Root dynamics of tropical forests in relation to nutrient availability

MISSING IMAGE

Material Information

Title:
Root dynamics of tropical forests in relation to nutrient availability
Physical Description:
xi, 154 leaves : ill. ; 29 cm.
Language:
English
Creator:
Ostertag, Rebecca, 1969-
Publication Date:

Subjects

Subjects / Keywords:
Roots (Botany)   ( lcsh )
Plants -- Nutrition   ( lcsh )
Soil fertility   ( lcsh )
Botany thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Botany -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1998.
Bibliography:
Includes bibliographical references (leaves 137-153).
Statement of Responsibility:
by Rebecca Ostertag.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 029246747
oclc - 39703554
System ID:
AA00017682:00001


This item is only available as the following downloads:


Full Text










ROOT DYNAMICS OF TROPICAL FORESTS IN RELATION TO NUTRIENT
AVAILABILITY












By

REBECCA OSTERTAG


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


1998


























In memory of Sylvia Ostertag














ACKNOWLEDGMENTS


Although my name alone is on the cover of this book, this dissertation is truly a

collaborative effort. Work in two tropical areas, far away from the institutional devices

of the University of Florida, would not have been possible without tremendous logistical

and financial support. Many people have contributed to the development and

implementation of the research described in this dissertation.

My most heartfelt thanks go to my advisor, Francis (Jack) Putz, whose imprint

and influence is present on every page of this document. During this dissertation process,

Jack has played multiple, intersecting roles as a teacher, advisor, mentor, and friend and I

am grateful for his kindness, patience, criticism, support, and the occasional colored star

on my papers. I will always remember how he unplugged the phone in his office to give

me his undivided attention when he was interviewing me as a prospective student. I

thank Jack for many things, but most particularly for 1) listening to me and valuing my

opinion, 2) giving me the freedom to challenge his knowledge and authority, and 3) never

ceasing to believe in my abilities.

My committee contributed much to my intellectual development during the

dissertation process. I particularly thank Jack Ewel for the considerable time he has spent

with me critiquing the fundamental questions and assumptions of this research. He also

provided considerable encouragement and logistical support, both in Costa Rica and in








Hawai'i. Kimberlyn Williams provided many thoughtful and thorough comments on my

writing, logic, and data analysis. I thank Nick Comerford for his advice on soils and roots

and for use of his laboratory facilities. I also thank Stephen Mulkey, Doria Gordon, and

Dave Eissenstat for their excellent questions, encouragement, and suggestions.

Besides my committee, I thank many people for intellectual inspiration. I

especially thank Peter Vitousek (Stanford University) for use of his field sites, vehicles,

and laboratory. Peter provided me great academic support, both in Hawai'i and in

California, and I am extremely grateful for the time he spent talking with me and

encouraging the fulfillment of my ideas. His laboratory also ran the nutrient analysis; I

thank Doug Turner for running the samples. I also thank Julie Denslow (Tulane

University) for her support and involvement; the conversations I had with her during the

early years of my research were critical to the development of this dissertation. I have

also had interactions over the years with many people that have greatly influenced my

thinking, writing, and research directions. I particularly wish to thank my colleagues Patti

Anderson, Martin Barker, Seth Bigelow, Mary Carrington, David Clark, Deborah Clark,

Ken Clark, Susan Cordell, Carla D'Antonio, Jeff Gerwing, Pauline Grierson, Manuel

Guariguata, Jeremy Haggar, Sharon Hall, Kim Heuberger, Ankila Hiremath, Sarah

Hobbie, Michael Keller, Debbie Kennard, Lianne Kurina, Michelle Mack, Adrienne

Nicotra, Diego P6rez Salicrup, Holly Pearson, Michelle Pinard, Zuleika Pinz6n, Carla

Restrepo, Claudia Romero, Buck Sanford, Ted Schuur, Phil Sollins, Margaret Torn,

Kathleen Treseder, Ed Veldkamp, Julia Verville, Antje Weitz, and Lisa Zweede. Oliver








Chadwick, Robin Harrington, and Kathleen Treseder provided access to unpublished

data. I thank John Cornell and Jay Harrison for statistical advice.

In Costa Rica, I thank the staff at La Selva Biological Station, and especially

Marian Sanchez, for hospitality and laboratory assistance; Jack Ewel, Jeremy Haggar, and

the Huertos project for use of office space, computers, and equipment; and the TRIALS

nursery for growing seedlings. I thank Isaac Bruck, Jack Putz, and Adrienne Nicotra for

help in the field.

In Hawai'i, I especially thank Heraldo Farrington for help with field work,

organization, and lab management. I am extremely grateful for the opportunity to work in

four forests there. I thank Tim Tunison and the staff at Hawai'i Volcanoes National Park

for access to the Ola'a and Thurston sites; the state of Hawai'i (Division of Forestry and

Wildlife) for access to Laupahoehoe; and Wayne Sousa for access to Koke'e State Park. I

thank Marcia Erickson and her staff at the Koke'e Museum for lodging and for incredible

hospitality. Robin Harrington, Jim Fownes, and Guillermo Goldstein facilitated my

research in several ways, particularly whenever I visited the University of Hawai'i.

Darrell Herbert introduced me to the site on the island of Kaua'i. I thank Jack and Kathy

Ewel for frequent use of their hotel and restaurant accommodations in Honolulu. And I

particularly thank those brave souls who assisted me with lab and field work. Sarah

Hobbie assisted with the decomposition study. Uma Sanghvi performed excellent soil

carrying duties, among her many talents, and Val Loh-Palumbo, Roy Sabate, and Gavin

Okano are much appreciated for tireless devotion to root sorting.








At the University of Florida, I thank the office staff of the Department of Botany

for help in many ways. Paula Rowe has been helpful in innumerable ways and was a

consistent source of information about the workings of the university. I thank Mary

McLeod for advising me on laboratory techniques. I thank my professors and fellow

graduate students for their constant help, feedback, prodding, and friendship.

This dissertation was funded by two grants from the National Geographic Society

(one to F.E. Putz, and one to Ostertag and Putz), a NSF dissertation improvement grant,

a Sigma Xi grant, and a fellowship from the College of Liberal Arts and Sciences at the

University of Florida. I thank the Department of Botany and the Graduate School for

providing initial air fare to Hawai'i. I also thank the Department of Botany for occasional

help with expenses such as traveling to meetings. I am grateful to USDA Forest Service,

Institute of Pacific Islands Forestry for providing several inter-island air tickets and the

assistance of Janice Haraguchi.

Finally, I thank my family and the many friends who accompanied me on this

journey. I thank all those who have entertained me, fed me, listened to my complaints,

and laughed with me or at me over the years. Their influence is also inscribed in these

pages.














TABLE OF CONTENTS




ACKNOWLEDGMENTS .............................. ....... ..... .. ...... iii

ABSTRACT ................ .......................................... x

CHAPTERS

1 INTRODUCTION .............................................. 1

2 BELOWGROUND EFFECTS OF CANOPY GAPS IN A TROPICAL WET
FOREST .................................................. 6

Introduction ..................................................... 6
Methods .................................................... 9
Study Area ................................................ 9
Site Selection and Experimental Layout ......................... 11
Root Gap Presence ......................................... 14
Soil M oisture .. ........................................... 15
Root Competition Bioassay ................................. 16
Ingrowth Cores .............................................. 18
Root Proliferation in Response to Nutrient-Rich Patches ............ 19
Statistical Analysis .......................................... 21
Results ......................................................22
Root Gap Presence and Soil Moisture ........................... 22
Root Competition Bioassay ................................... 27
Ingrowth Cores ................................. .... ....... 27
Root Proliferation in Response to Nutrient-Rich Patches ............ 28
D discussion ................................... .......... ...... .31
Root Gaps and Nutrient Cycling .............................. 31
Root Competition ........................................... 33
Nutrient Heterogeneity: Effects on Gap Dynamics ................. 35









3 FINE ROOT PRODUCTIVITY AND TURNOVER OF THREE HAWAIIAN
MONTANE FORESTS IN RELATION TO NUTRIENT
AVAILABILITY ........................................ 38

Introduction .................... ............................. 38
Methods ....................................................... 46
Study Sites and Experimental Design ...........................46
Root Depth Distribution Profiles ............................... 55
Annual Root Productivity and Turnover Rate ..................... 56
Nutrient Analysis of Fine Root Tissue ........................... 61
Statistical Analysis ......................................... 61
Results ..................... ............................... 62
Fine Root Biomass and Length Across the Natural Fertility Gradient .. 62
Fine Root Biomass and Length in Response to Fertilization .......... 69
Nutrient Concentrations in Fine Roots .......................... 73
Discussion ................................. .................. 73
Root Characteristics Along the Natural Fertility Gradient ........... 73
Effects of Fertilization on Root Dynamics ....................... 81
Limitations of BNPP and Fine Root Turnover Estimates ............ 82
Conclusions .............................................. 84

4 EFFECTS OF TISSUE QUALITY AND SOIL NUTRIENT AVAILABILITY ON
ROOT AND LEAF DECOMPOSITION IN HAWAIIAN FORESTS .. 86

Introduction .................................................... 86
Methods ....................................... ................. 89
Study Sites ................ .............................. 89
Collection and Preparation of Root and Leaf Material .............. 95
Experimental Set-up ................ ...................... 96
Processing and Chemical Analyses of Roots and Leaves ............ 98
Data Analysis ........................................... 99
Results ..................................................... 101
Decomposition of Leaves and Roots Along the Natural Fertility
Gradient .......................................... 101
Effects of Fertilization on Root Decomposition Rates ............. 106
Effects of Tissue Quality on Root Decomposition ................ 106
Root and Leaf Chemistry .................................... 112
Discussion ................... ............................. 112
Controls over Leaf and Root Decomposition Rates ................ 112
Nutrient Limitations to Decomposition: Effects of Fertilization ...... 119
Summary ................................................ 120









5 COMPARISON OF ABOVEGROUND AND BELOWGROUND PROCESSES
IN TROPICAL FORESTS ....................................... 122

Introduction ................................................... 122
Effects of Disturbances .......................................... 123
Effects of Soil Fertility ......................................... 124
Decomposition and Nutrient Cycling .......................... 124
Root and Leaf Biomass, Productivity, and Turnover Rates .......... 125
Effects of Fertilization ............................. ...... 134

LIST OF REFERENCES .......................................... ... 137

BIOGRAPHICAL SKETCH .............. ............................. 154














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

ROOT DYNAMICS OF TROPICAL FORESTS IN RELATION TO NUTRIENT
AVAILABILITY

By

Rebecca Ostertag

May 1998

Chairperson: Francis E. Putz
Major Department: Botany


I examined fine (< 2 mm diameter) root growth, root turnover rates, root

decomposition rates, and root competition to address how these belowground processes

are related to soil nutrient availability in tropical forests. These dynamics are often

inferred from observations of aboveground forest dynamics.

In a lowland wet forest in Costa Rica, I compared root processes in canopy gaps

and in undisturbed understory areas on two soil types (fertile and infertile). Fine root

biomass and length were reduced under canopy openings, especially on the infertile soil.

This reduction was unrelated to aboveground gap size or percent canopy openness in 1- to

18-month-old gaps. Exposure to root competition did not affect growth of seedlings of

the pioneer tree species Hampea appendiculata on either soil type. Creation of nutrient-

rich patches had little effect on root growth, except in canopy gaps on the infertile soil,








where fertilization caused root proliferation. In this forest, the belowground

consequences of gap formation depend on soil type, but cannot be predicted from

aboveground gap characteristics.

Belowground net primary productivity, root turnover rates, and root

decomposition rates were measured in three montane wet forests in Hawaii that differed

in nutrient availability but were dominated by the same tree species (Metrosideros

polymorpha). Belowground net primary productivity, root turnover rates, and root

decomposition rates were lowest in a forest where aboveground growth was limited by

phosphorus (P). These rates were similar in a forest where aboveground growth was

limited by nitrogen (N) and a forest with relatively higher levels of available N and P.

Fertilization with N or P had no effect on these belowground processes at the N-limited

site, but P-fertilization at the P-limited site increased the rates of these processes. At

these three sites, the dynamics of root growth, turnover, and decomposition were not

necessarily similar to aboveground growth, leaf turnover rates, or leaf decomposition

rates. Root dynamics were strongly influenced by P-availability but not by N-availability.

In both the Costa Rican and Hawaiian forests, belowground plant characteristics could

not be predicted from aboveground characteristics.














CHAPTER 1
INTRODUCTION


Our knowledge of forest dynamics is heavily biased towards aboveground

processes. This bias is understandable; aboveground plant parts are simply easier to

observe and measure. Relatively few data exist on the magnitude and importance of

belowground processes such as root growth, mortality, competition, herbivory, and

symbiotic associations. Therefore, many belowground processes are often inferred from

aboveground measurements. Part of the reasoning behind this assumption is that

aboveground and belowground plant growth are linked on a whole-plant level. The

carbon acquired by leaves is essential for root growth, and the nutrients and water

acquired by roots are essential for continued carbon gain. Nevertheless, the differences

between leaves and roots in their form, in the types of resources they capture, and in the

media in which they grow challenge this assumption. In this dissertation, I examine how

nutrient availability affects root productivity, turnover, decomposition, and competition.

Nutrient availability has been demonstrated to affect aboveground productivity in

many ecosystems, and it is assumed to affect belowground productivity in a similar

manner. In tropical forests, the focus of this dissertation, nutrients have been shown to

limit plant growth in both lowland and montane forest types (see Tanner et al. 1998 for a

review). Enhancements in tree diameter growth, aboveground net primary productivity,








2

foliar and litterfall nutrient concentrations, and leaf litterfall rates that occur after nutrient

addition provide direct evidence of nutrient limitation. For example, tree growth was

significantly increased after N-fertilization in Venezuelan forests (Tanner et al. 1992),

and leaf litterfall rates were increased after P-fertilization in a Hawaiian forests (Herbert

and Fownes 1995). Nutrient limitation in the absence of fertilization experiments is also

sometimes inferred from indirect evidence such as low nutrient concentrations in live

leaves, leaf litter, and soils (Tanner et al. 1998).

Despite demonstrations that nutrients limit aboveground growth in tropical

forests, belowground productivity in relation to nutrient availability is rarely measured

(but see Cavelier 1989). Instead, due to the difficulty involved in sampling roots,

inferences about root growth are usually made from one-time measurements of root

biomass. Root biomass is often compared among soil types or compared to aboveground

biomass as a root-to-shoot ratio. Greater amounts of root biomass or larger root-to-shoot

ratios are commonly reported on less fertile sites. Infertile sites have therefore been

hypothesized to have slow rates of root turnover (Chapin 1980). These slow rates are

analogous to patterns of leaf growth and turnover; as nutrient availability increases, rates

of leaf growth and turnover generally increase (Coley et al. 1985, Reich et al. 1992).

Based on this reasoning, aboveground and belowground growth are expected to be

positively related.

