Me

LMa

b

a

ARR2A

KWCNSL

1

aadc2lpMitet

i

2

0h

Applied Soil Ecology 76 (2014) 87– 94

Contents lists available at ScienceDirect

Applied Soil Ecology

journa l h om epage: www.elsev ier .com/ locate /apsoi l

ultistate assessment of wetland restoration on CO2 and N2Omissions and soil bacterial communities

aurel A. Klubera,∗, Jarrod O. Millera, Thomas F. Duceya, Patrick G. Hunta,egan Langb, Kyoung S. Roa

USDA-ARS Coastal Plains Soil, Water, and Plant Research Center, Florence, SC United StatesUSDA-FS Northern Research Station, Beltsville, MD United States

r t i c l e i n f o

rticle history:eceived 13 March 2013eceived in revised form4 December 2013ccepted 31 December 2013

eywords:etland restoration

O2

2Ooil bacteriaand use

a b s t r a c t

Over the last 200 years, wetlands have been converted to other land uses leading to the loss of approx-imately 53% of wetlands in the continental United States. In the late 1980’s, policies were instated tomitigate further wetland loss through wetland creation and restoration. Restored wetlands provideimportant ecosystem services, such as filtration of nutrients and wildlife habitat. However, these benefitscould be offset by increased greenhouse gas production. We assessed the impact of wetland conversion toagriculture and restoration on CO2 and N2O emissions and microbial communities in three land use types:wetlands with native vegetation (natural); wetlands converted to agricultural management (converted);and restored wetlands (restored). Soil properties varied among land use types. Most notably, soils fromrestored and converted sites had the lowest C and N, and higher pH. Multivariate analysis of soil prop-erties showed the pocosin wetlands in North Carolina separating from all other locations, regardless ofland use. Soil bacterial communities showed a similar trend with communities from North Carolina soils

separating from the others with no significant effect of land use or season. Furthermore, land use did nothave a significant effect on CO2 or N2O emissions, although there was significant temporal variation inCO2 emissions. These findings indicate that while wetland conversion and restoration may alter somesoil properties, these alterations do not appear to be great enough to override the underlying geographicand edaphic influences on soil bacterial communities. Furthermore, wetland restoration did not lead to

t the

increased N2O emission a

. Introduction

Wetlands across the United States have been drained to createrable land suitable for agriculture and timber production (Brinsonnd Eckles, 2011; DeSteven and Lowrance, 2011). Urban and ruralevelopment have exacerbated this phenomenon, replacing agri-ulture as the main driver of wetland loss (DeSteven and Lowrance,011). In the late 1980’s, concern over the dramatic loss of wet-

ands (Dahl, 1990) and increased recognition of ecosystem servicesrovided by wetlands led to a “no net loss” policy (Bendor, 2009;itsch, 1992). Since this time, restoration of drained wetlands has

ncreased, and in 2003 the United States Department of Agricul-ure (USDA) Conservation Effects Assessment Project (CEAP) beganxamining the benefits of wetland conservation practices across

he U.S. (Brinson and Eckles, 2011).

Wetland restoration aims to restore ecosystem services includ-ng the filtration and retention of nutrients, water storage, and

∗ Corresponding author. Kluber USDA-ARS, 2611 West Lucas Street Florence, SC9501, United States. Tel.: +843 669 5203.

E-mail address: laurel.kluber@ars.usda.gov (L.A. Kluber).

929-1393/$ – see front matter. Published by Elsevier B.V.ttp://dx.doi.org/10.1016/j.apsoil.2013.12.014

dates sampled.Published by Elsevier B.V.

wildlife habitat (Brinson and Eckles, 2011; DeSteven and Lowrance,2011; Mitsch and Gosselink, 1993). However, the success ofhydraulic restoration can vary (Zedler, 2003). In part, successdepends on the intent behind the restoration; wetlands thatprovide the best habitat and biodiversity may not be as effectiveat nutrient retention (Hansson et al., 2005; Zedler, 2000). Whilesome studies have shown that restored wetlands can reduce nutri-ent runoff (Ardón et al., 2010; Jordan et al., 2003; Tanner et al.,2005), others have found that hydraulic restoration can result inthe release of nutrients (Steinman and Ogdahl, 2011). Furthermore,like natural wetlands, anoxic conditions in restored wetlands havethe potential to produce significant amounts of N2O when N avail-ability is high, such as in agriculturally impacted areas (Verhoevenet al., 2006).

As mentioned, a common concern with wetland restoration isthe potential for increased greenhouse gas emissions, (Brinson andEckles, 2011; Verhoeven et al., 2006). Increased CO2 emissions isgenerally not a concern with wetland restoration (Whiting and

Chanton, 2001) as the typically high water content and low O2reduces aerobic respiration and thus the production of CO2. How-ever, ephemeral wetlands that are not permanently saturated maynot have conditions leading to a reduction in CO2 emissions, as

8 d Soil

Mwicdc(ratiiMrwrrof2ehr

iNslfppll2tabr

AWMCgwthwrpwtelst

2

2

sr

8 L.A. Kluber et al. / Applie

orse et al. (2012a) found in examining North Carolina pocosinetlands. However, these conditions create the potential for

ncreased N2O emissions through two microbially mediated N-ycling processes. The first is nitrification, which results in the oxi-ation of ammonia (NH4

+) to nitrite (NO2−). The second is denitrifi-

ation, the process of converting nitrate (NO3−) to gaseous nitrogen

N2). When denitrification is not carried to completion, N2O canesult as a byproduct, particularly when soils are not completelynoxic (Myrold, 1999). Due to its high reactivity and long residencyime, a relatively small increase in N2O flux can result in signif-cant greenhouse effect (Schlesinger, 1991). While denitrificationn soil is dependent on many factors (Hunter and Faulkner, 2001;

orse et al., 2012b), Peralta and colleagues (2010) reported thatestored wetlands had lower denitrification activity than naturaletlands. This finding suggests that wetland restoration may not

esult in increased N2O emissions from denitrification. However, aecent study using 15N tracers demonstrated that in soils with highrganic matter and moisture content, nitrification is responsibleor a significant portion of N2O emissions (Morse and Bernhardt,013). These findings highlight the complexities in predicting N2Omissions and the need for direct in situ monitoring of green-ouse gases when evaluating the environmental impact of wetlandestoration.

