Biogeochemistry (****) **:*** ***
O RI G IN AL PA PER
EVect of nutrient loading on biogeochemical
and microbial processes in a New England salt marsh
Jane M. CaVrey Michael C. Murrell
Cathleen Wigand Richard McKinney
Received: 13 September 2005 / Accepted: 1 December 2006 / Published online: 9 February 2007
Springer Science+Business Media B.V. 2007
from 0 to 8 mmol m 2 day 1. Fertilization had no
Abstract Coastal marshes represent an impor-
apparent eVect on soil oxygen consumption or
tant transitional zone between uplands and estu-
denitriWcation measured in the summer in intact
aries. One important function of marshes is to
assimilate nutrient inputs from uplands, thus pro- cores due to high core-to-core variation. P fertil-
viding a buVer for anthropogenic nutrient loads. ization led to increased pore water dissolved inor-
We examined the eVects of nitrogen (N) and ganic phosphorus (DIP) concentrations and
phosphorus (P) fertilization on biogeochemical increased DIP release from soils. In contrast the
control and N-only treatments had signiWcant
and microbial processes during the summer grow-
ing season in a Spartina patens (Aiton (Muhl.)) DIP uptake across the soil-water interface. The
marsh in the Narragansett Bay National Estua- results suggest that in the summer fertilization has
no apparent eVect on denitriWcation rates, stimu-
rine Research Reserve on Prudence Island (RI).
Quadruplicate 1 m2 plots were fertilized with N lates bacterial productivity, enhances pore water
and P additions, N-only, P-only, or no additions. nutrient concentrations and alters some nutrient
N-only addition signiWcantly stimulated bacterial Xuxes across the marsh surface.
production and increased pore water NH+ and 4
NO concentrations. DenitriWcation rates ranged Keywords Bacterial production Fertilization
3
DenitriWcation Nitrogen Phosphorus Salt
marsh
J. M. CaVrey Center for Environmental Diagnostics
and Bioremediation, University of West Florida,
11000 University Parkway, Pensacola, FL, 32514, USA
Introduction
e-mail: abpmuc@r.postjobfree.com
Eutrophication resulting from increased anthro-
M. C. Murrell
OYce of Research and Development, National Health pogenic nutrient loading is a serious concern
and Environmental EVects Research Laboratory,
nationally and internationally (NRC 2000). Salt
Gulf Ecology Division, US EPA, 1 Sabine Island Dr.,
marshes are at the interface between the uplands
Gulf Breeze, FL, 32561, USA
and the estuary and are frequently the Wrst recipi-
ents of high nutrient runoV via overland or
C. Wigand R. McKinney
OYce of Research and Development, National Health
groundwater Xow (Page 1995; Tobias et al. 2001).
and Environmental EVects Research Laboratory,
In estuaries with high nutrient concentrations,
Atlantic Ecology Division, US EPA, 27 Tarzwell Dr,
tidal exchange between the estuary and marsh is
Narragansett, RI, 02882, USA
13
252 Biogeochemistry (2007) 82:251 264
another source of nutrient enrichment. Studies inundation usually occurs during spring tides. The
relatively infrequent Xooding leads to reduced
examining the response of salt marsh vegetation
to fertilization have had consistent results, from removal of wrack and greater incorporation of
the Wrst studies in Great Sippewisset marsh (Vali- organic matter into soils compared to the low
marsh zone where S. alterniXora is dominant. S.
ela et al. 1975) and Delaware marshes (Sullivan
alterniXora decays about twice as fast as S. patens
and Daiber 1974) through the present day (Boyer
(Buschbaum et al. 1991). This critical diVerence in
et al. 2001; Sundareshwar et al. 2003; Wigand
organic matter cycling between S. alterniXora and
et al. 2004a). Most studies have shown nitrogen
S. patens marshes may also manifest as a diVeren-
limitation of aboveground marsh production (Sul-
livan and Daiber 1974; Valiela et al. 1975; Buresh tial response to fertilization between these
et al. 1980; DeLaune and Patrick Jr 1980; Boyer marshes. We anticipate that S. patens marshes
et al. 2001; Wigand et al. 2004a), although phos- should be particularly sensitive to fertilization by
phorus limitation can occur secondarily with enhancing the accumulation of organic material
greater marsh biomass in N + P treatments com- and the microbially mediated processing of this
pared to N-only (van Wijnen and Bakker 1999; material. S. patens is endomycorrhizal, that is it
Sundareshwar et al. 2003). harbors fungal symbionts in the roots (Cooke
The eVects of fertilization on bacterial produc- et al. 1993), and these root-fungi associations
tion and nutrient cycling in marsh soils are not have been demonstrated to facilitate primarily
well studied and the results of the studies to date phosphorus uptake in many land plants (Allen
are not as consistent as the plant response. Fertil- 1991; Smith and Read 1997). In addition,
ization enhanced soil respiration in a South Caro- researchers have shown that endomycorrhizal S.