In contrast, the greater root-to-shoot ratios observed on infertile sites have led

some authors to suggest that there are trade-offs in the relative allocation of resources

between roots and shoots, and that biomass allocation to roots and to leaves is negatively










related (Tilman 1988, Grime 1994). When belowground resources are limiting, plants

allocate relatively more biomass to roots to capture water and nutrients; when light is

limiting, more carbon is allocated to aboveground biomass. Based on this reasoning, it is

argued that increased aboveground growth will lead to decreased belowground growth,

and thus increases in nutrient availability should decrease belowground net primary

productivity.

Despite the inherent appeal of linking aboveground and belowground growth,

there are several reasons why belowground processes cannot necessarily be predicted

based on how aboveground plant tissues respond to nutrient availability. Although

aboveground growth may increase as nutrient availability increases, belowground growth

may not for several reasons. First, increases in root growth may not translate into greater

nutrient uptake. If the plants at a site already have high root length densities and have

formed depletion zones throughout the soil profile, then increases in root production may

not affect uptake. Second, nutrient uptake may be more related to mycorrhizal

associations than to the amount of root biomass. Although increased leaf surface area

may lead to increased carbon gain, it may be inappropriate to extend this reasoning to

roots. Thus, there may be no predictable relationship between aboveground and

belowground plant growth rates.

In this dissertation, I examine belowground processes in relation to nutrient

availability and demonstrate that inferring belowground processes from aboveground

processes often leads to incorrect conclusions. Root dynamics (e.g., root production, root

mortality, root decomposition), rather than one-time static measurements of root biomass,










will be emphasized. It is these dynamic measurements that define how plants respond to

changes in nutrient availability, while one-time measurements of root biomass are only a

snapshot from which root dynamics have to be inferred. These static are therefore

misleading, a point that will be highlighted in later chapters.

Each chapter in this dissertation addresses how root dynamics are affected by

nutrient availability. In a lowland wet tropical forest in Costa Rica, I examine how soil

fertility is related to root recovery after natural disturbance (Chapter 2). On two soil

types, the effect of canopy gaps on root growth, root competition, and root responses to

fertilization are measured. Canopy gaps are a natural disturbance in forests, and their

effects on aboveground processes have been much better studied than their effects on

belowground processes. In this chapter I ask whether the belowground consequences of

disturbance are dependent on soil fertility, and hypothesize that root growth on the

infertile soil type is more strongly affected by canopy opening.

I also test the hypothesis that leaves and roots behave similarly in response to

increasing nutrient availability, therefore implying a positive relationship between leaf

and root growth. To evaluate this idea, a series of experiments were conducted in the

Hawaiian Islands on sites that are similar in species composition, climate, and geology,

but that differed in soil nitrogen and phosphorus availability. I test whether belowground

net primary productivity and root turnover rates increase with increasing nutrient

availability (Chapter 3). The effects of soil nutrients on tissue quality and decomposition

rates are also compared between roots and leaves (Chapter 4). Finally, to determine

whether or not aboveground and belowground plant responses in relation to nutrient









5

availability are correlated, results from earlier chapters are related to data on aboveground

processes reported for tropical forests (Chapter 5).














CHAPTER 2
BELOWGROUND EFFECTS OF CANOPY GAPS IN A TROPICAL WET FOREST



Introduction



Canopy gap formation, a frequent small-scale disturbance in many forests, has

been hypothesized to be an important event in the structuring of many forest

communities. Canopy gaps increase environmental heterogeneity (Brokaw 1985,

Denslow 1987) by altering abundances and distributions of abiotic and biotic resources.

Resources that have been demonstrated to change after gap formation are light (e.g.,

Chazdon and Fetcher 1984), soil moisture (e.g., Vitousek and Denslow 1986, Uhl et al.

1988), nutrient availability (e.g., Matson and Boone 1984, Vitousek and Denslow 1986,

Mladenoff 1987, Uhl et al. 1988), fruit (e.g., Levey 1988) and seed availability (e.g.,

Alvarez-Buylla and Garcia Barrios 1991), seed germination (e.g., Putz 1983, Murray

1988), and herbivores (e.g., Braker and Chazdon 1992). Such changes in the spatial and

temporal heterogeneity of resources may affect forest dynamics by altering colonization

success and competitive outcomes, thereby linking natural disturbances to the

maintenance of plant species diversity (Ricklefs 1977, Denslow 1980, 1987, Connell

1989).










Most studies of gap dynamics have emphasized aboveground patterns and

processes (Denslow and Hartshorn 1994). Only recently have studies been undertaken

that focus on belowground changes after canopy gap formation (Sanford 1989, 1990,

Silver and Vogt 1993, Wilczynski and Pickett 1993, Parsons et al. 1994a,b, Ehrenfeld et

al. 1995) and this information has yet to be incorporated into theories of forest dynamics

or management practices based on gap dynamics models. Many studies on gap dynamics

elucidate the effects of the disturbance, without examining the disturbance in the context

of the environmental conditions (e.g., soil nutrients, water relations, microclimate,

topography) in which it occurs. An understanding of how the consequences of gap

formation vary in relation to background environmental conditions may be important for

predicting patterns of recovery after natural disturbance and for understanding how such

patterns might determine forest community structure.

I designed a series of experiments to address these issues by determining 1) the

effects of canopy gap formation on root length and biomass, and 2) whether or not the

consequences of canopy gap formation are dependent on soil type. To answer these

questions, I identified canopy gaps on two soil types of contrasting fertility in a lowland

tropical rain forest where canopy gap formation is frequent. Effects of canopy gap

formation were determined by examining patterns of root biomass in gaps of various ages

and sizes, while the consequences of canopy gap formation were investigated by

measuring root competition, root ingrowth rates, and root responses to nutrient

heterogeneity. In the first experiment I examined the belowground effects of gap

formation by asking whether canopy gaps alter belowground resources through the








8

formation of "root gaps." I defined root gaps as regions of lower root length and biomass

compared to undisturbed forest. These root gaps can be formed by small- and large-scale

disturbances that result in plant death, including animal burrowing, pathogen attacks

(Eissenstat and Caldwell 1989), hurricanes (Parrotta and Lodge 1991, Silver and Vogt

1993), and treefalls (Sanford 1989, Silver and Vogt 1993, Wilczynski and Pickett 1993).

Although root gap formation could have substantial effects on regeneration and forest

structure by altering soil resource availability and consequently affecting competitive

interactions and species composition, belowground gaps are not completely analogous to

canopy openings. Canopy gaps are distinct holes in the forest canopy, while root gaps

represent thinnings of root biomass of various intensities. Additionally, the spatial

distribution of canopy gaps and root gaps may vary; an opening overhead does not imply

a root gap directly below or vice versa.

To examine the consequences of root gaps on regeneration and community

structure, I examined root competition by creating artificial root gaps. Competitive

effects may become stronger after canopy opening if the increase in light levels is

accompanied by a decrease in fine root biomass and a change in spatial and temporal

distribution of available nutrients. Additionally, root growth may be limited under shaded

conditions (Bilbrough and Caldwell 1995), but may be accelerated at higher light levels.

Specifically, I investigated how the soil and light resources available affected the

aboveground growth rate of seedlings of a pioneer (light-demanding) tree, and whether

the growth rate of roots in the surrounding areas could explain variation in seedling

performance.











Finally, I increased spatial heterogeneity of soil nutrients to test whether root

responses, measured as length proliferation, are related to light levels and soil type.

Belowground responses to fertilization may be more sensitive to nutrient availability than

aboveground growth responses, which may be delayed until several years after

fertilization (Tanner et al. 1990, 1992). Root proliferation into fertile microsites has been

demonstrated (e.g., St. John 1983, Eissenstat and Caldwell 1989, Jackson and Caldwell

1989) and there is some evidence that plants can regulate the degree of root proliferation

according to the fertility of these patches (e.g., Jackson and Caldwell 1989, Friend et al.

1990). Rapid root proliferation may be characteristic of successful competitors (Jackson

and Caldwell 1989, Jackson et al. 1990), but it is unclear whether the benefits of

employing this response depend on levels of available soil nutrients and the spatial and

temporal patterning of these nutrients. I therefore hypothesized that root proliferation

into fertile microsites would be influenced by both changes in resources after natural

disturbance (comparison of canopy gap versus forest understory) and soil type.


Methods



Study Area


This study was conducted during portions of the wet (May-August 1993) and dry

seasons (Jan-April 1994) at La Selva Biological Station, Costa Rica (10026' N, 83059'

W). The forest is considered a tropical wet forest according to the Holdridge life-zone









10

system (Hartshorn 1983). Average monthly temperature is 25.80C and average rainfall is

3962 mm, with no month averaging less than 100 mm of rain (Sanford et al. 1994).

Gap formation is frequent in this forest, especially during the rainy season, and

these canopy gaps are most commonly formed by uprooted trees (Hartshorn 1980).

Estimates of average turnover time based on gap formation rates are 95 yr (Sanford et al.

1986) and 118 27 yr (Hartshorn 1980). Based on mortality rates of tagged trees, the

average stand half-life of trees >10 cm dbh is 34 yr, but not all of these trees create

canopy openings upon their death (Lieberman et al. 1985). About 75% of the gaps in this

forest are < 200 m2 (Sanford et al. 1986), with mean gap size estimates ranging from 87-

161 m2 (Hartshorn 1980, Sanford et al. 1986).

Soils are derived from volcanic parent material and represent a range of fertilities,

from relatively infertile Ultisols to more fertile Entisols and Inceptisols. The higher

elevation areas have been extensively weathered to form "residual" Ultisols, while many

of the lower elevation lava flows have been blanketed with more recent alluvial deposits.

This alluvial material, which originated from the two main rivers that traverse the reserve,

was deposited in several terraces, with the higher elevation sites containing older alluvial

material than lower ones (Sollins et al. 1994).

I conducted this study on fertile (Holdridge consociation) and less fertile soil

(Jaguar consociation) found in old growth forest at La Selva. Although some plant

species' distributions are related to edaphic conditions (Clark et al. 1995), forest structure

is similar between the two soil types and many of the common species can occur on both

alluvial and residual soil (pers. obs.). Both areas were dominated by Pentaclethra










macroloba (Fabaceae) and Welfia georgii (Arecaceae). The Holdridge consociation is a

middle terrace alluvial Andic Humitropept, consisting of an A horizon (0-22 cm)

characterized by 28% clay, 4.7 pH in H20, bulk density of 0.68 Mg/m3, and 8.91% organic

matter (Sollins et al. 1994). Base saturation is 51.2%, effective cation exchange capacity

(CEC) is 5.06 cmol (+)/kg, and acid-ammonium-fluoride-extractable P is 9.8 mg/kg

(Sollins et al. 1994). In contrast, the Jaguar consociation is a residual, volcanically

derived Typic Tropohumult with an A horizon (0-19 cm) with a similar bulk density (0.69

Mg/m3), and pH in H20 (4.5), but greater amounts of clay (40%), less organic matter

(6.75%), and much lower base saturation (27.9%), effective CEC (4.85 cmol (+)/kg) and

P (0.7 mg/kg) (Sollins et al. 1994). Due to high N mineralization rates, canopy dominance

of a nitrogen-fixing legume (P. macroloba), and high adsorptive potential of both of these

soils for P, it has been hypothesized that P rather than N is a limiting element in this

forest (Vitousek and Denslow 1986, 1987).


Site Selection and Experimental Layout


I located 10 canopy gaps per soil type in old growth forest (Table 2-1). Eight of

these 20 gaps were chosen (four on each soil type) in 1993, while the others were located

and sampled in 1994. I defined gaps and measured gap size based on the criteria of

Brokaw (1982) in which a gap is considered a hole in the forest canopy extending down

to an average height of 2 m. I chose only gaps that I estimated to be 2-18 mo old at the

time of sampling based on observations of resprouting, new seedling establishment, and

decomposition of woody debris. Subsequently, I observed regular gap formation over a










Table 2-1. Characteristics of the twenty canopy gaps used for this study.

ALLUVIAL SOIL


Gap size (m2)
9
18
28
38
38
39
75
90
92
116


Gap Age (mo)
10-12
12-18
10-12
12-18
2-3
3-4
12-18
2-3
9-12
1-2


Canopy Openness (%)
8.10
4.94
3.54
9.88
8.34
6.87
6.12
8.96
4.15
9.00


'Apparent' Cause
Branchfall
Snap
Branchfall, Snap
Dead Standing Tree
Snap
Snap
Partial Uproot, Snap
Dead Standing Tree, Branchfall
Uproot
Uproot, Snap










Table 2-1 (continued).


RESIDUAL SOIL


Gap size (m2)
23
39
45
58
84
86
88
94
94
154


Gap Age (mo)
6-8
3-4
2-3
3-4
6-8
2-3
2-4
3-4
6-12
2-3


Canopy Openness (%)
7.85
7.74
7.40
6.52
9.39
9.35
6.70
9.25
6.89
5.30


'Apparent' Cause
Branchfall, Snap
Branchfall
Snap, Branchfall
Snap, Dead Standing Tree
Uproot
Dead Standing Tree, Branchfall
Snap, Branchfall
Snap
Snap
Partial Uproot, Snap








14

several year period and found my estimates based on forest recovery consistent with gaps

of known age. The selected gaps were formed in several different manners (i.e., by

uprooted trees, snapped trees, or large branchfalls. I paired each gap site with a closed

canopy understory site, located about 40-50 m from an edge of the treefall gap. I defined

understory areas whenever canopy closure was ; 94% on average (range 93-99%) as

measured by a spherical densiometer (Lemon 1956) over the designated plot center.

Hemispherical photographs were subsequently taken at 50 cm above the center of each

gap and understory site and analyzed using Solarcalc 6.03 (Macintosh Computers).

Percent canopy openness was 7.31 1.82 in gaps (mean s.d.) and 3.90 1.62 in the

understory.


Root Gap Presence


I tested for the presence of root gaps by estimating root length densities and root

biomass in canopy gaps and understory sites on both soil types. Three soil cores, 8 cm in

diameter and 20 cm deep, were taken in each gap and understory site. Cores taken in

1993 were part of a root ingrowth experiment (see below), but were located in the gap

center within a 2 m diameter area that was relatively free of fallen logs and large tree

roots. In 1994, cores were taken at 3 random locations (azimuth and distance) within a 5

m radius of the gap center. In both cases, samples were taken in the gap center and

corresponded to the bole zone of a canopy gap (sensu Orians 1982). Understory cores

were taken in the same manner, with an area relatively free of debris being designated as

the "center" and cores taken randomly within that area.








15

Soil samples were frozen until they could be processed. I washed soil samples for

17-22 min with a hydropneumatic root elutriator (Gillison's Variety Fabrication Inc.,

Benzonia, Michigan, USA; Smucker et al. 1982) using 0.76 mm mesh filters and then

hand-picked roots out of the remaining organic debris. I separated live and dead roots

based primarily on texture and secondarily on color. Live roots were friable (stele did not

separate from epidermis) and often a light color (except for ferns and palms). All data

reported are for live roots. I measured root length density of four diameter classes (< 1

mm, 1 < 2 mm, 2 < 5 mm, and > 5 mm) using the line-intercept method (Newman

1966, Tennant 1975). Compared to root biomass, root length is considered to be a more

direct estimate of the nutrient and water uptake functions of fine roots (Nye and Tinker

1977). Roots were dried at 800C for 48 hours to correlate root length and biomass.


