40

60

80

100

A 5-10 cm dbh


0.0 0.5 1.0 1.5 2.0 2.5 3.0

Ave

rag

e p

rop

ort

ion o

f tr

ee

gro

wth

(% tre

es tha

t g

rew

yr-

1 p

lot-

1)

0

20

40

60

80

100

C >30 cm dbh


0.0 0.5 1.0 1.5 2.0 2.5 3.0

D All size classes

Figure 3-7. Mean (+ SE) percentage of trees that grew per plot for trees within (A) 5-10

cm DBH, (B) 10-30 cm DBH, (C) >30 cm DBH and (D) all size classes combined, for the four nutrient addition treatments measured between 0.4 and 2.7 yrs after initial fertilization.

102

Treatment

Control +N +P +NP

RG

R (

ln (

mm

yr

-1))

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2C > 30 cm dbh

RG

R (

ln (

mm

yr

-1))

-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5A 5-10 cm dbh

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

B 10-30 cm dbh

Treatment

Control +N +P +NP-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5D All size classes

combined

Figure 3-8. Box plots of relative growth rates (RGR) measured between 0.4 and 2.7 yrs

after initial fertilization, in the four nutrient addition treatments. The three size classes are shown separately (A-C) and then combined (D).

103

RG

R (

ln(m

m y

r-1))

0.0

0.2

0.4

0.6

0.8

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

RG

R (

ln(m

m y

r-1))

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Treatment

Control +N +P +NP

RG

R (

ln(m

m y

r-1))

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

Treatment

Control +N +P +NP-1.0

-0.5

0.0

0.5

1.0

1.5

2.0

2.5

A Dendropanax arboreus

F3,17 = 0.08, p = 0.97B Goethalsia meiantha

F3,46 = 3.48, p = 0.02

C Inga

F3,21 = 0.53, p = 0.66D Pentaclethra macroloba

F3,89 = 0.99, p = 0.40

E Protium

F3,39 = 0.79, p = 0.51F Socratea exohrriza

F3,152 = 3.71, p = 0.01

* *

* **

Figure 3-9. Box plots showing relative growth rates for six common tree species,

measured between 0.4 and 2.7 yrs. after initial fertilization, in the four nutrient addition treatments. Shown are also results from 1-way ANOVAs with treatment as dependent variable. An Asterisk indicates that the mean for the treatment differs from the control (Dunnett’s test).

104

% s

ee

dlin

gs a

live

0

20

40

60

80

100

1yr

2yr

% s

ee

dlin

gs t

ha

t gre

w

0

20

40

60

80

100

Treatment

Control +N +P +NP

% s

ee

dlin

gs in

cre

ase

d n

o.

lea

ve

s

0

20

40

60

80

100

Treatment

Control +N +P +NP

% s

ee

dlin

gs w

ith

he

rbiv

ory

0

20

40

60

80

100

(a) (b)

(c) (d)

Figure 3-10. Mean (+ SE) percent of seedlings (A) alive, (B) with increased stem length,

(C) with increased number of leaves, and (D) with evidence of herbivory for the four nutrient addition treatments one and two years after initial fertilization.

A B

C D

105

Fo

liar

litte

rfa

ll (M

g C

ha

-1 y

r-1)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

Control

+N

+P

+NP


1 yr 2 yr 3 yr

Re

pro

du

ctive

litte

rfa

ll (M

g C

ha

-1 y

r-1)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

(a)

(b)

Figure 3-11. Mean (+ SE) foliar litterfall production (foliage + sticks <2 mm in diameter) (A) and reproductive litterfall production (flowers and fruits) (B) for the four nutrient addition treatments during the first, second and third year after initial fertilization.

A

B

106

Interval after fertilization (yrs)

0.4-0.8 0.9-1.2 1.3-1.6 1.7-2.0

Co

ars

e litte

rfa

ll (M

g C

ha

-1 y

r-1)

0

2

4

6

8

10

12

14

Control

+ N

+ P

+ NP

Figure 3-12. Mean (+ SE) coarse litterfall production (palm fronds and woody debris >2 mm diameter for the four nutrient addition treatments in four intervals between 0.4 and 2 yrs. after initial fertilization.

107

Fin

e r

oo

ts (

Mg C

ha

-1)

0.5

1.0

1.5

2.0

2.5

3.0

(a)


Pre Fert. 1 yr 2 yr

Larg

e r

oo

ts (

Mg C

ha

-1)

0

1

2

3

4

5

(b)

Control+N+P+NP

Figure 3-13. Mean (+ SE) fine root biomass (roots <2mm diameter) (A) and large root

biomass (roots >2mm diameter) (B) collected at 0-15cm depth for the four nutrient addition treatments before, 1yr, and 2yrs after initial fertilization.

A

B

108

Treatment

Control +N +P +NP

Root P

rod

uctivity (

Mg C

ha-1

yr-1

)

0.0

0.2

0.4

0.6

0.8

Figure 3-14. Box plot (25th percentile, median and 75th percentile) for root production

from ingrowth cores installed from 0-10cm depth and left in the field for two years for the four different treatments.

109

CHAPTER 4

EFFECT OF NUTRIENT ADDITIONS ON FOLIAR, LITTER AND ROOT CHEMISTRY

Introduction

Tissue nutrient concentrations have been used extensively as indicators of plant

nutritional status (e.g., Stone 1968, Bowen and Nambiar 1984, Drechsel and Zech

1991, Marschner 1995).Relative abundances of nitrogen (N) and phosphorus (P) in

leaves have been used to infer nutrient limitation of terrestrial net primary productivity

(NPP; Koerselmann and Meuleman 1996, Güsewell 2004, Reich and Oleskyn 2004,

McGroddy et al. 2004, Ågren 2008, Cleveland et al. 2011). To grow and accumulate

biomass through photosynthesis (i.e., NPP) plants need proteins, which are largely

constituted by N. To construct these proteins, however, plants need ribosomes and

ATP, which are largely constituted of P. Thus, recent studies on ecological

stoichiometry have proposed that plants require a defined ratio of N to P (usually 14-16

mass based) and that deviations from this ratio indicate a relative abundance (or

limitation) of one of the nutrients relative to the other (Sterner and Elser 2002, Ågren

2004). For example, across large spatial scales, foliar N:P ratios increase towards the

tropics, which suggests that at low latitudes N is relatively more abundant than P. Lower

N:P ratios in high latitudes, however, indicate that P is relatively more abundant than N.

These differences in N:P ratios have been used to infer that in the tropics, NPP is likely

limited by P and in the temperate systems by N (McGroddy et al. 2004, Reich and

Oleksyn 2004, Hedin 2004, Kerkhoff et al. 2005).

Other factors in addition to soil nutrient supply, however, can influence tissue

nutrient concentrations (and their ratio). For example, seasonality, life history traits,

species identity, and herbivory, have been shown to influence foliar N:P ratios (reviewed

110

by Ågren 2008). Thus, in some communities, N:P ratios co-vary predictably with

relative nutrient abundance or limitation; but in other cases, intrinsic physiology and

external environmental factors decouple plant N:P ratios from relative nutrient

availability (e.g., Townsend et al. 2007). Therefore, interpretation of foliar (and other

tissue) N:P ratios can be challenging because of partitioning of control between nutrient

availability and other environmental and physiological factors (Ågren 2008).

One approach to elucidate the relationship between NPP and tissue nutrient

concentrations in a given system is to “calibrate” N:P values by comparing them to NPP

responses to fertilization (Vitousek et al. 2010, Cleveland et al. 2011). However, only a

few studies have tested if in practice, N:P ratios correlate with responses to nutrient

additions in natural systems (e.g., Koerselmann and Meuleman 1996, Vitousek and

Farrington 1997, Ostertag 2001). Moreover, there are few studies that have explored

the variation in responses to nutrient additions among tree species or tree size classes

(Ostertag 2010). In this study, I conducted a fertilization experiment in a wet, lowland

tropical forest in Costa Rica to explore if trees respond to nutrient additions by changing

their tissue (foliar, litter and root) N and P concentrations, and if these responses relate

to relative growth rates (RGR) of trees.

Although most research has focused on foliar nutrients, N and P concentrations

in litterfall can provide information regarding the nutritional status of plant communities.

High N and P concentrations in green leaves are typically associated with higher

concentrations in litter (Kobe 2005, Hättenschwiler et al. 2008, Wood et al. 2009).

However, the few fertilization studies conducted in the tropics have demonstrated,

regardless of nutrient limitation to NPP, that litterfall P is more likely to increase with P

111

fertilization than litterfall N with either N or P fertilization (Tanner et al. 1992, Vitousek

1998, Vitousek 2004, Kaspari et al. 2008, Wright et al. 2011). This can be partly

attributed to a higher increase in green leaf P than in green leaf N after fertilization

(Ostertag 2010). Relative differences in nutrient resorption efficiency–the withdrawal of

nutrients from senescing plant tissue and the transport of those nutrients to other plant

tissues (Killingbeck 1996) – can also affect litterfall N and P concentrations. For

example, in low P soils, P resorption is generally higher than N resorption, making

N:Plitter larger than N:Pgreen (e.g., Kerkhoff et al. 2005, Richardson et al. 2008, Wood

et al. 2011). Thus, interpretation of N:P litter in fertilization experiments must be

conducted within the context of the responses observed for N:Pgreen and for individual

N and P concentrations.

Root N and P concentrations have not been widely used as indicators of nutrient

limitation but they are generally correlated positively with foliar nutrient concentrations in

natural systems (Kerkhoff et al. 2006, Ågren 2008, Elser et al. 2010). Consistent with

green leaves, there is a decline in root N:P ratio with latitude, although this relationship

is exponential rather than linear (Yuan et al. 2011). Root responses to fertilization, in

contrast, have not always been consistent with aboveground responses (Wright et al.

2011, Yavitt et al. 2011 but see Ostertag 2001). Fundamental physiological,

morphological and functional differences between leaves and roots may be the cause

for such discrepancy. For example, high N demand for RuBisCO enzyme (Ribulose-1,5-

bisphosphate carboxylase oxygenase) in leaves –but not in roots– can lead to a greater

N concentration per unit carbon lost in dark respiration for leaves than roots and can

also lead to green leaves requiring more N relative to P than fine roots (Reich et al.

112

2008). This in turn may result in differences in N to P stoichiometry between these two

types of tissues. However, in a compilation from various studies that included multiple

sites, Yuan et al. (2011) found no significant differences among mean C:N:P ratios of

green leaves (1,212:16:1), litter (1, 016:19: 1), or live roots (1,053:16:1).

I conducted a full factorial NP fertilization experiment at the EARTH (Escuela de

Agricultura de la Región del Trópico Húmedo) University Forest Reserve, Costa Rica.

There, I investigated whether foliar, litter, and root N and P concentrations increased

after fertilization, and whether these increases were related to tree growth indices. In

Chapter 3, I reported that 2.5yrs after fertilization, small trees (5-10 cm diameter) had a

higher growth in plots where both N and P were added simultaneously (+NP treatment)

but there were no differences among treatments in litterfall production, or root biomass

and production. Thus, I did not expect to observe community-wide responses in foliar,

litter, and root N and P concentrations when N or P were added individually (+N or +P

treatments). However, I expected a potential increase in nutrient concentrations in the

+NP treatment. Moreover, I expected to find mean foliar N:P ratios between 14-16,

indicating N and P co-limitation (Sterner and Elser 2002, Ågren 2004). I expected litter

and root nutrients to track foliar nutrient concentrations (Ågren 2008).

Lowland tropical forests are characterized by a high variation in foliar and litter

chemistry (Townsend et al. 2007, Hättenschwiler et al. 2008). However, there is little

information on how this variation influences plant-soil-microbial feedbacks related to

nutrient limitation (Townsend et al. 2008). To explore how nutrient additions impact

different ecological groups within the ecosystem, I compared the effect of fertilization on

foliar N and P concentrations in trees from two size classes (trees with diameter at

113

breast height (DBH) between 5-10cm were considered “small trees” and trees with DBH

>10 cm were considered “large trees”). For small trees, I expected an increase in foliar

nutrient concentrations with fertilization because trees in this subcanopy size class are

usually light limited, which prevents them from incorporating added nutrients as new

growth (Lambers et al. 1998). In contrast, I expected that large trees, which have

access to light in the canopy, would incorporate added nutrients as new growth

immediately after each addition (probably as leaves or fruits). I therefore did not expect

an increase in foliar nutrient concentrations in large trees.

I investigated six taxa-specific responses to fertilizer additions including the palm,

Socratea exohrriza and the legume Pentaclethra macroloba (Table 3-1). Overall, I

expected that nutrient concentration responses would be consistent with tree growth

responses (e.g., Vitousek 2004). Thus, I expected Socratea to increase foliar P

concentrations after P additions because this species showed a significant increase in

growth after P additions (Chapter 3). In contrast, I did not expect legumes in general

and Pentaclethra (the most abundant species) in particular, to increase foliar N or P

concentrations after nutrient additions because neither “legumes” nor Pentaclethra

showed a RGR response after N or P additions.

Methods

Experimental Design

The study was conducted at the Forest Reserve of the EARTH University (Escuela

de Agricultura de la Región del Trópico Húmedo), in Guácimo, Limón, Costa Rica (10°

11’ N and 84° 40’ W). This private reserve is located approximately 30 m above sea

level and consists of 900 ha of mature and regenerating wet forest and wetlands. Mean

annual temperature is 25.1 °C and mean annual precipitation (MAP) is 3,464 mm. A

114

complete site description is provided in Chapter 3. In May 2007, I established 24 30 x

30 m plots and assigned them randomly to three fertilizer treatments or a control in a

complete block design (n = 6). Three blocks were located at the Rancho site and three

at the Rio site (Figure 3-3). Besides the control plots, the three treatments included +P

(47 kg ha-1yr-1 of P as super triple phosphate), +N (100 kg ha-1yr-1 of N applied as

ammonium nitrate and urea), and +NP (N and P added together in quantities as in +N

and +P plots). Fertilizer was broadcast by hand twice a year on the surface of the 900

m2 plots. All measurements were restricted only to the central 400 m2 of each plot (20 x

20 m) to reduce edge effects. In this plots, various soil measurements and

measurements of tree stem growth, litterfall productivity, root productivity and biomass,

were conducted (Chapter 3).

Foliage, Litterfall and Root Collection

I collected foliage, litterfall and root samples between July 2007 and September

2009 in the experimental plots at EARTH Forest (described in Chapter 3). In each plot I

collected fully expanded, sun-leaves from common tree species using a pole pruner, or

crossbow with a bolt affixed with monofilament line. One sample was composed of a

group of at least ten leaves per tree placed in a bag. Whenever possible, I attempted to

collect samples from the same taxa as described in Chapter 3 (Dendropanax arboreus,

Goethalsia meiantha, Pentaclethra macroloba, Socratea exohrriza, Inga sp., Protium sp.

and Virola sp.). Overall, I collected foliar samples from 286 trees representing 36

genera and 46 species. When possible, samples were collected from the same 4-8

large trees (>10cm DBH) and 4-8 small trees (4-9 cm DBH) per plot on three occasions:

prior to fertilization, 1 yr, and 2 yrs after initial fertilization between the months of July

and September. When it was not possible to collect leaves from the same trees

115

(because a tree had died, lost its leaves or showed significant herbivory), foliage was

collected from another tree in the same plot and size category.

Litterfall was collected using polyvinyl chloride (PVC) traps as described in

Chapter 3. From the litterfall collected during the 2 yr study, three representative

subsamples were used to conduct chemical analyses. The first subsample (pre-

fertilization) included foliar litterfall collected between 23 August and 9 September 2007.

The second subsample (1 yr after fertilization) included foliar litterfall collected between

8 October and 5 November 2008 and the third subsample (2 yrs after fertilization)

included foliar litterfall collected between 9 July and 10 August 2009. Each of these

subsamples was separated into species, when possible, and the rest of the leaves

analyzed collectively.

Roots were collected in each plot using a pound core, as described in Chapter 3,

in three occasions: prior to fertilization, 1 yr and 2 yrs after initial fertilization. In addition,

roots obtained from ingrowth cores installed at the beginning of the experiment and

retrieved after two years in the field (Chapter 3), were also included in the chemical

analyses.

