CHAPTER THREE
Groundwater Discharge and Spartina
Alterniflora Production in a Developing Barrier Island Salt Marsh
3.1 INTRODUCTION
The vegetation patterns on highly dynamic barrier
islands are controlled by the hydrogeomorphology of the immediate environment
(Hayden et al. 1995). The interaction between the land-surface elevation,
the fresh groundwater table and sea level is particularly apparent and
important in these environments. In general, local vegetation establishment
and growth is controlled by the physico-chemical makeup of the substrate.
The physico-chemistry of the substrate is controlled by subsurface hydrological
processes. On a barrier island, the subsurface hydrology is controlled
by the interaction between the static groundwater table and the dynamic
seawater table, and the geomorphology of the landscape controls where these
two water tables meet. Thus, the interaction between these three surfaces
(the groundwater table, the seawater table, and the land surface) controls
the vegetation. In an area where the landscape is such that the excursions
of the groundwater table dominate the subsurface physico-chemistry, the
salinities will be lower, and the vegetation will reflect the salinity;
alternatively, if the variations of the seawater table dominate, the salinity
will be higher, and halophytic plants will dominate. The principal aim
of this study is to show the connection between the vegetation, the physico-chemistry,
the hydrology and the landscape in the narrow fringing salt marshes that
are characteristic of the back-side of many Atlantic coast barrier islands.
In these marshes, the dominant primary producer is Spartina alterniflora
Loisel.
Many authors have studied the effect that high salinity
has on S. alterniflora (e.g. Phleger 1971, Nestler 1977, Linthurst
& Seneca 1981, Bradley & Morris 1991, Colmer et al. 1996). Salinity
interferes with S. alterniflora growth to the extent that Morris
and others have suggested that the interannual trends in S. alterniflora
productivity are the result of salinity fluctuations caused by annual changes
in mean sea level (Morris & Haskin 1990, Morris et al. 1990, Morris
1995). Porewater salinity is a function of landscape and position in the
marsh. In concert with the ceiling placed on plant growth by nutrient limitations
and redox potential, salinity adds a spatial aspect to the window of conditions
for optimum S. alterniflora production. In highly anoxic marsh sediments,
which are characteristic of older marshes, anoxia may be a more important
limiting factor (Linthurst & Seneca 1981); however, in young, sandy
marshes with low organic content, such as some back-barrier marshes, the
redox potential is most likely sufficiently high that salinity is a more
significant factor in S. alterniflora growth.
High pore water salinity has a two-fold effect on
S. alterniflora growth. First, it interferes with the plants ability
to take up ammonium. The sodium ion inhibits ammonium ion uptake through
competitive ion exclusion (Bradley & Morris 1991). Thus, even under
conditions where nitrogen is abundant, it is practically unavailable at
high salinities. Second, high salt concentrations create an osmoregulatory
problem for the plant. Halophytic plants utilize two mechanisms for maintenance
of osmotic balance: they can selectively take up ions, or they can produce
nitrogen containing organic solutes (Flowers et al. 1977, Colmer et al.
1996). Proline concentration in Salicornia fruticosa is positively
correlated with the salt concentration in the plant sap, and betaine concentration
in Spartina townsendii increases with increasing external
NaCl concentration (Flowers et al. 1977). S. alterniflora growing
at elevated salinity accumulates proline and glycinebetaine in leaf tissue
(Cavalieri & Huang 1981, Colmer et al. 1996). For plants with an already
limited nitrogen supply due to the inhibition of uptake, the use of available
nitrogen for osmoregulatory compounds causes even greater stress for the
plant.
Salts are brought into the marsh sediments by the
tides. Salts are subsequently removed by diffusion out of the sediment
at high tide (Morris 1995), flushing of the creekbank (e.g. Agosta 1985,
Yelverton & Hackney 1986), or by dilution from fresh groundwater input
(Nuttle & Harvey 1995). Removal of salt to the overlying water is a
small flux of material (Morris 1995), and flushing of the creekbank and
dilution by incoming groundwater may be of greater importance (Nuttle &
Harvey 1995). The drainage of porewater is a near creek phenomenon (Yelverton
& Hackney 1986, Jordan & Correll 1985). Further from the creek,
evaporation is the primary method for the removal of water from the sediment
(de Leeuw et al. 1991). Thus, as discussed previously, the interaction
between the sea level, the sediment surface elevation and slope and the
water table height is important. A site that has a shallower slope is likely
to have a higher salinity than one with a steeper slope due to the dominance
of evapotranspiration (ET) over subsurface drainage. A site with a greater
groundwater input will have lower salinities and greater S. alterniflora
production. Overall, the salinity of the sediments at the creekbank is
determined by the balance between the salt-concentrating effects of ET
and the diluting effects of tidal flushing and groundwater inflow.
In fringing back-barrier marshes, the inland portion
of the drainage area is much closer to the marsh edge than in other types
of marshes. Therefore, in these marshes the type of area that is drained
may have a significant influence on the sub-surface hydrology within the
marsh. A marsh with a large upland catchment area relative to the down-gradient
marsh area which receives the catchment drainage will have a greater discharge
of fresh water per area of marsh than a marsh with a very small upland
catchment relative to marsh area. In addition, topographic features such
as ponds, which act as freshwater reservoirs, or salt flats, which are
areas of little topographic relief, have an effect on the type (fresh or
saline) and quantity of drainage to the marsh.
Groundwater and upland discharge to salt marshes
has been estimated by several authors (e.g. Harvey et al. 1987, Nuttle
& Hemond 1988, Harvey 1990, Harvey & Odum 1990, Nuttle & Harvey
1995). The majority of these studies have estimated groundwater discharge
in areas where the tidal fluxes are very small (Nuttle & Hemond 1988),
or are non-existent (i.e., subtidal, Harvey 1990). These studies have estimated
the interaction between the groundwater table and the seawater table, and
have implied that the nature of this interaction is important in determining
the salinity and therefore the S. alterniflora production. However,
a direct link between the subsurface hydrology and the S. alterniflora
production has not been made. This study is an attempt to bridge the gap
between the hydrogeomorphology of the salt marsh/upland complex and the
physico-chemistry and biotic productivity within the marsh itself. The
interconnections which are considered in this study are high-lighted in
Figure 3.1; the relevance of this study to other
processes within the marsh can also be seen in this figure.
