Plant Species Diversity and Community Structure in a Louisiana Coastal Marsh

Home , High marsh, Sagittaria lancifolia

Louisiana State University LSU Digital Commons

LSU Historical Dissertations and Theses Graduate School

1996 Plant Species Diversity and Community Structure in a Louisiana Coastal Marsh. Laura Gough Louisiana State University and Agricultural & Mechanical College

Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_disstheses

Recommended Citation Gough, Laura, "Plant Species Diversity and Community Structure in a Louisiana Coastal Marsh." (1996). LSU Historical Dissertations and Theses. 6342. https://digitalcommons.lsu.edu/gradschool_disstheses/6342

This Dissertation is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Historical Dissertations and Theses by an authorized administrator of LSU Digital Commons. For more information, please contact [email protected]. INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type o f computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. UMI A Bell & Howell Information Company 300 North Zeeb Road, Ann Arbor MI 48106-1346 USA 313/761-4700 800/521-0600

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PLANT SPECIES DIVERSITY AND COMMUNITY STRUCTURE IN A LOUISIANA COASTAL MARSH

A Dissertation

Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Doctor of Philosophy

The Department of Plant Biology

by Laura Gough B.S., Brown University, 1990 December 1996

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 9720356

This microform edition is protected against unauthorized copying under Title 17, United States Code.

UMI 300 North Zeeb Road Ann Arbor, MI 48103

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS

First and foremost I would like to thank my parents, Lee and Michael Gough,

and my brother, Barry Gough, for their support and love. It is difficult to put into

words my gratitude to them and how blessed I feel to have these three wonderful

people as my family. Thank you for everything.

My academic achievements as a graduate student were guided by Dr. Jim

Grace, my major professor. Jim's excitement for ecology and field work was

contagious and the knowledge he shared with me was invaluable. I am prepared to

begin my postdoctoral research with all I have learned from him. I would also like

to single out Dr. Irv Mendelssohn and Dr. Jay Geaghan, two of my committee

members, who were frequently available to me for research advice and personal

support. In addition, I thank my other committee members, Dr. Bruce Williamson

and Dr. Bobby Harville. Dr. Shirley Tucker served on my committee through my

general exams and provided me with guidance as well.

The field work involved in this research could not have been accomplished

without the help of many people. Harry Haecker assisted me in the field and in the

lab on countless occasions; without his help I might still be sorting plant samples.

Of the students who aided me, Dr. Mark Ford and Dr. Andy Baldwin were the

standouts, weathering sun, intense heat, bugs, fence building, rain, boat breakdowns,

and tedious plant censuses with humor and patience. The three of us struggled

together through the first few years of our programs, and Mark and Andy helped me

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. physically, emotionally, and intellectually throughout that time. Several others

helped in the field in pleasant and unpleasant conditions: David Bordelon, Denise

D'Abundo, Jennifer Hayden, Terry Hedderson, Beth Schussler, and Judy Teague.

Two chapters benefitted from review before submission. I thank Glenn

Guntenspergen, Bill Shipley, and Phil Grime for reviewing Chapter 2. Comments by

Andy Baldwin, Mark Ford, Jay Geaghan, Bobby Harville, and Dennis Whigham

improved the manuscript of Chapter 3.

For access to the field site at the Pearl River Wildlife Management Area, I

thank the Louisiana Department of Wildlife and Fisheries. Funding for this research

was provided by a National Science Foundation Dissertation Improvement Grant

(DEB-9310890), the U.S. Fish and Wildlife Service, and the National Biological

Service, Southern Science Center (National Wetlands Research Center).

I would like to single out several graduate students and spouses who

supported me along the way: Bill and Flo Odum, Andy and Ellen Douglas, and Steve

and Ashley Brewer. Kevin and Clare Jones have been tried and true friends through

the last couple years who I could not have done without. Judy Teague and Susan

Meiers have been invaluable neighbors and listeners on the 3rd floor.

Many others contributed to my well being and progress while in graduate

school without directly impacting my research itself. The brave women of Students

for Reproductive Freedom, Women Organizing Women, and the LSU Women's

Center have kept me idealistic and hopeful these past years as we explored our

power to affect change. I have had outstanding role models in this, particularly

iii

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ronlyn Domingue, Michelle Masse, Dana Nelson, and Robin Roberts. In the past

couple years I have been lucky enough to find several more inspirational and

supportive friends: Marcie Fisher, Sara Evans, Susan Bryson, and Lori Bertman.

Barbara Davidson has been my feminist mentor and great friend while in Baton

Rouge; I would not be who I am today without her influence in my life.

Last but not least I would like to mention two people who have been

wonderful friends and kept me going on the roughest days. Jennifer Hayden

provided incredible support to me when I first started my field work and continues to

offer me her friendship. Henry Lamousin has been my patient and supportive

companion through the stressful end of this project, which has been a downhill ride,

but not without its bumps, potholes, and surprises. To all of you mentioned in this

long acknowledgement and others I may have neglected to mention, you have given

so much to me; thank you.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS

ACKNOWLEDGEMENTS...... ii

ABSTRACT ...... vi

CHAPTER 1 INTRODUCTION...... 1

2 THE RELATIONSHIP BETWEEN SPECIES RICHNESS AND COMMUNITY BIOMASS: THE IMPORTANCE OF ENVIRONMENTAL VARIABLES...... 7

3 THE INFLUENCE OF VINES ON AN OLIGOHALINE MARSH COMMUNITY: RESULTS OF A REMOVAL AND FERTILIZATION STUDY...... 33

4 THE EFFECTS OF NUTRIENT ENRICHMENT AND HERB IVORY ON SPECIES RICHNESS AND COMMUNITY STRUCTURE IN TWO LOUISIANA COASTAL PLANT COMMUNITIES ...... 61

5 THE INTERACTION OF HERB IVORY WITH FLOODING AND SALINITY STRESS IN A COASTAL MARSH: IMPLICATIONS FOR RELATIVE SEA LEVEL RISE ...... 93

6 AN EXPERIMENTAL TEST OF A STRUCTURAL EQUATION MODEL OF SPECIES RICHNESS IN A COASTAL MARSH ...... 137

7 SUMMARY AND CONCLUSIONS ...... 165

REFERENCES ...... 169

APPENDIX: LETTER OF PERMISSION ...... 188

VITA ...... 191

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT

The factors that control plant species diversity have been examined by many

researchers, but with little consensus as to the causes of diversity patterns. Plant

diversity has been correlated with disturbance, competition, soil resources, and other

variables. In the research presented here, I investigated the relationship between

plant species richness and biomass, competition, nutrient enrichment, flooding,

salinity, and herbivory in field studies in Louisiana coastal marshes.

Biomass alone was a poor predictor of richness in two Louisiana marshes.

When environmental variables (flooding, salinity, and soil fertility) were included

with biomass in a multiple regression model, 82% of the variance in species richness

was explained. When nutrients were added, a short-term study demonstrated no

significant change in richness though biomass increased substantially. When vines

were removed in one study, biomass increased but richness did not change.

The relationship between flooding and salinity and richness was also

investigated. Increasing flooding stress or salinity levels decreased richness.

Relative sea level rise is expected to increase salt water encroachment into fresh

coastal areas as well as increase water levels. The results here confirm other work

that relative sea level rise will have detrimental effects on coastal plant communities.

Herbivory in Louisiana coastal marshes is primarily by mammals. Alone,

herbivory did not affect species richness. However, in combination with flooding

treatments, herbivory caused a greater decrease in richness. When sods were

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. transplanted from a brackish to a fresh marsh, plants were consumed by the

herbivores resulting in an unexpected interaction. Herbivores prevented a decrease in

richness with fertilization.

Although in later data sets biomass was found to correlate better with species

richness, the addition of environmental variables to a statistical model increased

predictive ability. A structural equation model developed from descriptive data

predicted 47% of the variance in species richness. When the predictions were tested,

many were met, with the exception of the interactions described above. When

experimental treatments that had not been included in the original model sampling

were eliminated from the data set, the predicted richness values explained 63% of the

variance in observed richness in the experimental manipulations.

vii

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 1

INTRODUCTION

One of the most basic questions addressed by community ecologists is how is

species diversity maintained in a given habitat. Many researchers have correlated

plant species diversity with a variety of broad gradients and variables including study

area (Gleason 1925), latitude (Currie and Paquin 1987), disturbance (Connell 1978),

precipitation or moisture levels (Whittaker and Niering 1965, Whittaker 1975),

successional time (Pickett 1976), evapotranspiration (Currie 1991), and others

(reviewed in Huston 1994, Rosenzweig 1995). On somewhat smaller scales, plant

diversity has been correlated with above-ground biomass (Al-Mufti et al. 1977, Bond

1983, Puerto et al. 1990), soil drainage (Dix and Smeins 1967), and soil nutrient

levels (Beadle 1966, others summarized in Tilman 1986). Over 100 hypotheses have

been offered to explain patterns of species richness, with little consensus (Palmer

1994), making generalizations across habitat types next to impossible. While concern

for conserving species across the globe increases, our knowledge of what controls

species co-existence is still quite limited, despite the correlative studies listed above.

In order to predict how species diversity may be affected in a given habitat by

changing climate or other environmental conditions, we must understand the

mechanisms and causal relationships behind diversity.

The relationship between plant species diversity and productivity has been

investigated by many authors, particularly the relationship between species richness

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 (simply the number of species per unit area) and biomass (the amount of plant

biomass present at any one time, usually peak growing season) (Grime 1973, Tilman

1982, Huston 1994). Richness is predicted to peak at a low to intermediate level of

biomass, causing this "hump-shaped" pattern. This relationship is easily measured in

the field, but assumes that biomass is reflecting the environmental conditions present

at the site, acting as a surrogate variable for stress and/or disturbance (Grime 1979).

In systems where species may be adapted to particular stresses, such as flooded

habitats, they may not demonstrate a decrease in biomass with increasing "stress,"

therefore not demonstrating a clear decrease in biomass in stressful environments

(Gough et al. 1994). However, some authors believe this relationship is general

enough that richness may be predicted by biomass (Moore and Keddy 1989, Wisheu

and Keddy 1989, Shipley et al. 1991).

Despite general agreement that this hump-shaped pattern exists in many

communities, there is little consensus on the mechanisms behind this relationship (for

recent reviews see Rosenzweig and Abramsky 1993, Tilman and Pacala 1993,

Abrams 1995). Most researchers agree that the decrease in richness following the

peak in biomass is due to competitive exclusion, presumably by increased light

competition (Grime 1979, Campbell and Grime 1992, Grace 1993). However, there

is some argument over the importance of competition in the initial increase in species

richness at low biomass (Wilson and Tilman 1991, 1993; Huston and DeAngelis

1994).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3 If biomass was a good surrogate variable for all the forces acting to influence

plant community structure, the relationship between biomass and richness would be

simple. However, in plant communities each species probably has a specific

response to the different variables found in that habitat, and the relationships between

species may shift along an environmental gradient due not only to different tolerance

levels but also to differing competitive ability, both among species and under

different environmental conditions. Many factors contribute to a species'

establishment and survival. Competition (Newman 1973, Schoener 1983, Connell

1983, Aarssen and Epp 1990, Goldberg and Barton 1992), herbivory (McNaughton

1986, Milchunas et al. 1988, Milchunas and Laurenroth 1993, Cyr and Pace 1993),

disturbance (Collins et al. 1995, Connell 1978), facilitation (Bertness and Callaway

1994), and interactions of the above listed factors (Menge and Sutherland 1976, Sih

et al. 1985, Grime 1993, Gruob 1985, Belsky 1992) may affect all species in a given

area, one or two, or none at all. Limiting resources in the soil or other types of

disturbance may also play a role in determining survival of a given individual, and

may thus influence the interactions of the species with each other (e.g. Tilman 1986,

Huenneke et al. 1990). New experimental techniques are being developed to try to

understand how competition works within the environmental constraints found in

field settings (Goldberg et al. 1995), but we still lack basic understanding of the

interactions of all of these factors. Most experiments are conducted on one or two

species in artificial settings which may tell us a lot about how these particular

species respond to each other and to specific environmental conditions; however, it is

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 difficult to extrapolate such results to whole communities (Wilbur 1972, Neill 1974,

Wilbur and Fauth 1990, Wootton 1993, Goldberg 1994).

There have been many studies examining the response of species along

salinity, flooding and other gradients in wetland communities (Adams 1963,

Chabreck 1972, Bertness and Ellison 1987, Day et al. 1988, Brewer and Grace 1990,

Pennings and Callaway 1992, Grillas et al. 1993, Latham et al. 1994, Ellison and

Bedford 1995). Some of these studies have looked specifically at the richness-

biomass relationship and found some degree of correlation (Wheeler and Giller 1982,

Wilson and Keddy 1988, Wheeler and Shaw 1991, Garcia et al. 1993), while others

have not (Gough et al. 1994). Researchers have examined the importance of fire

(Taylor et al. 1994), herbivory (Hik and Jefferies 1990, Taylor 1992, Nyman et al.

1993, Taylor and Grace 1995, Ford 1996, van de Koppel et al. 1996), ice scouring

(Belanger and Bedard 1994), nutrient additions (Neely and Davis 1985; Grace 1988;

Neill 1990a, b, and c), and facilitation (Bertness et al. 1992, Bertness and Hacker

1994) in wetland habitats. Since wetlands contain plants which have adapted to

various flooding and salinity regimes, we do not expect a simple reflection of stress

levels in biomass production. Wetlands, especially coastal salt marshes, are some of

the most productive habitats in the world, despite the fact that most plant species

could not survive in them (Mitsch and Gosselink 1986).

In this dissertation, I have sought to explore experimentally, as well as with

multivariate analyses, the relative importance of biotic and abiotic factors in

controlling plant species diversity. One of these factors, competition, has been

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5 investigated in wetland habitats (e.g. Keddy 1989b, Pennings and Callaway 1992,

Keddy et al. 1994), usually by means of phytometer approaches (Gaudet and Keddy

1988, 1995), removal experiments (Silander and Antonovics 1982), or reciprocal

transplants (Bertness 1991a and b). Here, the role of vine species in a coastal marsh

community was examined using a removal experiment to determine the competitive

effect of this life form on the co-occurring herbaceous vegetation; results are reported

in Chapter 3. Fertilizer was applied to plots in the vine study and in other areas in

the same river basin to determine if competitive exclusion would increase when

nutrients were no longer limiting (Chapters 3 and 4).

The main stresses in coastal communities are salinity and flooding. The work

reported in Chapter 5 addressed these two factors over three growing seasons in a

field study. The results of this experiment have important implications for relative

sea level rise which is occurring along the Gulf Coast at a much higher rate than

elsewhere in the world due to regional subsidence. By conducting the research in the

field I hoped to mimic "natural conditions" as much as possible and thus be able to

apply my results to real world situations.

Herbivory, particularly by mammals, has been examined in many marsh

habitats. In Louisiana marshes, herbivory is frequently investigated in terms of its

effect on specific species rather than whole communities, although there are

exceptions. In Chapters 4 and 5 results are presented from an exclosure study, both

in terms of effects of herbivores alone and of the interaction of herbivory with the

fertilization, salinity, and flooding treatments.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 6 In order to synthesize the above information into a model to predict richness

in coastal wetlands, three studies were performed. Chapter 2 reports the results of a

preliminary study conducted in two coastal wetlands in Louisiana examining the role

of biomass in predicting species richness. After finding almost no correlation, we

added environmental variables into the model and found a much better explanation of

richness. This simple model was developed further into a structural equation model

(Chapter 6). This model was parameterized using descriptive data and was tested

using the results from the manipulations reported in Chapters 4 and 5.

The work presented in this volume represents the evaluation of the effects of

herbivory, nutrients, competition, flooding, and salinity on plant species richness and

community structure in a coastal wetland. The model presented in Chapter 6

combines the information gained from the experimental manipulations with

descriptive data and gives us a summary of how these factors interact to determine

diversity. With this knowledge, predictions of the effects of relative sea level rise

and other changing environmental variables can be made, and our understanding of

the factors controlling plant communities, specifically species diversity, is improved.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 2

THE RELATIONSHIP BETWEEN SPECIES RICHNESS AND COMMUNITY BIOMASS: THE IMPORTANCE OF ENVIRONMENTAL VARIABLES*

INTRODUCTION

One of the most-discussed attributes of plant communities is the "hump-

shaped" relationship between species richness (species density) and community

biomass (Grime 1979). Evidence of this relationship between biomass and richness

has been reported for a wide variety of communities. In one of the first studies to

reveal this as a quantitative, general relationship, Al-Mufti et al. (1977) examined

variation in biomass (above-ground live plus dead) and species richness for 13

terrestrial herbaceous communities; in this study, peak richness was found in

communities of low total biomass. The further establishment of this relationship as a

general phenomenon was aided by Tilman (1986), who presented evidence for such a

relationship from six different published studies including examples from tropical

forests, prairies, and herbaceous communities. Various other authors have reported

findings consistent with this relationship (e.g. Huston 1980, Wheeler and Giller 1982,

Austin 1987, Shipley et al. 1991).

Several theories have been proposed to explain patterns of species richness

(Whittaker 1977, Grime 1979, Huston 1979, Tilman 1982, and see discussion by Peet

et al. 1983). While different mechanisms have been proposed, certain points are in

general agreement. First, a hump-shaped relationship between richness and

’•'Reprinted by permission of Oikos.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 community productivity suggests that there are two distinct types of processes

operating. At very low levels of productivity, richness is primarily limited by the

inability of a species to survive the abiotic conditions. In this range, increasing

community productivity reflects a decrease in the harshness of the environment.

Above some point roughly corresponding to the peak species richness, the abiotic

environment is presumably amenable to most species. At higher levels of community

productivity, the decline in richness is believed to relate in some way to a greater

degree of competitive exclusion with increasing productivity. The exact processes

behind this decline, however, are in considerable dispute (Peet et al. 1983, Grace

1991, 1993, Wilson and Tilman 1991, Reader et al. 1994).

Of the communities that have been examined for the relationship between

richness and biomass, wetlands have received significant consideration. Wheeler and

Giller (1982) examined 34 stands of herbaceous fen vegetation for both species

richness and above-ground biomass. In this study they found species richness to be

negatively correlated with the amount of above-ground plant material, and they did

not find a biomass below which species richness declined. Thus, in their system it is

presumed that all sites were sufficiently productive to be to the right of the "hump"

of peak richness. Using multiple regression, they found that they could explain 76%

of the variance in species richness based on total above-ground material (live plus

dead) collected in September.

More recently, several studies have examined the relationship between species

richness and habitat relations in wetlands (Wilson and Keddy 1988, Moore and

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 Keddy 1989, Wisheu and Keddy 1989, Shipley et al. 1991, Wheeler and Shaw 1991,

Garcia et al. 1993). These studies have further illustrated that there exists a general

relationship between species richness and community biomass. However, the studies

have also shown that a number of factors can complicate our ability to predict

species richness from community biomass.

In this paper, we present the results of a study of the relationship between

species richness and community biomass for coastal marshes in Louisiana. In

addition to measuring community biomass and species attributes, we have

characterized certain habitat properties including elevation relative to the water table,

salinity, and soil organic matter. As will be shown below, these abiotic factors can

play a critical role in regulating species richness.

THE STUDY AREAS

Coastal marshes of the Gulf of Mexico, USA, are vast in extent, spanning

some 3.5 x 10s hectares and constituting some 60% of the total coastal wetlands in

the contiguous United States (Mitsch and Gosselink 1986). Various studies have

documented the vegetative structure of gulf coastal marshes (Gosselink 1984).

Within this region, there exist strong gradients in both salinity and elevation that

influence the distribution of species. In addition, variations in soil type occur on a

small scale associated with both marsh type and hydrologic events.

Within the Louisiana coastal marsh system we selected two drainage basins

for inclusion in this study. The first of these was the Tchefuncta River basin located

on the north shore of Lake Pontchartrain. At the mouth of the Tchefuncta River a

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 marsh system exists along the lower-most 2 km of the river. Previous studies of this

marsh revealed a gradient of community types as a function of distance from the

lake, presumably caused by occasional saltwater intrusions forced by storm events

(Brewer and Grace 1990).

The other drainage basin included in this study was the Pearl River basin

located at the border between Mississippi and Louisiana. The lower part of this

drainage basin consists of an anastomosing system of distributaries embedded in a

matrix of wetlands. As with the Tchefuncta, the lower Pearl River contains a

gradient of marsh types ranging from fresh (inland) to brackish (on the coast). Based

on their species composition, these wetlands represent a broad range of the coastal

wetlands in the northern Gulf of Mexico area.

METHODS

The sampling approach used in this study was similar in concept to that used

by Wheeler and Giller (1982) except that samples were collected across gradients of

salinity and elevation. Across both drainage systems, stands for sampling were

chosen so as to maximize the range of vegetation types included. Within the

Tchefuncta system, samples were collected at three areas of varying distances from

Lake Pontchartrain representing fresh, intermediate, and brackish marsh types. At

each site, stands were sampled along transects perpendicular to shore including low,

medium, and high elevations. At Tchefuncta this resulted in a total of 16 stands

sampled. Within the Pearl River Basin, samples were collected from four regions of

varying distances from the Gulf of Mexico representing fresh, intermediate, brackish,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and salt marsh zones. A total of 17 stands was sampled from this area. In addition,

data were obtained from a concurrent study in the Pearl River Basin. In this

concurrent study there were six replicate samples of each of three vegetation types.

For inclusion in this study, data were averaged across the replicate samples yielding

one estimate for each vegetation type. In addition, data sets were cross-compared to

correct for the effects of different observers on vegetation data. Combined, the total

for the Pearl River system was 20 stands and the total for both systems was 36.

For each stand sampled, a 1-m2 quadrat was surveyed during September of

1991 for species richness, percentage of areal coverage of ground surface, and

maximum height of each species. Within each sample, the center-most 0.25 m2 was

harvested for total above-ground biomass including both live and dead material.

Further, surface soil samples (approximately 10-cm deep) were collected for later

analysis. Harvested material was dried at 80C and weighed. Soil samples were

dried and ashed at 550C for determination of soil organic matter (Moore and

Chapman 1986).

Potential richness (defined as the density of species that could occur based on

their physiological tolerances) was estimated for each plot based on published studies

of the flora (Chabreck 1972, Gosselink 1984, Brewer and Grace 1990 - for the

Tchefuncte marsh, Taylor 1992 - for the Pearl River marsh), data from this study,

and extensive field observations. In order to estimate potential richness, each plot

was classified by its watershed, salinity type (fresh, oligohaline, brackish, or saline),

and elevation. The potential species richness was estimated by determining the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 species that have been reported for a given marsh type or are known based on

physiological studies to be able to grow in that marsh type. An extensive literature

exists on coastal Louisiana plant distributions including a survey of more than 1S00

vegetation samples examined along 39 helicopter transects (Chabreck 1972).

Specifically for the Tchefuncta marsh, Brewer and Grace (1990) examined the

distributions and dynamics of the marsh vegetation over a 13-month period. For the

Pearl River marsh, Taylor (1992) studied the temporal dynamics of vegetation as a

function of salinity over two growing seasons. Finally, Gosselink (1984) summarized

some of the physiological studies of marsh plants, characterizing their limits with

regard to flooding and salinity. The data from these sources were combined with the

data from this study to provide our best estimate of the number of species (I) known

to occur in that watershed and (2) known to tolerate any particular set of flooding

and salinity conditions.

Regression analyses were performed by using the Statistical Analysis System

(SAS Institute Inc., Cary, NC, USA) and Statistix program (Analytical Software, St.

Paul, MN, USA). Data were evaluated for colinearity, normality, and homogeneity

of variances. Salinity data were fitted to a nonlinear model (Table Curve, Jandel

Scientific, San Rafael, CA, USA) to achieve linearity. No additional transformations

were required to meet parametric assumptions for the models presented. The most

parsimonious models were selected by sequentially examining combinations of

variables using Mallow's Cp criterion in conjunction with R2 values (Neter et al.

1989).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13

RESULTS

Table 2.1 shows the species encountered in the study and their life-history

types. These species represent a significant proportion of the common and dominant

species found in Louisiana coastal marshes (Chabreck 1988). Most of these species

are clonal perennials; annuals were uncommon in our study plots. In addition, most

of the dominant species were graminoids.

The relationship between total biomass and species richness is shown in

Figure 2.1. While maximum richness was negatively related to biomass, the overall

relationship was veiy weak (R2 = 0.02, p<0.01). Instead, the variable best correlated

with richness was elevation (Figure 2.2A). A strong, linear, positive relationship

existed between elevation and species richness with an R2 = 0.52 (p<0.01). At the

lowest elevations, richness averaged only slightly above one species per square meter

while at the highest elevation, the average was nearly nine.

The relationship between salinity and species richness (Figure 2.3A) was

strongly negative. Salt marsh communities dominated by Spartina altemiflora and

Juncus Roemerianus (salinity type 4) averaged approximately 2 species/m2 while

fresh and intermediate marshes averaged densities of 6.5 - 8. The relationship

between richness and salinity was nonlinear and a curve fitting program (TableCurve,

Jandel Scientific, Corte Madera, CA) was used to determine the best-fit equation.

The best fit was obtained by the equation

y = 56.1 -21.2x -30.7/x + 2.4x2

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 Table 2.1. Species encountered during study plus an indication of life-history type and dominance type.

Altemanthera philoxeroides perennial subdominant Aster laterijlorus perennial subdominant Aster subulatus annual subdominant Aster tenuifolius perennial subdominant Calystegia sepium perennial subdominant Carex sp. perennial subdominant Cladium jamaicense perennial dominant Cyperus sp. perennial subdominant Distichlis spicata perennial dominant Eleocharis cellulosa perennial subdominant Eleocharis macrostachya perennial subdominant Erianthus giganteus perennial dominant Ipomoea sagittata perennial subdominant Iris hexagona perennial subdominant Juncus Roemerianus perennial dominant Kosteletzkya virginica perennial subdominant Leersia oryzoides perennial subdominant Lythrum lineare perennial subdominant Mikania scandens perennial subdominant Osmunda regalis perennial subdominant Panicum dichotomiflorum perennial subdominant Panicum virgatum perennial dominant Phragmites communis perennial dominant Phyla lanceolata perennial subdominant Polygonum punctatum perennial dominant Pontedaria cordata perennial dominant Ptilimnium capillaceum annual subdominant Sacciolepis striata annual subdominant Sagittaria lancifolia perennial dominant Scirpus califomicus perennial dominant Scirpus maritimus perennial dominant Scirpus olneyi perennial dominant Scirpus validus perennial dominant Setaria glauca annual subdominant Spartina altemiflora perennial dominant Spartina cynosuroides perennial dominant Spartina patens perennial dominant Thelypteris palustris perennial subdominant Typha domingensis perennial dominant Vigna luteola perennial subdominant Zizaniopsis miliacea perennial dominant

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15

inm LjJ 2 8 X CJ DC 6 in lu o 4 LU CL in 2

0 1000 2000 3000 4000 5000

BIOMASS

Figure 2.1. Relationship between total, above-ground community biomass (live + dead) per m2 and number of species per m2.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16

10 A

in 8 - cn LU 2 X O X in UJ o LJ a . cn

2500 B

2000 (/) in < 1500 o m 1000 o i— 500

ELEVATION

Figure 2.2. Relationships between elevation (1 = lowest, 2 = intermediate, 3 = highest) and (A) species richness per m2, or (B) total above-ground biomass (live + dead) per m2. Error bars represent one standard error.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17

2400 r

2000 -

GO 01 1600

m 1200

2 3 SALINITY TYPE

Figure 2.3. Relationships between marsh salinity type (1 = fresh, 2 = intermediate, 3 = brackish, and 4 = saline) and (A) species richness per m2, or (B) total above ground biomass (live + dead) per m2. Error bars represent one standard error.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 18 where y = richness and x = salinity type. When salinity was expressed in this form,

regression results were R2 = 0.30.