Chemical Analysis

To conduct chemical analyses, foliar, litterfall and root samples were dried at 60

°C and ground using a Wiley Mill (Thomas Scientific, Swedesboro, New Jersey, U.S.A.)

passed through a #40 screen or a coffee grinder. I measured total percent N and C with

an elemental analyzer (ECS 4010, Costech Analytical, Valencia, California,U.S.A.). I

measured P using an ash digestion (Jones and Case 1996) followed by colorimetric

determination of ortho-phosphate using a spectrophotometer microplate reader

116

(PowerWave XS Microplate Reader, Bio-Tek Instruments Inc., Winooski, Vermont,

U.S.A.).

To include nutrient concentrations in an area basis, and to test for changes in

leaf physical properties with fertilization, I calculated specific leaf area (SLA) for foliar

samples collected 1yr and 2 yrs after initial fertilization. I scanned five leaves from each

foliage sample using an image scanner and measured leaf area using Beta 4.0.3, Scion

Image software. I then obtained leaf dry weight for each sample and calculated SLA (g

cm-2) as dry weight (g)/leaf area (cm2).

Statistical Analysis

To test the effect of fertilization treatments on foliar, litter and root chemistry, I

used plot-averaged values (n = 24) in repeated measures MANOVAs with time specific

measurements as dependent variables, and treatment and block as independent

variables. I selected this approach over univariate repeated measures ANOVA because

in some cases the sphericity assumption was not met (Field 2009). When there was a

treatment*time interaction, I conducted a one-way ANOVA on each time point to

determine when there was a significant treatment effect. To test for treatment effects on

foliar nutrients from individual species, I used individual trees as the sampling unit and

did not include a block effect because there were not enough replicates from each

species in all blocks. To test the relationship between plot-averaged foliar nutrients and

soil parameters described in Chapter 3, I used simple linear regressions.

When analyzing litterfall data, it was challenging to find samples from a given

species in each plot and across the three time points. Therefore, I analyzed separately

only Pentaclethra macroloba (the most common species) and a species group (termed

“Tiliaceae”), which included primarily Goethalsia meiantha with traces of Apeiba

117

membranaceae or Luehea seemannii. Because of small sample sizes for these two

groups (Pentaclethra and Tiliaceae), I conducted T-tests for each time point and

grouped the treatments in +N (+N and +NP plots) versus -N (control and +P plots),

when comparing litterfall %N and in +P (+P and +NP plots) versus -P (control and +N

plots), when comparing litterfall P concentrations.

When analyzing the response of root P to fertilization (using repeated measures

MANOVA), I excluded data from plot 19. This plot was an outlier that had root P values

76% higher than the total average (mean root P including controls and fertilized plots =

0.56 + 0.02 mg g-1versus plot 19 P = 0.97 mg g-1). In all figures, tables, and text, means

(+ standard errors) are shown. All figures were constructed using Sigmaplot 12.0 and

analyses were conducted in JMP 8.0(SAS Institute Inc., Cary, NC, USA).

Results

Foliar Nutrients

Two years after initial fertilization, there was no difference among treatments in

plot-averaged foliar N or P concentrations but there was a significant difference among

blocks for both N and P (Table 4-1 and Figure 4-1F). Block effects were mainly driven

by differences among the two study sites. Mean foliar N was higher in the Rio site than

in the Rancho site before (T = 3.31, P < 0.01) and 2 yrs after fertilization (T = 2.89, P

<0.01). Mean foliar P was similar between sites before fertilization (t = 1.93, P = 0.07)

but higher in the Rio site 2 yrs after fertilization (T = 2.32, P = 0.03). Interestingly, this

difference disappears when considering foliar P by leaf area (T = 0.57, P = 0.57),

probably because SLA was slightly higher at the Rancho site (SLA Rancho = 89.72 g m

-2 versus SLA Rio = 87.27 g m-2). Finally, there was no significant difference among

treatments for foliar N (treatment F3,23 = 0.38, P = 0.77; block F5,23 = 0.76, P = 0.60) or P

118

(treatment F3,23 = 0.89, P = 0.47; block F5,23 = 0.72, P = 0.63) in the change of foliar

nutrients 2 yrs after initial fertilization.

Overall, large trees (>10cm DBH) had higher foliar P (T = 2.28, P = 0.02) but not

N (T = 0.082, P = 0.41) concentrations than small trees (5-10 cm DBH), both before and

2 yrs years after initial fertilization. In addition, 2 yrs after initial fertilization, large trees

had drastically higher N and P per unit leaf area than small trees (N: T = 6.89, P <0.01,

P: T = 6.72, P <0.01). I did not measure foliar nutrients by unit area prior to fertilization.

Two years after initial fertilization, large trees had higher mean foliar N in the +NP

treatment, although this result was not consistent across blocks (Table 4-1 and Figure

4-1C). Mean Foliar P, on the contrary, differed significantly among treatments only for

small trees, although this effect was likely driven by pre-fertilization differences among

treatments (Table 4-1 and Figure 4-1B). Overall, there was high variability in foliar N:P

ratios (Figure 4-2) and there was no difference across treatments, tree sizes, or sites

after fertilization (Table C-2).

I compared foliar nutrients among four species and two genera where replicates

were sufficient to conduct statistical analyses (Table 3-1). Foliar N, P and N:P ratios

differed significantly among taxa (Table 4-2) but only one species (Pentaclethra

macroloba) and one genus (Protium) differed among treatments in foliar N or P. Two

years after initial fertilization, the species Pentaclethra macroloba had higher foliar N in

the +NP treatment, (Table 4-3, Figure 4-3D; one-way ANOVA at 2 yrs post fertilization:

F3,41 = 3.08, p = 0.04). One year after initial fertilization, Protium sp. trees had higher

foliar N in the +N treatment (Table 4-3, Figure 4-3E) but this effect disappeared 2 yrs

after initial fertilization, probably due to the destruction of two +N plots in the Rio site

119

during a strong wind event (see methods section, Chapter 3). Although not statistically

significant, 2 yrs after initial fertilization the palm, Socratea exohrriza, had 15% and 19%

higher mean foliar P relative to the control in the +P and +NP treatments, respectively

(Table 4-4, Figure 4-4F). Foliar N:P ratios differed among treatments for Dendropanax

but this result was based on one individual and should therefore be interpreted with

caution (Figure 4-5A). For the rest of the study species N:P ratios did not differ among

treatments, blocks or sampling times (Figure 4-5; Table C-3).

Because community-averaged foliar nutrients were not strongly influenced by

fertilization additions, I conducted linear regressions on several soil parameters

measured before and 2 yrs after initial fertilization against foliar N and P to evaluate

which soil conditions were good predictors of foliar nutrient concentrations. Overall,

foliar nutrients were best predicted by soil pH, net nitrification rates and soil P (both

Melich and total P; Figure 4-6 and 4-7). Foliar N was negatively related to net

nitrification (Figure 4-6C) but this relationship disappeared with the fertilization

treatments (Figure 4-6D). Interestingly, soil P was not a good predictor of foliar %N

before fertilization (Figure 4-6E) but became positively correlated 2 yrs after fertilization

(Figure 4-6F). In contrast, soil P was a good predictor of foliar P before fertilization

(Figure 4-7E) but this relationship disappeared 2 yrs after initial fertilization (Figure 4-

7F). None of the other measured soil parameters were good predictors for foliar N or P.

Finally, mean foliar N:P ratios were not related to any of the soil parameters measured

(Figure C-1).

Litterfall Nutrients

Overall, litterfall nutrient concentrations were high (Figure 4-8) and varied over

time. This time effect probably indicates differences in the relative contribution of

120

species to the litter samples at different collection times, or differences in environmental

conditions among years. Mean litterfall N did not differ among treatments or blocks;

therefore I eliminated “block” from the model. When the non-significant block effect was

removed, the treatment effect became significant (Table 4-5). Plots where N was added

had higher litterfall N concentrations than plots where no N was added (Figure 4-8).

There was no significant treatment effect on mean litterfall P. However, there was a

significant block effect, which was caused by higher litterfall P in the Rio side (MANOVA

by side F1,19 = 13.34, P<0.01). There were no differences among treatments or blocks

over time in litterfall N:P ratios (Table 4-5), and mean foliar N:P ratios were not related

to mean litterfall N:P ratios (r2 = 0.01, P = 0.65). In addition, mean foliar nutrients (N and

P) were not related to mean litterfall nutrients either before or 2 yrs after initial

fertilization. Both Pentaclethra and Tiliaceae increased their litter N concentrations with

N additions but there was no increase in litter P after P additions (Figure 4-9).

Root Nutrients

There was no significant difference among treatments in root N but roots in the

+P and +NP treatments had higher P concentrations 2 yrs after initial fertilization (Table

4-6, Figure 4-10). There was also a significant block effect, which reflects the higher

root P in the Rio site, mainly 2 yrs after fertilization (mean P Rancho = 0.50 + 0.02 mg g-

1 and mean P Rio = 0.62+ 0.03mg g-1; T = 3.13 P< 0.01). This higher root P in the Rio

side also resulted in a significant block effect when comparing N:P ratios (Table 4-6,

Figure 4-10C). Roots that had grown into ingrowth cores over the course of the

experiment did not show differences among treatments or blocks in N (treatment F3,22 =

0.52, P = 0.67, block F5,22 = 0.85, P = 0.54) or P (treatment F3,22 = 3.08, P = 0.42, block

F5,22 = 0.45, P = 0.80) concentrations.

121

Prior to fertilization, there was a positive relationship between root N and P

concentrations (r2 = 0.41, d.f. = 23, P < 0.01) but one (r2 = 0.10, d.f. = 23, P = 0.10) and

two years (r2 = 0.01, d.f. = 23, P = 0.82) after fertilization this relationship disappeared.

However, in ingrowth cores installed at the beginning of the experiment and extracted 2

yrs after fertilization, there was a positive relationship between N and P concentrations

(r2 = 0.20, d.f. = 23, P = 0.03). Overall, root N could not be predicted by any soil

parameter (data not shown) but root P was strongly positively correlated to both Melich

P and total P before (Melich P vs. root P r2 = 0.45, d.f. = 23, P < 0.01; total P vs. root P

r2 = 0.31, d.f. = 23, P < 0.01) and 2 yrs after fertilization (Melich P vs. root P r2 = 0. 70,

d.f. = 23, P < 0.01; total P vs. root P r2 = 0.56, d.f. = 22, P < 0.01). In addition, root P

was positively correlated with foliar P before (r2 = 0.22, d.f. = 23, P = 0.02) and 2 yrs

after fertilization (r2 = 0.24, d.f. = 23, P = 0.02).

Discussion

Effects of Fertilization on Foliar Nutrients

Overall, both N and P concentrations at EARTH Forest were high relative to other

tropical forests where responses to fertilization were observed. In a P limited forest in

Hawaii (Kokee), mean foliar P was 0.55µg g-1 and increased to 1.88µg g-1 after long-

term fertilization. In my study plots, mean foliar P before fertilization was 1.25µg g-1.

This value is even higher than foliar P after long-term P fertilization at a N- limited forest

in Hawaii (Foliar P = 0.95µg g-1; Harrington et al. 2001). Foliar N concentrations at

EARTH Forest were high as well, with a mean of 2.65% and values up to 5%. High

foliar nutrient concentrations (reported here), in addition to high soil N and P values, and

a lack of response in total DBH increase or litterfall production after fertilization (Chapter

122

3), suggest that aboveground net primary productivity (ANPP) at EARTH forest is

probably not limited strongly by N or P availability.

Although most NPP components did not significantly change with N or P additions,

I observed a 37.4 % increase in mean tree RGR with +NP additions, which could be

interpreted as NP co-limitation (Chapter 3). Therefore, I expected to observe a mean

foliar N:P ratio between 14 and16 (mass based). However, mean foliar N:P ratio was

22.13, which is considered indicative of P limitation (Sterner and Elser 2002, Güsewell

2004, Ågren 2004, 2008). In fact, this value is higher than mean N:P ratio at Osa

Peninsula (16.4 + 4.7), a low P forest in the south of Costa Rica (Townsend et al. 2007).

Likely, the high N:P ratio at EARTH Forest reflects a high legume abundance (which are

high in N), and suggests that in this diverse forest other factors, such as species identity

are more important controls on leaf stoichiometry than relative nutrient supply.

Consistent with other tropical forests, there was great variation across species in

foliar nutrient concentrations (Townsend et al. 2007, Hattënschwiler et al. 2008; Figure

4-2). Species mean foliar P, for example, ranged from 0.67 to 2.16µg g-1, which is 57%

of the range of values for five lowland tropical forests combined (~ 0.25 – 2.75µg g-1;

Townsend et al 2007). This high variability reduces statistical power to detect

differences in foliar nutrient concentrations after fertilization. However, I conducted

measurements before, 1 yr and 2 yrs after fertilization, mostly in the same trees, and

thus it is likely that a strong and consistent across-species response to fertilization

would have been evident regardless of the variability (e.g., Ostertag 2010).

Alternatively, if there were strong but contrasting responses among species or even

among trees, these could have veiled any community response. In fact, my data

123

suggest that although at the community-level there may be no apparent response of

mean foliar nutrient concentrations to fertilization and therefore no apparent nutrient

limitation, there may be subtle and contrasting responses from individual taxa within the

community, which could have important consequences for ecosystem processes, such

as litter decomposition and nutrient turnover rate (Campo and Dirzo 2003, Vitousek

2004).

Influence of Tree Size on Foliar Nutrients

Large trees responded to fertilization by increasing their foliar nutrient

concentrations but not their RGR, and small trees responded by increasing their RGR

but not their foliar nutrients (Figure 4-1 and Chapter 3). These results indicate that,

contrary to expectations, small trees allocated the extra nutrients to growth while large

trees showed “luxury consumption” after fertilization and reveal a difference in nutrient

use strategies among age groups (Wright et al. 2011). In large trees, there was a 6.7%

increase in foliar N with +NP additions but no foliar P response. The significant increase

in foliar N after fertilization was unexpected given that high foliar N concentrations have

been associated with high levels of herbivory (Huberty and Denno 2006 but see Campo

and Dirzo 2003). Furthermore, after long term (> 10 yrs) fertilization, foliar P is more

likely to respond to changes in nutrient availability than foliar N, and it has been argued

that concentrations of foliar P are more plastic than concentrations of foliar N (Vitousek

2004, Ostertag 2010). I suspect that this apparent paradox is driven by the species-

specific response by the most abundant tree species in this forest, Pentaclethra

macroloba, which responded to fertilization by increasing foliar N after +NP additions.

124

Influence of Taxa on Foliar nutrients

Individual trees of Pentaclethra macroloba (mainly >10 cm DBH individuals or

“large trees”), had approximately 10% higher foliar N concentrations in the +NP

treatment, relative to the control. Interestingly, there was no difference in foliar N

between the +N treatment and the control. This response could be interpreted as N and

P co-limitation (Elser et al. 2007, Townsend et al. 2008); although in that case I would

have expected a simultaneous increase in foliar P in the +NP treatment. Possibly, this

species is allocating nutrients to different tissues, such as N to leaves and P to roots. I

did not separate roots by species, but there was a significant increase in mean root P

both in the +P and +NP treatments. To test this hypothesis, species-specific analyses of

nutrient concentrations in roots, leaves, and reproductive structures, should be

conducted.