Removal of nitrogen (N) is biologically driven; therefore, it ismportant to monitor and evaluate microbial communities when

removal is a restoration goal. Previous findings have shown thatoil bacterial communities in wetlands were correlated with pH andand use (Hartman et al., 2008). More recently, Peralta et al. (2010)ound that bacterial communities correlated with a number of soilroperties, including soil moisture, organic matter, and pH. Due inart to their agricultural legacies, restored wetlands usually have

ower carbon (C) content and higher pH compared to natural wet-ands (Bruland et al., 2003; Hogan et al., 2004; Morse and Bernhardt,013). Because both C and pH can influence microbial communi-ies (Lauber et al., 2009; Lauber et al., 2008), it is not surprising thatlthough hydraulic regimes can be restored almost immediately,iogeochemical functions of restored wetlands can take longer toe-establish (Bendor, 2009; Bruland et al., 2003).

The wetland component of the United States Department ofgriculture (USDA) Conservation Effects Assessment Project (CEAP-etland) is being carried out across the United States. In theid-Atlantic Region (the area between New England and North

arolina), the USDA CEAP-Wetland study is designed to investi-ate the effects and effectiveness of wetland conservation practices,ith a focus on the provision of ecosystem services. As part of

his interdisciplinary effort, our study was designed to test theypotheses that (1) the hydraulic restoration of converted wetlandsill alter soil chemical and physical properties, such that these

estored wetlands will begin to approximate their natural counter-arts and (2) the effects of hydraulic restoration on soil propertiesill likewise result in altered structure and function of soil bac-

erial communities. We achieved the objectives of this study byxamining soil properties in natural, converted, and restored wet-ands. Additionally, deep 16S sequencing was used to examine thetructure of bacterial communities and photoacoustic gas analysiso measure CO2 and N2O emissions.

. Materials and methods

.1. Site description and study design

As part of the Mid-Atlantic Region CEAP-Wetland project, thistudy examined three land use types throughout the Mid-Atlanticegion (Delaware, Maryland, Virginia, and North Carolina).

Ecology 76 (2014) 87– 94

• Natural-forested wetlands with native vegetation and no historyof drainage.

• Converted-wetlands that were drained and converted to farm-land.

• Restored-hydraulically restored wetlands that had been previ-ously drained and converted to farmland.

Restored sites were selected from a list provided by theNatural Resources Conservation Service (NRCS) consisting of farm-land restored under Conservation Practice Standard 657 (wetlandrestoration) or Conservation Practice Standard 646 (shallow waterdevelopment and management). Detailed information on theseconservation practice standards can be found on the NRCS FieldOffice Technical Guide website (http://www.nrcs.usda.gov/wps/portal/nrcs/main/national/technical/fotg/). After Restored siteswere identified, aerial photographs, digital elevation models, andNRCS personnel were used to select corresponding natural and con-verted sites within 1 to 4 km from the restored sites. This studyexamined twelve MIAR-CEAP sites, including one of each land usetype (i.e., natural, restored and converted) in each of four states.

Natural sites in Delaware, Maryland, and Virginia were clas-sified as depressional wetlands (confined to basins or hollows)and were managed by the state or the Nature Conservancy. InNorth Carolina, pocosin wetlands (acidic, peat soils) within PocosinLakes National Wildlife Refuge were used as natural sites. Domi-nant vegetation in the natural sites included mixed oak, loblollypine, and sweet gum. Soil series names and soil taxonomy for allsites are presented in Table 1. In all states, Converted sites weredrained, filled or leveled prior to 1985 and were typically man-aged under a corn–wheat–soybean rotation. Hydraulic restorationof the restored sites occurred in 2004 (DE and MD), 2008 (VA), and2005 (NC). The VA and MD restored sites were replanted, whileDE and NC were naturally re-vegetated. Visual inspection prior tosite selection insured that restored sites had the appeared similarto the corresponding natural wetlands. The natural and restoredwetlands used in this study were seasonally, but not permanentlysaturated.

2.2. Field sampling

Samples were collected at three time points: spring 2010, fall2010, and spring 2011. For each site, a digital elevation model inArcGIS (ESRI, Redlands, CA) and on-site visits were used to identifythe lowest wetland topographic position. An approximately 9 m2

area was randomly selected within the lowest topographic positionof each site. At each sampling date, four locations were randomlysampled within this area. One location was used to measure soilsurface gas emissions as described below, and the top 10 cm of min-eral soil was collected from the three additional locations to createa composite sample for analysis. If present, the organic horizon wasremoved prior to collecting soil samples. Soils for microbial com-munity analysis were stored at −80 ◦C, and soils used for pH, C, andN analysis were stored at 4 ◦C and processed within one week ofsampling.

2.3. Soil properties and gas emissions

Soil surface efflux rates were determined using an Innova 1412photoacoustic gas analyzer, fitted with a 987 optical filter (PAGA;LumaSense Technologies, Santa Clara CA). A static chamber wasconstructed from a metal beaker with three holes drilled into thebase. Two holes were used for inlet and outlet tubes and a rubber

stopper was inserted into the third. The chamber was wrapped in aninsulating material and covered with aluminium foil to minimizetemperature variations and the lip of the beaker was beveled to easeinsertion into the soil. For each sampling, the rubber stopper was

L.A. Kluber et al. / Applied Soil Ecology 76 (2014) 87– 94 89

Table 1Soil series and taxonomic classes for sites used in this study. Complete soil series descriptions are available at http://www.nrcs.usda.gov/wps/portal/nrcs/detailfull/soils/home/?cid = nrcs142p2 053587.