A positive relationship between richness and soil organic matter was found

(Figure 2.4A, R2 = 0.16). Considerable scatter occurred in this relationship but there

was no clear indication of a non-linear relationship between variables.

Additional tests were conducted for correlations between richness and either

maximum plant height, percent of the species in a plot that were perennial, or

evenness. These tests revealed no significant relationships with species richness.

When all significant independent variables were considered in a stepwise

multiple regression, an overall R2 of 0.82 was obtained (Table 2.2). As can be seen

by examining the relationship between observed values and those expected based on

the multiple regression (Figure 2.5), the combination of variables used in the model

achieved a good fit of the data with no obvious signs of nonlinearities.

Examination of the relationships between community biomass and

environmental conditions (Figures 2.2B, 2.3B, and 2.4B) showed an absence of the

clear associations that richness had with these variables. For elevation, there was no

significant difference in biomass between elevations; for salinity, there was no

overall trend in biomass; and for soil organic matter, there was no significant

correlation with biomass. Thus, biomass was not well correlated with environmental

conditions.

Potential richness as a function of elevation and salinity is shown in Tables

2.3 and 2.4. Mean potential richness increased strongly from low to high elevations.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 12

in in UJ z • • • X o x

o 20 40 60 BO 100

4000

3000 cn in < o 5 2000 _l < H- o H 1000

0 20 40 60 80 100 SOIL ORGANIC MATTER

Figure 2.4. Relationships between soil organic matter (%) and (A) species richness per m2, or (B) total above-ground biomass (live + dead) per m2. The line through the data for species richness indicates a significant positive relationship (p<0.05) while the absence of a line through the total biomass data indicates the absence of a significant relationship.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2.2. Multiple regression analysis of species richness based on environmental variables.

Predictor Cumulative Variables Coefficient Std Error R-square P <

CONSTANT -3.90 1.084 0.001

BIOMASS -0.0011 0.0003 0.02 0.001

ELEVATION 3.10 0.377 0.57 0.001

SALINITY 0.51 0.137 0.69 0.001

SOIL ORGANIC 0.052 0.011 0.82 0.001

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21

2 • •• 0

8 Q Id > c r id 6 • • tn m • • o 4

2 -• ► •

0 0 EXPECTED

Figure 2.5. Illustration of the fit of a multiple regression model (expected richness) to the actual data (observed richness). The multiple regression model (Table 2.2) was an expression of: richness = f(elevation, salinity, soil organic matter, biomass).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 Table 2.3. Potential richness as a function of elevation (elevations range from 1 = most flooded to 3 = least flooded).

Mean Potential Standard Elevation Richness Error

1 3.43 0.37

2 34.10 2.22

3 43.00 0.00

Table 2.4. Potential richness as a function of salinity (salinities range from 1 = fresh marsh to 4 = saline marsh).

Mean Potential Standard Salinity Richness Error

1 34.06 3.91

2 41.33 1.67

3 23.56 2.44

4 3.67 1.67

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 Additionally, it was greatest in fresh and oligohaline marshes and declined strongly

in brackish and saline marshes.

Regression analysis for potential richness found elevation, salinity, and soil

organic matter to combine to explain 89% of the variance (Table 2.5). Community

biomass did not correlate significantly with potential richness. Potential richness

correlated strongly with realized richness (R2 = 0.72) (Table 2.6). With potential

richness in the model, environmental factors were not significant; however, biomass

was able to explain an additional 9% of the variance for a combined R2 of 81%.

When the models in Table 2.2 and Table 2.6 were compared, there was no significant

difference in their abilities to explain variance in realized richness.

In order to gain insight into the factors controlling the importance of different

independent variables, data from the most environmentally extreme conditions was

excluded and regression analysis performed. When the lowest elevation and the two

highest salinity categories were eliminated from the analysis, the relationship between

environmental conditions and biomass improved. Further, biomass became a much

better predictor of realized richness (Figure 2.6) while all other variables became

nonsignificant (Table 2.7).

DISCUSSION

A rich body of literature has discussed the possible mechanisms controlling

species richness in a plant community (Grime 1979, 1980, 1985, Huston 1979,

Tilman 1982, 1988, Peet et al. 1983, Huston and Smith 1987, Keddy 1989a). Grime

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2.5. Multiple regression analysis of potential richness.

Predictor Cumulative Variables Coefficient Std Error R-sauare P <

CONSTANT -30.18 3.74 0.001

ELEVATION 16.18 1.39 0.63 0.001

SALINITY 2.77 0.51 0.77 0.001

SOIL ORGANIC 0.25 0.039 .89 0.001

Table 2.6. Multiple regression analysis of species richness including potential richness as a predictor variable.

Predictor Cumulative Variables Coefficient Std Error R-sauare P <

CONSTANT 2.20 0.66 0.01

POT. RICHNESS 0.18 0.015 0.72 0.001

BIOMASS 0.0011 0.0003 0.81 0.001

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25

12 r

10 • • C/i cn LU z 8 X o • • • Od cn UJ o UJ Q. cn

1000 2000 3000 4000 5 0 0 0

BIOMASS

Figure 2.6. Relationship between total, above-ground c o m m u n it y biomass (live + dead) per m2 and number of species per m2 for the narrow range of conditions (excludes lowest elevation, saline, and brackish marshes). R2 for relationship was 0.34 (Table 2.7).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2.7. Multiple regression analysis of species richness excluding the lowest elevation and highest two salinity categories.

Predictor Cumulative Variables Coefficient Std Error R-square P <

CONSTANT 13.96 6.49 0.049

BIOMASS -.0020 0.00054 0.270 0.002

POT. RICHNESS -0.126 0.17 0.263 0.477

ELEVATION 1.215 1.237 0.237 0.343

SALINITY -1.690 1.018 0.324 0.119

SOIL ORGANIC 0.050 0.024 0.446 0.057

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 (1979) has pointed out that the environment can control species richness in two

distinct ways, by regulating the expression of dominance and by affecting the

potential richness (pool of suitable species). Nonetheless, during the past few years,

several authors have relied primarily on standing biomass as an indicator of

dominance in order to predict species richness in wetlands (Wheeler and Giller 1982,

Wilson and Keddy 1988, Moore and Keddy 1989, Wisheu and Keddy 1989, Shipley

et al. 1991, and Wheeler and Shaw 1991). These studies were conducted in a variety

of locations and habitat types. Wheeler and Giller (1982) confined their studies to

strictly freshwater "rich fens" in which the main sources of environmental variation

were soils, hydrology, and prior management regime. Wilson and Keddy (1988)

examined richness-biomass relationships along a shoreline exposure gradient in small,

Canadian lakes. Moore and Keddy (1989) selected a wide range of wetlands

including open beach wetlands, sheltered riverine marshes, and shoreline habitats.

Wisheu and Keddy (1989) examined four lakeshore habitats while Shipley et al.

(1991), sampled freshwater shorelines of rivers and lakes ranging from exposed,

sandy sites to protected, organic sites. Finally, Wheeler and Shaw (1991) selected 86

stands encompassing a wide range of rich fen sites to represent the range of variation

found in lowland England and Wales.

Of the above-described studies, the study by Wheeler and Giller (1982) was

the most successful in explaining the variance in richness from standing crop of

biomass. They found that a slightly curvilinear, monotonic, negative relationship

between the variables could explain 76% of the variance. Wilson and Keddy (1988),

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 in contrast, found highest diversities at intermediate biomass and used a negative

quadratic equation which explained 48% of the variance. Moore and Keddy (1989)

found a bitonic relationship between biomass and richness with very low values of

richness at the lowest biomass, peak richness at crops less than 10% of the maximum

crop, and declining richness at higher crops. They fit this relationship with a second-

order polynomial that was able to discriminate 34% of the variance in richness.

Wisheu and Keddy (1989) likewise found a bitonic relationship that was able to

discriminate 40% of the variance in richness based on community biomass. Among

the first studies to incorporate additional independent variables in their analysis was

that of Shipley et al. (1991). Because they dealt with study areas that were exposed

to considerable disturbance, they found it necessary to use the percentage of species

in a plot that were annuals as an index of disturbance. Then, using both biomass and

percent perenniality as independent variables, they were able to explain 76% of the

variance within an area and 42% of the variance across areas. Recently, Wheeler

and Shaw (1991) examined both biomass and seasonal biomass production of live

and dead material as predictors of richness. Here they found productivity to explain

slightly more variance in richness (36%) than did biomass (32%).

The hypothesis tested by the above-described studies is that biomass standing

crop can act as an adequate predictor for richness. As can be seen by the results of

these studies, considerable predictive power can be ascribed to biomass but

substantial variation in richness often remains unexplained. One of the studies that

has reported the poorest success in using biomass to predict richness is Garcia et al.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 (1993). In their study, they examined the relationship between biomass and richness

for a saltmarsh and found salinity to be a better predictor of richness than was

biomass. In our study we found that biomass (living plus dead) was very poorly

correlated with species richness. Instead, we found that elevation (as indicative of

relative flooding depth) could explain 52% of the total variance in richness. Further,

we also found that an indicator of long-term average salinity (salinity type) could

explain 30% of the variance in richness. Finally, soil organic matter explained 16%

of the variance in richness.

In order to consider the causes for the inadequacy of biomass as a predictor

of richness in our study, we present a conceptual model of the possible relationships

amongst variables (Figure 2.7) based on previous theoretical discussions (Grime

1979). According to this model, environmental conditions act in a fundamental way

by determining both the potential richness (pathway A) and the standing biomass

(pathway B). Biomass, in turn, acts to indicate the amount of competitive exclusion

(pathway C). Finally, competitive exclusion acts on the potential pool of species

(pathway D) to determine the realized pool of species (see Grime 1979, Huston 1979,

Peet et al. 1983, and Tilman 1986 for a discussion of the possible mechanisms

involved).

As can be seen from examining Figure 2.7, biomass must act as an indicator

of both environmental stresses and competition in order to predict species richness.

Examination of Figures 2.2, 2.3, and 2.4 revealed clearly that in this study

environmental conditions were correlated with richness but biomass was not

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. U> O RICHNESS REALIZED RICHNESS ► ► POTENTIAL ------► ► aiUMAbb EXCLUSION ► ------Figure 2.7. General model of proposed relationships amongst variables controlling realized (observed) richness. See text for explanation. ENVIRONMENTAL B n.nMAQQ c COMPETITIVE D CONDITIONS

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 correlated with environmental conditions. Therefore, the influence of environmental

conditions on richness could not be assessed strictly from biomass.

There are several possible reasons why biomass may not reflect environmental

conditions. For example, flooding can be viewed as a stress for nonadapted species.

However, for those species that have evolved adaptations, increased water depth does

not necessarily result in greatly decreased biomass (Grace and Wetzel 1981, Grace

1989). Figure 2.2 shows that while richness declined with decreasing elevation,

biomass did not. Further, Table 2.3 shows that the potential richness also declined

with flooding stress. Therefore, biomass did not reflect that flooding was a stress for

the species adapted to it, only for those not adapted. Likewise, evidence suggests

that community biomass is not generally reduced by increasing salinity (Odum 1988)

and our results showed only a weak influence of salinity on biomass (Figure 2.3) but

a strong influence on both potential (Table 2.3) and realized (Figure 2.3) richness.

Analyses of potential richness are consistent with the above interpretation.

Potential richness can be a difficult variable to estimate. In this case, we simply

estimated potential richness as the species known to be able to occur under any

particular salinity and flooding regime. Nonetheless, our quantitative estimates of

potential richness yield results quite consistent with the qualitative model presented

in Figure 2.7. Further, when the most extreme environmental conditions were

deleted from the analysis, the predictive capacity of biomass was substantially

improved while the role of potential richness was effectively eliminated (Table 2.7).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 In conclusion, in this study we found that biomass was not an adequate

predictor of species richness. One reason for this inadequacy appears to be that

while stresses such as flooding and salinity may greatly reduce the pool of potential

species that can occur at a site, those species that have evolved adaptations to these

factors may not have substantially reduced biomasses. Thus, we recommend that

models developed to predict species richness should incorporate direct measurements

of environmental factors as well as community attributes such as biomass in order to

increase their applicability.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3

THE INFLUENCE OF VINES ON AN OLIGOHALINE MARSH COMMUNITY: RESULTS OF A REMOVAL AND FERTILIZATION STUDY

INTRODUCTION

The effects of vines on non-vine plants have frequently been observed but

seldom investigated experimentally (Gentry 1991). It has been suggested that vines

may have a number of effects on neighboring plants including competitive

suppression through shading and below-ground interactions (Whigham 1984,

Dillenburg et al. 1993a) and physical interference with the growth or survival of

supporting plants (Stevens 1987, Hegarty 1991). Hegarty (1991) has noted that leaf

area and fecundity of host trees may be reduced by abundant vines and that twining

vines may interfere with downward translocation in trees. Stevens (1987) removed

lianas from five Bursera simaruba trees and saw fecundity increase by an average of

148% as compared to controls and previous years' records of the manipulated trees.

Whigham (1984) removed vines from Liquidambar styraciflua hosts and documented

an increase in tree growth; he concluded this result was due to release from

belowground competition since when vines were removed from trunks and crowns

but not uprooted, there was no improvement in growth. Dillenburg et al. (1993a)

found an increase in allocation to leaf area and biomass by host trees when

competing aboveground, but not belowground, with vines. As Stevens (1987) points

out, observations of effects of vines must be tested with manipulative experiments in

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 order to discount the possible phenomenon of vines being present in greater numbers

on less healthy or less fecund host plants.

Most of the research investigating the effects of vines on their plant hosts has

been conducted in the tropics on the effects of lianas on trees (e.g. Putz 1984,

Stevens 1987, several chapters in Putz and Mooney 1991). Although there have been

several studies published on the effect of vines on trees in the temperate zone

(Siccama et al. 1976; Dillenburg et al. 1993a and b; Dillenburg et al. 1995), very few

have examined the effect of herbaceous vines on other herbaceous species. Friedland

and Smith (1982) found that reproductive effort of Solidago rugosa was not affected

by Lonicera japonica; however, those herbs with vines were taller and had greater

mean overall dry weight suggesting an increased growth rate by those herbs in

competition with vines. Penfound (1974) describes a case of a vine species,

Strophostyles helvola, eliminating Johnson grass from an old field and causing a shift

towards a forest community. Although the work was not conducted on vines,

Johnson (1994) recorded increased fruit production and leaf damage of Asclepias

syriaca (milkweed) plants bound by Spartina pectinata leaves as compared to

controls with no Spartina leaves. There has been little research conducted to

examine vine species effects on communities as a whole, except as part of larger

removal experiments in which many species are removed and community response is

noted (e.g. Allen and Forman 1976).

Vines are a prominent feature of many wetland communities (Gore 1983). In

the Gulf Coast of Louisiana, a number of vine species occur abundantly in coastal

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 marshes (Penfound & Hathaway 1938; Chabreck 1972; Conner et al. 1986). Many

studies have examined competitive interactions among wetland plants (e.g. Silander

and Antonovics 1982, Gaudet and Keddy 1988, Twolan-Strutt and Keddy 1996).

While the competitive interactions amongst other life forms have been experimentally

investigated in wetlands (clonal grasses vs. forb seedlings: Bertness and Ellison 1987;

shrub effects on herbs: Keddy 1989b), the effects of vines on non-vine species have

not. The objective of this study was to experimentally determine the effects of vines

on the growth, species composition, and diversity of non-vine species in an emergent

coastal marsh over two growing seasons.

METHODS

Study Area

The study area was located within the Pearl River Wildlife Management Area,

on the boundary between Louisiana and Mississippi. The Pearl River system is made

up of three main channels, the East, Middle and West Pearl Rivers, which empty into

the Gulf of Mexico. The study site is located on one of the branches of the East

Pearl in an oligohaline marsh dominated primarily by Sagittaria lancifolia and

Spartina patens. The area was chosen in early spring 1994 based on observations of

large numbers of vines wrapped around and growing on top of the senesced

vegetation. Three vine species occurred at the study site, Vigna luteola, Mikania

scandens, and Ipomoea sagittata. Salinity levels of water on the marsh surface range

from 0-3 ppt, and water levels are mostly contingent on wind and storm activity, but

typically low in winter and highest in early spring (Laura Gough, unpublished data).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 The three vine species present have different growth habits. Vigna luteola is

a nitrogen-fixing legume which produces nodules up to 1 cm in diameter on its roots

(Laura Gough, pers. obs.; Crozier et al. 1994). It grows upright from seed, climbing

over the canopy of the other plants present and twining around stems and leaves for

support. During the mid to late summer, V luteola can dominate the appearance of

this site such that the entire marsh appears covered in yellow flowers. Mikania

scandens typically grows as a trailing vine along the ground, rooting at nodes and

occasionally reaching the canopy top. Ipomoea sagittata, which has a growth form

similar to V. luteola, was the least common vine at the site. Although all three vines

are classified as perennials by some authors (Radford et al. 1968; Tiner 1993), we

could find no evidence that V. luteola or M. scandens maintains underground

structures over winter. These three vine species are found in marshes throughout the

Pearl River basin and coastal Louisiana with M. scandens restricted to fresh or

oligohaline sites while I. sagittata and V luteola are sometimes present in more

brackish areas (Chabreck 1972; White 1979; James B. Grace, unpublished data).

Experimental Design

The experimental design was a two by two factorial with two blocking

factors. The two treatments were vine removal and fertilization. The first blocking

factor was transect, and the second was plot location along the transect. Each

transect was positioned parallel to the shore of the East Pearl River, with #1 located

approximately 10 m from the shore and #8 farthest inland. The second block

("plot") was included because a 1 m wide channel bisects the study area: plots 2 and

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 37 3 were adjacent to the channel while 1 and 4 were farther away. One-meter-square

plots were set up along the transects at a distance of at least 1 m apart, depending on

proximity to the channel. The treatments were: RF: removal, fertilized; RC: removal,

control; NC: non-removal, control; and NF: non-removal, fertilized, with all four

treatment levels randomly located along each transect Permanent boardwalks were

established between transects in order to prevent disturbance of the marsh substrate.

The 32 plots were marked and censused prior to any treatments in early June 1994.

The removal of vines was conducted by unwinding them from their

supporting plants and removing stems and roots completely from the plot. The vines

rooted within a 0.5 m buffer surrounding the plot were also uprooted to prevent

invasion between removal treatments. Beginning in June, removals were conducted

every other week. After July, vines were not observed sprouting inside the plot but

the removal was continued as vines from surrounding plots easily grew over the

canopy top, in the case of V. luteola, or along the soil surface, like M. scandens.

The removal was discontinued after mid-September and resumed the following May

until the conclusion of the study. There was no evidence of persistent soil

disturbance following uprooting; the sediment is fine and loose at the study site and

sedimentation by periodic river flooding rapidly fills in any holes left from vine

roots.

The fertilization was conducted using Osmocote 20-10-5 tree and shrub

planting tablets (Grace-Sierra, Milpitas, CA) in June 1994 and April 1995. The

pellets are designed for one-year slow release. Due to the saturated soil conditions,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 release is probably faster, hence the reapplication less than a year following die first

treatment Twenty-four pellets were inserted in each plot to result in 17 g/m2 of

available nitrogen as ureaformaldehyde. The pellets were placed in each plot in a 5

x 5 grid with one missing from the middle row, modified as necessary by the plants

and rhizomes present. Pellets were inserted approximately 10 cm into the soil and

covered with sediment.

Data Collection

Censuses were conducted at two-month intervals from June 1994 until August

1995. At each census date, the species present along with their percent cover,

number of stems, average height, and average leaf blade width (for S. lancifolia only)

were recorded. Because vines in a plot may not be rooted there, stems cannot be

accurately counted due to their branching and rooting patterns. Their length was

nearly impossible to measure without damaging the other plants, so aerial percent

cover was used to reflect the vine biomass present. Also at each census date, light

levels 10 cm above the marsh surface were recorded by using a one-meter-long light

wand that integrates light levels along its length (LiCor Inc., Lincoln, NE). The

meter was placed with a consistent orientation diagonally across each plot. Ambient

light levels were recorded at both ends of each transect and as necessary (due to

temporary cloud cover) to calculate percentage of ambient light penetrating the

canopy.

At the end of the experiment, in August 1995, the center 0.25 m2 of each plot

was harvested for live and dead above-ground material, sorted to species, dried for

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 72 hours at 80° C and weighed. Sagittaria lancifolia biomass was individually

divided into live and dead categories since many plants contained a mixture of dead

and live ramets. Dead above-ground material that could not be identified to species

was grouped together as "dead." The 0.25 m2 values were multiplied by four to

estimate m2 biomass for analysis.

Data Analysis

Species richness, light levels, and percent cover of individual species were

analyzed using PROC GLM (SAS Institute, Cary, NC) with a repeated statement to

perform repeated measures analysis of variance with the main effects as outlined

above. ANOVA was performed on the biomass data at harvest Mean differences

were determined using Tukey’s HSD test with an alpha level of 0.05, unless

otherwise noted. A biomass regression was performed for S. lancifolia from data

collected at each census. The equation was calculated from an aboveground harvest

performed near the study site in 1993 for this purpose. The equation used was:

biomass = 0.54 + 0.0036(avght)(width)(stems)

where avght=average stem height measured in the field, width=average leaf blade

width, and stems=stem number in a 1 m2 plot. Normality was determined using the

Shapiro-Wilks test and homogeneity of variance was supported by examination of

residual plots. This equation explained 98% of the variance in biomass in the

original sample (p<0.0001, df=16).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 RESULTS

Vine Species

At the beginning of the experiment the non-removal and removal plots had

similar vine coverage and species richness. By the final harvest, removal plots had a

mean overall richness of five species per m2 compared to seven species in the non-

removal plots indicating that on average two vine species were removed. The most

frequently removed species were M. scandens and V. luteola', I. sagittata's mean

cover never exceeded 5% and was found in few plots within the study area.

The highest percent cover of vines was recorded in August 1994, with

approximately 80% of the aerial cover of the non-removal plots covered by vines

(Figure 3.1). Fertilization appeared to have a slight effect on the vines present in

these plots, postponing senescence. The peak was not seen again in August 1995

although the percent cover in June was similar to that of the previous year. Most of

the total coverage is caused by V luteola', in August 1994 Vigna composed 76% of

the total vine cover in non-removal plots. This is confirmed by the biomass data

from the harvest in whichM. scandens and I. sagittata were found in too few plots

to compare results (2 and 4 out of 16, respectively); V luteola was present in most

harvested plots (13 out of 16) but there was no significant difference between

fertilized and control plots (p=0.29).

In the non-removal plots, fertilization did not have a significant effect on

percent cover of the vine species except in December 1994 when the cover in the

fertilized plots exceeded that in the unfertilized, and in August 1995 when the control

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Removal, Control Removal, FeitiHzod Non-removal, Control “ Non-iemovol.Foitlllzod Census Date Jun 94Aug 940ct 94 Dec 94Feb 95 Apr 95 Jun 95Aug 95 Figure 3.1. Mean aerial percent cover by census date for the three vine species combined. Error bars represent means ±1 S.E. 20 8 0 100 -* 0 0 (C o >_ C a) CD c § 4 Q_ ° 6 0

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 plots had greater cover (Figure 3.1; p<0.05 both dates). The difference in December

was caused by the fertilizer treatment delaying senescence of M. scandens. In

August 1995, V luteola increased cover in the control plots but not in the fertilized;

the other two species did not change in percent cover over that time interval.

Light Measurements

Light measurements were recorded at all census dates except February 1995

when a series of high water events and low light levels due to thick cloud cover

prevented data collection. The main effect of census date was significant (p<0.000l)

over the course of the study. There was no significant interaction between the

fertilization and removal treatments from the repeated measures analysis (p=0.35) so

the results are shown separately in Figure 3.2 (see Figure 3.3a for individual

treatment means). As expected, the amount of light penetration reflects the seasonal

pattern of vegetation growth with the least amount of light during the growing season

and the most in spring before S. lancifolia and S. patens had resprouted fully and the

other species had grown beyond the initial seedling stage.

In the repeated measures analysis, fertilization caused a significant change in

light levels throughout the experiment (p=0.007). Fertilization caused a significant

decrease in light penetration at each date except June and December 1994, and April

1995 (Figure 3.2a; p<0.05). By August 1995 the fertilization treatment was causing

a 40% decrease in light penetration through the canopy relative to the control (Figure

3.2a, Table 3.1).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 100

JZa

e £o

Census Oate

Control -♦* Fertilized

b too

-> 60

Ju94 Au94 Oc94 De94 Fe95 Ap95 Ju9S Au95 Census Oate

-£- Non-removal -*• Removal

Figure 3.2. Mean light penetration by census date measured as percent of ambient light: (a) by fertilization treatment; (b) by removal treatment. Significant differences are indicated by asterisk for p<0.05 and m for 0.05

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. N C N F R T re a tm e n t

T re a tm e n t

C 1500 04 E 31000 CO CO ICO s 500

0 T re a tm e n t

Figure 3.3. Light, species richness, and biomass data from the final census and harvest, August 1995, by treatment: (a) mean percent ambient light penetration; (b) mean species richness; and (c) mean biomass at harvest. Treatment abbreviations: NC=non-removal, control; NF=non-removal, fertilized; RC=removal, control; RF=removal, fertilized. Different letters indicate means are significantly different at the p<0.05 level.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.1. ANOVA source table for percent ambient light penetration in August 1995.

Source df SS MS F-value p-value

transect 7 0.034 0.005 0.60 0.75

plot 3 0.007 0.002 0.29 0.83

removal 1 0.028 0.028 3.34 0.08

fertilization 1 0.058 0.058 6.99 0.02

removaI*fertiIization 1 0.005 0.005 0.60 0.44

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 A significant difference in light levels was not observed in the removal

treatments except in August 1994 (Figure 3.2b; p<0.05), when the vines were at their

maximum cover. In August 1994 only, there was a significant interaction between

fertilization and removal (p<0.05) because the difference between NF plots and RF

plots was less than the difference between die non-fertilized plots, therefore the main

effect of removal should be noted cautiously for this date. Although it is not a

significant change, it is interesting to note that between October and December 1994,

the light in the removal plots decreased relative to the non-removal plots for the

remainder of the study. The difference in light levels due to removal in August 1995

is marginally significant (Figure 3.2b, Table 3.1) (also see Figure 3.3a).

