HIGH MARSH-FOREST TRANSITIONS IN A BRACKISH
MARSH: THE EFFECTS OF SLOPE.
Joseph I. Hmieleski.
(Under the direction of Mark M. Brinson,
Ph.D.) Department of Biology, East Carolina University, May 1994.
ABSTRACT
Abiotic and biotic variables were assessed along brackish
marsh-upland continua at the Virginia Coast Reserve/LTER, to determine
the effect of slope on the position of vegetation zones through its
effect on physicochemical variables. Four transects through the high
marsh-upland transition were established to represent extremes of
slope and proximity to tidal creek (i.e., Flat Far, Flat Near, Steep
Far, Steep Near). Vegetation zones were delineated as high marsh,
transition, and forest based on differences in percent cover of grass,
shrub, and tree species. Vegetation cover, elevation, and soil
characteristics were sampled along all transects; hydrologic and pore
water variables were measured along two of the four transects, Steep
Near and Flat Near, from November 1991 to August 1993.
The transition and forest zones of both flat transects
occupied significantly lower elevations and were more hummocky as
compared with the steep transects, while the high marsh occupied
similar elevations. The flat transition zone had less drainage and a
higher mean annual water table, higher and more variable mean
groundwater and pore water salinities, and similar redox potentials
for the 10 cm depths in the transition and forest zones. However, the
steep slope showed significantly lower redox potential at 20 cm depth
for all zones which is hypothesized to be related to the discharge of
anaerobic groundwater. Hydrogen sulfide was very low throughout, but
was significantly higher for the flat transect high marsh. Results of
univariate analysis were supported by multivariate factor analysis
which identified three factors accounting for 69.4% of the variation
in the data: brackish water, redox potential, and sulfide variables.
Factor scores for brackish water and redox potential for the flat
transition and forest zones were significantly higher than for
corresponding steep zones, but similar in the high marsh zone. All of
these trends suggest that the dominant source of water is atmospheric
for the flat transition and drainage and groundwater discharge for the
steep transition.
Degree of slope affects transition zone size and
position, such that the steep transition zone is narrower and changes
more abruptly than the flat transition. Hydroperiod, salinity, and
redox potential are believed to directly influence the distribution of
plant species and their relative dominance. I could not determine
whether hydroperiod or salinity controlled the position of the
transition zone. However, the position of the transition zone roughly
corresponds to the distance at which mean salinity decreases most
rapidly. This occurs at the base of the forest area for the steep
slope due to drainage or groundwater discharge, but at a greater
distance from the tidal creek and at lower elevation for the flat
slope. Marsh vegetation may restrict overland flow over long
distances from the tidal creek during flood events, thus allowing the
glycophytic vegetation to inhabit lower elevations, further away from
the creek. Hummocks appear to allow glycophytic vegetation to colonize
closer to the tidal creek thus increasing the width of transition
zones.
HIGH MARSH-FOREST TRANSITIONS IN A BRACKISH MARSH: THE EFFECTS OF SLOPE
A Thesis
Presented to
the Faculty of the Department of Biology
East Carolina University
In Partial Fulfillment
of the Requirements for the Degree
Master of Science in Biology
by
Joseph I. Hmieleski
May 1994
DIRECTOR OF THESIS
________________________________________________
Mark M. Brinson, Ph.D.
COMMITTEE MEMBER
________________________________________________
Robert R. Christian,
Ph.D.
COMMITTEE MEMBER
________________________________________________
Claudia L. Jolls,
Ph.D.
COMMITTEE MEMBER
________________________________________________
Kevin F. O'Brien,
Ph.D.
CHAIR OF THE DEPARTMENT OF BIOLOGY
________________________________________________
Charles E. Bland, Ph.D.
DEAN OF THE GRADUATE SCHOOL
________________________________________________
Diane Jacobs, Ph.D.
...the present vegetation mosaic may be thought of
as a momentary expression, different in the past, destined to be
different in the future, and yet as typical as would be a photograph of
moving clouds.-- Miller and Egler(1950)
ACKNOWLEDGEMENTS
This work benefitted enormously from the help and
guidance of my committee: Drs. Mark M. Brinson, Robert R. Christian,
Claudia L. Jolls, and Kevin F. O'Brien. I would like to especially
acknowledge my thesis advisor, Dr. Mark Brinson, for his patience,
advice, energy, insight, and his ability to always make sure I was fed
(funded). I have many fond memories of personally guided tours
through wetlands from Edmonton, Canada, to New Orleans, Louisiana, of
which I will not soon forget. I have yet to meet a person who can
keep up with him in the marsh. Special thanks also go to Dr. Robert
Christian for his advice and help in the field; to Dr. Claudia Jolls
who continually reminded me of the vegetation aspects of my work; and
to my statistician guru, Dr. Kevin O'Brien.
So many people helped me along the way that I am sure
that I will never be able to thank them all. The hard working field
helpers include: James Taylor, who helped greatly in field collection
and installation of sampling instruments, John J. Russell, and Paul
Farley. The research at the VCR/LTER went most smoothly with the help
of Dr. Linda Blum, Dr. John Porter, Marcio Santos (the Sipper King),
Dave Osgood, Randy Carlson, Jimmy Spitler, and many UVA REUs. I would
also like to thank Dr. Tom Chenier for helping me through the panics
of factor analysis. Binita Kwankin provided a much needed review of
an early draft, and helped me through the turmoil of life on
Interstate 95. And lastly, I would like to thank my family for
emotional and fiscal support, who thought I was crazy for leaving a
great job.
This work was funded in part by a National Science
Foundation Grant DEB-9211772 and BSR-8702333-06 through the University
of Virginia, VCR/LTER, and East Carolina University.
TABLE OF
CONTENTS
Page
LIST OF FIGURES
vi
LIST OF TABLES
ix
1. INTRODUCTION
1
1.1 Factors Affecting Salt Marsh Vegetation
2
1.1.1 Oxygen Availability
2
1.1.2 Hydrogen Sulfide
3
1.1.3 Salinity
7
1.2 Effects of Slope
9
1.2.1 Effects on Hydrology
9
1.2.2 Effects on Vegetation
13
1.3 Objectives and Hypotheses
15
2. SITE DESCRIPTION AND METHODS
18
2.1 Site Description
18
2.2 Study Transects
22
2.3 Methods
24
2.3.1 Vegetation Sampling and Analysis
24
2.3.1.1 Vegetation Sampling
24
2.3.1.2 Vegetation Analysis
27
2.3.2 Surface Elevation
29
2.3.3 Soil Characteristics
31
2.3.4 Water Table Fluctuations
33
2.3.5 Groundwater Salinity
34
2.3.6 Pore Water
35
2.4 Data Analysis
36
3. RESULTS
40
3.1 Vegetation
40
3.1.1 Zone Detection and Delineation
40
3.1.1.1 Flat Near Transect
40
3.1.1.2 Flat Far Transect
43
3.1.1.3 Steep Near Transect
47
3.1.1.4 Steep Far Transect
48
3.1.2 Plant Species Presence
49
3.2 Elevation
52
3.2.1 Variation in Elevation by Transect
52
3.2.2
Variation in Elevation by Zone 55
3.2.3 Surface Macrotopography
58
3.3 Soil structure
58
3.3.1 Depth to mineral soil
58
3.3.2 Bulk Density and Organic Carbon
61
3.3.2.1 Bulk Density and
Organic Carbon by Slope 61
3.3.2.2 Bulk Density and
Organic Carbon by Zone 63
3.3.3 Chlorinity
65
3.3.3.1 Chlorinity by Slope
65
3.3.3.2 Chlorinity by Zone
65
3.4 Hydroperiod
66
3.4.1 Water Table Fluctuations
69
3.4.2 Precipitation, PET, and Water Table
Drawdown 72
3.4.3 Groundwater Salinity Patterns
74
3.5 Pore Water Chemistry
77
3.5.1 Pore Water Salinity
77
3.5.2 Redox Potential, Hydrogen Sulfide,
and pH 83
3.6 Factor Analysis
88
3.6.1 Analysis of Entire Data Set
88
3.6.2 Analysis of Transition Data Set
94
4. DISCUSSION
100
4.1 Factors Influencing Position of Transition Zone
101
4.1.1 Brackish Water Factor
101
4.1.1.1 Effect of Slope on
Hydroperiod 102
4.1.1.2 Effect of Slope on
Salinity 104
4.1.1.3 Effect of Slope on pH
106
4.1.2 Redox Factor
107
4.1.3 Sulfide Factor
108
4.2 Slope Effects on Vegetation
110
4.2.1 Position of Transition Zone
110
4.2.1.1 Steep Transitions
111
4.2.1.2 Flat Transitions
114
5. CONCLUSIONS
118
6. LITERATURE CITED
121
LIST OF FIGURES
Figure 1.Elements of a water budget in a tidal salt marsh as
they are
affected by slope.
10
Figure 2.Simple conceptual model of the effects of slope on the
hydrology
and pore water and vegetation used to formulate hypotheses.
14
Figure 3.Location and arrangement of study transects at the
Brownsville,
VA, marsh representing Steep Near (SN), Steep Far (SF),
Flat
Near (FN), and Flat Far (FF) transects in relation to
Phillips
Creek.
19
Figure 4.Frequency of occurrence for shoreline to upland soil
assemblages
for Northampton County, VA, using soil maps (USDA 1989).
20
Figure 5.Possible combinations of landscape position in relation
to proximity
to tidal creek and gradients of slope used to choose study
transects. 23
Figure 6.Layout of the Flat Near and Steep Near study transects
indicating
locations of the surficial groundwater wells, lysimeters,
and water
level recorders.
25
Figure 7.Sampling frame for vegetation analysis. (A) Frame size
and
geometry used for vegetation sampling. Frame size for
measuring
vegetation were: herbaceous (H) 1 m2, shrub (S) 4 m2, and
trees
(T) 8 m2.
26
Figure 8.Squared Euclidean distance computed using moving
split-window
analysis.
41
Figure 9.Peaks in the squared Euclidean distance for the Flat
Far transect
using window sizes of 10, 8, and 6 sampling points.
42
Figure 10.Percent of the total cover for each species grouped by
zone and
vegetation type.
44
Figure 11.Elevation contour relative to the creekbank benchmark
(HAYD). 53
Figure 12. Mean elevation for the high marsh zone, transition zone,
and forest
zone.
57
Figure 13.Topographic variation in
marsh surface and underlying mineral soil
from high marsh to terrestrial forest.
59
Figure 14.Mean soil structure data from core samples at 10 and
20 cm depths
for both Flat Far and Near (Flat) and Steep Far and Near
(Steep),
q standard error.
64
Figure 15.Mean monthly water table levels for the high marsh,
transition, and
forest zones determined by shallow (1.5 m) groundwater
wells. 67
Figure 16.Hydrograph during September 1992 showing the effects
of rainfall
and a storm surge due to Tropical Storm Diane for the Flat
Near
transition.
70
Figure 17.Monthly precipitation totals from 1990 to 1993
(Brownsville
Marsh) and 1949 to 1981 (Cheriton, VA).
73
Figure 18.Hydrograph during June and July 1993 comparing the
diurnal pulse
of water table drawdown due to evapotranspiration for flat
and
steep slopes.
75
Figure 19.Mean monthly salinity for the high marsh, transition,
and forest
zones determined by shallow (1.5 m) groundwater wells.
76
Figure 20.Box plot of the groundwater salinities along the Flat
Near and
Steep Near transects for the period beginning 4 November
1991 for
the Flat Near and 5 August 1992 for the Steep Near
transect. 78
Figure 21.Pore water salinity collected for the Flat Near and
Steep Near
transects from lysimeters at 10 and 20 cm depths for the
period 14
Jan to 29 July 1993.
80
Figure 22.Mean pore water data collected for the Flat Near and
Steep Near
transects from lysimeters at 10 and 20 cm depths for the
period 14
Jan to 29 July 1993.
81
Figure 23.Comparison between salinities measured using shallow
groundwater
and pore water salinities collected using lysimeter at 10
cm for the
Flat Near transect for the period January to July 1993.
84
Figure 24. Pore water redox potential collected for the
Flat Near and Steep Near
transects from lysimeters at 10 and 20 cm depths for the
period 14 Jan
to 29 July 1993.
85
Figure 25.Sulfide concentration collected from the 0 m lysimeter
at the 10 cm
depth and from the 150 m lysimeter at the 20 cm depth for
the Flat
Near transect.
87
Figure 26.Mean factor scores for the high marsh, transition, and
forest
vegetation zones for the period 14 January to 29 June 1993,
q
standard error.
91
LIST OF TABLES
Table 1.Algorithm used to convert dominance rankings to percent
cover. 28
Table 2.Vascular plant species in the Phillips Creek marsh.
50
Table 3.Slope summary table for study transects determined by
linear
regression.
54
Table 4.Summary of leveling transects separated by vegetation
zone. 56
Table 5.Mean depth (m) to organic layer for the Flat Near and
Steep Near
study transects, q standard error.
60
Table 6.Soil properties for study transects obtained from
sediment cores. 62
Table 7.Characteristics of hydroperiod at the transition zone
water level
recorders for the Steep Near and Flat Near transects.
68
Table 8.Mean pore water data collected from low-tension
lysimeters for the
period 14 January to 20 May 1993, q standard error.
82
Table 9.Correlations of physical and chemical variables.
89
Table 10.Correlations of physical and chemical variables with
factors. 90
Table 11.Two-way analysis of variance comparing factor scores
for
lysimeter data among slopes and sampling dates for the
entire data
set.