State & Land use Soil series Soil taxonomic class

DelawareNatural Hammonton-Fallsington-Mullica Complex Mesic Aquic Hapludult, Mesic Typic EndoaquultRestored Fallsington Mesic Typic EndoaquultConverted Carmichael Mesic Typic Endoaquult

MarylandNatural Corsica Mesic Typic UmbraquultRestored Hammonton-Fallsington-Corsica Complex Mesic Aquic Hapludult, Mesic Typic Endoaquult, Mesic Typic UmbraquultConverted Hammonton-Fallsington-Corisica Complex; Ingleside Mesic Aquic Hapludult, Mesic Typic Endoaquult, Mesic Typic Umbraquult

VirginiaNatural Myatt Thermic Typic EndoaquultRestored Roanoke Thermic Typic EndoaquultConverted Bojac Thermic Typic Hapludult

North Carolina

r1aNottwur

wn1Cs

2

sfrdpvummwbaociGtrysc

2

oA

axes scores using PROC GLM in SAS version 9.3 to test whether bac-

Natural Ponzer

Restored Ponzer

Converted Ponzer

emoved, and the chamber was inserted 5 cm into the soil, leaving150 cm3 headspace for sampling. The rubber stopper was replacednd headspace gas was pumped into the PAGA. Concentrations of2O, CO2, H2O vapor, and SF6 were measured every 60 s for a totalf ten readings per location. Water vapor measurements were usedo correct N2O concentration and manually injected SF6 was usedo monitor the system for leaks (Miller et al., 2012). Raw PAGA dataas converted from ppm V to mg m−3 and regressed against timesing a linear function using GraphPad Prism5 to determine fluxates.

Soil temperature, moisture, and electrical conductivity (EC)ere measured in situ using an ECTestr11 + meter (Spectrum Tech-ologies, East Plainfield, Ill.). Soil pH was measured after mixing0 g soil with 10 g deionized water. A TruSpec CN analyzer (Lecoorp., St Joseph, MI) was used to measure total C and N on air driedoils.

.4. Statistical analysis on soil properties and gas emissions

Analysis of soil properties and gas flux was performed in SAS ver-ion 9.3 (SAS Institute Inc.). Prior to analysis, variables were testedor normality and log transformed as necessary. Analysis was car-ied out using a PROC MIXED model with land use and samplingate as fixed effects and state (block) as a random effect. The appro-riate covariance structure was determined by comparing AICCalues, and the SLICES option was used to test for the effect of landse within each sampling date. Significant differences were deter-ined using post-hoc Tukey HSD tests for comparisons amongeans at a significance level of p < 0.05. Additionally, correlationsere performed with PROC CORR to examine the relationship

etween gas emissions and soil properties. To examine the over-ll influence of study parameters on edaphic conditions, a matrixf chemical and physical properties was constructed, data wereentered and normalized, and non-metric multidimensional scal-ng (NMS) was conducted with PC-ORD v6 (MjM Software Design,leneden Beach, OR). The Euclidian distance measure was used for

his analysis as it is compatible with the negative numbers thatesult from centering and normalizing the data. Multivariate anal-sis of variance (MANOVA) was then performed on the NMS axescores using PROC GLM in SAS version 9.3 to test whether edaphiconditions differed by land use, state, or season.

.5. DNA extraction and pyrosequencing

Bacterial community composition was assessed using a subsetf samples from two time points; the fall and the following spring.

total of 24 soil samples (3treatments × 4 states × 2 seasons, n = 4)

Thermic Terric HaplosapristThermic Terric HaplosapristThermic Terric Haplosaprist

were submitted to the University of Nebraska–Lincoln Core forApplied Genomics and Ecology for DNA extraction and sequenc-ing. Briefly, DNA was extracted using the procedure described byMartinez et al. (2009) and the V1,V2 region of the 16S rDNA genewas amplified using the 8FM and 357R primers (Benson et al.,2010) that were modified with 8-base barcodes and Roche-454Titanium adapters. The amplicons from the 24 reactions were quan-tified, pooled, gel purified, and quantified with a Qubit fluorometer(Invitrogen) before sequencing with Roche-454 GS FLX titaniumchemistry. Raw data were then quality screened, and parsed bybarcode. Detailed amplification, sequencing, quality control, andscreening methods can be found in Benson et al. (2010).

2.6. Bioinformatics and bacterial community analysis

The CloVR-16 S automated pipeline (Angiuoli et al., 2011) wasemployed to further quality screen reads, and analyze 16S rDNAsequences. Sequences shorter than 100 bp were removed from thedata set. Sequences containing ambiguous base calls or above thehomopolymer maximum of 8 bp were also removed from the dataset. Chimeric sequences were detected and removed from the dataset using UCHIME (Edgar et al., 2011). Sequences were then clus-tered into OTUs with a pair wise similarity of 97% using the UCLUSTfeature in QIIME (Caporaso et al., 2010), and taxonomy was assignedusing the RDP Bayesian classifier (Wang et al., 2007). Furthermore,OTU richness and diversity statistics were calculated by Mothur(Schloss et al., 2009).

To assess how bacterial richness and diversity were affectedby land use, state, or season, the Chao-1 richness estimator andShannon diversity index (H′) were tested with the mixed modeldescribed above in Section 2.4. To test whether dominant bacte-rial taxa varied by treatment, class-level abundance was Hellingertransformed to meet normality assumptions and analyzed usingthe previously described mixed model. A matrix of bacterial OTUabundance was constructed and singletons (present only in onesample) were removed from the data set to reduce the noise fromrare species (McCune and Grace, 2002). Non-metric multidimen-sional scaling (NMS) was conducted with the Sorensen distancemeasure in PC-ORD v6 (MjM Software Design, Gleneden Beach, OR)to examine the structure of the bacterial communities. Multivari-ate analysis of variance (MANOVA) was then performed on the NMS

terial communities were significantly clustered by land use, state,or season. The correlation between bacterial communities (OTUs)and soil properties was investigated with Mantel tests using MonteCarlo randomization to evaluate the test statistic (PC-ORD v6).

90 L.A. Kluber et al. / Applied Soil Ecology 76 (2014) 87– 94

Table 2The effect of land use and sampling date on soil chemical and physical properties reported as mean values and standard error; significant differences are denoted with letters.

