Nitrogen interception and export by experimental salt marsh plots exposed to chronic nutrient addition

Abstract

Mass balance studies conducted in the 1970s in Great Sippewissett Salt Marsh, New England, showed that fertilized plots intercepted 60 to 80% of the nitrogen (N) applied at several treatment levels every year from April to October, where interception mechanisms include plant uptake, denitrification and burial. These results pointed out that salt marshes are able to intercept land-derived N that could otherwise cause eutrophication in coastal waters. To determine the long-term N interception capacity of salt marshes and to assess the effect of different levels of N input, we measured nitrogenous materials in tidal water entering and leaving Great Sippewissett experimental plots in the 2007 growing season. Our results, from sampling over both full tidal cycles and more intensively sampled ebb tides, indicate high interception of externally added N. Tidal export of dissolved inorganic N (DIN) was small, although it increased with tide height and at high N input rates. NH4+ export was generally 2 to 3 times NO3– export, except at the highest N addition, where DIN export was evenly partitioned between NO3– and NH4+. Exports of dissolved organic N were not enhanced by N addition. Overall, export of added N was very small, <7% for all treatments, which is less than earlier estimates. Apparent enhanced tidal export of N from N-amended plots ceased when N additions ended in the fall. Nitrogen cycling within the vegetated marsh appears to limit N export, such that interception of added N remains high even after over 3 decades of external N inputs.

Full text

MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
Vol. 400: 3–17, 2010
doi: 10.3354/meps08460 Published February 11
INTRODUCTION
Salt marshes provide important ecological services
in coastal settings, including interception of land-
derived nitrate (NO3–) and export of energy-rich
reduced nitrogen (NH4+or dissolved organic nitro-
gen [DON]) (Valiela 1983). Urbanization in coastal
watersheds has increased land-derived nitrogen (N)
loads— primarily as NO3–— to coastal waters world-
wide (McClelland & Valiela 1998, Cloern 2001, Valiela
& Bowen 2002). These mainly anthropogenic N loads
(Valiela et al. 1997) have led to eutrophication and
associated detrimental effects in estuaries and other
shallow coastal waters (Baden et al. 1990, Valiela et al.
© Inter-Research 2010 · www.int-res.com*Email: lindsay_brin@brown.edu
FEATURE ARTICLE
Nitrogen interception and export by experimental
salt marsh plots exposed to chronic nutrient addition
Lindsay D. Brin1, 2,*, Ivan Valiela2, Dale Goehringer3, Brian Howes3
1Brown University, Department of Ecology & Evolutionary Biology, 80 Waterman Street, Box G-W, Providence,
Rhode Island 02912, USA
2The Ecosystems Center, Marine Biological Laboratory, 7 MBL St., Woods Hole, Massachusetts 02543, USA
3School of Marine Science and Technology, University of Massachusetts, Dartmouth, 706 South Rodney French Blvd.,
New Bedford, Massachusetts 02744, USA
ABSTRACT: Mass balance studies conducted in the
1970s in Great Sippewissett Salt Marsh, New England,
showed that fertilized plots intercepted 60 to 80 % of
the nitrogen (N) applied at several treatment levels
every year from April to October, where interception
mechanisms include plant uptake, denitrification and
burial. These results pointed out that salt marshes are
able to intercept land-derived N that could otherwise
cause eutrophication in coastal waters. To determine
the long-term N interception capacity of salt marshes
and to assess the effect of different levels of N input,
we measured nitrogenous materials in tidal water
entering and leaving Great Sippewissett experimental
plots in the 2007 growing season. Our results, from
sampling over both full tidal cycles and more inten-
sively sampled ebb tides, indicate high interception of
externally added N. Tidal export of dissolved inorganic
N (DIN) was small, although it increased with tide
height and at high N input rates. NH4+export was gen-
erally 2 to 3 times NO3–export, except at the highest N
addition, where DIN export was evenly partitioned
between NO3–and NH4+. Exports of dissolved organic
N were not enhanced by N addition. Overall, export of
added N was very small, < 7 % for all treatments, which
is less than earlier estimates. Apparent enhanced tidal
export of N from N-amended plots ceased when N
additions ended in the fall. Nitrogen cycling within the
vegetated marsh appears to limit N export, such that
interception of added N remains high even after over 3
decades of external N inputs.
KEY WORDS: Spartina salt marsh · New England ·
Nutrient addition · Nitrogen export · Nitrogen uptake ·
Dissolved inorganic nitrogen · Dissolved organic
nitrogen · Nitrate · Ammonium
Resale or republication not permitted without
written consent of the publisher
Salt marshes limit the amount of land-derived nitrogen
carried by ebbing tides out to receiving coastal waters
Photo: Ivan Valiela
O
PEN
PEN
A
CCESS
CCESS

Mar Ecol Prog Ser 400: 3–17, 2010
1990, 1992). Land-derived N loads may be intercepted
by fringing salt marshes located between terrestrial
sources of N and downstream ecosystems (Howes et al.
1996). In particular, biologically active NO3–can be
intercepted within salt marshes by plant uptake, deni-
trification and burial (White & Howes 1994, Valiela &
Cole 2002); at the ecosystem level, much interception
is through creek-bottom microbial processes (Howes et
al. 1996, Hamersley & Howes 2003).
In earlier work in Great Sippewissett salt marsh, MA,
Valiela et al. (1973) showed that salt marsh plots re-
ceiving moderate levels of N addition for 2 to 3 yr were
exporting 20 to 40% of added N, while the remaining
60 to 80% was intercepted. It has since been demon-
strated that experimental plots had increased denitrifi-
cation rates (Hamersley & Howes 2005) and increased
N uptake by plants, leading to greater biomass and
also a higher percentage of N in the biomass (Valiela et
al. 1975, Vince et al. 1981). However, denitrification
rate in estuaries has been shown to increase to an as-
ymptote, and as N input increases, the percent of N in-
put that is removed by denitrification often decreases
(Seitzinger 1990, Valiela 1995). Assuming that there is
a biological limit to the seasonal amount of denitrifi-
cation and plant biomass N storage that can occur in
salt marsh plots over a growing season, the degree
to which the interception capability could extend to
greater amounts of N is uncertain.
Furthermore, because anthropogenic, land-derived
N loading to natural marshes is continuing indefinitely,
the question was also raised as to whether the inter-
ception capability of these experimental plots could be
sustained under chronic, long-term, external N load-
ing, or whether this capacity might be degraded. As
salt marshes mature, they may shift from net import to
net export of dissolved and particulate matter (Dame &
Allen 1996); however, this is over very long time scales
and not necessarily in response to N loading, leaving
open the question of the longevity of the salt marsh
ecosystem service of N interception.
In the present study, we investigated the long-term
ability of salt marshes to intercept chronically added N,
as well as the potential effect of input rates on inter-
ception and export. To do so, we conducted a 2-part
study of the experimental salt marsh plots at Great
Sippewissett salt marsh, which have received N addi-
tions at different fertilization rates approximately
biweekly during the growing season for over 3 de-
cades. We first investigated the pattern of N import
and export over the entire tidal cycle by measuring
nutrient concentrations of flood and ebb water
exchanging through the tidal creek that bisects each
experimental plot. Second, we conducted time-course
sampling of NO3–, NH4+and DON in ebbing tidal
water, and used tidal volumes to determine the mass of
N exported from each replicate plot for all treatments.
These periodic field surveys were coupled with a plot-
specific tidal volume approach to estimate dissolved
inorganic N (DIN) export over the growing season. To
better refine the level of N retention, estimates of
seasonal export from N addition plots were compared
to parallel estimates for unamended control plots to
determine the mass of N exported relative to the
amount of N applied.
MATERIALS AND METHODS
Study site. We sampled 10 experimental plots in
Great Sippewissett salt marsh, Falmouth, MA (Valiela
et al. 1973). The plots are 10 m in radius and bisected
by small creeks, which are connected to Buzzards Bay
via a main creek (Fig. 1). Plots were established in sites
containing creek-bed, creek banks, low marsh and
high marsh habitats. Species composition of the vege-
tation has been mapped over the decades of study
(Valiela et al. 1985, Rogers et al. 1998, Fox 2007), and
we can generalize that tall-form Spartina alterniflora
grows on creek banks, while intermediate and short-
form S. alterniflora grows in low marsh areas. High
marsh habitats are dominated by S. patens and Dis-
tichlis spicata, and some S. alterniflora and Iva frutes-
cens may also be present (Hersh 1996, Fox 2007).
We examined 2 replicate plots of 5 treatment levels,
for a total of 10 plots. Control (C), low fertilization (LF)
4
Upland
Seaward
XF
XF
C
C
C
LF
LF
HF
HF
U
N
20 m
41°35’ 3.1’’N
70° 38’17.0’’ W
Upland
Fig. 1. Experimental plots in Great Sippewissett salt marsh.
Plots labeled C are controls; the other plots were fertilized
with mixed NPK fertilizer (low fertilization [LF]: 0.9; high fer-
tilization [HF]: 2.6; extra-high fertilization [XF]: 7.8 g N m–2
wk–1) or just urea (U: 2.6 g N m–2 wk–1) and superphosphate.
Upland areas are wooded. Modified from Fox (2007)

