DIRECTOR OF THESIS __________________________________________________

Dr. Robert R. Christian

COMMITTEE MEMBER _________________________________________________

Dr. Mark M. Brinson

COMMITTEE MEMBER _________________________________________________

Dr. Claudia L. Jolls

COMMITTEE MEMBER _________________________________________________

Dr. Kevin O'Brien

CHAIR OF THE

DEPARTMENT OF BIOLOGY _____________________________________________

Dr. Charles E. Bland

DEAN OF THE

GRADUATE SCHOOL ___________________________________________________

Dr. Thomas L. Feldbush

Page

LIST OF TABLES......................................................................................................vii

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

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

2. LITERATURE REVIEW........................................................................................3

2.1. Overview of Salt Marshes.........................................................................3

2.1.1. Salt Marsh Development and Stability........................................4

2.1.2. Salt Marsh Hydrology................................................................6

2.1.3. Salt Marsh Zonation...................................................................7

2.2. Defining Stress and Disturbance................................................................9

2.3. Responses of a Salt Marsh Plant Community to Disturbances..................12

2.4. The Impact of a Rise in Sea Level on the Community Structure

of Salt Marshes....................................................................................13

2.4.1. Increased Sea Water Inundation................................................15

2.4.2. Wrack Deposition.....................................................................18

2.5. Plants Associated with a High Salt Marsh Community.............................20

2.5.1. Juncus roemerianus Scheele (Juncaceae)..................................21

2.5.2. Spartina patens (Ait.) Muhl. (Pocaceae)...................................23

2.5.3. Distichlis spicata Raf. (Poaceae)..............................................25

2.5.4. Spartina alterniflora Loisel. (Poaceae).....................................26

2.5.5. Salicornia spp. L. (Chenopodiaceae)........................................27

2.7. Major Points of Literature Review (Summary)........................................28

3. GOALS OF RESEARCH, POSTULATED MECHANISM, AND NULL HYPOTHESES..............................................................................................29

3.1. Postulated Mechanisms...........................................................................29

3.2. Null Hypotheses......................................................................................32

3.2.1. Effect of Increased Inundation on Saltmarsh Vegetation............32

3.2.2. Effect of Wrack Deposition on Saltmarsh Vegetation................32

3.2.3. Combination of Increased Inundation and Wrack Deposition....32

3.2.4 Comparison between the Reference Juncus roemerianus

Populations and the Experimental Controls..............................32

3.3.5. Shading Experiment..................................................................33

3.3. Statement Regarding the Testing of the Hypotheses................................33

4. MATERIALS AND METHODS...........................................................................34

4.1. Site Description......................................................................................34

4.2. Overall Design of Experiment.................................................................36

4.3. Tracking of the Growth and Senescence of Juncus roemerianus Leaves..41

4.3.1. Reference Population of Juncus roemerianus...........................41

4.4. Aboveground Biomass Sampling of the Plots..........................................43

4.5. Vegetative Analysis of Ground Cover in the Plots...................................44

4.6. Annual Net Primary Production (ANPP) of Juncus roemerianus.............44

4.7. Shading Experiment................................................................................46

4.8. Statistical Analysis of Data......................................................................47

5. RESULTS.............................................................................................................51

5.1. Growth and Senescence of Juncus roemerianus......................................51

5.1.1. Maximum Green and Total Heights of Juncus roemerianus......51

5.1.2. Life Table Analysis (Survivorship) of Juncus roemerianus........59

5.1.3. Juncus roemerianus Leaf Densities...........................................82

5.2. Aboveground Biomass............................................................................88

5.2.1. Juncus roemerianus -- 1994 and 1995 Data..............................88

5.2.2. Spartina patens -- 1994 and 1995 Data...................................103

5.2.3. Distichlis spicata -- 1994 and 1995 Data.................................106

5.2.4. Spartina alterniflora -- 1994 and 1995 Data............................108

5.2.5. Scirpus sp. -- 1994 and 1995 Data...........................................108

5.2.6. Salicornia sp. -- 1994 and 1995 Data......................................109

5.2.7. Total Biomass -- 1994 and 1995 Data.....................................109

5.2.8. Reference Biomass -- 1995 Data.............................................111

5.3. Vegetative Analysis of Ground Cover in the Plots..................................111

5.4. ANPP of Juncus roemerianus................................................................115

5.5. Shading Experiment...............................................................................125

6. DISCUSSION......................................................................................................131

6.1. Effect of Increased Inundation on Salt Marsh Vegetation........................135

6.1.1. Increased inundation does not affect the growth

of J. roemerianus....................................................................135

6.1.2. Increased inundation does not affect the senescence

of J. roemerianus....................................................................142

6.1.3. Increased inundation does not affect the growth of S. patens

and D. spicata.........................................................................144

6.2. Effect of Wrack Deposition.....................................................................148

6.2.1. Wrack deposition does not affect the aboveground production

of the plants.............................................................................148

6.2.2. Wrack deposition does not affect the vegetative regrowth of a

plant community.......................................................................151

6.3. Combination of Increased Inundation and Wrack Deposition...................157

6.3.1. Increased inundation and wrack deposition are not

interacting stressors..................................................................157

6.3.2. Increased inundation will not affect the vegetative regrowth

during the first growing season after wrack deposition.............158

6.4. Comparison between the Reference Juncus roemerianus Populations

and the Experimental Controls.............................................................160

6.5. Shading Experiment................................................................................161

7. CONCLUSIONS...................................................................................................163

8. APPENDIX A: DETAILS ON ELECTRICAL AND PUMPING SYSTEMS......167

9. APPENDIX B: SUMMARY OF Juncus roemerianus TAGGED LEAF DATA....170

10. APPENDIX C: AMOUNT OF WATER PUMPED ONTO THE PLOTS............184

11. APPENDIX D: SALINITY AND WELL WATER DEPTH DATA

FROM JUNE TO AUGUST 1995...................................................................187

12. LITERATURE CITED........................................................................................191

Page

1. Dates of when leaves where tagged and how many leaves where tagged..................42

2. Abbreviations used in Results section and an example of GLM table.......................52

3. Means (standard deviations) of maximum green and total heights for

J. roemerianus................................................................................................54

4. Results of GLM for maximum green and total heights............................................56

5. Results of GLM for reference population versus control plots for

maximum heights.............................................................................................58

6. Life table data for time to maximum green heights and time to death from

maximum green heights for 1994 cohorts........................................................60

7. Life table data for time to maximum total heights and time to death from

maximum total heights for 1994 cohorts..........................................................61

8. Life table data for time to maximum green heights for 1995 cohorts.......................62

9. Life table data for time to maximum total heights for 1995 cohorts........................64

10. Results of LIFEREG procedure for 1994 cohorts................................................67

11. Results of LIFEREG procedure for 1995 cohorts................................................68

12. Results of Chi-square tests for 1994 cohorts.......................................................78

13. Results of Chi-square tests for 1995 cohorts........................................................79

14. Results of LIFEREG procedure for reference populations versus

the control plots..............................................................................................81

15. Results of Chi-square tests for reference populations versus the control plots......83

16. J. roemerianus leaf densities from the experimental plots.....................................84

17. Results of GLM for leaf densities from the experimental plots.............................85

18. J. roemerianus leaf densities from the reference populations and the 1995

nonwrack areas of the control plots................................................................86

19. Results of GLM for reference population leaf densities versus control plots.........87

20. Means (standard deviations) of 1994 biomass samples.........................................89

21. Means (standard deviations) of 1995 biomass samples.........................................91

22. GLM means by flooding and wrack treatments for 1994 biomass samples...........93

23. GLM means by flooding and wrack treatments for 1995 biomass samples...........95

24. Results of GLM for 1994 biomass samples.........................................................97

25. Results of GLM for 1995 biomass samples.........................................................99

26. Significant interactions from the GLM for 1994 and 1995 biomass samples.......101

27. Means (standard deviations) of 1995 reference biomass samples and 1995

nonwrack areas of the control plots biomass samples....................................112

28. Results for GLM reference populations biomass samples versus nonwrack

areas of the control plots biomass samples....................................................113

29. Means (standard deviations) of the percentage of ground cover change from

1994 to 1995 in the experimental plots by flooding and wrack treatments......114

30. Results of GLM for percentage of ground cover change....................................116

31. Means (standard deviations) of frequency, average height, and production-to-

biomass ratio for J. roemerianus...................................................................117

32. Means (standard deviations) of the ANPP for J. roemerianus.............................119

33. GLM means for frequency, average height, production-to-biomass, and

ANPP by flooding and wrack treatments.......................................................122

34. Results of GLM for frequency, average height, production-to-biomass,

and ANPP.....................................................................................................123

35. Means (standard deviations) of frequency, average height, production-to-

biomass, and ANPP for reference populations................................................126

36. Results of GLM for frequency, average height, production-to-biomass, and

ANPP for reference populations versus the control plots................................127

37. Means (standard deviations) for biomass samples from the

shading experiment........................................................................................128

38. Results of GLM for biomass from the shading experiment...................................130

39. Summary of the data collected and statistics used for this thesis..........................132

40. List of biomass and ANPP from other studies and this thesis...............................137

41. Means (standard deviations) of the green heights in the flooded plots for

each sampling date..........................................................................................171

42. Means (standard deviations) of the green heights in the border control plots for

each sampling date..........................................................................................173

43. Means (standard deviations) of the green heights in the control plots for each

sampling date..................................................................................................175

44. Means (standard deviations) of the total heights in the flooded plots for

each sampling date..........................................................................................177

45. Means (standard deviations) of the total heights in the border control plots for

each sampling date..........................................................................................179

46. Means (standard deviations) of the total heights in the control plots for each

sampling date..................................................................................................181

47. Means (standard deviations) of the green and total heights in the reference

populations for each sampling date.................................................................183

48. Pumping activities for 1995.................................................................................185

49. Results of GLM on well water depth and salinity from the experimental plots......190

Page

1. Map of the Virginia Coast Reserve Long-Term Ecological Research

(VCR LTER) site.............................................................................................35

2. Map of Brownsville marsh showing the location of the experimental plots and

the reference J. roemerianus populations.........................................................37

3. Design of the experimental plots.............................................................................38

4. Survivorship curves for time to maximum green height for 1994 J. roemerianus

leaves based on flooding treatment...................................................................70

5. Survivorship curves for time to death from maximum green height for 1994

J. roemerianus leaves based on flooding treatment...........................................71

6. Survivorship curves for time to maximum green height for 1995 J. roemerianus

leaves based on flooding treatment...................................................................72

7. Survivorship curves for time to maximum green height for 1995 J. roemerianus

leaves based on cohort number.........................................................................73

8. Survivorship curves for time to maximum total height for 1994 J. roemerianus

leaves based on flooding treatment....................................................................74

9. Survivorship curves for to time to death from maximum total heights for 1994

J. roemerianus leaves based on flooding treatment............................................75

10. Survivorship curves for the time to maximum total height for 1995

J. roemerianus leaves based on flooding treatment............................................76

11. Survivorship curves for the time to maximum total height for 1995

J. roemerianus leaves based on cohort number..................................................77

12. Depth of water below the ground surface in the experimental plots

from June to August 1995...............................................................................188

13. Ground water salinity in the experimental plots from June to August 1995...........189

I could not have accomplished this research without the help, support, and financial assistance of many people. First, I would thank my committee members, Bob, Mark, Claudia, and Kevin. To my advisor, Bob -- Thank-you for getting me "hooked" on research back in the summer of 1993, providing me with the opportunity to work on a thesis project that was perfectly suited for me, and for a copy of Cold Comfort Farm. To Mark -- Thank-you for allowing me to plunder through your enormous selection of papers and for making sure taking a class with you meant I was able to spend lots of time out in the field. To Claudia -- Thank-you for getting me "hooked" even more on research, for always being thorough with your questions and comments, and for allowing me to spend a week at "Bug Camp." To Kevin -- Thank-you for spending all those hours guiding me through the world of statistics and SAS with humor and patience, and I'm sure I would have ended up very lost without your help. To all my committee members -- Thank-you ever so much for being understanding and coming through for me time and time again. It has been my pleasure and honor to have worked with such a wonderful group of people.