92
Table 12.Mean factor scores for the Flat Near and Steep Near
transects
grouped by vegetation zone, q standard error.
93
Table 13.Correlations of physical and chemical variables with
factors using
data from transition zone only.
96
Table 14.Two-way analysis of variance comparing factor scores
for
lysimeter data among slopes and sampling dates for the
transition
data set only.
97
Table 15.Mean factor scores for
Transition zone dataset for the Flat Near
and Steep Near transects, q standard error.
99
1. INTRODUCTION
Salt and brackish marshes are characterized by sharp zonation
patterns of
vascular plants which have been the object of intensive study by many
researchers.
The vegetation patterns have been shown to be determined by the frequency and
duration of tidal flooding along with its related affects on substrate oxygen
levels,
pore water salinity, pore water sulfides, and nutrient availability. These soil
and pore
water characteristics have been shown to limit growth, reproduction, and
establishment of marsh vegetation.
Much of the work which has been done on the factors influencing
zonation and
growth of salt marsh vegetation has been conducted on the regularly flooded low
and
irregularly flooded high marsh vegetation zones. However, the transition from
marsh
to forested and upland areas at the landward edge of salt marshes has received
little
attention despite the fact that these areas are particularly susceptible to the
effects of
sea level rise. The transition from marsh to upland represents a continuum of
soil
and pore water characteristics which influence vegetation types and abundances.
The existing paradigms for coastal wetlands suggest that abiotic
variables
associated with inundation by brackish water and their influence on vegetation
should
decrease along the continuum from low marsh to upland as a function of
elevation.
Elevation has been shown to exert its influence mainly through the extent of
tidal
inundation. However, areas with little or no change in elevation from the marsh
landward may be influenced more by the distance to a tidal source than
elevation.
The effect of landscape slope on the marsh-forest soil characteristics, and
ultimately
the position of vegetation zones, has not been well documented.
The purposes of this section are to review the literature on the
factors known to
affect marsh and upland vegetation and to formulate hypotheses about the
patterns of
these factors along the continuum from marsh to upland along gradients of slope.
The introduction is organized into three sections which will discuss: (1) the
physical
and chemical factors affecting salt marsh and upland vegetation; (2) the
hydrologic
processes underlying the continuum from marsh to forested areas; and (3)
suggested
hypotheses for the physicochemical patterns as they are affected by slope.
1.1 Factors Affecting Salt Marsh Vegetation
The inundation by brackish water has been associated with changes
in sediment
physical and chemical characteristics which, in turn, affect the types of
vegetation
which can tolerate the potential stresses. The major physical and chemical
stresses
from flooding and its related effects include sediment oxygen availability,
hydrogen
sulfide, and salinity.
1.1.1 Oxygen Availability
The saturation of soil with water reduces the rate of oxygen
diffusion (Gambrell
and Patrick 1978) and can limit root and microbial oxygen availability. Plants
growing in submerged soils have two major adaptations to growth in anoxic
sediments: oxygen transport from the aerial parts to the roots and fermentation.
Oxygen transport from aerial parts to the roots in some marsh
plants occurs
mainly through gas spaces or aerenchyma (Conway 1940; Armstrong 1975). Marsh
plants have well developed aerenchyma and their root cells no longer depend on
diffusion of oxygen from the surrounding soil, which is the main source of root
oxygen in terrestrial plants. However, Bartlett (1961) found that terrestrial
plants
vary widely in their resistance to waterlogging and that this resistance was
linked with
the capacity of the root to oxidize the rhizosphere. Terrestrial plants respond
to
oxygen stress in the roots by forming intracellular gas spaces in the cortex
(Bryant
1934). Bryant (1934) suggests that limited amounts of oxygen may be transferred
from the shoot to the root to enable terrestrial plants to survive short periods
of
waterlogging.
Reduced soil environments have been associated with limited
productivity and
growth in Spartina alterniflora (Teal and Kanwisher 1965; Linthurst 1979;
Mendelssohn and Seneca 1980; Howes et al. 1981; Mendelssohn and McKee 1988)
and two European marsh species Spartina townsendii and Molinia caerulea (Goodman
and Williams 1961).
1.1.2 Hydrogen Sulfide
When oxygen is limiting, microbial communities use other terminal
electron
acceptors to break down organic matter (Turner and Patrick 1968), converting the
soil
to a biochemically reduced state and producing end products which are
potentially
toxic to wetland vegetation. Hydrogen sulfide (H2S), an end product of sulfate
reduction, has been shown to be a phytotoxin in wetland plants (Goodman and
Williams 1961; Havill et al. 1985; Koch and Mendelssohn 1989; Koch et al. 1990).
Linthurst (1979) found that growth of S. alterniflora in a greenhouse experiment
was
negatively correlated with hydrogen sulfide concentration of the sediment. Also
in a
greenhouse experiment, Koch and Mendelssohn (1989) directly added sulfide to
intact
cores of a salt marsh species, S. alterniflora, and a fresh marsh species,
Panicum
hemitomon, to determine the effects of sulfides on growth. Sulfide
significantly
reduced culm, root, and rhizome biomass in P. hemitomon, but only root biomass
in
S. alterniflora, suggesting interspecific difference in tolerance levels.
Although some of the effects of sulfide on the growth of some marsh
plants are
known, the precise physiological mechanism whereby sulfide exerts its effect is
not
clear. Recent investigations have attempted to clarify the growth limitation
mechanisms of sulfide toxicity. In a greenhouse experiment, Koch et al. (1990)
found
that the effect of sulfide results from the inhibition of the anoxic generation
of energy
via alcoholic fermentation and a concomitant reduction in plant nitrogen uptake.
They
conclude that sulfide has a significant negative effect on the anoxic production
of
energy in roots via alcohol fermentation, leading to inhibition of plant
nitrogen
uptake.
Sulfate reduction to sulfide occurs mainly in the top 30 cm of New
England salt
marsh soils (Howarth and Teal 1979), Georgia salt marshes (King et al. 1982;
Howarth and Giblin 1983), and Long Island Sound sediments (Martens and Berner
1977). Howarth and Teal (1979) found that the rates of sulfate reduction tended
to be
slightly lower in the top few centimeters than between 4 and 18 cm, even though
all
the sediments below the top few millimeters were reducing.
Spatial variability in the rates of sulfate reduction is thought to
be due to the
amount of organic matter available to fuel sulfate reduction (Martens and Berner
1977; Howarth and Teal 1979) and gradients in tidal water movement (King et al.
1982). Howarth and Teal (1979) suggest that the growth and decomposition of the
belowground biomass provide a large annual input of organic carbon throughout
the
top 20 cm of the peat resulting in high sulfate reduction rates over depth.
King et al.
(1982) found higher sulfide concentrations in sites which had the least water
movement, mainly areas away from the creekbank and at a higher elevation.
Sulfate reduction has also been shown to be temporally variable.
Sulfate
reduction rates in marsh soils have been shown to vary on a daily and monthly
basis
(Howarth and Teal 1979; Peterson et al. 1983). In a New England salt marsh,
Howarth and Teal (1979) showed that for three years the average sulfate
reduction for
all sediment depths had maxima during the late summer and fall, and minima in
the
winter. However, the seasonal trend was not entirely explainable by temperature
alone. The authors concluded that sulfate reduction in the peat is controlled
by
substrate availability (i.e., energy source), not by the amount of SO42-.
Almost no information exists in the literature
describing the effect of sulfate
reduction in forested areas fringing salt marshes. This is probably because
forest
sediments rarely experience reducing conditions conducive to sulfate reduction
despite
occasional sulfate influxes and flooded conditions from seawater during extreme
flooding events. Unlike strictly terrestrial systems where sulfate can be but
is rarely
limiting (Coleman 1966), forested areas surrounding marshes probably are never
sulfate limited due to these influxes, especially areas which experience
flooding more
frequently. No information exists on whether these flooding events contribute
to the
sulfur cycling in forests which fringe salt and brackish marshes and whether the
cycling is significant from an ecological perspective.
Very limited information exists on the effects of sulfate reduction
on the growth
of forest species. During photosynthesis, SO42- and HSO3- are reduced and many
species can emit the end product as sulfide gas (Winner et al. 1981). The
majority of
studies which have been conducted on sulfate reduction are primarily concerned
with
sulfide emission rates from the leaves of green plants (Winner et al. 1981), not
soil
sulfide toxicity. However, Spaleny (1977) added K2SO4 to spruce seedlings to
determine the effect of increased levels of sulfate in the soil on sulfide
production.
He found that the addition of sulfate caused a metabolic inhibition resulting in
a
brown necrosis of the needles, similar to field observation in sulfate rich,
polluted
areas. Spaleny (1977) concludes that increased levels of soil sulfate due to
pollution
would considerably damage spruce seedlings.
1.1.3 Salinity
High soil salinities (e.g., 35-50 ppt) are hostile to most vascular
plant species
thus excluding all but the most tolerant species known as halophytes. The
physiological mechanisms of halophytes differ from other plants since salt
accumulation in the cells can lead to toxic ionic imbalances. Halophytes can
withstand salinity by a combination of mechanisms of quantitative or qualitative
salt
exclusion at the roots (Smart and Barko 1978; Antlfinger and Dunn 1983), and the
leaves (Anderson 1974), and by a compartmentalization of salts within cells
(Jefferies
1981).
The dominant factors which control the salinity of salt marsh soils
are
frequency and duration of tidal flooding (Wells 1928; Chapman 1940; Hinde 1954;
Chapman 1960; Adams 1963), incidence and amount of rainfall, and evaporation
(Chapman 1960; Ranwell et al. 1964; Mahall and Park 1976). Other factors which
have been shown to influence soil salinity are soil texture (Chapman 1960; Smart
and
Barko 1978), depth to the water table (Penfound and Hathaway 1938), fresh
groundwater inflow (Chapman 1960; Lindberg and Harriss 1973; Harvey et al.
1979),
and proximity to a tidal creek (Chapman 1960). It is generally thought that
soil
salinity concentration at any one point in the marsh is influenced by a
combination of
one or more of these factors.
Research into the spatial patterns of soil salinity from the marsh
landward in
tidal salt marshes has shown that soil salinity increases landward to a maximum
at or
about mean high water and then gradually decreases (Gilham 1957a; 1957b; Kurz
and
Wagner 1957; Adams 1963). In contrast, other researchers indicate that salinity
decreases from the edge of S. alterniflora zone landward without an increase at
mean
high water (Penfound and Hathaway 1938; Reed 1947). However, Reed (1947)
studied marshes with sediments that had a thin layer (2-25 cm) of fine sand on
the
surface, and found that drainage increased landward. Reed (1947) suggests that
the
improved infiltration and drainage landward lead to decreased salinities.
The mechanisms behind the spatial patterns of sediment salinity are
thought to
be related to the extent of tidal flooding which gradually decreases from the
marsh
landward up to mean high water. Because the presence of salt in the soil above
mean
high water is transported by only the higher high tides, soil salinity must
decline with
increasing distance landward. However, the lower frequency of flooding above
mean
high water increases the influence of evapotranspiration and rainfall as
controls on
salinity.
These controls on salinity are highly variable temporally, and can
lead to
extremely high salinity in the summer when rainfall is minimal and
evapotranspiration
is high (Ranwell et al. 1964; Mahall and Park 1976). Nestler (1977) found that
interstitial salinities dropped in the Juncus portion of the high marsh
following a
rainfall, but increased later in drought situations. He concluded that high
interstitial
salinities indicated less frequent tidal flooding.
Studies of the soil salinity patterns in tidal marshes have also
been shown to
vary with depth of sediment (Christian et al. 1978). In a tidal salt marsh
along the
northeastern Gulf coast of Florida, Lindberg and Harriss (1973) found that the
sediment surface had the highest concentration of pore water salinity.
1.2 Effects of Slope
1.2.1 Effects on Hydrology
Slope has the potential to control the magnitude and direction of
surface and
subsurface water movement. The effect of slope on hydrology characteristics is
paramount since the transport of all nutrients, metabolic toxins, and solutes is
hydrologically mediated. In general, salt marsh sediments may receive water by
runoff or regional groundwater aquifer discharge from upland areas, direct
precipitation, or from tidal flooding (Figure 1). Water is lost in the sediment
by
evaporation, transpiration, or drainage through the sediment and out into the
tidal
creek. The degree of slope has the potential to influence water movement in the
form
of runoff or tidal flooding, drainage, and groundwater discharge.
Slope, which has both an elevation component and distance
component,
influences the extent of flooding by tidal water since surficial water is
physically
restricted by friction caused by vegetation as it travels over longer distances
from the
tidal creek.
Figure 1. Elements of a water budget in a tidal salt marsh as they
are affected by
slope. Slope has the potential to control water movement
in the form
of runoff, drainage, or discharge, and is indicated with an
asterisk
(modified from Nuttle 1988).
This restriction is analogous to the non-channelized overland sheetflow
situation for
flooded riparian systems and can be calculated using the Manning equation
(Mitsch
and Gosselink 1986). Because the slope component in the Manning equation is
represented by a square root, this suggests that very low gradients of slope
would
result in proportionately higher restrictions to sheet flow, all other variables
being
equal. The restriction due to vegetation may explain why Jackson (1952) found
that
the extent of land inundated by spring tides was less than based on elevations
alone.
Conversely, if the degree of slope is high, upland runoff would be expected to
increase.