Species Richness

The species list for all species recorded within the study plots throughout the

experiment is presented in Table 3.2. The maximum richness observed at any date in

any plot was 10 (12 including vines) and the minimum number present in any

treatment was 2. During the first growing season, the data for all four treatments

followed much the same pattern; richness steadily decreased as the growing season

progressed, to a low in December and February (Figure 3.4). In April there was a

sharp increase in species richness due to the appearance of spring annuals. These

species then dropped out of the plots during the summer leading to decreased

richness in August since few annuals appear in the late summer or fall. This

seasonal response caused census date to be a significant effect in the repeated

measures analysis (p<0.0001).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 Table 3.2. Species occurring within the study plots with Iife-history type and dominance type.*

Agalinis purpurea annual subdominant Altemanthera philoxeroides perennial subdominant Aster subulatus annual subdominant Aster tenuifolius perennial subdominant Baccharis halimifolia perennial subdominant Cyperus haspans perennial subdominant Eleocharis cellulosa perennial subdominant Galium tinctorium perennial subdominant Hydrocotyle bonariensis perennial subdominant Hymenocallis caroliniana perennial subdominant Ipomoea sagittata perennial subdominant Iris giganticerulea perennial subdominant Juncus romerianus perennial subdominant Kosteletzkya virginica perennial subdominant Leersia oryzoides perennial subdominant Lobelia cardinalis perennial subdominant Lythrum lineare perennial subdominant Mikania scandens perennial subdominant Panicum virgatum perennial subdominant Phyla nodiflora perennial subdominant Pluchea purpurascens annual subdominant Polygonum punctatum perennial dominant Ptilimnium capillaceum annual subdominant Sacciolepis striata annual subdominant Sagittaria lancifolia perennial dominant Scirpus validus perennial dominant Solidago spp. perennial subdominant Spartina patens perennial dominant Vigna luteola perennial subdominant

*Nomenclature based on Godfrey & Wooten (1979, 1981).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48

7 —

w - ®y 5 aei

Ju94 Au94 Oc94 De94 Fe95 ApS5 Ju95 Au95 Census Date

-e- Control Fertilized

w _ 3 5

Ju94 Au94 Oc94 De94 Fe95 Ap95 Ju95 Au95 Census Date

-6- Non-removal Removal

Figure 3 .4. Mean species richness (number of species/m2) by census date: (a) by fertilization treatment and (b) by removal treatment. Significant differences are indicated by m for 0.05

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 Repeated measures ANOVA showed no significant overall treatment effects

of fertilization or removal on richness throughout the study, nor was there a

significant main effect interaction (Figure 3.4; all p-values>0.05). Date by date

comparisons, however, did reveal a modest increase in richness due to vine removal

in April 1995 (Figure 3.4b; p=0.06), the period of highest richness, though that effect

did not persist. There was a marginally significant difference between removal

treatments in August 1995 (Figure 3.4b, Table 3.3).

Biomass

Examining the biomass of the non-vine species, there was no significant

interaction between the removal and fertilization treatments, but both vine removal

and fertilization significantly increased biomass (Table 3.4, Figure 3.3c). When

examining each species individually, the biomass means were greater in the removal

treatment than the non-removal for all species, but not significantly. The individual

species that demonstrated a significant increase in mass with fertilization were S.

lancifolia (both live and dead) (p<0.0001 for both) and Polygonum punctatum (Table

3.5; p=0.0062). All other species showed higher biomass in the fertilized plots but

not significantly. The other dominant species at the site, S. patens, showed no

significant effect of either fertilization or removal treatments on biomass (Table 3.5).

Treatment Effects on Sagittaria lancifolia

The non-vine species that contributed the greatest amount of biomass and

aerial percent cover was Sagittaria lancifolia, an herbaceous clonal perennial.

Sagittaria lancifolia responded to fertilizer with significant increases in cover and

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.3. ANOVA source table for species richness in August 1995 (number of species/m2).

Source df SS MS F-value p-value

transect 7 20.219 2.888 2.88 0.03

plot 3 2.594 0.864 0.86 0.48

removal 1 3.781 3.781 3.77 0.07

fertilization 1 2.531 2.531 2.52 0.13

removal *ferti lization 1 0.281 0.281 0.28 0.60

Table 3.4. ANOVA source table for biomass (g/m2) of non-vine species at harvest, August 1995.

Source df SS MS F-value p-v

transect 7 76,282 10,897 1.36 0.28

plot 3 39,207 13,069 1.63 0.22

removal 1 39,275 39,275 4.89 0.04

fertilization 1 95,479 95,479 11.89 <0.01

remova!*fertilization 1 751 751 0.09 0.76

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 51 Table 3.5. ANOVA results of biomass analysis for specific species (g/m2). Means with the same superscript are not significantly different at the 0.05 level except for those marked with an asterisk which are different at die 0.06 level.

Treatment

Species NC NF RC RF

Polygonum punctatum 32.52* 128.44* 32.44** 227.04b*

Sagittaria lancifolia—dead 229.2* 382.84* 225.32* 491.88"

Sagittaria lancifolia—live 283.84* 406.92"° 307.04“ 496.36"

Spartina patens 177.08* 222.4* 249.72* 307.28*

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 biomass (Figure 3.5; p<0.0001). The results of the analysis on percent cover showed

the same pattern as those of die biomass estimates, so only the biomass results will

be presented here. In the repeated measures analysis census date was significant

(pO.OOOl). Results indicate that fertilization caused the plants to senesce in late

October, as opposed to September in the non-fertilized plots, and to resprout earlier.

In the spring of 1995, before the second fertilizer application, those plants that had

been fertilized the prior year showed earlier and more robust growth indicating an

effect from the 1994 treatment. Fertilization increased cover and biomass in August

1994 and June 1995 (Figure 3.5) and significantly interacted with time in the

repeated measures analysis (pO.OOOl). At other dates there was less of a difference,

especially October through February, when the plants had largely senesced.

There was no significant effect of vine removal on the estimated biomass of

S. lancifolia. Only at the conclusion of the experiment was there a significant

difference in percent cover of S. lancifolia due to removal; in August 1995 the cover

in the removal plots was 35% and in the non-removal plots was 29% (p<0.05).

However, this difference was not significant in the biomass at harvest.

Fertilization treatment caused a significant increase in live and dead biomass

of S. lancifolia at harvest (Table 3.5; pO.OOOl) while the removal and removal by

fertilization interaction were not significant (p>0.10). These differences appear

among the four means for the live and dead biomass combined as those of the live

biomass only (p<0.05).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Removal, Fertilized at each census date. Biomass Census Date Sagittaria lancifolia Jun 94Aug 94Oct 94 Dec 94Feb 95 Apr 95 Jun 95Aug 95 calculated calculated using linear regression equation presented in Methods. Error bars Figure 3.5, Biomass estimates for represent means ±1 S.E. 50 1 0 0 150 25 0 3 0 0

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 DISCUSSION

Competition has been examined in many habitats and among many organisms.

Connell (1983), in his review of ecological field studies, found that 90% of field

studies designed to examine competitive effects found evidence of competition,

although this was true for only 40% of the experiments within the larger studies. A

similar review by Schoener (1983) also concluded that 90% of field studies showed

evidence of competitive interactions, while 57% of these demonstrated competition

under all conditions and for all species tested. In a more recent analysis specifically

aimed at interactions among plants, Aarssen and Epp (1990) reviewed more than 50

studies and found evidence of competition in all. Goldberg and Barton (1992) also

reviewed plant competition studies and pointed out that 63% of the 89 studies they

looked at only addressed the affect of competition on the individual fitness of a

single species at a single time and site, and most experiments examined short-term

responses (one growing season). Most studies included in Aarssen and Epp's review

(1990) were removal studies, but most examined the effects of removal on only one

target species and did not measure community response.

The majority of competition studies in wetland plant communities have found

significant competitive interactions. Bertness (1991a and b) demonstrated that the

distinct zonation in a New England salt marsh is a result of the competitively

subordinate species being displaced to the physically stressful tidal edge of the

marsh. Grace and Wetzel (1981) documented a similar phenomenon with two

species of Typha, where the less competitively successful species was displaced to

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55

deeper water. In a study examining effects of species removals alone or in

combination, Silander and Antonovics (1982) reported eight out of 48 species

removals caused a significant change in response of remaining species; these authors

suggest there are specific and diffuse effects of competition within a given wetland

community and these effects vary depending on the environmental conditions.

Diffuse competition has been found to be an important factor in species survival and

distribution along a Iakeshore gradient (Wilson and Keddy 1986).

To date, there have been no experimental investigations of the effects of vine

removals on wetland plant communities. In this study we found that continuous vine

removal over a 15 month period resulted in an increase in total biomass by the non

vine component of the community, indicating that vines suppressed other species.

Analysis of the responses of individual species showed no evidence that any specific

species was suppressed by vines more than any other. Thus, the effects of vines on

non-vine members of the community would appear to have been diffuse (Goldberg

and Werner 1983).

The increase in community biomass due to vine removal was clear even

though vine abundance was much lower in the second growing season when

compared to the first (Figure 3.1). Presumably the non-vine species are responding

with increased biomass in the second growing season due to the 1994 removal. The

reason for this carryover effect is not clear, although it may be related to the manner

in which the vines pull over the standing grasses in late fall, causing the senesced

vegetation to form a dense mat near the substrate surface. Throughout the growing

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 season the mechanism of suppression by vines of the non-vine species could be

through physical interference with leaves and stems, belowground competition,

shading higher up in the canopy than was measured, or a combination of these

factors. The importance of belowground competition by vines in limiting nutrient

acquisition by trees has been documented (Whigham 1984, Dillenburg et al. 1993a

and b), but has not been examined in herbaceous communities.

In an earlier study in the Pearl River basin, White (1979) did not record M.

scandens, but noted I. sagittata and V luteola at many locations in the lower portion

of the basin and in some of the nearby swamps. Vigna luteola has also been noted

as an early colonizer of the newly formed mud islands in the Atchafalaya Delta,

Louisiana with now stable abundance (Shaffer et al. 1992). In the same study, M.

scandens was noted as initially covering a wide area but disappearing from low

elevations. In a recent study attempting to restore Taxodium distichum (baldcypress)

to marsh areas, I. sagittata and V luteola were found to smother baldcypress

seedlings and prevent healthy growth, although the seedlings did grow longer in the

presence of vines (Myers et al. 1995). Therefore these vines may be important to the

vegetation structure of these marsh systems, but this study represents the first attempt

at addressing their effects on neighboring plants experimentally.

The release from competition resulting from removing a dominant species has

sometimes been found to result in increases in species richness (Armesto and Pickett

1986, Gurevitch and Unnasch 1989, Keddy 1989b), though in other cases changes

were not detected (Pinder 1975, Allen and Forman 1976). In one removal

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57

experiment that was part of a larger study examining species distributions in a

subalpine meadow, richness changed under some removal conditions but not under

others (del Moral 1983). In the study reported here, a modest increase in richness

(1 species/plot, p<0.05) due to vine removal was observed in April 1995, during the

season of highest richness (Figure 3.4b). Later that same year, after richness had

declined, values were found to be only marginally higher in the plots with vines

removed (p>0.05).

Measurements of light penetration at ground level in plots with and without

vines removed (Figure 3.2) provide some insight into how the community responded

to the removal. Early in the study, in August 1994, light penetration was somewhat

higher in plots that had vines removed (9% compared to 4% for intact plots,

pO.OOOl). This difference disappeared during the senescent period from October

1994 to April 1995 when light levels were higher. Rather, in August of 1995 at the

end of the study, light penetration to the ground was slightly lower for plots without

vines than for plots with vine coverage (14% versus 19%, p=0.08). Based on these

results, combined with the observation that vine removal led to a 25% increase in

biomass, it appears that removal of vines led to an increase in growth of non-vine

species that resulted in equivalent or even greater shading at the soil surface than was

found when vines were left intact. Thus, the evidence suggests that removal of vines

led to an initial release from competition and a modest increase in richness during

the spring of the second season. However, as the second growing season progressed,

the non-vine canopy developed to levels equal to or exceeding that of intact plots,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58

resulting in no difference in richness at the conclusion of the study. In a previous

study that included the Pearl River basin, Gough et al. (1994) found no relationship

between biomass and richness over the same range of biomass presented in this

paper.

Numerous authors support the contention that increased soil fertility leads to

increased competitive exclusion, based on natural and experimental fertility gradients

(Grime 1979, Silvertown 1980, Tilman 1982, Gurevitch and Unnasch 1989). In one

of the longest ecological experiments on record, richness, diversity, and evenness

have declined in fertilized plots when compared to controls for over 100 years in the

Park Grass Experiment (Silvertown 1980). Experimental evidence also indicates that

increased soil fertility and biomass leads to an intensification of aboveground

competition (Wilson and Tilman 1991, 1993), although some authors dispute these

findings (DiTommasso and Aarssen 1991, Campbell and Grime 1992). In this study,

fertilization led to an approximately 40% increase in community biomass by the end

of the second summer. Fertilization did not differentially favor vines but instead led

to a decrease in their abundance while resulting in an increase in S. lancifolia. It is

possible that fertilization may have selectively disadvantaged the most abundant vine,

V. luteola, which is known to produce nodules capable of nitrogen fixation (Crozier

et al. 1994). Previous competition studies involving nitrogen fixing and non-

symbiotic species have found the addition of nutrients to favor those species not

relying on nitrogen fixation (Harper 1977). Whether this explanation accounts for

the decrease in vines due to fertilization in this study is unknown. Gurevitch and

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 Unnasch (1989) noted the nitrogen fixing legume Vida cracca performed better at

low fertility levels although it was still present at high fertility levels along with the

dominants.

Light penetration was lowered during the first growing season by an increase

in cover due to fertilization; this occurred again in the second growing season. Part

of the effect of fertilization on light penetration resulted from delayed senescence and

earlier regrowth in fertilized plots. However, it is clear that the 40% increase in

biomass due to fertilization contributed to overall greater shading during the second

growing season. While it is likely that this enhanced biomass and decreased light

penetration may have resulted in increased competition in the community, the effect

was not great enough to lead to a significant decrease in species richness (reduction

in richness due to fertilization^ 6%, p=0.15). When considered with the results from

the vine removal treatments, it does not appear that fertilization led to a substantial

intensification of competitive exclusion or increased the suppressing effects of vines

on non-vine species.

The fertilization and removal treatments did not have a significant interaction

for species richness. Our results contrast with those of Gurevitch and Unnasch

(1989) who removed a dominant grass from an old field community and fertilized

plots in a design similar to ours. After two growing seasons had passed, they

recorded significant decreases in species richness from fertilization alone and removal

alone, and an increase in richness in plots that were subject to both treatments.

Surprisingly, fertilization did not cause an increase in biomass in their study. Overall

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 removal had a larger effect on biomass of individual species than fertilization. The

dominant grass in their system composed approximately 50% of the productivity at

the sites which is quite different from the biomass accumulated by vines at our site,

and may account for some of the differences from our results.

In conclusion, vine species appear to play an important role in wetland

communities of Louisiana. In the oligohaline marsh where this study was conducted,

vine removal caused an increase in total community biomass as well as an increase

in biomass of non-vine species individually. The removal effect carried over from

the first growing season to the second, since differences were noted in April 1995

before the vines grew enough to be important components of the community.

Fertilization also caused an increase in biomass, with corresponding lower light

levels, although it did not intensify the suppressive effect of the vines. The removal

appeared to release the other species from competition initially, but during the peak

of the growing season the increase in community biomass caused light levels to be as

low as in those plots with vines intact. This study demonstrated the suppressive

effects that species with a particular growth form, in this case vines, may have on co

occurring herbaceous vegetation.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 4

THE EFFECTS OF NUTRIENT ENRICHMENT AND HERBIVORY ON SPECIES RICHNESS AND COMMUNITY STRUCTURE IN TWO LOUISIANA COASTAL PLANT COMMUNITIES

INTRODUCTION

The importance of competitive exclusion as a major factor regulating species

diversity has been generally supported by studies of productivity-diversity

relationships. Numerous studies have shown decreases in plant diversity resulting

from increases in productivity associated with resource enrichment (Milton 1940,

Murphy 1960, Willis 1963, Bakelaar and Odum 1978, Vermeer 1986, Tilman 1987,

Gurevitch and Unnasch 1989, Carson and Pickett 1990, Huenneke et al. 1990,

DiTommaso and Aarssen 1991, Wilson and Tilman 1993). One of the earliest

reports that nutrient additions can lead to decreases in terrestrial plant diversity came

from the first years of the Park Grass Experiment at Rothamsted, England (Lawes et

al. 1882). More recent analyses of this ongoing experiment have confirmed the

negative effects of fertilization on species diversity (Thurston 1969, Williams 1978,

Huston 1979, Silvertown 1980, Tilman 1982). Consistent with these studies, several

authors have demonstrated a general hump-shaped correlation between diversity and

community biomass/productivity/soil fertility, with declining diversity at low and

high biomass values (Grime 1973, Al-Mufti et al. 1977, Wheeler and Giller 1982,

Wilson and Keddy 1988, Moore and Keddy 1989, Wisheu and Keddy 1989, Shipley

et al. 1991, Wheeler and Shaw 1991, Marrs et al. 1996).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 62 A number of theories have been proposed to explain more precisely how

competition and productivity relate to diversity (Grime 1977; Huston 1979, 1994;

Tilman 1982, 1988; Keddy 1990; Taylor et al. 1990; see also review of additional

theories in Rosenzweig 1995). Most, though not all, of these theories are in

agreement that at high levels of community biomass competitive exclusion acts to

limit species diversity. To a great degree, results from fertilization studies support

this general proposition.

An important factor believed to potentially affect the relationship between

productivity and plant diversity is herbivory (Rosenzweig 1971, Fretwell 1977, Grime

1979, Huston 1979, Oksanen et al. 1981, Crawley 1983, Oksanen 1990). For

example, Milton (1940, 1947) observed that periodic haying of long-term fertilization

studies acted to retard the loss of species, suggesting that herbivory would have a

similar effect. As Milchunas et al. (1988, 1993) and Louda et al. (1990) point out,

the effects of herbivores on competitive interactions and plant species diversity are

potentially complex. As these and other studies indicate, in some cases the effects of

herbivory will be simply to reduce the effects of competition (e.g. Reader 1992).

However in other cases the effects of herbivory will be influenced by coevolutionary

history, environmental gradients including productivity gradients, the selectivity of

herbivores, and of course the intensity of the herbivory itself (Bazely and Jefferies

1986, Swank and Oechel 1991, Belsky 1992, Anderson and Briske 1995).

The coastal wetlands of the northern Gulf of Mexico include marsh

ecosystems in which plant species diversity varies widely with variations in standing

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 biomass, disturbance regime, and abiotic stress factors such as salinity and flooding

(Chabreck 1972, Mitsch and Gosselink 1986, Gough et al. 1994, Grace and Pugesek

in press). Multivariate analyses of the factors associated with species richness have

indicated that competitive exclusion acts to limit richness at very high levels of

community biomass, while disturbance, including the grazing effects of vertebrate

herbivores, may play a major role in offsetting competitive effects (Grace and

Pugesek in press). Experimental studies of herbivory in these habitats have

consistently shown selective effects on species composition, but a range of effects on

species richness depending on the influences of additional factors including

community type, intensity of herbivory, and frequency of fire (Taylor et al. 1994,

Taylor and Grace 1995, Ford 1996). Little is known about the effects of fertilization

on community composition and species richness in these systems. Thus, while

multivariate models suggest an important, interactive influence of community

biomass and herbivory on species richness for coastal wetlands, experimental

evaluation of this effect is missing.

The objective of the study presented here was to address the following

questions for two coastal wetland communities:

(1) Does fertilization lead to a reduction in species richness, and if so, how

quickly does this response occur?

(2) How does fertilization influence the structure of the community? Which

species are favored or disfavored?

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 (3) Does the exclusion of herbivores affect species composition and richness?

Does it influence the response of the community to fertilization?

To examine these questions, nutrient additions, alluvial sediment additions, and

herbivore exclosures were applied to two marsh communities in coastal Louisiana

and responses were measured for three growing seasons from 1993 to 1995.

METHODS

Study Area

The study area was located within the Pearl River Wildlife Management Area,

on the boundary between Louisiana and Mississippi (White 1979). The Pearl River

system is composed of three main channels, the East, Middle, and West Pearl Rivers,

which empty into the Gulf of Mexico. Two study sites were chosen within the river

basin. The Sagittaria site was located along the Middle Pearl River in an area

dominated by Sagittaria lancifolia and Spartina patens with a variety of annual

species present. The S. patens site was located along the East Pearl River where it

widens to empty into Lake Borgne which drains into the Gulf of Mexico; this site

was dominated by S. patens and Scirpus americanus (previously known as Scirpus

olneyi), with mostly grasses and sedges making up the rest of the vegetation. There

are several mammalian herbivores present at both sites including wild boar (Sus

scrofa), rabbit (Sylvilagus sp.), muskrat (Ondatra zibethicus) and most abundantly,

nutria (Myocaster coypus). Although the introduced nutria have increased in number

over the past few years at this marsh, they consume vegetation in a manner similar to

the native muskrats.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 Experimental Design

This experiment was part of a larger study which also examined the effects of

salinity and flooding on species composition in these two study sites; those results

are reported elsewhere (see Chapter 5). There were three levels of fertilization

treatment applied to 1 m2 plots. First, the control was a marked, unmanipulated 1 m2

plot. Second, commercial fertilizer was applied to plots to obtain a level of 17 g/m2

available nitrogen as ureaformaldehyde. Twenty-four Osmocote 20-10-5 tree and

shrub planting tablets (Grace-Sierra, Milpitas, CA) were placed in each plot and

inserted approximately 10 cm into the ground in a uniform grid pattern, modified by

rhizomes and roots. These tablets are designed for one-year slow release but due to

saturated soil conditions, fertilizations were performed in August 1993, June 1994,

and April 1995. Third, sediment was applied to plots in August 1993 to mimic the

natural process of nutrient enrichment in coastal wetlands. Plots were covered with 2

cm of sediment collected from adjacent channel edges and bottoms. Numerous

studies have shown that substantial amounts of sediment are deposited in coastal

marshes in conjunction with floods, storms, and hurricanes (e.g. Guntenspergen et al.

1995). Experimental studies indicate that these sediment deposition events act as a

natural form of fertilization (DeLaune et al. 1990).

Fenced exclosures were constructed in June and July of 1993 to prevent

mammalian herbivory by nutria, muskrat, rabbit, and wild boar. Wooden comer

posts and 3 feet wide plastic coated fencing wire were used to construct the

approximately 7 x 7 m exclosures. The wire was sunk approximately 15 cm into the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 soil. All plots were located at least one meter from the fences to avoid edge effects.

Plots within each fenced or unfenced area were either fertilized, had sediment added,

or were unmanipulated. Eight replicate fenced or unfenced areas were located in

each of the two marshes for a total of 96 plots.

Data Collection

The plots were established in June 1993 and sampled in July 1993 for an

initial, pre-treatment census. Subsequently, they were censused in July, October, and

April of 1993, 1994, and through July 1995. This non-destructive sampling involved

recording the species found in each plot, estimating aerial percent cover, counting

number of stems, and measuring the average height of each species. For one species,

S. lancifolia, average leaf width was recorded as well. Vine species present were

only censused by percent cover as it is impossible during the growing season to

determine where the vines are rooted and how long the stems are.

In early August of 1995, destructive harvests were performed by removing all

aboveground biomass (including standing dead and litter) from the center 0.1 m2 of

the plot using a circular plot guide. Samples were brought to the laboratory and

sorted to species. Those species with easily identifiable dead ramets were divided

into live and dead, while litter that could not be identified to species was lumped into

one category. The plants were dried for at least 48 hours at 80° C and then weighed.

For biomass analysis, results were divided by 0.1 m2 to calculate results on a per m2

basis.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Data Analysis

Species richness was analyzed as a split-split plot design; the source table is

presented in Table 4.1. The whole plot was the marsh by fence interaction, since

each treatment was replicated within each marsh by fence combination. The

fertilization treatments were the first split and time was the second. Error terms are

shown in Table 4.1 directly beneath the effects they tested. PROC GLM (SAS

Institute, Cary, NC) with a repeated statement was used to perform the overall

analysis, determine sphericity and generate univariate results for each date of the

census. All effects were evaluated for significance using Type HI sums of squares,

and LSMEANS with Tukey's HSD test were used for pairwise comparisons. Alpha

levels were judged significant at 0.05 unless otherwise specified.

The same analysis was performed on biomass without repeated measures.

The analysis of variance was run on total, live, and dead biomass separately, and

individual species. Normality was evaluated using the Shapiro-Wilks statistic and

homogeneity of variances was determined with residual plots. Biomass data for

individual species was log transformed as necessary to meet model assumptions. To

determine rank abundance of each species at each site, grand means were calculated

for the log of percent cover plus one averaged over all plots and dates. Species were

grouped according to the log value determined using percent cover results from all

censuses combined and known life history characteristics, particularly the frequency

of occurrence of each species.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.1. Source table for repeated measures analysis of species richness over time.

Source df MS F value n value

marsh 1 1805.03 124.38 0.0001

fence 1 0.17 0.01 0.91

marsh*fence 1 1.78 0.12 0.73

rep(marsh*fence) 28 14.51

treatment 2 20.49 3.98 0.02

marsh *treatment 2 2.24 0.43 0.65

fence*treatment 2 15.66 3.04 0.06

marsh*fence*treatment 2 5.04 0.98 0.38

error 55 5.15

time 6 66.89 51.56 0.0001

time* marsh 6 5.36 2.46 0.03

time*fence 6 4.05 1.86 0.09

time*marsh*fence 6 3.76 1.72 0.12

time*rep(m*fence) 168 2.18

time*treatment 12 3.53 2.72 0.002

time*marsh*treatment 12 1.13 0.87 0.58

time*fence*treatment 12 1.73 1.33 0.20

time*marsh*fence*treatment 12 0.73 0.56 0.87

error(time) 330 1.30

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 RESULTS

Species Richness

As expected based on previous studies (Gough et al. 1994), species richness

was significantly higher in the S. lancifolia marsh than theS. patens marsh

throughout the entire experiment (Table 4.1; Figure 4.1). Fencing did not cause a

change in richness as a main effect, nor was there a significant interaction between

marsh type and fencing.

Since the repeated measures analysis met sphericity requirements (Mauchly's

criterion=0.56; p=0.06), the univariate results for effects of time are reported.

Richness peaked in all treatments in July of each year, causing the significant overall

time effect, and the different treatment responses at each date caused the significant

time by treatment interaction (Table 4.1, Figure 4.1). The only higher order

interaction with marsh that was significant was time by marsh, indicating the two

marshes responded differently over time although they responded to the treatments

similarly. In October 1993 and April 1994 the fenced Sagittaria plots had

significantly higher richness than the unfenced; this difference was not evident at any

other time.

In order to properly interpret the significant main effect of treatment (Table

4.1), the marginally significant (p=0.06) fence by treatment interaction was first

examined. The three treatments were not significantly different at any date in the

unfenced plots in either marsh (Figure 4.1b, d). However, in the fenced plots in both

marshes, the treatments diverged with the nitrogen plots having significantly fewer

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 a

•)O £c _oc ®M CL. oO

2 ----- 2 Control -*r Nitrogen -S- Sediment ♦ Control Nitrogen -0- Sediment

Jul 93 Oct93 Apr94 Jul94 Oct94 Apr95 Jul 95 Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Census Date Census Date

eo 4

Control Nitrogen -9- Sediment Control Nitrogen -c- Sediment

Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Jul 53 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Census Date Census Date

Figure 4.1. Species richness of the three treatments at each census date for a) S. patens site, fenced plots; b) S. patens site, unfenced plots; c) Sagittaria site, fenced plots; and d) Sagittaria site, unfenced plots.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 71 species than the control (Figure 4.1a, c). Thus over the length of the study,

treatment remained a significant effect. Richness in July 1995 showed a significant

overall treatment effect as well, with the nitrogen plots containing fewer species than

the sediment or control plots (p<0.01; Figure 4.2a).