I also would like to thank all the people who helped me in the field, in the lab, and with various other aspects of this thesis: ECU Interlibrary Loan Service people, R. Carlson, C. Christian, D. Christian, J. Contable, D. Daniel, G. Dickerson, L. (Peito) Gatti, S. Hamze, J. Holmes, T. Jones, D. Krovetz, J. Porter, D. Richardson, M. Spaulding, R. Stumpf, L. Stasavich, R. Tarnowski, J. Taylor, B. Tolley, J. Tolley, W. Tolley, J. Tomkins, S. Walston, Wetland Ecology Class of 1995, and D. Yozzo. Out of this long list, there are a few people who deserve an extra thanks for putting in overtime: Tammy Jones for spending many hours helping process biomass samples; Jed Tomkins for assisting me in the field during the summer of 1995; my father, Walt Tolley, for assistance with the pump and for building an assortment of items for me; and Dave Yozzo for being my full-time scribe, field partner, and reviewer.

Last, but certainly not least, I would like to thank my financial backers. Funding for this thesis came from the National Science Foundation's Long-Term Ecological Research and Research Experiences for Undergraduates programs, East Carolina University Department of Biology, and my parents.

Patricia M. Tolley. EFFECTS OF INCREASED INUNDATION AND WRACK DEPOSITION ON A SALTMARSH PLANT COMMUNITY. (Under the direction of Dr. Robert R. Christian) Department of Biology, June 1996.

Sea-level rise would increase the inundation frequency of high salt marshes. Furthermore, a rise in sea level may cause an increase in the amount of wrack deposited in a high marsh community. Since changes in sea level occur slowly, communities may also change gradually, in response to increased inundation. Therefore, abrupt disturbances, such as storm induced deposition of Spartina alterniflora wrack, may be necessary to induce rapid changes in saltmarsh plant communities. Previous ecological studies in this area have generally not focused on more than one type of disturbance. Therefore, there is a lack of knowledge on how disturbances interact with each other in eliciting community responses. Furthermore, few studies in salt marshes have investigated the impact of abrupt disturbances which result in change of community structure.

I experimentally manipulated inundation and wrack deposition to discern the individual and combined effects of these factors on high marsh plant species (i.e., Juncus roemerianus, Spartina patens, and Distichlis spicata) at a mainland marsh on the Eastern Shore of Virginia. This two year study involved (1) the experimental pumping of salt water onto the high marsh, (2) experimental placement of S. alterniflora wrack on the plant community, and (3) removing half of the wrack after six months of deposition. The experimental design was a randomized block that involved three flooding treatments (flooded, border control, and control), two wrack treatments (wrack and nonwrack), and three blocks (X, Y, and Z). The data collected from the experimental plots included (1) the tracking of the growth and senescence of J. roemerianus leaves, (2) estimations of growing and senescing J. roemerianus leaf densities, (3) quantification of aboveground biomass (g/m²), (4) vegetative analysis of the change in ground cover, and (5) estimation of the annual net primary productivity (ANPP, g/(m² x y)) of J. roemerianus. Furthermore, I monitored two additional J. roemerianus populations that were inundated naturally on a daily basis and compared them to the experimental control plots to study the effects of greater, natural inundation on J. roemerianus. I also conducted an experiment to study the effects of shading on the regrowth of J. roemerianus.

The effect of increased inundation on the plant community was species specific. Increased inundation did not significantly affect the aboveground biomass and ANPP of J. roemerianus, but it did result in slower growing and faster senescing J. roemerianus leaves. Furthermore, the reference J. roemerianus populations, which have undergone prolonged flooding (> 2 years), were significantly different from the control plots. Therefore, the growth and senescence rates of J. roemerianus are more sensitive to changes in inundation than aboveground biomass and ANPP; although aboveground biomass and ANPP probably decline with long-term inundation. Increased frequency of inundation decreased S. patens aboveground biomass, but ponding of water increased S. patens aboveground biomass. Increased inundation alone did not cause a significant effect on the aboveground biomass of D. spicata and did not promote the growth of minor species such as short S. alterniflora.

The effects of wrack deposition in reducing the aboveground biomass of the plant community were non-species specific, while regrowth and colonization after wrack deposition were species specific. Six months of wrack deposition significantly reduced the aboveground biomass of the three dominant species. One year after the removal of wrack, the aboveground biomass and leaf densities of J. roemerianus were still significantly lower in the post-wrack areas as compared to the nonwrack areas. In contrast, the aboveground biomass values for S. patens and D. spicata were not significantly lower in the post-wrack areas as compared to the nonwrack areas. Furthermore, living D. spicata biomass was significantly higher in the post-wrack areas. D. spicata also colonized the bare areas where J. roemerianus previously grew.

The interaction between the responses to increased inundation and wrack deposition occurs through time. During periods of deposition, the extreme effect of the wrack on the plants masked any interaction between flooding and deposition. However, after the removal of wrack, increased inundation inhibited D. spicata, a primary colonizing species, from colonizing some of the bare areas. When increased inundation inhibits the colonization of post-wrack areas, peat accumulation may decrease and erosion may increase in the bare areas. This would result in microtopographic changes that may eventually lead to the formation of marsh potholes.

Lastly, shading did not have a significant effect on the regrowth of J. roemerianus after clipping. However, shading inhibited the regrowth of species other than J. roemerianus. This sensitivity of other species to shading may be one reason why J. roemerianus stands are highly monospecific. The possibility of J. roemerianus being able to outcompete other high marsh species via shading deserves further study.

1. INTRODUCTION

The environmental conditions of a salt marsh affect the productivity and survival of halophytes. Chapman (1938) composed a list of factors that regulate the existence of plant species within a marsh. They included such things as tidal patterns, salinity, drainage, water table, precipitation, soil type, evaporation, temperature, and biota. Species survive in the areas of a marsh where they are physically adapted to these various environmental factors (Day et al. 1989).

Because salt marshes are coastal ecosystems, sea-level rise strongly affects them (Parkinson 1994). Thus, a significant rise in sea level would alter the species distribution and abundance, and the resultant community structure of a marsh (Oertel et al. 1992). The increased inundation of the high marsh community would result in longer periods of waterlogged soils which decreases the productivity of some plants (Bertness 1991b, Nyman et al. 1993). Furthermore, plants associated with the low marsh community may invade the high marsh (Adams 1963, Warren & Niering 1993).

Since sea-level rise is occurring slowly, communities may be able to resist the structural changes associated with the gradual increase in inundation. Some of the ways a community is able to resist change is through the accumulation of organic matter, baffle-like stems and leaves, and erosion-resistant roots and rhizomes. Since communities may be able to resist changes, disturbances more abrupt than rising sea level may be necessary to cause changes in high marsh communities (Brinson et al. 1995). An example of an abrupt disturbance is a storm depositing wrack onto a high marsh (Hayden et al. 1992, Brinson et al. 1995). Wrack deposition causes severe vegetation diebacks within the high marsh community (Reidenbaugh et al. 1983, Hartman 1988, Knowles et al. 1991). This type of disturbance creates open areas for new or neighboring species to invade. Distichlis spicata (L.) Greene (salt grass), Spartina alterniflora Loisel. (smooth cordgrass), and Salicornia europaea L. (glasswort) dominate the recolonization of bare soil patches regardless of what species were present originally (Reidenbaugh & Banta 1980, Bertness & Ellison 1987, Hartman 1988, Knowles et al. 1991).

Few studies focus on more than one type of disturbance. Therefore, there is a lack of knowledge on how systems respond to interacting disturbances such as increased inundation and wrack deposition (Lugo 1978, Turner 1988). Furthermore, few studies have investigated the impact of abrupt disturbances which result in a change in saltmarsh community structure (Brinson et al. 1995). This lack of knowledge about how combined disturbances affect the community structure of salt marshes lead to the topic of this thesis. The focus of my thesis research was to study the effects of increased inundation and wrack deposition, singularly and in combination, on the plants of a high marsh community.2. LITERATURE REVIEW

2.1. Overview of Salt Marshes

Intertidal salt marshes, inundated and drained by astronomical tides, are located along temperate coasts. Some salt marshes are among the most productive ecosystems in the world. Primarily, flowering halophytes dominate the vascular plant community of intertidal salt marshes; usually one or a few plants dominate (Mitsch and Gosselink 1993). The physical appearance of salt marshes differs among regions. Along the southeastern Atlantic and Gulf of Mexico coasts, marshes tend to have dense and robust vegetation. Higher-latitude coastal areas generally contain shorter and sparse vegetation (Day et al. 1989, Mitsch and Gosselink 1993).

One noticeable characteristic of salt marshes is the presence of vegetative zonation. This pattern is similar from one marsh to another. Day et al. (1989) cited three spatial scales on which zonation occurs: (1) latitudinal zonation, (2) zonation at the level of the coastal drainage basin, and (3) local zonation. Latitudinal zonation is the result of primary climatic differences. Zonation at the level of the coastal drainage basin is the result of the distance of the vegetation from the tidal creek. With this gradient of inundation, there also is a salinity gradient. Local zonation patterns are the result of local elevation, drainage, soil type, and sources of freshwater and saltwater. These factors affect the growth and reproduction of the different species of marsh plants. Different species vary in their tolerances to degree and duration of inundation and to salinity (Day et al. 1989).

Variations in abiotic factors along a topographic gradient lead to development of distinctive low elevation and high elevation plant communities. Monospecific stands of S. alterniflora often characterize the low marsh community along the Atlantic coast. The soil elevation in this community is low enough to allow for inundation on every high tide. The soil is typically mineral and has little to no peat accumulation. In contrast, Spartina patens (Aiton) Muhl. (saltmarsh hay) and D. spicata or short S. alterniflora typically dominate the high marsh community of Atlantic coast marshes. The high marsh community also may contain monospecific stands of Juncus roemerianus Scheele (black needlerush) or Juncus gerardi. Other taxa such as Salicornia spp. may be present in low numbers. The high marsh community is not inundated as frequently as the low marsh community because it is farther away from the tidal creek and often at a higher elevation. The soil of the high marsh often contains a thick layer of peat (Mitsch & Gosselink 1993).