The most abundant palm species at EARTH forest, Socratea exohrriza, was the

only taxon (from those observed individually) to show a strong response to fertilization

by increasing both RGR (Chapter 3) and foliar P concentrations in the +P and +NP

treatments. This can be interpreted as: (1) from the observed taxa, Socratea was the

only species clearly limited by P (sensu Chapin et al. 1986) or (2) Socratea has specific

life-history traits that influenced its responsiveness. For example, the response of

Socratea could have been influenced by its single stem architecture with no branches

(where a growth response would be missed by DBH measurements), relatively high

growth rate, and the formation of dense, superficial root mats (which would be beneficial

for fast nutrient uptake after fertilization (Henderson et al. 1995).It is likely that a

combination of both P limitation and life history traits, influenced the response of this

palm to fertilization. Interestingly, if Socratea was limited by P, as explained by the RGR

125

foliar P response, I would expect this species to have a high N:P ratio, at least relative

to other taxa in the same site. However, Socratea’s mean foliar N:P ratio was the lowest

among the studied taxa (16.93 + 0.37). This inconsistency suggests that in this system,

foliar N:P ratios may be more influenced by inherent species traits than by resource

availability (Townsend et al. 2007). Most importantly, these contrasting responses from

two of the most abundant tree species at EARTH Forest illustrate the importance of

considering species composition and life history traits when making interpretations

concerning nutrient limitation in diverse tropical forests (Townsend et al. 2007, Vitousek

et al. 2010, Cleveland et al. 2011).

Effects of Fertilization on Litterfall and Root Nutrients

Consistent with foliar nutrient responses to fertilization, there was an increase in

litterfall N concentrations in the +NP treatment. However, in contrast with foliar N

concentrations, which only increased when N and P were added together, litterfall N

also increased in the +N treatment. As postulated above, it is likely that the foliar

response was driven by the species-specific response of Pentaclethra. The litterfall

response, on the other hand, was determined by all leaves that fell in litter traps, and is

probably more representative of the community response to fertilization. This 13%

increase in litterfall N is equivalent to ~30 Kg N ha-1 yr-1 and is likely to have an impact

on decomposition and nutrient cycling (Aerts 1997, Hobbie and Vitousek 2000 but see

Kaspari et al. 2008). However, as for foliar N concentrations, most studies in the tropics

have reported a larger response in litter P than litter N after fertilization (Mirmanto et al.

1999, Hobbie and Vitousek 2000, Kaspari et al. 2008) and therefore, this result

deserves further investigation.

126

Contrary to foliage and litterfall, roots showed a significant increase in P

concentrations after P (and NP) fertilization. There was a 12% increase in root P with P

additions, which is equivalent to 0. 09 kg P ha-1 yr-1, calculated using root productivity

data obtained from root ingrowth cores (Chapter 3). Therefore, although the magnitude

of the response in litterfall N was similar to the response in root P (13% and 12%

increase respectively), lower root P concentrations and productivity make the additional

P input in roots significantly smaller than the additional N input in litterfall. To predict the

impact that these changes in nutrient concentrations will have on ecosystem dynamics,

further investigation that addresses the differences in decomposition rates between

litterfall and roots and potential seasonality of these responses, should be conducted

(e.g., Ostertag and Hobbie 1999, Cusack et al. 2009).

Inconsistent responses to fertilization between root and foliar concentrations can

be indicative of a decoupling between aboveground and belowground nutrient limitation

(Ostertag 2001, Kaspari et al. 2008, Wright et al 2011, Yavitt et al 2011). Further

measurements of belowground processes, including soil respiration, enzyme activities,

and microbial biomass and community composition, would be necessary to test this

hypothesis. Moreover, it has been shown that multiple nutrients regulate belowground

processes, such as decomposition rates, in a seasonal forest in Panama (Kaspari et al.

2008). Thus, future research focusing on belowground dynamics would ideally include

fertilization with other nutrients, in addition to N and P (Chapter 5).

Total Soil P as a Driver of Tissue Nutrient Concentrations

Traditionally, total soil P, or P that can only be extracted by strong acid assays

(Hedley et al. 1982), has been considered of no practical importance in tropical soils

(Sanchez 1976). Because these soils are generally highly weathered with high iron and

127

aluminum content, most P is found in crystalline or occluded forms associated with

secondary minerals (Cross and Schlesinger 1995, Tiessen 1998) and is considered

unavailable for plant use. Although this might be true in the short term, the tight

correlation between total soil P and foliar and root P observed across plots in this study,

suggests that total soil P is a strong predictor of foliar and root P at local (200 m2)

spatial scales and that in the long term some of this “occluded P” becomes available to

plant use (Turner and Engelbrecht 2011). Moreover, there was a strong correlation

between total P and available P (Chapter 3), and across a precipitation gradient

available P was a good predictor of mean foliar P (Chapter 2). In both the precipitation

gradient study and the fertilization study, foliar N could not be predicted by any soil

parameter and there was no strong relationship between soil N and P. Thus, at least for

these forests in Costa Rica, it could be stated that total P is a good indicator of the P

available for plant uptake. The lack of relationship between foliar P and tree growth,

however, deserves further investigation.

Conclusions

Results from this chapter support NPP results reported in Chapter 3, which

demonstrates that at EARTH Forest there is no strong limitation by N or P but that there

is some evidence of co-limitation by these two nutrients. However, the mechanism

driving this pattern is probably not a simultaneous scarcity of N and P but rather

“heterogeneous nutrient limitation” within the tree community. This is illustrated by the

foliar N increase in large Pentaclethra trees in the +NP treatment, and the RGR and

foliar P increase in Socratea palms in the +P and +NP treatments. Contrary to results

obtained in monospecific forests in Hawaii (Vitousek 2004), in this diverse tropical

forest, responses to experimental nutrient additions appear to be driven by the

128

interaction of functional traits (or at least species-specific traits) and resource availability

(Chapin, Vitousek and Van Cleve 1986). For example, N:P ratios at EARTH forest

reflect the high abundance of legumes (Pentaclethra in particular) more than the

nutritional status of the plant community. This highlights the importance of considering

both species identity and relative species abundance when making predictions related

to nutrient limitation in this forest. Moreover, the relative importance of a species-

specific response for the community-wide signal is going to depend not only on the

strength of the response but also on the relative abundance of that species in the

community.

Although it is clear that N:P ratios can provide insight into the relative availability of

these nutrients at larger scales (Koerselmann and Meuleman 1996, Güsewell 2004,

Reich and Oleskyn 2004, McGroddy et al. 2004, Ågren 2008, Cleveland et al. 2011), it

has been noticed that at local scales, variation in N:P ratios may reflect more

differences in intrinsic physiology than in the external environment (Townsend et al.

2007, Ågren 2008). This appears to be the case in my study; therefore it is advisable

that for diverse tropical forests, nutrient limitation assertions that result from

interpretation of N:P ratios are made at the functional group, genera, or species level,

rather than generalizing nutrient limitation from a single mean value.

Finally, further research is needed to explore how the contrasting responses to N

versus P additions observed in this experiment will affect nutrient cycling and carbon

storage in the long-term. For example, increased litter N and root P concentrations

could have drastically different effects by differentially changing nutrient turnover times,

microbial community composition, soil respiration, and/or greenhouse gas emissions. A

129

mechanistic understanding of how changes in nutrient availability affect nutrient uptake

and storage in plant tissues is critical for predicting the role that tropical forests will have

in mitigating increased N and P fluxes, which are caused by human alterations of global

biogeochemical cycles (Galloway and Cowling 2002, Galloway 2004, Okin et al. 2004,

Mahowald et al. 2008).

130

Table 4-1. Results from repeated measures MANOVAs for foliar chemistry by tree size class. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

dfn,d F Probability

Foliar %N 5-10 cm Treatment 3,15 2.60 0.09 Block 5,15 0.81 0.56 Time 2,14 0.80 0.47 Time*treatment 6,28 0.28 0.94 Time*block 10,28 0.57 0.82 > 10 cm Treatment 3,15 3.35 0.04* Block 5,15 8.22 <0.01* Time 2,14 0.52 0.61 Time*treatment 6,28 3.42 0.01* Time*block 10,28 0.94 0.51 Total Treatment 3,15 1014 0.41 Block 5,15 5.57 <0.01* Time 2,14 0.17 0.85 Time*treatment 6,28 1.73 0.15 Time*block 10,28 0.78 0.65 Foliar P (mg g-1) 5-10 cm Treatment 3,15 4.02 0.03* Block 5,15 2.58 0.07 Time 2,14 2.46 0.12 Time*treatment 6,28 2.25 0.07 Time*block 10,28 1.16 0.36 > 10 cm Treatment 3,15 0.72 0.56 Block 5,15 2.35 0.09 Time 2,14 1.66 0.23 Time*treatment 6,28 0.50 0.80 Time*block 10,28 0.65 0.76 Total Treatment 3,15 0.42 0.74 Block 5,15 3.79 0.02* Time 2,14 3.94 0.04* Time*treatment 6,28 1.36 0.27 Time*block 10,28 1.06 0.43

131

Table 4-2. Results from repeated measures MANOVAs comparing foliar %N, P (mg g-1) and N:P ratios among the six most common taxa (see table 3-1 for details). F-values for taxa and time were obtained from exact tests but time*taxa interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

dfn,d F Probability

Foliar %N Taxa 5,107 78.50 < 0.01* Time 2,106 0.22 0.81 Time*taxa 10,212 1.82 0.06 Foliar P (mg g-1) Taxa 5,99 5.68 < 0.01* Time 2,98 2.78 0.07 Time*taxa 10,196 2.98 < 0.01* Foliar N:P ratios Taxa 5,98 52.67 < 0.01* Time 2,97 3.41 0.04* Time*taxa 10,194 2.70 < 0.01*

132

Table 4-3. Results from repeated measures MANOVAs for foliar %N by species. F-values for treatment and time were obtained from exact tests but time*treatment interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

dfn,d F Probability

Dendropanax arboreus Treatment 3,6 3.71 0.08 Time 2,5 2.76 0.15 Time*treatment 6,10 2.46 0.10 Goethalsia meiantha Treatment 2,3 0.53 0.64 Time 2,2 0.91 0.52 Time*treatment 4,4 0.90 0.54 Inga Treatment 3,10 0.68 0.58 Time 2,9 0.18 0.84 Time*treatment 6,18 0.83 0.56 Pentaclethra macroloba Treatment 3,22 0.62 0.61 Time 2,21 2.16 0.14 Time*treatment 6,42 2.65 0.03* Protium Treatment 3,19 0.25 0.86 Time 2,18 0.76 0.48 Time*treatment 6,36 2.27 0.05* Socratea exohrriza Treatment 3,30 0.44 0.72 Time 2,29 3.73 0.04* Time*treatment 6,58 0.30 0.94

133

Table 4-4. Results from repeated measures MANOVAs for foliar P by species. F-values for treatment and time were obtained from exact tests but time*treatment interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

dfn,d F Probability

Dendropanax arboreus Treatment 3,6 2.24 0.18 Time 2,5 1.51 0.31 Time*treatment 6,10 1.61 0.24 Goethalsia meiantha Treatment 2,3 2.43 0.24 Time 2,2 0.59 0.63 Time*treatment 4,4 3.26 0.18 Inga Treatment 3,9 0.20 0.90 Time 2,8 1.33 0.32 Time*treatment 6,16 1.41 0.27 Pentaclethra macroloba Treatment 3,20 0.80 0.51 Time 2,19 1.48 0.25 Time*treatment 6,38 0.66 0.68 Protium Treatment 3,19 2.23 0.11 Time 2,18 1.15 0.34 Time*treatment 6,36 0.53 0.77 Socratea exohrriza Treatment 3,25 0.79 0.51 Time 2,24 3.18 0.05* Time*treatment 6,48 2.01 0.08

134

Table 4-5. Results from repeated measures MANOVAs for litterfall chemistry. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

dfn,d F Probability

Litterfall %N Treatment 3,15 2.43 0.111

Block 5,15 2.16 0.11 Time 2,14 15.13 <0.01* Time*treatment 6,28 1.97 0.10 Time*block 10,28 01.16 0.36 Litterfall P (mg g-1) Treatment 3,15 2.66 0.09 Block 5,15 2.98 0.04* Time 2,14 11.68 <0.01* Time*treatment 6,28 0.78 0.72 Time*block 10,28 0.56 0.83 Litterfall N:P Treatment 3,15 1.23 0.34 Block 5,15 0.63 0.68 Time 2,14 8.05 <0.01* Time*treatment 6,28 0.30 0.93 Time*block 10,28 0.85 0.59

1 Treatment effect becomes significant if non-significant block effect is removed from analysis

(Time*treatment F6,38 = 2.37 p= 0.04).

135

Table 4-6. Results from repeated measures MANOVAs for root chemistry. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

dfn,d F Probability

Root %N Treatment 3,15 0.81 0.51 Block 5,15 1.14 0.38 Time 2,14 5.56 0.02* Time*treatment 6,28 1.47 0.22 Time*block 10,28 1.15 0.36 Root P (mg g-1) Treatment 3,14 2.29 0.12 Block 5,14 3.35 0.03* Time 2,13 31.9 <0.01* Time*treatment 6,26 4.32 <0.01* Time*block 10,26 2.11 0.06 Root N:P Treatment 3,15 0.93 0.45 Block 5,15 3.20 0.04* Time 2,14 11.38 <0.01* Time*treatment 6,28 1.31 0.29 Time*block 10,28 1.34 0.26

136

%N

2.3

2.4

2.5

2.6

2.7

2.8

2.9

3.0A 5-10cm

P (

mg

g-1

)

1.0

1.1

1.2

1.3

1.4

1.5Control

+N+P+NP

B 5-10cm

%N

2.3

2.4

2.5

2.6

2.7

2.8

2.9

3.0C >10cm

P (

mg

g-1

)

1.0

1.1

1.2

1.3

1.4

1.5


Pre Fert. 1 yr 2 yr

%N

2.3

2.4

2.5

2.6

2.7

2.8

2.9

3.0

D >10cm

E All (f) All


Pre Fert. 1yr 2 yr

P (

mg

g-1

)

1.0

1.1

1.2

1.3

1.4

1.5F All

Figure 4-1. Mean (+ SE) foliar N and P for trees with DBH between 5-10 cm (A, B),

larger than 10cm (C, D) and both size classes combined (E, F), for the four nutrient addition treatments measured before, 1yr and 2 yrs after initial fertilization.

137

All t

empe

rate

tree

s

All t

ropi

cal tre

es

Res

erva

Sam

uel

Cau

axi

EARTH

Pre

Fer

t.

EARTH

2yrs

Fo

liar

N:P

0

10

20

30

40

50

Figure 4-2. Box Plot comparing variability in N:P ratios at EARTH forest before

fertilization and 2yrs after fertilization with other sites. Redrawn with permission from Townsend et al. (2007). Boxes denote median with 50th and 75th percentiles.

138

Fo

liar

%N

1.5

2.0

2.5

3.0

3.5

4.0

4.5A Dendropanax

arboreus

Control

+N+P+NP

B Goethalsia meianthaF

olia

r %

N

1.5

2.0

2.5

3.0

3.5

4.0C Inga


Prefert 1 yr 2 yr

Fo

liar

%N

1.5

2.0

2.5

3.0

3.5

4.0

4.5

D Pentaclethra macroloba

E Protium (f) All


Prefert 1 yr 2 yr

F Socratea exohrriza

Figure 4-3. Mean (+ SE) foliar %N for six common tree species for the four nutrient

addition treatments measured before, 1yr and 2 yrs after initial fertilization.

139

Fo

liar

P (

mg

g-1

)

0.8

1.0

1.2

1.4

1.6

1.8

2.0A Dendropanax

arboreus

Control

+N+P+NP

B Goethalsia meianthaF

olia

r P

(m

g g

--1)

0.8

1.0

1.2

1.4

1.6

1.8

2.0

C Inga


Prefert 1 yr 2 yr

Fo

liar

P (

mg

g--1

)

0.8

1.0

1.2

1.4

1.6

1.8

2.0


E Protium (f) All


Prefert 1 yr 2 yr


Figure 4-4. Mean (+ SE) foliar P for six common tree species for the four nutrient

addition treatments measured before, 1yr and 2 yrs after initial fertilization.

140

Fo

liar

N:P

10

15

20

25

30

35 A Dendropanax

arboreus

Control

+N+P+NP

B Goethalsia

meianthaF

olia

r N

:P

10

15

20

25

30

35 C Inga


Prefert 1 yr 2 yr

Fo

liar

N:P

10

15

20

25

30

35


E Protium (f) All


Prefert 1 yr 2 yr


Figure 4-5. Mean (+ SE) foliar N:P ratios for six common tree species for the four

nutrient addition treatments measured before, 1yr and 2 yrs after initial fertilization.