The amount of water movement in sediments depends on the sediment
hydraulic
conductivity and the hydraulic gradient. In sediments with relatively high
hydraulic
conductivity, slope is positively correlated with subsurface drainage. For
example,
Harvey (1990) found that drainage had a strong positive correlation with slope
for a
tidal freshwater marsh in Virginia. These findings supported modeling results
which
determined that horizontal hydraulic gradients highly correspond to topographic
gradients in tidal marshes (Harvey et al. 1987). However, these studies were
conducted in marshes which had relatively high soil conductivities (7.4 x 10-4
cm s-1;
Harvey et al. 1987). Sediments with lower hydraulic conductivities (0.05 x 10-4
cm
s-1, Harvey 1990) restrict water movement and require a larger hydraulic
gradient for
drainage to overcome this restriction (Harvey and Odum 1990).
Drainage in tidal salt marshes may be significant for sediment in
areas which
are near a tidal creek. For example, Nuttle (1988) found that there is
essentially no
horizontal water movement greater than 15 m from the creekbank in a New England
salt marsh. Without substantial drainage, the sediment at the interior of the
marsh is
mainly influenced by evapotranspiration (Hemond and Fifield 1982) and
precipitation
(Ranwell et al. 1964). Although drainage often constitutes a small proportion
of
water loss from tidal marsh soils when compared to evapotranspiration, solutes
(e.g.,
salt) are exported in drainage water and not evapotranspiration. This suggests
that
areas which experience less infiltration will have higher solute concentrations.
At the upland border, the influence of groundwater becomes
increasingly
important. Groundwater from regional aquifers is shown to have a large upward
hydraulic gradient beneath the hillslope and marsh (Harvey et al. 1988; Harvey
and
Odum 1990). However, the interaction between groundwater from hillslopes and
pore water from adjacent tidal marsh soils is highly dependent on sediment
composition. For example, highly conductive aquifers of glacial till deposits
in
Sippewissett, MA, which are composed of low organic coastal sands, had large
groundwater discharge (4397 L m-1 d-1) (Valiela et al. 1978).
In contrast, groundwater discharge is much less (6-10 L m-1 d-1)
(Harvey and
Odum 1990), and constitutes a small component (<10%) of the subsurface water
budget in sediments where clay or organic muds are present in substantial
proportions, as in the coastal plain of Virginia (Harvey et al. 1988). Because
of this
low sediment hydraulic conductivity, vertical gradients of groundwater discharge
were
not detected within 2 m of the surface of the marsh and the upward hydraulic
gradient
diminished exponentially as a function of distance into the marsh (Harvey et al.
1988).
1.2.2 Effects on Vegetation
Ultimately, landscape slope has the potential to affect the
distribution and
abundance of marsh and upland plant species by its effect on physicochemical
variables. Based on previous studies in low and high marshes, it is expected
that pore
water salinity, sulfides, and redox potential should play a significant role in
determining plant species composition in the high marsh-upland transition zone.
Since many of these abiotic variables are interrelated, a simple hierarchical
conceptual
model was created to illustrate the connections and relationships between the
abiotic
variables and vegetation as influenced by slope (Figure 2). Each of these soil
physicochemical variables is heavily influenced by differences in hydroperiod.
It
should be noted that the conceptual model was created to illustrate the
potential effects
that slope has on vegetation and is not meant to be complete since it lacks
obvious
feedback mechanisms.
Along the continuum from marsh landward, vegetation becomes less
adapted to
flooding by brackish water and its related effects. For example, halophytic
vegetation
(e.g., S. alterniflora) can withstand high pore water salinities, shrubs (e.g.,
Iva
frutescens L.) can tolerate intermediate salinities, and trees (e.g., Pinus
taeda L.)
Figure 2. Simple conceptual model of the effects of slope on the
hydrology and
pore water and vegetation used to formulate hypotheses.
have a limited capacity to tolerate saline water. Thus, the degree of slope
may
ultimately affect which types of vegetation can survive in a given area due to
the
effects on soil properties.
The degree of slope may also have a profound impact on the response
of
wetland and forested areas to a rising sea level. Brinson et al. (in
preparation)
suggest marsh migration over land (i.e., transgression) may be influenced by the
slope
of the antecedent landscape. Marsh areas which abut steep sloped uplands are
expected to stall the transgression process whereas marshes which abut gradual
or flat
sloped uplands would undergo more rapid transgression. Thus, the effects of
slope
are likely to play an integral role in the dynamics of marsh response to sea
level
fluctuations.
1.3 Objectives and Hypotheses
The overall objective of this research is to determine the effects
of slope on
abiotic and biotic variables along a brackish marsh-upland continuum.
Specifically,
the objectives are to assess the effects of slope on sediment physicochemical
variables
(i.e., hydroperiod, pore water salinity, redox potential, and sulfides) along
gradients
of steep and flat slopes; to assess the distribution of vegetation communities
along the
same continuum; and to correlate the soil variables with the vegetation
distribution.
Because vegetation distribution responds to abiotic variables, and
these variables
are affected by slope, slope effects are manifest in the position of vegetation
zones.
In particular, the position of the transition zone between marsh and upland is
indicative of the most landward extent of salt water influence and the most
seaward
extent of terrestrial influence. The position of the transition zone integrates
these
influences and thus represents a common point of reference to compare the
effects of
slope.
I used the position of the marsh-upland transition zone as a frame
of reference
to formulate testable hypotheses on the effects of slope on the following
variables: (1)
elevation of the transition zone, (2) hydroperiod, (3) salinity, (4) redox
potential, and
(3) sulfide concentration. Care was taken in constructing the hypotheses to
address
variables separately since many of the abiotic variables were expected to be
related.
The null hypotheses in each case were that each variable would be the same in
steep
and flat transition zones.
The transition zone of the flat transect is hypothesized to occupy
a lower
elevation, but at a greater distance from the source of tidal flooding for a
flat slope
because the resistance to tidal flow increases as distance increases. The
distance to
the tidal source is expected to ultimately affect many of the pore water
variables
because of its affect on hydrologic controls. The flat transition zone is
expected to
have a lower hydraulic gradient resulting in decreased potential for drainage
and thus
is hypothesized to exhibit a greater hydroperiod. Although not directly related
to
drainage, the flat transition is hypothesized to exhibit a higher flood
frequency from
extreme high water events because of its lower elevation in relation to mean
high
water. The effects of hydraulic gradient on drainage is expected to play a
major role
in influencing pore water characteristics. It is hypothesized that pore water
salinities
of a steep slope will be lower than those of a gradual slope because of lower
potential
for drainage and higher potential for evapotranspiration. The effect of slope
on
drainage potential is also hypothesized to result in lower pore water redox
potentials
and higher concentrations of hydrogen sulfide for a flat slope.
The study of the factors which affect vegetation dynamics is
usually conducted
at the ecological (i.e., less than a decade) or paleoecological (i.e., centuries
to
millennia) time scales. The vegetation dynamics between these time scales
(i.e.,
decades to centuries) is difficult to obtain since it is limited by methodology.
Long
term records by pollen analysis usually lack the spatial resolution of pattern,
and
vegetation analyses address the spatial scale at which individuals interact, but
lack the
historical perspective. Thus, this work establishes the initial effort of
characterizing
the abiotic and biotic patterns along transitions from marsh to upland within
the
framework of long term ecological research.
2. SITE DESCRIPTION AND METHODS
2.1 Site Description
The study area, Phillips Creek Marsh (37x 27', 75x 50'W), is
located at the
Virginia Coast Reserve (VCR-LTER) in Brownsville, VA, and is owned by The
Nature Conservancy (Figure 3). The site is an expansive brackish marsh on the
mainland side of the coastal lagoon complex on the Eastern Shore of Virginia.
Flooding of the marsh is from the southwest by the tidal creek, Phillips Creek,
which
has a tidal range of about 145 cm. The average monthly rainfall is 86 mm, with
annual maximum rainfall in July and August due to conductive storms (USDA 1989).
The stratigraphy of the coastal plain of Virginia is composed of
sands and clays
of Cretaceous age or younger that were deposited atop bedrock in fluvial,
estuarine,
or marine environments (Powars et al. 1988). The water table aquifer is
typically 3
to 20 m thick and is underlain by the semi-confining layers of clayey-sand
(Pliocene
age) that contain fresh interstitial water or groundwater (Cederstrom 1943, as
cited in
Harvey 1990). The surficial pore water of marsh soil is composed of deposits of
Holocene age or younger.
The soil types along the shoreline range from very poorly to poorly
drained and
consist of Chincoteague (low marsh), Magotha (high marsh), and Nimmo, Munden
and Bojac (upland) (USDA 1989). The dominant soil assemblage along the coast of
Northampton County, VA, consists of Chincoteague-Magotha-Nimmo and
Chincoteague-Magotha-Munden soil types (Figure 4). These soil assemblages
account
Figure 3. Location and arrangement of study transects at the
Brownsville, VA,
marsh representing Steep Near (SN), Steep Far (SF), Flat
Near (FN),
and Flat Far (FF) transects in relation to Phillips Creek.
Figure 4. Frequency of occurrence for shoreline to upland soil
assemblages for
Northampton County, VA, using soil maps (USDA 1989).
Capital
letters designate abbreviations of soil types from shore to
upland. Soil
abbreviations are: tidal marsh soil (low marsh)
Chincoteague (C), and
Fisherman (F); intertidal (high marsh) Magotha (M), Camocca
(Ca),
and Udorthents (U); and upland (forest) Nimmo (N), Munden
(Mu),
Bojac sandy (Bk), Bojac fine sandy (Bo), Dragston (D),
Seabrook (S),
Molena (Mo), and Polawanna (P). Soil assemblage
frequencies were
determined by drawing a perpendicular line from the marsh
edge to
upland for every 300 m of shoreline and recording soil
types. for over 60% of the shoreline soils. The principal soil types for the
study site consist
of Chincoteague, Magotha, Nimmo, and Munden.
The Magotha soil series is very deep, fine sandy loam, and poorly
drained.
Magotha soils are more loamy in the upper part of the soil than Chincoteague
soils,
have higher levels of sodium than Nimmo soils, and are grayer than Munden soils.
The Nimmo series are also very deep, poorly drained and are grayer than Munden
soils. Munden soils are moderately well drained and have moderately coarse
textured
sediments.
The brackish marsh is surrounded by forested uplands to the north
and west,
and by an agricultural field to the south (Figure 3). The marsh grades
gradually into
the forested areas to the north and steeply in the south.
Well developed low and high marsh plant communities occur in the
study area
with gradients of high marsh-upland transition zones. In general, the high
marsh is
dominated by Spartina patens, J. roemerianus, Panicum virgatum, and Distichlis
spicata. In the transition zone, these species are joined occasionally by the
shrubs I.
frutescens, Baccharis halimifolia, Myrica cerifera, and small (< 3 m) tree
species
Juniperus virginiana. The transition is also characterized by many dead red
cedar
trees and stumps, especially along gradual slopes, similar to high marsh forest
transitions from Long Island, New York (Clark 1986a; 1986b), to Florida (Kurz
and
Wagner 1957). The forest species are predominantly pines such as P. taeda, and
large J. virginiana.
2.2 Study Transects
Four linear transects perpendicular to the high marsh-upland
transition zone
were established to represent extremes of slope (i.e., flat or steep), and
distances
from a tidal source (i.e., far or near) (Figure 3). Each combination of slope
and
landscape position: Steep Near, Steep Far, Flat Near, and Flat Far was
represented by
a different transect (Figure 5). However, this designation does not represent a
complete array of slope and landscape positions since some arrangements (e.g.,
flat
transition immediately adjacent to a tidal creek) are improbable.
Transect lengths were established initially by choosing common end
members
(i.e., predominance of high marsh vegetation and distance to tidal creek for the
beginning of the transect, and lack of marsh vegetation cover for the end of the
transect) to represent starting and ending points. This designation inherently
may
misrepresent the actual length of the transect since starting and ending points
are
arbitrary. However, I felt that capturing the transition between marsh and
forested
vegetation types was more important than the actual length of the transect. As
such,
it should be noted that the actual length of the study transects could differ
considerably (q 20-50 m) if others had chosen it. The starting point was at
approximately 50 m away from Phillips Creek for Steep Near, 200 m for Steep Far,
150 m for Flat Near, and 250 m for Flat Far. Transect distances were numbered
to
begin at 0 and increased sequentially toward the forest area.
Vegetation cover, elevation, sediment bulk density, percent carbon,
percent
Figure 5. Possible combinations of landscape position in relation to
proximity to
tidal creek and gradients of slope used to choose study
transects. water, chlorinity, and depth to mineral sediment were assessed once
for all transects
and are described in section 2.3. Two of the four transects, Steep Near and
Flat
Near, were instrumented at sampling stations to measure hydrologic and pore
water
variables. Water table position, pore water salinity, redox potential, acidity,
and
sulfide concentration are described in section 2.3. Ten sampling stations per
transect
were installed and the distance between sampling stations was arbitrarily chosen
to
reflect the width of the transition zone and the total length of the transect
(Figure 6).
In general, sampling was conducted monthly during the cool season and twice
monthly for the warm season for up to 2 years.
2.3 Methods
2.3.1 Vegetation Sampling and Analysis
2.3.1.1 Vegetation Sampling
The percent cover of grasses, shrubs, and trees was assessed
continuously along
study transects using a plot method. A quadrat was created by placing meter
tapes on
the ground to create a 2 x 4 m frame. The frame was sub-divided into a 1 x 1 m
frame to measure herbaceous ground cover, 2 x 2 m frame to measure shrub cover,
and a 2 x 4 m frame to measure tree cover (Figure 7a) (after Mueller-Dombois and
Ellenberg 1974). Percent cover for trees and shrubs was directly estimated
within the
sampling frame by calculating the total area occupied for each species divided
by the
total area. For herbaceous species, cover was difficult to judge for
Figure 6. Layout of the Flat Near and Steep Near study transects
indicating
locations of the surficial groundwater wells, lysimeters,
and water level
recorders. (A) Flat Near transect. (B) Steep Near
transect.