2.1.1. Salt Marsh Development and Stability

Redfield (1967) cited several interacting factors that influence the development of a prograding salt marsh in New England. These factors included tidal range, physiology of the plants that produce peat in relation to tide levels, sedimentation rates on open tidal flats and within the stands of plants, and the relative change of sea level. Plants that can withstand limited submergence colonize barren areas located at the upper elevational limit for tidal inundation within a marsh. In New England salt marshes, these areas contain S. patens, D. spicata, short S. alterniflora, and a few other minor species. When these plant species die, they add to the peat layers of the high marsh. Tall S. alterniflora grows from the mid-intertidal to low intertidal areas of a marsh high. Mineral sediment accumulates within the low marsh stands of S. alterniflora. This accumulation of sediment eventually will reach the high-water level and become overgrown with high saltmarsh plants. Then organic peat from the death of high saltmarsh plants covers the sediment (Redfield 1967). This simplified outline of events has been the driving theory behind salt marsh development studies. However, Redfield's prograding marsh hypothesis is limited given salt marshes have a larger range of developmental patterns (Stevenson et al. 1986, Brinson et al. 1995).

Brinson et al. (1995) outlined four different patterns to explain how coastal marshes may respond to changes in sea level and sedimentation rates. The patterns were developed based on the slope of the marsh from the edge of the estuary to the terrestrial forest (gentle versus steep) and on the sediment supply (high versus low). The first pattern is a marsh with a gentle slope and a high sediment input, and is what Redfield (1967) described for New England marshes. This first pattern of community response hypothesizes the high marsh would migrate overland into the terrestrial forest. At the same time, the low marsh would be prograding into the estuary (Brinson et al. 1995). The second pattern involves a marsh with a gentle slope and a low sediment input. With these conditions, the high marsh would invade the upland forest while erosion occurred in the low marsh. Several marshes along the Mississippi Delta would fall into this second category. The other two patterns involve a marsh with a steep slope that would prevent the overland migration of the marsh into the upland forest. The marsh edge along the estuarine may be either prograding or eroding depending on the sediment supply. These four different patterns represent the extreme points along a gradient of possible patterns.

Elevational gradients, degree of tidal inundation, and sedimentation rate control the type of marsh environment that will develop for plant colonization. The rate of sea-level rise and sediment accumulation on the surfaces influence the maintenance of marshes (Oertel & Woo 1994). Marshes lining major estuarine systems (i.e., Hudson, Delaware, and Chesapeake estuaries) receive relatively small amounts of sediments from tidal waters. Therefore, the marshes are unable to keep pace with the relative rise in sea level and become submerged. Preservation of these marshes is possible by invasion of the marsh community into upland communities (Oertel & Woo 1994, Brinson et al. 1995). Continued inundation will cause changes in marsh topography and plant community structure (Oertel et al. 1992). However, long-term shifts in plant communities are not predictable, which makes the prediction of successional patterns difficult (Clark 1986, Nuttle et al., in prep).

2.1.2. Salt Marsh Hydrology

Hydrology plays an important role in determining the structure and productivity of coastal ecosystems. Inundation is important for the input of nutrients (Harvey & Odum 1990), the exchange of particulate organic matter (Hemminga et al. 1993), and the addition and removal of salts from the sediment (Hackney & de la Cruz 1978, Nuttle & Harvey 1995). A rise in sea level would alter the hydrology of a marsh by increasing the inundation of higher elevations and possibly raising the water table. By raising the water table, surface runoff will increase, as would duration and amount of saturated areas (Nuttle & Portnoy 1992).

The alteration of the flooding and drainage regimes of a marsh influences primary productivity (Parrondo et al. 1978, Linthurst & Seneca 1980, Mendelson & Seneca 1980, Zedler et al. 1980, Linthurst & Seneca 1981, Wiegert et al. 1983, Seneca et al. 1985, DeLaune et al. 1987, McKee & Patrick 1988, Morris & Haskin 1990, De Leeuw et al. 1990, Strakosch 1992, Reed & Cahoon 1992, Osgood & Zieman 1993). An increase in nutrients and a decrease in salinity may promote an increase in productivity(Parrondo et al. 1978, Linthurst & Seneca 1981, Osgood & Zieman 1993). A decrease in productivity may be the result of prolonged anoxic conditions caused by waterlogging (Mendelssohn & Seneca 1980, Linthurst & Seneca 1981, Weigert et al. 1983, Strakosch 1992). However, the positive increase in productivity associated with increased inundation is not consistent among studies. Some showed an increase in productivity(Strakosch 1992), while others showed a decrease (Zedler et al. 1980, DeLaune et al. 1987, Strakosch 1992, Reed & Cahoon 1992), or no change in productivity (De Leeuw et al. 1990).

2.1.3. Salt Marsh Zonation

Several interacting environmental factors influence the distribution of plants within a salt marsh. These factors include tides, salinity, drainage, aeration, water table, rainfall, soils, evaporation, temperature, and biota (Chapman 1938). Of these factors, tides are the primary factors influencing the distribution of plants within a marsh (Chapman 1938). Because of this idea, many researchers have studied the correlation between tidal inundation and plant distribution within a marsh. The frequency and amplitude of tides and elevation were determined to be the primary factors controlling the distribution of plant species in several North Carolina salt marshes (Adams 1963). Eleuterius & Eleuterius (1979) studied the frequency and duration of tidal inundation and exposure in relation to the dramatic change in vegetation in Davis Bay, Mississippi salt marsh. The change in plant species was from a monotypic zone of S. alterniflora to a monotypic zone of J. roemerianus. Inundation occurred more frequently in the S. alterniflora zone, which was located at a lower elevation than the J. roemerianus zone. Even though parts of these two different zones had different tidal relationships, the abrupt change in plant species did not correlate to the frequency or the duration of inundation. Eleuterius & Eleuterius (1979) suggested that factors other than tides were involved in determining the distribution of plant species.

Research on plant zonation conducted in a New England marsh involved factors associated with inundation such as salinity and substrate redox potential (Bertness & Ellison 1987). In New England salt marshes, S. alterniflora dominates the low marsh community. S. patens and J. gerardi dominate the high marsh (Bertness & Ellison 1987, Lefor et al. 1987, Bertness 1991a) and short S. alterniflora in some marshes (Redfield 1967). Each zone has different characteristic physical factors that are related to tidal inundation. Salinity and substrate redox potential decrease with increasing marsh elevation (Bertness & Ellison 1987). Furthermore, competitive displacement restricted S. alterniflora to the low marsh (Bertness 1991b). In contrast, Bertness (1991b) hypothesized that the restriction of S. patens to the high marsh is because of its limited aerenchyma tissue development.

Interspecific competition also may play an important role in determining plant zonation. Removal and transplant experiments done in a New England high marsh demonstrated that J. gerardi dominated S. patens and D. spicata. Furthermore, S. patens outcompeted D. spicata. However, this is not the case in bare soil patches. D. spicata and S. patens colonize bare soil patches, which are often hypersaline, before J. gerardi is able to grow in these areas. D. spicata and S. patens are more tolerant of high saline conditions then J. gerardi. Furthermore, the colonizing species appear to be able to lower the salinity of the bare soil, therefore, facilitating the growth of J. gerardi into the previously bare areas (Bertness 1991a).

2.2. Defining Stress and Disturbance

Ecologists widely apply the concept of stress to describe patterns occurring in different ecosystems (Lugo 1978). Lugo (1978) defined a stressor as "any condition or situation that causes a system to mobilize its resources and increase its energy expenditure." A stress is the response of an ecosystem to the stressor. An acute stressor is one that stresses an ecosystem for a short period time and the ecosystem is able to recover. Continuous or chronic stressors occur for long periods of time within an ecosystem (Lugo 1978, White 1979). Many organisms are able to adapt and survive within a stressed ecosystem. However, a major result of adaptation to a stressor is a decline in metabolic rates (Parsons 1989). Furthermore, organisms adapted to living in stressful environments are continuously overcoming the stress. For example, salt marsh plants are always dealing with the problems of inundation and salinity even though they may be adapted to survive flooding and salinity (Lugo 1978).

Some stressors are a normal part of an ecosystem while others are infrequent, acute events. Both normal and infrequent stressors cause a loss of energy in the system. Organisms within an ecosystem are usually able to adapt to normal stressors but are not as well adapted to infrequent stressors. Therefore, the effects of normal stressors are not as immediately obvious as infrequent stressors (Lugo 1978).

The effects a stressor has on an ecosystem are the result of several factors. These factors include such things as (1) the intensity of the stressor, (2) multiplicative (effect is greater than the sum of the individual stressors) or additive (effect equal to the sum of the individual stressors) effects on the ecosystem, (3) frequency, (4) ecosystem type, (5) condition of the ecosystem, and (6) the intensities, residual effects, and frequency of other stressors at the time of occurrence (Lugo 1978, Turner 1988). Chronic stressors are usually more harsh on an ecosystem than acute stressors. However, acute stressors still may be very damaging depending on their intensity, timing, and the ability of the ecosystem to recover. Furthermore, if more than one stressor occurs at the same time, the impact may be either additive or multiplicative and cause even more of an energy drainage on the system (Lugo 1978, Turner 1988).

The response of an ecosystem to a stressor may be either positive or negative. An increase in growth and net productivity may occur after some stresses. Disruptive stressors, such as tides and storms, help to remobilize resources within a system and may bring in new resources (Lugo 1978, Odum 1980). Furthermore, species diversity may increase after a disturbance. For example, a storm event may decrease salinity stress and allow for less saline tolerant plant species to invade an area. When normal salinity conditions return, the species diversity of the area will then decrease and return to pre-storm levels (Lugo 1978).

Disturbance is a physical stressor that is external to the community and causes a change in the community structure. A perturbation is the effect of a disturbance (Fahrig 1990). Most researchers do not make the distinction between disturbance and perturbation. Instead researchers tend to use the word "disturbance" for both the cause and the effect. Normally this practice does not cause difficulties (Fahrig 1990).

Traditionally, the term disturbance has been limited to describing infrequent, abrupt events that cause immediate structural changes in communities (White 1979). However, this approach is limiting since disturbances, as with other types of stressors, occur along the gradient of minor to major events (Lugo 1978, White 1979). Furthermore, the biotic structure of the community may promote some types of disturbances. For example, the severity of a fire depends not only on the source of the fire but also the amount and combustibility of organic matter within the community (White 1979).

Although physical factors may stimulate disturbances, the community type may determine the occurrences and the effects of disturbances (White 1979). The types of disturbances affecting a community may vary regionally and within a community as a function of the area's geomorphology and other local variables. Examples of the types of natural disturbances that affect vegetation include the following: fire, wind, ice, cryogenic soil movement, temperature changes, precipitation variability, alluvial processes, coastal processes, dune movement, saltwater inundation, landslides, lava flows, and biotic disturbances (White 1979). Two disturbances commonly associated with salt marshes are saltwater inundation and wrack deposition. I will discuss how these two types of disturbances relate to a rise in sea level in detail later.