More specifically, the purpose of this work is to
(1) characterize the role of variable upland surface morphology in controlling
the hydrological regime in the region of small creeks in a developing salt
marsh, (2) to show the connection between the hydrological regime and the
resulting sediment physico-chemistry, and (3) to demonstrate the effect
that this chemistry has on the dominant primary producer. This study makes
use of a 'natural' experiment where two creeks, of similar elevation, age
and sediment physical properties exist side by side. One of these creeks
drains a large upland area that contains a seasonally fresh/brackish water
pond, while the other drains a smaller catchment area which contains a
salt flat. It was hypothesized that the creekbank area which drains a large
upland area will have a consistently higher water table, larger discharge
of water, and lower salinities than the creekbank area which drains a salt
flat. It follows therefore, that the creek with the lower salinity will
have higher S. alterniflora production. Factors known to control
the growth of S. alterniflora were investigated at each site to
determine the limiting factor, and the resulting plant production and tissue
element composition were compared. This study provides two missing links:
first, it establishes the importance of the area directly adjacent to the
salt marsh in controlling processes that occur within the marsh by placing
the marsh in the larger context of the upland-marsh complex, and second,
it shows the direct connection between the subsurface hydrology and the
production of salt marsh vegetation.
3.2 METHODS
3.2.1 Site description
This study takes place on Hog Island (Figure
3.2), a mid-Atlantic barrier island off the coast of the Delmarva Peninsula,
Virginia. It is a part of the Virginia Coast Reserve Long Term Ecological
Research project. The island is a highly dynamic environment subject to
frequent disturbances (Hayden et al. 1991). This is exemplified by the
fact that the town of Broadwater, with nearly 250 inhabitants at the turn
of the century, is now underwater some distance from the present shoreline
of the island (Badger & Kellam 1989). In March of 1962 a large and
powerful coastal storm caused a washover event which deposited nearly a
meter of sand on the mature marshes on the eastern side of the southern
end of the island (Stewart 1962). Since this overwash event, the fringing
S. alterniflora marshes have gradually regrown on the bay side of
the island.
Two creeks were selected for this study. Both creeks
are approximately 5 years old, as determined by inspection of aerial photographs.
I have defined age as the time since S. alterniflora has been visible
on these photographs, and it is thereby the ecological age of the site.
Both sites, which are the result of the 1962 washover event, have the same
geological age. The topography of the region was mapped out using Pentax
Total Station survey equipment. The survey was tied-in to 3 benchmarks
which are part of the LTER benchmark system (which was established using
Trimble Survey Grade GPS), and is tied to USGS benchmarks on the mainland.
The reference datum is3.2 the USGS 1967 reduction level. The entire area
was mapped using a 10 m x 10 m grid, and the vegetation was noted at each
point. In addition, the areas closest to the creekbank were mapped using
a 1 m x 1 m grid. The overall accuracy of the survey is within 3-4 cm of
the 1967 USGS reduction level, and the within-survey accuracy is 1.5 cm
in the z-direction. A contour map of the study site that was generated
as a result of the Total Station survey is given in
Figure 3.3. The map shows the locations of the creeks relative to one
another, and relative to mean high water. It shows that the salt flat,
which drains into the southern creek has very little relief in comparison
to the northern creek, which receives drainage from region typical of a
barrier island upland. In this region there is some topographic relief,
which, although slight, is greater in this area than in the surrounding
areas.
Figure 3.4 is a surface plot of the study site which has the vegetation
types overlain in color. This figure illustrates the proximity of the upland
to each of the creeks more clearly. The northern creek is closer to the
upland, as is evidenced by the rapid transition from S. alterniflora
to plants characteristic of the high marsh, Spartina patens and
Distichlis spicata, and finally to upland shrub vegetation, represented
by Iva frutescens. The inland portion of the other creek is down-gradient
of an unvegetated salt flat.
3.2.2 Site Instrumentation
At each site, two 5-m transects were established
perpendicular to the creek. Past studies indicate that the range of influence
of the creek on the subsurface horizontal water 3.33.4movement does not
extend beyond approximately 2.5 m (Howes et al. 1994, Harvey et al. 1987,
Nuttle 1988, Agosta 1985) to 15 m (Nuttle 1988, Nuttle and Hemond 1988)
of the creekbank. Four suction lysimeters ("sippers", Chambers & Odum
1990) were placed along each transect at 0.5 m, 1 m, 2 m, & 3 m from
the marsh edge. Four piezometers were installed to approximate depths of
75 cm at 0.5 m, 1.5 m, 3 m and 5 m from the marsh edge. An additional piezometer,
at a depth of 50 cm, was installed 3 m from the creekbank in order to evaluate
the vertical hydraulic head gradient. The piezometers were constructed
of 2" PVC pipe, and were commercially slotted 20 cm from the bottom. The
site instrumentation for discharge measurements is shown in Figure
3.5. The site elevations and slopes are also shown. The elevation and
slope are similar between the two sites.
3.2.3 Hydraulic head measurement
Each piezometer was outfitted with a pressure transducer
(HOBO) attached to a 2 m standpipe. Upon installation, the standpipes,
with the HOBO already attached, were lowered into the piezometer, and fastened
onto the well. Figure 3.6 is a diagram illustrating
this instrumentation. By installing the stand pipes with the HOBO attached,
I allowed for pressure to build up in the pipe. The HOBOs were placed in
plastic bags containing dessicant, and then placed in sealed Rubbermaid
containers. The bags were not sealed, and the containers had a hole in
the side so that the tubing for the HOBO could reach the stand pipe, and
the HOBO could register ambient air pressure. The data loggers on the HOBOs
were started in the field using a laptop computer, and were set to measure
water level every 12 minutes for two week intervals. The horizontal gradients
at 75 cm were measured for 6 weeks, resulting in ~85 dicrete tidal cycles.
The vertical fluxes, measured between ~75 cm to ~50 cm below the surface,
were measured for a two week interval, resulting in ~25 discrete tidal
events.
Some adjustments to the HOBO data were made: (1)
Each HOBO was calibrated in the laboratory. The relationship between actual
and measured water height was linear, but the relationship is not 1:1.
(2) Nuttle and Harvey (1995) noted that a better model fit was obtained
using an adjustment for the temperature associated drift of the HOBOs.
In order to correct for this, an additional HOBO was installed in an unslotted
well containing a known height of seawater. The top of the well was sealed
sufficiently to prevent water evaporation, but allowed for the expansion
of air. This HOBO collected measurements every 12 minutes for 1 month.
A cyclic drift of ~4.5 cm was observed over a 24-hr period. Mean difference
between the measured and actual water level was calculated at hourly intervals,
and the piezometer data was adjusted accordingly. (3) Over time, there
may have been small leaks in the system which would allow water to rise
into the stand pipe. This may have caused an underestimation of the water
level. In order to correct for this, all measurements were corrected by
comparing the measured high tide values to the high tide from the tide
station on North Hog. (4) All data were adjusted relative to the survey
elevation for the individual well, and are reported relative to mean low
water (MLW), which is 33.9 cm below mean sea level (MSL).