Species Abundances and Responses

Log abundance curves are presented separately for each marsh in Figures 4.3

and 4.4 based on percent cover data from all census dates. To aid in discussing

community structure, species were classified as either dominant, codominant,

common subordinate, occasional subordinate, or rare based on knowledge of the

occurrence of each species over space and time. The S. patens marsh was dominated

by a grass, S. patens, and a sedge, S. americanus, with two co-dominants: a broad-

leaved monocot, S. lancifolia, and a grass, Spartina altemiflora (Figure 4.3; species

listed in Table 4.2). The common subordinates were annuals with the exception of

Spartina cynusoroides, Scirpus validus, and Aster tenuifolius-, this group of species

was distributed throughout the site. The occasional subordinates were found less

frequently while the rare species were highly infrequent. S. americanus, Aster

subulatus, and A. tenuifolius responded to herbivore exclusion by increasing percent

cover in 1994, but this effect did not continue. Other individual species results will

be discussed in terms of biomass accumulation.

The Sagittaria marsh dominants included two vine species ( Mikania scandens

and Vigna luteola), a broad-leaved monocot (S. lancifolia) and a clonal grass (S.

patens) (Figure 4.4). The common subordinates were seen throughout the marsh or

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72:

C o n tro l Nitrogen Sediment T re a tm e n t

— 1200

9 800

C o n tro l Nitrogen Sediment T re a tm e n t

Figure 4.2. July 1995 results for the fertilization treatments OT,a> and b) biomass. Different letters indicate means are significantly different (p 0. ).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. site. f r r-+ t - t S. S. patens -T—T- * , 5

5 • i5 5 555555 ------A ------4 ----- • • » i 3 Species —•—'I

3 3

3 -I -I 1 1 1 1 1 1 1 1 1- Figure 4.3. Log abundance curve for the species occurring at the Species Species abbreviations as in Table 4.2. Numbered categories as follows: 1: rare. dominants; 2; co-dominants, 3: common subordinates; 4: occasional subordinates; 5: - -I— h 1 SPP SCA SAL SPA ASS AST CYO UC VIL CYF SPC SCV ELF JUR ECW ELP SCR AMA IRV OIS HYC EUS KOV PLP PTC HYS IPS - - 0 1 2 ,6 .4 .2 .8 1.4 0 0 1.6 con O u § 0. (0 Qi O c < .Q ~o

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. site. Sagittaria S pecies m Figure 4.4. Log abundance curve for the species occurrring at the Species abbreviations as in Table 4.2. Numbered categories as follows: 1: dominants; 2: common subordinates; 3: occasionals; 4: rare. SAL SAL VII MIS SPP IPS POP PHN ASSSAS AST PAV CYH PTCGAT HYA HYB IRV AGM ELC KOV PIV LYL HYC ALP BAM m 0.2 ^ ^ 0.4 0.8 O) O) q 0.6 § § "O "O

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 Table 4.2. Species abbreviations as used in Figures 4.3 and 4.4.

AGM Agalinis purpurea ALP Altemanthera philoxeroides AMA Amaranthus australis ASS Aster subulatus AST Aster tenuifolius BAH Baccharis halimifolia BAM Bacopa monieri BIS Bidens sp. CYF Cyperus felicinus CYH Cyperus haspans CYO Cyperus oderatus DIS Distichlis spicata DIV Diodia virginica ECC Echinocloa crusgalli ELC Eleocharis cellulosa ELF Eleocharis fallax ELP Eleocharis parvula EUS Eupatorium sp. GAT Galium tinctorum HYA Hyptis alata HYB Hydrocotyle bonariensis HYC Hymenocallis caroliniana HYS Hydrocotyle sp. IPS Ipomoea sagittata IRV Iris virginica JUR Juncus romerianus KOV Kosteletzyfca virginica LIC Liliopsis chinensis LUS Ludwigia sp. LYL Lyihrum lineare MIS Mikania scandens PAV Panicum virgatum PHN Phyla nodiflora PLP Pluchea purpurea POP Polygonum punctatum PTC Ptilmnium capillaceum RAS Ranunculus sp. SAL Sagittaria lancifolia SAS Sacciolepis striata SCA Scirpus americanus SCR Scirpus robustus (table con'd.)

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 scv Scirpus validus SEG Setaria glauca SIS Sium suave SPA Spartina altemiflora SPC Spartina cynusuroides SPP Spartina patens VIL Vigna luteola

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 were abundant only in certain seasons (e.g. spring for Galium tinctorum and

Ptilmnium capillaceum). The occasional were also annuals except for Panicum

virgatum, and the rare species were widely interspersed, some occurring in only one

or two plots at the study site. Mikania scandens, I. sagittata, A. subulatus and P.

capillaceum all reponded initially to herbivore exclusion while S. lancifolia, I.

sagittata and Polygonum punctatum responded to fertilization in the first year; these

effects did not persist except for S. lancifolia and P. punctatum.

Biomass

Since there were several marginally significant interactions in the total

biomass analysis (Table 4.3), results are shown for the three way interaction broken

down into species group and litter (which includes standing dead and unidentifiable

plant parts) (Figure 4.5). The pairwise comparisons in the Sagittaria marsh were

not significant (Figure 4.5b). The fenced, S. patens plots that received nitrogen

showed significantly greater biomass accumulation than either the control or sediment

addition plots (both p<0.05) but this was not true in the unfenced, S. patens plots,

contributing to the three way interaction (Figure 4.5a). The fenced sediment plots in

the 5. patens marsh was also different from the unfenced nitrogen plots (p<0.05), but

other pairwise comparisons were not different.

The marsh by treatment interaction is clearly a result of the significant results

in the S. patens marsh compared with no treatment effect in the Sagittaria marsh.

The marsh by fence interaction depicts an opposite effect of fencing in the two

marshes: in theS. patens marsh the fenced plots contained less biomass than the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 Table 4.3. Source table for analysis of total, live, and dead biomass from August 1995 harvest. p-value Source df Total Live Dead

marsh 1 0.94 0.43 0.07

fence 1 0.85 0.29 0.01

marsh*fence 1 0.06 0.12 0.47

rep(marsh*fence) 28

treatment 2 0.0001 0.004 0.0001

marsh *treatment 2 0.07 0.03 0.52

fence*treatment 2 0.72 0.44 0.60

marsh*fence*treatment 2 0.08 0.13 0.40

error 54

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. -j \o 2 Treatment I } dominants QU] commondominants } QU] Qoccasional rareI f lu tte r site. Treatment FC UC FN UN FS US • 500 1000 Sagittaria 1500 2000 ! ! 1 site and b) S. S. patens Treatment |dominants occasional ([JJJco-domlnantsQ I common Blitter FC UC FN UN FS US in in Figures 4.3 and 4.4 for a) Figure Figure 4.5. Total biomass of plots harvested in July divided 1995 into categories as abbreviations are; FC: fenced, control; UC: unfenced, control; FN: UN: fenced, nitrogen; unfenced, nitrogen; FS: fenced, sediment; US: unfenced, sediment. Different letters indicate mean differences among total biomass for treatments within each marsh (p<0.05). 500 1000 1500 2000 o5 1

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission 80 unfenced while the opposite was true for the Sagittaria site. These interactions do

not affect die conclusion that overall, treatment was significant (Figure 4.2b, Table

4.3), with the sediment and control plots not different from each other, but containing

significantly less biomass than the nitrogen plots.

Examining live biomass separately, a similar pattern emerges. The marsh by

treatment interaction was significant, caused by greater differences among treatments

in the S. patens plots than in the Sagittaria plots (pairwise p<0.01; Table 4.3; Figure

4.5). The fenced sediment plots in the S. patens marsh were significantly different

from the fenced nitrogen plot (p=0.01). Overall the treatment effect was highly

significant for live biomass (Table 4.3) due to the dramatic increase in biomass in the

nitrogen plots in the S. patens marsh.

When the live biomass is divided into groups determined by the log

abundance curves, a more detailed picture emerges. Focusing first on the S. patens

marsh (Figure 4.5a), the dominants appeared to respond to fertilizer and also had a

positive response to being in unfenced plots. The co-dominants as well responded

positively to fertilizer, especially in the unfenced plots. The same pattern is seen for

the common species, but the occasional species accumulated biomass only in the

fenced nitrogen plots and the unfenced sediment plots. There was significantly more

litter in the fences than outside, and significantly more in the nitrogen treatment than

the other two treatments (p<0.01). This pattern is most evident when examining the

fenced nitrogen plots and unfenced nitrogen plots in relation to the two control plots.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 There was much less of a fertilization response at harvest in the Sagittaria

marsh. When divided into groups the live biomass did not differ among treatments

(Figure 4.5b). However, litter did differ, with the same pattern seen as in the S.

patens marsh (Table 4.3) of increased litter inside the fences and increased litter in

the fertilized plots relative to the controls.

Only a few species that demonstrated individual biomass differences due to

fertilization treatments occurred in enough plots be tested statistically. The only

species that occurred solely in the Sagittaria marsh that showed treatment effects was

P. punctatum, a common subordinate, with significantly greater biomass in the

nitrogen treatments than in the control or sediment plots (a three-fold increase;

p<0.05). A clonal dominant found only in the S. patens marsh, S. americanus was

divided into live and dead ramets for biomass analysis but results are reported for

total since there was no difference among the three categories. A marginal effect of

fencing was seen with slightly more biomass inside the fences than outside, due

primarily to more dead biomass inside the fences than out (p=0.08; Figure 4.6).

There was also more biomass in the nitrogen plots than in the other two treatments

causing a highly significant treatment effect (p=0.008), which appears to be due to

the dramatic increase in the fenced, nitrogen plots.

Two species occur at both sites in suitable abundances for detailed analysis:

S. patens and S. lancifolia. Sagittaria lancifolia is a dominant at the fresher site and

a co-dominant at the more brackish site (Figures 4.3, 4.4). It had greater biomass in

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82

_ 6 0 0 CM O)1 500 % 400 (0

FC UC FN UN FS US Treatment

Figure 4.6. Scirpus americanus biomass from the S. patens site. Treatment abbreviations as in Figure 4.5. Different letters indicate mean differences (p<0.05).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. FC UC FN UN FS US Treatment

600

£ 300

o 200

FC UC FN UN FS US Treatment

Figure 4.7. Sagittaria lancifolia biomass for a) S. patens site and the b) Sagittaria site. Treatment abbreviations as in Figure 4.5. Means were not significantly different.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 the Sagittaria marsh (p<0.01; Figure 4.7), but there was no significant difference

between fertilized and control plots in either marsh.

Spartina patens was a dominant in both marshes. The results for this species

are complex: the three way interaction was significant (p<0.05) although individual

pairwise comparisons within each marsh were not different (Figure 4.8). This

interaction can be better understood by examining the significant marsh by treatment

and marsh by fence interactions (both p<0.05) and the graph of the means (Figure

4.8). In the & patens marsh, fencing decreased biomass relative to the unfenced

plots; fertilization had a slight effect on the unfenced plots, but there was no

significant difference between the three treatments overall. In the Sagittaria marsh a

different pattern emerges. Fencing had a minimal effect on biomass; fertilization

caused a decrease in biomass relative to the controls, although not significantly. The

main effect of treatment was not significant, and was influenced by these complex

interactions.

DISCUSSION

This study was conducted to examine the effects of fertilization treatments on

the species composition of plots located in two marshes differing in salinity regimes,

and to determine how fertilization interacts with herbivore effects. Species richness

differed between these two marshes (mean was 6 species/m2 for the brackish marsh

and 9 for the fresher marsh in July of each year of the study), most likely due to

salinity constraints on germination and survival. This decrease in species number

along a salinity gradient has been documented in other studies at the Pearl River

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Treatment

b 800 700 ^ 6 0 0 04 s f.500O) ^ 4 0 0 as I 300 00 200 100 0 m m | v m m | V rn m t W m h | v m m x m m, FC UC FN UN FS US . Treatment

Figure 4.8. Spartina patens biomass at harvest for a) S. patens site and b) Sagittaria site. Treatment abbreviations are as in Figure 4.5. Means were not significantly different, however main effects were significant (see text).

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86

(White 1979, Gough et al. 1994) and throughout coastal Louisiana (Chabreck 1972).

Species richness also changed with the seasons, causing a significant effect of time,

as spring annuals increased richness in April and July and senescence decreased

richness in October (Figure 4.1). Both marshes behaved in similar ways in response

to the treatments, despite the average richness differences.

Previous fertilization studies in marshes have found that wetlands such as the

one studied here have limited nitrogen availability as evidenced by increased biomass

and productivity under enriched nutrient conditions (Mendelssohn 1979, Neely and

Davis 1985, Dougherty et al. 1990, Jordan et al. 1990). These results have been

produced by commercial fertilizer addition (e.g. Neill 1990a), sewage sludge (Payonk

1975) and sediment addition (DeLaune et al. 1990). A Canadian prairie marsh

showed a significant decrease in richness associated with increased biomass after two

years of nitrogen and phosphorus additions (Neill 1990b and c). In a previous study

conducted at the Pearl River, biomass was found to be significantly greater in

fertilized plots but richness did not significantly change during an 18-month period

(Gough and Grace, submitted; Chapter 2). When experimental nutrient additions

have been performed, species may drop out at high levels of fertilizer but this effect

can be offset by other environmental variables, particularly salinity and flooding (e.g.

Neill 1990a). In this study, fertilization alone caused an increase in total biomass in

the S. patens marsh but no overall change in the Sagittaria marsh at harvest (Table

4.3). Both marshes showed increased litter and live biomass accumulation with

nitrogen fertilization but no richness change in unfenced plots.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 In almost all cases the sediment treated plots behaved the same as the control

plots in terms of richness and biomass (with exceptions noted). A previous sediment

addition study in a Louisiana salt marsh used much greater amounts of sediment and

found an increase in S. altemiflora biomass and leaf area (DeLaune et al. 1990,

Pezeshki et al. 1992). In those studies up to 94 kg of sediment was used per m2 in

order to see definitive results.

Herbivory is a biotic factor known to influence plant community structure in

marshes (Mitsch and Gosselink 1986). Herbivores can function as disturbance agents

causing a similar unimodal species richness pattern based on disturbance level as that

seen along productivity gradients (Connell 1978, Grace and Pugesek in press).

Herbivores such as the nutria found in Louisiana marshes usually consume one or

two plant species while leaving others undisturbed, causing shifts in plant community

structure (Llewellyn and Shaffer 1993, Ford 1996).

In this study, herbivory alone as a disturbance agent had more of an effect on

dead biomass accumulation than on live or total and had no effect on richness in the

absence of fertilization. In both marshes more litter accumulated inside the fences

than outside. Live biomass differed significantly between fenced and unfenced plots

in the S. patens marsh but not in the Sagittaria marsh, likely due to greater herbivore

activity at the brackish marsh. Increased litter as a result of herbivore exclusion did

not appear to affect species numbers in this case (see also Taylor and Grace 1995).

Bazely and Jefferies (1986) found an increase in biomass, litter and species richness

inside snow geese exclosures in a Canadian salt marsh. A major difference between

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 their study and ours, however, is that snow geese tend to create complete eat outs

while in our case, herbivory effects on community biomass were less severe.

Fertilization of plots protected from herbivory caused richness to decrease in

both marshes but did not affect total biomass in relation to the unfenced, fertilized

plots. In Figure 4.5 it is clear that the total biomass did not differ between fenced,

nitrogen plots and unfenced, nitrogen plots in either marsh and only in the S. patens

marsh did the fenced, nitrogen plots differ from the unfenced, controls. Apparently

the fertilizer treatment did not attract herbivores at greater numbers as has been

demonstrated by other authors (e.g. Nams et al. 1996) and predicted in herbivore

optimization models (Oksanen et al. 1981). Since these marshes are productive

systems without nutrient enrichment, our results may support the findings of van de

Koppel et al. (1996) that herbivory is greatest at intermediate rather than high

productivities. However, there was greater litter accumulation inside the fences and

greater live biomass accumulation outside the fences in the nitrogen plots which

resulted in equal totals. When examining the number of species lost in the fenced,

nitrogen plots in each marsh, the effect in the Sagittaria marsh appears to be more

dramatic, perhaps because there were more annuals present in the fresher marsh. As

a proportion however, species lost due to fertilization were 23% in S. patens and

21% in Sagittaria at the conclusion of the study.

The loss of richness due to fertilization and herbivore exclusion may be

driven by slightly different forces in the two marshes. Both marshes exhibited

greater litter accumulation in the fenced, nitrogen plots which could have suppressed

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 germination or survival of other species. This litter may have contributed to the loss

of species in the nitrogen plots by further blocking light to the substrate or otherwise

interfering with plant growth as has been found in other studies (Neill 1990a, Day et

al. 1988). The increased dead biomass may have been a result of faster tissue

turnover in the nitrogen plots. In the Sagittaria marsh, the species that dropped out

of the fenced, nitrogen plots (Table 4.4), were mostly common subordinates and

occasionals that are annuals. In the S. patens marsh, S. altemiflora was lost from

five out of eight fenced, nitrogen plots; the others that dropped out were mostly

annuals (Table 4.4). S. lancifolia showed a minimal fertilization response at harvest

although it responded to fertilizer by emerging earlier in the growing season and

senescing later than unfertilized plants (L. Gough, unpublished data).

In the S. patens marsh a different phenomenon appears to be occurring. S.

americanus showed a two-fold increase in biomass in the fenced, nitrogen plots

(Figure 4.6). However, in Figure 4.5a, the dominants show greater biomass in the

unfenced, nitrogen plots, due to the increase of S. patens in the unfenced, fertilized

plots. These two species responded differently, and in combination with more litter

accumulating in the fences, there resulted no difference between total biomass in

fenced and unfenced nitrogen plots. This result suggests a complex interaction

between species abundance and herbivory in high fertility plots and supports the

ideas of Grover (1995) who proposed a model for edible and inedible plants in the

presence of herbivores which shows that when nutrients are added to plants, only the

inedible biomass will increase in the presence of herbivores, as S. patens did here.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 Table 4.4. Species lost from fenced, nitrogen plots in both marshes. Numbers indicate number of plots out of eight from which each species was lost

MARSH TYPE S. vatens Si Aster subulatus 2 5 A. tenuifolius 1 2 Cyperus haspans 0 2 C. oderatus 2 0 Eleocharis cellulosa 0 1 E. fallax 1 0 E. parvula 1 0 Galium tinctorum 0 5 Hydrocotyle honariensis 0 4 Ipomoea sagittata 0 3 Juncus romerianus 1 0 Lilaeopsis chinense 1 0 Lythrum lineare 0 1 Mikania scandens 0 2 Phyla nodiflora 0 2 Polygonum punctatum 0 1 Ptilimnium capillaceum 0 5 Sacciolepis striata 0 1 Sagittaria lancifolia 1 0 Scirpus americanus 1 0 S. robustus 1 0 S. validus 1 0 Spartina altemiflora 5 0 Vigna luteola 0 1

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 S. americanus is the preferred food species of nutria and muskrat, while S. patens

appears largely to escape herbivoiy. Thus, it appears that S. patens may have been

competitively suppressed by S. americanus inside the fences under fertilized

conditions.

When examining the individual species' responses in the Sagittaria marsh, a

slightly different story emerges. More S. lancifolia was present at this site with an

increase in fertilized plots in both fenced and unfenced areas. S. patens showed a

decrease in the fenced, nitrogen plots as compared to the unfenced and the control

plots. Again, S. patens is not a preferred forage species so is able to survive well in

the unfenced areas. The reason for the decrease inside the fences may be due to

competitive suppression by the increased biomass of S. lancifolia ramets or may be

related to the increased litter. S. patens has been eliminated by other species under

enriched nutrient conditions in a New England salt marsh (M.D. Bertness, pers.

comm.). In both marshes S. patens documented a decrease in biomass under fenced,

fertilized conditions, indicating it can be outcompeted here as well. The results for

S. patens indicate that this species which is widespread in Louisiana marshes, is not a

good competitor under increased nutrient regimes when herbivory does not suppress

other codominants such as S. americanus. A competition study performed with these

two species indicates S. patens is the better competitor in saline conditions while S.

americanus will dominate with increased flooding depths (Broome et al. 1995). Our

results suggest another set of factors which cause the relative dominance of these two

species to shift.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 Despite differences between individual species responses in fenced and

unfenced plots and plots in the two marshes, the treatment effects were significant

overall, with fertilization causing a decrease in species richness and an increase in

biomass. However, the dramatic biomass increase occurred only in the S. patens

marsh. Litter accumulation was the same in both marshes and likely contributed to

the significant loss of species in the fenced, nitrogen plots.

The results presented here confirm current models that as nutrient levels are

increased, species richness decreases. However, this only occurred when herbivores

were excluded from plots. Somewhat surprisingly, results were the same in both

marsh communities and occurred relatively quickly. Since more litter accumulated

inside the exclosures, it is difficult to separate the effects of litter from the

competitive effects of living tissue. However, this indicates that the interaction of

herbivory and fertilization is more complex than originally suspected; herbivory via

plant consumption or litter removal offset competitive exclusion in the unfenced,

fertilized plots.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 5

THE INTERACTION OF HERBIVORY WITH FLOODING AND SALINITY STRESS IN A COASTAL MARSH: IMPLICATIONS FOR RELATIVE SEA LEVEL RISE

INTRODUCTION

Relative sea level rise is occurring in Gulf Coast wetlands at a rate

approximately ten times higher than that of eustatic or worldwide sea level rise

(Gomitz 1995). This is occurring in most marshes because the compaction and

subsidence of recent alluvial sediments coincides with the prevention of new

sediment entering the marshes due to levees and control structures. This increase in

relative sea level is believed to be causing wetlands to disappear at an alarming rate

in the Mississippi delta plain of Louisiana. Other factors, including the construction

of oil and gas canals, and the control of water flow into the Mississippi and

Atchafalaya Rivers contribute to conversion of wetlands to open water and of fresh

marshes to brackish or saline marshes (Turner and Cahoon 1988, Boesch et al. 1994).

If vegetation is subjected to increasing flooding and salinity stress, plant growth may

be inhibited and plant mortality may occur (McKee and Mendelssohn 1989), thus

depriving marshes of root zones necessary to stabilize and accumulate sediment.

There has been a great deal of research focused on the specific effects of

salinity stress and flooding stress, individually or in combination, on individual

coastal plant species (Mendelssohn et al. 1981, Feijtel et al. 1989, Hellings and

Gallagher 1992, Naidoo and Mundree 1993, Bandyopadhyay et al. 1993, Rahman and

Ungar 1994, Glenn et al. 1995, Howard and Mendelssohn 1995, Allen et al. 1996,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 and others). Occasionally the response of two species is examined (e.g. Naidoo et al.

1992, Lenssen et al. 199S), and sometimes the treatments are applied to two species

in mixture (e.g. Broome et al. 1995). In even fewer studies the response to salinity

and flooding of marsh plants from natural sods has been measured to better

understand how communities may respond to these factors (Mendelssohn and McKee

1987, McKee and Mendelssohn 1989, Flynn et al. 1995).

McKee and Mendelssohn (1989) studied the effect of increasing salinity and

water depth on sods from a fresh marsh community both in the field and in the

greenhouse. This type of study is necessary to understand how salt water intrusion

may affect fresh plant communities, particularly whether or not these communities

will be able to shift to more salt tolerant and/or flood tolerant species or will simply

die out leading to increased erosion.

McKee and Mendelsshon (1989) found that increased flooding by 10 cm

reduced biomass of one species, Panicum hemitomum, but did not affect Leersia

oryzoides or Sagittaria lancifolia. When sods were transplanted to a marsh with

higher salinities, the plants were killed by an extreme salinity event but colonization

occurred, resulting in a more salt-tolerant community in the sods. Following this

study, Flynn et al. (1995) demonstrated that the recovery of freshwater marsh sods

after a high salinity event is dependent on the duration of flooding and salinity

following the event and showed that individual species responded differently to the

abiotic factors.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 Another factor involved with plant community survival and composition in

Gulf coast marshes is herbivory. Vertebrate herbivores have been shown to have

dramatic effects on vegetation composition in many marsh communities (Weller

1981, Smith and Kadlec 1985, Bazely and Jefferies 1986, Mitsch and Gosselink

1986). In Louisiana, nutria have been documented as having a significant impact on

wetland plants, and their population continues to increase, thus increasing the

importance of herbivory in structuring plant communities (Chabreck 1988, Wilsey et

al. 1991, Taylor 1992, Llewellyn and Shaffer 1993, Nyman et al. 1993, Taylor et al.

1994). By damaging plants and perhaps making them more susceptible to flooding

and/or salinity stress, herbivores may contribute to marsh degradation and loss. With

the exception of a study examining the effect of salinity and flooding and simulated

herbivory on a coastal marsh dominant (Grace and Ford 1996), there have been no

studies of the interaction of herbivory with these other processes in the coastal

marshes of Louisiana. However in a greenhouse study where marsh sods were

subjected to flooding and salt stress following clipping (simulated

disturbance/herbivory), most species survived unless they were clipped as well as

subjected to stress (Baldwin 1996).

The objectives of this study were to address the following questions in two

coastal plant communities:

(1) Does increased flooding stress reduce species richness? Does decreased

flooding stress increase species richness?

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 96 (2) Does increased salinity stress reduce species richness? Does decreased

salinity stress increase species richness?

(3) How do flooding stress and salinity levels affect community structure?

(4) Does the exclusion of herbivores affect species composition and richness?

Does it influence the response to salinity regime changes or water depth

changes?

To examine these questions, sods were raised, lowered, transplanted to sites of

different salinities and protected from herbivory in two marsh communities in coastal

Louisiana and responses were measured for three growing seasons from 1993 to

1995.

METHODS

Study Site

The study area was located within the Pearl River Wildlife Management Area,

on the boundary between Louisiana and Mississippi (White 1979). The Pearl River

system is composed of three main channels, the East, Middle, and West Pearl Rivers.

Two study sites were chosen within the river basin. The Sagittaria site was located

along the Middle Pearl River in an oligohaline marsh (water salinities ranged from 0

to 2 ppt) dominated by Sagittaria lancifolia and Spartina patens with a variety of

annual species present. The S. patens site was located along the East Pearl River

where it widens to empty into Lake Borgne and the Gulf of Mexico (water salinity

range: 6 to 14 ppt); this site was dominated by S. patens and Scirpus americanus,

with graminoids making up the rest of the vegetation. There are several mammalian

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 herbivores present at both sites including wild boar (Sus scrofa), rabbit (Sylvilagus

sp.), muskrat {Ondatra zibethicus) and most abundantly, nutria {Myocaster coypus).

Although the introduced nutria have increased in number over the past few years at

this marsh, they consume vegetation in a manner similar to the native muskrats.

Experimental Design

Flooding

The flooding treatments were as follows: raised, lowered, and control. For

the raised treatments, circular sods (0.33 m diameter, 15 cm deep) were cut out of

the marsh. Sods were placed in black plastic weed cloth and tied with string.

Sediment was used to fill in the hole left by the sod and the sods were then placed

on top of the sediment so that the soil surface of the sod was approximately 10 cm

above the marsh surface. Bamboo stakes were used to support the sods at this height

and to hold sods in place. The lowered sods were dug out and placed in weed cloth

in the same manner as the raised sods. Sediment was then removed from the hole

until the sod could be replaced with its surface 10 cm below the marsh surface.

Control sods were dug up, placed in weed cloth, and returned to the hole from which

they came. Undisturbed control plots of 0.33 m diameter were marked and regarded

as a separate treatment throughout the study. These treatments were carried out in

June 1993.

Salinity

In order to assess the effects of either increasing or decreasing ambient

salinity, reciprocal transplants of sods between marshes of contrasting salinity were

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 performed. These sods were the same size as those for the flooding experiment

Control and undisturbed control sods were the same as for the flooding experiment.

Sods to be transplanted were dug up and placed in weed cloth in the same manner.