2.3. Responses of a Salt Marsh Plant Community to Disturbances

Some disturbances create open areas that may allow for colonization by new or neighboring species. Revegetation often begins in the first growing season after the disturbance. Reidenbaugh and Banta (1980) documented new growth of S. alterniflora on bare soil patches within two years after the disturbance of wrack deposition. Two other species associated with recently disturbed sites are D. spicata and Salicornia europaea. D. spicata is a rapid colonizer, invading areas with vegetative runners. In contrast, S. europaea is a seed colonizer (Ellison 1987, Hartman 1988). However, in a New England marsh, the slower growing S. patens and J. gerardi eventually replaced the colonizing species (Bertness and Ellison 1987). In an irregularly flooded J. roemerianus marsh (Cedar Island, North Carolina), S. patens and other species excluded J. roemerianus from being able to grow in bare patches after a disturbance. Furthermore, J. roemerianus seeds were unable to germinate in open areas (1) with salinities greater than 20 ppt., (2) which were regularly flooded, or (3) which had reduced soils (Knowles et al. 1991).

Few studies focus on more than one type of disturbance (Turner 1988). However, Turner (1988) conducted a study involving multiple disturbances. She experimentally examined the combined effects of clipping, trampling, and fire on the productivity of S. alterniflora. Clipping combined with either trampling or fire reduced the net aboveground primary productivity (NAPP) of S. alterniflora more than predicted. The effect of the combination of trampling and fire was additive on S. alterniflora NAPP. Therefore not all combinations of disturbances produce additive responses. Because of this fact, it is important to understand how disturbances interact in order to predict how a community will respond to disturbances (Turner 1988).

2.4. The Effects of a Rise in Sea Level on the Community Structure of Salt Marshes

Because coastal ecosystems are located at the land-sea ecotone, they are strongly affected by changes in sea level (Parkinson 1994). There are several different ways a salt marsh may respond to a rise in sea level. An important consideration is the marsh's ability to maintain surface elevations relative to the mean high water level. Important factors that determine surface elevations include local submergence rates, sedimentation rates, density and composition of plants, hydrogeomorphic setting, and the type and intensity of human modifications. Furthermore, vertical accretion rates vary both spatially and temporally in marshes. This variation makes it difficult to predict how a marsh will respond to sea-level rise (Kearney et al. 1994).

Since sea-level rise is occurring slowly over a long period of time, rising sea level alone does not alter community structure. Changes in community structure occur with events more abrupt than rising sea level. Altered ecosystem states develop new self-maintaining properties, facilitate plant species replacement, and change sediment conditions (Brinson et al. 1995). For example, patterns of accretion and erosion in one Virginia salt marsh caused the marsh to change from a juvenile foreshore marsh to a creek-drained marsh (Reidenbaugh et al. 1983). Wrack deposition is an abrupt event that causes severe vegetation diebacks and erosion at higher elevations (Reidenbaugh et al. 1983).

The factors that determine and maintain the composition of plant species from estuaries to upland forests differ among the different zones in a marsh (Hayden et al. 1994, Brinson et al. 1995). Brinson et al. (1995) stated "Transitions between states are facilitated by disturbance or exposure to acute stress." The effects of rising sea level can be seen in the following mechanisms for the high marsh: (1) increased levels of brackish water at higher elevations, (2) tidal creek erosion of the marsh surface as it encroaches further into the marsh, and (3) increased water depth which causes increased waterlogging and anoxic conditions. Increased levels of brackish water and waterlogging result in osmotic stress and sulfide toxicity within anoxic conditions in the sediments. Tidal creek erosion of the high marsh causes greater drainage near the edge of the marsh and the movement and suspension of sediments. At lower elevations other mechanisms act to effect change (Brinson et al. 1995).

Currently, there are few studies that have investigated the effects of abrupt events which result in the change of the community structure of coastal marshes (Brinson et al. 1995). The lack of knowledge about ecological and geological processes hinders future predictions of the effect of sea level on coastal ecosystems (Brinson & Moorhead 1989). My work focuses on two processes in the high marsh community that are associated with an increase in mean sea level. These two processes are increased seawater inundation and wrack deposition.

2.4.1 Increased Seawater Inundation

A rise in sea level would cause the erosion of tidal creeks and an increase in the frequency of flooding at higher elevations (Hayden et al. 1994). As a result of the altered inundation patterns, salt marsh plants previously associated with more frequently inundated lower elevations may invade higher elevations (Adams 1963). Furthermore, increased inundation may result in longer periods of waterlogged soils if increased drainage does not occur. The anoxic stress caused by waterlogging may result in a decrease of aboveground primary production by macrophytes. A reduction in aboveground plant growth as a result of increased inundation as been suggested to be one cause of the loss of land in the Terrebonne Basin, Louisiana (Nyman et al. 1993). Plants prevent the loss of land by trapping fine-grain sediments (fine silt and clay) on marsh surfaces (Stumpf 1983, Nyman et al. 1993) and by accumulating organic matter (Moorhead & Brinson 1995). The trapping of sediments and the accumulation of organic matter aid in the vertical accretion of the marsh. A decrease in plant productivity results in limited vertical accretion, which causes an increase in flooding and even greater decreases in primary production (Nyman et al. 1993). A study with S. alterniflora demonstrated this decrease in productivity as a result of standing water. The productivity of S. alterniflora grown in 30 cm to 0 cm of standing water ranged from 0 to 1144 g/m², respectively (Linthurst & Seneca 1980).

The change in vegetation zonation in Wequetequock-Pawcatuck (Connecticut) tidal marshes over the last forty years has been linked to differential rates of marsh accretion and sea-level rise (Warren & Niering 1993). In the late 1940's, the marshes supported a J. gerardi-S. patens belting pattern characteristic of many New England salt marshes. The areas in most marshes that were once dominated by J. gerardi are now forbs, primarily dominated by Triglochin maritima. The zones previously dominated by S. patens are now composed of a mixture of short S. alterniflora, D. spicata, forbs, and relic stands of S. patens. The surfaces of the marshes are at lower elevations than they were in the 1940's relative to sea-level. The lower elevations caused greater frequency and duration of tidal inundation on the marshes. The change in tidal patterns also resulted in increased peat saturation, salinity, and sulfide concentrations, and a decrease in redox potential. These abiotic and biogeochemical responses to increased inundation initiated a change in plant species composition to one that favored wetter, more open vegetation types dominated by short S. alterniflora and forbs (Warren & Niering 1993).

The effects of increased surface flooding, increased subsurface drainage, and natural hydrology were studied experimentally for three years in a New Jersey salt marsh (Strakosch 1992). The study site was in an area dominated by short S. alterniflora. Although S. alterniflora remained dominant in all treatments, species richness decreased with flooding and increased with drainage relative to the controls. An increase in prolonged waterlogged conditions was presumed to be the main reason for the decrease in species richness. Several different but interrelated factors could have caused the increase in species richness in the drainage treatments -- removal of waterlogging, soil conditions, microtopography, seed source, and the adaptive ability of halophytes. Increased flooding did not have a significant effect on S. alterniflora stem density or belowground biomass. In contrast, increased drainage caused a decrease in stem density and belowground biomass. Increased inundation in areas of short S. alterniflora did not result in the production of tall S. alterniflora. However, the increased inundation did produce a plant community structure similar to the low marsh.

2.4.2. Wrack Deposition

Tidal wrack is composed of dead plant material that forms mats that are carried and deposited onto the marsh surface by winds, tides, and currents (Reidenbaugh and Banta 1980). In Atlantic coast salt marshes, tall S. alterniflora composes most wrack mats; but in some areas, there is a high percentage of Zostera marina present (Reidenbaugh and Banta 1980). In Gulf Coast marshes, dead J. roemerianus leaves compose the wrack mats (Stumpf, personal communication). Standing dead stems of plants break from the marsh surface and float together in mats. In cold climates, ice shearing may enhance this process. When high tides, storms, or wind deposit wrack in the marsh, the vegetation under the wrack dies as a result of compaction and smothering (Reidenbaugh and Banta 1980, Hartman 1984, Knowles et al. 1991). This killing of plants creates barren patches of soil, referred to as "secondary" bare soil since they were previously vegetated (Chapman 1940). A rise in sea level may result in increased disturbance of the high marsh by wrack deposition and intrusion of wrack farther into the marsh (Brinson et al. 1995).

Wrack is often created during the late fall and winter and deposited along the edge of the estuary. Then incoming tides carry the wrack into the marsh. Winter and spring storms produce higher tides that push the wrack farther into the marsh. Wind will bend S. alterniflora stalks toward the upland thus permitting wrack to move over the edge of the estuary and preventing it from being carried back to the estuary when the water recedes (Reidenbaugh and Banta 1980).

S. alterniflora wrack killed large amounts of aboveground vegetation in several studies (Reidenbaugh and Banta 1980, Hartman 1988, Knowles et al. 1991). Reidenbaugh and Banta (1980) found wrack caused complete devegetation in a low marsh on the Delmarva Peninsula during the spring. Furthermore, they found only partial devegetation in the transition zone between the low and high marsh, and partial or complete diebacks in the high marsh. However, the amount of devegetation by tidal wrack varied greatly from year to year depending on the tidal levels in the early spring. The higher the tidal amplitude in the spring, the more severe the dieback recorded. The degree of devegetation caused by wrack also depended on the elevation of the marsh and the time of year the deposition occurred. The most severe diebacks occurred in the low marsh during the spring (Reidenbaugh and Banta 1980).

A three year study of the recolonization of bare soil areas caused by wrack was conducted at the Great Sippewissett salt marsh (Cape Cod, Massachusetts). Hartman (1988) studied the presence of colonizers and abiotic factors in order to determine the causes of the rates and patterns of recolonization. S. alterniflora dominated the regrowth of the bare areas via vegetative expansion. Some colonization of the patches did occur from seed germination of Salicornia spp. The only other colonizer was D. spicata, but it was of minor importance compared to S. alterniflora. This is interesting because S. patens and D. spicata dominated the areas surrounding the open patches with S. alterniflora contributing little to the total biomass of the community. Salinity, sulfide concentrations, and ammonium concentrations from disturbed and undisturbed sites were not significantly different and could not explain the rates and patterns of recolonization. Therefore, the proximity of S. alterniflora and its ability for recolonization by vegetative expansion controlled the rate of recolonization of bare areas (Hartman 1988).

Knowles et al. (1991) concluded disturbance by wrack was the main cause of the mosaic vegetation patterns on Cedar Island (North Carolina), an irregular flooded marsh dominated by J. roemerianus. Mortality rates of plants under experimentally placed wrack were high. Light exclusion, leaf compression, and elevated sulfide levels in the soil were the main causes of the plants' death. Recolonization of the bare soil areas varied based on the location of the bare area. Bare areas that were located close to the edge of the island, and therefore more likely to be inundated, were colonized by D. spicata or persisted as algal pannes. A range of species other than J. roemerianus colonized the bare areas farther inland. These species included S. patens, D. spicata, Fimbristylis spadicea, and Panicum virgatum. Seedlings of J. roemerianus and several annuals were found colonizing bare patches located on elevated hummocks. Once bare patches were colonized by such species as D. spicata and S. patens, J. roemerianus slowly regrew back into these areas so that patch borders remained for decades (Knowles et al. 1991).