Soil Moisture


I measured soil moisture in gaps and understory sites using time-domain

reflectometry (TDR) (TDR 6050X1, Soil Moisture Trace Corporation, Santa Barbara,

California, USA). In late January 1994, after a 21 day dry period (27 mm precipitation as

measured by an automatic rain gauge; La Selva weather records), soil moisture to 20 cm

depth was sampled (Topp et al. 1980, 1982) in five random locations in 18 gap and 18

adjacent understory sites using a buriable TDR waveguide. Soil moisture, as a percent of

volume, was recorded and averaged over the five locations within each gap or understory

site at which the waveguide was buried.










Root Competition Bioassay


I examined root competition in the four canopy gaps per soil type located in May

1993. Within each gap and understory site, I chose five treatment areas within a 2 m

diameter area that were relatively free of woody debris and large tree roots and were fairly

homogeneous in litter depth (Figure 2-1). Three of these treatment areas were used for

the competition experiment and the remaining two were used for ingrowth core

measurements (see below). I took spherical densiometer readings at 1 m above each

treatment and at each site measured the diameter at 1.3 m above the ground (dbh) of all

trees, lianas, and palms in a 5 m radius from the circle's center.

The bioassay species was Hampea appendiculata Donn. Sm. (Malvaceae), a

dioecious early successional tree (Croat 1978). I chose this species because it has fast-

growing seedlings (Huston 1982) that are capable of germinating and surviving for short-

time periods in both canopy gap and understory sites (pers. obs.). It also responds to

nutrient addition under high-light conditions by increasing growth rates (Huston 1982),

suggesting that this species might be limited by nutrients.

In January 1993 I collected and mixed together fruit from 7 H. appendiculata

mother trees. Seedlings were grown in a nursery in alluvial soil that was inoculated with

mycorrhizal symbionts. Seedlings were planted in the field in June 1993. In one

treatment area a seedling was planted in otherwise undisturbed soil (control). In the other

two treatment areas I planted a seedling in the center of an experimentally created root

gap or trench (see Figure 2-1). One trench was lined with a double layer of root












Permanent Root Gac





SOpen Root Ga:


Ingrowth Core
(2-3 mo) /







Ingrowth Core
(7-8 mo)


No Root Gap


Figure 2-1. Experimental set-up for the root competition and root ingrowth experiments,
conducted in four gap and understory sites on each soil type.










restriction cloth (permanent root gap) while the other trench did not have this cloth and

was therefore subject to root ingrowth (open root gap). I created these 20 cm diameter x

25 cm deep circular trenches by using a shovel to cut through roots but some sub-soil was

overturned in the process.

The height and basal diameter of each seedling was monitored at 2, 7, and 12 mo

after planting to compare relative growth rates between the 3 root gap treatments and

between light and soil fertility levels. Relative growth rates were calculated as (In htl-ln

hto)/(time--timeo). Basal diameter did not change much over this period and these data

were therefore not analyzed. After 12 mo, surviving seedlings were excavated to

determine root to shoot ratios. Plant parts were dried until constant weight at 800C for

biomass measurements.


Ingrowth Cores


I estimated root growth by measuring root production into root-free areas of soil

using ingrowth cores in two of the treatment areas (see Figure 2-1). I created ingrowth

cores by removing a 8 cm diameter by 20 cm deep core and placing an ingrowth bag,

made of fiberglass window screening (1.5 mm2 mesh) in the resulting hole. Bags were

filled with hand-sieved (1 mm mesh) root-free soil taken from a nearby location; care was

taken to pack the soil in the bags to a bulk densities similar to original levels. One set of

bags (n=15) was harvested in August 1993 (72-90 days later); the other set of bags (n=14)

was harvested in January 1994 (223-241 days). I processed ingrowth samples similarly to

the root gap cores, but measured only root length and not biomass.











Root Proliferation in Response to Nutrient-Rich Patches


Within 5 m radius of the center of the 40 gap and understory sites, I located 6

debris-free areas, each separated by at least 1 m. Half of these experimental locations

were fertilized with a 50 ml solution of fertilizer (Peters' complete fertilizer, 20% N, 20%

P205, 20% K20 by mass) and half were treated with 50 ml of deionized water (controls).

Treatments were randomly assigned and were applied to a 10 cm diameter circle of soil.

Fertilizer was added at a concentration of 1000 kg/ha (equivalent to 200 mg/kg N, 87

mg/kg P, and 142 mg/kg K). Such high concentrations were necessary to ensure that a

lack of root proliferation would indicate no effect of the fertilizer, rather than the

confounding effect of the nutrient levels being too low to elicit a plant response. The

fertilizer and water solutions were sprayed as a fine mist directly on top of the soil with a

spray bottle at a steady, slow rate to saturate the soil pores but avoid too rapid infiltration.

To minimize post-treatment leaching by rain, each treatment was covered with a 15 cm

diameter circle of plastic window sheeting, elevated 1-2 cm above the manipulated area.

Thirty days after fertilization, a 2 cm x 20 cm deep core was taken at each

experimental location. Roots were washed for 8-10 min and measured as described

above. Few dead roots were found in these cores and therefore only live roots are

reported. Roots were originally separated into diameter classes, but almost all were < 2

mm so I combined all diameter classes together as total root length.

To analyze the data I created a root proliferation index (RPI) to account for site

differences between replicates (gap or understory sites). For each site I calculated the RPI










as the mean root length density of the three fertilized cores divided by the mean root

length density of three unfertilized control cores. An RPI = 1, therefore, corresponds to

equal root length density under fertilized and unfertilized conditions. These RPIs were

then used after log transformation in an ANOVA model (see Statistical Analysis) to test

the hypotheses that: 1) root length density in fertilized patches is greater than root length

density in unfertilized patches (i.e., RPI > 1), 2) the average RPI is greater on less fertile

soils than on more fertile soils, and 3) the average RPI is greater in canopy gaps than in

the understory.

To demonstrate that the fertilized soil had increased nutrient levels, I prepared one

extra fertilized and one extra control treatment area at half of the sites (randomly chosen

for both soil types). Set-up of the fertilized and control patches followed the same

procedures as above. Soil cores (2 cm diameter x 20 cm deep) were taken 24 hours after

treatment to ensure that all of the added nutrients had not been leached or taken up by

plants. Each sample was analyzed for N03-N and NH4-N by extracting 10 g of wet soil

with 100 ml of 1 M KCl and then analyzing the two duplicate samples of the supernatant

with a Technicon auto-analyzer (Scientific Instruments Corporation, Hawthorne, New

York, USA). Plant-available P04-P was estimated using sodium bicarbonate as an

extractant; each 2.5 g of wet soil was extracted with 50 ml of 0.5 M, pH 8.5 sodium

bicarbonate (Watanabe and Olsen 1965, Anderson and Ingram 1993). Absorption was

determined colorimetrically at 880 nm using spectrophotometry (Perkin-Elmer Lambda

3A UV/VIS Spectrophotometer) following procedures in Anderson and Ingram (1993).










Soil not used for nutrient analysis was oven-dried at 1050C for 48 hours to determine

water content.


Statistical Analysis


Data were analyzed using SAS 6.03 (SAS Institute 1988). Whenever comparisons

were made between gap and understory sites on the two soil types, data were analyzed

using a nested and crossed analysis of variance model. Sites (gap and understory

replicates) were nested within a soil type because the two soil types were not juxtaposed

in space and because gaps and their understory controls could not be chosen randomly

within each soil type. This basic model was expanded for the bioassay experiment, where

a split plot design was performed within the nested/factorial framework. Due to the

mortality of some seedlings the design is unbalanced. Other data sets were analyzed with

the appropriate parametric and non-parametric paired tests as described below. Data were

log-transformed where necessary and multiple comparisons were done using orthogonal

contrasts (see Montgomery 1991). To deal with problems of non-normality and

heteroscedascity in the root:shoot ratios and the root biomass (but not in the length) data,

data were analyzed parametrically using a rank transformation procedure (Potvin and

Roff 1993).










Results



Root Gap Presence and Soil Moisture


Based on orthogonal contrasts, canopy gaps on the residual soil had lower fine

root (< 2 mm diameter) length densities (F,,, = 33.01, p < 0.001; Table 2-2) and fine root

biomass (Fg,, = 33.94 p < 0.001; Table 2-3) than understory sites on the residual soil.

Understory areas on the residual soil also had greater fine length densities (F,,,s = 18.50, p

< 0.001; Table 2-2) and fine root biomass (F,,,s = 48.84, p < 0.001; Table 2-3) than

understory areas on the alluvial soil. Gap size, percent canopy openness, and median gap

age were poor predictors of fine and total root length and biomass (linear regressions

using each soil core, rather than the site average, as an independent sample: n=59, r2 <

0.12 in all cases).

Patterns of total root length were similar to fine root length (Table 2-2), although

for total biomass, only the soil type effect was significant (Table 2-3). The difference in

the fine and total root biomass is due to the fact that only 31.8% of total root biomass was

fine roots < 2 mm, while 96.0% of total root length was fine roots. For all four diameter

classes, biomass and length were positively correlated (Table 2-4). Although many root

studies measure biomass instead of root length, the length measurements showed stronger

patterns in this study.

Surface soils (0-20 cm) in canopy gaps were significantly wetter than adjacent

understory sites (paired t-test on log transformed data, t=3.2, p < 0.005, n=18).










Table 2-2. Mean ( standard deviation) root length density (cm root/cm3 soil) for the gap and understory sites on the alluvial (more
fertile) and residual (less fertile) soils.


<1 mm

GAPS 0.359 0.132

UNDERSTORY 0.449 0.059




GAPS 0.371 0.128

UNDERSTORY 0.610 0.151


ALLUVIAL SOIL
a
S-< 2 mm 2-< 5 mm

0.025 0.010 0.009 0.003

0.028 0.020 0.010 0.008


RESIDUAL SOIL

0.038 0.015 0.019 0.009

0.070 0.050 0.022 0.018


5 mm

0.004 0.005

0.002 0.001





0.007 0.008

0.006 0.005


Total
b
Roots
0.397 0.111

0.489 0.045





0.435 0.087

0.708 0.095


aFor fine roots < 2 mm, F,,, = 9.49, p <0.01 for soil type; F = 29.78, p < 0.0002 for site; and F = 7.11, p < 0.02 for soil*site
interaction.

b For all root diameters combined, Fg,, = 11.33, p < 0.004 for soil type; F,,g = 31.50, p < 0.0002 for site; and Flgs = 7.71, p < 0.02 for
interaction.









Table 2-3. Mean ( standard deviation) root biomass (g root/m2 soil) for the gap and understory sites on the alluvial (more fertile) and
residual (less fertile) soils.

ALLUVIAL SOIL

< 1 mm <2 mm 2 < 5 mm 5 mm Total

Rootsb
GAPS 80.410 30.309 35.994 13.933 63.608 39.973 138.555 147.512 318.567 207.779


UNDER- 81.687 11.493 37.803 18.115 53.403 37.508 97.239 112.955 270.132 131.863

STORY


RESIDUAL SOIL

GAPS 86.484 27.279 54.740 21.171 125.830 63.668 394.346 409.118


UNDER- 138.987 46.536 111.167 61.989 133.045 123.658 373.099 444.257

STORY


a For fine roots < 2 mm diameter, Fs = 16.88, p < 0.0008 for soil type; F = 5.72, p < 0.03 for site.

b For all root diameters combined, only the soil type effect is significant (F .8 = 13.28, p < 0.002).


661.40 416.973


756.298 523.359










Table 2-4. Correlation of root length (cm) and root biomass (g); n = number of soil cores containing roots of each size class. Data
were log transformed for normality.


Diameter Class


> 5mm
2 <5 mm
1 < 2 mm
<1 mm


66.5%
78.1%
72.2%
53.8%


<0.0001
<0.0001
<0.0001
<0.0001











0.12


0.10 -


0.08 -


0.06-


0.04 -


0.02 -


0.00 -
GAP UNDER GAP UNDER



ROOT GAP TYPE

II- PERMANENT

K OPEN
M NONE




Figure 2-2. Relative growth rates (means 1 SE) of Hampea appendiculata 2 mo after
planting in gaps and understory sites on both alluvial and residual soil types. The
permanent root gap treatment was a trench lined with root restriction cloth, the open root
gap treatment was a trench with no lining, and the no root gap treatment was a seedling
planted in undisturbed soil.


ALLUVIAL


RESIDUAL








27

Back-transformed means and standard deviations for percent soil moisture in canopy gaps

were 40.54 1.19% and 33.90 1.14% for the forest understory.


Root Competition Bioassay


No effect of root competition was demonstrated in this experiment. After 2 mo

seedlings growing in the canopy gaps had faster relative growth rates (RGR) than those in

the understory (F, = 21.10, p < 0.006), but there was no difference in RGR between the

two soil types or the three trenching treatments (Figure 2- 2). Seedlings in the open root

gaps tended to grow slower than those in the other two treatments, but this effect was not

significant (p < 0.07). Similar results were obtained after the 7 and 12 mo measurements;

only the main effect of light was significant. On a finer scale, RGR was not correlated to

percent canopy openness. Because there was no difference between the root gap

treatments, root:shoot ratios of the surviving seedlings (n = 32) were analyzed without

considering the effects of trenching. Seedlings growing in canopy gaps and understory

sites had similar root:shoot ratios, but those on the alluvial soil tended to have lower

ratios (F,6 = 4.43, p < 0.08).


Ingrowth Cores


After 2-3 mo there was no difference in accumulation of live or dead fine roots ( <

2 mm) between soil or light gap treatments (Table 2-5). Accumulation of live roots in the

ingrowth bags averaged 28 18.9% (range, 9-73%) of the length in the soil before the

ingrowth core was created. For the cores left in the soil for 7-8 mo, the ingrowth cores










had an average of 39 + 29% (range, 12-130%) of the initial root length. At 7-8 mo,

however, orthogonal contrasts indicated that 1) on the residual soil there were fewer dead

roots in canopy gaps than in the understory (FI,4 = 26.46, p < 0.025; Table 2-5); and 2)

understory sites on the residual soil had more dead roots than understory sites on the

alluvial soil (F,,4 = 71.97, p < 0.01; Table 2-5). Based on the accumulation rate of live

roots, I calculated how long it would take for root-free areas to reach root length densities

equivalent to understory areas on each soil type. Using the 2-3 mo ingrowth data, I

calculated that on the alluvial soil it would take an average of 211 days (range 128-603)

to reach understory root length densities and an average of 300 days (range 182-860) on

the residual soil. Using the 7-8 mo ingrowth rate, complete root gaps could be filled in an

average of 473 days (range 310-992) on the alluvial soil and in 673 days (range 442-

1414) on the residual soil.