Figure 7. Sampling frame for vegetation analysis. (A) Frame size and
geometry
used for vegetation sampling. Frame size for measuring
vegetation
were: herbaceous (H) 1 m2, shrub (S) 4 m2, and trees (T) 8
m2. (B)
The moving split-window for analysis of vegetation using a
six sample
window (asterisks designate sample points). The mean
vegetation
cover for each species is computed for each window half and
the
dissimilarity is calculated between windows. The window is
then
moved along the transect one sampling point at a time
(after Johnston
et al. 1992).
individual species and dominance was assessed instead. An algorithm converting
dominance to cover was then used. The algorithm was based
on field observations
between cover and dominance (Table 1). Vegetation layers
were chosen (after
Mueller-Dombois and Ellenburg 1974) to fall into 3
categories: (1) herbaceous layer
(<30 cm to 1 m); (2) shrub layer (>1 m to 3 m); and (3)
tree layer (>3 m).
Species were identified using Radford et al. (1968).
2.3.1.2 Vegetation Analysis
The moving split-window analysis was used to reveal vegetation
discontinuities
which were then used as a basis for partitioning study transects into three
zones: (1)
high marsh, (2) transition, and (3) forested/upland. The moving split-window
technique identifies boundaries by peaks in squared euclidean distance (SED)
(Wierenga et al. 1987; Ludwig and Cornelius 1987).
The SED is calculated by placing a double window over equally
spaced
sampling points (a sampling point is the total cover of all species types in one
2 x 4 m
frame), and the dissimilarity (distance) between attribute values in each window
half
is statistically compared (Figure 7b). The window is moved sequentially along
the
transect until all comparisons are made for the entire length of the transect.
The
mean attribute value is calculated for each window half and the SED is computed
as
the square of the difference between the means of each variable (i.e.,
individual
species cover) in adjacent windows, summed across all variables measured (i.e.,
all Table 1. Algorithm used to convert dominance rankings to
percent cover.
Algorithm was based on field observations between cover and
dominance.
Species Dominance Rank Multiply total
cover by:
Example:
total cover=80%
AA= 11.0A= 80% cover
A > BA= 1
B= 2 0.75
0.25 A=
60% cover
B=
20% cover
A > B > CA= 1
B= 2
C= 3 0.60
0.30
0.10 A=
48% cover
B=
24% cover
C=
08% cover
A = BA= 1
B= 1 0.50
0.50 A=
40% cover
B=
40% cover
species) (Brunt and Conley 1990; Johnston et al. 1992). Peaks in the SED are
then
used to indicate boundary locations (Wierenga et al. 1987).
Although the data collection methodology was identical for all
study transects, it
became apparent that the analytical window size would have to be sensitive with
regard to the overall transect length. Thus, the analysis of the steep gradient
transects
necessitated a smaller analytical window width (e.g., 4) than the flat transect
(e.g., 8
or 10) by nature of its shorter overall length. A smaller window width is more
sensitive to small changes in species composition (e.g., for one life form) and
can
cause greater noise (i.e., many peaks which do not correspond to vegetation
zones).
A larger window may include two or more boundaries and may obscure boundary
locations. Thus, after trying a range of window widths, I chose the analytical
window sizes to best represent field observations.
2.3.2 Surface Elevation
Elevations along study transects were determined using a laser
level (Pentax
Total Survey Station III, Model #5, Asahi Precision Co., Tokyo, Japan), data
logger
(Pentax, SC-5, Pentax Instruments, Englewood, CA) and reflecting prism. Optical
leveling used to augment the laser leveling in areas where views were obstructed
using an auto level (Topcon Model AT-F2, Tokyo Optical Company, Tokyo, Japan)
and stadia rod. All elevations were referenced to the Hayden Benchmark
(Virginia
Coast Reserve benchmark HAYD, N 372732.021 W 754958.036) located adjacent to
Phillips Creek creekbank water level recorder (Figure 6). This standardized
elevations for all transects relative to the benchmark and facilitated elevation
comparisons between transects.
Slope for the entire transect was calculated using linear
regression and
represents the change in elevation (m) divided by change in distance (m). Slope
is
reported as m m-1, and thus is unitless. The conversion to percent slope can be
calculated by multiplying the slope number by 100, and conversion to degree by
multiplying the percent slope by 0.9x.
The humocky nature of the surface elevations were assessed by
measuring
surface macrotopography (i.e., at the scale of 5-10 m). Macrotopography along
the
transects was determined by the departure from linearity of the linear
regression line.
The regression line would be expected to have a better fit (i.e., higher r2
value) to the
surface elevations if the transect profile departed little from the regression
line.
However, the r2 value from regressions of data having zero slope (e.g., the flat
slope)
is zero, so the root mean square error term was used instead. The regression
line
would be expected to have a better fit to the surface elevations (i.e., a lower
root
mean square error) if the transect departed little from the regression line. A
hummocky area would be expected to have a higher root mean square error. Thus,
the root mean square error value was used as a surrogate for actual small scale
macrotopographic measurement. It should be noted that this is a crude
estimation of
the surface macrotopography and is only used as a comparison between transects.
2.3.3 Soil Characteristics
Core samples were obtained for quantitative analysis using a thin
walled cork
borer (size #12, 2 cm o.d.) in August 1993. Samples were obtained by excavating
a
small soil pit (approximately 30x30x30 cm) by shovel and extracting cores from
the
vertical faces of the pit. Core samples were obtained with minimal compression
by
rotating the corer slowly into the sides of the pit at the appropriate depth
(i.e., 10 and
20 cm). Core samples were extruded and immediately placed in Ziploc bags.
Depth
to mineral soil was measured directly from the sides of the of the pit by visual
approximation and meter stick. Cores were taken every 20 m for the Flat Near
transect, every 10 m for the Flat Far, and every 5 m for the Steep Near and
Steep Far
transects.
Soil samples for bulk density and percent water were dried to a
constant mass
(105xC) for 48 h. Soil samples were allowed to cool inside a desiccator prior
to
weighing. Soil bulk density was calculated as the sample dry mass divided by
the
volume (SSSA 1986). Percent water was calculated by the mass of the wet soil
divided by the mass of the dry soil less 1, multiplied by 100 (SSSA 1986).
Soil samples for total carbon and chlorinity were air dried for 72
h, milled by
mortar and pestle, and sieved through 2 mm copper screen to remove macroorganic
matter (i.e., large roots and rhizomes). Total carbon was determined using a
H-C-N
elemental analyzer (Leeman Labs, Model CE 440, Lowell, MA) and was divided by
the sample mass to obtain percent total carbon. Percent total carbon was chosen
over
organic matter as a measure of organic carbon in the soil because: (1) the
amount of
inorganic carbon in the soil (i.e., mainly dolomite) is very low or non-existent
(as
determined by the Test for Presence of Inorganic Carbon using concentrated HCl,
SSSA 1986, p. 563); and, (2) the conversion value from organic carbon to organic
matter is inexact. If conversion to organic matter is desired, a rough
estimation
would be to multiply the total carbon figures by a factor of two (SSSA 1986).
An
attempt was made to mill all material of the soil sample, but large organic
clumps
(i.e., large rhizomal pieces and woody fragments) were unable to be used by the
HCN analyzer. It should be noted that the actual total carbon may be
underestimated
for soil samples which contained mostly large rhizomal material. This
underestimation would be most likely affect the 10 cm depth soil samples and at
locations closer to the transect beginning (e.g., 0 m).
Soil chlorinity was determined by placing a known volume of milled
sediment
into large (50 mL) centrifuge tubes. A known volume of distilled water (10-15
mL)
was added and then shaken for 48 h to assure mixture. The resulting mixture had
a
volumetric soil-water ratio of 1:2. The samples were then centrifuged for 2 h
at
3,000 rpm and allowed to settle. The supernatant was then automatically
titrated
using an automatic/amperometric Cl- titrator (Buchler Chloridometer Model
4-2500,
Fort Lee, NJ), calibrated with known standards. Each sample was titrated twice.
2.3.4 Water Table Fluctuations
The Flat Near and Steep Near transects were instrumented with
shallow wells to
monitor the position of the water table during sampling dates and with water
level
recorders to monitor fluctuations in the water table continuously. The wells
were
evenly spaced along the transect at 5 m intervals for the steep transect (n=11)
and at
approximately 25 m intervals (n=9) for the flat transect (Figure 6). Water
level
recorders were installed in the transition zone for both study transects (Figure
6).
The position of the water table was monitored using wells
constructed of PVC
pipe (4.8 cm diameter) slotted from top to bottom and inserted 1.5 m into the
substrate (after Bouma et al. 1980). Wells were screened with either multiple
wraps
of fiberglass window screening, or nylon stockings, and were augered to depth,
backfilled, and sealed with Bentonite to prevent water from entering from the
top.
Distance from the top of the tube to the water table was measured to the nearest
mm
by meter tape. The water table was rarely more than 1.2 m and could be easily
seen.
For continuous monitoring of the water table, one water level
recorder (Stevens
Type F) was installed in the transition zone for each of the Flat Near and Steep
Near
transects. The water level recorder was installed on 11 July 92 for the Flat
Near
transect, and 18 June 93 for the Steep Near transect. A stilling well (20.5 cm
o.d.)
slotted and screened with fiberglass window screening from top to bottom was
inserted 2 m into the substrate to allow the float to travel beneath the surface
undisturbed. Water table position was determined at the start and end of each
chart
period (i.e., 8-32 d) by measuring the distance from the water table to a fixed
point
on the water level recorder stand. The charts were then digitized and converted
to
date/times and water levels using a video recorder and JAVA software (JAVA,
Jandel
Scientific, San Rafael, CA).
Seasonal patterns of drawdown were estimated for the entire study
area using
simple water balance estimates. Periods with a high potential for water table
drawdown were identified on the basis of precipitation totals and mean monthly
potential evapotranspiration (PET) for two periods: (1) long term (1949-1981)
data set
for a town nearby the study site, Cheriton, VA (USDA 1989); and (2) a short term
(1990-1993) data set collected on-site by a meteorological station (Krovetz and
Porter
1993). PET was calculated using the Thornthwaite method (Thornthwaite and
Mather
1957). Differences between monthly precipitation totals, mean monthly PET and
the
frequency of deficits of precipitation relative to PET were calculated for all
months
for the study period (1990-1993).
2.3.5 Groundwater Salinity
The salinity of the surficial groundwater was assessed using the
same wells for
determining water table characteristics. Groundwater salinity was determined
using a
S-C-T meter (Yellow Springs Instrument, Model 33, Yellow Springs, OH) which was
calibrated with known standards. Salinities were taken at the water table
surface and
at the bottom. Values reported are the mean of top and bottom measurements.
2.3.6 Pore Water
Soil pore water was collected for chemical analysis using
low-tension lysimeters
(after Harvey 1990). The lysimeter allowed collection of pore water at a single
level
per instrument, at depths ranging between 10 and 60 cm. The body of the
lysimeter
was constructed of PVC pipe (5 cm o.d.) with the inlet capped with 70 fm porous
nylon fritware. Lysimeters were installed vertically into the soil with the top
stoppered and fitted with a sampling port and a gas port. Prior to sampling, a
hand
vacuum pump (Nalgene) was used to remove all water from the instrument and
discarded. The instrument was then flushed with nitrogen gas to replace the
headspace gas and evacuated. Pore water was collected about 12 h later with a
60
mL syringe. Shorter intervals between evacuation and collection were attempted,
but
failed to collect appreciable amounts of pore water due to very low soil
hydraulic
conductivity. Pore water samples were used to determine salinity, redox
potential,
hydrogen sulfide, and pH.
Pore water samples for salinity and pH determinations were
collected in 20 ml
glass scintillation vials, stoppered, and brought back to the lab and measured
immediately. Salinity was measured directly using a temperature compensated
refractometer (Reichert, Model 10419). The pore water pH was measured using an
automatically temperature compensated pH meter (Fisher Model 1002, Springfield,
NJ) and gel filled combination pH/ATC probe (Fisher 74613, Springfield, NJ). A
two point calibration method was used and the electrode was recalibrated for
each
new sample.
Redox potential was measured in the field using two kinds of
electrodes: (1) a
commercially produced redox combination electrode (Orion H4402, Cambridge, MA),
and (2) a brightened platinum electrode and a calomel reference probe (Fisher
74613).
Sampler headspaces were kept purged with N2 prior to sampling. Pore water was
removed from the sampler by syringe. The syringe was attached to a modified
syringe chamber which contained the electrode (after Marcio Santos, University
of
Virginia, personal communication). The measurement was recorded after
stabilization
of the reading which occurred within about one minute after immersion of the
electrode. Readings were standardized using Zobell's solution and referenced to
the
standard hydrogen electrode (SHE) by adding 244mV. The redox values were not
normalized with respect to pH.
Hydrogen sulfide was determined colorimetrically (after Cline
1969). Pore
water used for redox determinations was carefully added to stoppered test tubes
which
contained Cline's reagent and kept on ice. After collecting, samples were
brought
back to the lab and read immediately on the spectrophotometer.
2.4 Data Analysis
In order to evaluate the hypotheses for the physical and chemical
variables as
stated, statistical tests were performed on the data. Prior to the use of any
statistical
test, the data for all variables were tested for the necessary assumptions of
normal
distributions and homogeneity of variances; in all cases, assumptions were
violated.