Brin et al.: Salt marsh nitrogen loading and export
and high fertilization (HF) plots were established in
1970, urea (U) plots in 1971 and extra-high fertilization
(XF) plots in 1974 (Fig. 1). Fertilization treatments were,
and continue to be, applied to each plot approximately
every 2 wk throughout the April to October growing
season. LF, HF and XF plots receive a mixed fertilizer
sold commercially as Milorganite supplemented with
enough urea to be 10% N, 6% P and 4% K. Fertilizer
levels were set at 10, 30 and 90 times the recommended
annual dosage for oats (0.9, 2.6 and 7.8 g N m–2 wk–1,
respectively); over the season, this amounts to approxi-
mately 0.7, 2.2 and 6.5 times the annual short Spartina
alterniflora N demand, respectively (White & Howes
1994). Urea plots received the same amount of N as HF
plots, but entirely as urea instead of mixed fertilizer.
Furthermore, one half of each plot also received P as
superphosphate; this treatment was designed to exam-
ine the effect of N alone. In earlier studies, the effects of
N alone were indistinguishable from those of the mixed
fertilizer (Valiela et al. 1985).
Sampling phase 1: full tidal cycles. To define the
time course of nutrient concentrations of tidal water
entering and leaving the experimental plots during
entire tidal cycles, we took hourly water samples over
the course of a tidal cycle at the mouth of the creek in
each plot. During each flood tide, water level increased
in the creeks before spilling onto the marsh platform,
and the reverse occurred at ebb tide. We began sam-
pling at low tide, continued through flood and ebb tide,
and finished when water stopped flowing out of each
creek. Sample salinity was measured with a refrac-
tometer, and samples were filtered and then stored in
acid-washed bottles on ice until taken to the labora-
tory. Samples were analyzed for NO3–and NH4+con-
centration using a Lachat Autoanalyzer (QuikChem
FIA+ 8000 series) and standard colorimetric methods
for brackish water samples (Flow injection analysis:
QuikChem Method 31-107-04-1-E for NO3–and Quik-
Chem Method 31-107-06-1-B for NH4+). This sampling
program was carried out twice during June and July
2007. Both sampling dates were 10 d after the most
recent fertilization, and predicted high tide heights for
both dates were within 1 SD of the mean of all pre-
dicted high tide heights over the fertilization period.
To describe the time course of the tide, tide height
was measured using a meter stick attached to a perma-
nent tide stake in each plot. To convert tide height
measurements to absolute height above mean sea
level (MSL), the elevation of the base of each tide stake
relative to MSL was determined by use of a Trimble
GPS unit (Trimble 4800 receivers, Project Datum
NAD1983 [Conus], Geoid Model GEOID99 [Conus]);
3 measurements were taken at each point and an aver-
age elevation was used to convert measured tide
heights to MSL values.
Sampling phase 2: export and interception of NH4+,
NO3–and DON. Ebb tide sampling: The similarity of
flood tide nutrient concentrations between treatments,
as revealed in the Phase 1 results, indicated that analy-
sis of ebb tides associated with control and treated
plots was sufficient to determine the N loss resulting
from N amendment. This simplification of approach
allowed for additional data collection across a variety
of tide heights and weather conditions to capture the
large variation in nutrient concentrations and tidal vol-
umes due to seasonal changes and other environmen-
tal factors. Eleven tides were sampled between August
and November 2007; 9 were sampled during the fer-
tilization period and 2 were sampled in November, to
determine whether N export continued after fertiliza-
tion ceased for the year. Water sampling was con-
ducted at 20 min intervals throughout each ebb tide,
from high slack until there was no longer tidewater
outflow from the creek in each plot. Water samples
were integrated throughout the water column, with
care taken to avoid any potential freshwater at the sur-
face and porewater just above the sediment. Salinity
and tide height were measured and recorded for
each sample, and samples were collected, stored and
analyzed as described above. In addition, on 3 dates in
August and November, samples were analyzed for
DON concentrations using a Lachat Autoanalyzer
(QuikChem FIA+ 8000 series) on persulfate digested
samples with correction for NO3–content (D’Elia et al.
1977). These 3 dates were chosen to represent the
approximate degree of DON export during and after
the growing season (August and November) because
they had near-average high tide heights, which our
DIN analysis showed to be an important factor in
determining export.
Estimation of tidal volumes: To calculate export of
NO3–and NH4+, we needed not only measurements of
nutrient concentration, but also estimates of tidal water
exchange. To measure tidal volume, we used a
detailed description of the topography of each plot as
well as the tide height at the time of each sample. To
define the elevation of each plot, we entered detailed
contour maps of each plot into ArcGis 9.1 (ESRI), and
interpolated data to create files with approximately
7800 northing, easting, and elevation points for each
plot (L. Fox unpubl. data). The relative elevations from
these data files were corrected to elevation above MSL
by comparing the elevations of several known points
with measurements taken with a Trimble GPS unit, as
described above. From the elevation data files, we also
determined geometric mean elevation above MSL of
each plot.
To calculate the volume of water above a plot for any
given tide height, we created a MATLAB program
(MATLAB 6.5.0.180913a, Release 13; MathWorks) that
5