141

Soil pH

3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8

Folia

r %

N

2.3

2.4

2.5

2.6

2.7

2.8

2.9

3.0

2

7

8

14

19

24

1

10

1213

1722

35

11

18

20

23

4

6

9

15

1621

Soil pH

3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8

2

7

8

14

19

24

1

10

12

131722

3

511

18

2023

46 15

21

Net Nitrification (ug N g-1 d

-1)

-2 -1 0 1 2 3 4 5

Folia

r %

N

2.3

2.4

2.5

2.6

2.7

2.8

2.9

3.0

2

7

8

14

19

24

1

10

1213

1722

35

11

18

20

236

9

15

1621

-2 -1 0 1 2 3 4 5

2

7

8

14

19

24

1

10

12

13 1722

3

511

18

2023

46 915

21


-1)

Melich P (ug g-1)

0 1 2 3 4 5 6 7

Folia

r %

N

2.3

2.4

2.5

2.6

2.7

2.8

2.9

3.0

2

7

8

14

19

24

1

10

1213

1722

35

11

18

20

23

4

6

9

15

1621

0 1 2 3 4 5 6 7

2

7

8

1424

1

10

12

13 1722

3

511

18

2023

469 15

Melich P (ug g-1)

A R2 = 0.45, P < 0.001

B R2 = 0.17 P = 0.04

C R2 = 0.16 P = 0.05 D R

2 = 0.01 P = 0.89

E R2 = 0.10 P = 0.13 F R

2 = 0.21 P = 0.02

Figure 4-6. Relationship between several soil variables and plot-averaged foliar %N

before fertilization (A, C, E) and 2 yrs after initial fertilization (B, D, F). Symbols represent each plot and colors each treatment (Control= grey, +N = pink, +P= green, +NP = Cyan).

142

Soil pH

3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8

Folia

r P

(m

g g

-1)

1.0

1.1

1.2

1.3

1.4

1.5

1.6

2

7

814

1924

1

10

12

1317

223511

1820

234

69

15

16

21

Soil pH

3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8

2

78

14

19

24110

12

13

17223

5

11

18

20

23

4

615

16

21


-1)

-2 -1 0 1 2 3 4 5

Folia

r P

(m

g g

-1)

1.0

1.1

1.2

1.3

1.4

1.5

1.6

2

7

814

1924

1

10

12

1317

2235

11

1820

23

6 9

15

16

21

-2 -1 0 1 2 3 4 5

2

7

814

1924

1

10

12

1317

223511

1820

234

6 9

15

16

21


-1)

Melich P (ug g-1)

0 2 4 6 8 10 12 14

Folia

r P

(m

g g

-1)

1.0

1.1

1.2

1.3

1.4

1.5

1.6

2

7

814

1924

1

10

12

1317

223511

1820

234

69

15

16

21

0 2 4 6 8 10 12 14

2

78

14

19

24110

12

13

17223

5

11

18

20

23

4

6915

16

21

Melich P (ug g-1)

A R2 = 0.14, P = 0.07

B R2 = 0.01 P = 0.82

C R2 = 0.05 P = 0.29 D R

2 = 0.02 P = 0.49

E R2 = 0.19 P = 0.04 F R

2 = 0.15 P = 0.07

Figure 4-7. Relationship between several soil variables and plot-averaged foliar P (mg

g-1) before fertilization (A, C, E) and 2 yrs after initial fertilization (B, D, F). Symbols represent each plot and colors each treatment (Control= grey, +N = pink, +P= green, +NP = Cyan).

143

Litte

rfall

%N

2.0

2.1

2.2

2.3

2.4

2.5

2.6

2.7

Litte

rfa

ll P

(m

g g

-1)

0.7

0.8

0.9

1.0

1.1

1.2

Control

+N

+P

+NP

(a)


Pre Fert. 1 yr 2 yr

Litte

rfa

ll N

:P

20

22

24

26

28

30

32

34

(b)

(c)

Figure 4-8. Mean (+ SE) litterfall chemistry, including N concentration (A), P

concentration (B), and N:P ratios for the four nutrient addition treatments measured before, 1yr and 2 yrs after initial fertilization.

A

B

C

144

Figure 4-9. Mean litterfall nutrient concentrations by taxa, for the most common species

(A and C) and a common family (B and D) found in litterfall samples collected before, 1yr and 2 yrs after initial fertilization. Empty bars represent plots where no N (in A and B) or no P (in C and D) was added and filled bars represent plots where N (in A or B) or P (C and D) was added.

145

Ro

ot %

N

1.3

1.4

1.5

1.6

1.7

1.8

1.9

2.0

Ro

ot

P (

mg g

-1)

0.3

0.4

0.5

0.6

0.7

Control

+N

+P

+NP

(a)


Pre Fert. 1 yr 2 yr

Ro

ot

N:P

20

24

28

32

36

40

44

(b)

(c)

Figure 4-10. Mean (+ SE) root chemistry, including N concentration (A), P concentration

(B), and N:P ratios (C) for the four nutrient addition treatments measured before, 1yr and 2 yrs after initial fertilization. For root P analyses, an outlier (plot 19) was excluded (see methods for details).

C

B

A

146

CHAPTER 5 CONCLUSIONS AND LESSONS LEARNED

Conclusions

The purpose of this study was to investigate how environmental and biological

factors, such as mean annual precipitation (MAP) and tree species composition, can

influence nutrient limitation and availability in Costa Rican lowland tropical forests. I

used both large-scale observational and fine-scale experimental approaches to explore

patterns and mechanisms related to nutrient use. In the observational gradient study

(Chapter 2), I described patterns of soil and foliar Nitrogen (N) and Phosphorus (P)

across nine forest sites in Costa Rica. Overall, the results suggest that across the

precipitation gradient, soil N is relatively more abundant than P but it is also more

sensitive to changes in MAP. In addition, foliar P is a better predictor of soil P than foliar

N of soil N.

In the experimental fertilization study (Chapters 3 and 4), I tested N or P

limitation in a lowland tropical wet forest. After two years of fertilization treatment, no

significant effect of either N or P fertilization was detected on tree diameter increase,

litterfall productivity, or root biomass or productivity. However, there were interesting

and contrasting responses among tree species and size classes. For example, the

legume species, Pentaclethra macroloba, the most abundant tree in this forest, did not

grow more with N or P additions. Instead, this species increased mean foliar N

concentration when N and P were added collectively, indicating luxury consumption of N

and resulting in increased litterfall-N concentrations. In contrast, individuals from

Socratea exohrriza, the most abundant palm species in this forest, showed larger mean

diameter increase and higher mean foliar P concentration after P additions, which

147

suggest that this species is P limited (results are summarized in Table 5-1). Several

conclusions can be drawn from these results:

The Myth of P Limitation in the Tropics

Multiple lines of evidence suggest that tropical forests occurring in old soils are

limited by P (e.g., Walker & Syers 1976, Vitousek 2004, Reich & Oleksyn 2004,

Kerkhoff et al. 2004). However, tropical forests occur on multiple soil types (Townsend

et al. 2008), some of which are not old. Thus, the generalization that “tropical forests are

limited by P” is misleading. Though N was relatively more available than P, both in the

nine forests included in the observational gradient study and at EARTH Forest (where

the fertilization experiment was conducted), there was no distinct indication of P

limitation. Phosphorus limitation, as operationally described (Chapin et al. 1986), would

have required an increase in net primary productivity after P additions, but this was not

the case in my experiment. Moreover, other indicators, such as foliar nutrient

concentrations and stoichiometry (Güsewell 2004, Reich & Oleskyn 2004, McGroddy et

al. 2004, Ågren 2008, Cleveland et al. 2011), in most cases were more influenced by

species identity or by other environmental variables, than by nutrient availability. Thus,

either these forests are not strongly limited by P, or they are P limited but the research

approaches used in this study could not detect P limitation. If the latter is true, we need

to devise better techniques for accurately identifying nutrient limitation in a specific

system, within a reasonable time frame.

Heterogeneous Nutrient Limitation

Second, and probably the most relevant conclusion from this study, is the

assertion that in lowland tropical diverse forests a plant community is not a black box of

trees reacting in tandem to nutrient availability. On the contrary, “community nutrient

148

limitation” (Vitousek et al. 2010, Harpole et al. 2011) is a heterogeneous process,

resulting from the competing and conflicting responses of different biological and

biochemical processes. In the case of this study, I would expand to say that at EARTH

Forest there is “heterogeneous nutrient limitation”, not only driven by variability in

nutrient responses among tree species but also among size classes. This heterogeneity

highlights the importance of considering different aspects of the plant community, such

as forest structure and species composition, when making predictions concerning

nutrient limitation in these forests. Furthermore, the differential response of size classes

and species suggests that changes in nutrient availability could lead to changes in

forest structure or even diversity in the long term, and could have important implications

for plant-soil-microbial feedbacks as they relate to nutrient limitation.

Environmental and Biological Processes Influence Nutrient Limitation and Carbon Cycling

Finally, this study has shown that complex feedbacks and interactions among

environmental and biological factors make it difficult to predict how long-term changes in

climate (e.g., MAP) or plant resources (e.g., nutrient availability) will influence nutrient

limitation in lowland tropical forests. However, results from the gradient study (Chapter

2) suggest that a decrease in MAP in certain forests of Costa Rica, which has been

predicted by several climate models (Neelin et al. 2006), could lead to an increase in

available N from mineralization in the wetter sites of the gradient. Although I did not

measure P mineralization, it could be expected that results would be similar for P

availability (Schuur 2001). Thus, in wet sites (such as EARTH Forest), a decrease in

MAP could result in an increase in available nutrients (i.e., N and P).

149

Results obtained in the fertilization experiment (Chapters 3 and 4) suggest that at

EARTH Forest, an increase in N and P could cause an increase in foliar and litter N in

Pentaclethra macroloba (the dominant species). Returning to the model used in Chapter

1 (Figure 1-1) these results could be incorporated into a positive feedback where an

increase in available nutrients feeds back positively through an increase in foliar and

litter N that could further increase N availability at EARTH forest. This feedback,

however, would not happen for P availability because I did not record an increase in

foliar P in Pentaclethra after N or P additions. However, I did record an increase in foliar

P in the most abundant palm, Socratea (but did not measure litter P for this species).

Assuming that an increase in foliar P in Socratea resulted in an increase in litter P, then

the relative rate (or strength) of the feedback would be dependent on specific life history

traits of the two contrasting species, such as leaf life span, litter quality, and litterfall

productivity. If these rates differ between species, this could lead to a temporary

decoupling of N and P cycling, caused by heterogeneous responses within the tree

community (Figure 5-1).

Lessons Learned

To date, full factorial fertilization experiments have been viewed as the best way to

directly test nutrient limitation in a land ecosystem (Chapin et al. 1986, Vitousek and

Howarth 1991, Tanner et al. 1992, Vitousek 2004, Elser et al. 2007, LeBauer and

Treseder 2008, Cleveland et al. 2011). However, probably due to the methodological

and logistical challenges associated with these types of experiments, only a handful of

fertilization experiments have been conducted in the tropics (Table 1-1). After facing

many challenges during the course of my doctorate research, I have several

150

suggestions for improving study design and expanding working hypotheses in

fertilization experiments in diverse tropical forests.

Though funding and labor may be prohibitive, I suggest including as many

experimental treatments as possible, depending on the hypotheses for a particular site.

Given recent data from Panama (Kaspari et al. 2008, Wright et al. 2011), at least

potassium (K), and a treatment with micronutrients, should be included in a fertilization

experiment. An initial complete commercial soil analysis would provide a baseline to

formulate hypotheses regarding which nutrient may limit productivity at a given site.

When establishing a fertilization experiment in a tropical forest, I suggest planning

for a long-term experiment, acknowledging that although several biological processes

(e.g., soil respiration) may respond in a short term, other processes (e.g., tree stem

diameter increase) may exhibit limited responses during the first years. During the initial

2-3 years of study, I suggest collecting background data on as many parameters as

possible, and then running power analyses on all variables to verify that samples sizes

are sufficient to capture the immense heterogeneity characteristic of these forests

(Townsend et al. 2008, Field 2009). If certain parameters, such as soil nitrate, are too

variable to obtain reliable estimates, then they should be eliminated or improvements

made to the methodology in order to reduce the variability. During the initial

experimental stage, I suggest conducting P fractionations and P sorption curves (Lajtha

et al. 1999) to evaluate how much P fertilizer has to be added to increase available P by

a significant proportion (e.g., 10%). At EARTH forest, I suspect that a larger proportion

of “total P”, than captured by Melich extractions, was available to the plant community

thereby making the amount of fertilizer added in this study a smaller fraction of the

151

“available pool”. Thus, although time consuming, I recommend extensive P

measurements prior to fertilization (e.g., Sato and Comerford 2006).

I suggest including measurements of belowground processes, such as soil

respiration and microbial community ecology, because these processes are likely to

respond faster to changes in nutrient availability than tree growth or litterfall production

(e.g., Cleveland and Townsend 2006). To capture a representative sample of the litter

productivity in plots, I suggest installing both permanent and portable litter traps in each

plot. The portable traps should be moved to different sectors of the plot monthly or bi-

monthly to capture litter from trees not included in the footprint of the permanent traps.

Although excruciatingly slow and difficult, it would be greatly beneficial to separate

litterfall by species, at least two months per year, or at least from focus species. Ideally,

these months would coincide with green-leaf collection for chemical analysis. During

these months, the portable litter traps can be installed under target trees where foliar

samples are being collected. This design would enable calculation of retranslocation

measurements (Chapin et al. 2002). In my experiment, green-leaf N and P values were

obtained from specific trees within the study plots. However, litter N and P values were

obtained from samples collected in permanent litter traps. Because variability in foliar

nutrients between trees and between collection times was so large, and the litter

samples were not from the same trees or times as the green-leaf samples, I did not feel

comfortable calculating retranslocation values from these samples.

Finally, one of the most interesting and challenging aspects of this study were the

differential responses of tree species and size classes to N and P additions, or

“heterogeneous limitation” (Chapter 3). It is widely recognized that the traditional

152

definition of nutrient limitation can break down when applied across wide differences in

phylogeny, or climate (Chapin et al. 1986, Vitousek and Howarth 1991, Vitousek et al.

2010). Yet, there have been limited recommendations on how to address the situation.

Originally, Chapin et al. (1986) stated that “community nutrient limitation” should be

tested by comparing relative growth rates (RGR) of same species across a nutrient

availability gradient. For example, one could measure the maximum possible growth of

a group of species in a high P system. Then, one could measure RGR of the same

group of species in lower P sites after adding P fertilizer. If RGRs reached the levels of

the high P system, those species (or community) were P limited in the lower P sites.

This approach was successfully used in Hawaii, where valuable research on this topic

has been conducted (summarized in Vitousek 2004). However, this approach is

impractical in the continental tropics because of incredibly high species diversity, and

the difficulty in identifying natural nutrient gradients. Thus, the question remains: what is

the best way to test nutrient limitation in a diverse tropical forest?

Based on my experience at EARTH Forest, there is no simple solution to advice of

the optimal design for studying nutrient limitation in a diverse tropical forest. However,

there are several options depending on the objective of the individual study. If the main

objective is to explore how carbon (C) storage and cycling is affected by nutrient

availability in a forest, then not all tree species are weighted equally. At least in the short

term, only the most “important” species (species with higher importance value index

(IVI) values; Chapter 3) will control net primary productivity (NPP). For example at

EARTH Forest, although there were more than 100 species of trees within my study

plots (Table C-1), large trees (>10cm DBH) from the four most-important species

153

(Pentaclethra, Socratea, Goethalsia, and Dypterix) comprised 40% of the total IVI of the

forest. Thus, to test how NPP is influenced by N and P availability in this forest, one

could focus on these trees instead of on the entire tree community. In contrast, if the

objective of the study was to explore how different life history strategies interact with

nutrient availability to control C cycling, then it is advisable to focus on species that are

representative of those groups such as N fixers versus non-N fixers, light demanding

versus shade tolerant, etc. In any case, my study highlights that regardless of the

objective of the study, to understand nutrient limitation in a diverse tropical forest, it is

crucial to acknowledge diversity and select methodologies and hypotheses that

incorporate heterogeneity of responses within the community.