An airboat was used to transport the sods to the other marsh site and sods were

replaced randomly in holes left by the other sod removals. Transplants and controls

were established in June 1993.

Herbivory

Fenced exclosures were constructed following flooding and salinity treatments

in late June 1993 to prevent mammalian herbivory by nutria, muskrat, rabbit, and

wild boar. Wooden comer posts and 3-feet-wide plastic coated fencing wire were

used to construct the approximately 7 x 7 m exclosures. The wire was sunk

approximately 15 cm into the soil to inhibit animals from burrowing under the fence.

All plots within the exclosures were located at least one meter from the fences to

avoid edge effects.

Eight replicate fenced and unfenced treatment plots were located at each

marsh site. For the flooding experiment this resulted in a total of 108 sods; for the

salinity experiment this resulted in 96 sods.

Data Collection

The plots were sampled in July, October, and April of 1993, 1994, and

through July 1995. Non-destructive sampling during the course of the study involved

recording the species found in each plot, estimating aerial percent cover, counting

number of stems, and measuring the average height of each species. For one species,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 S. lancifolia, average leaf width was recorded for more accuarate biomass estimates.

Vine species present were only censused by percent cover as it is impossible during

the growing season to determine where the vines are rooted and how long the stems

are. Water levels in relation to sod height were also recorded for the flooding

treatments and signs of herbivory noted for all unfenced treatments.

In late July 199S elevation of the flooding treatment plots was recorded using

a laser survey system (Spectra-Physics Instruments, Dayton, OH). Salinity was

measured in standing water on the marsh surface using a portable salinity-

conductivity meter (YSI, Yellow Springs, OH).

In early August of 1995, destructive harvests were performed by removing all

aboveground biomass (including standing dead and litter) from the plot. Samples

were brought to the laboratory and sorted to species. Those species with easily

identifiable dead ramets were divided into live and dead, while litter that could not

be identified to species was lumped into one category. The plants were dried for at

least 48 hours at 80° C and then weighed to the nearest hundredth gram.

Data Analysis

The flooding and salinity experiments were analyzed separately following the

same procedures outlined here. Species richness was analyzed as a split-split plot

design; the source tables are presented in Tables 5.2 and 5.5. The whole plot was

the marsh by fence interaction, since each treatment was replicated within each marsh

by fence combination. The fertilization treatments were the first split and time was

the'second. Error terms are shown in Tables 5.2 and 5.5 directly beneath the effects

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 they tested PROC GLM (SAS Institute, Cary, NC) with a repeated statement was

used to perform the overall analysis, determine sphericity and generate univariate

results for each census date. All effects were evaluated for significance using Type

HI sums of squares, and LSMEANS with Tukey's HSD test were used for pairwise

comparisons. Alpha levels were judged significant at 0.0S unless otherwise

specified

The same analysis was performed on biomass without repeated measures.

The analysis of variance was run on total biomass, live and dead biomass separately,

and individual species. Normality was evaluated using the Shapiro-Wilks statistic

and homogeneity of variances was determined with residual plots. Biomass data for

individual species was log transformed as necessary to meet model assumptions. To

determine rank abundance of each species at each site, grand means were calculated

for the log of percent cover plus one averaged over all plots and dates. Species were

grouped according to the log value determined and known life history characteristics.

RESULTS

Species Abundances

Using log abundance curves, species were grouped by percent cover in each

marsh as dominant, codominant, common subordinate, occasional subordinate, and

rare based on occurrence at each site over space and time. In the S. patens marsh

(Figure 5.1; species abbreviations in Table 5.1), the four most abundant species were

divided into dominants and codominants. Both dominants are clonal monocots, S.

americanus (a sedge) and S. patens (a grass). Spartina altemiflora, a grass, and 5.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. © marsh. Group codes S. S. patens S p e c ie s Figure 5.1. Log abundance of species found at the are: l=dominant, 2=codominant, 3=common subordinate, 4=occasional subordinate, and 5=rare. Species abbreviations are as in Table 5.1. _ SCA SPP SAL SPA ASS ELF CYO SCV LIC CYF SPC ELP SCR VIL AST ECC DIS PLP. ECP AMA PTC 0.2 0.6 0 .4 I I

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 102 Table S.l. Species abbreviations used in Figures S.l and 5.2.

AGM Agalinis purpurea ALP AUemanthera philoxeroides AMA Amaranthus australis ASS Aster subulatus AST Aster tenuifolius BIS Bidens sp. CYF Cyperus felicinus CYH Cyperus haspans CYO Cyperus oderatus DIS Distichlis spicata DIV Diodia virginica ECC Echinocloa crusgalli ECP Eclipta prostrata ELC Eleocharis cellulosa ELF Eleocharis fallax ELP Eleocharis parvula GAT Galium tinctorum HYB Hydrocotyle bonariensis IPS Ipomoea sagittata IRV Iris virginica JUR Juncus romerianus LIC Liliopsis chinensis LUS Ludwigia sp. LYL Lythrum lineare MIS Mikania scandens PAV Panicum virgatum PHN Phyla nodiflora PLP Pluchea purpurea POP Polygonum punctatum PTC Ptilmnium capillaceum SAC Sabatia campanularis SAL Sagittaria lancifolia SAS Sacciolepis striata SCA Scirpus americanus SCR Scirpus robustus SCV Scirpus validus SIS Sium suave SPA Spartina altemiflora SPC Spartina cynusuroides SPP Spartina patens VIL Vigna luteola

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 lancifolia, a clonal broad-leaved monocot, were the codominants. Common

subordinates were present in most plots but in lower abundances; in the S. patens

marsh they were mostly annuals. Occasional subordinates included clonal species, a

perennial, an annual, and a vine. The rare species were infrequent annuals.

Two vines, Vigna luteola and Mikania scandens, and S. lancifolia were the

dominants at the Sagittaria site (Figure 5.2). The codominants were a vine species

(Ipomoea sagittata), a grass (S. patens), and an annual dicot ( Phyla nodiflora). The

common subordinates were mostly annuals as were the occasional subordinates,

except for Aster tenuifolius. The rare species were mostly annuals or perennials that

occur in only one or two locations at the site.

Flooding

Although the sods were initially established at 10 cm above or below the

marsh surface, the raised sods settled on averge 3 cm and the lowered sods were

gradually filled in an average of 3 cm by the conclusion of the study (Figure 5.3).

Since results for the undisturbed control plots were not significantly different

than the disturbed control plots, only the results of the disturbed control plots are

reported here. The disturbed control plots will be referred to as "controls"

throughout the remainder of this chapter.

Species Richness

The repeated measures analysis for species richness over the length of the

experiment revealed a consistently higher richness in the Sagittaria marsh (Table 5.2,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104

n marsh. Group codes Sagiitaria

i- 0 ■S C L Figure Figure 5.2. Log abundance of species found at the are. are. I—dominant, 2—codominant, 3—common subordinate,and 5=rare. 4=occasional Species subordinate, abbreviations are as in Table 5.1.

i ! ~T~ i CO CD CM CO CD o aouepunqv Boq

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 105

------iiiimimu------1------1------Lowered Control Raised Treatment I S. patens HI Sagittaria

Figure 5.3. Final elevations after 26 months for flooding experiment.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5.2. Source table for repeated measures analysis of species richness for flooding experiment.

Source df MS F-value D-value

marsh 1 549.37 84.71 0.0001

fence 1 25.09 3.87 0.06

marsh*fence 1 0.001 0 0.99

rep(marsh*fence) 28 6.48

treatment 2 245.89 72.62 0.0001

marsh*treatment 2 33.08 9.77 0.0002

fence*treatment 2 25.65 7.58 0.001

marsh * fence*treatment 2 5.67 1.68 0.20

error 55 3.38

time 6 46.12 32.96 0.0001

time*marsh 6 17.44 11.21 0.0001

time*fence 6 4.20 2.70 0.02

time* marsh* fence 6 0.91 0.58 0.74

time*rep(marsh*fence) 168 1.56

time*treatment 12 3.88 2.77 0.003

time*marsh*treatment 12 1.95 1.40 0.18

time*fence*treatment 12 1.86 1.33 0.22

time*marsh*fence*treatment 12 1.02 0.73 0.70

error 330 1.40

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 Figure S.4). Since the dataset did not meet sphericity requirements for univariate

analysis (Mauchly's criterion=0.528, p=0.03), within subject effects are reported using

Greenhouse-Geiser adjusted univariate p-values (Moser et al. 1990). Time was a

significant main effect, along with the time by marsh, time by fence, and time by

treatment interactions (Table 5.2).

For the significant effect of treatment, the significant two way interactions

must be examined first. The marsh by treatment interaction was due to the lowered

sods containing fewer species than the raised sods in the Sagittaria marsh throughout

the experiment, but only at April and July 1994 and July 1995 in the S. patens marsh

(Figure 5.4). The fence by treatment interaction was significant over the length of

the experiment due to less of a difference in the fenced plots than the unfenced; the

lowered treatments were significantly different from the controls at all dates in the

unfenced plots but only in the fenced plots at the final date (Figure 5.5). The raised

plots were only significantly different from the controls when averaged over all

marsh by fence combinations (Figure 5.6a).

The shift in species composition in the plots that caused these differences was

complex (Table 5.3). Many species were lost from some plots but established in

others within the same treatment. The loss of species in the lowered plots in the

Sagittaria marsh was mostly due to annuals dropping out, particularly P. nodiflora,

Galium tinctorum, and Polygonum punctatum, in addition to two vine species (Table

5.3). Species gained in the raised plots were mostly annuals as well, especially P.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108

a Lowered -®- Control Raised

Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Census Date

b Lowered -®- Control Raised

c 6

Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Census Date

Figure 5.4. Species richness for flooding experiment showing marsh by treatment interaction: a) S. patens', b) Sagittaria.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109

a Lowered Control Raised

« 7

Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Census Date

9 Lowered -®- Control 8 Raised « 7 CO c® 6e x: 5

CO A g> M d)° O *a Q. CO 2 1 0 Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Census Date

Figure 5.5. Species richness for flooding experiment showing fence by treatment interaction: a) fenced plots; b) unfenced plots.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110

a 7

ControlLowered Treatment

Lowered Control Raised Treatment

Figure 5.6. August 1995 treatment results of the flooding experiment: a) species richness; b) biomass. Different letters indicate means are different at p<0.05.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ill Table S.3. Species changes in the flooding experiment. Numbers indicate the number of plots out of eight in which that species was gained or lost For group number see Figures S.l and S.2.

LOWERED RAISED group lost gained lost pained Sagittaria marsh Fenced 1 Mikania scandens 3 3 1 3 1 Vigna luteola 3 3 0 2 2 Ipomoea sagittata 1 0 1 1 2 Phyla nodijlora 3 0 1 1 2 Spartina patens 1 2 0 2 3 Aster subulatus 1 0 0 4 3 Eleocharis cellulosa 3 3 0 1 3 Hydrocotyle bonariensis 2 0 1 2 3 Polygonum punctatum 0 2 0 5 3 Ptilimnium capillaceum 1 0 4 0 4 Aster tenuifolius 0 1 3 0 4 Galium tinctorum 3 0 4 1 5 Agalinis maritima 0 0 0 1 5 Altemanthera philoxeroides0 0 3 0 5 Ludwigia sp. 0 0 I 0 5 Lythrum lineare 0 0 0 1 5 Panicum virgatum 0 0 1 0 5 Pluchea purpurea 1 0 1 0 5 Sacciolepis striata 0 0 0 4

Unfenced 1 Mikania scandens 1 3 0 2 1 Vigna luteola 2 7 3 3 2 Ipomoea sagittata 1 2 3 2 2 Phyla nodiflora 0 3 1 4 2 Spartina patens 1 3 1 3 3 Aster subulatus 2 0 1 5 3 Eleocharis cellulosa 0 3 1 2 3 Hydrocotyle bonariensis 2 0 0 0 3 Polygonum punctatum 3 0 0 5 3 Ptilimnium capillaceum 0 0 2 2 4 Aster tenuifolius 0 1 1 1 4 Diodia virginica 0 0 0 3 4 Galium tinctorum 2 1 2 3

(table con'd.)

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 LOWERED RAISE eroup lost eained lost g; 5 Agalinis maritima 0 0 0 3 5 Altemanthera philoxeroides 0 1 1 0 5 Bidens sp. 0 0 0 1 5 Cyperus haspans 1 0 2 1 5 Sabatia campanularis 0 0 1 0 5 Sacciolepis striata 1 0 0 1

S. patens marsh Fenced 1 Scirpus americanus I 0 1 0 2 Sagittaria lancifolia 4 0 3 0 2 Spartina altemiflora 3 1 1 1 3 Aster subulatus 1 0 0 5 3 Cyperus oderatus 0 0 0 1 3 Eleocharis fallax 1 0 0 1 3 Liliopsis chinensis 0 0 2 0 3 Scirpus validus 0 1 3 1 4 Aster tenuifolius 1 0 0 1 4 Eleocharis parvula 0 0 2 0 4 Scirpus robustus 0 1 1 0 4 Spartina cynusoroides 0 1 0 1 4 Vigna luteola 0 0 0 1 5 Cyperus haspans 0 1 0 0 5 Distichlis spicata 1 0 0 0

Unfenced 1 Scirpus americanus 2 0 0 1 1 Spartina patens 2 1 0 0 2 Sagittaria lancifolia 2 2 1 0 2 Spartina altemiflora 0 1 2 0 3 Aster subulatus 1 0 0 5 3 Cyperus oderatus 0 0 0 1 3 Eleocharis fallax 1 0 1 1 3 Liliopsis chinensis 0 0 1 0 3 Scirpus validus 1 1 1 1 4 Aster tenuifolius 0 0 0 1 4 Eleocharis parvula 1 0 0 0 4 Spartina cynusoroides 0 0 0 1 5 Cyperus haspans 0 0 0 1 5 Echinocloa crusgalli 0 1 0 1 5 Vigna luteola 0 0 1 0

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 punctatum, P. nodiflora, and Aster subulatus. A. subulatus increased in the raised S.

patens sods while S. altemiflora and S. lancifolia dropped out of the lowered plots.

Biomass

The overall biomass results are presented in Figure 5.7. The analysis of total

biomass revealed a significant main effect of treatment (Table 5.4; Figure 5.6b).

Across both marshes and fence treatments, the raised plots had significantly more

biomass than the controls, and the controls had significantly more biomass than the

lowered sods; these differences were more pronounced in the Sagittaria marsh, but

not significantly so. When total biomass is broken down into live and dead, live

biomass also exhibits a significant treatment effect, and also has significant fence by

treatment and marsh by treatment interactions (Table 5.4). These interactions were

caused by the differences in the Sagittaria marsh among treatments. Dead biomass

(which includes standing dead and litter) was significantly different for many of the

main effects (Table 5.4). The significant three way interaction was caused by the

greater accumulation of litter inside the fences in some plots but outside in others,

and again by the response of the sods in the Sagittaria marsh (Figure 5.7). The

nonsignificant marsh effect for total biomass was caused by greater live biomass in

the S. patens marsh and greater amounts of dead biomass in the Sagittaria marsh.

Several species were examined individually for biomass accumulation in the

different treatments. S. americanus occured solely in the S. patens marsh where it

was a dominant species. The analysis of the log-transformed data resulted in a

significant fence effect (p=0.02) with greater biomass present inside the fences

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114

100

m 20

FC UC Treatment I Live HDead

m 20

FC UC Treatment I Live fflDead

Figure 5.7. Biomass results for flooding experiment: a) S. patens marsh; b) Sagittaria marsh. Treatment abbreviations are: FL=fenced, lowered; UL=unfenced, lowered; FC=fenced, control; UC=unfenced, control; FR=fenced, raised; UR=unfenced, raised. Different letters indicate means are significantly different at p<0.05. See text for explanation of significant main effects.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 115 Table 5.4. Source table for biomass analysis of flooding experiment.

p-value Source df Total Live Dead

marsh 1 0.73 0.08 0.02

fence 1 0.30 0.67 0.11

marsh*fence 1 0.61 0.51 0.97

rep(marsh*fence) 28

treatment 2 0.0001 0.0001 0.0001

marsh*treatment 2 0.24 0.74 0.004

fence*treatment 2 0.33 0.49 0.03

marsh * fence* treatment 2 0.29 0.78 0.01

error 56

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 116 (Figure 5.8). When biomass is divided into live and dead it can be seen that

significant effects resulted from greater accumulation of dead biomass inside the

fences; the live biomass alone was not significantly different across treatments.

S. patens was dominant or codominant in both marshes. The main effect of

treatment was significant (p=0.0008) as was the main effect of marsh (p=0.06).

There was more S. patens biomass present in the S. patens marsh, and more present

in the raised sods than the control sods (p=0.04); the control and lowered sods were

not different (p>0.10). There was no effect of fencing on biomass accumulation of

this species. The individual means are presented in Figure 5.9 and should be

examined with the above listed significant main effects in mind.

S. lancifolia was a dominant in both marshes. Analysis of total biomass

revealed an overall marsh effect (p=0.0001) and an overall treatment effect (p=0.03).

The overall treatment effect was a result of the raised sods having significantly

greater biomass than the lowered sods. Individual treatment means are presented in

Figure 5.10. When total biomass was divided into live and dead, the analysis results

produced the same pattern, although dead biomass showed a pairwise difference

between fenced, lowered and fenced, raised in the Sagittaria marsh (p=0.05).

Salinity

Species Richness

Many of the main effects of the repeated measures analysis on species

richness were significant (Table 5.5). The data maintained sphericity (Mauchly's

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 117

c?20 E 5 15 3 (0 8 1 0 E o ta '5

0 FL UL FC UC Treatment

Figure 5.8. Biomass results for S. americanus in flooding experiment. Treatment codes as in Figure 5.7. Different letters indicate means of total biomass are different at p<0.05.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. FL UL FC UC FR UR Treatment

b 40 35

w 30

5 2 5 - 2 0 CO CO g 15 5 10 5 0 FL UL FC UC FR UR Treatment

Figure 5.9. Biomass results for S. patens in flooding experiment for a) S. patens marsh and b) Sagittaria marsh. Treatment codes as in Figure 5.7. None of the means were significantly different within each marsh as indicated by the same letters.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 119

a 30

_ 2 5 CNI

Treatment

Hive I dead

FL UL FC UC FR UR Treatment

I live ■ dead

Figure 5.10. Biomass results for S. lancifolia in flooding experiment for a) S. patens marsh and b) Sagittaria marsh. Treatment codes as in Figure 5.7. None of the means were significantly different within each marsh as indicated by the same letters.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 120 Table 5.5. Source table for repeated measures analysis of species richness for salinity experiment.

Source df MS F-value D-value

marsh 1 239.12 77.77 0.0001

fence 1 99.12 32.24 0.0001

marsh*fence 1 68.75 22.36 0.0001

rep(marsh*fence) 28 3.07

treatment 1 477.37 85.68 0.0001

marsh*treatment 1 3.19 0.57 0.46

fence*treatment 1 81.87 14.69 0.0007

marsh*fence*treatment 1 113.46 20.36 0.0001

error 26 5.57

time 6 18.48 17.86 0.0001

time*marsh 6 4.52 3.74 0.002

time*fence 6 1.02 0.84 0.54

time*marsh*fence 6 1.14 0.94 0.47

time*rep(marsh*fence) 168 1.21

time*treatment 6 4.38 4.23 0.0006

time*marsh*treatment 6 8.12 7.84 0.0001

time*fence*treatment 6 0.63 0.61 0.73

time*marsh*fence*treatment 6 1.53 1.48 0.19

error(time) 156 1.04

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 121

criterion=0.472, p=0.060) so the univariate results for the within subject factors (time

and its interactions) are reported in Table S.5. Examining the main plots first, the

marsh by fence interaction was significant due to the similarity between the fenced

and unfenced plots in the Sagittaria marsh and the dramatic differences in the S.

patens marsh. The sods transplanted to the unfenced areas in the Sagittaria marsh

from the S. patens marsh were consumed and destroyed by animals at the beginning

of the experiment. This result caused the significant marsh and fence effects as well.

Time caused significant changes in richness in its interaction with marsh and

treatment, marsh, treatment, and alone.

The three way interaction of marsh by fence by treatment was significant

(Figure 5.11). The reason for this was again the response of the unfenced,

transplanted sods from the S. patens site. However, there was also a dramatic

difference between the fenced plots in each marsh which contributes to that

interaction. The fence by treatment interaction was significant because the unfenced

controls had more species than the unfenced transplants at all dates, but only in July

1994, April and July 1995 for the fenced plots (Figure 5.11). The marsh by

treatment interaction was not significant.

Despite there being no net effect on richness of transplanting sods from the S.

patens marsh to the fenced Sagittaria areas, species composition shifted (Table 5.6).

Many of the species from the S. patens marsh were lost throughout the course of the

experiment, with the exception of A. subulatus, which was present at both sites. The

transplanted sods gained quite a few species from the Sagittaria marsh which became

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 122

S. Transplantj Census Dale Census Dale Transplant fenced; b) Control S. S. patens, — Control Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 unfenced. Asterisks indicate 1 0 3 5 2 6 4 8 7 1 5 0 8 7 4 3 6 2 c u V) 0) c u •C £ a JC £ in Sagittaria, fenced; d) Sagittaria, Transplant Census Dale Census Dale unfenced; c) Control Control Transplant mean separation at p<0.05. Figure I. 5.1 Species richness for salinity experiment: a) patens, Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 Jul 93 Oct 93 Apr 94 Jul 94 Oct 94 Apr 95 Jul 95 1 0 2 5 4 3 7 6 8 1 0 3 2 5 4 6 8 7 & o co a <0

permission of the copyright owner. Further reproduction prohibited without permission. 123 Table 5.6. Species changes in the transplanted sods in the salinity experiment. Numbers indicate the number of sods out of 16 from which that species was gained or lost.

GAINED LOST Sagittaria marsh 1 Mikania scandens 0 2 1 Vigna luteola 4 4 2 Ipomoea sagittata 0 11 2 Phyla nodiflora 0 5 3 Aster subulatus 5 1 3 Eleocharis cellulosa 0 3 3 Hydrocotyle sp. 0 3 3 Polygonum punctatum 0 2 3 Ptilmnium capillaceum 0 4 4 Aster teniufolius 1 0 4 Galium tinctorum 0 2 P3 Cyperus felicinus 4 0 P3 Cyperus oderatus 2 0 P3 Eleocharis fallax 2 0 P3 Scirpus validus 3 0 P5 Echinocloa crusgalli 1 0

S. patens Marsh (Fenced) 1 Scirpus americanus 0 4 2 Spartina altemiflora 0 3 3 Aster subulatus 5 1 3 Eleocharis fallax 0 2 3 Lileaopsis chinensis 0 3 3 Scirpus validus 0 1 4 Aster tenuifolius 0 1 4 Eleocharis parvula 0 2 4 Scirpus robustus 0 1 4 Spartina cynusoroides 0 3 4 Vigna luteola 2 0 5 Juncus romerianus 1 0 SI Mikania scandens 2 0 S2 Ipomoea sagittata 3 0 S2 Phyla nodiflora 3 0 S3 Eleocharis cellulosa 1 0 S3 Ptilmnium capillaceum 2 0 S4 Galium tinctorum 1 0 S5 Sacciolepis striata 2 0 S5 Sium suave 1 0

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 124 established in the sods, presumably via seed dispersal from nearby plants. The

results of the Sagittaria sods being transplanted show a few species in the sods as a

result of local seed set, but also show many species dropping out, once again a

notable exception is A. subulatus. A. subulatus is present in the seed banks of both

sites and therefore could have been in the sod when it arrived at its new location, or

transported in from neighboring plants (L. Gough, unpublished data).

Biomass

Total biomass results were also driven by a significant three way interaction

of marsh by fence by treatment, resulting from no biomass present in the S. patens,

unfenced, transplanted sods (Table 5.7, Figure 5.12). The marsh by fence interaction

was caused by the more dramatic difference in the S. patens marsh. The main effect

of fence was also significant due to greater biomass in the fenced plots. Overall,

treatment was significant with greater biomass present in the controls (Figure 5.12).

Live biomass analyzed separately shows the same pattern as total. Dead biomass

alone only shows a significant main effect of fence, with greater litter in the fences,

and treatment, with greater litter in the control plots (Table 5.7).

Several species were examined individually for treatment effects on biomass.

Scirpus americanus occurred only in the S. patens marsh and was a dominant there.

The analysis of the log-transformed biomass reveals a significant main effect of

fence, with greater biomass inside the fences (p=0.0007), and a significant main

effect of treatment, with greater biomass in the control plots (p=0.0001). When live

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 125 Table 5.7. Source table for biomass analysis of salinity experiment.

p-value Source df Total Live Dead

marsh 1 0.51 0.98 0.10

fence 1 0.002 0.03 0.002

marsh*fence 1 0.02 0.05 0.12

rep(marsh*fence) 28

treatment 1 0.0003 0.0080 0.0001

marsh*treatment 1 0.22 0.14 0.92

fence*treatment I 0.83 0.30 0.13

marsh*fence*treatment 1 0.01 0.01 0.36

error 27

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. FC FT UC UT Treatment

I Live 11 Dead

b 80

Treatment

E Live HI Dead

Figure 5.12. Biomass results for salinity experiment: a) S. patens marsh; b) Sagittaria marsh. Treatment abbreviations are: FC=fenced, control; UC=unfenced, control; FT=fenced, transplant; UT=unfenced, transplant. Different letters indicate means of total biomass are different at p<0.05 within each marsh.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 127

UC FT Treatment

! live H dead

Figure 5.13. Biomass results for S. americanus in salinity experiment. Treatment codes as in Figure 5.12. Different letters indicate means of total biomass are different at p<0.05.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 128 biomass is examined separately, it has a significant treatment effect (p=0.0001) but

the fence effect is only marginally significant since there is almost no live biomass

present in the transplanted, fenced plots (p=0.10). Dead biomass shows the same

pattern as total. Figure S. 13 presents the results for individual treatment means.

S. patens was a dominant at the S. patens marsh and a codominant in the

Sagittaria marsh. Biomass of this species showed only a significant marsh by fence

by treatment interaction (p=0.01) (Figure 5.14). The cause of this interaction was the

increase in biomass when S. patens was transplanted from the S. patens marsh to the

fenced, Sagittaria marsh areas which contrasted with the loss of biomass in the

unfenced areas. The results for S. patens from the Sagittaria marsh were quite

different with no effect of transplanting into fenced areas but an increase in biomass

when transplanted into unfenced areas.

Sagittaria lancifolia was a dominant at the Sagittaria marsh and a codominant

at the S. patens marsh. Total biomass showed a significant marsh by treatment

interaction (p=0.05) which seemed to be driven by the difference in biomass between

the two marshes, not by the behavior of this species when transplanted since it

decreased biomass in both cases (Figure 5.15). The controls had greater biomass

than the transplanted plants (p=0.0001) and the Sagittaria marsh sods had greater

mass than theS. patens marsh sods (p=0.0001). When divided into live and dead,

live biomass shows only a significant marsh effect and treatment main effect. Dead

biomass shows a similar pattern but also had a significant fence effect (p=0.03), with

greater biomass accumulating inside the fences.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 129

Treatment

b 30

w w m —|—m m w sm —,—m m m m —|—m m m FC UC FT UT Treatment

Figure 5.14. Biomass results for S. patens in salinity experiment for a) S. patens marsh and b) Sagittaria marsh. Treatment codes as in Figure 5.12. Different letters indicate means of biomass are different at p<0.05 within each marsh.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. FC UC FT UT Treatment

Treatment

live H d ead

Figure 5.15. Biomass results for S. lancifolia in salinity experiment for a) S. patens marsh and b) Sagittaria marsh. Treatment codes as in Figure 5.12. Different letters indicate means of total biomass are different at p<0.05 within each marsh.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 131 DISCUSSION

Community Response

The maintenance of plant species diversity in marshes is controlled by the

relative influence of many factors including water depth, salinity level, herbivory,

other forms of disturbance, and competition (Chabreck 1972, Bakker 1985, Wilson

and Keddy 1986, Mitsch and Gosselink 1986, Day et al. 1988, Keddy 1990, Garcia

et al. 1993, Gough et al. 1994, Gaudet and Keddy 1995, Grace and Pugesek, in

press). The response of marsh communities to manipulations of these different

factors in the field was the focus of this study.