Mar Ecol Prog Ser 400: 3–17, 2010
used as input the tide height on the plot relative to
MSL, determined the height of water above each point
in the elevation data file for that plot, estimated the
area that each point represented, used water height
and representative point area to calculate an approxi-
mate volume and summed these volumes to approxi-
mate total water volume.
Horizontal flow between the creek bank and salt
marsh sediment is negligible compared to these mod-
eled flow volumes. Porewater draining from creek
banks has been found to be a small fraction of the
hydrologic budget of vegetated marsh sediments in
Great Sippewissett salt marsh (Howes & Goehringer
1994). Furthermore, flood tidal water enters marsh sed-
iments virtually entirely to replace water lost by evapo-
transpiration, and there is minimal transfer to adjacent
tidal creeks (Dacey & Howes 1984).
Calibration of flux model: We assessed the accuracy
of the tidal volume calculated from the topographic
model for each of the plots by comparison with mea-
sured flows in late October 2008. In each plot, ebb tidal
volume was directly measured from a time series of
water velocity and wetted cross-sections of each creek
where it enters the plot at the seaward edge. Flow was
measured by a Marsh McBirney electromagnetic flow
meter (FlowMate 2000) on a top-setting wading rod.
Velocities were measured using open channel flow
measurement techniques, with multiple measurements
across the tidal channel. Due to the channel configura-
tions and vegetation growing within some of the creeks,
accurate flow measurements, and therefore calibra-
tions, could be made on only 7 of the 10 plots, but did
include at least 1 replicate from each treatment.
Export calculations: To determine the amount of
DIN or DON exported from each plot over each sam-
pled ebb tide, we multiplied measured nutrient con-
centrations by the estimated water volume leaving
each plot over the relevant sampling interval, and
these values were summed for a total export over the
ebb tide. Specifically, the volume multiplier for a single
sample was calculated as the average of the water vol-
umes over the plot during the previous and the current
samples, minus the average of the water volumes over
the plot during the current and the subsequent sample.
For example, the amount of N as NO3–lost for sample y
was calculated as:
[NO3–]y×[average(voly–1
,voly) – average(voly,voly+1
)]
where [NO3–]yis the concentration of NO3–in water
sample y, volyis the water volume on the plot at the
time of sample y, voly– 1 is the water volume on the plot
at the previous sampling time and voly+1 is the water
volume on the plot at the subsequent sampling time.
For the first sample, the volume of water at high tide
was used in place of the first average, and for the last
sample, the volume of water left on the plot at low tide
was used in place of the second average.
Estimation of DIN export over the fertilization
period. The calculations described so far provided esti-
mates of N export during the tides that were actually
sampled. These data were extrapolated to the rest of
the experimental period, with the assumption that the
sampled tides were representative of the tides during
the fertilization period, April to October. To test this
assumption, we compared the frequency distribution
of tide heights on our sampling days to the distribution
over the entire fertilization period using a χ2goodness-
of-fit test.
To estimate DIN export over the fertilization period,
we first determined the history of tide heights for Great
Sippewissett salt marsh for April to October. This was
accomplished by relating available predictions of tide
heights for an adjacent site, West Falmouth Harbor, to
a series of tide heights that we measured at Great
Sippewissett salt marsh. We obtained tide height
predictions for West Falmouth Harbor, 1 km north of
Great Sippewissett salt marsh, from XTide version
2.8.2 (Flater 1998). To obtain measurements of actual
tide heights in Great Sippewissett salt marsh, we used
Onset HOBO Water Level Loggers, which use ambient
pressure, adjusted to account for barometric pressure,
to determine water height. The water level loggers
were deployed at several sites in Great Sippewissett
salt marsh, including one located in the high marsh
to record barometric pressure and one at a low point
at the mouth of an experimental plot, throughout
September 2007.
RESULTS
Changes in nutrient concentration across entire tidal
cycles
Concentrations of nitrate and ammonium changed
substantially (Fig. 2b,c) as tide water flooded onto and
ebbed from the experimental plots (Fig. 2a). There was
variation among plots subject to the different fertiliza-
tion treatments, but in general, NO3–concentrations
were initially relatively low, and increased somewhat
in flooding seawater. This pattern suggests that the
shallow layer of leading flood water initially picked up
nutrients as it moved onto the marsh plots. As larger
volumes of water flooded the plots, dilution occurred,
which led to near-minimum concentrations at the high-
est tide heights (Fig. 2b). As water ebbed, there was a
substantial increase in NO3–concentration in water
leaving the more highly fertilized plots, peaking at
mid-ebb and subsequently decreasing as the ebb tide
proceeded (Fig. 2b).
6

Brin et al.: Salt marsh nitrogen loading and export
The concentration of NH4+in flooding water became
lower with the incoming tide, reached a minimum or
near-minimum at high tide and then increased as the
ebb tide proceeded (Fig. 2c). The high NH4+concen-
trations in the residual tidal water before and after the
high tide likely derived from mixing with NH4+-rich
porewater.
Several notable features were apparent in the time
courses of both NO3–and NH4+concentrations across
the tidal cycle. First, concentrations of NO3–and NH4+
in flood water were similar among fertilization treat-
ments (Fig. 3, left) (Kruskal-Wallis test, NO3–: p =0.380,
NH4+: p =0.468), showing that flooding seawater had
consistent concentrations of DIN across the salt marsh.
Second, for C plots and plots receiving low N inputs,
concentrations of NO3–and NH4+in flood and ebb
water were not significantly different, except that NO3–
concentrations in flood water were higher than in ebb
water for C plots (Kruskal-Wallis test, p =0.021). This
suggests that there may have been a small net tidal
import of NO3–to C plots. In contrast, concentrations of
both NO3–(HF plots: p =0.018, XF plots: p =0.009) and
NH4+(HF plots: p =0.002, XF plots: p =0.011) were
higher in ebb water than in flood water for HF and XF
plots. These results therefore suggest that unamended
salt marsh plots intercepted NO3–to some degree, and
that highly amended plots (HF and XF) exported both
NO3–and NH4+in some proportion to input rates.
Intensive (time-course) ebb tide sampling
The results of the whole tidal cycle sampling showed
that there were substantial tidal water concentration
differences among treatments that could lead to differ-
ences in nutrient export, and that these differences
were manifest mainly during ebb tides, in water about
to leave the plots (Fig. 3). We now move on to the
second sampling phase of our research, in which we
used data from ebb tides sampled more intensively to
quantify NO3–and NH4+export, from which we could
estimate within-plot interception. It seemed reason-
able to concentrate on a more intensive sampling dur-
ing ebb tides because flood water DIN concentrations,
although variable, had a consistent range and variation
among the different plots (Fig. 3, left).
Export of NO3–and NH4+
There was substantial variation in the concentrations
of both NO3–and NH4+during the intensively sampled
ebb tides. Analysis across all fertilization regimes
showed that this variation was not due to season
(Brin 2008). The large scatter in the concentrations over-
whelmed whatever seasonal trends might be present.
There were evident differences in NO3–and NH4+
concentrations in ebb tidal water associated with the
different treatments (Brin 2008). Because the con-
centrations were highly skewed, we examined median
and maximum concentrations, both of which increased
with higher N inputs (Kruskal-Wallis test, p < 0.001).
To allow comparison of DIN exports between treat-
ments, we calculated NO3–and NH4+export by multi-
plying concentration data by associated water volumes
derived from the detailed topographic model. This pro-
cedure produced export values for each sample point
which were then summed to determine export of each
constituent over the complete ebb tidal period. The
resulting export per tide was used to evaluate the rela-
tive relationships of fertilization rate, time since appli-
7
0.01
0.1
1
10
100
–20
Concentration (µM)
C
LF
HF
XF
U
NO
3
–
0
20
40
60
80
Height (cm)
a
b
c
0.01
0.1
1
10
100
–4 –3 –2 –1 0 1 2 3 4
Hour from hi
g
h tide
Concentration (µM)
NH
4
+
Fig. 2. Mean (±SE) (a) tide height, (b) NO3–concentration and
(c) NH4+concentration in water on plots subject to each fer-
tilization treatment for full tidal cycle sampling. Data are
averaged over both sampling dates