154

Table 5-1. Summary of f responses ratios (RR) of treatments relative to the control in the fertilization experiment. Measurements were taken two years (or 2.7 yrs for RGR) after fertilization. Only RRs of statistically significant tests are shown. Refer to chapters 3 and 4 for measurement details and statistical tests. RR were calculated as ln (control/treatment).

Response Group

Variable Treatment

+N +P +NP

All trees RGR 0.34 Foliar N Foliar P Litterfall N 0.12 0.10 Litterfall P Root N Root P 0.22 0.15 Root N:P Large trees RGR Foliar N Foliar P Small Trees RGR 0.62 Foliar N Foliar P Pen Mac RGR Foliar N 0.09 Foliar P

Large RGR

PenMac Foliar N 0.09 Foliar P Small RGR

PenMac Foliar N Foliar P 0.12 Soc Exo RGR 1.04 0.99 Foliar N Foliar P 0.09 0.05

Large RGR Soc Exo Foliar N Foliar P 0.13 0.11 Small RGR 1.65 1.31

Soc Exo Foliar N Foliar P

155

Figure 5-1. Diagram representing how environmental and biological factors could interact to influence nutrient cycling in a diverse tropical forest. Environmental factors are shown in blue, biological factors in green, processes measured during this experiment in black and hypothesized processes in gray. MAP = mean annual precipitation, and RGR = relative growth rate. For a full description of the diagram refer to the text and Chapter 1. Model adapted from Vitousek 2004.

156

APPENDIX A

SPATIAL AND TEMPORAL VARIATION OF LIGHT REACHING THE UNDERSTORY

IN A WET TROPICAL FOREST IN COSTA RICA

Understanding how light and nutrient availability interact to affect primary

productivity is crucial to predict future effects of climate change on carbon storage

(Denslow et al. 1990, Wright et al. 2011). In this study, we explored the spatial and

temporal variability of transmitted diffuse light reaching the understory in 24, 20 x 20 m

plots that had been previously fertilized with nitrogen (N) phosphorous (P), N and P, or

left as controls in a lowland tropical wet forest in Costa Rica (see Chapter 3 for detailed

description of the experimental design). We utilized hemispherical photography to

analyze canopy structure and light reaching the understory, which can give insight into

below canopy photosynthetically active radiation (Frazer et al. 1997). Specifically, we

used transmitted diffuse light because calculations take the entire sky into account, and

this allows for the comparison of different sites (Madgwick and Brumfield 1969). The

specific objectives of this study were (1) to describe the spatial and temporal variation of

transmitted diffuse light to the understory in plots established in a wet tropical forest in

Costa Rica, and (2) to test the effect of increased nutrient availability by fertilization on

transmitted diffuse light to the understory, an indicator of canopy cover.

In February 2008, December 2008, and August 2009, we took four hemispherical

photographs in each of the 24 plots using a Nikon Coolpix 950 Limited Edition 2.11

megapixel digital camera with a fisheye lens set on a tripod, one meter above the forest

This study was conducted in collaboration with an undergraduate student, Laura Morales, and was funded through a SEAGEP- REU Minority Fellowship awarded to L. M

157

floor. All photographs were taken before 9 am or after 4 pm, to avoid distortions caused

by reflection of the sun on canopy leaves. When there was a large leaf from an

understory palm obstructing the canopy view, we removed the leaf before taking the

photograph. As an index of available light in the understory we used “transmitted diffuse

light” (which we refer to as light from here on). We used Gap Light Analyzer (GLA) free

software (http://www.ecostudies.org/gla/) to analyze our pictures. We conducted a

repeated measures MANOVA to test the effect of treatment, block and time on light. If

an interaction was significant we conducted a 1-way ANOVA to identify the time of the

effect. Statistical analyses were conducted in JMP 8.0(SAS Institute Inc., Cary, NC,

USA).

There was no significant difference in light levels among fertilizer treatments

through the course of the study. However, there was a significant increase in light with

time (Table A-1, Figure A-1). This was mainly due to the opening of several light gaps

after a particularly strong storm in August 2008. Figure A-2 demonstrates the difference

in canopy cover before and after the storm. It is not certain how this event could have

impacted the growth response of trees to fertilization. On one hand, trees exhibited

mechanical damage and a significant loss of leaves (personal observation), with an

evident peak in coarse litterfall after the event (Figure 3-12). This damage could have

repressed any fertilization effect because the trees would have probably used the extra

resources to repair damage and produce new leaves instead of increasing stem

diameter. On the other hand, the disturbance significantly increased light availability in

certain plots, which could have enhanced the response to fertilization. This is supported

158

by the significant contribution of light in explaining variation in relative growth rates of

trees larger than 10 cm DBH (Chapter 3).

Table A-1. Results from repeated measures MANOVAs for “transmitted diffuse light. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk. Plots 17 and 22 were excluded from the analysis (see text).

Parameter dfn,d F Probability

Treatment 3,13 2.68 0.09

Block 5,13 1.60 0.23

Time 2,12 18.33 <0.01

Time*treatment 6,24 0.52 0.78

Time*block 10,24 2.29 0.05


6 mo 1.5 yr 2 yr

Ln (

tra

nsm

itte

d d

iffu

se

lig

ht (m

ol m

-2 d

-1 )

)

0.0

0.2

0.4

0.6

0.8

1.0

Control

+N+P+NP

Figure A-1. Mean (+ SE) transmitted diffuse light for the four nutrient addition treatments

measured before, 1 yr and 2 yrs after initial fertilization.

159

Figure A-2. Hemispherical canopy photograph taken in the same position at three

successional dates. Intact forest is observed in (A), but during August 2008 there was a strong storm, which resulted in multiple gaps within the study plots, such as the one observed in (B) and (C).

A February 2008

2008

B December 2008

2008

C August 2009

160

APPENDIX B BIOTIC AND ABIOTIC FACTORS INFLUENCING TREE GROWTH IN A LOWLAND

TROPICAL WET FOREST: A MIXED MODEL APPROACH

The objective of this study was to explore how nutrient availability interacts with

other biotic and abiotic factors to control tree growth in a lowland tropical wet forest in

Costa Rica. A site description, experimental design, and methodology used to conduct

measurements are detailed in Chapter 3. Specifically, we tested three hypotheses:

(1) At the community level, nutrient availability limits tree “growth” in a lowland

tropical wet forest. Growth refers to relative growth rate (RGR) and is calculated as the

slope from a line traced from six log-transformed measurements of diameter at breast

height (DBH) of each tree bole, conducted over a 2.5 yr period (Chapter 3).

(2) There is an important effect of tree species on RGR response to

fertilization.

(3) There is a difference in the RGR response to fertilization depending on

tree size. We measured RGR in ten trees per plot with DBH between 5-10 cm DBH and

all trees with DBH larger than 10 cm (Chapter 3).

To test our hypotheses we fitted a series of mixed models and compared them

using AIC criteria. We used the lmer package and ML method in R (ver. 2.10.1; R

Corporation, Vienna, Austria) to calculate the AIC values and the REML method in lme

to calculate the parameter estimates for the best fitting model. The delta AIC (∆AIC)

provides a relative measurement of how good a model fits the data relative to other

The analysis included in this section was conducted in collaboration with Mollie Brooks and

included in an oral presentation at the Ecological Society of America (ESA) Meeting, Pittsburg, 2010.

161

models; the best model is assigned a value of zero and then the other models are

evaluated relative to that one, the lower the ∆AIC, the better the fit (Anderson 2008).

The specific parameters we used in our models are summarized in Table B1. We

treated both block and tree species as random effects and the rest of the parameters as

fixed effects.

To test our first hypothesis we compared model M5 (null hypothesis) against M2

and M3 (Table B-2). Delta AIC values revealed that there was a significant treatment

effect, but only in the Rio side (because there was a treatment * side interaction; Table

B-3; L-ratio test between M5 and M3 P = 0.01). The two sides represent two areas in

the forest where plots were established. Although they were separated by less than 1

km, the Rio side had more evidence of having been disturbed in the past. For example,

the Rio side was dominated by Goethalsia meinatha, a fast growing tree species

characteristic of disturbed areas (Hartshorn and Hammel 1994), which was not found in

the Rancho side. Therefore, we suspect that the stronger treatment effect in the Rio

side was caused by faster RGRs of species occurring in this area, rather than by

differences in nutrient availability among sides (Chapter 3).

To test our second hypothesis, which concerned the importance of tree species

identity to control RGRs, we compared model M8 (null hypothesis) with model M4.

There was a large decrease in ∆AIC when species was included in the model (Table B-

3; L-ratio test between M8 and M4 P <0.01). Thus, species identity contributes an

important portion of the variation in RGR at EARTH forest. However, because of the

high species diversity and low replicate trees from each species, it was difficult to detect

species-specific patterns (Figure 3-2). The canopy palm Socratea exorrhiza was one of

162

the few species where a clear fertilizer effect was observed (Figure B-2). This species

responded to fertilization by increasing RGR with P additions in the Rancho side and

with both P and NP additions in the Rio side. Differences among sides are likely due to

variation in other environmental controls, such as light or co-limitation with other soil

nutrients, potentially potassium (Wright et al. 2011).

To test our third hypothesis, which posed that initial tree size (represented by

DBH) was an important control on RGR, we compared models M4 (null hypothesis) and

M7. Initial DBH was, in fact, an important effect in determining RGR (Table B-3; L-ratio

test between M4 and M7 p <0.001). Overall, large trees grew less (relative to their size)

than small trees, regardless of fertilizer treatment. This pattern has frequently been

observed in other tropical forests (Losos and Leigh 2004, Wright et al. 2011).

From all the models tested, our best fit model was M1, which explained 18% of the

variation in RGR (Table B-3). This was a 15% improvement from our initial null model

(M8), which explained only 3% of the variation in RGR. In fact, the fit improved slightly

more if we include a three way interaction among treatment, initial DBH, and species in

the model. However, parameter estimates become non-significant due to a loss of

degrees of freedom caused by the three way interaction. Therefore we report as our

best fit model M1. From the parameter estimates (Table B-4) we conclude that RGR of

small trees was influenced by +NP additions in the Rio side. Although this analysis

showed no effect of light on tree RGR, this was probably due to methodological

constraints on light measurements (Appendix A). Integrated and accurate light

measurements are extremely difficult to obtain in tropical forests (Lambers et al. 1994).

163

Nonetheless, we suspect that these would probably contribute an important percentage

of the variability in RGR observed in this forest.

Results obtained with this analysis were consistent with those presented in

Chapter 3, where data were analyzed using repeated measures MANOVAs. One

difference is that here, we separated the “block effect” in two components, the random

block parameter and the “side” parameter. In the analyses presented in Chapter 3, all

the variation caused by this spatial heterogeneity was consolidated in the block effect.

Another difference is that because of the nature of the analysis, we could no incorporate

a block by treatment interaction in the MANOVA analyses. Here, the equivalent

treatment *side interaction proved to be important.

Table B-1. Parameters used in the mixed models. For measurement details refer to Chapter 3.

Parameter Description n

Growth Response variable. Refers to tree RGR ln( cm yr-1)

760

DBH 1 Initial DBH (cm) 760 Treatment fertilizer additions (C, N, P, NP) 4 Side side of the forest (Rio, Rancho) 2 (1 | block) Block. Treated as a random effect 6 (1| Sp.Code) Tree species. Treated as a random effect 130

Average Gap Qualitative measurement of % gap for each tree

760

Average trans. diffuse Index of available light (Appendix A) 24

165

Table B-4. Maximum likelihood estimates, their standard error, and T-value for parameters included in model M1 (Table B3), the best fitting model from those included in the analysis. A parameter estimate is considered significant if the confidence interval (estimate + std. error) does not include zero. Significant parameters are labelled with an asterisk.

Estimate Std. Error T

(Intercept) 1.08 e-02 1.77 e-03 6.10*

DBH1 -1.95 e-04 3.80 e-05 -5.13*

Treatment N: side Rancho 2.14 e-03 2.06 e-03 1.04

Treatment NP: side Rancho -9.43 e-04 2.16 e-03 -0.44

Treatment P: side rancho 1.46 e-03 2.12 e-03 0.69

Side Rio 1.37 e-03 2.14 e-03 0.64

Treatment N:side Rio -9.78 e-04 2.80 e-03 -0.35

Treatment NP: side Rio 7.92 e-03 2.93 e-03 2.71*

Treatment P: side Rio 2.47e-03 2.835e-03 0.873

Figure B-1. Box plots showing stem diameter increase (calculated as relative growth

rate, see methods) for the four fertilizer treatments, separated by “side”. There were three blocks on each of these two areas of the EARTH forest, which are separated by less than 1 km.

166

Figure B-2. Box plots showing stem diameter increase (calculated as relative growth

rate, see methods) for the most common canopy palm, Socratea exorrhiza in the four fertilizer treatments, separated by “side”.

Figure B-3. Relationship between initial diameter at breast height (DBH) and stem

diameter increase (calculated as relative growth rate, see methods).

167

APPENDIX C ADDITIONAL TABLES AND FIGURES

Table C-1. Species of trees found in the study plots. Trees were identified by Enrique

Rojas and Orlando Vargas.

Family Genus Species

Anacardiaceae Tapirira guianensis Annonaceae Anona subnubila Annonaceae Rollinia pittieri Annonaceae Unonopsis pittieri Annonaceae Xylopia sericophylla Anonaceae Guatteria sp1 Anonaceae Guatteria sp2 Anonaceae Guatteria sp3 Apocynaceae Tabernaemontana arborae c.f Aquifoliaceae Ilex skutchii Araliaceae Dendropanax arboreus Arecaceae Astrocarium alatum Arecaceae Chrysophylla warscewiczia Arecaceae Cryosophila warscewiczii Arecaceae Socratea exorrhiza Bignoniaceae Jacaranda Copaia Boraginaceae Cordia bicolor Boraginaceae Cordia lucidula Boraginaceae Cordia porcata Burseraceae Protium confusum Burseraceae Protium panamense Burseraceae Protium pittieri Burseraceae Protium ravenii Burseraceae Tetragastris panamensis Burseraceae Trattinnickia aspera Capparaceae Capparis pittieri Carycaceae Jacaratia Dolichaula Cecropicaceae Cecropia insignis Cecropicaceae Pourouma bicolor Chrysobalanaceae Licania manicarpa Clusiaceae Garcinia intermedia Dilleniaceae Doliocarpus dentatus Dilleniaceae Pinzona coriaceae Elaeocarpaceae Sloanea guianensis Euphorbiaceae Croton schiedeanus Euphorbiaceae Hyeronima alchorneoides

Table C-1. Continued

168


Fabaceae/Cae Bauhinia guinanensis Fabaceae/Mim Balizia elegans Fabaceae/Mim Inga alba Fabaceae/Mim Inga leiocalycina c.f Fabaceae/Mim Inga pezizifera Fabaceae/Mim Inga Thiboudiana Fabaceae/Mim Inga umbilifera Fabaceae/Mim Inga venusta Fabaceae/Mim Pentaclethra macroloba Fabaceae/Mim Stryphodendron microstachyum Fabaceae/Pap Dipteryx panamensis Fabaceae/Pap Dussia macroprophyllata Flacourtiaceae Cassearia arborea Flacourtiaceae Casearia commersoniana Flacourtiaceae Cassearia tacanensis Flacourtiaceae Laetia procera Flacourtiaceae Ryania especiosa Lauraceae Beilschmiedia sp Lauraceae Nectandra reticulata Lauraceae Ocotea laetevirens Lauraceae Ocotea leucoxylon Lecythidaceae Eschweilera costarricensis Lecythidaceae Eschweilera longirachis Malpighiaceae Byrsonima arthropoda Malpighiaceae Spachea correae Malvaceae Hampea appendiculata Melastomataceae Miconia affinis Melastomataceae Miconia elata Meliaceae Carapa guianensis Meliaceae Guarea guidonia Meliaceae Guarea rhopalocarpa Meliaceae Trichilia septentrionalis Moraceae Castilla elastica Moraceae Ficus maxima Moraceae Nauclopsis naga Myrcinaceae Ardisia fimbrillifera Myristicaceae Compsoneura Mexicana Myristicaceae Otoba novogranatensis Myristicaceae Virola koschnyi Myristicaceae Virola multiflora Myristicaceae Virola sebifera Myrsinaceae Parathesis trichogyne

169

Table C-1. Continued.