lina Spartina alterniXora (Lois.) marsh because patens also indirectly facilitate nitrogen Wxation in
microbial decomposers appeared to be nutrient the marsh soil possibly due to an increased
limited (Morris and Bradley 1999). Phosphorus release of organic exudates in the rhizosphere
additions stimulated bacterial biomass, bacterial (Burke 2001; Burke et al. 2002a). When phospho-
production, and pore water alkaline phosphatase rus is enriched in salt marsh plots, it may disrupt
activity suggesting phosphorus limitation of bac- the endomycorrhizal symbiosis of S. patens, and
terial growth (Sundareshwar et al. 2003). Denitri- cause a subsequent inhibition of the root fungal
Wcation was enhanced in plots fertilized with symbionts and diminish the release of organic
exudates and nitrogen Wxation (Wigand et al.
inorganic nutrients (Teal and Howes 2000). Fer-
tilization with organic nitrogen or sewage sludge 2004a).
enhanced denitriWcation in New England salt This study examined how biogeochemical and
marshes (Kaplan 1977; Hamersley and Howes microbial processes in a S. patens marsh were
aVected by fertilization with nitrogen and phos-
2005) and inhibited it in Georgia salt marshes
(Sherr and Payne 1981). DenitriWcation potential phorus. Measurements were made of bacterial
was inhibited by phosphorus additions to soils abundance and production (leucine incorpora-
tion), N2, O2, and inorganic nutrient (NH+, DIP,
from control plots (Sundareshwar et al. 2003). 4
Nitrogen Wxation was reduced in fertilized NO, NO ) Xuxes across the marsh surface and
3 2
marshes compared to controls (Van Raalte et al. pore water nutrients. An earlier study at this
1974; Sundareshwar et al. 2003). same site reported that fertilization with nitrogen,
Most fertilization studies to date have focused but not phosphorus led to an increase in above-
on S. alterniXora with little work in S. patens ground plant biomass and higher concentrations
marshes. S. patens is often dominant in the upper of N, P, and C in the above-ground tissues
salt marsh, although it also occurs in saline Xats, (Wigand et al. 2004a). In the present study, we
low dunes, and tidal shores from coastal, south- hypothesized that the increased plant production
west Newfoundland and eastern Quebec, south to in the nitrogen-fertilized plots would result in
Virginia and Florida, and west to eastern Texas. enhanced mineralization (the breakdown of
organic nitrogen to NH+ ) and other nitrogen
In New England estuaries, S. patens is the domi- 4
transformations, such as nitriWcation (the oxidation
nant plant in the high marsh zone, where tidal
13
Biogeochemistry (2007) 82:251 264 253
of NH+to NO ) and denitriWcation (the reduction removed with a wide bore pipet, further diluted
4 3
into 1 ml of phosphate buVered saline. A 100 l
of NO to dinitrogen gas). Further, we hypothe-
3
sized that P fertilization would result in an aliquot of this diluted sample was stained with
DAPI (1 g ml 1 Wnal concentration), Wltered
alteration of nutrient exchanges in the S. patens
rhizosphere, in part, because of inhibition of root onto black, 0.2 m pore size, 25 mm diameter,
polycarbonate membrane Wlter, mounted on
fungal symbionts and disruption of mycorrhizal-
mediated P uptake and N-Wxation leading to microscope slides between layers of immersion
oil. Bacteria were counted using a UV Wlter com-
increased competition between plants and bacte-
ria for nitrogen. bination (350 360 nm excitation, >400 nm emis-
sion) on a Nikon Microphot epiXuorescence
microscope at 1,250 (Porter and Feig 1980). A
minimum of 300 cells was counted distributed
Methods
over at least ten microscope Welds (typically 20
Study site 40). Duplicate preparations were periodically
counted and the coeYcient of variation among
This study was conducted in a S. patens marsh replicates averaged 28% (n = 6).