3.2.4 Sediment and hydrologic properties
Hydraulic conductivity (K) was measured both in
situ and in the laboratory. In the field, a slug test was performed
by placing a 'slug' into the piezometer, waiting for the water level to
equilibrate, and then rapidly removing the slug. The rate of rise of water
in the piezometer was measured using a HOBO, which was set at a measurement
frequency of 2 seconds. K values were calculated using the Bouwer and Rice
method (Bouwer & Rice 1976, Bouwer 1989). A comparison was made between
the Bouwer and Rice and the Hvorslev (1957, as described in Fetter 1994)
method for K estimation, and it was determined that the Bouwer and Rice
method yields more consistent results for replicate slug tests. Four replicate
tests were performed at each piezometer. The field values were checked
in the laboratory using a constant head permeameter test on 6.4 cm diameter
x 19.2 cm long repacked cores from each site. Because this method required
repacking the cores, an additional test of hydraulic conductivity was performed
using intact 26.5 cm diameter x 30 cm cores. A Marriotte bottle was employed
to maintain a constant head and the rate of discharge from an outlet approximately
5 cm from the bottom of the core was measured. Nine replicate tests were
performed on each core.
The specific yield of the sediment was determined
for each location using a modification of the method described in Osgood
(1996). Using the same 26.5 cm x 30 cm cores as described for the hydraulic
conductivity test, the specific yield of the sediment was measured for
drainage of water out of the bottom of the core, for addition of the water
to the surface of the core, and for addition of water to the manometer.
This was done to examine any hysteresis between drainage, infiltration
and vertical recharge. Following the removal or addition of water to the
core, the water level was allowed to stabilize for 6+ hours before the
reading on the manometer was taken. This time frame was chosen to simulate
the in situ situation where the sediment has only the time between
ebb and flood events to flood or drain. Specific yield (Sy)
was then calculated using:
where dV is the volume of water drained or added,
dh is the change in head following the removal or addition of water, and
A is the surface area of the core. Porosity was determined on intact saturated
cores. The mass of water held in the pores will be determined by the difference
between wet and dry weight of the core. Ten replicate cores from each site
were measured. Sediment grain size was determined using the method described
in Brower and Zar (1984), and the sediment organic content was determined
gravimetrically by loss upon combustion.
3.2.5 Evapotranspiration
Evapotranspiration (ET) for the site was estimated
indirectly and directly in order to obtain a realistic estimate. An indirect
estimate was made using the energy budget with a wind speed profile (EBWSP)
method with measurements at one level, which employs the Penman equation,
as described in Brutsaert (1982). This method estimates the potential evaporation
(PET) from a wet surface. Hussey and Odum (1992) reported that ET in a
salt marsh is not significantly different from evaporation, thus, this
method should provide a reasonable estimate of the PET. When available,
the weather data was obtained from the meteorological station on North
Hog Island. Hurricane Bertha, however, destroyed this station in mid-July,
and made it necessary to use data from the Phillips Creek meteorological
station, located on the mainland of the Delmarva Peninsula until the Hog
Island station was repaired in late August. These stations record temperature,
humidity, incoming solar radiation, and wind speed every six minutes, and
the data is available at hourly intervals (Krovetz et al. 1996, Porter
& Spitler 1996). Net solar radiation was assumed to be 70% of total
incoming radiation (Crabtree & Kjerfve 1978). It was assumed that the
rate of water loss is constant when the marsh surface is exposed. The amount
of water lost to ET over a tidal cycle was calculated by multiplying the
rate of loss by the average amount of time that the surface is exposed.
An independent field measurement of ET was also obtained.
A portion of intact sediment containing vegetation was placed in a large
plastic bucket. The bucket was put in a large hole in the marsh, such that
the surface of the bucket was level with the surface of the marsh. A piezometer
outfitted with a HOBO was inserted into the bucket, and the water level
in the bucket was measured every 12 minutes for 6 weeks. Because subsurface
drainage is impeded by the bucket, the observed change in water level in
the bucket while the marsh surface is exposed reflects the rate of actual
ET. A best fit relationship where (head loss) = (time since exposure of
surface) was used to evaluate the change in water level per time. The head
loss was converted to an actual rate of ET by multiplying by the specific
yield. Uptake by the plants and evaporation are most likely able to extract
a greater amount of water from the sediment than simple gravity drainage,
and thus this estimate of ET was hypothesized be on the low side. In addition,
the fact that drainage was impeded most likely led to the build up of salts
in the sediment. If the plants are salt stressed, they may exhibit different
patterns of transpiration.
3.2.6 Water Budget
The HOBO data, once corrected as described above,
were divided into discrete ~12.5 hour ebb-flood tidal events. Each tidal
event was then subdivided into 'stages', where each point, representing
the head over a 12 minute interval, was assigned a stage. This resulted
in 63 different tidal stages. All values for each stage were averaged,
such that there was a value of the average hydraulic head for each piezometer
at every stage in the tidal cycle, and from this, all vertical and horizontal
fluxes involved in the water budget were calculated. The various components
used in the computation of the water budget were: vertical, horizontal,
and ET. All fluxes, with the exception of ET were calculated based on Darcy's
law. The control volume used for calculation of water fluxes is shown in
Figure 3.7. The control volume used is positioned
such that the x axis is perpendicular to the creek, y is parallel to the
creek, and z is vertical. The x and y dimensions are fixed at 100 cm each.
Because it was assumed that there was no flow in the unsaturated zone,
the height of the control volume in the z direction was equal to the height
of the water table relative to MLW at each point in time. Thus, the control
volume changed in size depending on the tidal stage. When the water level
was greater than the marsh surface, the top of the control volume was equal
to the surface of the marsh. It was necessary to assume that there was
no movement of water in the y direction. This assumption may introduce
error because it is likely that there was some element of flow parallel
to the creek discharging directly into the lagoon. However, this provided
a necessary simplification, and allowed for the use of a 2-D model.
The equations used to calculate the fluxes of water
are in derived in Appendix A. The equations are based on Darcy's law, and
use a finite difference approximation to calculate the discharge at each
12 minute time step. The discharge was calculated for each point in the
vertical and horizontal datasets, and the mean discharge at each stage
in the tidal cycle was obtained by averaging all values at that stage.
The net fluxes in and out of the control volume for each of the fluxes
illustrated in Figure 3.7 were calculated by summing
all fluxes in that direction. The values are reported in L m-2
(tidal cycle)-1, which is equivalent to 103 m3
m-2 (tidal cycle)-1. Mean discharge values for each
flux were compared between sites using the paired t-test function in SPSS.