Myrtaceae Myrcia splendens Nyctaginaceae Neea laetevirens Piperaceae Piper colonense Quiinaceae Lacunaria panamensis Rubiaceae Coussarea hondensis Rubiaceae Faramea parvibractea Rubiaceae Miconia multispicata Rubiaceae Posoqueria latifolia Rubiaceae Warsewicsia coccinea Rutaceae Zanthoxylum panamensis Sapinadaceae Cupania sp Sapinadaceae Talisia nervosa Sapotaceae Chrysophyllum venezuelanense Sapotaceae Micropholis crotonoides Sapotaceae Pouteria calistophylla Simaroubaceae Simarouba amara Sterculiaceae Sterculia costaricana Tiliaceae Apeiba membranacea Tiliaceae Goethalsia meiantha Tiliaceae Luehea seemannii Ulmaceae Ampelocera macrocarpa Violaceae Rhinorea hummeli

170

Table C-2. Results from repeated measures MANOVAs for foliar N:P ratios by tree size class. F-values for treatment, block and time were obtained from exact tests but time*treatment and time*block interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

dfn,d F Probability

5-10 cm Treatment 3,15 0.45 0.72 Block 5,15 0.93 0.49 Time 2,14 0.52 0.61 Time*treatment 6,28 0.59 0.74 Time*block 10,28 0.48 0.89

> 10 cm Treatment 3,15 0.53 0.67 Block 5,15 0.90 0.51 Time 2,14 1.79 0.20 Time*treatment 6,28 0.23 0.96 Time*block 10,28 0.40 0.94

Total Treatment 3,15 0.47 0.71 Block 5,15 0.57 0.72 Time 2,14 0.27 0.77 Time*treatment 6,28 0.87 0.91 Time*block 10,28 0.88 0.99

171

Table C-3. Results from repeated measures MANOVAs for foliar N:P ratios by species. F-values for treatment and time were obtained from exact tests but time*treatment interactions are F-value approximations resulting from Wilk’s lambda multivariate tests. In these cases degrees of freedom (df) are approximated as well. Significant effects are signaled with an asterisk.

dfn,d F Probability

Dendropanax arboreus

Treatment 3,6 5.17 0.04* Time 2,5 0.06 0.94 Time*treatment 6,10 0.40 0.57

Goethalsia meiantha Treatment 2,3 0.10 0.90 Time 2,2 14.77 0.06 Time*treatment 4,4 6.85 0.04*

Inga Treatment 3,9 0.60 0.63 Time 2,8 1.45 0.29 Time*treatment 6,16 0.79 0.59

Pentaclethra macroloba Treatment 3,19 1.20 0.34 Time 2,18 1.71 0.21 Time*treatment 6,36 1.40 0.25

Protium Treatment 3,19 2.01 0.14 Time 2,18 0.71 0.50 Time*treatment 6,36 1.24 0.31

Socratea exohrriza Treatment 3,25 0.12 0.95 Time 2,24 0.29 0.76 Time*treatment 6,48 0.70 0.65

172

Soil pH

3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8

Folia

r N

:P

18

20

22

24

26

28

30

2

7

8 141924

1

10

12

13172235

11

18

20

23

46

9 15

16

21

Soil pH

3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8

27 814

1924

1

10

12

131722

3 5

11

18

20

23

46

15

1621


-1)

-2 -1 0 1 2 3 4 5

Folia

r N

:P

18

20

22

24

26

28

30

2

7

8141924

1

10

12

13 17 22 35

11

18

20

23

6

9 15

16

21

-2 -1 0 1 2 3 4 5

2781419

241

10

12

131722

35

11

18

20

23

46 9

15

1621


-1)

Total P (ug g-1)

500 1000 1500 2000 2500 3000 3500

Folia

r N

:P

18

20

22

24

26

28

30

2

7

8141924

1

10

12

13 172235

11

18

20

23

46

9 15

16

21

500 1000 1500 2000 2500 3000 3500

27 814

1924

1

10

12

131722

35

11

18

20

23

469 16

21

Total P (ug g-1)

(a) R2 = 0.04, P = 0.36

(b) R2 = 0.04 P = 0.36

(c) R2 = 0.01 P = 0.63 (d) R

2 = 0.04 P = 0.38

(e) R2 = 0.01 P = 0.56 (f) R

2 = 0.14 P = 0.77

Figure C-1. Relationship between several soil variables and plot-averaged foliar N:P ratios before fertilization (a, c, e) and 2 yrs after initial fertilization (b, d, f). Symbols represent each plot and colors each treatment (Control= grey, +N = pink, +P= green, +NP = Cyan).

173

Me

lich

P (

ug

g-1

)

0

1

2

3

4

5

6

7

2

7

8

14

19

24

1 10

1213

17

22

35

11

18

20

23

46

9

15

16

21

Total P (ug g-1)

500 1000 1500 2000 2500 3000 3500

Me

lich

P (

ug

g-1

)

-2

0

2

4

6

8

10

12

14

27

814

19

24

110

12131722

35

11

18

20

23

4 6

9

16

21

A R2 = 0.50, P < 0.001

B R2 = 0.73, P < 0.001

Figure C-2. Relationship between soil Total P and Melich P before fertilization (A) and 2

yrs after initial fertilization (B). Symbols represent each plot and colors each treatment (Control= grey, +N = pink, +P= green, +NP = Cyan).

174

LIST OF REFERENCES

Aerts, R. 1997. Climate, leaf litter chemistry, and leaf litter decomposition in terrestrial ecosystems: a triangular relationship. Oikos 79:439–49.

Aerts, R., and F. S. Chapin. 2000. The mineral nutrition of wild plants revisited: a re-evaluation of processes and patterns. Advances in Ecological Research 30:1–67.

Ågren, G. I. 2004. The C:N:P stoichiometry of autotrophs – theory and observations. Ecology Letters 7:185–191.

Ågren, G. I. 2008. Stoichiometry and nutrition of plant growth in natural communities. Annual Review of Ecology Evolution and Systematics 39:153–170.

Alvarez-Clare, S., and M. C. Mack. 2011. Influence of precipitation on soil and foliar nutrients across nine Costa Rican forests. Biotropica 43:433–441.

Amundson, R., A. T. Austin, E. A. G. Schuur, K. Yoo, V. Matzek, C. Kendall, A. Uebersax, D. Brenner, and W. T. Baisden. 2003. Global patterns of the isotopic composition of soil and plant Nitrogen. Global Biogeochemical Cycles 17:1031–1041.

Andersen, K. M., M. D. Corre, B. L. Turner and J. W. Dalling. 2011. Plant–soil associations in a lower montane tropical forest: physiological acclimation and herbivore mediated responses to Nitrogen addition. Functional Ecology 24:1171–1180.

Anderson, D. R. 2008. Model Based Inference in the Life Sciences: A Primer on Evidence. Springer, New York, USA.

Arnold, J., M. D. Corre, and E. Veldkamp. 2008. Cold storage and laboratory incubation of intact soil cores do not reflect in-situ Nitrogen cycling rates of tropical forest soils. Soil Biology and Biochemistry 40:2480–2483.

Austin, A. T. and P.M. Vitousek. 1998. Nutrient dynamics on a precipitation gradient in Hawai’i. Oecologia 113:519–529.

Baraloto, C., D. Bonal and D. E. Goldberg. 2006. Differential seedling growth response to soil resource availability among nine Neotropical tree species. Journal of Tropical Ecology 22: 487–497.

Bazzaz, F. A. 1998. Tropical forests in a future climate: Changes in biological diversity and impact on the global carbon cycle. Climatic Change 39:317–336.

Bouyoucos, C. Y. 1950. Recalibration of hydrometer method for making mechanical analysis of soil. Agronomy Journal 43:434¬–439.

175

Bowen, G. D. and E. K. S. Nambiar. 1984. Nutrition of Plantation Forests. Academic Press, London, UK.

Bunker, D. E., and W. P. Carson. 2005. Drought stress and tropical forest woody seedlings: effect on community structure and composition. Journal of Ecology 93:794–806.

Burslem D. F. R. P., P. J. Grubb, and I. M. Turner. 1995. Responses to nutrient addition among shade-tolerant tree seedlings of lowland tropical rain forest in Singapore. Journal of Ecology 83: 113–122.

Campo, J., and R. Dirzo. 2003. Leaf quality and herbivory responses to soil nutrient addition in secondary tropical dry forest of Yucatan, Mexico. Journal of Tropical Ecology 19: 525–530.

Ceccon, E., S. Sanchez, and J. Campo. 2004. Tree seedling dynamics in two abandoned tropical dry forests of differing successional status in Yucatan, Mexico: a field experiment with N and P fertilization. Plant Ecology 170:277–285.

Cernusak, L. A., K. Winter and B. L.Turner. 2010. Leaf Nitrogen to phosphorus ratios of tropical trees: experimental assessment of physiological and environmental controls. New Phytologist 185:770–779.

Chandler, G. 1985. Mineralization and nitrification in three Malaysian forest soils. Soil Biology and Biochemistry 17:347–353.

Chapin, F. S. 1980. The mineral-nutrition of wild plants. Annual Review of Ecology and Systematics 11:233–260.

Chapin, F. S. III, A. J. Bloom, C. B. Field, and R. H. Waring. 1987. Plant response to multiple environmental factors. Bioscience 37:49–57.

Chapin, F. S. III, P. A. Matson, H. A. Mooney. 2002. Principles of Terrestrial Ecosystem Ecology. Springer, New York, USA.

Chapin III, F. S., P. M. Vitousek, and K. Van Cleve. 1986. The nature of nutrient limitation in plant communities. The American Naturalist 127: 48–58.

Clark, D. A. 1994. Plant Demography. Pages 90-105 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: Ecology and natural history of a neotropical rain forest. The University of Chicago Press, Chicago, Illinois, USA.

Clark, D. A., S. Brown, D. W. Kicklighter, J. Q. Chambers, J. R. Thomlinson, J. Ni, and E. A. Holland. 2001a. Net primary production in tropical forests: An evaluation and synthesis of existing field data. Ecological Applications 11:371–384.

176

Clark, D. A., S. Brown, D. W. Kicklighter, J. Q. Chambers, J. R. Thomlinson, and J. Ni. 2001b. Measuring net primary production in forests: Concepts and field methods. Ecological Applications 11: 356–370.

Clark, D. A. and D. B. Clark. 1992. Life history diversity of canopy and emergent trees in a neotropical rain forest. Ecological Monographs 62: 315–344.

Clark, D. A. and D. B. Clark. 2011. Assessing tropical forests’ climatic sensitivities with long-term data. Biotropica 43: 31–40.

Clark, D. B., M. W. Palmer, and D. A. Clark. 1999. Edaphic factors and the landscape-scale distributions of tropical rain forest trees. Ecology 80:2662–2675.

Clark, D. A., S. C. Piper, C. D. Keeling, and D. B. Clark. 2003. Tropical rain forest tree growth and atmospheric carbon dynamics linked to interannual temperature variation during 1984–2000. Proceedings of the National Academy of Sciences of the United States of America 100: 5853–5857.

Cleveland, C. C., and A. R. Townsend. 2006. Nutrient additions to a tropical rain forest drive substantial soil carbon dioxide losses to the atmosphere. Proceedings of the National Academy of Sciences of the United States of America 103:10316–10321.

Cleveland, C. C., A. R. Townsend, P. Taylor, S. Alvarez-Clare, M. M. C. Bustamante, G. Chuyong, P. Grierson, K. Harms, B. Z. Houlton, A. Marklein, W. Parton, S. Porder, S. C. Reed, C. A. Sierra, W. L. Silver, E. J. R. Tanner, W. R. Wieder. 2011. Relationships among net primary productivity, nutrients and climate in tropical rain forest: a pan-tropical analysis. Ecology Letters 14:939–947.

Cornwell, W. K., W. K. Cornwell, J. H. C. Cornelissen, K. Amatangelo, E. Dorrepaal, V. T. Eviner, O. Godoy, S. E. Hobbie, B. Hoorens, H. Kurokawa, N. Pérez-Harguindeguy, H. M. Quested, L. S. Santiago, D. A. Wardle, I. J. Wright, R. Aerts, S. D. Allison, P. van Bodegom, V. Brovkin, A. Chatain, T. V. Callaghan, S. Díaz, E. Garnier, D. E. Gurvich, E. Kazakou, J. A. Klein, J. Read, P. B. Reich, N. A. Soudzilovskaia, M. V. Vaieretti, and M. Westoby. 2008. Plant species traits are the predominant control on litter decomposition rates within biomes worldwide. Ecology Letters 11:1065–1071.

Crews, T. E. 1999. The presence of nitrogen fixing legumes in terrestrial communities: Evolutionary vs. ecological considerations. Biogeochemistry 46: 233–246.

Crews,T. E., K. Kitayama, J. H. Fownes, R. H. Riley, D. A.Herbert, D. Mueller-Dombois, P. M. Vitousek. 1995. Changes in soil phosphorus fractions and ecosystem dynamics across a long chronosequence in Hawaii. Ecology 76:1407–1424.

Cross A. F. and W. H. Schlesinger. 1995. A literature review and evaluation of the Hedley fractionation scheme: applications to the biogeochemical cycle of soil phosphorus in natural ecosystems. Geoderma 64:197–214.

177

Cuevas, E. and E. Medina. 1988. Nutrient dynamics within Amazonian forests: fine root-growth, nutrient availability and leaf litter decomposition. Oecologia 76: 222–235.

Cusack, D. F., W. W. Chou, W. H. Yang, M. E. Harmon, W. L. Silver, and The LIDET Team. 2009. Controls on long-term root and leaf litter decomposition in Neotropical forests. Global Change Biology 15:1339–1355.

Davidson, E. A., C. J. R. de Carvalho, I. C. G. Vieira, R. D. Figueiredo, P. Moutinho, F. Davidson, E. A., C. J. R. de Carvalho, I. C. G. Vieira, R. D. Figueiredo, P. Moutinho, F. Y. Ishida, M. T. P. dos Santos, J. B. Guerrero, K. Kalif, and R. T. Saba. 2004. Nitrogen and Phosphorus limitation of biomass growth in a tropical secondary forest. Ecological Applications 14:S150–S163.

Denslow, J. S., and G. S. Hartshorn. 1994. Tree-fall gap environments and forest dynamic processes. Pages 120–127 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: Ecology and natural history of a neotropical rain forest. The University of Chicago Press, Chicago, Illinois, USA.

Denslow, J. S., J. C. Schultz, P. M. Vitousek, and B. R. Strain. 1990. Growth responses of tropical shrubs to tree fall gap environments. Ecology 71: 165–179.

Denslow, J. S., and G. S. Hartshorn. 1994. Tree-fall gap environments and forest dynamic processes. Pages 120–127 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: Ecology and natural history of a neotropical rain forest. The University of Chicago Press, Chicago, Illinois, USA.

Denslow, J. S., J. C. Schultz, P. M. Vitousek, and B. R. Strain. 1990. Growth responses of tropical shrubs to tree fall gap environments. Ecology 71:165–179.

Dixon, R. K., S. Brown, R. A. Houghton, A. M. Solomon, M. C. Trexler, and J. Wisniewski. 1994. Carbon pools and flux of global forest ecosystems. Science 263:185–190.

Dominy, N. J., P. W. Lucas, and S. J. Wright. 2003. Mechanics and chemistry of rain forest leaves: canopy and understorey compared. Journal of Experimental Botany 54:2007–2014.

Douville, H., D. Salas-Melia, and S. Tyteca. 2006. On the tropical origin of uncertainties in the global land precipitation response to global warming. Climate Dynamics 26:367–385.

Drechsel, P., and W. Zech. 1991. Foliar nutrient levels of broad-leaved tropical trees- a tabular review. Plant and Soil 131:29–46.