Basal area of surrounding trees, palms, and lianas within a 5 m radius had little

effect on root ingrowth rate. Basal area of lianas was weakly related to the accumulation

of live roots after 2-3 mo (r2 = 0.35, p < 0.02), but had no effect on root growth rate after

7-8 mo. Similarly, the basal area of neighboring trees and palms within a 5 m radius

could not be used to predict root ingrowth rate at either time period.


Root Proliferation in Response to Nutrient-Rich Patches


Levels of N03-N, NH4-N, and P04-P were elevated over distilled water controls

(Wilcoxon signed rank test, p < 0.0001 for all three nutrients). The root proliferation

index (RPI) was greater on the residual soil (F,,g = 4.92, p < 0.04) and tended towards a










Table 2-5. Accumulation rate of live and dead roots < 2 mm diameter into ingrowth bags left in the soil for 2-3 mo or 7-8 mo. Data
are presented as means and standard errors x 10'4, in cm cm"3 d'.


LIVE ROOTS


2-3 MO


7-8 MO


ALLUVIAL SOIL


GAP


UNDERSTORY


14.37 1.01

17.62 1.03


7.06 2.28

11.09 3.58


RESIDUAL SOIL


GAP


UNDERSTORY


18.42 0.91

26.27 0.89


9.53 2.42

12.57 3.10


back transformed means after log transformation









Table 2-5 (continued).


DEAD ROOTS


2-3 MO"


ALLUVIAL SOIL

GAP

UNDERSTORY



RESIDUAL SOIL

GAP

UNDERSTORY


4.51 0.85

8.72 1.25






11.25 0.80

6.56 1.18


6.81 2.57

6.01 2.78






10.10 2.49

16.40 5.12


back transformed means after log transformation

b for roots after 7-8 mo, F,,6 = 10.01, p < 0.03 for site; and F,,6 = 5.69, p < 0.08 for soil*site interaction.


7-8 MOb










lower value under canopy gaps (F,,8 = 3.73, p < 0.07; nested and factorial ANOVA on

log-transformed data). RPI, however, was only greater than a value of 1 (no proliferation

response) on the residual soil in canopy gaps (one tailed t-test on log transformed data, t =

2.19, p < 0.05). On the alluvial soil, under both canopy gaps and in the understory, as

well as in the understory on the residual soil, the RPI was not significantly different than

1 (Figure 2-3).


Discussion



Root Gaps and Nutrient Cycling


Canopy gaps on the more infertile site type had significantly greater soil moisture

and lower fine root length and biomass, suggesting that belowground changes

accompanied changes in light resources after canopy opening. The decreased root

biomass in canopy gaps on the residual soil also corroborates previous investigations of

root gaps in this (Sanford 1989) and other temperate and tropical forests (Sanford 1990,

Silver and Vogt 1993, Wilczynski and Pickett 1993, Parsons 1994b, Ehrenfeld et al.

1995, Cavelier et al. 1996).

Although soil moisture was only measured on one sampling date, soils in gaps

have been found to be wetter at La Selva (Vitousek and Denslow 1986) and in other

neotropical forests (Becker et al. 1988, Uhl et al. 1988). Besides the obvious effects on

plant water use, soil moisture may also interact with nutrient cycling by altering

decomposition (e.g., Silver and Vogt 1993) or nutrient loss rates. The loss of root

















1.4

1.2-

1.0 -

S0.8 -

0.6 -

0.4 -

0.2 -

0.0
GAP UNDER GAP UNDER

ALLUVIAL RESIDUAL









Figure 2-3. Back-transformed means (means + 1 SD) of the root proliferation index (RPI,
mean root length density in fertilized patches divided by mean root length density in
unfertilized patches) in gaps and understory sites on the two soil types. The only RPI
value significantly greater than 1 was on the residual soil under canopy gaps. Actual non-
transformed means (+ 1 SD) for RPI values are 1.06 + 0.27 for canopy gaps, alluvial soil;
0.90 + 0.33 for understory, alluvial soil; 1.30 + 0.42 for canopy gaps, residual soil; and
1.11 + 0.28 for understory, residual soil.










biomass in canopy gaps may depress water and nutrient uptake by plants, potentially

leading to increased leaching of mobile elements, an effect that may be exacerbated by

nutrient pulses (Lodge et al. 1994). For example, in a Puerto Rican forest, increased root

mortality due to trenching led to increased concentrations of soil N03-N within 2 mo,

along with a subsequent decrease in cations (exchangeable Ca and Mg), presumably by

leaching (Silver and Vogt 1993). Similarly, in a Rocky Mountain coniferous forest, NOx-

N was higher in gaps created by tree felling (Parsons et al. 1994a), coupled with a

decrease in active ectomycorrhizal root tip densities (Parsons et al. 1994b).

The reason for the more pronounced root gaps on the residual soil is not obvious

because, even though the chosen gaps on the residual soil tended to be larger and younger

(Table 2-1), there was no relationship between canopy gap characteristics and root length

or biomass measurements. There was also no difference in net growth rates of live roots

into ingrowth bags between the two soils or between gap and understory sites. Based on

the ingrowth rates, root gaps on both soil types should have been completely filled with a

1-2 years after formation. On the residual soil, however, root mortality is reportedly

faster than on alluvial soils (R.L. Sanford, pers. comm.) and this differential mortality rate

may partly account for greater loss of root biomass and slower recovery. Root mortality

could also have an effect on nutrient cycling because, given similar the N and P

concentrations of root litter between the two soil types (Parker 1994), decay of a larger

quantity of dead roots would lead to greater cycling of these elements on the residual

soils. It is still unclear whether the larger root gaps and potential differences in nutrient








34

cycling patterns between the two soil types may affect plant performance and competitive

interactions of species regenerating in canopy gaps.


Root Competition


Although changes in belowground environments after natural disturbances have

been demonstrated to decrease competition intensity (Wilson and Tilman 1993),

decreased root competition in canopy gaps was not evident on either soil type. Light,

rather than nutrients, seemed to be the primary limiting factor of aboveground growth of

the bioassay species, Hampea appendiculata, despite previous indications that this

pioneer species may be limited by nutrients (Huston 1982). Alternatively, trenching may

have had little effect in canopy gaps due to the already lowered fine root length density

and biomass caused by the canopy gap formation. Trenching was not done in areas of

higher light that lacked root gaps, allowing for the possibility that the lower root biomass

in canopy gaps may have contributed to enhanced seedling growth. Furthermore, the P

that should be available after trenching may not have been made more available in these

high P-fixing soils (Sollins et al. 1994). The trenching treatment may also not have

reduced root length densities significantly to affect inter-root competition or it may not

have severed mycorrhizal connections.

The apparent lack of root competition found in this study agrees with studies on

dipterocarp seedlings (Turner et al. 1993) and two previous studies conducted in this

forest on the residual soil, in which neither fertilized shrub seedlings in treefall gaps

(Denslow et al. 1990) nor trenched understory tree seedlings (Denslow et al. 1991)










experienced growth increases. Phosphorus concentrations in leaves of the fertilized

species, however, were increased (Denslow et al. 1990), and root competition between

trees and underplanted shrubs has been demonstrated on a plantation growing on alluvial

soils of this forest (Gerwing 1995).

The lack of growth of the bioassay seedling after trenching may be explained by

the choice of species (see also Denslow et al. 1987) or by the practice of measuring

aboveground growth when studying root competition rather than measuring root

parameters. For example, on a stand level, roots in canopy gaps on the residual soil

proliferated into nutrient-enriched patches, suggesting that there is some evidence for

nutrient-limitation on the residual soil. Thus, belowground and aboveground responses to

nutrient addition may differ. Conclusions about the nutrient limitation of a site based on

plant bioassays may be premature if both above- and belowground components are not

considered at appropriate temporal scales.


Nutrient Heterogeneity: Effects on Gap Dynamics


Nutrient pulses may be common after disturbances due to changes in the spatial

scale of aboveground and belowground biomass distribution after disturbances. Short-

term temporal pulses in N03-N have been noted after treefalls (Uhl et al. 1988, Denslow

and Hartshorn 1994), as well as after trenching or hurricanes (Silver and Vogt 1993).

Root proliferation into fertile microsites appears to be dependent upon the nutrient

concentration of the patches (Jackson and Caldwell 1989), the ion involved (Drew 1975),

duration and timing of the nutrient pulse (Crick and Grime 1987, Campbell and Grime










1989, Pregitzer et al. 1993), prior nutrient status of the responding plants (Friend et al.

1990), and, as some results from this study demonstrate, soil conditions of the non-patch

areas. The differential root proliferation response on the two contrasting soil types, with

similar community composition and structure, suggests that the consequences of gap

formation may depend on background levels of soil fertility. On the more fertile site,

heterogeneity in soil resources resulting from or associated with canopy gap formation

apparently does not elicit a root growth response. On the more infertile site, this

heterogeneity may be important in determining the success of species that regenerate in

the gap; similar root proliferation responses into fertilized microsites have been reported

in other nutrient-limited ecosystems (Cuevas and Medina 1988, Raich et al. 1994, Riley

and Vitousek 1995). What cannot be determined conclusively, however, is whether the

increased root proliferation index in canopy gaps on the residual soil was due also to the

lower root densities after canopy opening.

Nutrient heterogeneity between zones of a canopy gap (i.e., bole zone, crown

zone, and root throw zone) has been hypothesized to be important in allowing differential

colonization success of species and thereby maintaining diversity (Orians 1982, Brandani

et al. 1988). Differences in physical conditions between gaps may be a more appropriate

scale for use in determining patterns of root biomass, root foraging, and subsequent

regeneration, especially because differences in fine root biomass between gaps are greater

than those within gaps (Sanford 1990). Although this study demonstrates the pattern of

root gap formation after canopy opening, the consequences of these root gaps appear to be

dependent on both site characteristics and the response variables measured. Under










conditions of reduced competition or increased nutrient availability, neither root

proliferation nor aboveground growth of the indicator species increased on the more

fertile soil, but root growth was stimulated by fertilization on the infertile soil.

Additionally, the larger root gaps on the residual soil irrespective of gap size or age

suggest that the size of root gaps created after canopy opening may be related more to site

conditions than to gap characteristics (see also Ehrenfeld et al. 1995). It is therefore

unlikely that a natural disturbance will have the same effect on both soil types. Nutrient

addition does not appear to affect plant growth on the more fertile soil in this high

diversity tropical forest. Thus, the hypothesis that high species diversity can be

maintained by niche partitioning of resources along interacting soil and light gradients

(Ricklefs 1977, Denslow 1980) may not be applicable for fertile sites.














CHAPTER 3
FINE ROOT PRODUCTIVITY AND TURNOVER IN THREE HAWAIIAN
MONTANE FORESTS IN RELATION TO NUTRIENT AVAILABILITY



Introduction


Availability of mineral nutrients is a key factor controlling forest productivity.

Interest in understanding the relationship between soil fertility and plant growth has

therefore led to the development of classification schemes linking plant characteristics to

soil fertility (e.g., Monk 1966, Grime 1979, Chapin 1980, Coley et al. 1985, Tilman 1988;

Table 3-1). These generalizations are based on the economic analogy that tissue or plant

level traits are directly related to construction costs and resource return times (e.g.,

Chapin 1980). Tissues that are expensive to build due to scarcity of mineral nutrients or

slow acquisition of carbon should be long-lived because it will take longer for them to

return to the plant the resources spent in constructing them. Consequently, plants growing

on infertile sites often have long-lived leaves with low carbon gaining capacities.

Conversely, plants on fertile sites often have leaf tissues that require a shorter time period

before the resources spent in building them are returned to the plant.

While the patterns reported in Table 3-1 suggest that nutrients affect plant

processes in predictable and testable ways, development of these generalizations were











Table 3-1. Relationships of plant characteristics to idealized categories of soil fertility,
based on Chapin et al. (1993) but also compiled from Chapin (1980), Chabot & Hicks
(1982), Vitousek (1982), (1984), Coley et al. (1985), Reich et al. (1992), and Turner
(1994). The definition of nutrient-use efficiency follows Vitousek (1982): g organic
matter produced/unit nutrient taken up by plant.





Trait Infertile Fertile


Physiological
Stomatal Conductance Low High
Rate of Net Carbon Gain Low High
Leaf Tissue Nitrogen Low High
Nitrogen-Uptake Potential Low High

Leaf
Leaf Type Sclerophyllous Non-sclerophyllous
Specific Leaf Area Low High
Leaf Longevity High Low

Plant
Carbon-based Defenses High Low
Root-to-Shoot Ratio High Low
Nitrogen-based Defenses Low High
Maximum Relative Growth Rate Low High
Morphological Plasticity Low High

Ecosystem
Nitrogen and Phosphorus in Litterfall Low High
Nutrient Resorption from Leaves High Low
Nutrient-use Efficiency High Low








40

largely based on studies of aboveground plant measurements. Few studies have examined

how belowground plant characteristics and processes are related to nutrient availability.

This lack of data is disconcerting because fine root productivity can often account for

more than half of the annual net primary productivity in forest ecosystems, and fine root

turnover may contribute more to soil organic matter and nutrient supply than leaf litterfall

(Vogt et al. 1986a). Furthermore, understanding the effects of nutrient supply on

belowground processes is important in both natural and managed ecosystems, as nutrient

availability often changes during the course of succession and human activities.

The observation that plants growing on nutrient-poor sites consistently have

greater root-to-shoot biomass ratios (Table 3-2) has led to the inference that allocation to

roots is greater under nutrient-poor conditions. Similarly, the root weight ratio (i.e., the

ratio of root weight to total plant weight), reportedly decreased with increased nitrogen in

75% of the 206 cases considered in a literature review examining root responses to

nutrient availability (Reynolds and D'Antonio 1996). When only root (but not shoot)

biomass is measured in both field and species-specific pot studies (see Friend et al. 1994

for a review), there also appears to be an inverse relationship between root biomass and

nutrient availability. At the community level, the pattern of greater root biomass under

nutrient-poor conditions appears to hold among communities where the assemblage of

plant species differ (Jordan & Escalante 1980, Chapin 1980, Clarkson 1985, Nadelhoffer

et al. 1985, Cavelier 1992) as well as for situations in which most of the same species

occur on both fertile and infertile soils (Keyes & Grier 1981, Gower 1987, Chapter 2).










Table 3-2. Belowground processes along natural fertility gradients. F = root measurement at the more fertile site(s), I = root
measurement at the less fertile site(s), NG = not given.

Fine Root New Root Standing Root-to- Root Root
Diameter Production Stock Shoot Ratio Mortality Longevity Methods Reference

NG FI F>I Soil coring; Jordan & Escalante


Root screens

Soil coring;
Observation
windows

N-budgeting


Soil coring

Various


Soil coring

Soil coring

Soil coring


(1980)

Keyes & Grier (1981)


Nadelhoffer et al.
(1985)

Gower (1987)

Medina & Cuevas
(1989)

Kellman (1990)

Cavelier (1992)

Ostertag (Chapter 2)


F


F>I


F

<2 mm



<3 mm


<1 mm

NG


<2 mm

<2 mm

<2 mm


F>I


F









Table 3-3. Effects of fertilization on fine roots. NG = not given; fert. = fertilization.