Mar Ecol Prog Ser 400: 3–17, 2010
cation of fertilizer and tide height with tidal
export of NO3–and NH4+.
Fertilization rate. During the period of
application of fertilizer, exports of NO3–and
NH4+were higher in plots treated with higher
dosages (Fig. 4, left) (linear regression using
log-transformed values, NO3–: p < 0.001,
NH4+: p =0.008, DIN: p < 0.001; without U
plots, NO3–: p < 0.001, NH4+: p =0.01, DIN:
p < 0.001). Export of NO3–and NH4+was of
similar magnitude in plots receiving high N
additions (Fig. 4, left). Export of DIN was
small but measurable from plots receiving
low or no experimental input (Fig. 4, left), but
because export measurements did not in-
clude tidal import, this minimal tidal DIN
export may be from the small amount of N
carried onto the plots by the flooding tide
(Fig. 3, left).
Time since fertilization. During the fertil-
ization period, sampled tides ranged from
1 to 19 d after the most recent fertilization.
Nonetheless, there was no significant rela-
tionship between export of NO3–or NH4+and
number of days since the most recent fertil-
ization (Brin 2008), implying that other fac-
tors had a greater effect on variation of DIN
export.
Export of NO3–and NH4+diminished
markedly soon after experimental fertilization
ended (Fig. 4, right). This change suggested
that it was unlikely that there was sustained
year-round tidal export of experimentally
added N from the plots, which allowed us to
constrain our estimates of treatment-related N
export to the fertilization period.
Tide height effects. Different volumes of water
flooded each plot at any given tidal height, owing to
differences in elevation of plots relative to MSL. To ac-
count for these differences, we defined ‘effective tide
height’ as the difference between high tide height rel-
ative to MSL and geometric mean elevation of the plot
relative to MSL (Brin 2008 and associated Appendix 1).
In all fertilization treatments, NO3–and NH4+
export from plots increased in relation to the effec-
tive tide height (Fig. 5). This greater export of N by
greater tidal volumes may be a result of increased
time of tidal submergence of the plots. Furthermore,
the increase in export with effective tide height was
greatest for XF plots, as demonstrated by the slope of
this relationship (Fig. 5). These results suggest that
salt marshes subject to larger external inputs of N
will export proportionally larger amounts of DIN to
downstream waters, although the fraction exported
will be small.
Export of DON
DON analysis of samples from the 3 analyzed
sampling dates showed that DON export was highly
variable, and constituted approximately 80 to 90%
of gross total dissolved N (TDN) export from C plots and
approximately 10 to 75% of gross TDN from XF plots.
This is a measurement of gross rather than net export,
because it included any DON present in the water of
the incoming tide, which may or may not have been
variable throughout the marsh and among sampling
dates. A comparison between fertilized and control
plots indicated that DON export from the N-amended
plots was equal to or less than export from the control
plots (Table 1). Although there were no significant dif-
ferences between treatments within this data set, there
was a trend towards decreased gross DON export from
XF plots. This suggests that an estimate of dissolved N
loss based upon DIN measurements is more likely to
overestimate than underestimate the amount of treat-
8
0.1
1
10
10
–2 0 2 4 6 8 –2 0 2 4 6 8
–2 0 2 4 6 8 –2 0 2 4 6 8
Fertilization rate (
g
N m
–2
wk
–1
)
0.1
1
10
10
NH
4
+
NH
4
+
0.1
1
10
100
0.1
1
10
100
NO
3
–
NO
3
–
EbbFlood
Concentration (µM)
Fig. 3. NO3–and NH4+concentrations in flood and ebb water during
full tidal cycles. Dots show medians, vertical lines show first and third
quartiles, and asterisks show minimum and maximum concentrations,
by treatment. Treatments are C, LF, U, HF and XF, which received 0.0,
0.9, 2.6, 2.6 and 7.8 g N m–2 wk–1, respectively. Median ebb water
concentrations were higher on HF plots than on U plots. See ‘Results’
for statistics

Brin et al.: Salt marsh nitrogen loading and export
ment N exported, and can thus be used to
constrain the upper limit of N export and the
lower limit of N retention.
Estimates of N export over the experimen-
tal period relative to N inputs
To estimate the magnitude of DIN export
from the salt marsh plots and compare this
export to the total amount of N added, we
extrapolated export data from the sampled
tides to the entire series of tides over the fer-
tilization period (April– October). To allow
extrapolation of export data over the fertil-
ization period, we first tested the assumption
that tide heights on our sampling days were
representative of the tides that occurred dur-
ing the entire fertilization period. No signifi-
cant difference between frequency distribu-
tions of these 2 sets of tide heights was
evident (χ2goodness-of-fit test, p =0.47).
Second, we tested whether high tide heights
measured in Great Sippewissett salt marsh
could be represented by available high tide
height predictions for West Falmouth Har-
bor, and found that the relationship between
these 2 sets of tide height data was highly
significant (F1, 62 = 1037.5, p < 0.001, r2= 0.94)
(Fig. 6). This result was expected, given the
proximity of the tide stations and the
absence of tidal restrictions in both systems.
Based upon the results of the tidal analysis,
we used 2 methods to calculate the export of
NO3–and NH4+from the experimental plots
over the fertilization period. First, we used
the linear regressions of NO3–and NH4+ex-
port against effective tide height (Fig. 5), and
the history of tide heights in Great Sippewis-
sett salt marsh as estimated from tide heights
in West Falmouth Harbor (Fig. 6), to recon-
struct the likely export of NO3–and NH4+for
each tide and plot, and we then averaged ex-
port between plots for each treatment (Fig. 7).
We then summed across all tides to calculate
cumulative exports of NO3–and NH4+from
each plot over the fertilization period, and
averaged export between plots for each treat-
ment (Fig. 8, Table 2).
With this linear regression method, the
tight positive relationship between tide height
and DIN export inherent in the regression
equations is visible as the coupling of NO3–
and NH4+export to tide phases, with higher
exports during high tides (Fig. 7). One of the
9
0.1
1
10
0.1
1
10
0.1
1
10
100
0.1
1
10
0.1
1
10
0.1
1
10
100
100 100
100 100
After
During
NO
3
–
NO
3
–
NH
4
+
NH
4
+
DINDIN
Mean export (g N tide
–1
)
–2 0 2 4 6 8 –2 0 2 4 6 8
–2 0 2 4 6 8 –2 0 2 4 6 8
–2 0 2 4 6 8 –2 0 2 4 6 8
Fertilization rate (
g
N m
–2
wk
–1
)
Fig. 4. Mean (±SE) measured export per tide of NO3–, NH4+and dis-
solved inorganic nitrogen (DIN) from plots receiving each treatment,
during and after the fertilization period. A log scale is used due to the
log-normal distribution of the data. Mean exports were higher from HF
plots than from U plots, except for NH4+export after the fertilization
period. See Table 1 for treatment abbreviations and levels
Table 1. Gross export of dissolved organic nitrogen (DON) for several
representative sample dates during and after the fertilization period.
Mean (±SE) exports are over 2 plots for each treatment. C: control;
LF: low fertilization; U: urea; HF: high fertilization; XF: extra-high
fertilization
Treatment DON (g N tide–1)
(g N m–2 wk–1) August 13 August 31 November 27 Mean
C (0) 29.9 ± 8.5 38.4 ± 11.3 16.1 ± 11.3 28.2 ± 6.2
LF (0.9) 33.9 ± 7.0 48.0 ± 4.1 28.2 ± 2.5 36.7 ± 4.3
U (2.6) 24.5 ± 1.4 31.7 ± 0.8 12.4 ± 6.7 22.9 ± 4.0
HF (2.6) 18.3 ± 7.6 32.0 ± 8.8 12.0 ± 3.8 20.8 ± 4.9
XF (7.8) 15.0 ± 0.7 20.0 ± 9.7 3.3 ± 0.4 12.8 ± 4.0