The sum of the fluxes into and out of the control volume were assumed to
balance, and the degree to which the water budget does not close is assumed
to be the error associated with the measurements.
It was assumed that, on average, the water that entered
the subsurface during flooding left during the subsequent ebb event. A
means for estimating the groundwater discharge was the net flux of water
at low tide. I estimated the time when the change in head per time was
lowest, and took the average discharge at this point. Assuming a constant
groundwater discharge, this rate was applied over the entire tidal cycle
to estimate the groundwater discharge. Thus, the gradient present at low
tide was the "static" water table, and all other excursions of hydraulic
head gradients were assumed to be due to tidally induced fluctuations.
Nuttle and Harvey (1995) found that the best model for predicting groundwater
discharge to the marsh was obtained using a constant rate of discharge.
In the current study, their approach may have resulted in an overestimate
of the flux because it is likely that the head gradients due to groundwater
discharge were lower when there was a reverse gradient due to a rising
tide, and when the surface of the marsh was flooded. As before, the difference
in 'groundwater' flux between the two sites was evaluated using a paired
t-test for the discharge measurements at low tide (stages 36-40).
3.2.7 Pore water collection, nutrient and salinity
determination
Salinity of water in all of the piezometers was sampled
on a regular basis throughout July and August of 1996. The well was quickly
evacuated using a hand operated vacuum pump, and allowed to refill. Thus,
the water sampled was water that had recently entered the well from the
surrounding sediment. Salinity was measured using a temperature compensating
hand-held refractometer. Pore water samples were collected from the sippers
on a monthly basis from April to October 1996. Prior to collection of pore
water, any old water in the sipper was evacuated using N2 gas
to maintain an anoxic environment. A slight vacuum pressure was applied
to the sipper using a hand-pump, and pore water was allowed to refill the
sipper for a few hours. The water was extracted from the sipper using a
syringe. Redox potential (Eh, or platinum electrode potential), pH, and
temperature were determined immediately following collection by injecting
the sample into an anaerobic chamber fitted with Corning electrodes and
temperature probe attached to a Beckman 12 pH/ISE meter. The measured redox
potential in millivolts was adjusted by adding 199 mV to correct for the
Ag/AgCl, saturated KCl electrode used.
Water for NH4+ and PO4-3
analysis was immediately filtered through 0.45 m membrane filters into
vacutainer tubes containing 0.1 ml 6N HCl, and kept on ice. The acid was
added to prevent the volatilization of NH4+ during
the removal of H2S. The samples were brought back to the field
laboratory, and analyzed within 6 hours. Because H2S interferes
in the analysis of both NH4+ and PO4-3,
each sample was bubbled for 5-10 minutes with N2 gas in order
to drive off the H2S. Immediately prior to pipeting the samples
for analysis and addition of the color reagents, the pH of the samples
was readjusted with 0.1 ml 6N NaOH. Occasionally the pH of the sample was
tested using a pH meter to ensure that it was in the proper range for color
development. Both NH4+ and PO4-3
were determined spectrophotometrically as described in Grasshoff et al.
(1983). Ammonium was determined by the addition of 0.3 ml each of trisodium
citrate, phenol/nitroferricyanide, and hypochlorite reagents to 5ml of
sample and standard. The samples were incubated in the dark for at least
6 hr, and the absorbance was determined on a spectrophotometer at 630 nm.
The pH of the PO4-3 samples was adjusted to 8.0 by
titration with 6N NaOH and 1N H2SO4 using phenolphthalein
as an indicator. One ml of a combined color reagent containing ammonium
molybdate, sulfuric acid, ascorbic acid and potassium antimonyl tartrate
solutions was added to 5.0 ml of sample and standard. Color development
was allowed to proceed for 30 min, and the absorbance was read at 885 nm.
Sulfide concentration was determined using a method
described by Cline (1969) as modified by Otte and Morris (1994). Five ml
of unfiltered porewater was added immediately to a vacutainer tube containing
5 ml ZnAc. If necessary, dilutions were made, and 0.4 ml N,N-dimethyl-p-phenelynediamine
sulfate + ferric chloride dye was added to each sample and standard. At
least 20 min was allowed for color development, and the absorbance was
measured spectrophotometrically at 670 nm.
3.2.8 Spartina alterniflora production
and tissue element composition
Production was measured at each site in September,
which is the end of the growing season. Within each zone, (0-0.5 m, 0.5-1.0
m, 1.0-2.0 m and 2.0-3.0 m), three 1/16th m2 haphazardly
placed quadrats were thrown. The species composition was determined, and
all of the above ground S. alterniflora was clipped. The plants
were brought back to the laboratory and frozen until biomass determinations
were made. Each plant was cleaned of sediment and all dead leaves removed.
The plant height was measured to the tallest point (leaf or flower), oven
dried and weighed individually. Using the density and the average weight
per plant, the biomass per square meter of marsh surface was calculated.
S. alterniflora leaf samples were collected
from each site in June, August and October for tissue element determination.
Ten to fifteen leaves from different plants were collected and combined
in to a single sample from each site. The leaves were washed free of sediment,
and stored in a freezer. The samples were then lyophilized, and ground
to homogeneity using a Krups coffee mill. The carbon and nitrogen composition
of the tissue was then analyzed using a Carlo-Erba NA1500 Elemental Analyzer.
Three replicates of each sample were run in order to assess the analytical
error of the instrument, and the homogeneity of the samples.
3.2.9 Data Analysis
Differences between sites for porewater variables
and S. alterniflora production and tissue element composition were
evaluated using the General Linear Model - General Factorial function in
SPSS. Tranformations to achieve normality were done when necessary. For
the porewater variables, all monthly data was pooled. Likewise, the bimonthly
tissue element data were pooled.
3.3 RESULTS
3.3.1 Pore water nutrients and salinity
The porewater chemistry from the sippers can be found
in Table 3.1. For the porewater nutrients, NH4+ was
not significantly different between sites, but the PO4-3
is different (p<0.001). The sulfide concentration is significantly higher
(p=0.02) at the upland catchment; however, there is not a significant difference
between the Eh values between sites. There is a 12 ppt difference in mean
salinity between the two sites, with the upland catchment averaging 22
ppt and the salt-flat catchment averaging 35 ppt. There was a slight increase
in the salinity in August, probably due to an increase in ET resulting
from higher temperatures. July was extremely rainy, however (293 mm of
precipitation were received at the Phillips Creek met station), and it
is likely that this depressed the salinity during that month. However,
the depression in salinities was only noticeable at the upland catchment
site. For example, following Hurricane Bertha, which hit on July 12 and
13, the mean salinity at the upland catchment was 18 ppt, and several sippers
were as low as 2-5 ppt. However, at the salt-flat catchment the mean salinity
was 36 ppt. The mean distribution of salinity with depth is shown in Figure
3.8. This figure represents all measurements taken during July and August.
The salinity at depth was remarkably consistent over the sampling period.