Fine Root New Root Standing Root-to- Root Root
Diameter Production Stock Shoot Ratio Mortality Longevity

NG Decrease


Decrease No change


Decrease


Increase


No change
in biomass;
Increase in length

Decrease Decrease

Decrease


Decrease


<5 mm


<2 mm



<2 mm

<2 mm


<2 mm


NG


NG


Increase


Increase


Increase


Methods


Coring in plots
5 yr after fert.

Monthly cores
for 2 yr after fert.

Ingrowth cores



Coring

Cores taken
15 mo after fert.

Cores taken
11 mo after fert.

Video camera
for 7-14 d

Video camera
for 82 d


Reference

Farrell & Leaf
(1974)

Alexander & Fairley
(1985)

Ahlstrom et al. (1988)



Cavelier (1989)

Gower & Vitousek
(1989)

Gower et al. (1992)


Gross et al. (1993)


Pregitzer et al. (1993)


Increase


Increase


I








43

A limitation of many of these studies relating roots to nutrient availability is that

they have focused on static (e.g., root-to-shoot ratios) rather than dynamic (e.g.,

productivity and turnover) measurements. These one-time measurements of root biomass

ignore root dynamics; the amount of live root biomass at a given time (the standing stock)

is a balance between the processes of root productivity and longevity. Changes in

productivity can also affect root turnover rates, the number of times live roots are

replaced per year (yr'), which under steady-state conditions can be calculated as the

standing stock of live biomass divided by the net productivity. In this chapter, I focus on

fine root productivity (hereafter, belowground net primary productivity, BNPP) and fine

root turnover rates.

Although distinct patterns seem to characterize aboveground traits (Table 3-1), it

is unclear whether fine roots exhibit responses similar to leaves in relation to nutrient

availability (Hendricks et al. 1993). If fine roots behave as leaves, then increasing

nutrient availability should lead to higher BNPP and root turnover rates (Chapin 1980).

For example, in a study of 13 temperate hardwood and pine forest sites, fine root turnover

rates were positively related to the N-availability of a site (Aber et al. 1985). When

species composition is more constant along a soil fertility gradient, however, fine root

turnover does not always increase with increasing nutrient availability. In one case,

BNPP and root turnover rates were reportedly faster on more infertile soils (Keyes &

Grier 1981). Root turnover rates can also be slower after fertilization (Table 3-3)

apparently due to decreased mortality (Alexander & Fairley 1985, Ahlstrom et al. 1988,

Fahey & Hughes 1994).










Along natural fertility gradients, I suggest that some of the conflicting results

regarding fine root production and turnover rates may be related to the characteristics of

species that tend to inhabit infertile and fertile sites, rather than the direct effects of

nutrient availability on root growth. In the studies where species composition was

relatively constant (e.g., Keyes and Grier 1981), the soil fertility gradients on which they

were conducted were not particularly pronounced. Plant species usually replace each

other along such gradients (Chapin et al. 1986), even under similar climatic conditions.

Thus, the observation that plants growing in fertile sites have greater root turnover rates

(e.g., Nadelhoffer et al. 1985, Aber et al. 1985) can be accounted for simply by the fact

that species typical of nutrient-rich sites tend to have traits associated with fast growth

and fast tissue turnover rates. Within a species, some of these traits may be linked; in

three grass species in Switzerland, rates of leaf and root turnover were positively

correlated (Schlhipfer and Ryser 1996). Among these grasses, the species characteristic of

more nutrient-rich meadows had faster relative growth rates and faster leaf and root

turnover rates than the species characteristic of nutrient-poor meadows. When a species

was transplanted into a habitat of different fertility, there was little change in root and leaf

turnover rates (Schlipfer and Ryser 1996). Furthermore, tissue turnover rates of roots

and leaves were negatively correlated with species-specific traits such as tissue density

(Ryser and Lambers 1995, Ryser 1996). Both of these results suggest that rates of tissue

turnover may be more strongly influenced by a species' life history rather than by

exogenous nutrient supply.










In this study, I examine how nutrient availability affects root dynamics and ask

whether root dynamics are similar to leaf dynamics. I focus on root processes in relation

to both N-and P-availability in tropical forests; previously most similar studies have

examined N-availability in temperate forests. To investigate the relationship between

BNPP and fine root turnover rates along a strong fertility gradient with similar species

composition, I conducted this study along a chronosequence in the Hawaiian Islands. The

sites in this chronosequence form a soil fertility gradient that varies greatly in nitrogen

(N) and phosphorus (P) availability, but where climate, geology, and species composition

are relatively constant (Crews et al. 1995). In most ecosystems, species composition

changes completely over gradients of soil fertility, but it remains similar in these forests

due to the broad ecological ranges of the limited number of tree species in the Hawaiian

Islands (Carlquist 1980). Therefore, this soil fertility gradient is appropriate for

examining how root productivity and turnover rates vary with the availability of N and P.

Nonetheless, despite similar climate and species composition among sites on this

gradient, the differences in soil ages among the sites interact with other environmental

variables besides nutrient availability. I therefore also examined root dynamics within a

site in response to manipulation of N and P by fertilization. These two experiments are

complementary but not equivalent. Examining the effects of fertilization within a site

reduces some of confounding inter-site environmental factors. Root responses after

fertilization, however, may not be directly comparable to root responses along the natural

soil fertility gradient. Plants growing at each site may be adapted to the nutrient

conditions at that site and therefore may respond to short-term changes in nutrient










availability differently due to varied background soil conditions and evolutionary

histories. Taken together, these two approaches eliminate some factors that are

confounding in many studies to address how nutrient availability can influence

belowground plant productivity.



Methods



Study Sites and Experimental Design


Natural fertility gradient. I compared root production and turnover rates along a

natural fertility gradient that consisted of sites differing in soil age and consequently in N-

and P-availability (Table 3-4). These sites are three of six sites described as the "long

substrate age gradient" by Crews et al. (1995). Each site has soils derived from volcanic

ash, receives a mean annual rainfall of about 2500 mm, has a mean annual temperature of

160C, is located between 1122 and 1210 m elevation, and is dominated by Metrosideros

polymorpha (Myrtaceae) (Crews et al. 1995). This tree species, known for its

morphological variation, is widespread throughout the Hawaiian Islands (Carlquist 1980).

It is the dominant canopy tree in mesic to wet forests and is one of the earliest pioneers on

recent lava flows (Aradhya et al. 1990), where it tends to form even-aged stands that are

subject to synchronous dieback (Mueller-Dombois 1985).

The youngest site is adjacent to Thurston Lava Tube in Hawai'i Volcanoes

National Park on the island of Hawai'i. The soil at this site consists of 200-400-yr-old










Table 3-4. Characteristics of the study sites along the natural fertility gradient. Samples sizes (n) are given whenever possible.


Site


N-limited


Fertile


P-limited


Physical Characteristics a
Site Name
Parent Material Age (yr)
Island
Elevation (m)
Soil Type

Soil Characteristics
In situ Resin Bags (pg bag'' d') a
N03-N
NH4-N
P


Thurston
300
Hawai'i
1176
Hydric Dystrandept



0.22 0.12 SE (n=5)
3.09 1.44 SE (n=5)
0.20 0.08 SE (n=5)


Laupahoehoe
20,000
Hawai'i
1170
Typic Hydrandept



4.25 1.27 SE (n=8)
8.12 2.05 SE (n=8)
1.21 0.28 SE (n=8)


Total Nutrients (upper 100 cm mineral soil) b
N (g/kg) 0.10 0.03 SE (n=7)
P (g/kg) 0.05 0.01 SE (n=7)


Gross N Mineralization (mg m"2 d")b

Gross Nitrification (mg m-2 d-') b


195 70 SE (n=8)

33 9 SE (n=8)


Koke'e
4,100,000
Kaua'i
1134
Plinthic Acrudox



10.20 4.91 SE (n=6)
4.12 2.29 SE (n=6)
0.41 0.17 SE (n=6)


0.53 0.04 SE (n=4)
0.04 0.02 SE (n=4)

647 114 SE (n=4)

112 52 SE (n=4)


pH in H20'


5.02


3.57


3.99









Table 3-4 (continued).


Plant Characteristics
Mean Maximum Tree Height (m)

Community Basal Area (m2/ha)a

Foliar nutrients c
N (%)
P (%)


Litterfall a
N(%)
P(%)
Lignin (%)


Net Primary Productivity (g m-2 yrl) d
Leaf
Twig
Stem
Total Aboveground
Root (based on soil respiration
measurements and not soil coring)
Total


16.5 0.4 SE (n=5)

35.8 (81% Metrosideros)


0.87 0.04 SE
0.060 0.006 SE


0.40 0.01 SE
0.026 0.001 SE
25.8 2.2 SE


381
166
506
1053


1444 117 SE


24.7 1.0 SE (n=5)

33.6 (83% Metrosideros)


1.42 0.05 SE
0.101 0.006 SE


0.80 0.04 SE
0.053 0.003 SE
36.0 0.7 SE


430
118
422
970
360


1453 88 SE


13.7 0.4 SE (n=5)

38.0 (88% Metrosideros)


0.86 0.04 SE
0.061 0.002 SE


0.37 0.015 SE
0.022 0.001 SE
36.7 (bulked sample)


305
107
408
820
560


1380 128 SE


a Data from Crews et al. (1995)
b Data from Riley and Vitousek (1995)
C foliar nutrients on Metrosideros polymorpha leaves (glabrous variety); data from Vitousek et al. 1995
d Data from Herbert and Fownes (in preparation)


N-limited


Fertile


P-limited









coarse tephra deposits (Crews et al. 1995) overlaying an older pahoehoe (smooth) lava

flow (Vitousek et al. 1993). Vegetation at this site is dominated by Metrosideros with a

conspicuous tree fern understory/sub-canopy of Cibotium spp.

The intermediate-aged site, located in Laupahoehoe State Forest Reserve (island

of Hawai'i) contains 10,000-30,000-yr-old tephra deposits from Mauna Kea (Crews et al.

1995). Metrosideros trees at this site are much larger with tree ferns and shrubs

dominating the understory.

The oldest site is located within Koke'e State Park on the island of Kaua'i. It is

difficult to determine conclusively whether soils here were derived from tephra or lava,

but the parent material has been estimated to be 4.1 million yr (Crews et al. 1995). While

Metrosideros is also dominant here, the trees are considerably shorter and Cibotium is

almost completely absent from study plots, perhaps due to harvesting. Other ferns,

particularly Elaephoglossum spp., are common in the understory.

Nutrient limitation to aboveground net primary productivity (ANPP) of

Metrosideros at the three sites has been documented through fertilization experiments

(Vitousek and Farrington 1997). These nutrient addition experiments (N, P, and

micronutrients in a complete factorial design) demonstrated that plants at the youngest

site are limited by N-availability due to absence of N in primary minerals. Older soils

have accumulated higher levels of N through atmospheric deposition or N-fixation, but

often have low levels of total P due to occlusion of P into unavailable forms (Crews et al.

1995). Along this chronosequence, ANPP was enhanced by N but not by P at the

youngest site (hereafter N-limited site) and only by P addition at the oldest site (hereafter











P-limited site). I denote the intermediate-aged site as the fertile site because it has trees

of the greatest diameter, height, and foliar N and P concentrations (Crews et al. 1995) and

it has soils with relatively high levels of both available N and P (Table 3-4). Growth at

the fertile site, however, is still limited by nutrients. Metrosideros trees increased ANPP

when N and P were applied together, but not when they were applied singly; this site is

co-limited by these two elements.

As discussed by Crews et al. (1995), this chronosequence is not flawless, and the

fact that present day climatological and topographical conditions are similar among the

sites does not imply that identical conditions occurred throughout the development of

each system. The oldest site may have accumulations of dust that has blown over from

Asia; this dust may be a source of base cations and perhaps P. Temperatures and

elevation were also not constant through time because these islands are undergoing

subsidence (Crews et al. 1995).

Present day factors besides nutrient availability that vary among the sites include

some differences in understory species composition among sites, soil texture, and natural

disturbance regime. For example, fertilization with the limiting nutrient appears to

increase stem densities of some non-native understory species (Ostertag, Verville, and

Vitousek, in preparation). The soil texture differences among the sites (Table 3-5) will

result in differences in the soil moisture regimes, despite the fact that climate is similar

across sites. These soil texture differences are difficult to control for due to the

differences in soil age. Soil at the N-limited site is a gravelly sandy loam, soil at the

fertile site is a silty loam, and soil the P-limited site is a silty clay loam (Table 3-5). The












Table 3-5. Description of the soils at the three sites following USDA-Soil Conservation Service
format (unpublished data). Bulk density and effective cation exchange capacity (ECEC, sum of
bases + aluminum) are from O. Chadwick, unpublished data.

N-limited Site

Described by: O. Chadwick, G. Kelly, T. Crews, and T. Hilinski
Soil Series: Puhimau
Latitude: 19 degrees, 25 minutes, 14 seconds N
Longitude: 155 degrees, 14 minutes, 37 seconds W
Slope: 1 percent
Classification: medial, isothermic, Lithic Hapludands
Moisture Regime: Udic
Notes: Site is borderline isothermic/isomesic. Setting: tephra over Kilauea lava; undissected
pahoehoe lava flow; ohia-fern forest. All textures are "apparent field textures."

Horizons:
Oe--0-4 cm; dark reddish brown (5YR 3/2) muck; weak very fine and fine granular and weak
medium granular structure; very friable, slightly sticky and nonplastic; many very fine and fine
roots and few medium and coarse and few very coarse; few very fine pores; clear smooth
boundary. Bulk density = 0.22 g/cm3.

Oa-4-10 cm; very dark gray (5YR 3/1) muck; weak fine and medium subangular blocky
structure; very friable; slightly sticky and slightly plastic; many very fine and fine roots; few very
fine and fine pores; clear smooth boundary. Bulk density = 0.38 g/cm3.

A--10-18 cm; black (10YR 2/1) sandy loam; weak fine subangular blocky structure; firm,
nonsticky and nonplastic; common very fine and fine roots; clear wavy boundary. Bulk density =
1.00 g/cm3; ECEC = 4.4 cmol (+)/kg.

Bw-18-26 cm; dark brown (7.5YR 3/3) sandy loam; weak fine and medium subangular blocky
structure; friable, slightly sticky and slightly plastic; common very fine and fine roots; clear wavy
boundary. Bulk density = 1.00 g/cm3; ECEC = 3.7 cmol (+)/kg.

2Ab--26-29 cm; dark brown (10YR 3/3) sandy loam; weak fine and medium subangular blocky
structure; slightly sticky and slightly plastic; clear wavy boundary. Bulk density = 1.02 g/cm3;
ECEC = 4.8 cmol (+)/kg.