Transformations were attempted using a Box-Cox power transformation test, but
none
adequately met these assumptions. Thus, the data were rank-transformed (PROC
RANK, SAS User's Guide 1985) and all subsequent statistical tests were performed
on the ranked data. The RANK procedure produces rank scores across observations
which creates a normal distribution because the sapling distribution of ranks
are
equal.
Univariate analyses of variances (ANOVA) were performed separately
on the
variables that were measured once (i.e., elevation, carbon content, chlorinity,
percent
water, and bulk density) to test slope (i.e., flat, steep) and zone effects
(i.e, high
marsh, transition, and forest). Because of the small sample size for individual
transects, data were pooled for vegetation zone by slope (i.e., flat, steep) and
treated
as replicates within zone. The assumption that data from separate transects
could be
pooled by zone was supported by a randomized block design using the variable
elevation. Individual contrasts showed that no significant difference occurred
between
the transition zones for each type of slope. Significant differences occurred
between
the forest zone of the flat transects and the high marsh zone of the steep
transects, but
these areas had very small sample sizes (e.g., 2-4 sampling points) making
comparisons difficult. It should be noted that this assumption was tested for
elevation
only, but is probably tenable for other variables because they are heavily
influenced
by elevation.
Differences in the variables that were measured
repeatedly over time (i.e., pore
water salinity, acidity, redox potential, hydrogen sulfide, and water table
position)
were assessed using a randomized block design blocked on date. I chose to block
on
date because many of these variables would be expected to respond similarly to
date
effects. It should be noted that the assumption of repeated measures as
replicates is
somewhat problematic since the argument can be made that the replicates are not
strictly independent from each other. However, changes in these variables occur
at
time scales much less than the two week sampling interval and can, for the most
part,
be considered independent.
Many of the pore water variables and physical variables were
thought to be
interdependent so a factor analysis was used to determine common factors
underlying
the composition of multiple variables. Unlike the univariate analysis which
used
ranked data, the factor analysis was performed on actual scores. However, the
factor
analysis was used primarily for data reduction and distribution assumptions were
not
completely necessary. An orthogonal rotation method, Varimax factor rotation,
was
used to clarify factor interpretation (Duntemann 1984). Factor scores were then
used
as variables in ANOVA to determine if factor scores differed across zones and
sampling dates. Tukey's (HSD) test was used to isolate which means were
different,
if any. Unless specified, significant refers to p < 0.05, and highly
significant refers
to p < 0.01.
Two subsets of the data were analyzed using factor analysis: data
set for all
zones, and a subset consisting of the transition zone only. The subset of the
entire
data were analyzed to determine if properties unique to the transition zone
could be
found using a smaller scale. 3. RESULTS
3.1 Vegetation
3.1.1 Zone Detection and Delineation
The moving split-window analysis revealed vegetation
discontinuities which
were used as a basis for partitioning study transects into three zones: (1) high
marsh,
(2) transition, and (3) forested/upland. In general, the flat transects have
considerably
more peaks than the steep transects (Figure 8), making boundary determination
more
difficult. Thus, it became necessary to use field notes and observations to
"ground
truth" and augment the window analysis for the flat transects. A window width
of 4
sampling points was used for the steep transects and 8 sampling points for the
flat
transects. The effect of window width on peak identification is illustrated by
data for
the Flat Far transect. Varying the window width from 6 to 10 sampling units did
not
appreciably affect the interpretation of the boundary locations, only the
emphasis of
certain peaks (Figure 9).
3.1.1.1 Flat Near Transect
For the Flat Near transect, the analysis revealed 10 peaks between
0 m and 120
m (Figure 8a), but these were considered to represent mostly shifts in the types
of
herbaceous ground cover. For this transect, the zones were delimited as high
marsh
(0-120 m), transition (120-140 m), and forested (140-190 m).
Figure 8. Squared Euclidean distance computed using moving
split-window analysis.
Open circles represent sampling points. Window size for
analysis was 8 for
the Flat transects, and 4 for the Steep transects. (A) Flat
Near transect: high
marsh (0-120 m), transition zone (120-140 m), and forest
(140-190 m). (B)
Flat Far transect: high marsh (0-62 m), transition zone
(62-116 m), and forest
(116-130 m). (C) Steep Near transect: high marsh (0-12 m),
transition zone
(12-22 m), and forest (22-50 m). (D) Steep Far transect:
high marsh (0-20
m), transition zone (20-28 m), and forest (30-50 m).
Figure 9. Peaks in the squared Euclidean distance for the Flat Far
transect using
window sizes of 10, 8, and 6 sampling points.
The high marsh zone was predominantly comprised of D. spicata, S. patens,
and J. roemerianus which accounted for the majority of the total cover (Figure
10a).
In this zone, the shrubs I. frutescens and B. halimifolia, occasionally
appeared, but do
not make up a large part of the total cover. The landward extent of the high
marsh
and beginning of the transition zone was harder to detect over such a gradual
change.
A peak at approximately 120-135 m corresponded with field notes as the beginning
of
tree cover, as well as continued shrub cover and was felt to represent the
beginning of
the transition zone. The transition zone was comprised of the tree species P.
taeda
and J. virginiana, and the shrubs M. cerifera, and B. halimifolia (Figure 10a).
However, the herbs S. patens and D. spicata continued to be dominant along with
P.
virgatum especially in higher, hummocky areas.
The landward extent of the transition zone and the beginning of the
forested
area was delimited by a sharp peak around 140 m (Figure 8a). The forested
species
was predominantly P. taeda, although an occasional J. virginiana was also found
mostly in the marshward extent of the forested area. The shrub species M.
cerifera
was present along the edge of the forested area and was the only shrub species
found
in this zone. The ground cover was sparse and consisted predominantly Rhus
radicans and Uniola laxa.
3.1.1.2 Flat Far Transect
The Flat Far transect had similar numerous, steep peaks as the Flat
Near
Figure 10. Percent of the total cover for each species grouped by zone
and
vegetation type. Species abbreviations are as follows:
herbaceous:
A=Spartina alterniflora, D=Distichlis spicata, P=Spartina
patens,
J=Juncus roemerianus, V=Panicum virgatum, and O= Other;
shrubs:
I=Iva frutescens, B=Baccharis halimifolia, J=Juniperus
virginiana
(< 3 m), and M=Myrica cerifera; and forest: P=Pinus taeda,
and
J=Juniperus virginiana (> 3 m). (A) Flat Near transect. (B)
Flat Far
transect. (C) Steep Near transect. (D) Steep Far transect.
Figure 10. Concluded.
transect (Figure 8b). Thus for the Flat Far transect, the zones were delimited
as high
marsh (0-62 m), transition (62-116 m), and forested (116-130 m). The steep peak
at
the 70-75 m location was primarily due to a large bare patch and was not
considered
a transition area.
For the high marsh zone, the predominant ground cover was similar
to the Flat
Near transect and comprised of D. spicata, S. patens, and J. roemerianus (Figure
10b). Also present were P. virgatum, S. viridis, and Juncus spp., which were
not
found in the Flat Near transect. The shrubs present in the high marsh zone were
B.
halimifolia, M. cerifera, and small (<3 m) J. virginiana.
The landward extent of the high marsh zone and the beginning of the
transition
zone was delimited by a sharp peak at approximately 58 m (Figure 8b). The peak
corresponds to the emergence and dominance of tree and shrub species. In the
transition zone, the dominant tree species were P. taeda and large J. virginiana
(Figure 10b). The shrub M. cerifera and small (<3 m) P. taeda and J. virginiana
comprised all of the shrub species in the transition zone. The herbaceous
ground
cover was predominantly P. virgatum, S. patens, and Setaria viridis, although D.
spicata and J. roemerianus were present.
The large, steep peak at 116 m marked the end of the transition
zone and the
beginning of the forested zone (Figure 8b). The tree P. taeda and the shrub M.
cerifera, comprised the majority of cover in the forested zone (Figure 8b). The
ground cover was mainly R. radicans, and occasionally Campsis radicans.
3.1.1.3 Steep Near Transect
The peaks in the SED for the Steep Near transect were less numerous
and
broader than both of the flat transects (Figure 8c), making zone boundaries
easier to
define. The zones were delimited as high marsh (0-12 m), transition (12-22 m),
and
forested (22-50 m). The large peak at 44 m was primarily due to patchy bare
areas
and very dense pockets of vine species (e.g., R. radicans), and was not
considered to
be a separate zone.
For the Steep Near transect, the high marsh cover consisted of
mostly D.
spicata and S. patens with S. alterniflora appearing at the marsh end of the
transect
(Figure 10c). The herbaceous ground cover was joined by the emergence of the
shrub
species I. frutescens and B. halimifolia at 10 m.
The landward extent of the high marsh and the start of the
transition zone was
signalled by a large peak around 12 m (Figure 8c). The peak was primarily due
to a
change in the ground cover species, but also signalled the beginning of cover
for
forested species. S. patens and P. virgatum became the dominant herbaceous
species
(Figure 10c) in the transition zone. Shrub cover was predominantly B.
halimifolia
and M. cerifera. Tree cover was predominantly comprised of P. taeda and J.
virginiana, although these accounted for only a small amount of the overall
cover.
The edge of the forested ecotone and the marsh transition was
delimited by
another sharp peak at approximately 22 m (Figure 8c). The peak was mostly due
to
the shift of cover from shrub species to forest species, with all shrub species
terminating at 32 m. The dominant tree cover species was P. taeda, although few
very large (>3 m) J. virginiana were present (Figure 10c). The shrubs B.
halimifolia, M. cerifera and small J. virginana accounted for very little of the
total
cover. The ground cover was mostly R. radicans, Sagittaria latifolia, and P.
virgatum. The percent cover of these ground species was very sparse, with large
bare
patches and occasional dense pockets.
3.1.1.4 Steep Far Transect
The peaks in the SED for the Steep Far transect were very distinct
and less
numerous than any of the study transects. Peaks were delimited at 22 m and at
28 m,
and correspond to high marsh (0-20 m), transition (20-28 m), and forested (30-50
m)
vegetation zones (Figure 8d).
The high marsh consisted of mostly herbaceous ground cover with S.
patens
and D. spicata as the sole species (Figure 10d). The shrub I. frutescens was
also
present, but contributed little to the total cover.
The peak delimiting the landward extent of the high marsh and
beginning of the
transition zone was mostly due to the shifting of the total cover from
herbaceous
species to the shrub I. frutescens, which was the sole shrub member (Figure
10d). In
the transition zone, the herbaceous cover was dominated by S. patens and D.
spicata
with traces of P. virgatum.
The tree cover of the forested zone which began at 30 m was
comprised of
exclusively P. taeda (Figure 10d). The shrub species cover was comprised of M.
cerifera and small (<3 m) J. virginiana. The herbaceous ground cover was
similar
to the Steep Near, with P. virgatum, R. radicans, and S. latifolia. A vine, C.
radicans, and the grasses, U. laxa, and S. viridis, were also present and
together
comprised most of the ground cover.
3.1.2 Plant Species Presence
In total, twenty-four species were identified along the four
transects (Table 2).
Seven species could be found in all three zones and included the graminoids D.
spicata, P. virgatum, and S. patens, the shrubs I. frutescens, B. halimifolia,
M.
cerifera, and a tree species J. virginana. One species was unique to the high
marsh
(S. virginica) and one was unique to the transition zone (S. bona-nox). Eight
species
were unique to the forested areas and generally represent grasses and trees
which
cannot tolerate saline water. In general, the high marsh were the least rich,
the
transition were the next rich, and the forested areas were the most rich.
Comparisons of species presence and diversity between the gradients
of slope,
the flat slopes were slightly less rich than the steep slopes. J. roemerianus
was found
in the high marsh and transition on both flat slopes, but was noticeably absent
on the
steep slopes. Similarly, Limonium carolinianum was found along the steep
slopes, but
was absent on the flat transects.
Table 2. Vascular plant species in the Phillips Creek marsh. Zones
are
designated as HF: High marsh, flat transect; HS: High
marsh, steep
transect; TF: Transition zone, flat transect; TS:
Transition zone, steep
transect; FF: forest, flat transect; FS: forest, steep
transect. Presence of
a species in a zone is indicated by a "1" in the zone
columns. Species
nomenclature follows Radford et al. (1968).
Family Species
Common Name
Zone
HFHSTFTSFFFS
Poaceae
Distichlis spicata Greene
Panicum virgatum L.
Uniola laxa (L.) BSP.
Setaria viridis (L.) Beauv.
Spartina alterniflora Loisel
Spartina patens (Aiton) Muhl
Juncaceae
Juncus roemerianus Scheele
Juncus spp.
Cyperaceae
Fimbristylis spadicea Vahl
Scirpus olneyi Gray
Anacardiaceae
Rhus radicans L.
Plumbaginaceae
Limonium carolinianum
(Walter) Britton
Alismataceae
Sagittaria latifolia Willd.
Salt grass
Switch grass
Foxtail grass
Saltmarsh cordgrass
Saltmeadow cordgrass
Black needlerush
Marsh fimbristylis
Three-square sedge
Poison ivy
Sea lavender
Broad leaved arrowhead
1 1 1
1 0 1
1 0 1
1 1 1
0 0 0
0 1 0
1 0 0
0 1 0
0 1 0
1 0 0
1 1 1
1 1 1
1 0 1
0 0 0
1 1 0
0 1 0
1 0 1
1 0 0
0 0 1
1 0 1
0 0 1
0 1 1
0 1 0
1 0 0
0 0 0
0 1 1
Bignoniaceae
Campsis radicans (L.) Seeman
Chenopodiaceae
Salicornia virginica L.