At the upland catchment the salinity decreased from 19 ppt at 10 cm to
8ppt at 90cm, while at the salt flat catchment the salinity increased with
depth, from 38 ppt at 10 cm to 55 ppt at 104cm.
|
upland
catchment
|
salt-flat
catchment
|
NH4+ (mol
l-1)
|
11 [4]
|
8 [1]
|
PO4-3 (mol
l-1)
|
14 [2]
|
2 [1]
|
S-2 (mol l-1)
|
30 [10]
|
8 [1]
|
Eh
|
78.2 [10.6]
|
75.9 [5.2]
|
salinity (ppt)
|
22 [1]
|
35 [1]
|
Table 3.1. Porewater chemistry at 10 cm. Data
are mean values of eight samples from each site, taken monthly April through
October 1996 (n=56). Values are means for each site, and represent all
samples taken from each zone at the creekbank: 0-0.5 m, 0.5-1 m, 1-2 m,
and 2-3 m. Numbers in brackets represent the standard error of the mean,
and indicates a significant difference between sites (ANOVA, p<0.05
at = 0.05).
Table 3.2 contains the S. alterniflora tissue
element composition and the production measurements. The stem density between
sites is very similar, but the total biomass is significantly different
(p=0.02). There is >300% more biomass at the upland catchment site than
the salt-flat site. The standard error of the mean for these sites is largely
due to the range of production values along the transect. The biomass at
0.5 m is 1850 g m-2 and 531 g m-2 for the upland
and salt-flat sites, respectively, while the biomass at 3 m is 547 g m-2
and 126 g m-2. In general, the plants at the upland site are
heavier, taller and more robust than those growing at the salt flat site.
In addition, the plants at the upland site are putting more energy into
sexual reproduction: 13.9% of these plats were flowering as opposed to
5.6% at the salt flat site. The upland site plants have a higher C:N ratio,
and a lower %N than the salt flat plants.
3.8
|
upland
catchment
|
salt-flat
catchment
|
biomass (g m-2)
|
920.9 [223.1]
|
296.4 [62.2]
|
density (plants m-2)
|
443 [33]
|
432.0 [45.0]
|
flowering (%)
|
13.9 [2.2]
|
5.6 [0.7]
|
plant height (cm)
|
45.5 [10.6]
|
26.8 [1.4]
|
plant weight (g)
|
1.9 [0.3]
|
0.6 [0.1]
|
N (%)
|
1.05 [0.02]
|
1.21 [0.04]
|
C:N
|
48.3 [1.4]
|
41.2 [0.8]
|
Table 3.2. Spartina alterniflora
end-of-season production and tissue element composition. Production was
estimated in September 1996. The tissue element composition is a pooled
mean of all samples taken in June, August and October 1996. Values are
means for each site, and represent all samples taken from each zone at
the creekbank: 0-0.5 m, 0.5-1 m, 1-2 m, and 2-3 m. Numbers in brackets
represent the standard error of the mean, and indicates a significant difference
between sites (ANOVA, p<0.05 at = 0.05).
3.3.2 Sediment Hydrological Properties
Table 3.3 contains the results of the sediment properties
examination. The hydraulic conductivity values obtained for both the slug
test and the large core constant head test are shown. For all slug tests
the water level recovered to its initial level within 2-3 minutes, indicating
the high permeability of the sediment. All K values calculated from the
slug tests were averaged for each site. A log transformation of the data
was performed to assess if there is a log-normal distribution, but this
did not change the mean significantly, and is not reported. There is a
five fold discrepancy between the slug test K and the constant-head K,
with the field derived values being larger. This is most likely due to
the difference between the vertical and horizontal K. The laboratory measurement
represents primarily vertical flow, whereas the field experiment represents
flow in all directions from an approximately cylindrical volume of sediment
around the slotted portion of the piezometer (Bouwer & Rice 1976),
and perhaps represents a slight bias towards the horizontal. Although a
direct comparison between the two methods cannot be made, it is probable
that there is a difference between the vertical and horizontal K. For the
purposes of this study, the field derived measures were used, but it must
be kept in mind that this may provide an overestimate of the vertical discharge.
|
upland
catchment
|
salt-flat
catchment
|
K (cm s-1) field test
|
0.01 [0.0005]
|
0.01 [0.001]
|
K (cm s-1) lab test
|
0.003 [0.00005]
|
0.003 [0.00002]
|
porosity
|
0.45 [0.01]
|
0.43 [0.01]
|
specific yield
|
0.01 [0.0008]
|
0.01 [0.001]
|
organic matter (%)
|
0.74 [0.13]
|
0.66 [0.05]
|
sand (%)
|
88 [1]
|
87 [2]
|
silt (%)
|
2 [1]
|
3 [1]
|
clay (%)
|
10 [1]
|
10 [1]
|
Table 3.3. Sediment properties of each site.
The field derived hydraulic conductivity (K) was obtained using a slug
test, and the laboratory K using a constant head test. Values are pooled
means for each site, and the numbers in brackets represent the standard
error of the mean. The standard error for K and specific yield are included
to demonstrate the precision, or repeatability of the test, and do not
indicate the overall accuracy of the measurement. The indicates a significant
difference between sites (ANOVA, p<0.05 at = 0.05).
The specific yield is also shown in Table 3.3. No
hysteresis between drainage or recharge was found, and all replicate tests
were averaged for each site. The specific yield, which is the volume of
water loss per unit area for a given change in hydraulic head is 0.01 for
both sites. This value is at the low end of the expected range (Freeze
& Cherry 1979), but is similar to other measurements at neighboring
sites (Osgood 1996, Santos 1996). The porosity, ~0.45, is within the range
of values expected for sand (Freeze and Cherry 1979). The sediment organic
content is less than 1%, and the sand content is ~88% at both sites.
3.3.4 Evapotranspiration
The results of the Penman equation analysis yielded
an hourly rate of potential evaporation of 0.18 mm hr-1. This
value falls within the range of evaporation rates reported by Hussey and
Odum (1992). When applied over the time when, on average, the marsh surface
above the control volume is not flooded, this yields a total ET of 1.9
L m-2 (tidal cycle)-1. The relationship between the
time since the water level fell below the surface of the bucket and the
head loss was significant (r2=0.35, F=818.3, p<0.0001). The
measured ET was 1.2 L m-2 (tidal cycle)-1, which,
as stated above, is probably on the low side due to the use of the specific
yield to determine the relationship between the head loss and the actual
volume of water lost. However, these two estimates are close, and their
agreement lends confidence that my estimate of ET is in the correct range.