178

Eklund, T. J., W. H. Mcdowell, and C. M. Pringle. 1997. Seasonal variation of tropical precipitation chemistry: La Selva, Costa Rica. Atmospheric Environment 31:3903–3910.

Elser, J. J., M. E. S. Bracken, E. E. Cleland, D. S. Gruner, W. S. Harpole, H. Hillebrand, J. T. Ngai, E. W. Seabloom, J. B. Shurin, and J. E. Smith. 2007. Global analysis of Nitrogen and Phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems. Ecology Letters 10:1135–1142.

Elser, J. J., W. F. Fagan, A. J. Kerkhoff, N. G. Swenson, and B. J. Enquist. 2010. Tansley review: Biological stoichiometry of plant production: metabolism, scaling and ecological response to global change. New Phytologist 186:593–608.

Engelbrecht, B. M. J., L. S. Comita, R. Condit, T. A. Kursar, M. T. Tyree, B. L. Turner, and S. P. Hubbell. 2007. Drought sensitivity shapes species distribution patterns in tropical forests. Nature 447:80–82.

Enquist, C. A. F. 2002. Predicted regional impacts of climate change on the geographical distribution and diversity of tropical forests in Costa Rica. Journal of Biogeography 29:519–534.

Espeleta, J. F., and D. A. Clark. 2007. Multi-scale variation in fine-root biomass in a tropical rain forest: a seven-year study. Ecological Monographs 77:377–404.

Fetcher, N., S. F. Oberbauer, and R. L. Chazdon. 1994. Physiological Ecology of Plants. Pages 198-141 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: Ecology and natural history of a neotropical rain forest. The University of Chicago Press, Chicago, Illinois, USA.

Field, A. 2009. Discovering statistics using SPSS. SAGE Publications Ltd. California, USA.

Field C., and H. A. Mooney. 1986. The photosynthesis-nitrogen relationship in wild plants. Pages 22–55 in T. J. Givnish, editor. On the economy of plant form and function. Cambridge University Press, NewYork, USA.

Frazer, G. W., J. A. Trofymow, and K. P. Lertzman. 1997. A method for estimating canopy openness, effective leaf area index, and photosynthetically active photon flux density using hemispherical photography and computerized image analysis techniques. Report BC-X-373. Forest Ecosystem Processes Network, Pacific Forestry Center. Canadian Forestry Service, Canada.

Galloway, J. N., and E. B. Cowling. 2002. Reactive nitrogen and the world: 200 years of change. Abstracts of Papers of the American Chemical Society 31: 64–71.

179

Galloway, J. N., F. J. Dentener, D. G. Capone, E. W. Boyer, R. W. Howarth, S. P. Seitzinger, G. P. Asner, C. C. Cleveland, P. A. Green, E. A. Holland, D. M. Karl, A. F. Michaels, J. H. Porter, A. R. Townsend, C. J. Vorosmarty. 2004. Nitrogen cycles: past, present, and future. Biogeochemistry 70: 153–226.

Garten, C. T. 1993. Variation in foliar N-15 abundance and the availability of soil-nitrogen on walker branch watershed. Ecology 74:2098–2113.

Gleeson, S. K., and D. Tilman 1992. Plant allocation and the multiple limitation hypothesis. American Naturalist 139:1322–1343.

Grace, J., Y. Malhi, N. Higuchi, P. Meir. 2001. Productivity of tropical rainforests. Pages 401-426 in Mooney, H., J. Roy, and B. Saugier, editors. Terrestrial global productivity: past, present and future. Academic Press, San Diego, California, USA.

Grace J, San Jose J, Meir P, Miranda HS, Montes RA (2006) Productivity and carbon fluxes of tropical savannas. Journal of Biogeography, 33, 387–400.

Güsewell, S. 2004. N:P ratios in terrestrial plants: variation and functional significance. New Phytologist 164:243¬–266.

Haggar J. P., and J. J. Ewel 1997. Primary productivity and resource portioning in model tropical ecosystems. Ecology 78:1211–1221.

Handley, L. L., and J. A. Raven. 1992. The use of natural abundance of nitrogen isotopes in plant physiology and ecology. Plant and Cell Environment 15:965–985.

Harpole, W. S., J. T. Ngai, E. E. Cleland, E. W. Seabloom, E. T. Borer, M. E. S. Bracken, J. J. Elser, D. S. Gruner, H. Hillebrand, J. B. Shurin, and J. E. Smith. 2011. Nutrient co-limitation of primary producer communities. Ecology Letters 14:852–862.

Harrington, R. A., J. H. Fownes, and P. M. Vitousek. 2001. Production and resource use efficiencies in N- and P-limited tropical forests: A comparison of responses to long-term fertilization. Ecosystems 4:646–657.

Hartshorn, G. S. 1983. Plants. Pages 118–350 in D. H. Jansen, editor. Costa Rican natural history. The University of Chicago Press, Chicago, USA.

Hartshorn, G. S., and B. E. Hammel. 1994. Vegetation types and floristic patterns. Pages 73–89 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: Ecology and Natural History of a Neotropical Rain Forest. The University of Chicago Press, Chicago, Illinois, USA.

Hattenschwiler, S., B. Aeschlimann, M. M. Couteaux, J. Roy, and D. Bonal. 2008. High variation in foliage and leaf litter chemistry among 45 tree species of a Neotropical rainforest community. New Phytologist 179:165–175.

180

Hedin, L. O. 2004. Global Organization of Terrestrial Plant-Nutrient Interactions. Proceedings of the National Academy of Sciences of the United States of America 30:10849–10850.

Hedin, L. O., P. M. Vitousek, and P. A. Matson. 2003. Nutrient losses over four million years of tropical forest development. Ecology 84:2231–2255.

Hedley, M. J., J. W. Stewart, and B. S. Chauhan.1982. Changes in inorganic and organic soil phosphorus fractions induced by cultivation practices and by laboratory incubations. Soil Science Society of America Journal 46:970–976.

Henderson, A., G. Galeano, and R. Bernal. 1995. Field Guide to the Palms of the Americas. Princeton University Press, New Jersey.

Herbert, D. A., and J. H. Fownes. 1995. Phosphorus limitation of forest leaf-area and net primary production on a highly weathered soil. Biogeochemistry 29:223–235.

Hiremath, A. J., and J. J. Ewel. 2001. Ecosystem nutrient use efficiency, productivity, and nutrient accrual in model tropical communities. Ecosystems 4:669–682.

Hobbie S. E. 1992. Effects of plant species on nutrient cycling. Trends in Ecology and Evolution 7:336–339.

Hobbie, S. E., and P. M. Vitousek. 2000. Nutrient limitation of decomposition in Hawaiian forests. Ecology 81:1867–77.

Hogberg, P. 1997. Tansley review no 95 - N-15 natural abundance in soil-plant systems. New Phytologist 137:179–203.

Holdridge, L. R., W. C. Grenke, W. H. Hatheway, T. Liang, and J. A. Tosi. 1971. Forest environments in tropical life zones: A pilot study. Pergamon Press. Oxford, UK.

Holste, E. K., R. K. Kobe, and C. V. Vriesendorp. 2011. Seedling growth responses to soil resources in the understory of a wet tropical forest. Ecology 92: 1828–1838.

Hooper, D. U., and P. M. Vitousek. 1998. Effects of plant composition and diversity on nutrient cycling. Ecological Monographs 68:121–149.

Houlton, B. Z., Y. Wang, P. M. Vitousek and C. B. Field. 2008. A unifying framework for dinitrogen fixation in the terrestrial biosphere. Nature 454: 327–331.

Huberty, A. F. and R. F. Denno. 2006. Consequences of nitrogen and phosphorus limitation for the performance of two plant hoppers with divergent life-history strategies. Oecologia 149: 444–455.

Idol, T., P. J. Baker, and D. Meason. 2007. Indicators of forest ecosystem productivity and nutrient status across precipitation and temperature gradients in Hawaii. Journal of Tropical Ecology 23:693–704.

181

IPCC (Intergovernmental Panel for Climate Change). 2007. IPCC climate change 2007 working group I to the fourth assessment report of the IPCC 2007: the physical science basis. Cambridge University Press, New York, USA.

Jenny, H. 1941. Factors of soil formation. McGraw-Hill, New York, USA.

John, R., J. W. Dalling, K. E Harms, J. B. Yavitt, R. F. Stallard, M. Mirabello, S. P. Hubbell, R. Valencia, H. Navarrete, M. Vallejo, R. B. Foster. 2007. Soil nutrients influence spatial distributions of tropical tree species. Proceedings of the National Academy of Sciences of the United States of America 104:864–869.

Jones, J. B., and B. W. Case. 1996. Soil testing and plant analysis no. 3. Pages 389–415 in D. L. Sparks, editor. Methods of soil analysis. Part 3: Chemical methods. Soil Science Society of America, Madison, Wisconsin.

Kaspari, M., M. N. García, K. E. Harms, M. Santana, S. J. Wright, and J. B. Yavitt. 2008. Multiple nutrients limit litterfall and decomposition in a tropical forest. Ecology Letters 11:35–43

Keeland, B. D., and P. J. Young. 2004. Construction and installation of dendrometer bands for periodic tree-growth measurements. U.S. Geological Survey, National Wetlands Research Center. http://www.nwrc.usgs.gov/Dendrometer/index.htm

Kerkhoff, A. J., B. J. Enquist, J. J. Elser, and W. F. Fagan. 2005. Pleant allometry, stoichiometry and the temperature-dependence of primary productivity. Global Ecology and Biogeography 14:585–598.

Kerkhoff, A. J., W. F. Fagan, J. J. Elser, and B. J. Enquist 2006. Phylogenetic and functional variation in the scaling of nitrogen and phosphorus in the seed plants. American Naturalist 168: E103–E122.

Killingbeck, K. T. 1996. Nutrients in senesced leaves: keys to the search for potential resorption and resorption proficiency. Ecology 77:1716–27.

Kleber, M., L. Schwendenmann, E. Veldkamp, J. Rossner, and R. Jahn. 2007. Halloysite versus gibbsite: Silicon cycling as a pedogenetic process in two lowland Neotropical rain forest soils of La Selva, Costa Rica. Geoderma 138:1–11.

Kobe, R. K, C. A. Lepczyk, and M. Iyer 2005. Resorption efficiency decreases with increasing green leaf nutrients in a global data set. Ecology 86:2780–2792.

Koerselmann, W., and A. F. M. Meuleman. 1996. The vegetation N:P ratio: a new tool to detect the nature of nutrient limitation. Journal of Applied Ecology 33:1441–1450.

Kuo, S. 1996. Phosphorus. Pages 869–919 in D. L. Sparks, editor. Methods of soil analysis. Part 3: Chemical methods. Soil Science Society of America, Madison, Winsconsin, USA.

182

Lambers, H., F. S. Chapin III, and T. L. Pons. 1998. Plant physiological ecology. Springer-Verlag, New York, USA.

Lajtha, K., C. T. Driscoll, W. M. Jarrell, and E. T. Elliot. 1999. Soil Phosphorus. Characterization and Total Element Analysis. Pages 115–142 in G. P. Robertson, D. C. Coleman, C. S. Bledsoe, and P. Sollins, editors. Standard soil methods for long-term ecological research. Oxford University Press, New York, USA.

Lawrence, D. 2003. The response of tropical tree seedlings to nutrient supply: meta-analysis for understanding a changing tropical landscape. Journal of Tropical Ecology 19:239–250.

LeBauer, D. S., and K. K. Treseder. 2008. Nitrogen limitation of net primary productivity in terrestrial ecosystems is globally distributed. Ecology 89:371–379.

Lieberman, M., and D. Lieberman. 1994. Patterns of density and dispersion of forest trees. Pages 106-119 in L. A. McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: Ecology and natural history of a Neotropical rain forest. The University of Chicago Press, Chicago, Illinois, USA.

Liebig, J. 1842. Animal chemistry or organic chemistry in its application to physiology and Pathology. Johnson Reprint Corporation, New York, USA.

Losos, E. C., and E. G. Leigh, Jr. 2004. Tropical forest diversity and dynamism. The University of Chicago Press, Chicago, Illinois, USA.

Madgwick, H. A. I., and G. L. Brumfield. 1969. The use of hemispherical photographs to assess light climate in the forest. Journal of Ecology 57:537–542.

Mahowald N., T. D. Jickells, A. R. Baker, P. Artaxo, C. R. Benitez-Nelson, G. Bergametti, T. C. Bond, C. Tami, Y. Chen, D. D. Cohen, B. Herut, N. Kubilay, R. Losno, C. Luo, W. Maenhaut, K. A. McGee, G. S. Okin, R. L. Siefert, S. Tsukuda. 2008. Global distribution of atmospheric phosphorus sources, concentrations and deposition rates, and anthropogenic impacts. Global Biogeochemical Cycles 22:GB4026.

Marklein, A. R., and B. Z. Houlton. 2011. Nitrogen inputs accelerate phosphorus cycling rates across a wide variety of terrestrial ecosystems. New Phytologist 193:696–704.

Marschner, H. 1995. Mineral Nutrition of higher plants. Second edition. Elsevier, USA.

Martinelli, L. A., S. Almeida, I. F. Brown, M. Z. Moreira, R. L. Victoria, S. Filoso, C. A. C. Ferreira, and W. W. Thomas. 2000. Variation in nutrient distribution and potential nutrient losses by selective logging in a humid tropical forest of Rondonia, Brazil. Biotropica 32:597–613.

183

Martinelli, L. A., M. C. Piccolo, A. R. Townsend, P. M. Vitousek, E. Cuevas, W. Mcdowell, G. P. Robertson, O. C. Santos, and K. Treseder. 1999. Nitrogen stable isotopic composition of leaves and soil: tropical versus temperate forests. Biogeochemistry 46:45–65.

McGroddy, M. E., T. Daufresne, and L. O. Hedin. 2004.Scaling of C:N:P stoichiometry in forests worldwide: Implications of terrestrial Redfield-type ratios. Ecology 85:2390 –2401.

McKey, D. 1994. Legumes and nitrogen: the evolutionary ecology of a nitrogen-demanding lifestyle. Pages 211–228 in: J. L. Sprent, and D. McKey, editors. Advances in legume systematics: Part 5 - The Nitrogen factor. Royal Botanic Gardens, Kew, England.

Meijer E. L. and P. Buurman. 2003. Chemical trends in perhumid soil catena on the Turrialba volcano (Costa Rica). Geoderma 117:185–201.

Miller, A. J., E. A. G. Schuur, and O. A. Chadwick. 2001. Redox control of phosphorus pools in Hawaiian montane forest soils. Geoderma 102: 219–237.

Mirmanto, E., J. Proctor, J. Green, L. Nagy, and Suriantata. 1999. Effects of nitrogen and phosphorus fertilization in a lowland evergreen rainforest. Philosophical Transactions of the Royal Society of London Series B-Biological Sciences 354:1825–1829.

Murphy, J., and J. P. Riley. 1962. A modified single solution method for the determination of phosphate in natural waters. Annual Chemical Acta 27: 31–36.

Neelin, J. D., M. Münnich, H. Su, J. E. Meyerson, and C. E. Holloway. 2006. Tropical drying trends in global warming models and observations. Proceedings of the National Academy of Sciences USA 103:6110–6115.

Nepstad, D. C., P. Moutinho, M. B. Dias–Filho, E. A. Davidson, G. Cardinot, D. Markewitz, R. Figueiredo, N. Vianna, J. Q. Chambers, D. G. Ray, J. B. Guerreiros, P. A. Lefebvre, L. Da S. L. Sternberg, M. Z. Moreira, L. Barros, F. Y. Ishida, I. Tohlver, E. Belk, K. Kalif, and K. Schwalbe. 2002. The effects of partial throughfall exclusion on canopy processes, aboveground production, and biogeochemistry of an Amazon forest. Journal of Geophysical Research 107(D20):8085.

Newbery, D. M., G. B. Chuyong, J. J. Green, N. C. Songwe, F. Tchuenteu, and L. Zimmermann. 2002. Does low phosphorus supply limit seedling establishment and tree growth in groves of ectomycorrhizal trees in a central African rainforest. New Phytologist 156: 297–311.