Moisture (%) pH Temperature (◦C) C (%) N (%) C:N Ratio EC (�S cm−1)

Land useNatural 49.01a (3.75) 4.15b (0.13) 20.34b (1.16) 15.35a (5.13) 0.59a (0.13) 20.53a (2.55) 66.60a (6.29)Restored 36.69b (5.14) 5.76a (0.21) 23.35ab (1.55) 7.75b (2.71) 0.34b (0.06) 17.09b (2.95) 107.63ab (15.28)Converted 29.27b (3.55) 6.13a (0.19) 23.75a (1.96) 6.89b (2.50) 0.37b (0.09) 13.9b (1.87) 170.15b (37.22)p-value <0.01 <0.01 0.04 <0.01 <0.01 <0.01 0.03

DateSpring 2010 32.88 (4.30) 5.57a (0.29) 27.23a (1.22) 9.46 (3.69) 0.42 (0.10) 17.00 (1.86) 103.02 (21.31)Fall 2010 45.25 (4.84) 5.15b (0.30) 17.78c (1.07) 10.14 (3.74) 0.49 (0.12) 16.25 (2.24) 107.37 (15.21)Spring 2011 38.31 (4.64) 5.15b (0.33) 21.72b (1.10) 11.18 (4.44) 0.42 (0.09) 18.72 (3.59) 133.09 (38.42)

T

3

3

ssgpaawpFgDt

Fv

p-value 0.06 0.01 <0.01

otal C and N as percent dry mass.

. Results

.1. Soil properties and gas emissions

Land use had a significant effect on all soil properties mea-ured. Although soil properties from restored wetlands were notignificantly different from converted sites, natural wetlands hadreater soil moisture, C, N, and C:N ratio, as well as decreased soilH and temperature (Table 2). Soil temperature and pH variedmong sampling dates with the highest mean soil temperaturend greatest pH occurring in the spring of 2010 (Table 2). Thereere no significant land use by date interactions for any soilroperties. The NMS ordination of soil properties (stress = 4.2;ig. 1A) showed that edaphic conditions were strongly grouped

eographically with NC being significantly different than theE, MD, and VA (p < 0.001). Although not visually discernible on

he ordination, the blocked MANOVA also indicated significant

ig. 1. NMS ordinations of (A) soil properties and (B) bacterial OTUs. As indicated in the lectors represent significant correlations with soil properties and percent variance expla

0.83 0.41 0.74 0.96

treatment difference with the Natural wetlands being sig-nificantly different than the Converted and Restored sites(p < 0.001).

There were no significant differences in N2O or CO2 emissionsamong the land use types (Fig. 2A). While the Converted andRestored sites had the greatest mean N2O emissions, they werehighly variable, and thus not significantly greater than the naturalsites. Similarly, the high variability contributed to there being nosignificant effect of date on N2O emissions. Sampling date did havean effect on the CO2 flux rate with spring 2010 having the greatestand fall 2010 having the lowest CO2 emissions (Fig. 2B). There wasnot a significant land use by date interaction for either CO2 or N2O.There was a significant correlation between CO2 and N2O emissions(r = 0.47, p = 0.01), although only CO was correlated with soil tem-

2perature (r = 0.41, p = 0.02). Neither CO2 or N2O were significantlycorrelated with other soil properties (pH, moisture, C, N, C:N ratio,or EC).

egend, states are coded by color, and treatment is coded by shape. Join plot overlayined by each axis is indicated.

L.A. Kluber et al. / Applied Soil Ecology 76 (2014) 87– 94 91

Fig. 2. Means and standard error of N2O (black bars) and CO2 (grey bars) effluxrd

3

s5Gt2OTatrRwstAiss

TSmt

ates coded by land use type (A) and sampling date (B). Letters denote significantifferences among bars of same color (p < 0.05).

.2. Bacterial communities

A total of 582,179 sequence reads were returned from the 24oil samples. After removing low quality and chimeric sequences,35,636 sequences remained for subsequent analysis (Table 3).rouping by 97% sequence similarity resulted in 13,685 OTUs. Of

hese, 8,562 (62%) were found in only one land use type, while,765 (20%) were found in all three land use types (Fig. 3). Chao-1TU richness and bacterial diversity (H′) are reported in Table 3.here were no significant differences in OTU richness or diversitymong land use types or between the two seasons. Furthermore,here was not a significant land use by season interaction for eitherichness or diversity. Based on taxonomic lineage assigned by theDP Bayesian classifier Firmicutes, Proteobacteria, and Bacteroidetesere the most abundant phyla accounting for 53, 22 and 11% of all

equences, respectively. Surprisingly, there were few differences inaxa abundance among the states, treatments and dates (Table 4).ctinobacteria were the only taxa to differ among treatments, hav-

ng greater abundance in the Converted sites compared Restored

ites (p = 0.01). Deltaproteobacteria differed between fall and springamples, with the spring having a greater abundance (p < 0.01).

able 3equencing results and OTU richness and diversity for each land use reported asean and standard error. Diversity indices were not significantly different among

reatments.

Natural Restored Converted

Total sequences 24,325 (2,158) 25,815 (3,738) 22,633 (2,397)Final high quality sequences 22,423 (2,002) 23,812 (3,435) 20,720 (2,127)Chao-1 OTU richness 2,912 (55) 2,838 (76) 2,830 (82)OTU diversity (H′) 5.17 (0.03) 5.06 (0.03) 5.26 (0.01)

Fig. 3. Venn Diagram showing the distribution of bacterial OTUs among the threetreatments.

The NMS ordination of bacterial OTUs explained 95.6% of thecommunity variance (stress = 8.7, Fig. 1B) and the blocked MANOVAindicated that microbial communities varied by state (p < 0.01), butnot treatment (p = 0.17) or season (p = 0.44). The geographic separa-tion of the communities was also confirmed in the MANOVA resultswith communities from the southernmost samples (NC) being sig-nificantly different from the northernmost samples (DE and MD;p < 0.01 & p = 0.03, respectively). Accordingly, bacterial communi-ties from the VA samples were not significantly different from thesurrounding states (NC and MD) but significantly different fromDE which it does not border (p = 0.03). The NMS ordinations forsoil properties and bacterial communities (Fig. 1A & 1B) showedsimilar trends and the Mantel test confirmed that soil bacterialcommunities were indeed significantly correlated with soil prop-erties (r = 0.24, p = 0.03).