3.3.5 Hydraulic Head Gradients
There were significant differences in hydraulic head
between the two sites. Figures 3.9 and 3.10 show the head at 75 cm for
each of the sites during an ebbing tide and flooding tide. These values
are mean hourly heads, and represent the average at that point in the tidal
cycle over 85 tidal events. The mean head data for each piezometer, along
with the standard error of the means can be found in Appendix C. As shown
in Figure 3.9, the tide falls rapidly at both sites, such that 3 hours
following the high tide the water level has reached a nearly constant position.
At both sites, the water level remains constant until about 2.5 hours before
the next high tide, at which point a rapid rise is seen (Figure 3.10).
At the upland catchment, the water table at low tide is consistently higher,
by approximately 10cm, than the water table at the salt-flat catchment.
In addition, there is a noticeable gradient in hydraulic head at low tide
at the upland site, while there is no such gradient at the salt-flat site.
Upon inspection of the gradients from 4-9hr after high tide, it becomes
evident that the horizontal gradient at the upland site is greatest between
0.5 m and 1.5 m, and is least between 3 m and 5 m. Indeed, there is a slight
decrease from 3 m to 5m. At the salt-flat site there is no difference between
the heads at 50 and 75 cm at low tide, whereas the upland site exhibits
higher hydraulic head at 75 cm than at 50 cm at low tide. Thus there is
upward vertical discharge at the upland site at low tide, and no discharge
at the salt flat site.
3.3.5 Discharge/Recharge
The mean instantaneous discharge values over a tidal
cycle are shown in Figure 3.11. The net discharge
estimates can be found in Table 3.4, and mean discharge for each of the
fluxes defined in Figure 3.7 are in Appendix D.
Figure 3.11 illustrates that the head gradients
seen at the upland site in Figures 3.9 and 3.10 forced a substantial discharge
of water from the marsh, even at low tide. The discharge rate at low tide
remained at a constant for nearly 6 hr. In contrast, at the salt flat site,
the tidal water seemed to drain rapidly, and the discharge fluctuated around
zero for several hours. At both sites, the predominant direction tidal
recharge is vertical rather than horizontal. Even while there is a net
discharge of water from the marsh horizontally, there is a net recharge
of water into the subsurface vertically. The vertical recharge peaks approximately
1.5 hours before high tide, and then decreases. Approximately 2 hr before
high tide the horizontal flux at both sites switches so that there is recharge
rather than discharge. However, in both cases the horizontal recharge is
small compared to the vertical recharge.
3.3.6 Water Budget and Groundwater Fluxes
Table 3.4 contains the elements of the water budget,
and Table 3.5 contains the estimates of groundwater flux obtained by carrying
the discharge at low tide over the entire tidal cycle. The greatest influx
of water to the salt flat site, 37.9 L m-2 (tidal cycle)-1
comes from downward vertical recharge during the flooding tide. The greatest
influx of water to the upland site ( 92.1 L m-2 (tidal cycle)-1)
is also vertical, but it is upward flow, presumably from groundwater drainage
from the upland. The greatest efflux of water at both sites occurs as horizontal
discharge towards the creekbank. This estimate assumes that there is no
change in storage over the tidal cycle, and thus the budget, ideally, should
balance. However, neither water budget closes perfectly, and the difference
between the flux of water into the control volume and the flux out of the
control volume may be taken as the error of the estimate. Thus, there is
an approximate error of 15 L m-2 (tidal cycle)-1.
The net flux of groundwater over a tidal cycle, assuming a constant rate
of discharge, is 75.8 L m-2 (tidal cycle)-1 and 13.2
L m-2 (tidal cycle)-1 for the upland and salt-flat
sites, respectively. This estimate does not assume that there is no change
in storage within the control volume. Given the error estimate above, the
discharge of groundwater from the upland site is probably not significantly
different from zero.
|
upland catchment
|
salt flat catchment
|
equation
|
Qv - up
|
92.1 [19.5]
|
14.3 [14.3]
|
B-3a
|
Qv - down
|
28.1 [14.8]
|
37.9 [29.5]
|
B-3b
|
QCBA - in
|
0.4 [5.3]
|
7.4 [10.6]
|
B-7c
|
QDCB - in
|
40.2 [5.9]
|
24.6 [4.6]
|
B-7a
|
total in
|
160.8
|
84.2
|
B-8a
|
QCBA - out
|
173.7 [10.6]
|
53.2 [10.4]
|
B-7d
|
QDCB - out
|
2.2 [5.2]
|
15.3 [9.2]
|
B-7b
|
ET
|
1.9 [0.1]
|
1.9 [0.1]
|
|
total - out
|
177.8
|
70.4
|
B-8b
|
in-out
|
-17.0
|
13.8
|
|
Table 3.4. Discharge measurements. All values
are in L m-2 (tidal cycle)-1, which is equivalent
to 10-3 m3 m-2 (tidal cycle)-1. Figure
3.7 illustrates these fluxes. Qv - up is where the head is greater
at 75 cm than at 50 cm, and Qv - down is where the head is greater
at 50 cm than at 75 cm. Both are considered positive fluxes (inflows).
For the horizontal fluxes, QCBA and QDCB, the in
and out fluxes are denoted relative to the control volume. The difference
between the total-in and the total-out (in-out) can be taken as the error
of the measurements. Numbers in brackets are the standard error of the
mean. A indicates a significant difference between the two sites in a paired
t-test, and a indicates that no statistical test was performed. The equation
number indicates the equation from Appendix B used to calculate each flux.
|
Upland catchment
|
salt flat catchment
|
Qv - up
|
104.2 [10.5]
|
-
|
Qv - down
|
-
|
18.0 [8.1]
|
QDCB - in
|
50.2 [6.8]
|
14.0 [1.7]
|
total in
|
154.4
|
32.0
|
QCBA - out
|
230.2 [4.5]
|
45.2 [4.5]
|
total - out
|
230.2
|
45.2
|
Qnet
|
-75.8
|
-13.2
|
Table 3.5. Estimated groundwater fluxes from
each site. All values are in L m-2 (tidal cycle)-1,
which is equivalent to 10-3 m3 m-2 (tidal cycle)-1.
See Figure 3.7 for a definition of the fluxes. Both up and down vertical
flows represent inflows. Numbers in brackets are the standard error of
the mean. A indicates a significant difference between the two sites in
a paired t-test, and a indicates that no statistical test was performed.
GW is the groundwater flux estimated, as described in the text, by applying
the mean discharge at low tide over the entire tidal cycle.