2.5. Plants Associated with a High Salt Marsh Community

2.5.1. Juncus roemerianus Scheele (Juncaceae), needlerush or black needlerush

J. roemerianus generally grows in areas of infrequent inundation, such as high marshes, along the Atlantic coast from Virginia to Florida and along the coast of the Gulf of Mexico, between the latitudes of 26o N and 42o N (de la Cruz & Gabriel 1974, Eleuterius 1976). However, sparse populations occur as far north as New Jersey, Long Island, New York, Connecticut, and as far south as Lauguna Madre, Texas, Mexico, and on Caribbean Islands (Eleuterius 1976). J. roemerianus forms patches of monospecific stands that may dominate a marsh or be within marshes where other species dominate. For example along the Gulf Coast, these J. roemerianus stands usually occur in areas where the tidal amplitude is rarely more than 1 m. J. roemerianus grows with S. patens, D. spicata, Spartina cynosuroides, and Scirpus americanus (de la Cruz & Gabriel 1974).

J. roemerianus leaves are tough, long, cylindrical (3-5 mm in diameter), slightly tapered, and terminate with extremely, sharply pointed tips, thus its common name. J. roemerianus grows continuously throughout the year. New growth occurs in the basal region from perennial underground rhizomes (Foster 1968, Seibert 1969, Williams & Murdoch 1972, Anderson 1974, Eleuterius 1975, Giurgevich & Dunn 1978, Snogerup 1993). The maximum rates of growth for J. roemerianus in Mississippi occur from May through July (de la Cruz and Gabriel 1974) and in North Carolina from May to June (Foster 1968). Annual net primary productivity (ANPP) and aboveground biomass of J. roemerianus have been estimated for several different marshes. In North Carolina marshes, ANPP has been estimated between 560 and 1038 g/(m² x y) (Foster 1968, Murdoch & Williams 1972, Blum et al. 1978, Christian et al. 1990). ANPP for J. roemerianus in one Georgia marsh was from 1500 to 2800 g/(m² x y) (Gallagher et al. 1980). The aboveground biomass of J. roemerianus in Mississippi marshes ranges from 400 g/m² to 2000 g/m² (Eleuterius 1972). Total aboveground J. roemerianus biomass in North Carolina marshes ranges from 35 g/m² to 2294 g/m² with standing dead leaves making up most of the biomass (Murdoch & Williams 1972, Cooper & Waits 1973, Blum et al. 1978, Christian et al. 1990, Hook 1991). The large range of numbers reported for marshes located within the same geographic area may be in part the results of differences in the methods used to estimate productivity (Kaswadji et al. 1990).

The gynodioecious flowers of J. roemerianus are unusual for both Juncaceae and salt marsh angiosperms where flowers are normally only perfect. J. roemerianus has both perfect and pistillate flowers but no stamimate flowers. The timing of flowering varies throughout the range of J. roemerianus (Eleuterius 1975). For J. roemerianus populations on the eastern Atlantic, flowering occurs from May until October (Radford et al. 1968). Sexual reproduction may be important in the colonization of new areas while vegetative reproduction is the driving force in maintaining an established population (Eleuterius 1975).

The ionic and anoxic stresses of a salt marsh affect the survival and productivity of J. roemerianus. Ionic stress, caused by high salinities, and anoxic stress, caused by waterlogged soils, have a negative effect on the growth of J. roemerianus (Eleuterius 1984, 1989a & b). In addition to reducing the growth of J. roemerianus, ionic toxicity reduces the number of flowers and seeds produced (Eleuterius 1989b). One adaptation J. roemerianus has to combat ionic stress is the ability to grow deep penetrating specialized roots that go down to areas of lower salinities (Eleuterius 1984). J. roemerianus does not have the ability to secrete salts or block the uptake of ions (Anderson 1974). To counterbalance anoxic stress, J. roemerianus develops aerenchyma (Anderson 1974). Furthermore, one species of Juncus, J. gerardi, is able to actively oxidize the sediments around the rhizomes and roots (Hacker & Bertness 1995). J. roemerianus also may be able to oxidize its rhizomes and roots.

2.5.2. Spartina patens (Aiton) Muhl. (Poaceae), saltmarsh hay, saltmeadow cordgrass, saltmeadow grass

The geographic range of S. patens is greater than the range of J. roemerianus with populations occurring from New England to Florida marshes along the Atlantic coast and in marshes along the Gulf of Mexico. The aboveground biomass of S. patens in a North Carolina marsh was estimated between 230 and 440 g/m² (Cooper & Waits 1973) and 430 g/m² for a New England marsh (Nixon & Oviatt 1993) where this species is dominant. Hook (1991) estimated the aboveground biomass for S. patens to range from 3.0 to 21 g/m² within a J. roemerianus dominated marsh in North Carolina.

Several factors probably interact to influence the primary production of S. patens and limit it to the high marsh communities (Burdick et al. 1989). From recent research with S. patens populations, Bertness (1991b) suggested that this species is restricted to the high marsh because of limited aerenchyma development. Anoxic conditions decrease the growth of S. patens (Smart & Barko 1978). A reduction in soil oxygen levels can initially reduce the photosynthetic activity of S. patens. However, S. patens is able to recover this loss of photosynthetic activity (DeLaune et al. 1990). Furthermore, anoxic conditions stimulate the production of alcohol dehydrogenase (ADH) in S. patens. However, hydrogen sulfide, which accumulates in anoxic soils, inhibits ADH in some plants. The decrease of ADH activity results in lower total adenine nucleotides, adenylate energy charge ratio (AEC), nitrogen uptake, and growth (Koch et al. 1990).

Ionic stress also has a negative effect on the growth of S. patens (Smart & Barko 1978, Pezeshki et al. 1987, Pezeshki & DeLaune 1993, Bandyopadhyay et al. 1993, Broome et al. 1995). The experimental increase of soil salinity from 0 ppt. to 22 ppt. caused a 54 % reduction in the stomatal conductance and a 43 % reduction in net photosynthesis. Furthermore, salt excretion from leaves occurred within three to five days after the salinity was increased (Pezeshki et al. 1987). When tested over a range of salinities, soil hypoxia was determined to be the most important factor controlling nutrient uptake in S. patens. Therefore in submerging coastal ecosystems, increased inundation and decreased soil redox potential are the primary stress factors causing the displacement of S. patens unless salinity levels are extremely high (Bandyopadhyay et al. 1993, Broome et al. 1995).

S. patens has developed some adaptations to ionic stress and anoxic conditions. First, S. patens has the ability to secrete salt out of its leaves through salt glands (Anderson 1974, Queen 1974, Broome et al. 1995) and through its roots (Broome et al. 1995). Furthermore, S. patens is selective in its ion uptake and is able to actively exclude certain ions (Smart and Barko 1978). S. patens does have aerenchyma in its roots (Anderson 1974, Burdick & Mendelssohn 1987, Cooke et al. 1993) though its development is limited (Gleason & Zieman 1981, Bertness 1991b).

2.5.3. Distichlis spicata Raf. (Poaceae), saltgrass or spike grass

The geographic range of D. spicata is similar to the range of S. patens with D. spicata growing in marshes along the Atlantic and Gulf coasts. D. spicata is often found in association with S. patens. The only estimate of aboveground biomass I could find for D. spicata was for two North Carolina marshes where D. spicata ranged from 190 to 240 g/m² when it was a dominant species (Cooper & Waits 1973) and 0.2 g/m² within J. roemerianus stands (Hook 1991).

Ionic stress and anoxic conditions have a negative effect on the growth of D. spicata (Smart and Barko 1978). However, D. spicata does have several adaptations for growing and surviving in a salt marsh. It is able to secrete salt (Anderson 1974, Liphschitz & Waisel 1982), is capable of ion exclusion, and has selective ion uptake (Smart & Barko 1978, 1980). D. spicata also is able to increase its succulence (Flower 1985). Under low nitrogen conditions, D. spicata will decrease its aboveground production and increase its belowground production (Smart & Barko 1980). It is often one of the first plants to colonize bare soil patches (Hansen et al. 1976, Hartman 1988, Bertness 1991a, Knowles et al. 1991).

2.5.4. Spartina alterniflora Loisel. (Poaceae), smooth cordgrass

S. alterniflora has been the most widely studied marsh plant. While the tall form of S. alterniflora dominates the low marsh, a short form of S. alterniflora often occurs in the high marsh. The presence of other plants excludes S. alterniflora from the high marsh (Bertness 1988). One of the differences between short S. alterniflora and tall S. alterniflora, besides their location within a marsh, is the amount of nitrogen present within the plants. The short form tends to have less nitrogen and a lower nitrogen metabolism than the tall form (Mendelssohn 1979). High sulfide concentrations (King et al. 1980), low sediment oxidation (Linthurst 1980, King et al. 1982, Howes et al. 1986), and high salinities (Morris & Haskin 1990, Osgood & Zieman 1993) may limit nitrogen uptake by S. alterniflora. The result of a decrease in nitrogen uptake is a reduction in aboveground productivity.

Furthermore, an increase in the peat density of soil may cause a decrease in aboveground production (Bertness 1988). Tall S. alterniflora plants collected from a silt-clay area and a sandy area of the same creek had similar biomasses. Differences in plant size did exist with the silt-clay soil having taller and more robust plants than plants collected from the sandy soil. This shows the importance of substrate type on the allocation of resources (Christian et al. 1983).

Aboveground productivity of S. alterniflora increases with subsurface drainage (Mendelssohn & Seneca 1980, Wiegert et al. 1983, Strakosch 1992), and stagnant conditions decrease productivity (Linthurst & Seneca 1980, Mendelssohn & Seneca 1980, Strakosch 1992). S. alterniflora grown in a greenhouse with aerobic soil had an average of 6.3 times higher biomass than plants grown in an anoxic substrate (Linthurst 1980). In a Louisiana marsh, researchers found a positive correlation between marsh elevation and S. alterniflora growth (Reed & Cahoon 1992).

2.5.5. Salicornia spp. L. (Chenopodiaceae), glasswort

Salicornia spp. are annuals and are associated with high salinity and disturbed areas (Ellison 1987). Salicornia europaea is able to colonize bare patches in salt marshes through the germination of seeds (Ellison 1987, Ungar 1987, Hartman 1988). Provided other species do not invade the patch, S. europaea is able to maintain its productivity. However, other marsh species will rapidly outcompete S. europaea and eventually overgrow the patches if salinity is low (Ellison 1987). Its main adaptation to ionic stress is increased succulence (Flower 1985). However, soil salinity levels do control the growth and survival of Salicornia spp. The main response to high salinity stress is a reduction in plant growth. Mortality occurs at extreme levels of salinities. Density-dependent factors can cause high mortality in some populations (Jefferies et al. 1981) but not in others (Ungar 1987).

2.6. Major Points of Literature Review (Summary)

Several major points were made in this introduction which relate to the theme of this thesis. First, salt marshes have characteristic zonation of macrophytes based on the elevation of the marsh and the frequency of inundation. A rise in sea level could alter the distribution of plants in a marsh. Furthermore, wrack deposition also can alter the plant composition of a salt marsh. However, it is not known whether or not the combination of inundation and wrack would accelerate the change in macrophyte composition.

3. GOALS OF RESEARCH, POSTULATED MECHANISMS, AND NULL HYPOTHESES

The main goal of this investigation was to determine the effects of experimental tidal inundation and wrack deposition on the plants of an organic high marsh. Specifically, I investigated the effects of increased inundation and wrack deposition, individually and in combination, on aboveground growth, senescence, and biomass accumulation and ground cover change after an experimental disturbance. The experiment was set-up as a random block design that contained both J. roemerianus dominated areas and S. patens and D. spicata dominated areas. Estuarine saltwater was pumped onto a third of the plots during the growing seasons. Wrack was placed onto half of each plot and was partially removed after six months. Furthermore, a reference population of J. roemerianus leaves was monitored for comparison with the experimental control. A shading experiment also was set-up to determine the effects of taller J. roemerianus leaves on shorter, younger leaves. The following paragraphs contain the postulated mechanisms for this thesis. The specific null hypotheses are listed at the end of this section.

3.1. Postulated Mechanisms

Christian et al. (1990) and Hook (1991) found no overall significant correlation between production of J. roemerianus along a gradient of hydroperiods and salinity in an irregularly flooded marsh. Nuttle and Harvey (1995) reported flooding in tidal marshes helps to remove salts from marsh soils during periods of reduced evaporation. If the pumping of water onto the plots is able to remove salts from the soil, therefore, reducing the salinity, the growth of J. roemerianus may be greater in the inundated plots than in the border control and control plots. However, since J. roemerianus does not normally grow in regularly flooded areas, a decrease in the production of J. roemerianus in the flooded plots may occur. Furthermore, the inundation may actually increase the salinity, which has been shown to inhibit the growth of J. roemerianus (Eleuterius 1984) and the growth of J. gerardi (Bertness et al. 1992).

Hook (1991) found disturbed sites that receive a greater amount of inundation have a greater susceptibility to decreased growth and increased mortality. Furthermore, J. roemerianus seeds are only able to germinate in bare patches if the areas have salinities less than 20 ppt., are not regularly flooded, and do not have reduced soils (Knowles et al. 1991). Therefore, J. roemerianus regrowth and production in post-wrack areas that are being inundated may be lower than post-wrack areas that are not being flooded.

S. patens and D. spicata are not usually found in regularly flooded areas of salt marshes (Niering & Warren 1980, Bertness 1991a). S. patens appears to be restricted from the low marsh habitat because of unsuitable physical conditions and probably not because of competition with S. alterniflora (Bertness 1991b). Therefore, increased inundation may reduce the aboveground biomass of S. patens and D. spicata in this experiment.

S. patens and D. spicata are two of the dominant colonizing species in disturbed areas in New England salt marshes (Hartman 1988, Bertness & Ellison 1987, Bertness & Shumway 1993). S. patens and D. spicata may competitively exclude J. roemerianus from being able to regrow in patches created by disturbance (Knowles et al. 1991). Furthermore, D. spicata is more tolerant of wrack burial than other plant species and may actually survive short term wrack deposition (Bertness & Ellison 1987). Therefore, S. patens and D. spicata may dominate the recolonization of post-wrack areas regardless of the species dominant in the areas before wrack deposition.

S. alterniflora also may persist after wrack burial (Bertness & Ellison 1987). Furthermore, S. alterniflora is adapted to and thrives in anoxic areas of the marsh that receive frequent flooding (Bertness 1991b). This combination of tolerances may facilitate the establishment of S. alterniflora in disturbed areas especially if those areas already have S. alterniflora present and are being flooded.

Wrack deposition causes high mortality of vegetation through the increase of sulfide levels under the mat (Hartman 1988, Knowles et al. 1991). Even though S. alterniflora is not as sensitive to wrack deposition as other marsh macrophytes, substantial short-term decreases in its production after wrack deposition can occur (Reidenbaugh & Banta 1980). However, if D. spicata and S. alterniflora are able to survive the wrack deposition, their production may be greater in the post-wrack areas than other species since the wrack deposition would reduce competition by killing the more disturbance sensitive species.

3.2. Null Hypotheses

3.2.1. Effects of Increased Inundation on Saltmarsh Vegetation

H_O: Increased inundation does not affect the growth of J. roemerianus.

H_O: Increased inundation does not affect the senescence of J. roemerianus.

H_O: Increased inundation does not affect the growth of S. patens and D. spicata.

3.2.2. Effects of Wrack Deposition on Saltmarsh Vegetation

H_O: Wrack deposition will not affect the aboveground production of the plants.

H_O: Wrack deposition will not affect vegetative regrowth of a plant community.

(i.e., The pre-wrack species will return in the same abundance in a location

after wrack deposition as compared to the nonwrack areas.).

3.2.3. Combination of Increased Inundation and Wrack Deposition

H_O: Increased inundation and wrack deposition are not interacting stressors.

H_O: Increased inundation will not affect the vegetative regrowth during the first

growing season after wrack deposition.

3.2.4. Comparison between the Reference J. roemerianus Populations and the Experimental Controls

H_O: The growth dynamics (i.e., total height, biomass, survivorship, etc.) of the reference population of J. roemerianus does not differ from the experimental control.

3.2.5. Shading Experiment

H_O: Experimental shading does not promote or inhibit the growth of J. roemerianus and other marsh species.

3.3. Statement Regarding the Testing of the Hypotheses

The growth and senescence of J. roemerianus were determined by measuring the total and green heights of tagged leaves in the field through time and aboveground biomass collected at the end of the growing season. The production of S. patens and D. spicata was inferred from aboveground biomass measured at the end of the growing season. The vegetative regrowth of the plots was determined by aboveground biomass at the end of the growing season and vegetative ground cover analysis. The comparison of reference J. roemerianus populations and the experimental controls involved the monitoring of two populations of J. roemerianus that were inundated daily. The effects of shading on marsh plants involved the experimental shading of marsh pants.

4. MATERIALS AND METHODS

4.1. Site Description

The Virginia Coast Reserve - Long-Term Ecological Research site (VCR-LTER) is a 110 km long coastal marine landscape located on the southern end of the Delmarva Peninsula (USGS Nassawadox Quadrangle). The VCR-LTER consists of 14 barrier islands, coastal wetlands, deciduous and evergreen forest, and agricultural fields. The mean elevation for the VCR-LTER is less than 2 m above sea level (Hayden et al. 1994) (Fig. 1).

Brownsville Marsh (37^o 27' N, 75^o 50' W) is located within the VCR-LTER site. The marsh is an expansive, mainland brackish to salt marsh, whose edge is flooded twice a day by Phillips Creek, which has a tidal range of about 145 cm (Hmieleski 1994). Phillips Creek is part of the Machipongo drainage system. The change in elevation from the marsh to the forested upland is gradual on the north side and steep on the west side. The principal soil types for Brownsville Marsh are Chincoteague, Magotha, Nimmo, and Munden. All of these soil series are characteristic of moderately to poorly drained areas (Hmieleski 1994).

Distinctive plant zonation occurs throughout Brownsville Marsh. The upland forest is dominated by Pinus taeda (loblolly pine). S. patens and D. spicata dominate the organic high marsh of Brownsville. J. roemerianus also grows in the high marsh on peaty

Fig. 1. Map of Virginia Coast Reserve Long-Term Ecological Research site (VCR-LTER).

deposits (Hmieleski 1994). Short S. alterniflora grows toward Phillips Creek (Brinson et al. 1995). A salt flat has developed in the high marsh. Along the edges of the salt flat, Salicornia sp. is growing with S. patens and D. spicata. The high marsh area is not regularly flooded, being inundated only by extreme high tides. The mineral low marsh of Brownsville is regularly flooded and dominated by S. alterniflora (Blum 1993, Brinson et al. 1995).

4.2. Overall Design of Experiment

A randomized block design was used to test the effects of increased inundation and wrack deposition on a high salt marsh plant community from March 1994 to October 1995. The experimental site was located about 30 m away from Phillips Creek, in an area that is flooded by Phillips Creek only during periods of extreme high tides (i.e., storm tides and extremely high spring tides) (Fig. 2). The experimental area contained both a J. roemerianus dominated community and a S. patens/D. spicata dominated community. The experimental design was composed of three blocks (i.e., block X, block Y, and block Z) with each block containing three 4 m x 3 m plots. Block X contained plots 1-3, block Y contained plots 4-6, and block Z contained plots 7-9. Each plot consisted of a 2 m x 3 m area dominated by J. roemerianus and a 2 m x 3 m area dominated by S. patens/D. spicata. Each block contained three treatments for flooding (i.e., flooding, border control, and control), which I describe in detail later (Fig. 3). The mean elevation above mean sea level of block X was 1.01 m, block Y was 0.975 m, and block Z was 1.03 m.

Fig. 2. Map of Brownsville marsh showing the location of the experimental plots and the reference J. roemerianus populations.

Fig. 3. Design of experimental plots modified from Taylor (1995).

Overall, the mean elevation of the entire experimental site was 1.00 m above mean sea level (Taylor 1995).

Increased inundation was accomplished by pumping water from Phillips Creek through 1.27 cm (3/4 in) PVC pipes up to the experimental plots. Two different types of pumps were used during the experiment. The first pump was a Cimaron 4 in (10.16 cm) SolarSub pump. This pump was used from April 7, 1994, to the beginning of July 1994 when it stopped functioning. The second type of pump was a SolarJack 4 in the SDS series. The SolarJack pump operated from July 23 to September 25, 1994, when it was removed for the winter. The SolarJack pump was reinstalled on March 10, 1995, for the second growing season. A float switch (Thomas Product LTD. model 4200 P/N 24251) was used to activate the pump during both day and night high tides. During 1994 and from March 10, 1995, to June 8, 1995, the pumps were powered by four 12 V DC batteries (Reliant GPR-1285). The batteries were recharged by solar panels (Siemens M55 Solar Electric Module). During 1995, the batteries were no longer fully recharging, and I chose to run the pump directly off the solar panels. Therefore, from June 8, 1995, until the end of the 1995 growing season, the pump moved creek water through the pipes only during daytime high tides. The three plots receiving water from Phillips Creek (i.e., flooded plots) were enclosed by a wood border in order to slow surface runoff. The wood border consisted of strips of plywood 0.90 cm (3/8 in) thick and 20 cm high plywood. Three coats of Thompson's Water Seal were applied to the plywood to prevent decomposition. The strips were buried 10 cm into the ground leaving a 10 cm high border around the plots. Space was left between the strips to allow for the slow drainage of water. To determine if the wood borders and PVC pipes laid across the flooded plots had any effect in the flooded plots, half of the nonflooded plots had wood borders and PVC pipes laid across (i.e., border control plots). The other half of the nonflooded plots did not have a wood border surrounding them or PVC pipes (i.e., control plots). The control plots were delineated by placing PVC pipes at the corners of the plots (Taylor 1995). Therefore, each block contained one flooded plot, one border control plot, and one control plot. Pumping occurred from April 7, 1994, to the first week of July 1994; from July 23, 1994, to September 25, 1994; and from March 10, 1995, to August 20, 1995. The gaps in the pumping were the result of mechanical failures and waiting for the arrival of new parts. Appendix A describes the pumping design in more detail. Appendix C lists the amount of water pumped onto the plots throughout the experiment.