Mar Ecol Prog Ser 400: 3–17, 201010
0.01
0.1
1
10
100
1000
C
LF
U
HF
XF
XF
HF
LF C
U
NO
3
–
0.01
0.1
1
10
100
1000
–0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0
–0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0
XF
C
HF
LF
U
NH
Effective tide hei
g
ht (m)
4
+
XF F = 33.58*** r
2
= 0.68
HF F = 15.07** r
2
= 0.52
U F = 14.06** r
2
= 0.47
LF F = 25.07*** r
2
= 0.61
C F = 5.14* r
2
= 0.24
XF F = 22.72*** r
2
= 0.59
HF F = 3.64 r
2
= 0.21
U F = 7.95* r
2
= 0.33
LF F = 8.59** r
2
= 0.35
C F = 18.97*** r
2
= 0.54
Export from plots (g N tide
–1
)Export from plots (g N tide
–1
)
Fig. 5. Linear regressions of nitrogen export as NO3–and NH4+against effective tide height. F-values are presented for the model
utility test of the slope of each relationship. See Table 1 for treatment abbreviations and levels (see ‘Result: Tide height effects’
for definition). *p ≤0.05; **p ≤0.01; ***p ≤0.001

Brin et al.: Salt marsh nitrogen loading and export
most prominent effects of N addition was the effect of
tide range on export at the highest level of N addition
(XF). The XF plots showed lower export rates at low
tide heights and higher export rates at high tide
heights (Fig. 7) than the other plots, as suggested by
the regression slopes in Fig. 5. However, the exports
from XF plots during higher tides so exceeded exports
from C or LF plots that greater amounts of DIN were
exported seasonally; total export of NO3–and NH4+
during the fertilization period increased with N input
rate (linear regression using log-transformed values for
NO3–, NO3–: p < 0.00, NH4+: p =0.01; without U plots,
NO3–: p =0.001, NH4+: p =0.02) (Fig. 8), with signifi-
cance driven by the XF plots.
As a check on the N exports calculated from the lin-
ear regression approach, we used a second (multiplica-
tive) method to estimate NO3–and NH4+export over
the period of fertilization. The lowest high tide height
for all sampled ebb tides was approximately 25 cm, at
which point there was almost no flooding of the marsh
surface. Based on the tide record for Great Sippewis-
sett salt marsh, there were 291 high tides ≥25 cm over
the fertilization period. Multiplication of the mean
export of DIN per tide (Fig. 4) by 291 tides resulted in
DIN exports that were comparable to estimates using
effective tide height regressions (Table 2).
It is important to differentiate between mean DIN ex-
port, which was derived from direct measurements from
sampled tides, and extrapolated export, which repre-
sents estimates for the entire fertilization period. Both
mean DIN export and extrapolated export are estimates
of gross DIN export, i.e. they do not consider imports.
These seasonal export predictions, as well as the
measured export on individual tides, depend on accu-
rate estimates of the volume of ebb tidal water leaving
each plot. Calibrations of the tidal volume model were
based upon measured and predicted tides in West Fal-
mouth Harbor for the tides when direct measurements
were made. Modeled ebb tidal volumes were higher
than values from measurements, with an average ratio
of 1.6 (modeled to measured, SE = 0.2, n = 7, Fig. 9).
The 2 sets of volumetric predictions aligned when pre-
dicted Great Sippewissett tide heights were adjusted
by approximately 6 cm (mean factor difference = 1.0,
SE = 0.1), suggesting that the discrepancy may be in
the conversion from tide height predictions in West
Falmouth Harbor to those in Great Sippewissett salt
marsh. Therefore, to account for variability in tide
height conversions such as that due to wind and other
meteorological factors, the calibrated model was also
used to calculate the Great Sippewissett salt marsh ebb
tidal flows for each plot over the growing season,
resulting in a lower bound of N export estimates (data
not shown), which were used for calculations of per-
cent N interception (see below). Because the model
was constructed from site-specific measurements, and
only one calibration tide could be performed per plot
for the flow measurements, the unadjusted modeled
volumes were used to provide a conservative measure
of N export from the plots and the proportion of added
N exported. Since the measured volume estimates
were less than the modeled volume estimates, N exp-
ort based upon the adjusted model was lower, and
hence N interception higher. Actual rates are probably
somewhere in between these 2 estimates of N inter-
ception.
Difference in NO3–and NH4+responses to increased
fertilization
Estimates of export over the entire fertilization
period showed notable differences in the response of
exports of NO3–and NH4+to increased fertilization
regimes (Fig. 8). NH4+exports increased linearly with
fertilization rate, which contrasted with the acceler-
ated response of NO3–export as N fertilization rate
increased (Fig. 8). As noted above, the significance of
these regressions was driven by XF plots. This differ-
ential regime resulted in greater export of NH4+than
NO3–when N inputs were low, but comparable NO3–
and NH4+exports from plots receiving the highest N
inputs (Table 3).
Interception capacity of salt marsh plots
Using the export estimates, it is possible to estimate
the fraction of N exported relative to the amount of N
added over the range of N loading rates (Table 2); this
11
0
0.2
0.4
0.6
0.8
1.0
0.8 1 1.2 1.4 1.6 1.8
y = 1.074x – 0.84898
F = 1037.5***
Measured high tide (m)
Predicted high tide (m)
Fig. 6. Linear regression of high tide height in West Falmouth
Harbor, as predicted from the XTide program (Flater 1998),
against high tide height in Great Sippewissett salt marsh, as
measured by HOBO Water Level Loggers. ***p < 0.001

Mar Ecol Prog Ser 400: 3–17, 2010
export fraction and the interception fraction sum to 1.
From each estimate of gross DIN export (as NO3–and
NH4+), we first subtracted the mean total DIN export
from C plots, assuming that export from C plots was a
measure of export from natural, unfertilized marshes.
This calculation provided an estimate of DIN export
attributable to the experimentally added N for each
plot, which could be compared to experimental N input.
Even though we measured significantly higher ex-
ports from plots subject to higher N additions, the more
ecologically significant finding was that DIN export
from N-amended plots constituted only a small fraction
of N addition. This low level of N export, even under
very high external loading rates, was evident in esti-
mates obtained from both methods of calculation.
Using estimates of N export from the linear regression
method, and the original and unadjusted flux models,
percent interception of added N ranged among treat-
ments from 93.9 to 105.9%; with the multiplicative
method, percent interception ranged among treatments
12
10
1
0.1
10
1
0.1
10
1
0.1
10
1
0.1
100
10
1
0.1
100
10
1
0.1
100
10
1
0.1
0.01
100
10
1
0.1
0.01
10
1
0.1
0.01
10
1
0.1
0.01
A M J J A S O A M J J A S O
N export (g)
NO3–NH4+
CC
LF LF
UU
HF HF
XF
Month
XF
Fig. 7. Export of NO3–and NH4+over the period of fertilization, predicted from fertilization rate and high tide height. Export
values are means over the 2 replicate plots per treatment. See Table 1 for treatment abbreviations and levels