3.4 DISCUSSION
3.4.1 Limitation of S. alterniflora growth
by high salinity
The difference in S. alterniflora biomass
between the two sites is significant. While there is no difference in stem
density, the creek with the upland drainage area has ~300% greater end-of-season
biomass than the one attached to the salt flat owing to the fact that the
plants are 3x heavier. In addition, the plants at the upland site have
2.5x the percentage of flowers, indicating that they have a higher level
of sexual reproduction. It is possible that the difference in porewater
P concentration could explain this difference, however, Osgood & Zieman
(1993a) showed that S. alterniflora in these young barrier island
marshes is N limited. In addition, the N:P, which is <1 at the upland-attached
site and 3 at the salt-flat site are well below levels where we would expect
that P may limit growth. Sulfide is higher at the site with greater production,
thus that can be ruled out as a limiting factor as well. Thus, salinity,
which, on average over the growing season is 12 ppt higher at the salt
flat site than at the upland site, stands alone as the potential limiting
factor.
The C:N ratio, as well as the %N in the plant tissue,
can be used as further evidence to support the idea that salinity is limiting
growth at the salt flat site. As discussed earlier, S. alterniflora
and other halophytes, will accumulate solutes in the form of nitrogenous
organic compounds such as proline and glycinebetaine (Flowers et al. 1977,
Cavalieri & Huang 1981, Colmer et al. 1996). This usurps valuable N
and makes it unavailable for growth related compounds such as chlorophyll
and RuBP carboxylase. The %N in the plant tissue is significantly higher,
and the C:N is significantly lower at the salt flat site than at the upland
site. This indicates that although the overall availability of N is low,
the plants under salt stress are concentrating N in their tissues. However,
at the upland attached site there is 9.7 g N m-2 in aboveground
biomass versus 3.6 g N m-2 at the salt flat site. Therefore,
even though the observed N concentrations are the same, the plants at the
upland site are better able to use it. There are two potential reasons
for this. First, the excess cations at the salt flat site could be interfering
with the plants ability to take up NH4+, as suggested
by Bradley and Morris (1991). Second, this could lend support to Osgood's
(1996) theory that greater flushing and through-flow of water creates more
total N available for the plants, even though these concentrations may
not be observed in "snap-shot" concentration profiles. Regardless of the
mechanism, it can be stated that the plants at the upland site were able
to procure a greater amount of N over the course of the growing season,
and that this additional N was used for the photosynthetic production of
a greater amount of above-ground structural material. Further, it is likely
that the higher salinities at the salt flat site stressed the plants to
the point that they were unable to grow as well, such that only ~6% of
the plants were sexually reproductive. Below-ground biomass was not directly
measured, however, coring into the sediment at the upland site was much
more difficult than at the salt flat site due to the number and size of
the roots and rhizomes present. This can be taken as potential evidence
that the below ground biomass at the upland site is greater.
3.4.2 Subsurface hydrology and groundwater
discharge
Given that a difference in salinity is the factor
determining the relative S. alterniflora health between these two
sites, it follows that there must be differences in the subsurface hydrology
between them which is controlling the salinity. As described previously,
the two sites do not differ greatly in local elevation or in slope. Thus,
the interaction between the marsh surface and the seawater surface cannot
be controlling the hydrology in the creekbank region, and it is necessary
to look further to determine the cause of the variability in salinities.
In this case, it is the geomorphological difference of the catchments in
terms of size and type that is the apparent difference between the two,
and that controls the position of the interface between the freshwater
and the salt water surfaces.
Figure
3.12 is a schematic diagram illustrating the relative magnitude of the
groundwater fluxes between the two sites. This diagram illustrates that
the larger catchment is responsible for greater fluxes into the marsh.
In addition, because this catchment contains a pond, there is a reservoir
of freshwater that provides a continual source to the marsh. The effect
of the difference in catchments can be seen in the head gradients, the
discharge, and the resulting salinity within the marsh.
The water table at low tide is consistently 9 cm
higher at the upland site than at the salt-flat site, resulting in a smaller
unsaturated zone. The ratio of the volume of unsaturated sediment per m2
of marsh surface between the two sites (upland:salt-flat) at low tide is
0.77. Thus, more seawater can infiltrate into the sediment at the salt-flat
site per tidal cycle simply because there is a greater void volume. Indeed,
the volume of water 3.12entering due to tidal fluxes (CBAin
+ Qv-down) is 28.5 and 45.3 L m-2 (tidal cycle)-1
at the upland and salt flat sites, respectively. The ratio of these
discharges is 0.63 (upland:salt-flat). Thus ~35% more salt water enters
the subsurface at the salt-flat site due to tidal influxes, and this recharge
is proportional to the amount of 'space' available in the subsurface. The
higher 'static' water table at the upland catchment site is presumably
due to perpetual discharge from the upland. Evidence for this comes from
the fact that at low tide there are both horizontal and vertical hydraulic
head gradients which are causing flow out of the control volume within
the creekbank zone at the upland site. In opposition to this, the low tide
water table at the salt-flat site is virtually flat, indicating little
or no flow out of the marsh.
The net groundwater flux at the salt-flat site, 13.2
L m-2 (tidal cycle)-1 is sufficiently small that
it is probably not different from zero. However, the flux at the upland
site, 75.8 L m-2 (tidal cycle)-1 is sufficiently
large to be distinguished from zero. It is likely that this is an overestimate
of the real groundwater discharge into the marsh due to the probable variation
in the groundwater discharge rate over a tidal cycle. When the surface
of the marsh is flooded, the rate of discharge is less than when the surface
is exposed. Thus, these estimates must be used carefully. However, the
net upward discharge of groundwater (104.2 L m-2 (tidal cycle)-1)
is similar in magnitude to the total upward discharge (92.1 L m-2
(tidal cycle)-1) at the upland site.
As shown in Table 3.4, the water budget at each site
did not close perfectly, which indicates that there are some errors associated
with the reported fluxes. It is likely that there is the greatest error
in the vertical discharge estimates because head was only measured at two
depths. A greater number of piezometers would have increased the accuracy
of the vertical flux measurements. In addition, the piezometer transects
were located perpendicular to the creek, and there is probably discharge
parallel to the creek (a divergence of flow in the y direction) that is
unaccounted for. Finally, as mentioned previously, there may be differences
between the vertical and horizontal permeabilities that lead to an overestimate
of the vertical fluxes. Despite these potential sources of error, there
is a significant difference in the magnitude of the fluxes in nearly all
directions between the two sites. The estimates of the groundwater fluxes
are substantially different between sites, and this difference is great
enough to suggest that it is the cause of the observed differences in salinity.