near the Nags Creek (41 37.546 N and Bacterial production was measured as the
incorporation of 4,5-[3H]-L-leucine into cold
71 19.223 W) in the Narragansett Bay National
TCA/ethanol insoluble pools using a modiWcation
Estuarine Research Reserve on Prudence Island,
RI. Sixteen 1 m2 plots were established with a of the methods developed for water samples
2 2 factorial design with nitrogen and phospho- (Smith and Azam 1992) and sediments (Van Duyl
and Kop 1994). Ten aliquots of ca. 0.1 cm3 were
rus as the treatments. Plots were randomly
located in vegetated patches at least 3 m apart. removed from the homogenized soil (0 1 cm
layer) using a 1 cm3 syringe, placed into 2 ml pre-
When the randomization process placed a plot in
a bare patch, the plot was moved to the nearest weighed micro-centrifuge tubes, and reweighed.
vegetated patch. At low tide, dissolved Ca(NO3)2 The L-leucine primary stock was diluted using
0.2 m Wltered seawater collected from Nags
and P2O5 were sprinkled on the soil surface twice
per month for the growing season, May August, Creek. This working stock (100 l) was added
(2 Ci at 20 Ci mmol 1) to each tube and vor-
but monthly thereafter. Fertilizer application
rates were 2 g N m 2 and 0.2 g P m 2 with a total texed to make a soil slurry. Additional aliquots of
addition of 80 g N m 2 and 8 g P m 2 for the working stock (20 50 l) were pipeted in tripli-
entire 28 months of the experiment (from May cate into micro-centrifuge tubes for total activity
2000 to September 2002). determination. Five replicates were immediately
Wxed for time zero; the other Wve were Wxed after
Bacterial abundance and production ca. 1 h incubation (21 C, dark). Several prelimi-
nary experiments established that isotope uptake
was linear for the Wrst 2 h (data not shown). Sam-
During July 2002, three 2.5 cm core samples of
ples were Wxed by addition of 1 ml of ice-cold
soil were taken from each plot and the top 1 cm
was pooled and homogenized. Soil bacterial 80% ethanol amended with a high concentration
of unlabeled leucine (1 g l 1), to dilute the 3H L-
abundance was determined by collecting dupli-
leucine speciWc activity and to facilitate extraction
cate samples (0.1 cm3) from the homogenized
soil, which were placed in pre-weighed micro-cen- of unincorporated label. The samples were pro-
trifuge tubes and re-weighed. The samples were cessed a total of three times by extracting in etha-
Wxed with 1 ml of 2% formaldehyde in seawater nol/leucine on ice for 2 h, centrifuging at 13,000 g
(pre-Wltered through a 0.2 m syringe Wlter) and for 5 min, and aspirating the supernatant. The
stored at 4 C until analysis within 3 weeks. Sam- tubes were placed into carrier vials with 1 ml of
ples were sonicated with a Branson micro-tip ScintiSafe gel cocktail and assayed for radioactiv-
soniWer (Model 250, setting #2) with 10 15 1-s ity with a Packard TR 2500 liquid scintillation
bursts. Aliquots (50 100 l) of the slurry were analyzer using transformed spectral index quench
13
254 Biogeochemistry (2007) 82:251 264
correction. This cocktail was designed to suspend and frozen until analysis. A Lachat Instruments
the particles into a gel-like matrix, but soil still QuikChem 8000 FIA+ automated ion analyzer
settled signiWcantly, causing the measured activity was used for NH+ NO plus NO (hereafter,
4 3 2
referred to as NO in the text), NO, and PO3
to decline over time. To account for this time- 3 2 4
dependent decline in activity, the samples were (referred to as DIP) analyses (Diamond 1997a, b;
vortexed and counted twice, recording the time Huberty and Diamond 1998; Schroeder 1997). A
positive Xux represents release from the soil to
interval between the two counts. The activity at
water, while a negative Xux represents uptake by
the time zero (the time of vortexing) was calcu-
lated by linear extrapolation. The correction was the soil from the water.
usually small (10% or less), but removed a known
and variable bias. Leucine incorporation rates Pore water sampling
were scaled to carbon units using conversion fac-
ProWles of pore water nutrients (ammonium,
tor 3.1 kg C mol 1 leucine incorporated, recom-
mended by Simon and Azam (1989). Bacterial nitrate, and phosphate) were sampled in situ in
abundances and production were normalized to each plot with PVC equilibrators (Hesslein 1976;
soil volume using a wet weight to volume density Bottomley and Bayley 1984; Wigand et al. 1997).
of 0.83 gww cm 3. Equilibrators (one per plot) were prepared in the
laboratory using deoxygenated, deionized water
Oxygen consumption, nutrient Xux and denitriW- in each sampling cavity and set in the Weld for
cation 5 days (June 5 10, 2002) to allow for equilibration
before retrieving. Dissolved constituents in the
soil diVused from the soil pore water across the
Duplicate intact cores (8 cm ID) were collected
for denitriWcation, oxygen consumption and nutri- cellulose membrane and into the sampling cavi-
ent Xux measurements from each experimental ties. Each equilibrator had six cavities, 2.5 cm in
diameter every 1.2 cm on the PVC. The Wrst cav-
plot at low tide between July 15 and 18, 2002.