During the week of April 24, 1994, S. alterniflora wrack was placed on the plots to simulate wrack deposition. The wrack covered the left side of each plot facing from a boardwalk (Fig. 3). The S. alterniflora wrack was collected from a deposit near the experimental site from a storm in March 1994. The wrack covered half the J. roemerianus community and half of the S. patens/D. spicata community and was secured with 1.905 cm plastic mesh bird netting and wire to prevent removal. Therefore, each plot had a 4 m x 1.5 m area of stimulated wrack deposition and a 4 m x 1.5 m area of no wrack deposition (Fig. 3). On October 15, 1994, approximately 1 m x 1.5 m of wrack was removed from either end of the 4 m x 1.5 m wrack deposition area within each plot. The remaining wrack within the plots was left and resecured with bird netting and wire.

4.3. Tracking of the Growth and Senescence of Juncus roemerianus leaves

To determine the effects of increased inundation on the growth of J. roemerianus and its regrowth after wrack deposition, I recorded the progress of tagged J. roemerianus leaves throughout the experiment (Christian et al. 1990, Williams and Murdoch 1972). I used a meter stick to measure the height of green portion and the total height of each leaf to the nearest 0.1 cm. Measurements were taken approximately every four weeks from October through April and every two weeks from May through September. Table 1 describes the timing, number, and designation of each cohort. I tagged leaves twice a year since J. roemerianus produces new leaves throughout the spring and summer months. Therefore, tagging twice a year would more accurately reflect the growth pattern of the plants. Appendix B has the details on the exact dates measurements were taken and the means and standard deviations for each cohort and treatment. I measured the leaves on 31 different days over the course of 20 months.

4.3.1. Reference Population of Juncus roemerianus

I also tagged two reference populations of J. roemerianus leaves in areas that were regularly inundated, and, therefore, flooded more often than the control plots (Fig. 2).

Table 1. Dates of when I tagged each cohort of Juncus roemerianus leaves, the cohort's designation, and the number of leaves tagged.

Cohort Number	Date Tagged	Initial Height of Leaves	Area where Leaves were Tagged	Number of Leaves Tagged per Plot	Number of Leaves Tagged per Flooding Treatment
1	March 26, 1994	< 30 cm	Nonwrack	5	15
2	March 26, 1994	> 30 cm	Nonwrack	5	15
3	July 11, 1994	< 30 cm	Nonwrack	5	15
4	March 4, 1995	< 30 cm	Nonwrack	5	15
5	March 4, 1995	< 30 cm	Post-wrack	5	15
6	July 6, 1995	< 30 cm	Nonwrack	5	15
7	July 6, 1995	< 30 cm	Post-wrack	5	15
			TOTALS	35	105
4	March 4, 1995	< 30 cm	Reference Areas	20	20
6	July 6, 1995	< 30 cm	Reference Areas	20	20

The reference populations were designated as reference population A and reference population B. Ten leaves were tagged in each reference population on March 4, 1995, and ten were tagged in each reference population on July 6, 1995 (Table 1). I took measurements of green and total heights at the same time as the leaves in the experimental plots. Appendix B contains the exact dates and means and standard deviations for the leaves from the reference populations.

4.4. Aboveground Biomass Sampling of the Plots

I collected aboveground biomass samples in the plots at the end of the first and second growing seasons. A total of eight 0.0625 m² aboveground biomass samples were collected in every plot at the end of each growing season. On October 15, 1994, two 0.0625 m² samples were collected without bias in the following areas: (1) Juncus roemerianus-wrack, (2) J. roemerianus-nonwrack, (3) S. patens/D. spicata-wrack, and (4) S. patens/D. spicata-nonwrack. On September 30, 1995, two 0.0625 m² samples were collected without bias in the following areas: (1) J. roemerianus-regrowth from wrack deposition, (2) J. roemerianus-nonwrack, (3) S. patens/D. spicata-regrowth from wrack deposition, and (4) S. patens/D. spicata-nonwrack. I did not collect samples from the remaining wrack areas in 1995 because of the lack of plant material in those areas. Also in 1995, four biomass samples were collected in the J. roemerianus reference population. I sorted samples by plant species and into living, dying, and dead plant material. For each sample taken in the J. roemerianus-nonwrack areas, the total and green heights of 20 living (>95% green) and 20 senescing (<95% green) J. roemerianus leaves were measured. The total numbers of living and senescing J. roemerianus leaves also were recorded to determine leaf densities. Once sorted, the plant material was placed in brown paper bags and dried at 85 ^oC for at least 72 h to ensure complete dryness before weighing. Each type of plant material was weighed inside its bag, and then the bag was weighed. I subtracted the mass of the bag from the total mass (i.e., plant material + bag) to yield the mass of the dried plant material. I converted dried plant masses to g/m² and leaf densities to number of leaves/m² by dividing the numbers by 0.0625 m².

4.5. Vegetation Analysis of Ground Cover in the Plots

Ground cover analysis was done twice (July 11, 1994, and July 24, 1995) and involved placing a 1 m² quadrat, divided into 100 cm² grids, over each plot. I would place the quadrat down on the plots, record the species present in each grid, and then move the quadrat down 1 m. I repeated this on both the wrack and nonwrack sides of the plots. This analysis of vegetation included the wrack and nonwrack sides in (1) the area dominated by J. roemerianus, (2) the transition area between the two vegetation types, and (3) the areas dominated by S. patens and D. spicata. I recorded the presence and the absence of each plant species in each grid. The total number of individuals and the relative importance of each species were not recorded.

4.6. Annual Net Primary Production (ANPP) of Juncus roemerianus

I calculated annual net primary production (ANPP, g/(m² x y)) of J. roemerianus as the product of the annual frequency of replacement of growing leaves (FREQ, number/y), production-to-biomass ratio (P/B, dimensionless), and aboveground biomass (AVE BIO, g/m²). FREQ and P/B were determined from the tagged leaf data, and AVE BIO was determined from the aboveground biomass collections (Christian et al. 1990). I estimated ANPP for both the leaves tagged in the experimental plots and for the leaves tagged in the reference populations.

In order to determine when the leaves began to grow, I extrapolated backwards to T₀ assuming a linear relationship between height and time:

T₀ = T₁ - [H₁ x (T₂ - T₁) / (H₂ - H₁)]

where T₁ = date of tagging; T₂ = next sampling date; H₁ = height at the day of tagging; and H₂ = height at the next sampling. I only used total height measurements to determine T₀.

The difference between T₀ and the day at which the total height was maximum (T_m) was designated as the replacement time for a leaf and was divided by 365 days/y in order to determine FREQ (Christian et al. 1990).

I considered the total height at T_m as the maximum height (H_m) for each leaf. P/B was calculated by dividing H_m by the average total height (H_avg) during the study. H_avg was calculated as:

H_avg = Sum { [H_i-1 + (H_i - H_i-1) / 2] x (T_i - T_i-1) / (T_m - T₀) }

for i = 0 to m

where H_i (cm) = total height at data T_i; and m = sampling date for H_m. I calculated H_avg for each leaf from the different cohorts (Christian et al. 1990).

I used the biomass collections from 1994 and 1995 to estimate AVE BIO of growing leaves (g/m²). Since the J. roemerianus leaves reached their maximum green height before they reached their maximum total height, I combined the growing leaf biomass and a percentage of the senescing leaf biomass for AVE BIO. I determined that leaves tended to be approximately 76 % green when they reach their maximum total height using the tagged leaf data. From the measurements of the random senescing leaves from the biomass collections, I was able to estimate that about 55 % of the senescing biomass may have actually been growing leaves. Therefore, I multiplied 55 % by the senescing leaf biomass for each sample and added the product to the biomass of the growing leaf biomass. The sum of these two numbers was used as AVE BIO for the ANPP calculations.

The means of FREQ and P/B for each cohort were used to estimate ANPP. For the cohorts tagged in 1994 (cohorts 1-3), the 1994 AVE BIO was used for the ANPP calculations; and the 1995 AVE BIO was used for the cohorts tagged in 1995 (cohorts 4-7). I used the following equation to calculate ANPP (g/(m² x y)):

ANPP = FREQ (number/y) x P/B (dimensionless) x AVE BIO (g/m²).

4.7. Shading Experiment

To examine the possible effects of shading by older J. roemerianus leaves on younger J. roemerianus leaves, I performed a shading experiment in two patches of J. roemerianus near the experimental plots. In May 1995, I clipped five 0.0625 m² areas of J. roemerianus from the edges of each patch. Shading was accomplished by using wooden frames with two layers of window screen on them. This corresponded to approximately a 75 % reduction in light intensity reaching the leaves. Furthermore, two additional wooden frames were made without screen in order to control for the effect of the wooden frames. The placement of the treatments was as follows: shaded area, non-shaded area (i.e., control), wooden frame control, non-shaded area, shaded area. This incomplete block design was used for both patches. An incomplete block design was used because this was only a preliminary study.

I harvested aboveground biomass twice from the shading experiment, first on July 24, 1995, and then on October 14, 1995. Because of the small amount of plant material collected, I sorted the samples into only J. roemerianus material and non-J. roemerianus material. Plant material was placed in a brown paper bag and dried at 85 ^oC for 72 h to ensure complete dryness before weighing. Each type of plant material was weighed inside its bag, and then the bag was weighed. I subtracted the mass of the bag from the total mass (i.e., plant material + bag) to yield the mass of the dried plant material. Dry plant masses were converted to g/m² by dividing each mass by 0.0625 m²

4.8. Statistical Analysis of Data

I used the general linear model (GLM) procedure of SAS to analyze most of the data (SAS Institute Inc. 1988). The model for the three-way ANOVA was as follows: BLOCK, FLOODING TREATMENT, WRACK TREATMENT, BLOCK * FLOODING TREATMENT, and FLOODING TREATMENT * WRACK TREATMENT. Significance was accepted at the 0.05 level. Specific relationships among means were determined using the least square means test. The GLM with a three-way ANOVA was used to analyze the following measurements: (1) maximum green heights, maximum total heights, and leaf densities of J. roemerianus; (2) the biomass of each type of plant material, total biomass of each species, and total plant biomass for 1994 and 1995; and (3) FREQ, AVE HT, P/B, and ANPP.

For the comparison of the maximum green and maximum total heights and ANPP of reference populations of J. roemerianus and of the control plots, a two-way ANOVA model was used with GLM and least square means (SAS Institute Inc. 1988). The model for the ANOVA was TREATMENT, COHORT, TREATMENT*COHORT. I compared the biomass samples collected from the reference populations and the control plots using a one-way ANOVA model that examined the effect of TREATMENT only.