The salinity-depth profile (Figure 3.8) supports
the idea that there is fresh groundwater flowing upwards into the marsh
at the upland site. At this site, the salinity at depth is much lower than
it is at the surface, indicating a dilution of the salts brought in by
the tides. The salts are most likely removed from the sediment adectively
by the groundwater flow. The discharge of freshwater at this site is evidently
sufficient to counteract the salt-concentrating effects of ET, and is able
to maintain the salinity at a level below that of the flooding lagoon (~33
ppt, pers. obs.). In contrast, the salinity profile at the salt flat site
indicates that there is very little dilution of salts, and any water coming
from the up-gradient catchment must have high salinities. The salinities
at 10 cm below the surface within the salt flat are 70-80 ppt (pers. obs.),
and most likely increase with depth, as they do at the creekbank. Thus,
water flowing into the creekbank area adds salt to the system. At this
site, the only mechanisms of salt removal are tidal flushing of the creekbank
and diffusion through the marsh surface when it is inundated. ET increases
the salinity, as does the discharge from the salt flat.
Nuttle and Harvey (1995) estimated the discharge
of groundwater to a mainland Virginia marsh based on the change in hydraulic
head and the change in storage in the sediment. They argue that this approach
is inherently more accurate than estimates of discharge based on hydraulic
conductivity. One reason for this argument is the variability and uncertainty
associated with estimates of hydraulic conductivity. However, this is relevant
for cases where the conductivity is low, and exhibits a great deal of spatial
variability. In this case, the sediment consists of unconsolidated sands,
and the measured K is somewhat consistent. They also cite that macropores
and roots act to increase the vertical K, and thus estimates of discharge
using a bulk K are likely to be underestimates. This is a possibility in
back barrier marshes as well. However, as discussed previously, the K measured
in situ and that measured in the lab differed by approximately one order
of magnitude. This may be due to differences in the vertical and horizontal
conductivity. If this is the case, the vertical K is substantially less
than the horizontal K, and thus estimates of vertical discharge using the
in situ K will be biased. Thus, my estimate is biased towards the high
side. Finally, Nuttle and Harvey (1995) argue that the osmotic potential
created by flow from one geological stratum with a low solute concentration
to a stratum with a high solute concentration can result in osmotic 'head'
that is much greater than measured head gradients. As discussed in Freeze
and Cherry (1979, p. 105) this type of osmotic pressure differential is
observed when there is a clay, which acts as a semi-permeable membrane
between the two geological strata. The upland, which is the source of fresh
groundwater on Hog Island, has sediments of the same type and source as
those present in the marsh. Thus, the biases which may lead to the inherent
inaccuracy of discharge estimates made from Darcy's law in highly organic,
silt/clay marsh sediments where there is groundwater discharge from a very
deep aquifer are unlikely to be important here. In addition, estimating
discharge from the change in storage alone would neglect water that flows
through the marsh, but which does not result in a change in storage in
the sediment. At the upland site there is a head driven flux of water out
of the sediment even when there is no change in storage, i.e. when the
water table is static, at low tide.
The published values for groundwater discharge at
estuarine shorelines vary from 0.2 liters m-2d-1
to 105 liters m-2d-1 (from summary in Harvey &
Odum 1990). At the salt flat site, the net discharge of water per m2
per day is 25.3 L; the net discharge at the upland site is 145.5 L. The
upland site has a much higher discharge than any previously reported value.
There are a few reasons why this is so. First, these sediments, which are
characteristic of barrier islands have a high conductivity compared to
other marsh sediments. For example, the Belle Isle marsh, in Massachusetts
has a K=1.4 x 10-4 cm sec-1 (Nuttle & Hemond
1988), and the Carter Creek marsh in Virginia has a K=1 x 10-3 cm
sec-1 (Harvey & Odum 1990). These values are 10-100 times
lower than the conductivity of the marshes on Hog Island. Such conductivities
would change the discharge by 1-2 orders of magnitude. Second, this study
takes place in a very narrow, fringing back-barrier marsh. The upland,
which is the source of the groundwater, is much closer to the marsh than
in the more expansive marshes characteristic of the mainland. Finally,
the summer of 1996 was unseasonably wet. Hurricane Bertha in July, and
Hurricane Fran in September, were responsible for providing a great deal
of freshwater to the upland catchment. The previous year, which was unseasonably
dry, resulted in higher mean salinities: 33 ppt and 39 ppt for the upland
and salt flat sites, respectively (Krovetz et al. 1996). The pond that
drains, through the subsurface, into the upland catchment site is traditionally
an ephemeral pond which dries up during the summer (pers. observ.). During
1996 there was fresh to brackish water in the pond over the entire summer,
while in 1995 the pond was dry by July. Thus, this estimate of groundwater
discharge is perhaps higher than that seen on average at this site.
3.4.4 Conclusions
Regardless of the discrepancy between the 1996 season
and an 'average' year, the comparison between the two marshes for this
particular season is robust. It is difficult to say with certainty what
the actual discharge of groundwater is out of this marsh, but it is quite
evident that there is a difference in discharge between the two systems,
and that this difference is due to the geomorphological difference between
the two catchments. The marsh that drains an upland catchment has higher
discharge, lower salinities, and greater S. alterniflora production
than the marsh which has only a salt-flat as its source of extra-tidal
water. Due to the geomorphology of the upland site, the interface between
the fresh water table surface and the seawater table surface exists within
the marsh proper, whereas at the salt-flat site, these two surfaces do
not interact within the marsh. Thus, the link between the landscape, the
subsurface hydrology, the sediment physico-chemistry, and the production
of S. alterniflora is established.
3.4.5 Implications
In recent years, there has been an increased effort
made to mitigate salt marsh destruction by restoring damaged marshes, and
by creating new ones (Race & Christie 1982, Mitsch & Wilson 1996).
Narrow, fringing marshes are particularly important in stabilizing shorelines
(Lugo & Brinson 1979). Studies such as this one, which examine the
structure and function of the salt marsh ecosystem during the early stages
of marsh development, are important in facilitating an understanding of
how a 'natural' young marsh works. If the S. alterniflora production
is high, there will be an accelerated build-up of organic matter in the
marsh. The decomposition of this organic matter will provide nutrients
to fuel greater production in following seasons. Often, the failure of
mitigation projects is due to improper hydrologic conditions (Mitsch &
Wilson 1996). In addition, failure can also be attributed to the establishment
of created wetlands in inappropriate settings, where the landscape and
larger ecosystem have been ignored (Race & Fonseca 1996)). The implications
of this study are that the subsurface hydrology, which is evidently important
in controlling the production of S. alterniflora, is controlled
to a great extent by the large scale landscape of the area, and that if
the hydrology is established for optimal productivity, then the marsh will
develop more rapidly, and will more quickly achieve functional maturity.