Subjecting sods of marsh vegetation to increased flooding stress caused a

significant loss of species in the fresher marsh community (Figure 5.4) and a loss of

species in both communities when sods were also exposed to herbivores (Figure 5.5).

There was a great deal of shifting of species within these sods over the course of the

experiment (Table 5.3) to result in an overall loss. Biomass decreased in die

lowered sods across both marshes and fence combinations, likely as a result of

flooding inhibition of growth.

The effect of alleviating flooding stress on species richness was less clear,

although biomass significandy increased in raised sods (Figure 5.6b). Within each

marsh by fence combination the raised sods had the same number of species as the

controls. Individual species shifted a great deal in these plots resulting in no overall

change in richness. A slight trend towards increasing richness in the raised sods was

seen in the overall treatment means in July 1995 when the raised sods had

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 132 significantly more species than the controls (Figure 5.6a). However, this difference

was only present at the final census date.

Transplanting sods from a fresh marsh to a more brackish marsh caused a

decrease in species richness overall, although some species from the brackish marsh

were able to establish in the sods, likely due to seed dispersal from nearby plants

(Table 5.6). Transplanting also caused a decrease in biomass.

Transplanting sods from the brackish marsh to the fresher marsh and

protecting sods from herbivores caused no change in overall richness, even though

most of the brackish species died off and many fresh species established in the plots

(Table 5.6). Many of these fresh species are present in the seed bank at the site and

can produce viable seed that is either water or air dispersed into the plots (L. Gough,

unpublished data). Those sods transplanted to the unfenced areas were completely

consumed by the herbivores (most likely nutria), presumably because they contained

5. americanus, the preferred food of these animals, which is not normally found at

this site. Biomass was not affected in the transplanted sods inside the fences.

The herbivores had no significant influence on overall biomass accumulation

for any of the treatments (except the transplanted sods noted above) indicating that

herbivory pressure did not interact with growth of the species although it had some

effect on individual species (see below). Herbivory did intensify the loss of species

from flooded and transplanted sods, so perhaps the removal of certain species

allowed increased biomass accumulation by those species remaining, resulting in no

overall biomass change.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 133 Individual Species

The dominant species at the two sites behaved quite differently from each

other in the different treatments, as did the rare species as evidenced by the changes

in abundances in most plots (Tables S.3 and S.6). Scirpus americanus showed a

significant increase in biomass in all treatments when herbivores were excluded

(Figures 5.8 and 5.13) which supports other researchers findings that this species is

severely impacted by nutria herbivory (Ford 1996, other chapters here). This species

showed no detrimental effect of flooding, confirming other work that has found S.

americanus to be able to withstand flooding when other species may not (Broome et

al. 1995). Somewhat surprisingly, S. americanus ramets transplanted to the fresher

marsh died before the conclusion of the study, despite the alleviation of salinity

stress. This was likely due to the competitive exclusion by S. patens, which

responded positively to lowered salinity (see below).

Spartina patens increased biomass in the raised sods and decreased in the

lowered in the S. patens marsh without being affected by herbivory; no effect of

flooding treatment was seen in the fresher marsh. The decrease may have been a

result of a shift in dominance with S. americanus which did not decrease biomass in

flooded plots, supporting the conclusions of Broome et al. (1995) that increased

flooding favors S. americanus over S. patens. In contrast with their findings

however, we found that S. patens transplanted from the brackish marsh to the fresh

marsh increased production overall while S. americanus died out. It is difficult to

separate the effects of competition from the inability to tolerate stressful abiotic

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 134 conditions in this study, but as has been found elsewhere, these two species appear to

be in constant battle with one another and factors such as herbivory or increased

water depth may tip the balance towards one species or the other (see Chapter 4).

Sagittaria lancifolia responded to altered elevations in both marshes with the

raised sods accumulating more biomass than the lowered. The lowered sods did not

differ in biomass from the control sods, as others have documented for this species

(Howard and Mendelssohn 1995). This difference was more pronounced in the

Sagittaria site where it was significantly more abundant. Herbivory had no effect on

biomass accumulation, as has been seen before (see Chapter 4). When sods were

transplanted from the brackish site to the fresher site, biomass decreased, perhaps

because of the relative increase of S. patens. A similar result was seen for the fresh

sods transplanted to the brackish marsh, which was likely due to salinity stress that

restricted growth. Sagittaria was present at the brackish marsh, but does not

naturally accumulate as much biomass at that site, likely due to the higher salinity

levels (Figure 5.15).

Implications for Sea Level Rise

As others have found, the combination of a stress such as increased salinity or

increased flooding with a disturbance such as herbivory may cause increased

mortality in marsh species that otherwise could withstand such stresses (Sale and

Wetzel 1983, Herndon et al. 1991, Grace and Ford 1996, Baldwin 1996). Although

not examined here, the combination of salinity and flooding may also severely

impact vegetation by increasing mortality when niether factor alone would cause

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 135 plant death (McKee & Mendelssohn 1989, Flynn et al. 1995, Baldwin 1996). The

encroachment of sea level rise into fresh areas bringing deeper saline water may

cause entire communities to migrate (Warren and Niering 1993, Boesch et al. 1994),

requiring a substantial replacement of species.

Many of the questions surrounding the issue of relative sea level rise effects

deal with the method of response of the plants. Vegetation regrowth following

clipping may occur much faster through clonal resprouting than seedling re

establishment (Baldwin 1996). Seedlings of many of the dominants are rarely seen

in these marsh communities although there is a large viable seedbank (L. Gough,

unpublished data). Unless disturbances such as herbivory clear space on the marsh

substrate, seedling establishment is severely limited. In the areas studied here

herbivory pressure was rarely severe enough to result in bare mud, and if it did,

plants colonized the areas within months. As has been suggested elsewhere and

demonstrated here, species may drop out of stressful habitats more quickly than they

may be gained in less stressful areas (Marrs et al. 1996). In an experiment testing a

model simulating response to flooding, clonal marsh species failed to migrate above

the water line to remain emergent (de Swart et al. 1994); the rate at which species

are able to migrate in response to environmental change will greatly affect the ability

of ecosystems such as marshes to maintain themselves.

Individual species' responses to these treatments varied considerably in our

study and in others (Mendelssohn and McKee 1987, Flynn et al. 1995, Baldwin

1996). If the tolerance of salinity or flooding involves a trade-off with competitive

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 136 ability as has been seen for some wetland species (e.g. Grace and Wetzel 1981,

Bertness 1991a and b), changing abiotic variables may not only limit die species that

can survive in a given habitat but cause shifts in community composition as a result

of competitive interactions. This may be true for those species selectively subjected

to herbivory like S. americanus', although it is flood tolerant it may not survive under

increased herbivory pressure. Herbivory rates fluctuate in these marshes and may

produce a variety of results in terms of sediment deposition and plant species

composition (Ford 1996). At the moderate herbivory rates present in this study,

plants were not as able to withstand flooding treatments and transplants as when they

were protected from herbivory. In these marshes where the dominant species mosdy

reproduce clonally, it is unclear how communities might shift to being more salt

tolerant if the salt tolerant species lack appropriate dispersal mechanisms and may be

more susceptible to herbivory. Thus the initial composition of the community, the

availability of propagules, and the presence of herbivores will have an important

effect on marsh response to sea level rise.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 6

AN EXPERIMENTAL TEST OF A STRUCTURAL EQUATION MODEL OF SPECIES RICHNESS IN A COASTAL MARSH

INTRODUCTION

Perhaps the greatest task facing community ecologists today is the ability to

understand the factors influencing plant community structure in order to predict how

communities will react to changing environmental conditions, human caused or

otherwise. A survey of the literature reveals a great many models of species

diversity with application to a variety of communities (see Huston 1994, Palmer

1994, Rosenzweig 1995 for recent reviews). However, there is little agreement as to

what variables are necessary to parameterize these models as well as how to measure

the variables themselves. Many of these models are constructed in one habitat with

hope of pertaining to others; rarely are the models tested outside the initial parameter

space (for an exception see Shipley et al. 1991).

One of the few general community patterns that is agreed on by many

researchers is the hump-shaped productivity-diversity curve between plant species

richness and community biomass (Grime 1973). This relationship is characterized by

peak richness at a low to intermediate level of biomass. This pattern has been seen

in many different types of communities, although the variance explained by the

relationship varies and may be somewhat scale dependent (Wheeler and Giller 1982,

Wilson and Keddy 1988, Wheeler and Shaw 1991, Garcia et al. 1993, Shipley et al.

1991, Moore and Keddy 1989). Theoretical discussions have tended to interpret the

137

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 138 x-axis of this relationship as productivity (e.g. Huston 1979), though in practice a

number of different variables have been used to represent the environmental axis

including biomass (most commonly), litter, soil nutrients, elevation, and moisture

gradients (e.g. Tilman 1986). In general, it is agreed that this relationship is caused

by colonization and growth limitation on the left side of the peak, and competitive

exclusion on the right side of the peak (but see Taylor et al. 1990, Abrams 1995).

However, substantial disagreement exists on the mechanism of competition (see

Grace 1990, 1991, 1993 for reviews).

Not all studies of the relationship between species richness and habitat

productivity have found a strong relationship (Garcia et al. 1993, Abrams 1995). In

an earlier study in two Louisiana coastal marshes (Gough et al. 1994, Chapter 2),

richness was not found to be explained by biomass. Rather, environmental variables

had to be included in a multiple regression model in order to explain the observed

variation in richness. With salinity, flooding, and soil fertility included as

independent variables, 82% of the variance in species richness was explained;

biomass accounted for only 3%. We concluded that, in systems such as coastal

marshes where plant species may be differentially adapted to environmental

gradients, the relationship between biomass and richness is weak because of

environmental gradients in the species pool. As a result of this interpretation, we

proposed a simple conceptual model (Figure 2.7) which included the environmental

variables in addition to biomass as a means of explaining variation in species

richness.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 139 In order to further evaluate the model proposed by Gough et al. (1994)

(Chapter 2), Grace and Pugesek (in press) developed a structural equation model of

the factors controlling plant species richness in a coastal wetland landscape.

Structural equation modeling (SEM) is a multivariate technique designed to evaluate

complex hypotheses using simultaneous partial regression analyses (Hayduk 1987,

Bollen 1989). Through the analysis of covariance relationships, SEM allows for (1)

the evaluation of alternative models, (2) the partitioning of direct and indirect effects,

and (3) the prediction of dependent variable responses to changes in independent

variables, singly or in combination (Johnson et al. 1991, Wesser and Armbruster

1991, Mitchell 1992). Using this method, Grace and Pugesek (in press) proposed

both a general model whereby species richness is controlled by environmental stress,

disturbance, and plant density, and a specific version of the model for a coastal

wetland.

While structural equation models are typically evaluated using non-

experimental data, experimental tests are possible. For example, Wootton (1994)

provided a partial test of a path analysis model (a related technique) using

experimental procedures. Wootton (1994) found that the combination of path

analysis and experimenation provided a powerful approach to evaluating hypotheses

that are normally beyond the scale of complexity feasible for traditional hypothesis

testing. In this paper, the results from field experimental manipulations are used to

test both qualitative and quantitative predictions from a SEM of a coastal wetland

system.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 140

METHODS

Study Area

The research reported here was conducted in the Pearl River basin, located at

the border between Louisiana and Mississippi (White 1979, other chapters this

volume). The system consists of three main channels of the Pearl River with braided

distributaries throughout. The wetlands studied here were herbaceous communities

ranging from fresh to salt marsh.

Data Collection

Model

The SEM model presented in this chapter was constructed using data

collected in the Pearl River basin (Grace and Pugesek, in prep.). Twenty sampling

sites were located within the basin using a stratified random procedure. Ten 1 m2

plots were established within each site during May-July of 1993. One site was

determined to be impacted by a roadway and was eliminated from the study leaving

a total of 190 plots. Data was collected in September 1993 and September 1994.

Only the 1994 data is reported in this chapter since it is deemed most appropriate for

comparison to the experimental data available. Elevation, salinity, soil organic

content, percent carbon, percent nitrogen, disturbance, and light levels at the ground

were recorded for each plot (for detailed description see Grace and Pugesek in press).

Representative 0.25 m x 0.25 m plots adjacent to each permanent plot were harvested

as estimates of above-ground biomass. The species composition and species richness

of each plot were determined as well.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 141

Experimental Testing

The experiments to test the model were established in the summer of 1993

and were maintained until August 1995. They were located at two sites within the

Pearl River basin, one a fresh/oligohaline marsh dominated by Sagittaria lancifolia

and the other a mesohaline/brackish marsh dominated by Spartina patens and Scirpus

americanus (for detailed descriptions see Chapters 4 and 5, METHODS). Several

manipulations were conducted. One-meter-square plots were fertilized with

commercial nitrogen, had sediment added, or were left undisturbed as controls (see

Chapter 4). Smaller sods (0.33 m diameter by 0.15 m deep) were dug up and placed

back in the ground (disturbed control), transplanted to the other site (transplant),

placed on top of sediment so that they were 10 cm above the marsh surface (raised),

deposited in a hole so that they were 10 cm below the marsh surface (lowered), or

simply marked and not manipulated (undisturbed control) (see Chapter 5). One set

of each treatment was placed inside a 7 x 7 m mammalian exclosure and one set was

placed outside. This was replicated eight times at each site for a total of 256 plots.

Two plots were eliminated from analysis as they were covered by heavy wooden

debris at the conclusion of the study so 254 plots were included in this study.

In August 1995, all plots were harvested for above ground biomass. Light

was measured in all plots using an integrating light wand that measures

photosynthetic radiation striking a strip 1 cm wide and 1 m long (LiCor Inc.,

Lincoln, NE). Elevation was measured for lowered, rasied, and control plots at the

conclusion of the study using a laser survey system (Spectra-Physics Instruments,

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 142 Dayton, OH). Carbon and nitrogen content were determined for a composite sample

for each marsh site using a CHN Elemental Analyzer (Control Equipment

Corporation, Lowell, MA). Disturbance was determined by visual estimation of the

percent of the plot surface that was bare mud. Species richness was determined as

the number of species found in the harvested sample when the samples were sorted

to species before being dried for biomass analysis.

Analysis

SEM Model

Data were transformed prior to analysis to achieve linearity with richness

(Table 6.1). The equations to be used for transformations were determined using the

TableCurve program (Jandel Scientific, San Rafael, CA). The simplest equation that

provided a near-maximal fit to the data was chosen; equations were restricted to

second order or less. Indicators used to estimate the conceptual (latent) variables of

interest are shown in Table 6.1.

The LISREL program (Joreskog and Sorboom 1993) was used to fit data from

the 190 plots to hypothesized dependence models (see Grace and Pugesek in press

for more detailed discussion of modelling method and Bollen 1989, Hair et al. 1995,

and Hayduk 1996 for additional documentation on LISREL). The protocol used for

evaluating alternative models was that of a nested analysis (Hair et al. 1995). In a

nested analysis a theoretically based initial model is evaluated for both fit and

parsimony in comparison to all other models containing the same causal order of

variables. Comparisons between the specified model and all other nested models is

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 143 Table 6.1 Transformations used for variables in SEM analysis. Shapes of relationships that were not linear are shown in Figure 6.1. Latent variables refers to the conceptual variables shown in Figure 6.2 while indicator variables refers to the measured variables. RELLOW and RELHIGH refer to the lowest and highest elevations in a plot relative to the water's edge. PCTORG refers to the percentage of soil organic matter. PCTC refers to the percentage of soil carbon. PCTOPEN refers to the surface area of a plot that was bare mud, and MASSM2 refers to the dry biomass per meter square.

Latent Indicator Variables Variables Transformation

Salinity saltl untransformed

salt2 untransformed

Flooding floodhi 60 - RELHIGH

floodlo 60 - RELLOW

Infertility soil organic 7.405 + 0.232*PCTORG - 2.536*SQRT(PCTORG)

soil carbon 7.228 + 0.545*PCTC - 3.858*SQRT(PCTC)

Disturbance percent disturbed 1.639 - 0.0975*PCTOPEN + 0.00143*SQR(PCTOPEN)

Biomass mass 2.17 - 0.00332*MASSM2 + 0.284*SQRT(MASSM2)

Richness number of species un transformed

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 144

accomplished within LISREL through evaluation of modification indices and the

significance of pathways. Through this process, the model with the best fit,

parsimony, and biological justification is identified and estimated.

Experimental Testing

Bivariate plots of the experimental data were made to compare the observed

relationships between richness and other variables derived from the data used to

parameterize the SEM. Quantitative predictions were made using the structural

equation model to determine if biomass and richness would increase or decrease

given the changes in the environmental variables manipulated in the experiments.

The predictions were simply a directional change taken from the coefficient of the

total effect of the manipulated variable on either biomass or richness. The total

effect is estimated to describe the net relationship between variables that are

connected by both direct and indirect pathways. In this case, for bitonic relationships

(those that include increasing and decreasing phases) predictions were based on the

mean values of predictor variables. For example, unfenced plots had a mean percent

of exposed plot surface of approximately 10 and fenced plots had a mean of 0.

Examination of the relationship between disturbance and richness (Figure 6.1) shows

that these values both occur to the left of the peak and suggest that fencing might

lead to a decline in richness, albeit a small one. For qualitative assessment, these

predictions were evaluated using grand means for each treatment without applying

any statistical analysis.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 A »—4 50 • ••• 40 • «M •• • 30 (fig. con'd.) Elevation (cm) • • • 20 10 0 C 6 Salinity (ppt) • • •• M • M •• • • •» •• •» 0 5 15 20 8 c S S 1 « 2 10 •• richness versus: a) biomass, b) salinity, c) elevation, d) percent open (bare mud) and e) percent soil carbon. Figure 6.1 Bivariate plots of the 1994 data used to parameterize the model for •• Biomass (g/m2) 1000 1000 2000 3000 4000 5000 6000 7000 • ••• 20

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 146

c. J3o

CMo to o o ssauipiu sapeds

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 147

Quantitative predictions were also made for species richness using the total

effects estimates from the SEM model. These values are the net effect of one latent

variable on another and give the per unit expected change in biomass or richness;

e.g. the change in biomass per cm the sod is lowered (cm increased flooding). The

predictions were tested by examining the grand means of the treatments and

comparing the experimental manipulations to their respective controls. Total effects

could not be used for the fertilization treatment as the fertilization treatment itself

was not summarized in the infertility variable. Therefore, predicted changes for

fertilization were made based on the observed changes in biomass that accompanied

fertilization.

Finally, the overall fit of the model to the experimental data was evaluated.

The LISREL regression equation does not include an intercept. The grand means for

the predicted values was subtracted from the grand means of the observed values to

obtain an estimate of the intercept. This value was subsequently subtracted from

each individual predicted value. Due to this procedure, the predicted and observed

values were not expected to have a one to one relationship. Using the regression

equation from the LISREL output to explain richness, a predicted richness value was

calculated for each of the 254 plots and compared to the observed richness with

simple linear regression. This process was repeated for subsets of the data in order to

determine the applicability of the model relationships to those observed in the

experimental results.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 148 RESULTS

SEM Model

Bivariate Relationships for Data Used to Build SEM

Biomass explained 25% of the variance in richness (Figure 6.1). Light was

also reasonably well correlated with richness, yielding an R2 of 0.28. However, light

was not included in the SEM model due to lack of reliability of the experimental

light data (see below). Salinity and flooding both significantly fit linear relationships

(R2=0.02 and 0.18, respectively). Disturbance was strongly correlated with richness

(R2=0.31) while percent carbon was less so (R2=0.11).

Model Fit

The structural model (Figure 6.2) explained 47% of the variance in species

richness. The equation explaining richness was:

richness=0.84(biomass)-0.98(salinity )-0.043(flooding)+1.3 2(infertility )

and biomass was explained by the following equation:

mass = -0.026(flooding) + O.lO(infertility) + 0.7652(disturbance)

which accounted for 86% of the variance. Note that in these equations, biomass,

infertility, and disturbance are in transformed units (Table 6.1). The model fit the

data well with a Root Mean Square Error of Approximation (RMSEA) of 0.067

(df=20, p=0.19) and a Goodness of Fit Index (GFI) of 0.96. These index values

show that there was a strong correspondence between predicted and observed

covariances. For the RMSEA, a p-value greater than 0.05 indicates no significant

deviation between expected and observed values. For the GFI, values of 0.90 or

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 149

R2= .4 7 . RICHNESS ' .44(+/-) .28( + /-) + .28( -.83 -.16 -.25 -.34 -.25 -.32 SALINITY FLOODING DISTURBANCE INFERTILITY .40 .39 relationship (+/-). R2 values are given for biomass and richness as the endogenous coefficients ("r" values); signs variablesindicate beingpositive predicted or negative by theircorrelation respective or pathways.a bitonic correlations between variables. Coefficients along arrows are partial regression Figure 6.2. The structural SEM model of the 1994 data. Arrows indicate significant

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 150

better generally correspond to a high degree of fit for the structural model (see Hair

et al. 1995 for more information on these indices). The values shown in Figure 6.2

are the completely standardized values resulting from a covariance analysis and can

be interpreted as partial regression coefficients ("R" values). These values, therefore,

denote the degree to which one variable is a strong predictor of another.

Predictions

The quantitative predictions are made in Table 6.2. For each of the

manipulations conducted, a simple direction of change prediction is indicated. The

specific quantitative predictions for species richness are made in Table 6.3.

Experimental Testing

Bivariate Relationships for Experimental Data

Before the experimental data was transformed to be linearized, the same

bivariate relationships were examined (Figure 6.3). These plots are quite similar to

those for the descriptive data (Figure 6.1) so the descriptive transformation equations

were used without reservation. Salinity and soil carbon are not presented as bivariate

plots because these two variables were the same for all plots within a marsh site.

Although there was no pattern seen in the plot of richness against elevation (Figure

6.3c), when the elevational changes due to raising and lowering sods were examined,

these plots were different (see Chapter 5). Biomass and light were well correlated

with richness on their own, but the relationship between biomass and light was very

weak in the experimental plots (R2=0.08). We believe this relationship was a result

of a problem with the light data collection, that surrounding vegetation was allowed

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 151

Table 6.2. General responses predicted for experimental manipulations based on total effects results of SEM model. Values in parentheses indicate observed changes.

Responses Manipulations Biomass Richness

1. fertilization + (+) - (-)

2. decrease elevation - (-) - (-)

3. increase elevation + (+) + (+)

4. decrease salinity O(-) + (-)

5. increase salinity 0 (0) - (-)

6. exclude herbivores + (+) - (0)

Table 6.3. Specific quantitative predictions for species richness for experimental treatments based on total effects from SEM model. Observed change was calculated using means from the ANOVA analysis. All non-zero observed change values were significantly different from appropriate controls (see Chapters 4 and 5).

Manipulation Dredicted chance observed chance

fertilization -0.2 -1.2

decrease elevation 10 cm -0.6 -2.0

increase elevation 10 cm +0.6 +0.8

increase salinity 2 ppt -1.9 -4.2

decrease salinity 2 ppt +1.9 -1.6

exclude herbivores -1.1 -0.03

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 152 100 100 R2-0.21 R2-0.26 60 80 60 40 40 Percent Bare Mud Percent Ambient Light 20 20 0 0 0 2 8 4 6 8 4 6 2 0 10 10 c ■6 i£

R2-0.37 Elevation (cm) Biomass (g/m2) 500 1000 1500 2000 2500 3000 3500 Figure 6.3 and Bivariated) percent plots bareof themud. experimental data: a) biomass, b) light, c) elevation, 0 5 10 15 20 25 30 0 8 2 4 8 6 4 2 0 6 10 10 C & -6

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 153 to shade plots and thus lower light so that many low biomass plots had very low

light levels. In other studies in the Pearl River (Grace and Pugesek in prep., Grace

and Pugesek in press, Chapters 2 and 3), light was measured somewhat differently

and was well correlated with biomass. Due to the problems with this data, light was

removed from the analysis. This had very little effect on the SEM model chosen as

the best fit since biomass was a strong predictor of richness and light was not.

Qualitative Predictions

Simply examining the grand means of the treatment manipulations, it is clear

that the model predicted most of the qualitative changes in biomass and richness

correctly (Table 6.2). Fertilization led to an increase in biomass and a decrease in

richness as predicted. Manipulations of elevation also resulted in the changes in

biomass and richness that were predicted. Increasing salinity resulted in no change

in biomass but a decline in richness, while excluding herbivores led to an increase in

biomass. When salinity was decreased by transplanting sods from a brackish marsh

to a fresher marsh, biomass was not predicted to change and richness was predicted

to increase; however, both biomass and richness decreased (Table 6.2). Further,

excluding herbivores was predicted to reduce richness but no reduction was observed.

Quantitative Predictions

The experimentally observed results for species richness are compared to the

predicted values from the model in Table 6.3. Similar results are seen here as for the

qualitative predictions, except that, as expected, differences in observed versus

predicted could be seen at a finer scale. For example, while decreasing elevation did

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 154

cause a decrease in richness, the observed decrease was greater than predicted.

Increasing elevation caused a gain of about one species, very close to the predicted

value. Increasing salinity had a greater effect than predicted, with sods losing on

average four species, while decreasing salinity had the opposite effect as predicted,

with a loss of one species despite alleviating salinity stress. Excluding herbivores

caused no change in overall richness which was slightly less than predicted. Using

biomass means for predicted response to fertilization, there was little change in

richness but in the observed results there was a loss of about one species.

Model Applicability to Experimental Data

When all the experimental data are considered together, the predictions

explain 35% of the variance in the observed richness numbers (Figure 6.4a). This is

a reasonable fit considering the SEM model explained 46% of the variance in

richness. However, examination of the responses of individual treatments revealed

that not all treatment responses were predicted equally well.

Since the fertilization experiment involved increasing biomass in a manner not

normally encountered in this system, the fertilization experiment plots were

eliminated from the results to determine if the predictions would improve. The

undisturbed control plots were also removed from the dataset as they would

undoubtably be quite different from the manipulated sods. The model improved,

with 49% of the variance explained (Figure 6.4b).

The fencing treatment was another externally imposed condition which was

not sampled in order to construct the model originally so a similar analysis was run

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LA LA (fig. con'd.) #• #• Predicted Richness R2=0.49 1 1 2 3 4 5 6 1 3 0 8 5 2 6 9 4 7 10 d 0 w u>

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 156

OfflCONCDlO'sfOCM o sssuipjy ps/UQsqo

OffiCQNDlfi^nCMv-O

ssauipiy ps/uesqo O

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 157 with only the unfenced data. The model improved with an R2 of 0.53 (Figure 6.4c).

Finally, both the fenced data and the fertilized and undisturbed control data were

eliminated and the highest R2 was obtained, with 63% of the variance in richness

explained (Figure 6.4d).

DISCUSSION

Similar to what was found by Grace and Pugesek (in press), the SEM of 1994

data and its underlying relationships indicate that species richness at the Pearl River

is controlled by a combination of direct and indirect environmental effects (Figure

6.2). Several variables were observed to have effects on biomass which then

represented indirect effects of these environmental variables on richness. Flooding

and infertility are both believed to have potential effects on biomass but in this

system, it appears that disturbance by waves and intense herbivore activity are by far

the most important factor controlling variations in biomass. To a great extent, the

overwhelming effect of biomass acts to overshadow relationships that would likely be

strong under less disturbed conditions (particularly salinity and infertility). Direct

effects of environmental variables on richness were also conspicuous in the model

and it has been argued that these represent environmental gradients in the species

pool that can strongly limit potential species richness (Gough et al. 1994, Grace and

Pugesek in press).

The suite of experiments whose effects are summarized in this paper (see also

Chapters 4 and 5) found that indeed, nearly all these variables can affect richness in

ways that are consistent with predictions. In general, the predictions made using the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 158 structural equation model corresponded fairly well with observed experimental

results, with some exceptions.

In general, the elevation treatments responded as the predictions directed.

However, when elevation was decreased, two species were lost on average when only

one species was predicted to be lost. This was likely due to the interaction of

herbivory with increased flooding; the ANOVA results indicated a significant

interaction between fencing and treatment which showed that when plots were

lowered and exposed to herbivory they lost more species than plots that were

protected (Chapter 5).

The salinity results behaved according to the predictions when salinity was

increased, although more species were lost than were predicted. When salinity was

decreased, species numbers also decreased, counter to predictions. This was due to a

fencing effect; the sods that were transplanted into unfenced areas were consumed

by animals. The sods contained Scirpus americanus, the preferred food species of

the nutria and muskrat, and when they were placed in a marsh where this species

does not normally occur, they were selectively consumed. However, in the fenced

plots, richness increased by one species, as predicted (Chapter 5).

This interaction between herbivory and the experimental treatments was not

detected by the model. When an interaction term was placed in the SEM model, it

was not significant. The fences themselves created conditions not sampled for the

model. The prediction of one species being lost when herbivores are excluded (Table

6.3) was not met. This prediction is based on a disturbance estimate of 10% for the

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 159

unfenced plots and of 0% for the fenced plots. However, the disturbance variable for

1994 included other elements in addition to herbivory, such as wave action and

wrack deposits. No change was observed in richness between fenced and unfenced

areas overall, or within any treatment (Chapters 4 and 5). Nutria activity in the

Sagittaria marsh was minimal, while in theS. patens marsh, some areas would be

tom up by hogs or subjected to severe herbivory but would recover within months.

This constant disturbance followed by recovery period cycle may have prevented an

overall change.

Like the fencing treatments, the fertilization treatments represented an

alteration not sampled for the model. Adding fertilizer to the soil caused an increase

in biomass and therefore would be mediated through biomass in the model.

However, it was not factored into the infertility variable so the expected increase in

biomass with increasing fertility was not part of the model to explain richness. The

predictions for the fertilization treatment were made by taking grand means of the

predicted richness numbers and observed richness numbers. Using that method, no

change is predicted based on fertilization, but using the total effects, species should

be lost. This occurred in the study, much more quickly and significantly in the

fenced plots than the unfenced (Chapter 4).

The two aforementioned problems with the model predictions become more

apparent when the predicted vs. observed data are plotted (Figure 6.4). Removing

the fertilization data improved the fit of the model, as did removing the fenced plots.

The best plot was achieved for the unfenced flooding and salinity plots with an R2 of

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 160 63%. Although this was not expected initially, as the analysis was completed it

became clear that artificial treatments, those not found naturally in the system, were

not well predicted from the model, likely because they lie outside the inital parameter

space. Disturbance was recorded in a similar manner in the descriptive data and the

experimental data, but the plots being fenced imposed a restriction on herbivore

activity that was not found in the sampling, despite sites being included in the dataset

that most likely were experiencing minimal herbivory.

The other issue which became clear from the experiments was that the model

could not predict where interactions would be important. There was no way to

predict that sods moved from the saltier marsh to the fresh marsh would be

consumed by animals. However, the response of most plots was predicted by the

model in terms of overall richness predictions.

Despite the abovementioned problems with the testing of the model, on the

whole, our results were quite encouraging. The relationships between richness and

the environmental variables sampled were similar for both the 1994 data and the

experimental results. As the experimental manipulations were conducted in only two

areas within the marsh and the 1994 data included 19 such locations, and since

methods differed somewhat between studies, it is no surprise that some of the

relationships were not quite the same. Despite the fact that both salinity and

flooding are important components of the model, the experimental data which had

little spread for those two variables still fit the model well.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 161 In the case of this study, the approach of combining an SEM model with

experimental testing has advantages over either approach individually. The

experimental results revealed interactions, particularly with herbivory, that were not

detected by the model. However, the model describes more generally how the

environmental variables cause biomass and richness to change over a broad area,

providing insight into the relationships among variables that would be virtually

impossible to discover using standard field manipulations.

General Implications

A number of recent critiques of the current state of ecological research have

argued that ecology has become too focused on reductionist and hypothetical-

theoretical approaches, and have called for a greater emphasis on the search for

empirical generalities (Bradshaw 1987, Grubb 1988, Hall 1988, Keddy 1989, Grace

1991, Peters 1991, Weiner 1995, Grimm 1994, Brown 1995). This call for change

seems to stem from a number of different origins but among the most obvious would

be (1) the difficulty we face in attempting to predict responses of ecological systems

to environmental change, (2) frustration with theoretical approaches to the discovery

of general principles, (3) the desire to test competing evolutionary and ecological

theories convincingly, (4) an increasing recognition of the limitations of experimental

approaches (particularly for larger scale or longer term processes), and (5) the

intensifying need to apply ecological principles to conservation and restoration

efforts.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 162 A variety of approaches have been recommended for achieving more

empirically-based generalities. “Comparative ecology” (Bradshaw 1987), “predictive

ecology” (Peters 1991), and “macroecology” (Brown 1996) are all approaches that

have been proposed. These and other similar efforts have an essential element of

reliance on regression approaches to develop, evaluate, and test generalizations.

Within these applications, bivariate regressions and sometimes multivariate regression

techniques have proven highly effective in discovering, elucidating and exploring

broad general relationships.

Sustained efforts to explore general ecological phenomena, however, can be

expected to lead to multivariate examination of processes. While traditional multiple

regression approaches can be powerful, an infusion of more powerful multivariate

techniques from other disciplines has occurred in the last few years. A prime

example has been the increasing use of path analysis and the more general family of

methods known as structural equation modeling. Statistically, these methods provide

for a general and versatile suite of applications in multivariate analysis (e.g. partial

regression, latent variable analysis, covariance modeling, factor analysis).

Philosophically, however, these methods are often presented as means of theory

development and non-experimental hypothesis testing (“causal analysis,” see for

example Bagozzi 1980, Bentler 1980, Spirtes et al. 1993). A number of elegant

ecological applications (Maddox and Antonovics 1983, Johnson et al. 1991, Wesser

and Armbruster 1991, Mitchell 1992, Walker et al. 1994, Wootton 1994, Pugesek and

Tomer 1996) have indicated the potential of these approaches for our field.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 163 Structural equation modeling (which includes path analysis as a special case)

has a number of capabilities that can serve as advantages over traditional multiple

regression. Aside from these statistical advantages, however, its most prominent

feature is the testing and validation of hypothesized dependence models. Stated in

another way, structural equation models that have been tested and validated are

patterns of covariances or correlations that are presumed to represent some

underlying fundamental truth about how a system functions. While a great deal of

debate of this issue has taken place in other fields (e.g. Spirtes et al. 1993),

ecologists have yet to grapple fully with the question of the degree to which

structural equation models actually reflect system function. The single exception to

this is the elegant study by Wootton (1994) who utilized a limited set of

experimental manipulations to evaluate the predictability of a path analysis of

community interactions.

In this paper, an extensive set of experimental treatments involving nearly all

the major variables in the structural equation model of Grace and Pugesek (in press)

have been used to address the question, “to what degree does such a model

accurately reflect the underlying processes?” The results indicate that the SEM

formulated and evaluated in this paper accurately reflects a substantial set of

underlying processes by which environmental conditions control species richness. To

a great extent, when it is considered that the model was formulated for an entire

landscape yet tested at only two sites within that landscape, the degree of

correspondence between predicted and observed species richness values is

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 164

encouraging. At the same time, experimentation revealed a number of specific

underlying processes that are not currently represented in the model, particularly

interactions of herbivory with other variables. What is clear, however, is that SEM

can contribute to our ability to develop and assess hypotheses about ecological

systems, and that experimental tests can and should contribute to our evaluation and

refinement of such models.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 7

SUMMARY AND CONCLUSIONS

Attempts at understanding the factors that control species diversity in plant

communities frequently use natural history knowledge to develop a basic

understanding of those variables likely to influence plant survival and incorporate this

knowledge into experimental tests of hypotheses. The relative importance of factors

such as herbivory or competition in controlling species composition varies from

habitat to habitat, therefore studies must be designed accordingly.

The results of the work presented here describe several phenomena heretofore

uninvestigated. Competition in the Pearl River system does occur, but is mediated

by herbivory. When herbivores were present, species richness did not decline with

increasing soil fertility. However when herbivores were absent, species dropped out

of plots after three growing seasons. Vines were demonstrated to have a suppressive

effect on the growth of co-occurring herbaceous species in one area within this

coastal marsh complex. Surprisingly, vine removal from the previous growing

season allowed an initial increase in species richness which then declined by the end

of the summer due to increased biomass by the non-vine species. Thus the

community as a whole responded quickly to the removal, providing evidence of a

diffuse effect of competition from the vines.

As relative sea level rise begins to have more of an impact on Gulf Coast

marshes, flooding and salinity stress are predicted to increase. Other work has

demonstrated that both flooding and salinity may decrease species richness, as seen

165

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 166 in broad gradients across the area and in experimental tests. In this study, I found

that this was the case and that when these stresses were alleviated, richness

increased.

Herbivory provided one of the most interesting set of results. Alone,

herbivory caused no overall change in species richness after three growing seasons.

However, when combined with flooding, it caused a more severe loss in species.

When sods were transplanted into a fresh marsh, herbivores selectively consumed the

plants from the brackish marsh causing an overall decrease in richness.

The use of prediction models was also shown to be successful. Expanding a

multiple regression model into a structural equation model, I achieved an

understanding of the relative importance of environmental variables in determining

both biomass and richness throughout the Pearl River basin. I discovered that the

model is not good at predicting outside of its original parameter space, but that the

testing was generally successful overall. For future studies of this kind I would

recommend conducting the experimental tests at more than two locations where

sampling was made for the model. I would also recommend better coordination

between model design and experimental design so that factors such as herbivory

might be better accounted for in the model.

Further research in coastal habitats is greatly needed. We are only just

beginning to understand how herbivory, flooding, and salinity stress interact to

govern plant community structure in a healthy sediment-rich marsh such as the Pearl

River. In areas of degrading marsh throughout Louisiana, the interaction of these

/-*

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 167 factors may have quite different effects. However, examining interactions is difficult

logistically. Mammal exclosures require a great deal of effort to install and maintain.

Despite these difficulties, I recommend that manipulations conducted within these

marsh areas be replicated inside and outside such fences in order to understand how

the herbivores are affecting plant survival. It is difficult to analyze such experiments

without delving into complex multivariate statistical designs. This is one reason why

SEM is a good technique to use, since it combines a great deal of data to generate

several predictive regression equations.

The results reported here provide us with a great deal of information

regarding abiotic and biotic processes in the Pearl River marsh. Through models

which can be generalized, like the SEM model presented in Chapter 6, these

relationships can be tested in other marshes in coastal Louisiana. The practical

applications of these results are clear, although not easily addressed. My results

suggest that herbivores will play an important role in the response of coastal

communities to relative sea level rise. Nutria populations are already known to be a

problem and steps are being taken to control their numbers. These results give added

urgency to the issue of nutria eradication.

Through a combination of observations, descriptive data, experimental testing,

and statistical modelling, I have shown that the relationships between environmental

variables and species community structure and richness may be teased apart and

better understood. There is still much to be learned about the relative importance of

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 168 competition, herbivory, disturbance and stress on plant communities; this document

represents an attempt to further our understanding.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES

Aarssen, L.W. and E.A. Epp. 1990. Neighbor manipulations in natural vegetation: a review. Journal of Vegetation Science 1:13-30.

Abrams, P. A. 199S. Monotonic or unimodal diversity-productivity gradients: what does competition theory predict? Ecology 76:2019-2027.

Adams, D.A. 1963. Factors influencing vascular plant zonation in North Carolina salt marshes. Ecology 44:445-456.

Allen, E.B. and R.T.T. Forman. 1976. Plant species removals and old-field community structure and stability. Ecology 57:1233-1243.

Allen, J.A., S.R. Pezeshki and J.L. Chambers. 1996. Interaction of flooding and salinity stress on baldcypress (Taxodium distichum). Tree Physiology 16:307- 313.

Al-Mufti, M.M., C.L. Sydes, S.B. Furness, J.P. Grime, and S.R. Band. 1977. A quantitative analysis of shoot phenology and dominance in herbaceous vegetation. Journal of Ecology 65:759-791.

Anderson, V.J. and D.D. Briske. 1995. Herbivore-induced species replacement in grasslands: is it driven by herbivory tolerance or avoidance? Ecological Applications 5:1014-1024.

Armesto, J.J. and S.T.A. Pickett. 1986. Removal experiments to test mechanisms of plant succession in oldfields. Vegetatio 66:85-93.

Austin, M.P. 1987. Models for the analysis of species response to environmental gradients. Vegetatio 69:35-45.

Bagozzi, R.P. 1980. Causal Models in Marketing. John Wiley Publishers, New York.

Bakelaar, R. G. and E. P. Odum. 1978. Community and population level responses to fertilization in an old-field ecosystem. Ecology 59:660-665.

Bakker, J.P. 1985. The impact of grazing on plant communities, plant populations and soil conditions on salt marshes. Vegetatio 62:391-398.

169

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 170

Baldwin, A.H. 1996. The Role of Seed Banks, Disturbance, and Sea Level Rise in Determining the Plant Community Structure of Oligohaline Coastal Marshes. Ph.D. Dissertation, Louisiana State University, Baton Rouge.

Bandyopadhyay, B.K., S.R. Pezeshki, R.D. DeLaune and C.W. Lindau. 1993. Influence of soil oxidation-reduction potential and salinity on nutrition, N-15 uptake, and growth of Spartina patens. Wetlands 13:10-15.

Bazely, DR. and R.L. Jefferies. 1986. Changes in the composition and standing crop of salt-marsh communities in response to the removal of a grazer. Journal of Ecology 74:693-706.

Beadle, N.C.W. 1966. Soil phosphate and its role in molding segments of the Australian flora and vegetation, with special reference to xeromorphy and sclerophylly. Ecology 47:992-1007.

Belanger, L. and J. Bedard. 1994. Role of ice scouring and goose grubbing in marsh plant dynamics. Journal of Ecology 82:437-445.

Belsky, A.J. 1992. Effects of grazing, competition, disturbance and fire on species composition and diversity in grassland communities. Journal of Vegetation Science 3:187-200.

Bentler, P.M. 1980. Multivariate analysis with latent variables: causal modeling. Annual Review of Psychology 31:419-456.

Bertness, M.D. 1991a. Interspecific interactions among high marsh perennials in a New England salt marsh. Ecology 72:125-137.

Bertness, M.D. 1991b. Zonation of Spartina patens and Spartina altemiflora in a New England salt marsh. Ecology 72:138-148.

Bertness, M.D. and R. Callaway. 1994. Positive interactions in communities. Trends in Ecology and Evolution 9:191-193.

Bertness, M.D. and A.M. Ellison. 1987. Determinants of pattern in a New England salt marsh community. Ecological Monographs 57:129-147.

Bertness, M.D., L. Gough, and S.L. Shumway. 1992. Salt tolerances and the distribution of fugitive salt marsh plants. Ecology 73:1842-1851.

Bertness, M.D. and S.D. Hacker. 1994. Physical stress and positive associations among marsh plants. American Naturalist 144:363-372.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 171 Boesch, D.F., M.N. Josselyn, A.J. Mehta, J.T. Morris, W.K. Nuttle, C.A. Simensted, and D.J.P. Swift. 1994. Scientific assessment of coastal wetland loss, restoration and management in Louisiana. Journal of Coastal Research SI 20.

Bollen, K.A. 1989. Structural Equations with Latent Variables. John Wiley and Sons, New York.

Bond, W. 1983. On alpha diversity and the richness of the Cape flora: a study in southern Cape fynbos. In F.J. Kruger, D.T. Mitchell and J.U.M. Jarvis (eds.). Mediterranean-type Ecosystems: The Role of Nutrients. Springer-Verlag, Berlin, pp. 337-356.

Bradshaw, A.D. 1987. Comparison - its scope and limits. In I.H. Rorison, J.P. Grime, R. Hunt, G.A.F. Hendry, and D.H. Lewis (eds.). Frontiers of Comparative Plant Ecology. The New Phytologist 106(supplement). pp. 3- 22 .

Brewer, J.S. and J.B. Grace. 1990. Plant community structure in an oligohaline tidal marsh. Vegetatio 90:93-107.

Broome, S.W., LA. Mendelssohn, and K.L. McKee. 1995. Relative growth of Spartina patens (Ait.) Muhl. and Scirpus olneyi Gray occurring in a mixed stand as affected by salinity and flooding depth. Wetlands 15:20-30.

Brown, J.H. 1995. Macroecology. Chicago University Press, Chicago.

Campbell, B.D. and J.P. Grime. 1992. An experimental test of plant strategy theory. Ecology 73:15-29.

Carson, W.P. and S.T.A. Pickett. 1990. Role of resources and disturbance in the organization of an old-field plant community. Ecology 71:226-238.

Chabreck, R.H. 1972. Vegetation, water and soil characteristics of the Louisiana coastal region. Bulletin No. 664, Louisiana State Univeristy Agricultural Experiment Station, 75 pp.

Chabreck, R.H. 1988. Coastal Marshes. University of Minnesota Press, Minneapolis.

Collins, S.L., S.M. Glenn, and D.J. Gibson. 1995. Experimental analysis of intermediate disturbance and initial floristic composition: decoupling cause and effect. Ecology 76:486-492.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 172

Connell, J.H. 1978. Diversity in tropical rain forests and coral reefs. Science 199:1302-1310.

Connell, J.H. 1983. On the prevalence and relative importance of interspecific competition: evidence from field experiments. American Naturalist 122:661- 696.

Conner, W.H., C.E. Sasser, and N. Barker. 1986. Floristics of the Barataria Basin Wetlands, Louisiana. Castanea 51:111-128.

Crawley, M.J. 1983. Herbivory. University of California Press, Los Angeles.

Crozier, C.R., J.A. Nyman, and R.D. DeLaune. 1994. Symbiotic N fixation by Vigna luteola, a legume of low-salinity Gulf Coast marshes. Northeast Gulf Science 13:145-147.

Currie, D.J. 1991. Energy and large-scale patterns of animal- and plant- species richness. American Naturalist 137:27-49.

Currie, D.J. and V. Paquin. 1987. Large-scale biogeographical patterns of species richness of trees. Nature 329:326-327.

Cyr, H. and M.L. Pace. 1993. Magnitude and patterns of herbivory in aquatic and terrestrial ecosystems. Nature 361:148-150.

Day, R.T., P.A. Keddy, J. McNeill, and T. Carleton. 1988. Fertility and disturbance gradients: a summary model for riverine marsh vegetation. Ecology 69:1044- 1054.

DeLaune, R.D., S.R. Pezeshki, J.H. Pardue, J.H. Whitcomb, and W.H. Patrick Jr. 1990. Some influences of sediment addition to a deteriorating salt marsh in the Mississippi River deltaic plain: a pilot study. Journal of Coastal Research 6:181-188.

del Moral, R. 1983. Competition as a control mechanism in subalpine meadows. American Journal of Botany 70:232-245.

de Swart, E.O.A.M., A.G. van der Valk, K.J. Koehler, and A. Barendregt. 1994. Experimental evaluation of realized niche models for predicting responses of plant species to a change in environmental conditions. Journal of Vegetation Science 5:541-552.

Dillenburg, L.R., A.H. Teramura, I.N. Forseth, and D.F. Whigham. 1995. Photosynthetic and biomass allocation responses of Liquidambar styraciflua

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 173 (Hamamelidaceae) to vine competition. American Journal of Botany 82:454- 61.

Dillenburg, L.R., D.F. Whigham, A.H. Teramura, and I.N. Forseth. 1993a. Effects of below- and aboveground competition from the vines Lonicera japonica and Parthenocissa quinqefolia on the growth of the tree host Liquidambar styraciflua. Oecologia 93:48-54.

Dillenburg, L.R., D.F. Whigham, A.H. Teramura, and I.N. Forseth. 1993b. Effects of vine competition on availability of light, water, and nitrogen to a tree host {Liquidambar styraciflua). American Journal of Botany 80:244-252.

DiTommaso, A. and L.W. Aarssen. 1991. Effect of nutrient level on competition intensity in the field for three coexisting grass species. Journal of Vegetation Science 2:513-522.

Dix, R. and F. Smeins. 1967. The prairie, meadow, and marsh vegetation of Nelson County, North Dakota. Canadian Journal of Botany 45:21-58.

Dougherty, K.M., I.A. Mendelssohn, and F.J. Monteferrante. 1990. Effect of nitrogen, phosphorus, and potassium additions on plant biomass and soil nutrient content of a swale barrier strand community in Louisiana. Annals of Botany 66:265-271.

Ellison, A.M. and B.L. Bedford. 1995. Response of a wetland vascular plant community to disturbance: a simulation study. Ecological Applications 5:109- 123.

Feijtel, T.C., P.A. Moore, and K. McKee. 1989. Salinity and flooding level as determinants of soil solution composition and nutrient content in Panicum hemitomum. Plant and Soil 114:197-204.

Flynn, K.M., K.L. McKee, and I.A. Mendelssohn. 1995. Recovery of freshwater marsh vegetation after a saltwater intrusion event. Oecologia 103:63-72.

Ford, M.A. 1996. Impacts of Fire and Herbivory on Plant Characteristics and Soil Processes in a Coastal Marsh of Louisiana, USA. Ph.D. Dissertation, Louisiana State University, Baton Rouge.

Fretwell, S.D. 1977. The regulation of plant communities by food chains exploiting them. Perspectives on Biology and Medicine 20:169-185.

Friedland, A.J. and A.P. Smith. 1982. Effects of vines on successional herbs. American Midland Naturalist 108:402-403.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 174 Garcia, L.V., T. Maranon, A. Moreno, and L. Clemente. 1993. Above-ground biomass and species richness in a Mediterranean salt marsh. Journal of Vegetation Science 4:417-424.

Gaudet, C.L. and P.A. Keddy. 1988. A comparative approach to predicting competitive ability from plant traits. Nature 334:242-243.

Gaudet, C.L. and P.A. Keddy. 1995. Competitive performance and species distribution in shoreline plant communities: a comparative approach. Ecology 76:280-291.

Gentry, A.H. 1991. The distribution and evolution of climbing plants. In F.E. Putz and H.A. Mooney (eds.). The Biology of Vines. Cambridge University Press, New York. pp. 3-52.

Gleason, H.A. 1925. Species and area. Ecology 6:66-74.

Glenn, E., T.L. Thompson, R. Frye, J. Riley and D. Baumgartner. 1995. Effects of salinity on growth and evapotranspiration of Typha domingensis Pers. Aquatic Botany 52:75-91.

Godfrey, R.K. and J.W. Wooten. 1979. Aquatic and Wetland Plants of Southeastern United States: Monocotyledons. University of Georgia Press, Athens.

Godfrey, R.K. and J.W. Wooten. 1981. Aquatic and Wetland Plants of Southeastern United States: Dicotyledons. University of Georgia Press, Athens.

Goldberg, D.E. 1994. The influence of competition at the community level: an experimental version of the null models approach. Ecology 75:1503-1506.

Goldberg, D.E. and A.M. Barton. 1992. Patterns and consequences of interspecific competition in natural communities: a review of field experiments with plants. American Naturalist 139:771-801.

Goldberg, D.E., R. Turkington, and L. Olsvig-Whittaker. 1995. Quantifying the community level effects of competition. Folia Geobotanica Phytotaxonomica 30:231-242.

Goldberg, D.E. and P.A. Wemer. 1983. Equivalence of competitors in plant communities: a null hypothesis and a field experimental approach. American Journal of Botany 70:1098-1104.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 175 Gore, AJ.P. 1983. Ecosystems of the World 4B: Mire, Swamp, Bog, Fen and Moor. Elsevier Publishing, New York.

Gomitz, V. 1995. Sea-level rise: a review of recent past and near-future trends. Earth Surface Processes and Landforms 20:7-20.

Gosselink, J.B. 1984. The Ecology of Delta Marshes of Coastal Louisiana: A Community Profile. U.S. Fish and Wildlife Service, Biological Services, FWS/OBS-84/09, Washington, D.C.

Gough, L., J.B. Grace, and K.L. Taylor. 1994. The relationship between species richness and community biomass: the importance of environmental variables. Oikos 70:271-279.

Grace, J.B. 1988. The effects of nutrient additions on mixtures of Typha latifolia and Typha domingensis along a water-depth gradient. Aquatic Botany 31:83- 92.

Grace, J.B. 1989. Effects of water depth on Typha latifolia and Typha domingensis. American Journal of Botany 76:762-768.

Grace, J.B. 1990. On the relationship between plant traits and competitive ability. In J.B. Grace and D. Tilman (eds.). Perspectives on Plant Competition. Academic Press, New York. pp. 51-65.

Grace, J.B. 1991. A clarification of the debate between Grime and Tilman. Functional Ecology 5:583-587.

Grace, J.B. 1993. The effects of habitat productivity on competition intensity. Trends in Ecology and Evolution 8:229-230.

Grace, J.B. and M.A. Ford. 1996. The potential impact of herbivores on the susceptibility of the marsh plant Sagittaria lancifolia to saltwater intrusion in coastal wetlands. Estuaries 19:13-20.

Grace, J.B. and B.H. Pugesek. in press. A general model of plant species richness and its application to a coastal wetland. American Naturalist.

Grace, J.B. and R.G. Wetzel. 1981. Habitat partitioning and competitve displacement in cattails (Typha): experimental field studies. American Naturalist 118:463-474.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 176 Grillas, P., C. van Wijck and A. Bonis. 1993. The effect of salinity on the dominance-diversity relations of experimental coastal macrophyte communities. Journal of Vegetation Science 4:453-460.

Grime, J.P. 1973. Control of species density in herbaceous vegetation. Journal of Environmental Management 1:151-167.

Grime, J.P. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. American Naturalist 111:1169-1194.

Grime, J.P. 1979. Plant Strategies and Vegetation Processes. John Wiley, London.

Grime, J.P. 1980. An ecological approach to management. In I.H. Rorison and R. Hunt (eds.). Amenity Grassland: An Ecological Perspective. John Wiley, London, pp. 13-55.

Grime, J.P. 1985. Dominant and subordinate components of plant communities: implications for succesion, stability and diversity. In A. Gray, P. Edwards and A. Crawley (eds.). Colonisation, Succession and Stability. Blackwell, Oxford, pp. 413-428.

Grime, J.P. 1993. Stress, competition, resource dynamics and vegetation processes. In L. Fowden, T. Mansfield and J. Stoddart (eds.). Plant Adaptation to Environmental Stress. James and James, Guernsey, pp. 45-63.