I also used a two-way ANOVA with GLM and least square means for the shading experiment (SAS Institute Inc. 1988). The model for the two-way ANOVA was BLOCK, SHADING TREATMENT, and BLOCK*SHADING TREATMENT.

Statistical analysis of growth and senescence data of J. roemerianus leaves involved the use of life table analysis. I used the SAS procedures LIFEREG and LIFETEST (SAS Institute Inc. 1988). The LIFEREG procedure uses parametric models to compare failure-time data. The LIFEREG procedure can handle more complex models than the LIFETEST procedure, which is a nonparametric test. I compared the life table data using the Weibull model command. Since LIFEREG did not reveal any higher order interactions in my data set, I used a Chi-square with one degree of freedom to compare differences of the log-rank values for each variable from LIFETEST. Log-rank values are calculated from the distribution of the data. Furthermore, I ran the stratum (i.e., variable) FLOODING TREATMENT separate from the stratum COHORT; and analysis for the cohorts tagged in 1994 (cohorts 1-3) were done separately from the cohorts tagged in 1995 (cohorts 4-7). The separation of FLOODING TREATMENT and COHORT and the 1994 and 1995 cohorts were done to simplify the model. I compared the amount of time required for a leaf (1) to reach maximum green height, (2) to reach maximum total height, (3) to die from the point of maximum green height, and (4) to die from the point of maximum total height. SAS used censored data up to the point of the censorship. All analyses were performed at a 0.05 level of significance.

I determined the frequency of change in ground cover from 1994 to 1995 for the wrack and nonwrack sides of each plot as a percentage of change using the SAS procedures SORT, PLOT, and FREQ. I compared the percentage of change among the plots using the GLM procedure of SAS (SAS Institute Inc. 1988). The model for the three-way ANOVA for the vegetative analysis was BLOCK, FLOODING TREATMENT, WRACK TREATMENT, BLOCK*FLOODING TREATMENT, and FLOODING TREATMENT*WRACK TREATMENT. Significance was accepted at the 0.05 level. Specific differences among means were determined by use of the least square means test.

5. RESULTS

The results reported here include the analyses of the flooding and wrack treatments and block effect. I discuss interactions between flooding treatment and block and between flooding and wrack treatments only when the components of the interactions were significant by themselves. Table 2 lists the abbreviations I used on the tables and figures and an example of how to read the GLM tables in this Results section.

5.1. Growth and Senescence of Juncus roemerianus

5.1.1. Maximum Green and Total Heights of Juncus roemerianus

Table 3 contains the means and standard deviations for the maximum green and total heights of the J. roemerianus leaves from the experimental plots and the reference populations. Mean maximum green heights ranged from 36.8 to 69.4 cm, and mean maximum total heights ranged from 41.5 to 76.7 cm (Table 3).

For the maximum green and total heights from the experimental plots, the flooding treatment was not significant (Table 4). The cohorts were significantly different from each other. However, most of the differences among the cohorts were the result of when I tagged and not because I tagged the cohort tagged in a nonwrack or post-wrack area. Since the purpose of this thesis was to study flooding and wrack treatments, I will

Table 2. List of abbreviations used in tables and figures and an example of how to read the GLM tables.

Abbreviation	Refers To
F	Flooded plots
B or BC	Border control plots
C	Control plots
N	Nonwrack areas
W	Wrack areas
R	Areas recovering from wrack (post-wrack)
X	Block X
Y	Block Y
Z	Block Z
1	Cohort 1
2	Cohort 2
3	Cohort 3
4	Cohort 4
5	Cohort 5
6	Cohort 6
7	Cohort 7

Table 2, continued

Key:
	Block	Flooding Treatment	Wrack Treatment	Block * Flooding	Flooding * Wrack
name of data being tested	p-value for block effect (Significant differences among the blocks)	p-value for flooding treatment (Significant differences among the flooding treatments)	p-value for wrack treatment (Significant differences between the wrack treatments)	p-value for block and flooding treatment interaction	p-value for flooding treatment and wrack treatment interaction
Example:
Biomass	0.08	0.01 (F > C)	0.01 (N > W)	0.90	0.54
	no significant differences	Flooded plots have significantly more biomass than the control plots	Nonwrack areas have significantly more biomass than the wrack areas	no significant interaction between block and flooding treatment	no significant interaction between flooding treatment and wrack treatment

Table 3. Means and standard deviations (in parenthesis) of the maximum green and total heights (cm) for each J. roemerianus cohort among the different flooding treatments and for the reference J. roemerianus population. The year and the wrack treatment for each cohort are listed in parenthesis. (94 = leaves tagged in 1994 and 95 = leaves tagged in 1995; N = leaves tagged in the nonwrack areas and R = leaves tagged in the areas recovering from wrack deposition)

Flooding Treatment	Cohort Number	Maximum Green Heights	Maximum Total Heights
Flooded	1 (94, N)	68.9 (26.4)	73.3 (25.9)
	2 (94, N)	69.4 (19.1)	76.7 (18.7)
	3 (94, N)	56.9 (18.8)	61.8 (21.6)
	4 (95, N)	57.7 (19.4)	66.2 (14.2)
	5 (95, R)	52.7 (20.0)	65.3 (16.1)
	6 (95, N)	46.5 (16.5)	48.4 (16.6)
	7 (95, R)	47.3 (10.8)	50.4 (10.4)
	Total	57.0 (20.6)	63.1 (20.3)
Border Control	1 (94, N)	55.8 (16.8)	61.9 (16.0)
	2 (94, N)	66.3 (12.4)	74.1 (10.9)
	3 (94, N)	52.7 (8.4)	60.8 (10.1)
	4 (95, N)	51.6 (15.3)	57.3 (13.8)
	5 (95, R)	57.1 (24.1)	66.9 (22.9)
	6 (95, N)	39.5 (10.3)	41.5 (11.2)
	7 (95, R)	49.5 (11.6)	53.1 (10.4)
	Total	56.5 (18.9)	59.4 (16.9)

Table 3, continued.

Flooding Treatment	Cohort Number	Maximum Green Heights	Maximum Total Heights
Control	1 (94, N)	62.2 (23.9)	68.1 (23.8)
	2 (94, N)	67.8 (8.9)	73.5 (8.7)
	3 (94, N)	52.2 (18.9)	57.8 (22.8)
	4 (95, N)	64.1 (21.8)	70.0 (22.2)
	5 (95, R)	48.8 (21.5)	64.0 (22.2)
	6 (95, N)	52.2 (11.7)	56.0 (10.3)
	7 (95, R)	48.0 (12.2)	53.9 (14.0)
	Total	55.5 (17.9)	63.3 (19.5)
Reference A	4 (95, N)	64.8 (19.9)	69.8 (19.4)
	6 (95, N)	48.6 (19.9)	50.3 (10.2)
	Total	56.7 (17.4)	60.0 (18.1)
Reference B	4 (95, N)	36.8 (15.7)	46.0 (15.5)
	6 (95, N)	46.3 (13.4)	49.6 (13.8)
	Total	41.6 (15.0)	47.8 (14.4)

Table 4. Results of GLM for maximum green and total J. roemerianus leaf heights. Specific significant differences were determined by least square means and are listed below the GLM values in parentheses.

	Block	Flooding Treatment	Cohort	Block * Flooding	Flooding * Cohort
Maximum Green Heights	< 0.01 (Z > Y & X; Y > X)	0.21	< 0.01 (1 > 3, 5, 6, & 7; 2 > 3, 4, 5, 6, & 7; 3 > 6; 4 > 6 & 7)	0.35	0.36
Maximum Total Heights	< 0.01 (Z > Y & X; Y > X)	0.15	< 0.01 (1 > 3, 6, & 7; 2 > 1, 3, 4, 5, 6, & 7; 3, 4, & 5 > 6 & 7)	0.34	0.50

only be discussing the significant differences between the nonwrack and post-wrack cohorts that were tagged at the same time (i.e., cohort 4 vs. 5 and cohort 6 vs. 7). The cohorts tagged in the spring of 1995 in the nonwrack (cohort 4) and post-wrack (cohort 5) areas did not have statistically different maximum green and total heights (Table 4). The cohorts tagged in the summer of 1995 in the nonwrack (cohort 6) and post-wrack (cohort 7) areas did not have statistically different maximum green and total heights (Table 4).

Block effect was significant for the maximum heights. Block Z had significantly higher heights than block X and Y. Furthermore, block Y had significantly higher heights than block X (Table 4).

5.1.1.1. Control vs. Reference Population Comparison for Maximum Heights

The treatment effect was significant for the comparison of the maximum green and total heights of the control plots and the reference population. For the maximum green heights, the leaves in reference A and all the control plots were significantly taller than the leaves in reference B. The maximum total heights for reference A and the control plots in block X and Y were significantly higher than reference B (Table 5). The cohort effect was significant for maximum total heights with cohort 4 having significantly taller leaves than cohort 6.

Table 5. Results of GLM for maximum green and total heights of J. roemerianus from the reference populations and from the control plots. Specific significant differences were determined by least square means and are listed under the GLM values in parenthesis.

	Treatment	Cohort	Treatment * Cohort
Maximum Green Heights	0.01 (Reference A, X, Y & Z > Reference B)	0.05 (4 > 6)	0.14
Maximum Total Heights	0.01 (Reference A, X & Y > Reference B)	< 0.01 (4 > 6)	0.17

5.1.2. Life Table Analysis (Survival Data Analysis) of Juncus roemerianus Leaves

Tables 6 through 9 contain the means and standard deviations of the time (days) required for the cohorts to reach their maximum green and total heights from T₀ and for the time (days) required for the cohorts to go from the maximum green and total heights to death, respectively. The tables also list the number of censored observations for the time to maximum green and total heights and for the time to death from the maximum heights. Censored observations include leaves that were lost during the study and leaves which were either still growing or senescing when the studied ended in October 1995. In this section, I am using the term "survivorship" in the statistical sense of the word. Therefore, survivorship is the probability of not reaching either a maximum height (green or total) or death from either maximum green or total heights. Tables 8 and 9 contain this information for both the experimental plots and the reference populations.

When studying the data, it is important to keep two facts about the data in mind. First, cohort 2 was tagged in the spring of 1994 and consisted of older leaves. Therefore, the leaves in cohort 2 may have not grown much beyond their tagged heights. Furthermore, since the study ended in the fall of 1995, many of the leaves tagged during 1995, did not completely senesce or even begin to senesce. Because of this, there were not enough data to accurately determine the effects of inundation or wrack deposition on the senescence of the 1995 cohorts. Therefore, the senescing data of the 1995 cohorts are not presented in this thesis.

Table 6. Means and standard deviations (in parenthesis) of the time (days) for the 1994 cohorts to reach the maximum green heights from T₀ and the time for the 1994 cohorts to completely senesce from their maximum green heights. The number of censored points for each mean also is listed. The cohorts are listed by flooding treatment. In 1994 leaves were only tagged in nonwrack areas.

Flooding Treatment	Cohort	Time to Max. Green Height	Max. Height Censored	Time to Death	Death Censored
Flooded	1	>175 (53)	1	>210 (144)	2
	2	347 (131)	0	>188 (87)	1
	3	>228 (138)	2	>170 (123)	8