4. Discussion

Land use had a significant impact on soil properties with Natu-ral and Converted wetlands having the greatest differences. Thisfinding was expected given that conversion of forested land toagriculture has been shown to influence soil properties (Murtyet al., 2002). Similar to the findings of others (Bruland et al., 2003;Hogan et al., 2004; Morse and Bernhardt, 2013), soil properties inrestored wetlands reflected their agricultural legacy and had higherpH and lower C and N compared to natural wetlands. Our find-ings, and those of others (Bruland et al., 2003; Hogan et al., 2004),demonstrate that there is a considerable lag time in the responseof soil properties to hydraulic restoration. Thus, it is important toexamine long-term trends in soil properties. For example, whileRestored wetlands may always have lower C than natural wetlands,an increase in soil C compared to the converted sites would indicatea positive ecosystem service.

Here, we report that soil CO2 emissions were not significantlydifferent among the land use types. This finding is in accordancewith those of Morse et al. (2012a) who examined pocosin wet-lands in North Carolina and found no effect of land use on CO2 fluxrates. While our study included pocosin wetlands, we also exam-ined depressional wetlands and were able to confirm that this trendholds true across a larger geographic area and multiple soil types.Although both soil temperature and moisture content can influence

CO2 production (Fierer et al., 2003), we found that temperature wasthe only soil property that correlated with CO2 emissions. The cor-relation between CO2 and temperature is likely the driver behindthe seasonal variation in soil CO2 fluxes as well.

92 L.A. Kluber et al. / Applied Soil Ecology 76 (2014) 87– 94

Table 4Relative abundance of bacterial taxa for each land use type (mean and standard deviation) and p-values from the mixed model analysis of the Hellinger transformed taxaabundance.

Taxa % Abundance p-values

Natural Restored Converted State Treatment Date

Acidobacteria 6.72 (3.67) 4.11 (0.64) 4.86 (0.81) 0.15 0.05 0.56Actinobacteria 2.35 (0.55) 2.17 (0.43) 3.02 (0.70) 0.10 0.01 0.82Bacteroidetes 12.29 (0.85) 12.37 (1.45) 12.46 (1.58) 0.67 0.96 0.88

Bacteroidetes 9.76 (0.72) 9.73 (1.31) 9.7 (1.49) 0.79 0.94 0.80Flavobacteria 0.86 (0.36) 0.99 (0.51) 0.93 (0.27) 0.07 0.75 0.06Sphingobacteria 0.36 (0.12) 0.27 (0.09) 0.44 (0.16) 0.97 0.11 0.41Other 0.11 (0.03) 0.13 (0.04) 0.14 (0.32) 0.73 0.43 0.86

Chloroflexi 0.13 (0.04) 0.15 (0.04) 0.14 (0.05) 0.93 0.56 0.15Cyanobacteria 1.07 (0.19) 1.10 (0.19) 1.11 (0.05) 0.11 0.98 0.84Firmicutes 54.13 (4.59) 53.80 (5.8) 54.43 (1.25) 0.22 0.50 0.67

Bacilli 34.43 (2.59) 33.22 (3.95) 33.10 (3.14) 0.05 0.39 0.10Clostridia 18.62 (2.13) 18.10 (2.81) 18.46 (2.98) 0.34 0.80 0.29Erysipelotrichi 0.72 (0.18) 0.86 (0.43) 1.11 (0.53) 0.01 0.06 0.31Other 0.36 (0.06) 0.32 (0.15) 0.35 (0.11) 0.35 0.62 0.05

Fusobacteria 0.49 (0.09) 0.55 (0.23) 0.64 (0.12) 0.71 0.88 0.41Gemmatimonadetes 0.51 (0.07) 0.58 (0.10) 0.62 (0.19) 0.75 0.41 0.93Nitrospira 0.10 (0.02) 0.17 (0.06) 0.15 (0.08) 0.23 0.07 0.08Proteobacteria 19.55 (1.15) 24.08 (7.71) 21.69 (1.66) 0.69 0.22 0.79

Alphaproteobacteria 2.88 (0.78) 2.40 (0.40) 2.71 (0.53) 0.37 0.24 0.11Betaproteobacteria 3.78 (0.44) 7.14 (7.31) 4.42 (0.74) 0.31 0.22 0.23Deltaproteobacteria 1.93 (0.32) 1.88 (0.33) 1.97 (0.41) 0.92 0.80 0.01Epsilonproteobacteria 0.15 (0.05) 0.15 (0.05) 0.15 (0.11) 0.32 0.95 0.06Gammaproteobacteria 9.98 (0.76) 11.70 (2.40) 11.44 (0.11) 0.81 0.19 0.63Other 0.83 (0.16) 0.81 (0.14) 1.04 (0.28) 0.89 0.13 0.51

tclrelwremBan2Nr2agfiweete

tFNMipttpe

Other bacteria 3.45 (0.43) 3.04 (0.63)

Soils are the source of 70% of N2O produced (Conrad, 1996), andhe high reactivity and long life span of N2O make it a primaryoncern for those evaluating the environmental impacts of wet-and restoration (Verhoeven et al., 2006). Our findings show thatestored wetlands along the Mid-Atlantic Coastal Plain do not havelevated N2O emissions and that N2O flux rates from restored wet-ands are similar to those found in agricultural soils and natural

etlands. However, there was large variation in N2O flux rates fromestored and converted soils, potentially due to site to site differ-nces in N fertilization and/or runoff. This high variation could beasking treatment difference and contributing to a type 2 error.

ecause N2O can be produced by both aerobic (nitrification) andnaerobic (denitrification) processes, and each is influenced by aumber of factors (Morse et al., 2012b; Sahrawat, 2008; Vilain et al.,010), it is difficult to identify environmental variables that predict2O flux rates across ecosystems. It has been suggested that C:N

atio may be used to predict potential N2O emissions (Hunt et al.,007); however, we found no significant correlations between C:Nnd soil surface N2O measurements. Vilain et al. (2010) attributedreater N2O flux from low landscape positions to greater water-lled pore space, and others have found that N2O flux was greatesthen water tables were high (Berglund and Berglund, 2011; Flessa

t al., 1998). Thus it was somewhat unexpected that we saw noffect of water content on N2O flux rates. This finding underscoreshe difficulty in predicting how hydraulic regimes influence N2Omissions.