2Bwb--29-35 cm; dark brown (7.5YR 3/3) silt loam; weak fine and medium subangular blocky
structure; friable, slightly sticky and slightly plastic, clear wavy boundary. Bulk density = 0.98
g/cm3; ECEC = 5.9 cmol (+)/kg.











3Bwb--35-38 cm; brown (10YR 4/3) silt loam; massive; firm, slightly sticky and slightly plastic;
abrupt wavy boundary. No sample.

3Cr--38 cm; pahoehoe lava.


Fertile Site

Described by: O. Chadwick, G. Kelly, T. Crews, and T. Hilinski
Soil Series: Maile
Latitude: 19 degrees, 56 minutes, 0 seconds N
Longitude: 155 degrees, 18 minutes, 20 seconds W
Slope: 11 percent
Classification: hydrous, isomesic Typic Hydrudands
Moisture Regime: Udic
Notes: Setting: tephra over Mauna Kea lava; constructional ridge on pahoehoe lava low; ohia-
fern forest. All textures are "apparent field textures."

Horizons:
0e--0-5 cm; dark reddish brown (5YR 3/2) muck; weak fine granular structure; nonsticky and
slightly plastic; many very fine and fine roots and few medium and coarse and few very coarse;
few very fine pores; clear smooth boundary. Bulk density = 0.25 g/cm3.

Oa-5-12 cm; black (10YR 2/1) muck; weak fine and medium subangular blocky structure;
nonsticky and slightly plastic; many very fine and fine roots; few very fine and fine pores; clear
smooth boundary. Bulk density = 0.32 g/cm3.

A--12-20 cm; dark reddish brown (5YR 3/2) loam; weak fine and medium subangular blocky
structure; slightly sticky and slightly plastic; common very fine and fine roots; clear wavy
boundary. Bulk density = 0.34 g/cm3; ECEC = 9.5 cmol (+)/kg.

Bwl--20-27 cm; dark reddish brown (5YR 3/3) loam; weak fine and medium subangular blocky
structure; slightly sticky and slightly plastic; common very fine and fine roots; clear wavy
boundary. Bulk density = 0.34 g/cm3; ECEC = 5.7 cmol (+)/kg.

Bw2--27-39 cm; dark brown (10YR 3/3) silt loam; weak fine and medium subangular blocky
structure; slightly sticky and slightly plastic; clear wavy boundary. Bulk density = 0.34 g/cm3;
ECEC = 5.3 cmol (+)/kg.

Bw3--39-52 cm; dark brown (7.5YR 3/4) silt loam; weak fine and medium subangular blocky
structure; slightly sticky and slightly plastic; clear wavy boundary. Bulk density = 0.48 g/cm3;
ECEC = 3.5 cmol (+)/kg.











Bw4--52-71 cm; brown (10YR 4/3) silt loam; weak fine and medium subangular blocky
structure; slightly sticky and slightly plastic; clear wavy boundary. Bulk density = 0.45 g/cm3;
ECEC = 2.7 cmol (+)/kg.

Bw5--71-94 cm; dark brown (10YR 3/3) silt loam; weak medium subangular blocky structure;
slightly sticky and slightly plastic; clear wavy boundary. Bulk density = 0.60 g/cm3; ECEC = 2.8
cmol (+)/kg.

2Cr--94-1 10 cm; weak medium subangular blocky structure; highly weathered lava.


P-limited Site

Described by: O. Chadwick, J. Hsiek, T. Crews, and M. Torn
Soil Series: Kumuweia
Latitude: 22 degrees, 8 minutes N
Longitude: 159 degrees, 37 minutes W
Slope: 3 percent
Classification: clayey, ferritic, isomesic Plinthic Acrudox
Moisture Regime: Udic
Notes: Parent material: lava of the Olokele Formation. Slope length: 100 ft.

Horizons:
Oe--0-7 cm; dark reddish brown (5YR 3/2) peat; common fine to coarse roots; 10 percent gravel;
clear smooth boundary. Bulk density = 0.25 g/cm3; ECEC = 23.8 cmol (+)/kg.

A--7-11 cm; very dark gray (5YR 3/1) loam; weak very fine granular structure; slightly sticky
and slightly nonplastic; common fine to coarse roots; 10 percent gravel; clear wavy boundary.
Bulk density = 0.60 g/cm3; ECEC = 12.0 cmol (+)/kg.

Bhs-11-16 cm; very dark grayish brown (10YR 3/2) sandy loam; yellowish brown (10YR 5/6)
and dark olive brown (2.5Y 3/3) mottles; strong medium subangular blocky structure; slightly
sticky and slightly plastic; common fine and medium roots; much of horizon is composed of
indurated peds; 60 percent gravel; clear wavy boundary. Bulk density = 1.12 g/cm3; ECEC = 6.0
cmol (+)/kg.

Bwl--16-23 cm; brown (10YR 4/3) sandy clay loam; yellowish brown (10YR 5/6) mottles; weak
fine and medium subangular blocky structure; friable, sticky and plastic; common fine and
medium roots; gravel is composed of 50 percent weathered lava and 50 percent plinthite nodules;
10 percent gravel; clear wavy boundary. Bulk density = 0.91 g/cm3; ECEC = 5.6 cmol (+)/kg.

Bw2--23-32 cm; dark brown (10YR 3/3) silty clay loam; weak fine angular blocky structure;
friable, sticky and plastic; common fine and medium roots; sand is from plinthite nodules; 0











percent gravel; clear wavy boundary. Bulk density = 0.81 g/cm3; ECEC = 4.4 cmol (+)/kg.

Bw3--32-43 cm; dark brown (10YR 3/3) silty clay loam; moderate fine and medium subangular
blocky structure parting to moderate fine and medium angular blocky; friable, sticky and plastic;
few fine and medium roots; 20-30 percent plinthic/placic nodules and laminae; 5 percent gravel;
clear wavy boundary. Bulk density = 0.91 g/cm3; ECEC = 4.5 cmol (+)/kg.

Bw4--43-54 cm; dark yellowish brown (10YR 3/4) silty clay loam; moderate fine angular blocky
structure; friable, sticky and plastic; few fine and medium roots; 20-30 percent plinthic/placic
nodules and laminae; 30 percent gravel; clear wavy boundary. Bulk density = 1.16 g/cm3; ECEC
= 2.6 cmol (+)/kg.

Bw5--54-72 cm; dark brown (10YR 3/3) silty clay loam; moderate fine angular blocky structure
parting to moderate fine subangular blocky; friable, sticky and plastic; few fine roots; 10 percent
plinthic/placic nodules and laminae; 30 percent gravel; clear smooth boundary. Bulk density =
1.20 g/cm3; ECEC = 3.5 cmol (+)/kg.

Bgl--72-89 cm; dark yellowish brown (10YR 3/4) silty clay; yellowish brown (10YR 5/8)
mottles; moderate fine angular blocky structure parting to moderate fine subangular blocky; firm,
sticky and plastic; 10 percent gravel; gradual wavy boundary. Bulk density = 1.20 g/cm3; ECEC
= 3.2 cmol (+)/kg.

Bg2--89-109cm; dark brown (10YR 3/3) silty clay; yellowish brown (10YR 3/3) silty clay;
yellowish brown (10YR 5/4) mottles; moderate medium subangular blocky structure; firm, sticky
and plastic; not as intense mottles as below; 20 percent gravel; gradual wavy boundary. Bulk
density = 1.20 g/cm3; ECEC = 1.7 cmol (+)/kg.

Crg-109-125 cm; very dark grayish brown (10YR 3/2) silty clay; dark yellowish brown (10YR
4/6) mottles; massive; firm, sticky and plastic; mottles are along the vesicle and fracture planes
of weathered rock; 30 percent gravel. Bulk density = 1.20 g/cm3; ECEC = 3.4 cmol (+)/kg.









55

N-limited site also has soils that are much shallower than the other two sites (Table 3-5).

Therefore, the soils at the N-limited site probably have the lowest water holding capacity

and available water. These expected differences in moisture availability have the

potential to affect both aboveground and belowground growth. A final difference among

sites to consider is that many trees on the P-limited site were damaged or killed by

hurricanes in 1982 and 1992, but fine root biomass had recovered to its pre-hurricane

levels before the initiation of this study (Herbert et al., in press).

Fertilized plots. To examine the effect of fertilization on fine root production and

turnover, I compared root dynamics between fertilized and unfertilized plots at both the

N-limited and P-limited sites. Long-term factorial fertilization experiments have been

ongoing at the N-limited site since October 1985 (Vitousek et al. 1993) and at the P-

limited site since March 1991 (Herbert and Fownes 1995). For this experiment, I only

used the N, P, and control plots. A 15 x 15 m area of each plot was fertilized biannually

at a rate of 100 kg ha' yr' of N (half as urea, half as ammonium nitrate) or 100

kg ha' yr' P (as triple superphosphate). I did not examine the effects of fertilization at

the fertile site because the trees at this site were fertilized individually rather than in plots.


Root Depth Distribution Profiles


I initially characterized the root distribution at each site (unfertilized areas only)

by taking soil cores to the depths at which I noticed a significant amount of fine roots.

Sampling depth was limited at the N-limited site by the presence of lava at approximately

35-40 cm. At the P-limited site, soil sampling was partially limited to the top 15 cm of











soil due to the formation of plinthite (a secondary mineral) below that depth. Using a 5

cm diameter soil corer, I took 10 initial cores per site, separated into 5 cm increments. I

washed soil samples, picked out roots, and measured root length and biomass, as

described below. I used these initial depth distribution profiles to decide on sampling

depths for the monthly soil cores used in measuring annual root productivity and turnover

rate. The final sampling depths were chosen based on field observations, previous

investigations (Herbert and Fownes, in preparation), and data from these initial cores. I

tried to encompass approximately 85% of the fine root (< 2 mm diameter) length and

biomass in each soil. The final sampling depths chosen were 20 cm at the N-limited site,

30 cm at the fertile site, and 15 cm at the P-limited site.


Annual Root Productivity and Turnover Rate


Sequential soil sampling. To test the hypothesis that root production decreases as

nutrient availability increases I monitored fine root ( < 2 mm diameter) biomass and

decomposition at all sites for one year. I made comparisons on two levels: 1) among

sites--comparing the unfertilized control plots across the three sites; and, 2) within sites--

comparing the N- and P-fertilization treatments with control plots at both the N-limited

and the P-limited sites. For fine root production to be estimated, it is necessary to know

the change in live root mass, the change in dead root mass, and the fine root

decomposition rate. The first two were measured using sequential soil coring, the latter

was determined by a root decomposition study using litterbags, as described in Chapter 4.









57

I was unable to separate roots by species in the soil cores; the data are therefore presented

on a stand level.

At the N-limited and P-limited sites, one corer (7.5 x 7.5 m) of the larger 15 x 15

m plot was used for sequential soil coring. In this corer I set up a grid of 49 points, each

1 m apart. I randomly chose without replacement four of these points for monthly

sampling; these four cores are considered sub-replicates and the plots are considered true

replicates (n = 4/treatment). At the fertile site, there were no pre-existing plots so I set up

four 7.5 x 7.5 m plots, with 49 points within each plot, in areas of the forest that were

unfertilized. These four plots were then sampled in the same manner as the plots at the

other two sites. Between October/November 1995 and September/October 1996, sites

were sampled at approximately monthly intervals, with the N- and P-limited sites

sampled 10 times and the fertile site sampled 9 times over the 12 mo interval. At each

sampling date, four cores were taken per plot using a 5 cm diameter corer.

In the laboratory, I sub-sampled each core by using only 1/4 of the core (by

weight), after I determined that the variation within cores was less than the variation

between cores. Cores used for determining fine root biomass and length were frozen until

they could be processed. Processing involved washing soil through a 0.5 mm sieve, and

then placing the resulting material in a tray of deionized water to separate roots from

other organic matter. I separated roots (from all depths combined) into live and dead

components and quantified length and mass for every core. All of the calculations

presented here use only data from fine roots (< 2 mm diameter) because soil coring is not

an appropriate method for sampling coarse roots (Vogt and Persson 1991). Root length









58

was determined using the line-intercept method (Newman 1966, Tennant 1975), and mass

was determined after drying at 70 C for 48 hours. Live and dead roots were

distinguished based on texture and color (Vogt and Persson 1991). Live roots were

flexible and friable and often a light color (except for fern roots which were black). In

dead roots, the stele often separated from the cortex or these roots were extremely rigid.


Calculation of annual fine root productivity and turnover. I calculated annual fine root

productivity using a modification of the compartment-flow model of Santantonio and

Grace (1987). This equilibrium model allows calculation of mortality and productivity

using data on decomposition rate and the standing stock mass of live and dead roots. The

decomposition rate for each sampling interval is used to calculate the amount of live fine

roots needed to restock the dead fine root compartment (mortality rate) and the amount of

biomass needed to restock the live fine root compartment (production rate). This model

makes the assumption that death and decay are temporally segregated; fine roots that die

during one interval do not decompose until the next interval.

I modified this approach slightly because the decomposition equation in the

compartment flow model is based on an exponential rate of decay. Fine roots in the

forests I studied had a linear decay rate over the 12 mo period, perhaps because they were

still in the early stages of decomposition (Chapter 4). For a linear rate of decay, k (in

mo"') is simply the slope of the relationship between the fraction of roots remaining in the

litterbags and time. Therefore, the equation of a linear rate for decomposition is:

(1) R = 1-kt,









59
where R is the fraction of roots remaining after decomposing for time t. Therefore, 1- R

is equal to the fraction of roots that have decomposed over the time interval. Thus,

(2) 1- R = 1- (- kt) = kt

Substituting this linear rate of decay equation for the exponential equation (1- e"")

described by Santantonio and Grace (1987) yields a decomposition amount (D),

(3) Dj = y, (1- ekt) = y, (kt),

where y, is the standing stock mass of dead roots at interval i and Dj is the decomposition

over the interval j.

I therefore used the equation Dj = y, (kt) to calculate the amount of decomposition

over the interval j. This gives the amount of roots decomposed in units of g/m2, since y,

is in g/m2, k is in mo' and t is in mo. Annual rates of decomposition were determined by

summing the calculations for each sampling interval over the 12 mo period to obtain a

decomposition rate (g m'2 yr').

To convert these decomposition rates into belowground net primary productivity

values, Santantonio and Grace (1987) used two other equations. Publicover and Vogt

(1993) have pointed out that for the compartment flow model, the calculation of

productivity can be simplified. Changes in live and dead root mass in the intermediate

months cancel each other out, and only the changes between the first and last month's

sampling are important. In addition, changes in root mass between months 1 and 12 are

only considered when they are significantly different, an approach recommended by many

authors to eliminate adding random sampling error (e.g., Vogt et al. 1986b, Publicover









60

and Vogt 1993). The productivity calculations can be simplified to (Publicover and Vogt

1993):

(4) BNPP,j, = (Xf.a Xinia)) + (Yinal- initial) + I Dannua

where xf, and xinia, are the standing stock mass of live roots at months 12 and 1,

respectively,

Yfn and y, u are the standing stock mass of dead roots at months 12 and 1,

respectively,

Dannua is the annual decomposition rate.