Grimm, V. 1994. Mathematical models and understanding in ecology. Ecological Modelling 75:641-651.

Grover, J.P. 1995. Competition, herbivory, and enrichment: nutrient-based models for edible and inedible plants. American Naturalist 145:746-774.

Grubb, P.J. 1985. Plant populations and vegetation in relation to habitat, disturbance and competition: problems of generalization. In J. White (ed.). The Population Structure of Vegetation. Dr. W. Junk Publishers, Dordrecht, pp. 595-621.

Grubb, P.J. (ed.). 1988. Toward a More Exact Ecology. Blackwell Scientific, Oxford.

Guntenspergen, G.R., D.R. Cahoon, J. Grace, G.D. Steyer, S. Foumet, M.A. Townson, and A.L. Foote. 1995. Disturbance and recovery of the Louisiana coastal marsh landscape from the impacts of Hurricane Andrew. Journal of Coastal Research SI No. 21:324-339.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 177 Gurevitch, J., and R. S. Unnasch. 1989. Experimental removal of a dominant species at two levels of soil fertility. Canadian Journal of Botany 67:3470-3477.

Hair, J.F., Jr., R.E. Anderson, R.L. Tatham, and W.C. Black. 1995. Multivariate Data Analysis. 4th ed. Prentiss Hall, New York.

Hall, C.A.S. 1988. An assessment of the historically most influential theoretical models used in ecology and the data provided in their support. Ecological Modelling 43:5-31.

Harper, J.L. 1977. Population Biology of Plants. Academic Press, London.

Hayduk, L.A. 1996. LISREL: Issues, Debates, and Strategies. John Hopkins University Press, Baltimore.

Hegarty, E.E. 1991 Vine-host interactions. In F.E. Putz and H.A. Mooney (eds.). The Biology of Vines. Cambridge University Press, New York. pp. 357-376.

Hellings, S.E. and J.L. Gallagher. 1992. The effects of salinity and flooding on Phragmiies australis. Journal of Applied Ecology 29:41-49.

Herndon, A., L. Gunderson, and J. Stenberg. 1991. Sawgrass (Cladium jamaicense) survival in a regime of fire and flooding. Wetlands 11:17-28.

Hik, D.S., and R.L. Jefferies. 1990. Increases in the net above-ground primary production of a salt-marsh forage grass: a test of the predictions of the herbivore-optimization model. Journal of Ecology 78:180-195.

Howard, R.J. and I.A. Mendelsssohn. 1995. Effect of increased water depth on growth of a common perennial freshwater-intermediate marsh species in coastal Louisiana. Wetlands 15:82-91.

Huenneke, L.F., S.P. Hamburg, R. Koide, H.A. Mooney, and P.M. Vitousek. 1990. Effects of soil resources on plant invasion and community structure in California serpentine grassland. Ecology 71:478-491.

Huston, M.A. 1979. A general hypothesis of species diversity. American Naturalist 113:81-101.

Huston, M.A. 1980. Soil nutrients and tree species richness in Costa Rican forests. Journal of Biogeography 7:147-157.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 178 Huston, M.A. 1994. Biological Diversity: the Coexistence of Species on Changing Landscapes. Cambridge University Press, Cambridge.

Huston, M.A. and D.L. DeAngelis. 1994. Competition and coexistence: the effects of resource transport and supply rates. American Naturalist 144:954-977.

Huston, M.A. and T. Smith. 1987. Plant succession: life history and competition. American Naturalist 130:168-198.

Johnson, M.L., D.G. Huggins, and F. deNoyelles, Jr. 1991. Ecosystem modeling with LISREL: a new approach for measuring direct and indirect effects. Ecological Applications 4:383-398.

Johnson, S.R. 1994. Wind induced leaf binding by prairie cordgrass onto common milkweed in northeastern Kansas. Prairie Naturalist 26:301-305.

Jordan, T.E., D.F. Whigham, and D.L. Correll. 1990. Effects of nutrient and litter manipulations on the narrow-leaved cattail, Typha angustifolia L. Aquatic Botany 36:179-191.

Joreskog, K.G. and D. Sorbom. 1993. LISREL 8 User's Reference Guide. Scientific Software International, Chicago.

Keddy, P.A. 1989a. Competition. Chapman and Hall, New York.

Keddy, P.A. 1989b. Effects of competition from shrubs on herbaceous wetland plants: a 4-year field experiment. Canadian Journal of Botany 67:708-716.

Keddy, P.A. 1990. Competitive hierarchies and centrifugal organization in plant communities. In J.B. Grace and D. Tilman (eds.). Perspectives on Plant Competition. Academic Press, New York. pp. 266-290.

Keddy, P.A., L. Twolan-Strutt, and I.C. Wisheu. 1994. Competitive effect and response rankings in 20 wetland plants: are they consistent across three environments? Journal of Ecology 82:635-643.

Latham, P.J., L.G. Pearlstine, and W.M. Kitchens. 1994. Species association changes across a gradient of freshwater, oligohaline, and mesohaline tidal marshes along the Savannah River. Wetlands 14:174-183.

Lawes, J.B., J.H. Gilbert, and M.T. Masters. 1882. Agricultural, botanical and chemical results of experiments on the mixed herbage of a permanent meadow. II. The botanical results. Philosophical Transactions of the Royal Society 713:1181-1413.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 179 Lenssen, G.M., W.E. Van Duin, P. Jak and J. Rozema. 1995. The response of Aster tripolium and Puccinellia maritima to atmospheric carbon dioxide enrichment and their interactions with flooding and salinity. Aquatic Botany 50:181-192.

Llewellyn, D.W. and G.P. Shaffer. 1993. Marsh restoration in the presence of intense herbivory: the role of Justicia lanceolate (Chapm.) Small. Wetlands 13:176-184.

Louda, S.M., K.H. Keeler, and R.D. Holt. 1990. Herbivore influences on plant performance and competitive interactions. In J.B. Grace and D. Tilman (eds.). Perspectives on Plant Competition. Academic Press, New York. pp. 414-444.

Maddox, G.D. and J. Antonovics. 1983. Experimental ecological genetics in Plantago: a structural equation approach to fitness components in P. aristata and P. patagonica. Ecology 64:1092-1099.

Marrs, R.H., J.B. Grace, and L. Gough. 1996. On the relationship between plant species diversity and biomass: a comment on a paper by Gough, Grace & Taylor. Oikos 75:323-326.

McKee, K.L. and I.A. Mendelssohn. 1989. Response of a freshwater marsh plant community to increased salinity and increased water level. Aquatic Botany 34:301-316.

McNaughton, S.J. 1986. On plants and herbivores. American Naturalist 128:765- 770.

Mendelssohn, I. A. 1979. The influence of nitrogen level, form, and application method on the growth response of Spartina altemiflora in North Carolina. Estuaries 2:106-112.

Mendelssohn, I.A. and McKee, K.L. 1987. Experimental field and greenhouse verification of the influence of saltwater intrusion and submergence on marsh deterioration: mechanisms of action. In R.E. Turner and D.R. Cahoon (eds.). Causes of Wetland Loss in the Coastal Central Gulf of Mexico. Volume II: Technical Narrative. Final report to the Minerals Management Service, New Orleans, Louisiana. OCS Study/MMS 87-0120. pp. 8/1-8/34.

Mendelssohn, I.A., K.L. McKee, and W.H. Patrick. 1981. Oxygen deficiency in Spartina altemiflora roots: metabolic adaptation to anoxia. Science 214:439- 441.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 180 Menge, B.A. and J.P. Sutherland. 1976. Species diversity gradients: synthesis of the roles of predation, competition, and temporal heterogeneity. American Naturalist 110:351-369.

Milchunas, D.G., O.E. Sala, and W.K. Laurenroth. 1988. A generalized model of the effects of grazing by large herbivores on grassland community structure. American Naturalist 132:87-108.

Milchunas, D.G. and W.K. Laurenroth. 1993. Quantitative effects of grazing on vegetation and soils over a global range of environments. Ecological Monographs 63:327-366.

Milton, W.E.J. 1940. The effect of manuring, grazing and liming on yield, botanical and chemical composition of natural hill pastures. Journal of Ecology 28:326-356.

Milton. W.E.J. 1947. The yield, botanical and chemical composition of natural hill herbage under manuring, controlled grazing and hay conditions. I. Yield and botanical section. Journal of Ecology 35:65-89.

Mitchell, R.J. 1992. Testing evolutionary and ecological hypotheses using path analysis and structural equation modeling. Functional Ecology 6:123-129.

Mitsch, W.J. and J.G. Gosselink. 1986. Wetlands. Van Nostrand Reinhold, New York.

Moore, D.R.J. and P.A. Keddy. 1989. The relationship between species richness and standing crop in wetlands: the importance of scale. Vegetatio 79:99-106.

Moore, P.D. and S.B. Chapman. 1986. Methods in Plant Ecology. Blackwell Scientific, New York.

Moser, E.B., A.M. Saxton, and S.R. Pezeshki. 1990. Repeated measures analysis of variance: application to tree research. Canadian Journal of Forest Research 20:524-535.

Murphy, W.E. 1960. Ecological changes induced in moorland pastures by different fertilizer treatments. Proceedings of the International Grasslands Congress 8:86-89.

Myers, R.S., G.P. Shaffer and D.W. Llewellyn. 1995. Baldcypress (Taxodium distichum (L.) Rich.) restoration in southeast Louisiana: the relative effects of herbivory, flooding, competition, and macronutrients. Wetlands 15:141-148.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 181 Naidoo, G., K.L. McKee and I.A. Mendelssohn. 1992. Anatomical and metabolic responses to waterlogging and salinity in Spartina altemiflora and S. patens (Poaceae). American Journal of Botany 79:765-770.

Naidoo, G. and S.G. Mundree. 1993. Relationship between morphological and physiological responses to waterlogging and salinity in Sporobolus virginicus (L.) Kunth. Oecologia 93:360-366.

Nams, V.O., N.F.G. Folkard, and J.N.M. Smith. 1996. Nitrogen fertilization stimulates herbivory by snowshoe hares in the boreal forest. Canadian Journal of Zoology 74:196-199.

Neely, R.K. and C.B. Davis. 1985. Nitrogen and phosphorus fertilization of Sparganium eurycarpum and Typha glauca stands. I. Emergent plant production. Aquatic Botany 22:347-361.

Neill, C. 1990a. Effects of nutrients and water levels on emergent macrophyte biomass in a prairie marsh. Canadian Journal of Botany 68:1007-1014.

Neill, C. 1990b. Effects of nutrients and water levels on species composition in prairie whitetop (Scolochloa festucacea) marshes. Canadian Journal of Botany 68:1015-1020.

Neill, C. 1990c. Nutrient limitation of hardstem bulrush ( Scirpus acutus) in a Manitoba interlake region marsh. Wetlands 10:69-76.

Neter, J., W. Wasserman and M.H. Kutner. 1989. Applied Linear Regression Models. 2nd ed. Irwin Publishers, Boston.

Newman, E.I. 1973. Competition and diversity in herbaceous vegetation. Nature 244:310-311.

Nyman, J.A., R.H. Chabreck and N.W. Kinler. 1993. Some effects of herbivory and 30 years of weir management on emergent vegetation in brackish marsh. Wetlands 13:165-175.

Odum, W.E. 1988. Comparative ecology of tidal freshwater and salt marshes. Annual Review of Ecology and Systematics 19:147-176.

Oksanen, L. 1990. Predation, herbivory, and plant strategies along gradients of primary productivity. In J.B. Grace and D. Tilman (eds.). Perspectives on Plant Competition. Academic Press, New York. pp. 445-474.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 182 Oksanen, L., S.D. Fretwell, J. Arruda, and P. Niemala. 1981. Exploitation ecosystems in gradients of primary productivity. American Naturalist 118:240-261.

Palmer, M.W. 1994. Variation in species richness: towards a unification of hypotheses. Folia Geobotanica Phytotaxanomica 29:511-530.

Payonk, P.M. 1975. The response of three species of marsh macrophytes to artificial enrichment at Dulac, Louisiana. M.S. Thesis. Louisiana State University, Baton Rouge.

Peet, R.K., D.C. Glenn-Lewin and J.W. Wolf. 1983. Prediction of man's impact on plant species diversity. In W. Holzner, J.J.A. Werger and I. Ikusima (eds.). Man's Impact on Vegetation. Dr. W. Junk Publishers, Boston, pp. 41-54.

Penfound, W.T. 1974. The role of vines in plant communities. Advancing Frontiers of Plant Sciences 17:187-192.

Penfound, W.T. and E.S. Hathaway. 1938. Plant communities in the marshlands of southeastern Louisiana. Ecological Monographs 8:1-56.

Pennings, S.C. and R.M. Callaway. 1992. Salt marsh plant zonation: the relative importance of competition and physical factors. Ecology 73:681-690.

Peters, R.H. 1991. A Critique for Ecology. Cambridge University Press, Cambridge.

Pezeshki, S.R., R.D. DeLaune, and J.H. Pardue. 1992. Sediment addition enhances transpiration and growth of Spartina altemiflora in deteriorating Louisiana Gulf Coast salt marshes. Wetlands Ecology and Management 1:185-189.

Pickett, S.T.A. 1976. Succession: an evolutionary interpretation. American Naturalist 110:107-119.

Pinder, J.E., III. 1975. Effects of species removal on an old-field plant community. Ecology 56:747-751.

Puerto, A., M. Rico, M.D. Marias and J.A. Garcia. 1990. Variation in structure and diversity in Mediterranean grasslands related to trophic status and grazing intensity. Journal of Vegetation Science 1:445-452.

Pugesek, B.H. and A. Tomer. 1996. The Bumpus house sparrow data: a reanalysis using structural equation models. Evolutionary Ecology, in press.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 183 Putz, F.E. 1984. The natural history of lianas on Earro Colorado Island, Panama. Ecology 68:1713-1724.

Putz, F.E. and H.A. Mooney (eds). 1991. The Biology of Vines. Cambridge University Press, New York.

Radford, A.E., H.E. Ahles and C.R. Bell. 1968. Manual of the Vascular Flora of the Carolinas. University of North Carolina Press, Chapel Hill.

Rahman, M. and I.A. Ungar. 1994. The effect of competition and salinity on shoot growth and reporductive biomass of Echinocloa crus-galli. Aquatic Botany 48:343-353.

Reader, R.J. 1992. Herbivory as a confounding factor in an experiment measuring competition among plants. Ecology 73:373-376.

Reader, R.J., S.D. Wilson, B. Beisner, J.W. Belcher, I. Wisheu, P.A. Keddy, D. Tilman, C. Morris, J.B. Grace, J.B. McGraw, R. Turkington, E. Klein, Y. Leung, B. Shipley, R. van Julst, M.E. Johansson, C. Nilsson, J. Gurevitch, and K. Grigulis. 1994. Plant competition in relation to neighbor biomass: an intercontinental study with Poa pratensis. Ecology 75:1753-1760.

Rosenzweig, M.L. 1971. Paradox of enrichment: destabilization of exploitation ecosystems in ecological time. Science 171:385-387.

Rosenzweig, M.L. 1995. Species Diversity in Space and Time. Cambridge University Press, New York.

Rosenzweig, M.L. and Z. Abramsky. 1993. How are diversity and productivity related? In R. Ricklefs and S. Schluter (eds.). Species Diversity in Ecological Communities: Historical and Geographical Perspectives. University of Chicago Press, Chicago, pp. 52-65.

Sale, P.J.M. and R.G. Wetzel. 1983. Growth and metabolism of Typha species in relation to cutting treatments. Aquatic Botany 15:321-334.

Schoener, T.W. 1983. Field experiments on interspecific competition. American Naturalist 122:240-285.

Shaffer, G.P., C.E. Sasser, J.G. Gosselink, and M. Rejmanek. 1992. Vegetation dynamics in the emerging Atchafalaya Delta, Louisiana, USA. Journal of Ecology 80:677-687.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 184 Shipley, B., P.A. Keddy, C. Gaudet, and D.R.J. Moore. 1991. A model of species density in shoreline vegetation. Ecology 72:1093-1110.

Siccama, T.G., G. Weir, and K. Wallace. 1976. Ice damage in a mixed hardwood forest in Connecticut in relation to Vitis infestation. Bulletin of the Torrey Botanical Club 103:180-183.

Sih, A., P. Crowley, M. McPeek, J. Petranka and K Strohmeier. 1985. Predation, competition, and prey communities. Annual Review of Ecology and Systematics 16:269-311.

Silander, J.A. and J. Antonovics. 1982. Analysis of interspecific interactions in a coastal plant community - a perturbation approach. Nature 298:557-560.

Silvertown, J. 1980. The dynamics of a grassland ecosystem: botanical equilibrium in the Park Grass Experiment. Journal of Applied Ecology 17:491-504.

Smith, L.M. and J.A. Kadlec. 1985. Fire and herbivory in a Great Salt Lake marsh. Ecology 66:259-265.

Spirtes, P., C. Glymour, and R. Schemes. 1993. Causation, Prediction and Search. Springer-Verlag Lecture Notes in Statistics, Vol. 81.

Stevens, G.C. 1987. Lianas as structural parasites: the Bursera simaruba example. Ecology 68:77-81.

Swank, S.E., and W.C. Oechel. 1991. Interactions among the effects of herbivory, competition, and resource limitation on chaparral herbs. Ecology 72:104-115.

Taylor, D.R., L.W. Aarssen, and C. Loehle. 1990. On the relationship between r/K selection and environmental carrying capacity: a new habitat templet for plant life history strategies. Oikos 58:239-250.

Taylor, K.L. 1992. Effects of Vertebrate Herbivory on Louisiana Coastal Marshes. Ph.D. Dissertation, Louisiana State University, Baton Rouge.

Taylor, K.L., and J.B. Grace. 1995. The effects of vertebrate herbivory on plant community structure in the coastal marshes of the Pearl River, Louisiana, USA. Wetlands 15:68-73.

Taylor, K.L., J.B. Grace, G.R. Guntenspergen, and A.L. Foote. 1994. The interactive effects of herbivory and fire on an oligohaline marsh, Little Lake, Louisiana, USA. Wetlands 14:82-87.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 185

Thurston, J.M. 1969. The effect of liming and fertilizers on the botanical composition of permanent grassland, and on the yield of hay. In I. H. Rorison (ed.). Ecological Aspects of the Mineral Nutrition of Plants, Symposium of the British Ecological Society Edition. Blackwell Scientific Publications, Oxford, pp. 3-10.

Tilman, D. 1982. Resource Competition and Community Structure. Princeton University Press, Princeton.

Tilman, D. 1986. Resources, competition and the dynamics of plant communities. In M.J. Crawley (ed.). Plant Ecology. Blackwell Scientific Publications, Boston, pp. 51-75.

Tilman, D. 1987. Secondary succession and patterns of plant dominance along experimental nitrogen gradients. Ecological Monographs 57:189-214.

Tilman, D. 1988. Plant Strategies and the Dynamics and Structure of Plant Communities. Princeton University Press, Princeton.

Tilman, D. and S. Pacala. 1993. The maintenance of species richness in plant communities. In R.E. Ricklefs and D. Schluter (eds.). Species Diversity in Ecological Communities: Historical and Geographical Perspectives. University of Chicago Press, Chicago, pp. 13-25.

Tiner, R.W. 1993. Field Guide to Coastal Wetland Plants of the Southeastern United States. University of Massachusetts Press, Amherst.

Turner, R.E. and D.R. Cahoon (eds.). 1987. Causes of Wetland Loss in the Coastal Central Gulf of Mexico. Volume II. Minerals Management Service, New Orleans, Louisiana OCS Study/MMS 87-0120.

Twolan-Strutt, L. and P.A. Keddy. 1996. Above- and belowground competition intensity in two contrasting wetland plant communities. Ecology 77:259-270.

van de Koppel, J., J. Huisman, R. van der Wal, and H. Olff. 1996. Patterns of herbivory along a productivity gradient: an empirical and theoretical investigation. Ecology 77:736-745.

Vermeer, H.J.G. 1986. The effect of nutrients on shoot biomass and species composition of wetland and hayfield communities. Acta Oecologia/Oecologia Plantarum 7:145-155.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 186 Walker, M.D., P.J. Webber, E.H. Arnold, and D. Ebert-May. 1994. Effects of interannual climate variation on aboveground phytomass in alpine vegetation. Ecology 75:393-408.

Warren, R.S. and W.A. Niering. 1993. Vegetation change on a northeast tidal marsh: interaction of sea-level rise and marsh accretion. Ecology 74:96-103.

Weiner, J. 1995. On the practice of ecology. Journal of Ecology 83:153-158.

Weller, M.W. 1981. Freshwater Marshes. University of Minnesota Press, Minneapolis.

Wesser, S.D. and W.S. Armbruster. 1991. Species distribution controls across a forest-steppe transition: a causal model and experimental test. Ecological Monographs 61:323-342.

Wheeler, B.D. and K.E. Giller. 1982. Species richness of herbaceous fen vegetation in Broadland, Norfolk in relation to the quantity of aboveground material. Journal of Ecology 70:179-200.

Wheeler, B.D. and S.C. Shaw. 1991. Above-ground crop mass and species richness of the principal types of herbaceous rich fen vegetation of lowland England and Wales. Journal of Ecology 79:285-301.

Whigham, D. 1984. The influence of vines on the growth of Liquidambar styraciflua L. (sweetgum). Canadian Journal of Forest Research 14:37-39.

White, D. A. 1979. Plant Communities of the Lower Pearl River Basin, Louisiana. Ph.D. Dissertation. Tulane University, New Orleans, LA.

Whittaker, R.H. 1975. Communities and Ecosystems. 2nd ed. MacMillan, New York.

Whittaker, R.H. 1977. Evolution of species diversity in land communities. Evolutionary Biology 10:1-67.

Whittaker, R.H. and W.A. Niering. 1965. Vegetation of the Santa Catalina Mountains, Arizona, n. A gradient analysis of the south slope. Ecology 46:429-452.

Wilbur, H.M. 1972. Competition, predation and the structure of the Ambystoma- Rana sylvatica community. Ecology 57:3-21.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 187 Wilbur, H.M. and J.E. Fauth. 1990. Experimental aquatic food webs: interactions between two predators and two prey. American Naturalist 135:176-204.

Williams, E.D. 1978. The Botanical Composition of the Park Grass Plots at Rothamsted 1856-1976. Rothamsted Experimental Station, Harpenden.

Willis, A.J. 1963. Braunton Burrows: the effects on the vegetation of the addition of mineral nutrients to the dune soils. Journal of Ecology 51:353-374.

Wilsey, B.J., R.H. Chabreck and R.G. Linscombe. 1991. Variation in nutria diets in selected freshwater wetlands of Louisiana. Wetlands 11:263-278.

Wilson, S.D. and P.A. Keddy. 1986. Measuring diffuse competition along an environmental gradient: results from a shoreline plant community. American Naturalist 127:862-869.

Wilson, S.D. and P.A. Keddy. 1988. Species richness, survivorship, and biomass accumulation along an environmental gradient. Oikos 53:375-380.

Wilson, S.D. and D. Tilman. 1991. Components of plant competition along an experimental gradient of nitrogen availability. Ecology 72:1050-1065.

Wilson, S.D. and D. Tilman. 1993. Plant competition and resource availability in response to disturbance and fertilization. Ecology 74:599-611.

Wisheu, I.C. and P.A. Keddy. 1989. Species richness - standing crop relationships along four lakeshore gradients: constraints on the general model. Canadian Journal of Botany 67:1609-1617.

Wootton, J.T. 1993. Indirect effects and habitat use in an intertidal community: interaction chains and interaction modifications. American Naturalist 141:71- 89.

Wootton, J.T. 1994. Predicting direct and indirect effects: an integrated approach using experiments and path analysis. Ecology 75:151-165.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX

LETTER OF PERMISSION

188

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 189

L o u i s i a n a S t a t e U n i v e r s i t y c i Oecsrfr.en: Qt o-c«: Bioi^gy

M ay S. 1996

Dr Pehr Enckell. Managing Editor Oikos Department of Ecology Ecology Building S-22362 Lund Sw eden

Dear Dr Enckell

i am uruing to request written permission to use a paper 1 had published m (h k a \ in my Ph D dissertation at Louisiana State University The citattor. for the paper is

Gough. L . J B Grace, and K L Taylor. 1994 The relationship between species richness and community biomass the importance of environmental variables Oikos 70 271-279

In order to include tms paper as a chapter in my dissertation, i must have a letter from you allowing me to do so The LSU requirements state simply that as long as I am first author, me paper may be included If you could write a letter to grant me permission in the next month or so. 1 wouid be most appreciative, as I intend to complete my disseration by the middle of July

Thank you for your time

Sincerely yours.

Laura Gough Ph.D Candidate

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 190

Edoor-tfvChief: Professor Nds Maimer. Ph.D.

Etfttonal Office: Penr H. Enckell. PttO. Department ot Ecology Ecology Budding S*223 62 Lund. Sweden Lund Phone *’46 46 148188 Fax *46 46 104716 lOur ref. no. | ( P le a s e guae)

Ms Laura Gough Department of Plant Biology Louisiana State University 502 Life Sciences Building Baton Rouge, LA 70803-1705 USA Lund 13 May 1996

PERMISSION'

Permission is hereby granted for you to use your Oikos paper (The relationship between species richness and community biomass. Oikos 70:271-279.1994) in your Ph D dissertation.

Yours sincerely.

Pehr H Enckell Managing Editor Oikos

Swed»sn postal account (sostgrroi 62 25 32-7 Suescnpfcons Murnsgaa*: PC So* 2:45 8anker Skandmavtsfca ensb’ida Barken QK-1016 Copennagcn K Denman account 5668*33 005 63

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VITA

Laura Gough was bom in Houston, Texas, in 1969 to Lee Hackler Gough and

Michael Gough. Her brother Bany was bom in Houston in 1970. After she turned

three, her family moved to Stony Brook, New York where Laura wrote in a second

grade assignment that she would grow up to be "a ballerina or a biologist." In 1976

her family moved again, to Silver Spring, Maryland. The family moved once more

in Silver Spring and then to Bethesda, Maryland in 1979. Laura enjoyed math and

science classes and decided while in high school that she would either become a

doctor or an ecologist/environmental scientist. She graduated from Walt Whitman

High School in June 1986. That summer she participated in a Student Conservation

Association trip to the Great Smoky Mountains where she volunteered for the

National Park Service for five weeks.

In the fall of 1986 Laura began her undergraduate studies at Brown

University in Providence, Rhode Island. At the end of her junior year she became

involved with what would become her honors thesis research in a New England salt

marsh; this experience convinced her to pursue ecology as a career instead of

medicine. Laura received her Sc.B. in Biology with Honors from Brown in May

1990. From August 1990 until July 1991, Laura lived in Santa Cruz, California,

working for an environmental consulting firm as a word processing specialist and

occasional field technician. In the fall of 1991, Laura began her graduate study at

LSU. Five years later in the summer of 1996, she completed her dissertation on

191

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 192

Louisiana coastal marshes and begins her postdoctoral position in Woods Hole,

Massachusetts, studying plant species interactions in the Alaskan low arctic.

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DOCTORAL EXAMINATION AND DISSERTATION REPORT

Candidate: Laura Gough

Major Field: Plant Biology

Title of Dissertation: Plant Species Diversity and Community Structure in a Louisiana Coastal Marsh

Approved:

jor Professor and Chairman

Graduate School

EXAMINING COMMITTEE:

Date of Examination:

August 30, 1996____

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.