Liliaceae
Asparagus officinalis
Smilax bona-nox L.
Asteracea
Iva frutescens L.
Baccharis halimiflora L.
Myricaceae
Myrica cerifera L.
Cupressaceae
Juniperus virginiana L.
Pinaceae
Pinus taeda L.
Rosaceae
Prunus serotina Ehrh.
Aquifoliaceae
Ilex opaca Aiton
Trumpet vine
Perennial saltwort
Asparagus
Greenbriar
Marsh elder
Groundsel tree
Wax myrtle
Red cedar
Loblolly pine
Black cherry
American holly
0 0 0
0 1 1
0 1 0
0 0 0
0 0 0
0 0 1
0 0 0
1 0 0
1 1 1
1 0 1
1 0 1
1 0 1
1 0 1
0 1 1
1 0 1
0 1 1
0 0 1
0 1 1
0 0 0
0 0 1
0 0 0
0 0 1
TOTALS24 Species11712101115
3.2 Elevation
3.2.1 Variation in Elevation by Transect
The study transects were chosen to represent flat and steep slope
gradients. In
general, the flat transects had negligible slopes and more topographic variation
than
the transects with steep slopes (Figure 11).
The Flat Near transect had a very negligible slope (0.29 x 10-3,
Table 3) with
numerous hummocks along the entire length of the transect (Figure 11a). The
general
profile of the transect was flat with a large depression between 150-160 m. The
topographic high at 60 m was represented by a spoil probably from the
construction
of ditches for mosquito control. Other than the ditch, the topographic high was
around 100 m which was at the edge of a small pond (radius approximately 5 m).
The topographic low along the transect was located at 160 m which is well into
the
forested area. This low spot represented the lowest elevation measured for all
transects.
The Flat Far transect also had a negligible slope (0.81 x 10-3
Table 3), but a
smoother surface and was less hummocky than the Flat Near transect (Figure 11b).
The general profile of the transect was flat with a slight high elevation
between 25-50
m followed by a depression around 60 m. The topographic high was at 85 m and
was
located on a large hummock in the transition zone. Unlike the Flat Near
transect,
there was only a slight depression immediately after the beginning of the
forested
area.
Figure 11. Elevation contour relative to the creekbank benchmark
(HAYD). Open
circles represent sampling points; a linear regression line
used to
determine slope is indicated. (A) Flat Near transect. (B)
Flat Far
transect. (C) Steep Near transect. (D) Steep Far
transect. Table 3. Slope summary table for study transects determined by
linear regression.
Regression analysis was computed from elevation measurements.
Transect N Slope (x 10-3) r2 MSEa
P
Flat Near
Flat Far 69
20 0.29
0.81 0.042
0.135 44.5
39.3
0.0932
0.1103
Steep Near
Steep Far 33
12 13.56
7.28 0.892
0.668 5.0
8.3
0.0001
0.0010
aDesignates root mean square error.
In contrast to the flat transects, the Steep Near transect had an
appreciable
slope (13.6 x 10-3, Table 3) and contained fewer hummocks than either of the
flat
transects (Figure 11c). The change in elevation was continuous, with the
topographic
low at the start of the transect (0 m) and the high near the end of the transect
(40 m)
in the forested zone. There was a slight depression between 42-50 m which
occurs
well into forest cover.
The Steep Far transect also had an appreciable slope (7.3 x 10-3,
Table 3), but
the change in elevation was not as continuous as the Steep Near transect. The
profile
of the transect was relatively flat until 32 m where the change in elevation was
the
greatest (Figure 11d). Similar to the Steep Near transect, the Steep Far
transect had
fewer hummocks than the flat transects.
3.2.2 Variation in Elevation by Zone
The study transects were grouped by gradient of slope (i.e., flat,
steep) and
divided into zones (i.e., high marsh, transition, and forested) for comparison
of mean
elevations between zones (Table 4).
The elevations for the high marsh zone were similar for both flat
and steep
transects. For this zone, the flat transects had a slightly higher elevation
than the
steep transects, but the difference (2.5 cm) was not significantly different
(Figure 12).
However, for the transition and forested zones, the difference
between mean
elevations for the flat and the steep transect was highly significant (Table 4).
The Table 4. Summary of leveling transects separated by vegetation
zone.
Transect
N Elevation (m)
Standard
Deviation
Mean Maximum Minimum
High marsh
Flat Near
Flat Far
Steep Near
Steep Far
Flat
Steep 52
7
7
5
59
12 1.034
1.044
0.963
1.080
1.037
1.012 1.310
1.110
1.040
1.110
1.310
1.110 0.880
0.940
0.910
1.050
0.880
0.910
0.074
0.057
0.053
0.025
0.071
0.073
Transition
Flat Near
Flat Far
Steep Near
Steep Far
Flat
Steep* 12
10
7
2
22
9 1.045
1.065
1.153
1.120
1.054
1.146 1.150
1.280
1.220
1.140
1.280
1.220 0.967
0.970
1.080
1.100
0.967
1.080
0.065
0.107
0.055
0.028
0.085
0.051
Forest
Flat Near
Flat Far
Steep Near
Steep Far
Flat
Steep* 5
3
19
3
8
22 1.015
1.160
1.453
1.360
1.070
1.440 1.140
1.220
1.594
1.440
1.220
1.594 0.809
1.130
1.280
1.280
0.809
1.280
0.142
0.052
0.072
0.080
0.133
0.078
* p < 0.001
Figure 12. Mean elevation for the high marsh zone, transition zone,
and forest
zone. Data were pooled data for both Flat Far and Near
(Flat), and
Steep Far and Near (Steep). steep transect transition zone
was 9.0 cm higher than that for the flat transect. The
forested zone for the steep transect was 37.0 cm higher than that for the flat
transect.
3.2.3 Surface Macrotopography
In general, the flat transects had more topographic variation than
the steep
transects which corroborated field notes and observations. Using linear
regressions,
the Flat Near and Flat Far transects had very high root mean square errors
compared
with the steep transects (Table 3). The Flat Near transect had the most
topographic
variation and highest root mean square error (44.5), whereas the Steep Near
transect
had the least topographic variation and lowest root mean square error (5.0).
3.3 Soil structure
3.3.1 Depth to mineral soil
The depth to mineral soil along both flat and steep transects was
highly
variable. In general, the organic layer became thinner along the transect from
marsh
landward (Figure 13). The mean organic thickness for the Flat Near and Steep
Near
transect was similar when compared over the entire transect (Table 5). The
overall
mean for the Steep Near transect was probably overestimated due to the thick
organic
layer at 40 m in the forested zone (Figure 13b). The organic layer in the
forested
zone is comprised of leaf litter of terrestrial origin and is not strictly
depositional peat
material. When only the depositional peat material was considered (i.e., high
marsh
Figure 13. Topographic variation in marsh surface and underlying
mineral soil
from high marsh to terrestrial forest. (A) Flat Near
transect. (B) Steep
Near transect. Table 5. Mean depth (m) to organic
layer for the Flat Near and Steep Near study
transects, q standard error.
Zone
N Mean Depth to Organic
Layer (m)
Flat TransectSteep Transect
High and Transition120.087 q 0.180.071 q 0.14
Total140.078 q 0.180.079 q 0.19
and transition zones), the Flat Near transect had a mean 1.6 cm thicker organic
layer
than the Steep Near transect (Table 5). The grand mean (8 q 2 cm) for both
transects suggests that the 10 cm depth is mostly organic and the 20 cm depth is
mostly mineral.
The thickness of the organic layer smoothed the hummocky nature of
the
underlying mineral layer and created less variable surface elevations (Figure
13). The
organic matter accumulation on the antecedent landform for the flat slope
resulted in a
flattening of surface elevations. If considered separately from the surface
elevations,
the slope of the underlying mineral layer for the Flat Near transect was 1.4 x
10-3 and
the slope of the surface was slightly flatter at 0.29 x 10-3. The difference is
small,
but acts to flatten an already flat slope. However, organic matter accumulation
for
the Steep Near transect resulted in a steeper surface slope (13.6 x 10-3) than
the
underlying mineral layer (12.8 x 10-3). The thickness of the organic layer was
weakly
negatively correlated with elevation (r= -0.259, n= 50), but not significant (p
=
0.07), suggesting that as elevation increased, the organic layer became thinner.
3.3.2 Bulk Density and Organic Carbon
3.3.2.1 Bulk Density and Organic Carbon by Slope
The mean bulk density of soil samples from the flat slopes were
slightly less
and the organic carbon content slightly more than for the steep slopes (Table
6). For
the 10 cm depth, the mean bulk density of the flat slope was significantly less
than for
Table 6. Soil properties for study transects obtained from sediment
cores. Values
reported are means of pooled values: Flat represents both Flat
Far and Flat
Near, and Steep represents both Steep Near and Steep Far, q
standard
error.
Variable Zones
Transect
Mean
High MarshTransitionForest
Bulk Density (g cm-3)
Flat 10 cm
Steep 10 cm
0.29 q0.03
0.38 q0.03
0.35 q0.04
0.52 q0.03
0.42 q0.11
0.51 q0.02
0.35 q0.251
0.47 q0.20
Flat 20 cm
Steep 20 cm 0.55 q0.03
0.59 q0.02 0.62 q0.03
0.52 q0.03 0.60 q0.02
0.52 q0.05
0.59 q0.20
0.55 q0.20
% Carbon
Flat 10 cm
Steep 10 cm
13.0 q2.7
2.1 q0.8
8.5 q1.8
2.1 q0.4
2.6 q0.1
1.9 q0.7
8.3 q2.33
4.4 q2.0
Flat 20 cm
Steep 20 cm 2.1 q0.8
1.4 q0.6 1.0 q0.2
1.8 q0.7 2.0 q0.6
1.0 q0.2
1.7 q0.78
1.4 q0.62
% Water
Flat 10 cm
Steep 10 cm
55.4 q8.0
33.5 q4.0
37.6 q7.7
7.3 q1.5
17.4 q1.9
3.6 q1.4
36.8 q4.43
14.8 q4.0
Flat 20 cm
Steep 20 cm 17.0 q4.0
5.8 q1.0 10.3 q2.1
9.1 q4.5 9.8 q3.0
3.3 q1.1
12.4 q2.03
6.1 q1.7
Chlorinity (mM)
Flat 10 cm
Steep 10 cm
151.4 q22.0
188.6 q24.2
107.6 q27.0
56.8 q14.2
26.0 q12.8
10.4 q 4.0
95.0 q11.23
85.2 q13.6
Flat 20 cm
Steep 20 cm 78.6 q 9.2
98.2 q10.2 43.2 q8.2
72.0 q6.2 23.6 q5.6
6.6 q1.8
48.4 q7.43
59.0 q9.8
1 p < 0.05
2 p < 0.01
3 p < 0.001 the steep slope. The organic carbon content was also
significantly higher for the flat
slope. However, the 20 cm samples were essentially the same for both bulk
density
and organic carbon content for both steep and flat slopes (Table 6).
3.3.2.2 Bulk Density and Organic Carbon by Zone
Soil samples at 10 cm in all zones of the flat transects had lower
bulk densities
and higher carbon content than for the steep transects (Figure 14a, c). None of
the
differences in bulk density were significant, although the transition zone of
the flat
transect was marginally significantly lower (p= 0.06) than that for the steep
transect.
The organic carbon content of the high marsh and transition zone for the flat
transects
were significantly higher than that for the steep transects (Table 6). The
forested
zone of the flat transects was only slightly higher in organic content than for
the steep
transects.
The mean bulk densities at 20 cm for the flat transect were higher
for the
transition and forested zones, but lower for the high marsh (Figure 14b).
Again,
none of the differences were found to be significant, although the transition
zone of
the flat transect was marginally significantly higher (p= 0.06) than for the
steep
transect. The organic content at 20 cm was low for both the steep and flat
transects
(Figure 14d) with no significant differences.
Figure 14. Mean soil structure data from core samples at 10 and 20 cm
depths for
both Flat Far and Near (Flat) and Steep Far and Near
(Steep), q
standard error. (A) Bulk density at 10 cm. (B) Bulk
density at 20 cm.
(C) Total carbon at 10 cm. (D) Total carbon at 20 cm. (E)
Chlorinity at
10 cm. (F) Chlorinity at 20 cm. 3.3.3 Chlorinity
3.3.3.1 Chlorinity by Slope
The sediment chlorinity ranged from a high of 310 mM for the steep
high
marsh at 10 cm depth to 0 mM for the steep forest at 60 cm depth. In general,
the
chlorinty was highest for the high marsh and decreased toward the forest for
steep and
flat slopes. For both steep and flat slopes, chlorinity also decreased with
depth.
The chlorinity of soil samples for the 10 cm depth was higher than that for the
20 cm
depth for both steep and flat transects.
3.3.3.2 Chlorinity by Zone
The chlorinity of the samples from the high marsh, transition, and
forested
zones at 10 cm was generally higher for the flat transect than for the steep
transect,
but the differences were not significant (Figure 14e).
The soil samples at the 20 cm depth of the flat transect were
highest for the
high marsh, and essentially the same for transition and forested zones. The
chlorinity
was highest for the high marsh, intermediate for the transition zone, and lowest
for
the forested zone of both flat and steep transects (Figure 14f). The chlorinity
of the
high marsh and forested zones of the flat transects was higher than that for the
steep
transects; in the transition zone, it was lower in the flat transect. Despite
the large
apparent differences in mean chlorinity, none were not significantly different
between
zones.