I used this approach to calculate annual belowground net primary productivity

(BNPP) for every plot. I determined whether standing stocks of live and dead root mass

changed with time using a repeated measures analysis of variance model, using the PROC

MIXED Procedure of SAS 6.12 (SAS 1997). If the main effect of time was significant,

comparisons between months were made using a Tukey's HSD test. Only one plot had a

any significant difference between month 1 and month 12 in either live or dead roots;

therefore for the rest of the plots productivity is calculated as the annual sum of each

month's decomposition amount, yikt (Equation 3). In the decomposition study used to

calculate the decomposition constant (k), root samples from plots within the same

treatment were homogenized before placement in the field (Chapter 4). Therefore, the

value of k was calculated on a treatment (site or fertilization) basis rather than on a plot

basis, so that the four replicate plots within a treatment have the same k value. The

variance among plots within a treatment is based on differences in standing stock mass of









61

dead roots (y,). Root turnover rate was calculated as BNPP divided by the standing stock

of live roots.


Nutrient Analysis of Fine Root Tissue


To determine if fine root tissue nutrient concentration varied with time since

fertilization, among fertilization treatments, and among sites, roots were analyzed for

total N and P. Although the sites were fertilized biannually, they were fertilized at

different times of the year. The N-limited and fertile sites were fertilized every January

and July, and the P-limited site was fertilized every October and April. I therefore chose

to sample nutrient concentrations in roots collected from the three sites in November

1995 and August 1996, allowing for comparisons to be made between roots collected 1

mo and 4 mo after fertilization at all three sites.

Tissue nutrients were analyzed only in live roots < 2 mm in diameter. I pooled

dried roots from the 4 sub-replicate cores taken in a given plot so that all nutrient values

represent a plot average (n = 4 plots per fertilization treatment per site). These composite

samples were ground in a Wiley mill (40 mesh) and analyzed for total N and P using a

peroxide persulfate procedure to acid-digest the samples in a block digester. N and P

concentrations were determined with an Alpkem autoanalyzer at Stanford University.


Statistical Analysis


Fine root production, root turnover rate, standing stock mass of live roots, and

standing stock mass of dead roots were each analyzed using one-way analysis of variance











(ANOVA); treatment differences were determined by Fisher's LSD test. Whenever

necessary, as determined by Bartlett's F test, data were log transformed before analysis to

make variances more homogeneous. In a few cases, I used the appropriate non-

parametric test when the homogeneity of variance assumption could not be satisfied.

Comparisons were made 1) among unfertilized control plots at the three sites

along the natural fertility gradient; 2) among the N, P, and control fertilization treatments

within the N-limited site; or, 3) among the N, P, and control fertilization treatments

within the P-limited site. I used a two-way ANOVA to analyze nutrient concentrations in

roots at the two sampling dates. When comparing controls across the three sites, site and

sampling date were the main effects; when comparing fertilization treatments within a

site, fertilization treatment and sampling date were the main effects. All data were

analyzed using SYSTAT (SYSTAT 1992) or SAS 6.12 (SAS 1997).


Results



Fine Root Biomass and Length Across the Natural Fertility Gradient


Fine roots were more concentrated in the upper soil layers (Figure 3-1a-c).

Seasonal changes in the amount of either live or dead root mass (Figure 3-2a) or length

(Figure 3-2b) were small. I averaged root mass over the 12 mo sampling period to obtain

an average standing stock mass for each site. The average standing stock mass of live

fine roots was greater at the P-limited site than at the N-limited site, but the fertile site

was not different from the other two sites (Table 3-6). The standing stock mass of dead










I I' Live
0-5 Dead


5-10


10-15


15-20


20-25


0 100 200 300 400 500
Root Length (cm)


0-5


5-10


10-15


15-20


20-25


0.0


I
0.2
Mass (g)


I
0.4


Figure 3-la. Length and biomass of fine roots (< 2 mm diameter) in the top 25 cm at the
N-limited site. Values are mean + 1 SE (n = 10 cores).











0-5

5-10

10-15

15-20

20-25

25-30

30-35




I -
U





- 0-5

5-10


10-15

15-20

20-25

25-30

30-35


"' //////I//////////////////////A//,----I

"' ///////M///////////^///// --




~y//////>Sy///////----





0 45 90 135
Root Length (cm)


"' ///////////////^--

~ /////////////////////A----
I I
Y//////////////r=




I I
----
///// -

~y//>wr

~ y////////.I


0.00


0.05


0.10


0.15


Mass (g)


Figure 3-lb. Length and biomass of fine roots (< 2 mm diameter) in the top 35 cm at the
fertile site. Values are mean + 1 SE (n = 10 cores).


I I Live
E Dead


_
























I I I I I I
0 100 200 300 400 500
Root Length (cm)


0-5


-1


5-10 -


10-15 -


15-20 -


0.0


0.4
Mass (g)


0.6


Figure 3-1c. Length and biomass of fine roots (< 2 mm diameter) in the top 20 cm at the
P-limited site. Values are mean + 1 SE (n = 10 cores).


I I Live
Dead


~;izii~-i


0-5 -


5-10 -


10-15 -


15-20 -















P-LIMITED


225

150

75

0
225

150

75

0
225

150

75

0


0 5 10 0 5 10 0 5 10


Month
Figure 3-2a. Monthly mass of fine live and dead roots at the three sites. Sampling began
in late October/early November 1995 (month 1) and continued through late
September/early October 1996 (month 12). Values are given as standing stock 1 SE.
Each point represents the average of four replicate plots, and each plot average is based
on four sub-replicates averaged for the plot.


N-LIMITED


FERTILE

















N-LIMITED


0



0o


e0
1



Er


0 5 10 0 5 10 0 5 10



Month

Figure 3-2b. Monthly length of fine live and dead roots at the three sites. Sampling
began in late October/early November 1995 (month 1) and continued through late
September/early October 1996 (month 12). Values are given as mean standing stock 1
SE. Each point represents the average of four replicate plots, and each plot average is
based on four sub-replicates averaged for the plot.


FERTILE


P-LIMITED









Table 3-6. Fine (< 2 mm) root dynamics in unfertilized plots (n=4 plots/site) along the natural fertility gradient. Values are means 1
SD. For each variable, F and p values are presented at the bottom of the table based on one-way analysis of variance. Different letters
within a column represent statistically significant differences using Fisher's LSD test.


Site Standing Stock Standing Stock Production Turnover
of Live Roots of Dead Roots Rate* Rate
(g/m2) (/m (g m2 yr) (y')*


90 33

109 45

70 16


F2,9 = 1.4

p = 0.298


33a 13

39a 16


F2,, = 5.5

p = 0.028


0.5 0.2

0.4 a 0.1

0.1 b 0.02


F2,9 = 15.3

p = 0.001


* ANOVA on log transformed data


N-limited


Fertile


63 12

89 ab 30


P-limited


F,9, = 5.6

p = 0.026











roots averaged over the 12 mo period did not vary among sites (Table 3-6).

Belowground net primary productivity (BNPP) of fine roots was lower at the P-

limited site than at the other two sites (Table 3-6). The differences in standing stocks of

live roots and BNPP caused fine root turnover rate to be greatest at the N-limited and

fertile sites, and lowest at the P-limited site (Table 3-6). Fine root length density (cm

roots/cm3 soil) was greatest at the P-limited site but similar at the N-limited and fertile

sites (F29 = 7.816, p = 0.011; Figure 3-3).


Fine Root Biomass and Length in Response to Fertilization

Fertilization at the N-limited site. The standing stock mass of live fine roots was

greater in the N-fertilized plots than in control plots, but the P-fertilized plots were not

different from the other two treatments (Table 3-7). There were no differences among

fertilization treatments in the standing stock mass of dead roots, BNPP, or root turnover

rates. Root length density tended to be greater in the N- and P-fertilized plots than in

control areas but this effect was not significant (F2,9 = 3.5, p = 0.074, Figure 3-3).

Fertilization at the P-limited site. Standing stock of live or dead roots was not

different among the fertilized plots, but BNPP in the P-fertilized plots was greater than in

control plots (Table 3-7). Because of this increase in BNPP, turnover rates tended to be

greater and turnover times tended to lower P-fertilized plots. Root length density was not

different among the fertilization treatments (F2.9 = 1.5, p = 0.277; Figure 3-3).
















1.0-
SI I +P
E Control
0.8-

.w,,
5 0.6











N-Limited Fertile P-limited
SITE













Figure 3-3. Mean root length density + 1 SD in fertilized and unfertilized plots. Each
bar represents the average of four replicate plots; plot averages were based on cores taken
over the entire 12 mo sampling period (approximately 40 cores).









Table 3-7. Fine (< 2 mm) root dynamics at the N-limited and P-limited sites. Plots (n=4 plots/site) received either N-fertilization, P-
fertilization, or no fertilization (control) at the N-limited and P-limited sites. Values are means 1 SD. Statistics used were either
one-way analysis of variance or Kruskal-Wallis tests. Different letters within a column represent statistically significant differences
using Fisher's LSD test (ANOVA) or non-parametric multiple comparisons (Kruskal-Wallis) following Daniel (1990).


N-limited Site


Treatment Standing Stock Standing Stock Production Turnover
of Live Roots of Dead Roots Rate Rate
(g/m2) (g/m2) (g m2yr-) (yr-)


+Nitrogen

+Phosphorus

Control


p = 0.724 p = 0.484


83 a 9

73 ab 5

63 b 12



F2.9 = 4.9

p = 0.037


103' 18

88 35

90 33



F2.9 = 0.3


46 11

39 17

33" 13



F2,9 = 0.8


0.6 0.1

0.5 a 0.2

0.5 0.2



F2,9 = 0.02

p = 0.977










Table 3-7 (continued).


P-limited site


Treatment Standing Stock Standing Stock Production Turnover
of Live Roots of Dead Roots Rate Rate
(g/m2) (g/m2) (g m-2 y-1) (y')


+Nitrogen

+Phosphorus

Control


117 26

136" 37

114a 20



F29 = 0.7

p = 0.521


74a 12

80 a 24

70a 16



F2.9 = 0.3

p = 0.736


21 ab 4

81 107

16 b4



H =6.5

p = 0.039


0.2 ab 0.03

0.3 a 0.2

0.1 b 0.02



H =7.0

p = 0.030











Nutrient Concentrations in Fine Roots


Natural fertility gradient. Roots from the unfertilized plots did not vary in tissue

N concentration among the three sites or among sampling dates (Table 3-8 and Figure 3-

4). P concentrations also did not differ among sites but there was more P in roots 1 mo

after fertilization than after 4 mo (Table 3-8 and Figure 3-5).

Fertilization at the N-limited site. N concentrations did not differ among

fertilization treatments or by sampling date (Table 3-9 and Figure 3-4). P concentrations

were significantly greater in the P-fertilized plots and 1 mo after fertilization (Table 3-9

and Figure 3-5).

Fertilization at the P-limited site. N concentrations were not affected by

fertilization treatment but were greater 1 mo after fertilization (Table 3-9 and Figure 3-4).

P concentrations were greater in P-fertilized plots but were not different between

sampling dates (Table 3-9 and Figure 3-5).


Discussion



Root Characteristics Along the Natural Fertility Gradient


At these three sites, species composition, climate, and geology were held constant,

so that inferences about the effects of soil fertility on root dynamics could be made.

While I suggest that N and P affect root dynamics differently, other differences among the

three sites should not be ignored. For example, differences in root growth among sites












Table 3-8. F and p values from a two-way analysis of variance of fine root-tissue N and P
concentrations. Concentrations were compared among unfertilized plots at the three sites along
the natural fertility gradient and at two sampling dates.


Source Natural Fertility Gradient

%N %P

df F p F p


Site

Date


2 2.4 0.127

1 1.3 0.271

2 0.8 0.466


Site*Date


2.5 0.110

10.2 0.005

1.1 0.365


Error 18










Table 3-9. F and p values from two-way analysis of variances of fine root-tissue N and P concentrations. The effects of fertilization
were compared within both the N-limited and P-limited sites. Concentrations were compared among fertilized plots and at two
sampling dates.


Source N-limited Site P-limited Site

%N %P %N %P

df F p F p F p F p


2 0.9 0.429

1 1.3 0.269

2 0.7 0.508


88.3 <0.001

14.3 0.001

11.4 0.001


2.5 0.112

9.9 0.005

0.3 0.743


45.9 <0.001

0.5 0.831

1.6 0.237


Error 18


Fert.

Date


Fert*Date













1.5


1.0


. 0.5


0.0
Z

o 1.5


1.0


0.5


Figure 3-4. Nitrogen concentration of fine roots collected 1 mo and 4 mo after
fertilization. Values represent means + 1 SD (n = 4 plots).


1 mo W +N
111 +p
E Control









4 mo



- T T







N-Limited Fertile P-limited
SITE


0.0













0.15


0.10


0.05

o




0

0.10


0.05


0.00


N-Limited


Fertile


P-limited


SITE


Figure 3-5. Phosphorus concentration of fine roots collected 1 mo and 4 mo after
fertilization. Values represent means + 1 SD (n = 4 plots).


1 mo +N
I +p
T Control





4 mo
-1





4 mo





- T

SM


""' """ ""' """ """


-











may be related to ecotypic differences among conspecific plants. Metrosideros is a

species with high genetic and morphological variability (Aradhya et al. 1990), and trees

on the oldest site on Kaua'i probably have had a longer time to adapt or acclimate to local

conditions than those on the younger sites on the island of Hawai'i. Soil texture also

differs among sites and will affect both water availability and the mechanics of root

growth. The N-limited site probably had the lowest amount of available water because of

its sandy loam texture and shallow depth (Table 3-5). Other factors that might affect root

dynamics include root age (Barber 1984), root competition for water and nutrients,

internal plant demands (i.e., the steepness of the concentration gradient from roots to

leaves; Cooper 1984), and mycorrhizal associations. These are factors that unknown in

almost all studies that relate root dynamics to nutrient availability. The strength of this

study is therefore in the comparison of three sites of vastly different soil fertility that

contain similar species composition. While ecotypic differences and soil water regimes

may affect the root growth patterns reported, I suggest that root dynamics will

undoubtedly be affected by soil fertility as well.

In contrast to the pattern observed for leaves, infertile sites do not necessarily have

low belowground net primary productivity (BNPP) and slow fine root turnover rates.

While the P-limited site on the island of Kaua'i had lower BNPP and slower root turnover

rates than the fertile site, this result was not the case for the N-limited site. In most of the

belowground processes measured, the N-limited site and the more fertile site were

similar. The two sites had comparable amounts of live and dead standing stock root











mass, BNPP, and root turnover rates, as well as equivalent rates of fine root

decomposition presumably due to similar tissue qualities (Chapter 4).