Brin et al.: Salt marsh nitrogen loading and export
from 93.3 to 104.8% (Table 4). In both cases, the frac-
tion of the added N that was exported increased with
the level of N amendment for the more highly fertilized
plots. LF plots exported less DIN than the C plots, such
that the calculated fraction of the addition that was
exported was negative; however, the differences and
values involved were small, and treatment N export
was low for all plots.
DISCUSSION
Net import and export of N from unfertilized marsh
The similarity in flood and ebb DIN concentrations
for control plots and for plots receiving lower levels of
N amendment is consistent with studies of natural
marsh indicating negligible losses of DIN from the
vegetated marsh surface during June and July (Wolaver
et al. 1983, Whiting et al. 1989). This suggests that if
there are any natural inputs of N to the salt marsh, such
as through atmospheric deposition, elevated tidal N
concentrations or internal N sources to tidal water,
such as leaching from plants, it is entering the N cycle
within the plots and is largely not exported in outgoing
tidal waters at this time of year.
Difference in NO3–and NH4+responses to increased
fertilization
Several inferences can be made from the differences
among fertilization treatments in the ratios of NO3–to
NH4+tidally exported over the experimental period.
First, there was greater export of NH4+than of NO3–,
except at the highest N addition, where DIN export
was equally divided between these 2 forms. This may
suggest that in marsh plots where N inputs are rela-
tively low, as is the case in a natural system, NH4+
13
Table 2. N addition and export over the April to October period of fertilization. Fertilizer was applied 10 times over the growing
season, approximately biweekly. N added is the total N applied over the fertilization period to each plot receiving the specified
treatment. Mean (± SE) exports (NO3–, NH4+and total dissolved inorganic N [DIN]) were estimated using linear regression equa-
tions describing the relationship between NO3–and NH4+export and effective tide height for each treatment, as well as by multi-
plying average tidal export of DIN by the number of tides likely to remove DIN during the fertilization period. See Table 1 for
treatment abbreviations
Treatment N NO3–export (kg N plot–1)NH
4+export (kg N plot–1) DIN export (kg N plot–1)
(g N m–2 added Linear Mean Linear Mean Linear Mean
wk–1) (kg plot–1) regression tidal export regression tidal export regression tidal export
C (0) 0.0 0.18 ± 0.03 0.19 ± 0.37 0.58 ± 0.12 0.63 ± 0.36 0.76 ± 0.14 0.82 ± 0.51
LF (0.9) 5.5 0.11 ± 0.02 0.16 ± 0.37 0.32 ± 0.04 0.40 ± 0.37 0.43 ± 0.06 0.56 ± 0.53
U (2.6) 16.5 0.23 ± 0.03 0.27 ± 0.36 0.76 ± 0.11 0.88 ± 0.37 0.99 ± 0.14 1.14 ± 0.52
HF (2.6) 16.4 0.46 ± 0.22 0.59 ± 0.37 1.08 ± 0.40 1.33 ± 0.39 1.54 ± 0.62 1.93 ± 0.54
XF (7.8) 49.2 2.00 ± 0.77 1.45 ± 0.44 1.75 ± 0.60 1.49 ± 0.43 3.76 ± 1.38 2.94 ± 0.62
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0 2 4 6 8
NO
3–
F = 36.130***
NH
4+
F = 12.360**
Fertilization rate (g m
–2
wk
–1
)
N export (kg season
–1
)
Fig. 8. Total estimated export (±SE) of N as NO3–and NH4+
versus total N added over the fertilization period. Export val-
ues were estimated using linear regression equations relating
export and high tide height, by fertilization rate. Mean
exports were higher from HF plots than from U plots. See
Table 1 for treatment abbreviations and levels
0
20
40
60
80
Modeled water flux (m
3
)
Flow-measured water flux (m
3
)
0 20 40 60 80
Fig. 9. Ebbing water flux from experimental plots as modeled
from elevation data and as calculated from directly measured
water flow speeds, October 2008. Values are for the 7 plots for
which accurate flow data were obtained

Mar Ecol Prog Ser 400: 3–17, 2010
regeneration and possibly leaching from marsh grasses
are the major sources; in addition, there may be either
effective denitrification or minimal nitrification, such
that NO3–is not available to be exported. The lesser
importance of NH4+export from highly fertilized plots
might also be related to the increased presence of high
marsh plant species (Valiela et al. 1985, Hersh 1996, L.
Fox et al. unpubl. data); Jordan et al. (1983) found that
low marsh sites exported NH4+, whereas high marsh
sites actually imported a small amount of NH4+.
Second, because of this increase in the relative
importance of NO3–export at XF plots, there was an
accelerated response of NO3–export with increasing
fertilization rate, which contrasted with the near-linear
response of NH4+export (Fig. 8, Tables 2 & 3). This
implies that some biogeochemical mechanism results
in the export of proportionally more NO3–, but not
NH4+, when large amounts of N are added. One possi-
ble mechanism involves air entry to salt marsh sedi-
ments, which occurs when evapotranspiration removes
water from the sediment, and also by drainage on
creek banks, leading to greater oxidation potentials
(Howes et al. 1986). It is possible that increased growth
at high fertilization levels causes greater oxidation of
the sediments, resulting in accelerated rates of nitrifi-
cation (Howes et al. 1986, Howes & Teal 1994). Al-
though denitrification rates have been shown to be
substantially higher in high fertilization than control
plots (Hamersley & Howes 2005), denitrification rates
may only be able to rise to a point (Valiela et al. 2000).
In this case, NO3–from any source, including nitrifica-
tion, could be exported by tidal waters. This contrasts
with whole marsh studies, where it is thought that
NO3–is intercepted and NH4+is exported (Nixon 1980,
Valiela et al. 2000, Valiela & Cole 2002). The distinc-
tion is thought to stem from consistent differences in N
cycling within the marsh, where emergent marsh is
dominated by internal cycling and creek bottoms
are dominated by external inputs and denitrification
(Howes et al. 1996).
N export and interception over the experimental
period
Tidal export of NO3–and NH4+increased with N ad-
dition for all but the LF plots, suggesting that N export
from salt marshes depends to some degree on the
amount of N received. The significance of this trend
was driven by the XF plots, which have substantially
higher denitrification, biomass and plant percent N.
Notably, these plots also are higher in elevation, re-
ducing the amount of time the marsh surface is under-
water during each tide. Had this not been the case, we
may have seen slightly higher export, in which case
the difference between XF and other plots would have
been greater. However, longer submersion time would
also affect denitrification rates and plant growth.
In spite of significant export of N from plots receiving
higher inputs, overall percent interception of added N
was high for all experimental plots (Table 4), empha-
sizing the importance of salt marshes in reducing N
inputs to coastal waters. This result is complemented
by observations that tidal export of N from the plots
decreased sharply after experimental addition of N
ceased (Fig. 4, right). The degree and rapidity of N
interception is unsurprising in light of recent studies in
14
Table 3. Ratio of NO3–export to NH4+export over the April to
October period of fertilization. Ratios were calculated using
estimates of export derived from linear regression equations
describing the relationship between NO3–and NH4+export
and effective tide height for each treatment, using the unad-
justed tidal volume model, as well as by multiplying average
tidal export of dissolved inorganic nitrogen (DIN) by the
number of tides likely to remove DIN during the fertilization
periods. See Table 1 for treatment abbreviations
Treatment NO3–:NH4+export
(g N m–2 wk–1) Linear Mean tidal
regression export
C (0) 0.31 0.30
LF (0.9) 0.36 0.40
U (2.6) 0.30 0.31
HF (2.6) 0.43 0.44
XF (7.8) 1.15 0.97
Table 4. Mean percent interception of dissolved inorganic ni-
trogen (DIN) over the April to October period of fertilization.
Percent interception is the ratio of the difference between
DIN loss from the specified treatment plot and the control to
the total N added to the treatment plot, i.e. the amount of
added N that is not exported by tidewater. Percent intercep-
tion is calculated using estimates of export derived from lin-
ear regression equations describing the relationship between
NO3–and NH4+export and effective tide height for each treat-
ment, as well as from multiplying average tidal export of DIN
by the number of tides likely to remove DIN during the fertil-
ization period. For the linear regression method, estimates
from the original and adjusted volume models are presented
as upper and lower bounds, with the estimate from the ad-
justed model in italics (SE is presented for treatments with ad-
justments for both replicate plots). Also shown is percent in-
terception as measured by Valiela et al. (1973). nd: no data.
See Table 1 for treatment abbreviations
Treatment Present study Valiela
(g N m–2 Linear Mean tidal et al.
wk–1) regression export 1973
LF (0.9) 102.3– 105.9 (2.8) 104.8 (13.8) 77.2– 93.6
U (2.6) 97.7 (1.1) – 98.6 (1.2) 98.1 (4.6) nd
HF (2.6) 95.2 (3.9)–96.0 (2.0) 93.3 (4.7) 59.5 – 79.7
XF (7.8) 93.9 (2.8)–99.2 95.7 (1.7) nd

Brin et al.: Salt marsh nitrogen loading and export
Great Sippewissett. Hamersley & Howes (2005) in-
jected 15NH4+into Great Sippewissett salt marsh sedi-
ments underlying Spartina alterniflora and found mini-
mal, non-significant losses of 15N to overlying tidal water
by sediment–water exchange. Similarly, rapid losses of
15N through coupled nitrification–denitrification and
quick incorporation of DIN into salt marsh plant bio-
mass has been observed in natural stands of S. alterni-
flora (White & Howes 1994, Hamersley & Howes 2005).
These studies further found that > 40 % of the added N
was retained in the sediments after 7 yr, and the domi-
nant pathway of N loss was denitrification. The present
study corroborates the conclusion that, even on the
larger spatial scale of experimental plots, DIN export
by emergent marsh is extremely low.
N interception in experimental plots also occurs by
plant uptake, as evidenced by increases in biomass
(Valiela et al. 1975, Vince et al. 1981) and percent N in
the tissues of several species of marsh grasses (Vince et
al. 1981, Turner et al. 2009). Further visual evidence of
this can be found in the still-distinct boundary between
fertilized and unfertilized vegetation at the edges of
plots (Turner et al. 2009, I. Valiela pers. obs.). Recent
findings by Turner et al. (2009) indicate that N enrich-
ment does not generally lead to belowground organic
or inorganic accumulation, although the burial fate of
added N is being further assessed by E. Kinney et al.
(unpubl. data). Thus N interception in experimental
plots is likely due to high denitrification rates and
incorporation into plant tissue.
On the dates on which DON concentrations were
measured, gross DON exports were variable, and
there appeared to be no significant differences in DON
export associated with the level of N addition. Never-
theless, DON export tended to be lowest at the highest
levels of N addition, which contrasts with DIN export,
which was highest from XF plots. The mechanism for
this pattern of DON export is presently unclear, but
may relate to inundation time, higher rates of organic
matter turnover and/or less immersion of the taller
plants in the HF and XF plots versus controls. What-
ever the mechanism, the lower DON gross export from
XF plots suggests that adding treatment N does not
increase DON export to coastal systems.
Although the present study examined DIN and DON
export from experimental plots, an undetermined N
loss may have occurred via particulate organic N
(PON) export. PON flux varies substantially by tide
and the direction of the flux is often inconsistent (Va-
liela et al. 1978, Wolaver et al. 1983, Whiting et al.
1989). An N balance constructed for unfertilized marsh
plain vegetated with short Spartina alterniflora in
Great Sippewissett salt marsh indicated that particu-
late N losses were about 1.6 g N m–2 yr–1 (White &
Howes 1994). Valiela et al. (1978) estimated an annual
particulate export from Great Sippewissett salt marsh
of 1462 kg yr–1, which, divided over 483800 m2of total
marsh area (submerged and emergent) (Valiela & Teal
1979), gives an estimate of 3.0 g N m–2 yr–1. These esti-
mates are equivalent to 0.50 and 0.95 kg N yr–1 plot–1
for unfertilized plots. Thus PON fluxes may be impor-
tant to coastal waters receiving exports from pristine
marshes. N in particulate form is unavailable for up-
take by primary producers, unlike NO3–or NH4+,
which have been shown to increase growth of macro-
algae (Naldi & Wheeler 1999, Cohen & Fong 2004, Te-
ichberg et al. 2008), leading to detrimental effects
in eutrophic systems. Thus small DIN export from N-
amended plots suggests that salt marshes mediate
coastal eutrophication by intercepting added N before
it reaches coastal waters in a potentially harmful form.
Our calculations of percent interception of treatment
N may be conservative because of possible overesti-
mates of ebb tidal volume from the unadjusted model,
leading to a higher N export. The difference between
modeled and directly measured tidal volumes may be
due to several factors. First, the measured flux estimate
would be lowered if ebbing water bypassed the mea-
surement point by flooding across the edges of the plot
once the main creek is bank full. Although this effect
does not seem to occur substantially, it might vary
depending on the height of the tide. While this would
introduce error, the derived N export would still be an
overestimate, leading to a conservative estimate of
percent N interception. Second, as already suggested,
error could be introduced into the conversion of Great
Sippewissett tide heights from West Falmouth Harbor
predictions by variation in weather or wind. This
would lead to an overestimate of seasonal DIN export
with the volume model, tempered by use of adjusted
seasonal tide height predictions which led to a bounded
estimate of N interception (Table 4).
Such a low amount of N exported relative to the high
amount of N added to vegetated marsh plots is a notable
result. After more than 3 decades of N enrichment, these
plots have retained their ability to sequester and deni-
trify added N. The low rates of N export are lower than
earlier estimates at the same site (Valiela et al. 1973), but
methodological differences prevent more critical com-
parison (Table 4). Nonetheless, our measurements of N
interception suggest remarkable adaptation of the salt
marsh N cycle in response to large differences in N addi-
tion over a long period of time, leading to a near com-
plete retention or interception of added N.
The lack of dissolution and export of fertilizer N is
corroborated by the independent observation that after
more than 3 decades of experimental N addition, the
edges of the plots remain to within ~25 cm of the orig-
inal boundary (Turner et al. 2009). The permanence of
the bounds of the experimental plots attests to the lack
15

Mar Ecol Prog Ser 400: 3–17, 201016
of mobility of N on the marsh plain, most likely result-
ing from uptake by edaphic algae or rooted plants, or
denitrification. Thus, despite the fact that increased N
input into salt marsh ecosystems leads to increased N
export as NO3–and NH4+, salt marshes maintain the
ability to intercept and remove N even when exposed
to decades of N input.
Acknowledgements. We thank J. Fill, B. Barber and Y. Olsen
for participation in fieldwork and laboratory analysis, L. Fox
for providing data from her work and A. Giblin and B. Peter-
son for comments on earlier versions of this manuscript. We
thank the 200+ students and interns who helped to maintain
the experimental plots over the last several decades. Support
for this analysis and for site maintenance was provided by
many federal agencies, especially the National Science Foun-
dation (OCE-0453292, DEB-0516430) and, for the past 12 yr,
through the institutional support of the Coastal Systems
Program SMAST-UMD. We are much indebted to E. F. X.
Hughes and family for access and permission to use their salt
marsh property for our experimental plots.
LITERATURE CITED
Baden SP, Loo LO, Pihl L, Rosenberg R (1990) Effects of
eutrophication on benthic communities including fish:
Swedish west coast. Ambio 19:113– 122
Brin L (2008) Nitrogen retention and export in experimental
salt marsh plots exposed to chronic nutrient addition.
MA thesis, Boston University, Boston, MA
Cloern J (2001) Our evolving conceptual model of the coastal
eutrophication problem. Mar Ecol Prog Ser 210:223 –253
Cohen RA, Fong P (2004) Nitrogen uptake and assimilation in
Enteromorpha intestinalis (L.) Link (Chlorophyta): using
15N to determine preference during simultaneous pulses
of nitrate and ammonium. J Exp Mar Biol Ecol 309:67– 77
D’Elia CF, Steudler PA, Corwin N (1977) Determination of
total nitrogen in aqueous samples using persulfate diges-
tion. Limnol Oceanogr 22:760– 774
Dacey JWH, Howes BL (1984) Water uptake by roots controls
water table movement and sediment oxidation in short
Spartina marsh. Science 224:487– 489
Dame RF, Allen DM (1996) Between estuaries and the sea.
J Exp Mar Biol Ecol 200:169– 185
Flater D (1998) XTide: harmonic tide clock and tide predictor.
Available at http://www.flaterco.com/xtide/
Fox L (2007) The effect of nutrient enrichment and sea level
rise on salt marsh vegetation. MA thesis, Boston Univer-
sity, Boston, MA
Hamersley MR, Howes BL (2003) Contribution of denitrifica-
tion to nitrogen, carbon and oxygen cycling in tidal creek
sediments of a New England salt marsh. Mar Ecol Prog Ser
262:55– 69
Hamersley MR, Howes BL (2005) Coupled nitrification– deni-
trification measured in situ in a Spartina alterniflora marsh
with a 15NH4+tracer. Mar Ecol Prog Ser 299:123– 135
Hersh DA (1996) Abundance and distribution of intertidal and
subtidal macrophytes in Cape Cod: the role of nutrient
supply and other controls. PhD dissertation, Boston Uni-
versity, Boston, MA
Howes BL, Goehringer DD (1994) Porewater drainage and
dissolved organic carbon and nutrient losses through the
intertidal creekbanks of a New England salt marsh. Mar
Ecol Prog Ser 114:289–301
Howes BL, Teal JM (1994) Oxygen loss from Spartina alterni-
flora and its relation to salt marsh oxygen balance. Oeco-
logia 97:431– 438
Howes BL, Dacey JWH, Goehringer DD (1986) Factors con-
trolling the growth form of Spartina alterniflora: feedback
between above-ground production, sediment oxidation,
nitrogen and salinity. J Ecol 74:881–898
Howes BL, Weiskel PK, Goehringer DD, Teal JM (1996) Inter-
ception of freshwater and nitrogen transport from uplands
to coastal waters: the role of saltmarshes. In: Nordstrom
K, Roman C (eds) Estuarine shores: hydrological, geomor-
phological and ecological interactions. Wiley Interscience,
New York, p 287–310
Jordan TE, Correll DL, Whigham DF (1983) Nutrient flux in
the Rhode River: Tidal exchange of nutrients by brackish
marshes. Estuar Coast Shelf Sci 17:651– 667
McClelland JW, Valiela I (1998) Linking nitrogen in estuarine
producers to land-derived sources. Limnol Oceanogr 43:
577– 585
Naldi M, Wheeler PA (1999) Changes in nitrogen pools in
Ulva fenestrata (Chlorophyta) and Gracilaria pacifica
(Rhodophyta) under nitrate and ammonium enrichment.
J Phycol 35:70– 77
Nixon SW (1980) Between coastal marshes and coastal
waters: a review of twenty years of speculation and
research on the role of sale marshes in estuarine produc-
tivity and water chemistry. In: Hamilton P, MacDonald KB
(eds) Estuarine and wetland processes with emphasis on
modeling. Plenum Press, New York, p 437– 526
Rogers J, Harris J, Valiela I (1998) Interaction of nitrogen
supply, sea level rise, and elevation on species form and
composition of salt marsh plants. Biol Bull 195:235– 237
Seitzinger SP (1990) Denitrification in aquatic sediments. In:
Revsbech NP, Srensen J (eds) Denitrification in soil and
sediment. Plenum Press, New York, p 301– 322
Teichberg M, Fox SE, Aguila C, Olsen YS, Valiela I (2008)
Macroalgal responses to experimental nutrient enrich-
ment in shallow coastal waters: growth, internal nutrient
pools, and isotopic signatures. Mar Ecol Prog Ser 368:
117–126
Turner RE, Howes BL, Teal JM, Milan CS, Swenson EM,
Goehringer-Toner DD (2009) Salt marshes and eutrophi-
cation: an unsustainable outcome. Limnol Oceanogr 54:
1634– 1642
Valiela I (1983) Nitrogen in salt marsh ecosystems. In:
Carpenter EJ, Capone DG (eds) Nitrogen in the marine
environment. Academic Press, New York, p 649– 678
Valiela I (1995) Marine ecological processes, 2nd edn.
Springer-Verlag, New York
Valiela I, Bowen JL (2002) Nitrogen sources to watersheds
and estuaries: role of land cover mosaics and losses within
watersheds. Environ Pollut 118:239–248
Valiela I, Cole ML (2002) Comparative evidence that salt
marshes and mangroves may protect seagrass meadows
from land-derived nitrogen loads. Ecosystems 5:92 –102
Valiela I, Teal JM (1979) The nitrogen budget of a salt marsh
ecosystem. Nature 280:652–656
Valiela I, Teal JM, Sass W (1973) Nutrient retention in salt
marsh plots experimentally fertilized with sewage sludge.
Estuar Coast Mar Sci 1:261–269
Valiela I, Teal JM, Sass WJ (1975) Production and dynamics of
salt marsh vegetation and the effects of experimental
treatment with sewage sludge. Biomass, production and
species composition. J Appl Ecol 12:973– 981
Valiela I, Teal JM, Volkmann S, Shafer D, Carpenter EJ (1978)
Nutrient and particulate fluxes in a salt marsh ecosystem:
tidal exchanges and inputs by precipitation and ground-
➤
➤
➤➤
➤
➤
➤
➤
➤
➤
➤
➤
➤➤
➤
➤
➤
➤

Brin et al.: Salt marsh nitrogen loading and export 17
water. Limnol Oceanogr 23:798– 812
Valiela I, Teal JM, Cogswell C, Hartman J, Allen S, Van Etten
R, Goehringer D (1985) Some long-term consequences
of sewage contamination in salt marsh ecosystems. In:
Godfrey P.J, Kaynor ER, Pelczarski S, Benforado J (eds)
Ecological considerations in wetlands treatment of munic-
ipal wastewaters. Van Nostrand Reinhold, New York,
p 301– 316
Valiela I, Costa J, Foreman K, Teal JM, Howes B, Aubrey D
(1990) Transport of groundwater-borne nutrients from
watersheds and their effects on coastal waters. Biogeo-
chemistry 10:177– 197
Valiela I, Foreman K, LaMontagne M, Hersh D and others
(1992) Couplings of watersheds and coastal waters:
sources and consequences of nutrient enrichment in
Waquoit Bay, Massachusetts. Estuaries 15:443–457
Valiela I, McClelland J, Hauxwell J, Behr PJ, Hersh D, Fore-
man K (1997) Macroalgal blooms in shallow estuaries: con-
trols and ecophysiological and ecosystem consequences.
Limnol Oceanogr 42:1105– 1118
Valiela I, Cole ML, McClelland J, Hauxwell J, Cebrian J, Joye
S (2000) Role of salt marshes as part of coastal landscapes.
In: Weinstein MP, Kreeger DA (eds) Concepts and contro-
versies in tidal marsh ecology. Kluwer Academic Pub-
lishers, Dordrecht, p 23 –38
Vince SW, Valiela I, Teal JM (1981) An experimental study of
the structure of herbivorous insect communities in a salt
marsh. Ecology 62:1662– 1678
White DS, Howes BL (1994) Long-term 15N-nitrogen retention
in the vegetated sediments of a New England salt marsh.
Limnol Oceanogr 39:1878– 1892
Whiting GJ, McKellar HN Jr, Spurrier JD, Wolaver TG (1989)
Nitrogen exchange between a portion of vegetated salt
marsh and the adjoining creek. Limnol Oceanogr 34:
463– 473
Wolaver TG, Zieman JC, Wetzel R, Webb KL (1983) Tidal
exchange of nitrogen and phosphorus between a meso-
haline vegetated marsh and the surrounding estuary in
the lower Chesapeake Bay. Estuar Coast Shelf Sci 16:
321– 332
Editorial responsibility: Kenneth Heck Jr.,
Dauphin Island, Alabama, USA
Submitted: June 27, 2009; Accepted: December 17, 2009
Proofs received from author(s): January 25, 2010
➤
➤
➤