Niklas, K. J., T. Owens, P. B. Reich, and E. D. Cobb. 2005.Nitrogen/phosphorus leaf stoichiometry and the scaling of plant growth. Ecology Letters 8:636–642.

184

Okin, G. S., N. Mahowald, O. A. Chadwick, and P. Artaxo. 2004. Impact of desert dust on the biogeochemistry of phosphorus in terrestrial ecosystems. Global Biogeochemical Cycles 18:GB2005.

Olander, L. P., and P. M. Vitousek 2000. Regulation of soil phosphatase and chitinase activity by N and P availability. Biogeochemistry 49: 175–191.

Olander, L. P., and P. M. Vitousek. 2004. Biological and geochemical sinks for phosphorus in soil from a wet tropical forest. Ecosystems 7:404–419.

Ostertag, R. 2001. Effects of nitrogen and phosphorus availability on fine-root dynamics in Hawaiian montane forests. Ecology 82:485–499.

Ostertag, R. 2010. Foliar nitrogen and phosphorus accumulation responses after fertilization: an example from nutrient-limited Hawaiian forests. Plant Soil 334:85–98.

Ostertag, R. and S. E. Hobbie. 1999. Early stages of root and leaf decomposition in Hawaiian forests: effects of nutrient availability. Oecologia 121:564–573.

Palow, D. T. and S. F. Oberbauer. 2009. Soil type affects seedling shade response at low light for two Inga species from Costa Rica. Plant Soil 319:25–35.

Pandey, C. B., R. C. Srivastava, and R. K. Singh. 2009. Soil nitrogen mineralization and microbial biomass relation, and nitrogen conservation in humid-tropics. Soil Science Society of America Journal 73:1142–1149.

Porder, S., D. A. Clark and P. M. Vitousek. 2006. Persistence of rock derived nutrients in the wet tropical forests of La Selva, Costa Rica. Ecology 87:594–602.

Powers, J. S., R. A. Montgomery, E. C. Adair, F. Q. Brearley, S. J. Dewalt, C. T. Castanho, J. Chave, E. Deinert, J. U. Ganzhorn, M. E. Gilbert, J. A. Gonzalez-iturbe, S. Bunyavejchewin, H. R. Grau, K. E. Harms, A. Hiremath, S. Iriarte-vivar, E. Manzane, A. A. De oliveira, L. Poorter, J. B. Ramanamanjato, S. Salk, A. Varela, G. D. Weiblen, and M. T. Lerdau. 2009. Decomposition in tropical forests: a pan-tropical study of the effects of litter type, litter placement and mesofaunal exclusion across a precipitation gradient. Journal of Ecology 97:801–811.

Proyecto Atlas Digital de Costa Rica. 2008. Instituto Tecnológico de Costa Rica (ITCR), editors. Escuela de Ingeniería Forestal. Cartago, Costa Rica. 1 DVD.

Radulovich R. and P. Sollins 1991. Nitrogen and Phosphorus leaching in zero-tension drainage from a humid tropical soil. Biotropica 23:84–87.

Rastetter, E. B., and G. R. Shaver. 1992. A model of multiple-element limitation for acclimating vegetation. Ecology 73:1157–1174.

185

Reed, S. C., C. C. Cleveland, and A. R. Townsend. 2007. Controls over free-living nitrogen fixation in a lowland tropical rain forest. Biotropica 39:585–592.

Reed, S. C, C. C. Cleveland, and A. R. Townsend. 2008. Tree species control rates of free-living nitrogen fixation in a tropical rain forest. Ecology 89:2924–2934.

Reich, P. B., and J. Oleksyn. 2004. Global patterns of plant leaf N and P in relation to temperature and latitude. Proceedings of the National Academy of Sciences (USA) 101:11001–11006.

Richardson, S. J., R. B. Allen, and J. E. Doherty. 2008. Shifts in leaf N:P ratio during resorption reflect soil P in temperate rainforest. Functional Ecology 22:738–745.

Riley, R. H., and P. M. Vitousek. 1995. Nutrient dynamics and nitrogen trace gas flux during ecosystem development in montane rain-forest. Ecology 76:292–304.

Rinklebe, J., and U. Langer. 2006. Microbial diversity in three floodplain soils at the Elbe River (Germany). Soil Biology and Biochemistry 38:2144–2151.

Robertson, G. P., D. Wedin, P. M. Groffman, J. M. Blair, E. A. Holland, K. J. Nadelhoffer, and D. Harris. 1999. Soil carbon and nitrogen availability, nitrogen mineralization, nitrification, and soil respiration potentials. Pages 115–142 in G. P. Robertson, D. C. Coleman, C. S. Bledsoe, and P. Sollins, editors. Standard soil methods for long-term ecological research. Oxford University Press, New York, USA.

Sack, L., and P. J. Grubb. 2002. The combined impacts of deep shade and drought on the growth and biomass allocation of shade-tolerant woody seedlings. Oecologia 131:175–185.

Sanchez, P. A. 1976. Properties and management of soils in the tropics. Wiley-Interscience Publications, New York.

Sancho, F., R. Mata, E. Molina, and R. Salas. 1990. Estudio de suelos finca de la Escuela de Agricultura de la Región Tropical Húmeda. Reporte para la Escuela de Agricultura de la Región Tropical Húmeda (EARTH), Limón, Costa Rica.

Santiago, L. S, K. Kitajima, S. J. Wright, and S. S. Mulkey. 2004. Coordinated changes in photosynthesis, water relations and leaf nutritional traits of canopy trees along a precipitation gradient in lowland tropical forest. Oecologia 139:495–¬502.

Santiago, L. S., E. A. G. Schuur, and K. Silvera. 2005. Nutrient cycling and plant-soil feedbacks along a precipitation gradient in lowland Panama. Journal of Tropical Ecology 21: 461–470.

Sato, S., and N. B. Comerford. 2006. Assessing methods for developing phosphorus desorption isotherms from soils using anion exchange membranes. Plant and Soil 279:107–117.

186

Schuur, E. A. G. 2001. The effect of water on decomposition dynamics. Ecosystems 4:259–273.

Schuur, E. A. G. 2003. Productivity and global climate revisited: The sensitivity of tropical forest growth to precipitation. Ecology 84: 1 165–1170.

Schuur, E. A. G., O. A. Chadwick, and P. A. Matson. 2001. Carbon cycling and soil carbon storage in mesic to wet Hawaiian montane forests. Ecology 82:3182–3196.

Schuur, E. A. G., and P. A. Matson. 2001. Net primary productivity and nutrient cycling across a mesic to wet precipitation gradient in Hawaiian montane forest. Oecologia 128:431–442.

Sheil, D. Growth assessment in tropical tree: large daily diameter fluctuations and their concealment by dendrometer bands. 2003. Canadian Journal of Forest Research 33: 2027–2035.

Smil, V. 2000. Phosphorus in the environment: Natural flows and human interferences. Annual Review of Energy in the Environment 25: 53–88.

Sollins, P., M. F. Sancho, Ch. R. Mata, and R. L. Sanford, Jr.1994. Soils and soil process research. Pages 34–53 in L. A.McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, editors. La Selva: ecology and natural history of a neotropical rain forest. University of Chicago Press, Chicago, Illinois, USA.

Sterner, R. W., and J. J. Elser. 2002. Ecological stoichiometry: the biology of elements from molecules to the biosphere. Princeton, NJ, USA: Princeton University Press.

Stone, E. L. 1968. Microelement nutrition of forest trees: A review. Pages 132–175 in Forest fertilization, theory and practice. Tennessee Valley Authority. N.F.D.C., Muscle Shoals, AL.

Tanner, E. V. J., and V. Kapos. 1992. Nitrogen and phosphorus fertilization effects on Venezuelan montane forest trunk growth and litterfall. Ecology 73:78–86.

Tanner, E. V. J., P. M. Vitousek, and E. Cuevas. 1998. Experimental investigation of nutrient limitation of forest growth on wet tropical mountains. Ecology 79: 10-22.

Tiessen, H. 1998. Resilience of phosphorus transformations in tropical forest and derived ecosystems. Pages 92–98 in A. Schulte and D. Ruhiyat, editors. Soils of tropical forest ecosystems: characteristics, ecology and management. Springer, New York.

Tilman, D., J. Fargione, B. Wolff, C. D'Antonio, A. Dobson, R. Howarth, D. Schindler, W. H. Schlesinger, D. Simberloff, and D. Swackhamer. 2001. Forecasting agriculturally driven global environmental change. Science 292:281–284.

187

Tosi, J. A. 1969. Mapa Ecológico, República de Costa Rica: Según la clasificación de zonas de vida del mundo de L.R. Holdridge. Tropical Science Center, San José, Costa Rica.

Townsend, A. R., G. P. Asner, and C. C. Cleveland. 2008. The biogeochemical heterogeneity of tropical forests. Trends in Ecology and Evolution 23:424–431.

Townsend, A. R., G. P. Asner, C. C. Cleveland, M. E. Lefer and M. M.C.Bustamante. 2002. Unexpected changes in soil phosphorus dynamics along pasture chronosequences in the humid tropics. Journal of Geophysical Research 107(D20): 8067.

Townsend, A. R., C. C. Cleveland, G. P. Asner, and M. M. C. Bustamante. 2007. Controls over foliar N:P ratios in tropical rain forests. Ecology 88:107–118.

Townsend, A. R., C. C. Cleveland, B. Z. Houlton, C. B. Alden, and J. W. C White. 2011. Multi-element regulation of the tropical forest carbon cycle. Frontiers in Ecology and the Environment 9:9–17.

Treseder, K. K., and P. M. Vitousek. 2001. Effects of soil nutrient availability on investment in acquisition of N and P in Hawaiian rain forests. Ecology 82:946–954.

Turner, B. L. and B. M. J. Engelbrecht. 2011. Soil organic Phosphorus in lowland tropical rain forests. Biogeochemistry 103:297–315.

Vance, C. P., C. Uhde-Stone, and D. L. Allen. 2003. Phosphorus acquisition and use: critical adaptations by plants for securing a nonrenewable resource. New Phytologist 157:423–447.

Vandecar , K. L., D. Lawrence, T. Wood, S. F. Oberbauer, R. Das, K. Tully, and L. Schwendenmann. 2009. Biotic and abiotic controls on diurnal fluctuations in labile soil phosphorus of a wet tropical forest. Ecology 90: 2547–2555.

Vitousek, P. M. 1984. Litterfall, nutrient cycling, and nutrient limitation in tropical forests. Ecology 65: 285–298.

Vitousek, P. M. 1998. Foliar and litter nutrients, nutrient resorption, and decomposition in Hawaiian Metrosideros polymorpha. Ecosystems 1:401–407.

Vitousek, P. M. 2004. Nutrient cycling and limitation. Hawai'i as a model system. Princeton University Press, New Jersey, USA.

Vitousek, P. M. and J. S. Denslow. 1986. Nitrogen and Phosphorus availability in tree fall gaps of a lowland tropical rainforest. Journal of Ecology 74: 1167–1178.

Vitousek P. M. and J. S. Denslow. 1987. Source differences in extractable phosphorus among soils of the La Selva Biological Station, Costa Rica. Biotropica 19:167–170.

188

Vitousek, P. M., and H. Farrington. 1997. Nutrient limitation and soil development: Experimental test of a biogeochemical theory. Biogeochemistry 37:63–75.

Vitousek, P. M, and R. W. Howarth. 1991. Nitrogen limitation on land and in the sea: How can it occur? Biogeochemistry 13:87–115.

Vitousek, P.M. R. Lawrence, R. Walker, L. D. Whiteaker and P. M. Matson. 1993. Nutrient limitations to plant growth during primary succession in Hawaii Volcanoes National Park. Biogeochemistry: 197–215.

Vitousek, P. M., S. Porder, B. Z. Houlton, and O. A. Chadwick. 2010. Terrestrial phosphorus limitation: mechanisms, implications, and nitrogen-phosphorus interactions. Ecological Applications 20:5–15.

Walker, T. W., and J. K. Syers. 1976. Fate of phosphorus during pedogenesis. Geoderma 15:1–19.

Wieder, W. R., C. C. Cleveland, and A. R. Townsend. 2008. Tropical tree species composition affects the oxidation of dissolved organic matter from litter. Biogeochemistry 88:127–138.

Wieder, W. R., C. C. Cleveland, and A. R. Townsend. 2009. Controls over leaf litter decomposition in wet tropical forests. Ecology 90: 3333–3341.

Wood, T. E., D. Lawrence, D. A. Clark and R. L. Chazdon. 2009. Rain forest nutrient cycling and productivity in response to large-scale litter manipulation Ecology 90: 109-121.

Wood, T. E., D. Lawrence, and J. A. Wells. 2011. Inter-specific variation in foliar nutrients and resorption of nine canopy tree species in a secondary neotropical rain forest. Biotropica 43:544–551.

Wright, R. B., B. G. Lockaby, and M. R. Walbridge. 2001. Phosphorus availability in an artificially flooded southeastern floodplain forest soil. Soil Science Society of America Journal 65:1293–1302.

Wright, S. J., J. B., Yavitt, N. Wurzburger, B. L. Turner , E. V. Tanner , E. J. Sayer, L. S. Santiago, M. Kaspari, L. O. Hedin, K. E. Harms, M. N. Garcia, and M. D. Corre. 2011. Potassium, Phosphorus or nitrogen limit root allocation, tree growth and litter production in a lowland tropical forest. Ecology 92:1616–1625.

Wu, C. C., C. C. Tsui, C. F. Hseih, V. B. Asio, and Z. S. Chen. 2007. Mineral nutrient status of tree species in relation to environmental factors in the subtropical rain forest of Taiwan. Forest Ecology and Management 239:81–91.

189

Yavitt, J. B., K. E. Harms, M. N. Garcia, M. J. Mirabello, and S. J. Wright. 2011. Soil fertility and fine root dynamics in response to 4 years of nutrient (N, P, K) fertilization in a lowland tropical moist forest, Panama. Austral Ecology 36:433–445.

Yavitt, J. B., and S. J. Wright. 2008. Seedling growth responses to water and nutrient augmentation in the understorey of a lowland moist forest, Panama. Journal of Tropical Ecology 24:19–26.

Yuan, Z. Y., Y. H. H. Chen, and P. B. Reich. 2011. Global-scale latitudinal patterns of plant fine-root nitrogen and phosphorus. Nature Communications 2:344.

Zarin, D., R. Janaki, R. Alavalapati, F. E. Putz, and M. Schmink. 2004. Working Forests in the Neotropics: Conservation through sustainable management. Columbia University Press, New York, USA.

190

BIOGRAPHICAL SKETCH

Silvia Alvarez Clare was born in San José, Costa Rica in 1977. She attended Saint

Francis High School and upon graduation, decided to explore the world by backpacking

through Asia for a year. After her return, she studied biology at the Universidad de

Costa Rica (UCR), where she obtained a bachelor’s degree in July 2001. She briefly

taught biology at Lincoln High School, in San José but decided to continue her studies

in Tropical Plant Ecology. Silvia became a Gator in 2002, when she started a master’s

degree with Dr. Kaoru Kitajima. She conducted her master’s research on seedling

biomechanics in Barro Colorado Island, Panama. Silvia obtained her MSc degree in

botany with a minor in statistics in May 2005. Silvia then decided to expand the scope of

her research from plant ecophysiology to ecosystems ecology, and started a PhD in

interdisciplinary ecology with Dr. Michelle Mack, in the Department of Biology at the

University of Florida. During her PhD, Silvia conducted her research at EARTH

(Escuela de Agricultura de la Región del Trópico Húmedo) University Forest Reserve, in

the Caribbean Slope of Costa Rica. During these years, Silvia also got married, moved

to Chicago, and became a mom. Silvia obtained her PhD with a concentration in soil

and water science in May 2012 and will continue her work at EARTH as an NSF

Postdoctoral Fellow, measuring the effect of nutrient additions on greenhouse gas

emissions from tropical forest soils.

biological processes influencing nutrient...

Documents