3.4 Hydroperiod
Hydroperiod evaluations were conducted for the Steep Near and Flat
Near
transects. The water table levels showed differences between months, seasons,
and
years of record (Figure 15a,b). The spring and summer of 1993 had the lowest
water
tables for all zones during the study period. Based on the long-term data set
(USDA
1989), lower than seasonal rainfall occurred during the late spring and early
summer
which resulted in lower water tables for both transects. The lowest water table
levels
of the study period were recorded in July and August 1993. Higher than usual
rainfall for August may have prevented the summertime drawdown for 1992. The
water table levels were generally highest in January and February.
Differences between seasons are reinforced by summary statistics
of
hydroperiod from continuous water level recorders placed in the transition zone
of
both the near steep and flat transects (Table 7). The percent of time flooded
and
mean water level were considerably lower during the warm period than the cool
period for the flat slope. The warm period showed variation between 1992 and
1993,
with higher water levels and percent of time flooded in 1992. This may be the
result
of the higher rainfall in the summer of 1992. For the warm period, the
transition
zone for the flat transect experienced more flooding and a higher mean water
level
than the steep transition zone. Because of the extremely dry summer, no
flooding
occurred for the period of record for the steep transition water level recorder.
However, data collection for the steep transition zone water level recorder
Figure 15. Mean monthly water table levels for the high marsh,
transition, and forest
zones determined by shallow (1.5 m) groundwater wells.
Sample points
represent the mean of three wells for each zone, q standard
error. The water
table position was measured twice in August 1992 to October
1992, and from
March to August 1993. For these dates, the mean represents
six wells. (A)
Flat Near transect. (B) Steep Near. (C) Mean water table for
the period
October 1992 to August 1993 for Flat Near (Flat), and Steep
Near (Steep)
transects. Table 7. Characteristics of hydroperiod at
the transition zone water level recorders
for the Steep Near and Flat Near transects.
Characteristics Flat1 (1991-1993) Steep2
(1992-1993)
WarmaCoolaWarm
% of time flooded
Mean depth (cm)b
(S.D.)
Total wks in period
# wks precipitation was received
# floods from storm events 13.4
- 18.83
(5.43)
27
21
2 31.0
0.88
(2.13)
18
17
3
0
-
81.84
(3.73)
11
7
0
1 Total period from 11 July 92 to 31 Aug 93.
2 Total period from 18 June 93 to 31 Aug 93.
a Warm season = April through September; Cool season = October through March.
b Calculated from hourly value. commenced 18 June 93, which was already into
the warm period. When the identical
period was examined, the flat transition zone still experienced higher mean
water
levels than the steep (e.g., -45.3 cm for the flat, and -81.8 cm for the steep),
but the
percent of time flooded decreased to 1% from 10.7%. Comparison among slopes for
the cool period was not possible because data were collected only during the
summer
1993 for the steep transition zone.
Flooding from the estuary occurred more often during the cool
period and
mainly during extratropical storm events. The number of storm events during the
warm period was higher than would be expected and due to events which occurred
during the beginning or end of the warm period (e.g., a storm on September 30).
Storm events are characterized by very large, steep rises in the water level
which are
greater than expected due to rainfall alone (Figure 16). Storm surges were
often
influenced by astronomical tides once water levels are above the surface and
result in
a quick succession of peaks in water level about 12 h apart.
3.4.1 Water Table Fluctuations
Overall, the high marsh zone experienced water level fluctuations
at or near
the surface for most of the study period. The transition zone had a somewhat
lower
water table with slightly more seasonal variation. The forested zone
experienced the
largest seasonal and yearly variation.
The overall trends in seasonal water table fluctuations were
generally the same
Figure 16. Hydrograph during September 1992 showing the effects of
rainfall and a
storm surge due to Tropical Storm Diane for the Flat Near
transition.
Position of the marsh surface is indicated by the 0 cm
water level.
Asterisk indicates beginning of storm. for both steep and
flat transects for the period of record. However, the drawdown
during the summer 1993 was more pronounced for the steep transect, with lowest
water levels recorded in August (Figure 15b). The drop in water levels was more
severe for the steep slope, especially between May and June. Rainfall from
convection storms in August resulted in a rise in water levels for all zones of
the flat
slope, but the steep slope only showed slight increases in water levels.
Although the general patterns of seasonal water table fluctuations
were similar
for both steep and flat slopes, the extent of these characteristics was
different among
slopes. For both steep and flat slopes, the mean water table was lowest in the
forest
(Figure 15c). However the mean water table was highest for the steep transect
in the
high marsh, but highest in the transition zone for the flat transect. The
monthly water
table levels for the flat slope were actually higher in the transition zone than
the high
marsh for the most of the spring and summer of 1993, even during periods of
drawdown.
The mean water table level for the high marsh showed little
difference between
flat and steep slopes and was less than 10 cm from the surface (Figure 15c).
However, highly significant differences were observed between the transition and
forested zones of flat and steep slopes. The mean water table levels of both
transition
and forested zones were closer to the surface for the flat slope than for the
steep
slope. The mean water table in the transition zone for the flat transect ranged
from 5
cm above the surface to 50 below the surface. For the steep transition zone,
the
mean water table ranged from at the surface to 75 cm below the surface. For the
forested zone, the mean water table of flat transect ranged from 10 to 80 cm
below
the surface whereas the range of the steep transect forested zone ranged from 20
to
100 cm below the surface.
3.4.2 Precipitation, PET, and Water Table Drawdown
From 1949-1981, total precipitation averaged 103 cm y-1, virtually
all as rain.
Monthly means were relatively constant throughout the year and were generally
highest in July through October and lowest from April through June (Figure 17).
>From 1990 to May 1993, the mean monthly precipitation was highest in August
(21.8
cm mo-1) and lowest in June (2.93 cm mo-1). For the three year period, the
total
precipitation was highest in 1992 (113.4 cm y-1) and lowest in 1990 (94.7 cm
y-1).
Based on data for 1949-1981, mean monthly potential
evapotranspiration (PET)
ranged from 1 cm in January to 14 cm in July and was less than mean monthly
precipitation in all months except for May through September (Figure 17). Thus,
these months are predicted to experience drawdown due to precipitation deficits.
However, over the period January 1990 to August 1993, only June, July and
October
had precipitation deficits for all three years. Due to the unusually high
rainfall during
the study period for August, rainfall exceeded the predicted PET between 1990
and
1992, but not in 1993.
Despite PET being less than precipitation in November 1992, the
water table
Figure 17. Monthly precipitation totals from 1990 to 1993 (Brownsville
Marsh) and
1949 to 1981 (Cheriton, VA). The dotted line connects
monthly means
for 1949 to 1981. The small open and solid circles connect
mean
Thornthwaite evapotranspiration for the same period.
Points below the
line connected with small circles represent months with
precipitation
deficits. decreased from October to November for the steep
slope for all zones and increased
during the same period for the flat slope. The water table did increase
substantially
in December for the steep slope. Drawdown during the summer of 1993 began in
May for both transects. A typical drawdown for flat and steep slopes shows that
the
recharge period and extent for the steep slope is more pronounced than for the
flat
slope (Figure 18).
3.4.3 Groundwater Salinity Patterns
Salinity patterns show variation seasonally, yearly, and among
vegetation
zones and gradients of slope (Figure 19). Average salinities ranged from a high
of 30
ppt in the high marsh to a low of 2 ppt in the forested zone.
In general, seasonal and yearly variations in salinity can be
attributed to
specific environmental conditions (e.g., flooding by the estuary, rainfall).
For
example, the Halloween Storm (31 October 1991), and a Northeaster (4 January 92)
flooded the marsh and forested areas, and resulted in increased salinity which
remained elevated until March 1992 (Figure 19a). Unusually high rainfall during
the
spring and summer of 1992 lowered salinities throughout the summer and fall.
Another storm on 25 September 92 (Tropical Storm Diane) flooded the marsh and
forest with saline water (salinity of the flood water at the forest zone was
10-20 ppt,
measured during the event) and resulted in increased salinity of the transition
and
forest zone and decreased salinities for the high marsh. Salinities increased
and
Figure 18. Hydrograph during June and July 1993 comparing the diurnal
pulse of
water table drawdown due to evapotranspiration for flat and
steep
slopes. Depth is shown relative to the marsh surface. (A)
Hydrograph
for the Flat Near transition water level recorder. (B)
Hydrograph for the
Steep Near transition water level recorder.
Figure 19. Mean monthly salinity for the high marsh, transition, and
forest zones
determined by shallow (1.5 m) groundwater wells. Sample
points
represent the mean of three wells for each zone, q standard
error. The
salinity was measured twice in August 1992 to October 1992
(for the
Flat Near transect), and from March to August 1993. For
these dates,
the mean represents six wells. (A) Flat Near transect. (B)
Steep Near. remained high following another flood by a strong Northeaster in
December 1992.Salinity was found to be highest in the high marsh, declining both
toward the
forest and creekbank (Figure 20). Although the general patterns of salinity
with
distance appear to be similar for both flat and steep slopes, the nature of the
change is
different. The flat transect showed a very gradual decrease in salinity with
increasing
distance from the marsh, with much variation (Figure 20a). In contrast, the
steep
transect had a very abrupt change in salinity with increasing distance, and
comparatively little variation (Figure 20b).
The area with the largest drop in salinity with distance
corresponds with the
transition zone for both flat and steep transects. The largest drop in salinity
for the
flat slope occurred between 123 to 150 m which is the location of the transition
zone.
Similarly, for the steep slope, the greatest drop in salinity with distance
occurred
between 20 to 25 m for the steep slope which is also the location of the
transition
zone.
3.5 Pore Water Chemistry
The lack of precipitation during the spring and summer of 1993
resulted in
periods of severe drawdown. Consequently, pore water was unable to be collected
after June and only data collected from January to June 1993 were used for
analysis.
3.5.1 Pore Water Salinity
The salinity of the pore water ranged from a high of 34 ppt for
the high marsh
Figure 20. Box plot of the groundwater salinities along the Flat Near
and Steep
Near transects for the period beginning 4 November 1991 for
the Flat
Near and 5 August 1992 for the Steep Near transect. Dotted
bars
represent the mean, solid bars represent the median, and
the box and
error bars encompass the range. (A) Flat Near transect. (B)
Steep Near
transect. zone and a low of 1 ppt for the forested zone.
Pore water salinities were stable for
all zones of both steep and flat slopes, and for 10 and 20 cm during the period
of
study (Figure 21). As the summer began, salinity increased for the high marsh
and
transition zones for the flat slope, but only for the high marsh zone of the
steep slope.
Variation was the greatest for the forest zone of the flat slope and the high
marsh
zone for the steep slope. For the lysimeter data collection period mean
salinity
decreased with distance landward for both steep and flat slopes and for 10 and
20 cm
(Figure 22a, b). Mean salinity increased with depth for all zones and slopes.
The 20
cm depth was about 3 ppt more saline than the 10 cm depth for the high marsh
zone.
The flat transect experienced higher mean pore water salinities at
10 and 20
cm than the steep transect for all vegetation zones (Figure 22a, b). The
differences
between the transects were highly significant for the transition and forest
zones. The
mean salinity for these zones was more than double the mean salinity for the
same
zones of the steep transect (Table 8).
A comparison between the salinities obtained from lysimeters and
shallow
groundwater wells shows that the wells measure slightly higher salinities
(Figure 23).
The wells average about 3 ppt higher than the lysimeters for the high marsh and
transition zones, and about 1 ppt higher than the lysimeters for the forest.
This
discrepancy may be due to increased salinity with depth as was found between 10
and
20 cm lysimeters.
Figure 21. Pore water salinity collected for the Flat Near and Steep
Near transects
from lysimeters at 10 and 20 cm depths for the period 14
Jan to 29 July
1993. Points represent the mean of three lysimeters for
the Flat
transect and six lysimeters for the Steep transect, q
standard error. (A)
Flat Near transect at 10 cm. (B) Flat Near transect at 20
cm. (C) Steep
Near transect at 10 cm. (D) Steep Near transect at 20 cm.
Figure 22. Mean pore water data collected for the Flat Near and Steep
Near transects
from lysimeters at 10 and 20 cm depths for the period 14 Jan
to 29 July 1993.
Bars represent the mean of three lysimeters for the Flat
transect and six
lysimeters for the Steep transect, q standard error. (A)
Mean pore water
salinity at 10 cm. (B) Mean pore water salinity at 20 cm.
(C) Mean pore water
redox potential at 10 cm. (D) Mean pore water redox
potential at 20 cm. (E)
Mean pore water sulfide concentration at 10 cm. (F) Mean
pore water sulfide
concentration at 20 cm. (G) Mean pore water pH at 10 cm. (H)
Mean pore
water pH at 20 cm.
Table 8. Mean pore water data collected from low-tension lysimeters for
the period
14 January to 20 May 1993, q standard error. Data were used as
a basis
for Figure 21.
Variable Depth1 (cm)Zone2 Flat3
Steep4
Salinity
(ppt)
10 H
T*
F* 16.7 q 0.6
14.4 q 0.4
7.5 q 0.7
16.3 q 0.8
7.0 q 0.3
3.0 q 0.2
20 H
T*
F* 19.8 q 0.5
14.1 q 0.3
8.4 q 0.6
19.5 q 0.9
7.1 q 0.3
4.4 q 0.2
Redox potential
(mV)
10 H
T
F 195 q 11
211 q 6
219 q 7
182 q 12
195 q 10
199 q 9
20 H*
T*
F* 206 q 9
210 q 6
218 q 7
157 q 11
166 q 11
170 q 8
Hydrogen
Sulfide (fM)
10 H*
T
F 105.6 q 44.9
0.2 q 0.1
3.8 q 2.3
34.1 q 9.6
0.4 q 0.1
4.0 q 0.5
20 H
T
F 2.4 q 1.0
0.4 q 0.2
0.1 q 0.1
1.6 q 0.4
0.3 q 0.2
1.5 q 0.4
pH
10 H
T*
F* 6.5 q 0.2
5.8 q 0.2
3.5 q 0.1
6.4 q 0.1
4.4 q 0.1
4.2 q 0.1
20 H
T*
F 5.8 q 0.3
5.2 q 0.2
3.4 q 0.1
5.9 q 0.1
4.0 q 0.8
3.6 q 0.6
* p< 0.01
1 Depths were 10 and 20 cm for High marsh and Transition zones, 20 and 60 cm for
Forest zone.
2H= high marsh zone; T= transition zone; F= forest zone.
3n=24 for all zones; 4n=48 for H and T; n=64 for F
3.5.2 Redox Potential, Hydrogen Sulfide, and pH
Mean redox potential increased over time during the study period
for all
vegetation zones and depths (Figure 24). Values were lowest in January, then
rose
and remained high in February through June. For both slopes, redox potentials
ranged from a low of -50 mV for the high marsh to a high of +260 mV for the
forest. Very little variation was seen among zones and depths within type of
slope.
It should be noted that the low readings on 14 January 93 may have been due to
instrument malfunction, since it was the first measurement. However, sediment
redox
potentials measured for two months prior to the lysimeter sampling period, which
are
not reported here, were similarly low (e.g., approximately 100 mV).
In general, the redox potential was higher for the flat transect
than the steep
transect for all zones and depths (Figure 22c, d). The mean redox potential for
the
flat transect was significantly higher than the steep transect at 20 cm for all
zones.
The flat transect was roughly 50 mV more positive than the steep transect for
all
zones at 20 cm. Although higher for the flat transect, the differences between
flat
and steep transects at 10 cm were not significant.
The range of redox potentials during the study period was often
low enough to
create conditions favorable to hydrogen sulfide production. However, with the
exception of the high marsh lysimeter at 10 cm, very little sulfide was measured
during the study period (Figure 22e, f). Sulfide ranged from a high of 1200 fM
for
the high marsh to below the level of detection.
Figure 23. Comparison between salinities measured using shallow
groundwater and
pore water salinities collected using lysimeter at 10 cm
for the Flat
Near transect for the period January to July 1993.
Figure 24. Pore water redox potential collected for the Flat Near and
Steep Near
transects from lysimeters at 10 and 20 cm depths for the
period 14 Jan
to 29 July 1993. Points represent the mean of three
lysimeters for the
Flat transect and six lysimeters for the Steep transect, q
standard error.
(A) Flat Near transect at 10 cm. (B) Flat Near transect at
20 cm. (C)
Steep Near transect at 10 cm. (D) Steep Near transect at 20
cm. Mean pore water sulfide concentration was significantly higher
for the flat
transect at 10 cm in the high marsh zone. The high marsh zone had a mean
sulfide
concentration three times that of the same zone and depth of the steep transect.
The
mean concentrations for other depths and zones were highly variable making
comparisons difficult.
An examination of two lysimeters (0 m at 10 cm depth and 150 m at
20 cm
depth) of the flat slope shows that a time lag exists between peaks in sulfide
production with distance landward (Figure 25). The 0 m lysimeter showed peak
sulfide production between 20 May and 17 June, whereas production peaked around
17 June for the 150 m lysimeter. Although the differences in concentrations are
about
two orders of magnitude, the sulfide production shows a time lag of about two
weeks.
The decline in sulfide concentrations coincided with the beginning of the
summertime
drawdown.
Sulfide detection by the lysimeters was very patchy in nature; at
times, one
lysimeter would have high concentrations of hydrogen sulfide and the replicate
lysimeters would not. This resulted in very large standard errors and made it
difficult, if not impossible, to compare means.
Mean pH ranged from a high of 8.1 for the high marsh zone to a low
of 2.5
for the forest zone. The mean pH decreased from the marsh landward at both
depths
(Figure 22g,h). The flat transect had significantly higher pH for the
transition zone at
10 and 20 cm than the steep transect (Table 8). The forest zone at 10 cm had a
Figure 25. Sulfide concentration collected from the 0 m lysimeter at
the 10 cm
depth and from the 150 m lysimeter at the 20 cm depth for
the Flat
Near transect. Bars represent sampling dates from 8 April
to 13 August
1993.
significantly lower pH for the flat transect, but was not significant at 20 cm.
The
high marsh zone was similar for both transects and depths.
3.6 Factor Analysis
3.6.1 Analysis of Entire Data Set
Six of the 11 physical and chemical variables had correlations
greater than
0.48 (Table 9) suggesting factor analysis would be appropriate. The factor
analysis
constructed 11 factors of which three had eigenvalues greater than 1.0, a
criterion
used in deciding the number of factors to extract (Dunteman 1984). The fourth
factor
had an eigenvalue of 0.81 indicating the three factor solution was satisfactory.
The
three factors accounted for 69.4% of the variation in the data (Table 10).
After
rotation, all factors had at least two variables with a loading score whose
absolute
value was 0.5 or greater.
The first factor was interpreted as a brackish water factor.
Salinity and pH at
10, and 20 cm, and water table had high positive loadings on this factor, while
distance and elevation had large negative loadings (Table 10). From the
analysis of
the univariate data, it follows that as distance and elevation increased (from
high
marsh toward forest), salinity, pH, and water table levels decreased (Figure
26a).
Analysis of variance for factor one showed significant zone
differences, but
not significant date differences (Table 11). The transition and forest zones of
the flat
transect had higher mean factor scores for the study period (Table 12),
reflecting Table 9. Correlations of physical and chemical variables1.
Correlations with absolute values greater than 0.48 are
significant (p < 0.05).
P10
P20
E10
E20
W
S10
S20
H10
H20
D
E
P10
1.00
P20
0.72
1.00
E10
-0.27
-0.21
1.00
E20
-0.25
-0.18
0.86
1.00
W
0.41
0.38
-0.23
-0.13
1.00
S10
0.52
0.46
-0.10
-0.11
0.36
1.00
S20
0.62
0.48
-0.10
-0.11
0.35
0.86
1.00
H10
0.22
0.28
-0.06
-0.01
0.10
0.19
0.23
1.00
H20
0.25
0.33
0.01
-0.06
0.06
0.06
0.20
0.25
1.00
D
-0.75
-0.67
0.18
0.23
-0.42
-0.58
-0.66
-0.30
-0.29
1.00
E
-0.47
-0.50
0.12
0.04
-0.41
-0.65
-0.63
-0.19
-0.10
0.53
1.00
1Variable abbreviations are: P = pH; E = Eh (Redox potential); W = water table;
S = pore water salinity; H =
hydrogen sulfide; D = distance from creekbank to lysimeter; E = elevation at
lysimeter; 10 = 10 cm depth; 20 = 20 cm
depth. Table 10. Correlations of physical and chemical
variables1 with factors.
Variable Factor 1 Factor 2
Factor 3
S10
S20
P10
P20
W
D
E
E10
E20
H20
H10 0.88
0.87
0.71
0.62
0.58
- 0.74
- 0.81
- 0.11
- 0.07
0.01
0.14 0.02
0.01
- 0.24
- 0.18
- 0.19
0.16
- 0.03
0.95
0.95
0.00
0.03
0.00
0.14
0.38
0.50
-
0.01
-
0.42
-
0.04
0.01
-
0.03
0.82
0.65
% variance explained by
each factor
42.10
16.26
11.06
1Variable abbreviations are: P = pH; E = Eh (Redox potential); W = water table;
S = pore water salinity; H = hydrogen sulfide; D = distance from creekbank to
lysimeter; E = elevation at lysimeter; 10 = 10 cm depth; 20 = 20 cm depth.
Figure 26. Mean factor scores for the high marsh, transition, and
forest vegetation
zones for the period 14 January to 29 June 1993, q standard
error. (A)
Brackish water factor. (B) Redox factor. (C) Sulfide
factor. Table 11. Two-way analysis of variance comparing factor
scores for lysimeter
data among slopes and sampling dates for the entire data
set.
Factor Variation df SS F
P
Factor 1Model
Date
Slope x Date
Error 5
13
38
90 96.6
4.2
3.2
31.1 55.97
0.93
0.25
0.0001
0.5273
1.0000
Factor 2Model
Date
Slope x Date
Error 5
13
38
90 2.9
76.9
23.9
5.4 9.50
98.15
10.45
0.0001
0.0001
0.0001
Factor 3Model
Date
Slope x Date
Error 5
13
38
90 30.8
9.2
13.6
92.8 5.98
0.69
0.35
0.0001
0.7729
0.9998
Table 12. Mean factor scores for the Flat Near and Steep Near
transects grouped
by vegetation zone, q standard error.
Factor Zone1 Flat Transect Steep
Transect
1 H 0.49 q0.07 0.72
q0.10
T* 0.07 q0.05-1.01 q0.10
F*-1.07 q0.12-2.08 q0.07
2
H- 0.04 q0.16- 0.15
q0.14
T* 0.07 q0.22-0.17 q0.16
F* 0.57 q0.28-0.01 q0.35
3 H 0.61 q0.31 0.24
q0.13
T-0.41 q0.09-0.29 q0.06
F-0.77 q0.08 0.12 q0.06
* p< 0.05 (Tukey's)
1H = high marsh; T = transition zone; and F = forest area higher salinities, pH,
and water tables. The high marsh was similar for both
transects. The differences in factor one over time were not significant and
probably
reflect the stable nature of salinity and pH during the study period (see, for
example,Figure 21).
The second factor was interpreted as the redox factor because the
redox at 10
and 20 cm was the dominant loading variable (Table 10). No other variable had
high
loadings on this factor. The redox factor generally increased with distance
landward
for both slopes (Figure 26b), although variability was very high.
The redox factor showed significant zone and date differences over
the period
of study (Table 11). The transition and forest zones of the flat transect had
significantly higher mean factor scores (Table 12) which is in agreement with
the
univariate data for redox potential. As in the univariate analysis of redox
potentials,
the redox factor increased from January to February and remained high for the
rest of
the study period.
The third factor was interpreted as the sulfide factor since
sulfide at 10 and 20
cm had the only dominant loading value for the factor (Table 10). The sulfide
factor
scores decreased with distance (Figure 26c), although no significant differences
between zones or dates existed (Table 11). The variability of this factor was
very
high and reflects the high variability of the sulfide variable (Table 12).
3.6.2 Analysis of Transition Data Set
A factor analysis was conducted using the data set
for the transition zone to
determine if this zone contained any unique properties. Similar to the analysis
for the
entire data set, the factor analysis constructed 11 factors of which three had
eigenvalues above 1.0. The factor loadings were very similar to the loading for
the
entire data set with differences mainly between the loading scores of distance
and
water table (Table 13). Distance remained as a loading variable on the first
factor,
but changed its value from a positive to negative loading. The water table
variable
remained a positive, but changed position to load on the third factor. The
factors
were interpreted as being identical to the factors described for the entire data
set and
are brackish water, redox potential, and sulfide.
For factor one, the brackish water factor, it follows that as
distance increased
landward, the salinity of the pore water also increased. The 2 y well data for
salinity
support this trend for the transition zone (see, for example, Figure 20a). The
finding
that as distance increases for the transition, salinity also increases is the
reverse of
what was found using the entire data set. Although pH contributed to the
loading on
this factor, the values were much smaller, suggesting that the influence due to
pH
may play a smaller role.
For factor three, the sulfide factor, water table was found to be
positively
associated with sulfide production (Table 13). It follows that as the water
table
increases, sulfide concentrations would also increase, especially during the
period of
study (i.e., spring and summer). Table 13. Correlations of
physical and chemical variables1 with factors using data
from transition zone only.
Variable Factor 1 Factor 2
Factor 3
S10
S20
D
P20
P10
E
E20
E10
H10
W
H20 0.90
0.79
0.78
0.73
0.68
- 0.86
0.10
- 0.07
- 0.13
0.41
0.23 0.05
0.20
0.00
- 0.23
- 0.21
- 0.12
0.95
0.94
0.05
0.05
- 0.40
0.26
0.27
-
0.23
0.12
-
0.12
-
0.35
-
0.02
0.06
0.78
0.57
0.48
% variance explained by
each factor
39.00
19.15
11.59
1Variable abbreviations are: P = pH; E = Eh (Redox potential); W = water table;
S = pore water salinity; H = hydrogen sulfide; D = distance from creekbank to
lysimeter; E = elevation at lysimeter; 10 = 10 cm depth; 20 = 20 cm depth.
Table 14. Two-way analysis of variance comparing factor scores for
lysimeter
data among slopes and sampling dates for the transition
data set only.
Factor Variation df SS F
P
Factor 1Model
Date
Slope x Date
Error 1
11
6
23 28.8
1.1
0.3
1.6 414.96
1.39
0.61
0.0001
0.2445
0.7170
Factor 2Model
Date
Slope x Date
Error 1
11
6
23 0.2
28.9
2.2
1.7 2.15
35.59
4.99
0.1563
0.0001
0.0021
Factor 3Model
Date
Slope x Date
Error 1
11
6
23 2.1
13.5
6.4
20.0 2.45
1.41
1.23
0.1313
0.2352
0.3279
MARK M. BRINSON (919) 757-6718
BIOLOGY DEPARTMENT
EAST CAROLINA UNIVERSITY
GREENVILLE, NC 27858
BIBRINSO@ECUVM.CIS.ECU.EDU