Adding to the difficulty in predicting N2O emission is the facthat denitrification and N loss do not occur at a steady rate in soil.or example, Groffman and Tiedje (1989) found that 80% of annual

loss occurred over a three to six week period. More recently,orse and colleagues (2012b) developed a model incorporat-

ng CO2 flux, soil temperature, ammonium, and denitrificationotential to predict N2O emissions across land use types. While

he model performed well in the fall, it performed poorly inhe summer months, thus highlighting the difficulty of N2O fluxrediction over time. Because precipitation and agricultural runoffvents occur in pulses, more extensive year-round sampling may

2.98 (0.39) 0.38 0.18 0.49

be necessary to better understand soil processes involved inregulating N2O emissions. Understanding the processes regulatingN2O flux may become more important in the future as Dijkstraet al. (2012) reported that elevated temperature and atmosphericCO2 can increase N2O emissions, particularly when combined withN fertilization. Thus, suggesting that agriculturally impacted andrestored wetlands with high N content may have greater N2Oemissions as atmospheric CO2 levels increase.

The dominant bacterial taxa in this study (Firmicutes, Proteobac-teria, Bacteroidetes, and Acidobacteria) are common in a variety ofsoils (Lauber et al., 2009; Nacke et al., 2011) including wetland soils(Hartman et al., 2008). Soil bacterial communities are known tovary with soil properties, disturbance, and land use (Chaer et al.,2009; Lauber et al., 2009; Lauber et al., 2008). Therefore, giventhe differences in soil properties among the natural, restored, andconverted land uses, it is somewhat surprising that bacterial com-munities varied so little across treatments. While we found thatsoil bacterial communities were correlated with soil properties, theeffects of treatment on these communities was minimal comparedto the differences by state. As such, the C rich pocosin wetlandsin NC had significantly different bacterial communities than thedepressional wetlands in DE, MD, or VA, regardless of the currentland use. This contrasts with previous work showing that floodingagricultural land led to altered microbial communities (Bossio et al.,2006), as well as the findings of Hartman and colleagues (2008) whoreported that bacterial communities are significantly different inagricultural, restored and reference wetlands and correlate withsoil pH. It should be noted, however, that our study encompasseda larger geographic region compared to the studies by Bossio et al.(2006) and Hartman et al. (2008). As such, it is possible that treat-ment effects in our study were impacted by regional differences insoil properties.

In examining effect of land use on pocosin wetlands, Hartman

et al. (2008) found that bacterial diversity was lowest in naturalwetlands; thus, restored sites having lower diversity than agricul-tural sites was an indication of ecosystem recovery. Furthermore,the authors reported that Acidobacteria abundance was greatest in

d Soil

ravitfidr

gvavw(h(cciwtehhepeocc

n(1wtcNttotgcetiparwt

A

faofsUoU

L.A. Kluber et al. / Applie

eference wetlands while Proteobacteria were most abundant ingricultural sites. Here, we found that bacterial diversity did notary with land use. Additionally, only Actinobacteria were greatern converted than restored sites, and no other taxa showed areatment difference. The similarities and differences between ourndings and those of Hartman et al. (2008) are intriguing and likelyue to differences in hydrology, soil properties, and time sinceestoration inception.

Interestingly, natural and restored wetlands did not have areater abundance of N-cycling organisms compared to Con-erted wetlands. Nitrosomonas, Nitrospira and other ammoniand nitrite oxidizing bacteria (NOB) are commonly found in aariety of habitats including soils (Freitag et al., 2005). Alongith denitrifying bacteria, NOB can produce N2O as a byproduct

Goreau et al., 1980). Previous work in agricultural soils examinedow NOB populations correlated with potential nitrite oxidationPNO). Interestingly, while Nitrobacter-like NOB were positivelyorrelated with PNO, Nitrospira-like NOB had a weak, negativeorrelation with PNO (Attard et al., 2010). Although nitrifications responsible for a significant portion of N2O production in some

etlands (Morse and Bernhardt, 2013), further study is necessaryo determine how the relative abundance of these organisms influ-nce N cycling and N2O flux in Restored wetlands. Recent workas suggested that microbial predictors (gene abundance) aloneave little explanatory power for N-cycling processes (Grahamt al., 2014) and that across seasons, edaphic factors had betterredictive power. Within a season, however, a combination ofdaphic and microbial predictors increased the explanatory powerf their model. The findings of Graham et al. (2014) highlight theomplexity of separating the effect of soil properties and microbialommunities on N-cycling processes.

In conclusion, wetland restoration can improve filtration ofutrient and sediment loads, water storage, and wildlife habitatBrinson and Eckles, 2011; DeSteven and Lowrance, 2011; Mitsch,992). For example, Jordan et al. (2003) reported that a restoredetland reduced 52% of nitrate and 25% of ammonium in agricul-

ural runoff over a two year period. While some have expressedoncern that these positive effects could be offset by increased2O emissions (Verhoeven et al., 2006), our findings demonstrate

hat under the conditions we examined in the Coastal Plain region,his concern may be unwarranted. Indeed, the short hydroperi-ds, incomplete saturation, and bacterial communities present inhe restored wetlands studied here do not appear to exacerbatereenhouse gas production. Results from this study underscore theomplex interplay between regional soil properties, land use, andnvironmental conditions and the effects they have on the struc-ure and function of soil microbial communities. However, they alsondicate that hydraulic restoration creates conditions that allow soilroperties and biogeochemical processes to recover from years ofltered land use. Although biogeochemical processes experience aesponse lag after hydraulic restoration, soil properties in restoredetlands appear to be in a transitional state and are progressing

owards properties indicative of natural wetlands.

cknowledgements

We thank Anthony Shriner, William Brigman, and Katie Lewisor their field and laboratory assistance, Dr. Zaid Abdo for statisticaldvice, and Drs. Ariel Szogi and Ken Stone for thoughtful discussionsn the manuscript content. We also thank the property ownersor their cooperation with this study and allowing access to field

ites. Funding was provided by the Wetland Component of theSDA National Conservation Effects Assessment Project. Mentionf trade or firm names does not constitute an endorsement by the.S. Department of Agriculture.

Ecology 76 (2014) 87– 94 93

References

Angiuoli, S.V., White, J.R., Matalka, M., White, O., Fricke, W.F., 2011. Resources andcosts for microbial sequence analysis evaluated using virtual machines and cloudcomputing. PloS One 6, e26624.

Ardón, M., Morse, J.L., Doyle, M.W., Bernhardt, E.S., 2010. The water quality conse-quences of restoring wetland hydrology to a large agricultural watershed in thesoutheastern coastal plain. Ecosystems 13, 1060–1078.

Attard, E., Poly, F., Commeaux, C., Laurent, F., Terada, A., Smets, B.F., Recous, S.,Roux, X.L., 2010. Shifts between nitrospira- and nitrobacter-like nitrite oxidiz-ers underlie the response of soil potential nitrite oxidation to changes in tillagepractices. Environ. Microbiol. 12, 315–326.

Bendor, T., 2009. A dynamic analysis of the wetland mitigation process and its effectson no net loss policy. Landscape Urban Plan. 89, 17–27.

Benson, A.K., Kelly, S.A., Legge, R., Ma, F., Low, S.J., Kim, J., Zhang, M., Oh, P.L., Nehren-berg, D., Hua, K., Kachman, S.D., Moriyama, E.N., Walter, J., Peterson, D.A., Pomp,D., 2010. Individuality in gut microbiota composition is a complex polygenic traitshaped by multiple environmental and host genetic factors. Proc. Nat. Acad. Sci.USA 107, 18933–18938.

Berglund, Ö., Berglund, K., 2011. Influence of water table level and soil properties onemissions of greenhouse gases from cultivated peat soil. Soil Biol. Biochem. 43,923–931.

Bossio, D.A., Fleck, J.A., Scow, K.M., Fujii, R., 2006. Alteration of soil microbial commu-nities and water quality in restored wetlands. Soil Biol. Biochem. 38, 1223–1233.

Brinson, M.M., Eckles, S.D., 2011. U.S. Department of agriculture conservation pro-gram and practice effects on wetland ecosystem services: a synthesis. Ecol. App.21, S116–S127.

Bruland, G.L., Hanchey, M.F., Richardson, C.J., 2003. Effects of agriculture and wetlandrestoration on hydrology, soils, and water quality of a carolina bay complex.Wetlands Ecol. Manage. 11, 141–156.

Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello,E.K., Fierer, N., Pena, A.G., Goodrich, J.K., Gordon, J.I., Huttley, G.A., Kelley, S.T.,Knights, D., Koenig, J.E., Ley, R.E., Lozupone, C.A., McDonald, D., Muegge, B.D.,Pirrung, M., Reeder, J., Sevinsky, J.R., Turnbaugh, P.J., Walters, W.A., Widmann,J., Yatsunenko, T., Zaneveld, J., Knight, R., 2010. Qiime allows analysis of high-throughput community sequencing data. Nat. Meth. 7, 335–336.

Chaer, G.M., Fernandes, M.F., Myrold, D.D., Bottomley, P.J., 2009. Shifts in microbialcommunity composition and physiological profiles across a gradient of inducedsoil degradation. Soil Sci. Soc. Am. J. 73, 1327–1334.

Conrad, R., 1996. Soil microorganisms as controllers of atmospheric trace gases (H2,CO, CH4, OCS, N2O, and NO). Microbiol. Rev. 60, 609–640.

Dahl, T.E., 1990. In: Fa, W.S. (Ed.), Wetlands Losses in the United States 1780s to1980s. Department of the Interior U.S., Washington DC, p. 13.

DeSteven, D., Lowrance, R., 2011. Agricultural conservation practices and wetlandecosystem services in the wetland-rich piedmont-coastal plain region. Ecol.Appl. 21, S3–S17.

Dijkstra, F.A., Prior, S.A., Runion, G.B., Torbert, H.A., Tian, H., Lu, C., Venterea, R.T.,2012. Effects of elevated carbon dioxide and increased temperature on methaneand nitrous oxide fluxes: evidence from field experiments. Front. Ecol. Environ.10, 520–527.

Edgar, R., Haas, B., Clemente, J.C., Quince, C., Knight, R., 2011. Uchime improvessensitivity and speed of chimera detection. Bioinformatics 27, 2194–2200.

Fierer, N., Allen, A.S., Schimel, J.P., Holden, P.A., 2003. Controls on microbial CO2

production: a comparison of surface and subsurface soil horizons. Glob. ChangeBiol. 9, 1322–1332.

Flessa, H., Wild, U., Klemisch, M., Pfadenhauer, J., 1998. Nitrous oxide and methanefluxes from organic soils under agriculture. Eur. J. Soil Sci. 49, 327–335.

Freitag, T.E., Chang, L., Clegg, C.D., Prosser, J.I., 2005. Influence of inorganic nitrogenmanagement regime on the diversity of nitrite-oxidizing bacteria in agriculturalgrassland soils. Appl. Environ. Microbiol. 71, 8323–8334.

Goreau, T.J., Kaplan, W.A., Wofsy, S.C., McElroy, M.B., Valois, F.W., Watson, S.W., 1980.Production of N2O2-and N2O by nitrifying bacteria at reduced concentrations ofoxygen. Appl. Environ. Microbiol. 40, 526–532.

Graham, E.B., Wieder, W.R., Leff, J.W., Weintraub, S.R., Townsend, A.R., Cleveland,C.C., Philippot, L., Nemergut, D.R., 2014. Do we need to understand microbialcommunities to predict ecosystem function? A comparison of statistical modelsof nitrogen cycling processes. Soil Biol. Biochem. 68, 279–282.

Groffman, P.M., Tiedje, J.M., 1989. Denitrification in north temperate forest soils:Spatial and temporal patterns at the landscape and seasonal scales. Soil Biol.Biochem. 21, 613–620.

Hansson, L.-A., Bronmark, C., Anders Nilsson, P., Abjornsson, K., 2005. Conflictingdemands on wetland ecosystem services: nutrient retention, biodiversity orboth? Freshwater Biol. 50, 705–714.

Hartman, W.H., Richardson, C.J., Vilgalys, R., Bruland, G.L., 2008. Environmental andanthropogenic controls over bacterial communities in wetland soils. Proc. Nat.Acad. Sci. 105, 17842–17847.

Hogan, D., Jordan, T., Walbridge, M., 2004. Phosphorus retention and soil organiccarbon in restored and natural freshwater wetlands. Wetlands 24, 573–585.

Hunt, P.G., Matheny, T.A., Ro, K.S., 2007. Nitrous oxide accumulation in soils fromriparian buffers of a coastal plain watershed carbon/nitrogen ratio control. J.Environ. Qual. 36, 1368–1376.

Hunter, R.G., Faulkner, S.P., 2001. Denitrification potentials in restored and naturalbottomland hardwood wetlands. Soil Sci. Soc. Am. J. 65, 1865–1872.

Jordan, T.E., Whigham, D.F., Hofmockel, K.H., Pittek, M.A., 2003. Nutrient and sed-iment removal by a restored wetland receiving agricultural runoff. J. Environ.Qual. 32, 1534–1547.

9 d Soil

L

L

M

M

M

M

MM

M

M

M

M

4 L.A. Kluber et al. / Applie

auber, C.L., Hamady, M., Knight, R., Fierer, N., 2009. Pyrosequencing-based assess-ment of soil ph as a predictor of soil bacterial community structure at thecontinental scale. Appl. Environ. Microbiol. 75, 5111–5120.

auber, C.L., Strickland, M.S., Bradford, M.A., Fierer, N., 2008. The influence of soilproperties on the structure of bacterial and fungal communities across land-usetypes. Soil Biol. Biochem. 40, 2407–2415.

artinez, I., Wallace, G., Zhang, C., Legge, R., Benson, A.K., Carr, T.P., Moriyama, E.N.,Walter, J., 2009. Diet-induced metabolic improvements in a hamster model ofhypercholesterolemia are strongly linked to alterations of the gut microbiota.Appl. Environ. Microbiol. 75, 4175–4184.

cCune, B., Grace, J.B., 2002. Analysis of Ecological Communities. MjM SoftwareDesign, Glendale Beach, Oregon.

iller, J.O., Hunt, P.D., Ducey, T.F., Glaz, B.S., 2012. Denitrification and gas emissionsfrom organic soils under different water table and flooding management. Trans.ASABE 55, 1793–1800.

itsch, W.J., 1992. Landscape design and the role of created, restored, and nat-ural riparian wetlands in controlling nonpoint source pollution. Ecol. Eng. 1,27–47.

itsch, W.J., Gosselink, J.G., 1993. Wetlands. Van Nostrand Reinhold, New York.orse, J.L., Ardón, M., Bernhardt, E.S., 2012a. Greenhouse gas fluxes in South-

eastern US coastal plain wetlands under contrasting land uses. Ecol. Appl. 22,264–280.

orse, J.L., Ardón, M., Bernhardt, E.S., 2012b. Using environmental variables and soilprocesses to forecast denitrification potential and nitrous oxide fluxes in coastalplain wetlands across different land uses. J. Geophys. Res. 117, G02023.

orse, J.L., Bernhardt, E.S., 2013. Using 15 N tracers to estimate N2O and N2 emissionsfrom nitrification and denitrification in coastal plain wetlands under contrastingland-uses. Soil Biol. Biochem. 57, 635–643.

urty, D., Kirschbaum, M.U.F., McMurtrie, R.E., McGilvray, H., 2002. Does conversion

of forest to agricultural land change soil carbon and nitrogen? A review of theliterature. Global Change Biol. 8, 105–123.

yrold, D.D., 1999. Transformations of nitrogen. In: Sylvia, D.M., Fuhrmann, J.J., Har-tel, P.G., Zuberer, D.A. (Eds.), Principles and Applications of Soil Microbiology.Prentice Hall, New Jersey.

Ecology 76 (2014) 87– 94

Nacke, H., Thürmer, A., Wollherr, A., Will, C., Hodac, L., Herold, N., Schöning, I.,Schrumpf, M., Daniel, R., 2011. Pyrosequencing-based assessment of bacterialcommunity structure along different management types in german forest andgrassland soils. PloS One 6, e17000.

Peralta, A.L., Matthews, J.W., Kent, A.D., 2010. Microbial community structureand denitrification in a wetland mitigation bank. Appl. Environ. Microbiol. 76,4207–4215.

Sahrawat, K.L., 2008. Factors affecting nitrification in soils. Commun. Soil Sci. PlantAnal. 39, 1436–1446.

Schlesinger, W.H., 1991. Biogeochemistry: An Analysis of Global Change, Second ed.Academic Press, San Diego, CA.

Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister,E.B., Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., Sahl, J.W.,Stres, B., Thallinger, G.G., Van Horn, D.J., Weber, C.F., 2009. Introducingmothur: open-source, platform-independent, community-supported softwarefor describing and comparing microbial communities. Appl. Environ. Microbiol.75, 7537–7541.

Tanner, C.C., Nguyen, M.L., Sukias, J.P.S., 2005. Nutrient removal by a constructedwetland treating subsurface drainage from grazed dairy pasture. Agric. Ecosyst.Environ. 105, 145–162.

Verhoeven, J.T.A., Arheimer, B., Yin, C., Hefting, M.M., 2006. Regional and globalconcerns over wetlands and water quality. Trends Ecol. Evol. 21, 96–103.

Vilain, G., Garnier, J., Tallec, G., Cellier, P., 2010. Effect of slope position and land useon nitrous oxide (N2O) emissions (Seine Basin, France). Agric. Forest Meteorol.150, 1192–1202.

Wang, Q., Garrity, G.M., Tiedje, J.M., Cole, J.R., 2007. Naïve bayesian classifier for rapidassignment of rrna sequences into the new bacterial taxonomy. Appl. Environ.Microbiol. 73, 5261–5267.

Whiting, G.J., Chanton, J.P., 2001. Greenhouse carbon balance of wetlands: methane

emission versus carbon sequestration. Tellus B 53, 521–528.

Zedler, J.B., 2000. Progress in wetland restoration ecology. Trends Ecol. Evol. 15,402–407.

Zedler, J.B., 2003. Wetlands at your service: reducing impacts of agriculture at thewatershed scale. Front. Ecol. Environ. 1, 65–72.