Each day 8 of the 16 plots were sampled, so a ity was near surface with the last cavity at a depth
total of 32 cores were used for Xux measurements. of about 20 cm. The equilibrators were sampled
in the Weld within minutes upon removal using a
Aboveground vegetation was removed by cutting
syringe and Wltered (0.8 m, GF/F) into 20 ml
stems at the soil surface prior to coring. Cores
were returned to the laboratory each day and vials containing 20 l of 6 N HCL. Water samples
overlying water supplied from nearby Nags Creek were transported on ice to the lab and then frozen
was gently siphoned over the surface for a total until analysis.
volume of 1 l. An additional core with water
only was incubated to account for changes occur- Statistics
ring in the overlying water. The cores were incu-
bated for 8 h in the dark at ambient water Bacterial abundance, soil oxygen consumption,
denitriWcation, nutrient Xux, and pore water
temperature (20 C). Overlying water samples (ca.
45 ml) were collected at 1 2-h intervals for dis- nutrient values were natural logarithm trans-
solved gas (N2, Ar, and O2) analysis and pre- formed to homogenize variance prior to running a
two-way ANOVA to examine for the main eVects
served with 10 l of 50% saturated solution of
HgCl2. Samples were stored underwater and ana- of nitrogen or phosphorus and for an N P inter-
lyzed within 2 weeks on a membrane inlet mass action. The ANOVA for pore water nutrient data
spectrometer (MIMS), as described in Kana et al. also included depth as a covariable. Bacterial pro-
(1994). In addition to oxygen analysis by MIMS, duction values were square root transformed.
dissolved oxygen samples (7 ml) were Wxed with DiVerences between oxygen consumption mea-
Winkler reagents and analyzed within 24 h by surements using oxygen measured by Winkler
thiosulfate titration (Parsons et al. 1984). Nutrient method and MIMS were compared using a
samples collected at the same intervals as dis- paired t-test. Correlation analysis examined rela-
solved gases were Wltered through GF/F Wlters tionships among rate measurements, pore water
13
Biogeochemistry (2007) 82:251 264 255
Concentrations of both nutrients were highest in
Bacterial Production g C cm-3 d-1
9
N + P treatments, next highest in N-only, then P-
only and lowest in control treatments. Pore water
6
DIP and NO concentrations were signiWcantly
2
enhanced in the phosphorus treatments, but not
the nitrogen treatments. In addition, NO concen-
3 2
trations had a signiWcant N P interaction, while
pore water DIP concentrations did not. Pore water
DIP concentrations and Xuxes across the marsh
0
N
C P NP
surface were signiWcantly correlated (r = 0.81). NH+4
and NO concentrations signiWcantly covaried
8.E+09 3
Bacterial abundance # cm-3
with depth, but neither DIP nor NO concentra-
2
tions. NH+ concentrations increased with depth to
4
7.E+09
a maximum at 8 cm (4 cm for the N + P treatment)
(Fig. 2a). In contrast, NO concentrations were
3
highest in the surface layer (Fig. 2c), reaching
6.E+09
13 M in the N + P treatment, higher than the
concentrations found in the creek.
5.E+09
In general, nutrient exchange across the marsh
N
C P NP
surface was low, making it diYcult to resolve sta-
tistically signiWcant diVerences among the experi-
Fig. 1 (a) Bacterial production as measured by leucine
incorporation in g C cm 3 day 1. (b) Bacterial abundance
mental treatments. Of the nutrient exchanges
in cells cm 3, plot mean + SE (n = 4)
measured, NH+ Xux was highest, ranging from
4
0.9 to 1.7 mmol m 2 day 1 in individual plots.
concentrations and above and belowground bio-
Soil uptake of NH+ tended to occur in control
mass. Above and belowground biomass values 4
and N-only treatments, while NH+ release
are reported in Wigand et al. (2004a). The proba- 4
occurred in P treatments (Fig. 3a); however these
bility for signiWcance is reported at p