Based the observed differences in belowground dynamics between the N- and P-

limited forests, the simple dichotomy between fertile and infertile sites is probably

unjustified. Instead, differences inherent in N- and P-limitation may influence

belowground allocation patterns. As N-availability increased (i.e., comparing the N-

limited site to the fertile site), BNPP and root turnover rates did not change. In contrast,

as P-availability increased (i.e., comparing the P-limited site to the fertile site), BNPP and

fine root turnover rates increased. In these forests with moderately high root length

densities, a new root produced is likely to be competing with other roots for nutrients, but

the strength of this competitive interaction is dependent on the ion involved. Nitrate

(NO,) is very mobile in the soil and enters the root through mass flow, which leaves large

depletion zones, and its uptake is limited by the surface area of root tissue available for

absorption (Nye and Tinker 1977, Cooper 1984). In contrast, both ammonium (NH4) and

phosphate (P04) are fairly immobile in the soil and the depletion zones formed are much

smaller (Nye and Tinker 1977). PO4 uptake is also enhanced by the presence of

mycorrhizae; its uptake is limited by the rate of diffusion to the root or hyphal surface

(Nye and Tinker 1977, Cooper 1984; Eissenstat and Yanai 1997).

Based on the root length density of these sites (Figure 3-3) and the diffusion

coefficients for NO, and PO4 (Nye and Tinker 1977), I calculated that a new root

produced at any of the three sites would probably be growing within the NO, depletion

zone but outside of the PO4 depletion zone. For example, given typical diffusion










coefficients of 106 cm2/s for NO3 and 10-9 cm2/s for PO4 (Nye and Tinker 1977), I

calculated that the size of a depletion zone over a 1 mo period would be 10.2 cm for NO3

and 0.3 cm for P04. Based on the root length densities of these sites, the average inter-

root distance between competing roots is 1.0 cm at the N-limited site, 1.1 cm at the fertile

site, and 0.7 cm at the P-limited site. Thus, roots at all three sites are competing for NO3

but not for PO4 and so new root production will not necessarily lead to increased NO3

uptake but should enhance uptake of PO4.

Root turnover rates may also be related to a plant's mycorrhizal investment. If

newly produced roots need to be reinfected, fast turnover rates may not be advantageous.

At present, the mechanisms of infection for VA mycorrhizal species (which comprise

approximately 90% of the endemic Hawaiian flora including Metrosideros; Koske et al.

1992) are poorly understood (Wilcox 1996). There is some evidence that infection

affects lifespans of ectomycorrhizal roots; colonization of root tips increased root

longevity by increasing the lifespan of cortical cells (Harley 1969; Harley and Smith

1983). While the percent root length colonized with VA mycorrhizae is low at the P-

limited site, mycorrhizal alkaline phosphatase activity is high at this site (K. Treseder,

unpub. data). It has been hypothesized that this enzyme is involved in phosphate transfer

from soil to roots by arbuscular mycorrhizal fungi (Tisserant et al. 1993), which suggests

that a large fraction of the mycorrhizae that are present at the P-limited site are active. At

present, little is known about the surface area of hyphae available for absorption, whether

plant tissue nutrients or soil nutrients regulate infection, and how the carbon costs of

mycorrhizal associations vary with soil fertility (Cooper 1984, Wilcox 1996, Johnson et











al. 1997). These types of questions need to be answered before it will be possible to

evaluate the role of mycorrhizae in relation to soil fertility, but if the mycorrhizal

associations in the N- and P-limited ecosystems differ it may partly explain why P has

more of an effect on root growth and turnover rates than N.


Effects of Fertilization on Root Dynamics


Fertilization affected BNPP and root turnover rates at the N-limited and P-limited

sites in two different ways. At the N-limited site neither variable was affected by N- or P-

fertilization. At the P-limited site, fertilization with N caused no change in BNPP or

turnover rates, but fertilization with P increased BNPP and root turnover rates.

Therefore, increases in N-availability do not appear to affect root dynamics, but rates of

root growth and turnover are increased as P-availability is increased at the P-limited site.

This pattern is similar to the responses of roots along the natural fertility gradient.

Increases in NPP and tissue nutrient concentrations after fertilization are

interpreted as evidence for nutrient limitation (Tanner et al. 1998). BNPP and root-tissue

N concentrations were not increased by N-fertilization at the N-limited site, despite its

enhancement of tree diameter increments, leaf litterfall rates, and foliar N (Vitousek et al.

1993, Vitousek and Farrington 1997). At the P-limited site, BNPP and root-tissue P

concentrations increased after P-fertilization, suggesting that P limits belowground and

aboveground growth. Tissue concentrations of P after P-fertilization were also enhanced

at the N-limited site (Figure 3-5), suggesting that either P limits some aspect of root

performance or that the roots are engaged in luxury consumption. Based on these results,










N does not appear to limit root growth at either site, but P does affect root growth and

root tissue concentrations.

The conclusion that N does not affect root growth contradicts two previous studies

at the N-limited site. N was previously determined to limit root growth because roots

proliferated more into ingrowth cores fertilized with N than those with P, Ca + K, or

distilled water (Raich et al. 1994). In another study at the N-limited site, N-fertilization

was found to decrease standing stocks of live fine (< 2 mm) root mass (Gower and

Vitousek 1989), while I found a small increase. Root responses to nutrient patches are

probably not equivalent to root responses to fertilization across entire root systems

occupying large soil volumes. Conclusions about the effects of N on root growth may be

scale-dependent and are difficult to make without examining BNPP and root turnover

rates. In addition, changes in BNPP after fertilization ideally should be examined in

concert with ANPP. P-fertilization in a Eucalyptus pauciflora forest in Australia altered

aboveground and belowground carbon allocation but only had a small stimulatory effect

on total productivity (Keith et al. 1997). Thus, conclusions about the effects of

fertilization on roots should be tempered by the possibility that some of the changes in

BNPP after fertilization may be due to the reallocation of carbon (Tanner et al. 1998).


Limitations of BNPP and Fine Root Turnover Estimates


Calculations of BNPP are problematic for any forest and there remains much

debate over which methodology is best (e.g., Vogt et al. 1986b). I chose the compartment

flow model for the calculations because it tends be more accurate than other calculation










methods when either there are lack of seasonal differences in root biomass or there is

simultaneous root production and mortality (Publicover and Vogt 1993). Both of these

conditions occurred in these Metrosideros forests and would be expected in other tropical

forests (Figure 3-2). The limitation of this method is that the BNPP estimates in this

study are only as good as the estimates of decomposition rates (Chapter 4) and the ability

to sort live and dead root biomass (Publicover and Vogt 1993). The two other common

methods of calculating BNPP were developed for situations in which root decomposition

has not been measured. The maximum-minimum method calculates production by taking

the difference between the maximum and minimum peaks in biomass over the course of

the sampling period (BNPP = A live + A dead; Vogt and Persson 1991). Because of the

lack of seasonal peaks in live or dead fine root biomass at my study sites, the maximum-

minimum method would give a BNPP value of 0 for these three forests. Similarly,

estimating BNPP based on summing positive increments of biomass, rather than just the

maximum and the minimum, (e.g., Fairley & Alexander 1985) would also yield an BNPP

of zero.

It is still possible that the monthly sampling of roots was insufficient to detect

changes in root biomass in response to fertilization. In a slash pine (Pinus elliottii var.

elliottii) plantation in Florida growing on a nutrient-poor Spodosol, a modeling exercise

demonstrated that a 5% decrease in fine root biomass could account for the entire

aboveground biomass increase after fertilization (Cropper and Gholz 1994). Based on

their sampling scheme, they could only detect a 25% change in root biomass. In this








84

study, because I was limited by having only four plots per treatment, I calculated that only

a 64-104% change in BNPP rates could be detected (a = 0.05, P = 0.95 power).


Conclusions


Understanding how nutrient availability affects root dynamics is complicated by

several factors including 1) methodogical differences among studies; 2) comparisons

among sites of contrasting fertility that also differ in species composition and other

ecosystem properties; and 3) disregard for the differences between N- and P-limitations.

One methodological strength of this study was that species composition and climate were

kept relatively constant over a gradient of N- and P-availability. If the assumption holds

that ecotypic differences are smaller than species differences, it appears that both along a

natural fertility gradient and in response to fertilization, N-availability had little effect on

root dynamics whereas P-availability increased BNPP and root turnover rates. The

effects of increasing P-availability on root dynamics are analogous to the effects of

increased nutrient availability on leaves (but N and P not considered separately; Table 3-

1).

Some of the inconsistent patterns in the literature may also result from the

consideration of static rather than dynamic measures of roots. Typically, belowground

allocation patterns in relation to soil fertility are inferred from one-time measurements of

biomass, a practice that would lead to faulty conclusions for these forests. Based on

standing stocks of root biomass, plants on infertile sites have been predicted to allocate

more carbon to roots than plants on more fertile sites. For example, in this study, the P-










limited site has a greater standing stock of live root biomass than the other two sites; it

might therefore be inferred that plants at this site invest more into roots than at the other

two sites. In contrast, BNPP and root turnover were slower at the P-limited site than at

the other two sites. Thus, due to lower BNPP and slower turnover rates, the P-limited site

actually had lower carbon investment into root biomass (not considering the costs of

respiration, exudates, herbivory, or mycorrhizae) than the N-limited and fertile sites. The

processes of root growth and turnover cannot be predicted based on standing stocks of

live biomass and also appear to be influenced by P- but not necessarily by N-availability.

Consideration of whether a site was infertile due to N- or P-limitation had a strong impact

on root dynamics in these Hawaiian montane forests.














CHAPTER 4
EFFECTS OF TISSUE QUALITY AND SOIL NUTRIENT AVAILABILITY
ON ROOT AND LEAF DECOMPOSITION IN HAWAIIAN FORESTS



Introduction



Root decomposition is an often ignored, yet potentially important regulator of

nutrient availability. While many studies have investigated rates of mass and nutrient

loss from decaying leaves, and the factors that control these rates, comparable

information for roots is lacking for most ecosystems (Bloomfield et al. 1993). Based on

leaf decomposition studies, site fertility is hypothesized to be one factor controlling

decomposition rates. When climate is constant, plants growing on nitrogen-rich sites tend

to produce leaf litter of higher quality that decomposes more quickly than the leaf litter of

plants from nitrogen-poor sites (Gosz 1981). Due to the limited data available on root

decomposition rates, however, it is unknown whether nutrient availability controls rates

of root decomposition in an analogous manner.

In leaves, initial tissue nitrogen (N) concentration often correlates positively with

decomposition rate (Melillo et al. 1982, Taylor et al. 1989), at least during the initial

stages of decomposition (Berg et al. 1982), suggesting that nitrogen limits decomposition

in these stages. The purported influence of soil nutrient availability on decomposition








87

rates of leaves, however, has been challenged by studies in which nutrient supplies were

manipulated by fertilization (reviewed by Fog (1988)). In these types of experiments,

fertilization increases exogenous nutrient supply and also affects the nutrient

concentration of leaf tissues. In some of these investigations, fertilization increased

decomposition rates (Salonius 1972, Gill 1983, Prescott et al. 1992), which would be

expected if decomposition rates were positively related to nutrient availability. In other

cases, fertilization either decreased leaf decomposition rate (Titus and Malcolm 1987,

Kemp et al. 1994, Wright and Tietema 1995, Prescott 1995) or had no effect (Staaf 1980,

MacKay 1987, Theodorou and Bowen 1990, Slapokas and Granhall 1991, van Vuuren

and van der Eerden 1992, Prescott 1995). Root decomposition in relation to fertilization

has only been examined in one study, and results varied with the successional age of the

vegetation. On 24- and 36-yr-old barrier dune islands in Virginia, root decomposition

rates were faster with N-addition, but N-fertilization had no effect on root decomposition

on 120-yr-old dunes (Conn and Day 1996).

Within a single species, it therefore appears that exogenous nutrient supply

particulary N) does not always exert a controlling influence on decomposition rates (Fog

1988). Instead, nutrient availability may have a larger influence on decomposition rates

through its effects on tissue quality. Faster rates of decomposition on more fertile sites

may therefore be faster mainly due to the presence of species with more readily

decomposable tissues on fertile sites than on infertile sites, a consequence rather than a

direct effect of exogenous nutrient supply (Hobbie 1992, Prescott 1995).










To examine the effects of soil nutrient availability and tissue quality on

decomposition of both roots and leaves, I conducted this study in a chronosequence in the

Hawaiian Islands. The sites in this chronosequence form a soil fertility gradient that

varies greatly in N- and P-availability, but where present climate, geology, and species

composition are relatively constant (Crews et al. 1995). In most ecosystems, species

composition changes completely over gradients of soil fertility, but due to the broad

ecological ranges of tree species in the Hawaiian Islands (Carlquist 1980), similar species

dominate across these sites. Additionally, the sites are part of a long-term fertilization

experiment (Vitousek and Farrington 1997), allowing for questions of nutrient limitation

to be addressed.

In this study, I describe three experiments that investigate how nutrient

availability affects decomposition rates. Decomposition was measured in unfertilized and

fertilized plots at sites across this soil fertility chronosequence. In the first approach, I

examined how natural variation in soil N- and P-availability affected decomposition of

leaves and roots by comparing decay rates in unfertilized plots along the chronosequence

(natural fertility gradient). In the second experiment, I assessed whether N or P limited

microbial decomposition of roots by comparing rates of root decomposition in

unfertilized and fertilized plots. These roots were harvested from both fertilized and

unfertilized plots and were placed back in their plot of origin (i.e., in situ decomposition).

Therefore, roots decomposing in the fertilized plots may be affected both by the enhanced

nutrient availability in the soil and any effect that such soil nutrient increases may have on

the tissue quality of roots. To examine the effects of tissue quality on root









89
decomposition, I conducted a third experiment in which roots from all sites and fertilizer

treatments were placed in a common site. Because roots at the common site were

decomposed in the same soil environment, the influence of tissue quality on

decomposition rates can be inferred. Together, these three experiments highlight the

mechanisms by which N- and P-availability can potentially influence decomposition

rates.



Methods



Study Sites


Natural fertility gradient. I compared root and leaf decomposition rates along a

natural fertility gradient that consisted of sites differing in soil age and consequently, in

N- and P-availability (Table 4-1). These sites are three of six sites described as the "long

substrate age gradient" by Crews et al. (1995). Each site has soils derived from volcanic

ash, receives a mean annual rainfall of about 2500 mm, has a mean annual temperature of

160C, is located between 1122 and 1210 m elevation, and is dominated by Metrosideros

polymorpha (Myrtaceae) (Crews et al. 1995). This tree species, known for its

morphological variation, is widespread throughout the Hawaiian Islands (Carlquist 1980).

It is the dominant canopy tree in mesic to wet forests and is one of the earliest pioneers on

recent lava flows (Aradhya et al. 1990), where it tends to form even-aged stands that are

subject to synchronous dieback (Mueller-Dombois 1985).




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ELF10OZNG_KI4ETE INGEST_TIME 2014-04-14T19:47:01Z PACKAGE AA00017682_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES