Biogeosciences 14 2611ndash2626 2017wwwbiogeosciencesnet1426112017doi105194bg-14-2611-2017copy Author(s) 2017 CC Attribution 30 License

Flooding-related increases in CO2 and N2O emissions from atemperate coastal grassland ecosystemAmanuel W Gebremichael123 Bruce Osborne23 and Patrick Orr13

1UCD School of Earth Sciences University College Dublin Belfield Dublin 4 Ireland2UCD School of Biology and Environmental Sciences University College Dublin Belfield Dublin 4 Ireland3UCD Earth Institute University College Dublin Belfield Dublin 4 Ireland

Correspondence to Amanuel W Gebremichael (amanuelgebremichaelucdconnectie)

Received 30 November 2016 ndash Discussion started 3 January 2017Revised 10 April 2017 ndash Accepted 17 April 2017 ndash Published 23 May 2017

Abstract Given their increasing trend in Europe an under-standing of the role that flooding events play in carbon (C)and nitrogen (N) cycling and greenhouse gas (GHG) emis-sions will be important for improved assessments of local andregional GHG budgets This study presents the results of ananalysis of the CO2 and N2O fluxes from a coastal grasslandecosystem affected by episodic flooding that was of either arelatively short (SFS) or long (LFS) duration Compared tothe SFS the annual CO2 and N2O emissions were 14 and 13times higher at the LFS respectively Mean CO2 emissionsduring the period of standing water were 144plusmn 1818 and111plusmn 951 mg CO2ndashC mminus2 hminus1 respectively for the LFSand SFS sites During the growing season when there wasno standing water the CO2 emissions were significantlylarger from the LFS (244plusmn 2488 mg CO2ndashC mminus2 hminus1) thanthe SFS (183plusmn 1490 mg CO2ndashC mminus2 hminus1) Fluxes of N2Oranged from minus037 to 065 mg N2OndashN mminus2 hminus1 at the LFSand fromminus050 to 055 mg N2OndashN mminus2 hminus1 at the SFS withthe larger emissions associated with the presence of stand-ing water at the LFS but during the growing season at theSFS Overall soil temperature and moisture were identifiedas the main drivers of the seasonal changes in CO2 fluxes butneither adequately explained the variations in N2O fluxesAnalysis of total C N microbial biomass and Q10 valuesindicated that the higher CO2 emissions from the LFS werelinked to the flooding-associated influx of nutrients and al-terations in soil microbial populations These results demon-strate that annual CO2 and N2O emissions can be higher inlonger-term flooded sites that receive significant amounts ofnutrients although this may depend on the restriction of dif-fusional limitations due to the presence of standing water to

periods of the year when the potential for gaseous emissionsare low

1 Introduction

The frequency of flooding events has increased in Europein the last 3 decades and is likely to increase further in awarmer climate as a consequence of climate change (Benis-ton et al 2007 Christensen and Christensen 2007) It isrecognised that flooding could have significant implicationsfor greenhouse gas (GHG) emissions for example due towater acting as a barrier to gaseous diffusion as well as viaalterations to soil biological and physio-chemical processes(Hansen et al 2014 Peralta et al 2013) Recent studies onthe impact of longer-term flooding (LFS) have attributed dif-ferences in GHG emissions to a range of factors includingnutrient availability (Juutinen et al 2001 Samaritani et al2011) soil microbial activity (Unger et al 2009 Peralta etal 2013) oxygen concentration (McNicol and Silver 2014)and vegetation characteristics (Lewis et al 2014) Howeverthere is less information on short-to-medium-term floodingevents and it is unclear how flooding impacts on ecosystemsthat experience variable periods of inundation The impact offlooding will also depend on the timing extent and durationof the period of inundation as well as on site-related varia-tions in topography and hydrology The extent of freshwaterflooding will also depend on the frequency and magnitude ofrainfall events

The natural topographic and hydrological conditions ofmany coastal plain or floodplain wetlands increases the pos-

Published by Copernicus Publications on behalf of the European Geosciences Union

2612 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

sibility of these ecosystems being rapidly inundated withfreshwater following extreme precipitation events or ex-tended periods of rainfall Episodic flooding is a commonfeature of these ecosystems with potential sources of wa-ter from inland or marine incursions although inundationfrom freshwater sources is probably the more common(Doornkamp 1998) Many assessments of GHG emissionsand an understanding of the factors that control them arebased on information from ecosystems less prone to floodingepisodes These data may not be directly applicable to coastalareas or floodplains where both the volumes involved and thefrequency and duration of water incursions can be high Forexample soil water availability can affect the emissions ofCO2 and N2O by influencing the rates of C and N mineral-isation (Alongi et al 1999 Noe et al 2013) but it is lessclear how variations in standing water levels influence theseprocesses although this likely depends on the extent and theduration of inundation (Lewis et al 2014) Previous studiesprovide contrasting evidence as to whether soil inundationenhances or impedes organic matter mineralisation Wilsonet al (2011) and Kim et al (2015) showed that the rate of or-ganic matter mineralisation can be enhanced after a short hy-droperiod (the period of time the soil area was waterlogged)whereas longer inundation periods suppressed decomposi-tion by limiting oxygen supply (Lewis et al 2014) Flood-ing generally promotes anaerobic conditions by excludingair from the pore spaces in the soil that would reduce min-eralisation of organic matter (Altor and Mitsch 2008) De-spite lower mineralisation rates under these conditions CO2emissions are still possible from anoxic soils (Glatzel et al2004) A number of factors such as the quality and quantityof substrates available for microbial processes temperaturesoil microbial activity and oxidation-reduction potential po-tentially impact on the magnitude of CO2 or N2O exchangein intermittently flooded ecosystems High soil organic mat-ter concentrations in combination with warmer conditions inflooded soil can enhance CO2 production (Oelbermann andSchiff 2008 Kim et al 2015) When associated with suf-ficient oxygen CO2 emissions in wetlands increase as a re-sult of accelerated organic matter decomposition whilst CH4production decreases because of aerobic methane oxidation(Smith et al 2003) Although CH4 is often considered to bethe more significant GHG in wet or flooded ecosystems thismay not be the case where there are only temporary hydrope-riods (Audet et al 2013 Batson et al 2015) where CO2fluxes are still likely to dominate the annual budget Finallythe effect of hydroperiod duration on N2O fluxes has receivedless attention We therefore focussed on N2O and CO2 fluxeswhich are likely to be the more important in many terrestrialecosystems

Denitrification and nitrification are often considered to bethe main mechanisms by which N2O is produced in soil al-though denitrification may be the dominant process for therelease of N2O both under aerobic and anaerobic soil con-ditions (Bateman and Baggs 2005) Denitrification and in-

creased N2O production is enhanced by high soil water con-tents (Machefert and Dise 2004 Maag and Vinther 1996)although the frequency and duration of flooding may alsobe important In riparian systems a shorter flooding dura-tion correlated with the highest N2O emissions which di-minished the longer the period of inundation (Jacinthe et al2012) In addition to water availability temperature is alsorecognised as an important variable in determining the pro-duction and consumption of N2O and CO2 via its effects onthe metabolic activity of microorganisms and plants (David-son and Janssens 2006 Butterbach-Bahl et al 2013 Kirwanet al 2014 Kim et al 2015)

Flooding stimulates changes in the structure of soil mi-crobial communities (Bossio and Scow 1998 Unger et al2009 Wilson et al 2011) and this in turn affects the rate ofdecomposition of organic material (Van Der Heijden et al2008) The extent of any change in microbial biomass (MB)andor microbial populations could also vary with floodingduration (Rinklebe and Langer 2006) For example Wilsonet al (2011) showed significant changes in microbial struc-ture and increases in soil enzymatic activity after short-term(24-day) inundation of floodplain soil Nevertheless an in-crease in microbial activity and GHG emissions as a responseto short-term flooding (SFS) is not always found (Unger etal 2009 Jacinthe 2015) and the timing as well as the fre-quency of flooding events may be important in determin-ing both the microbial community change and the associ-ated GHG emissions Organic substrate availability (via thebreakdown of plant litter or the introduction of organic com-pounds in flowing water) is a key factor regulating the activ-ities of soil microorganisms and the mineralisation and im-mobilisation of carbon and nitrogen (Badiou et al 2011 Liet al 2015) The availability of organic substrates influencesthe activity of carbon-cycling extracellular enzymes and asa result can impact on the rate of CO2 emissions and poten-tially other GHGs (Li et al 2015)

Owing to the episodic nature of flooding events captur-ing their impact on GHG emissions is quite challenging anda possible explanation for the lack of data from field studiesGiven the projected increase in flooding events an evaluationof their impact on GHG emissions is clearly required Theidentification of sites with flooding potential before floodingevents and the establishment of the appropriate measurementprotocols are crucial steps required to capture the dynam-ics of GHG fluxes in response to real flooding events Thiswould enable the capture of the full pattern of GHG fluxesprior to during and after a flooding event In this paperwe assess the impact of freshwater inundation with differ-ent hydroperiods on CO2 and N2O fluxes before during andafter real flooding events in a coastal grassland ecosystemover sim 2 years The main objective was to assess the effectsof short- and long-term flooding on the dynamics of CO2and N2O fluxes and how this impacts on the annual budgetsfor each gas We also evaluate the significance of a number

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2613

of soil physical chemical and biological parameters whichmay influence the fluxes of CO2 and N2O

2 Materials and methods

21 Site description

The study site (53

05prime87primeprime N 604prime07primeprimeW) was situated on alow lying area (sim 0 m asl) of coastal grassland that formspart of the East Coast Nature Reserve (ECNR) a portion ofthe larger coastal wetland complex called the Murrough wet-lands on the east coast of Ireland near Newcastle CountyWicklow (Fig 1) The site is owned and managed by Bird-Watch Ireland predominantly as a habitat for a wide vari-ety of birds The grassland area lies between a long drainageditch that runs in a northndashsouth direction and a shingle beachbordering the Irish Sea In the past 12 years restoration meth-ods including water management low-intensity grazing andcrop planting have been undertaken to maintain the site Thestanding water level varies spatially across the site and fluctu-ates seasonally in response to rainfall and the water-retainingcapacity of the drainage ditch The site experiences near tocomplete saturation with localised flooding in autumn andwinter but drains and dries out in spring and summer Despitebeing located in a coastal area the ditch water is generallycharacterized as being freshwater with the majority sourcedfrom the landward side

Differences in elevation (sim 20ndash25 cm vertical height) overthe site result in variations in hydrological connectivity be-tween different portions of the grassland and the ditch watersystem Initial monitoring of the general area indicated thatthis resulted in spatial hydroperiodic differences The sitewas therefore divided into two areas based on these hydrope-riod characteristics The site on the more elevated groundis characterized by less-frequent and shorter-term floodingwhilst the site located at lower elevations becomes inun-dated seasonally for an extended period during the winter(LFS) The plant community of the SFS is mainly dominatedby creeping bent (Agrostis stolonifera) (48 ) velvet grass(Holcus lanatus) (27 ) hair grass (Eleocharis acicularis)(11 ) meadow grass (Poa) (10 ) and couch grass (Elymusrepens) (7 ) The LFS is dominated by grasses and rushesincluding common couch grass (Elytrigia repens) (80 )sharp-flowered rush (Juncus acutiflorus) (12 ) and curleddock (Rumex crispus) (4 ) The mean annual rainfall of thestudy area is 756 mm (based on data from the Dublin Air-port station 50 km north of our study site data source MetEacuteireann)

22 Greenhouse gas measurements

The fluxes of the greenhouse gases CO2 and N2O were mea-sured using closed chambers The chambers were made from16 cm diameter acrylic tubing cut to a length of 23 cm witha flat cap of similar material fitted over the top with an open

base In order to increase the opacity and reflectivity of thechambers they were painted dark grey inside and outsideand additionally their outer surfaces were wrapped in sil-ver duct tape The chambers were placed on top of collars(sim 16 cm diameter) with rubber lips around the top whichwere inserted into the soil to 5 cm depth The purpose of therubber lip was to create a secure seal between the chamberand the collar during measurements After establishment ofthe LFS and SFS sites in September 2013 six collars wereinstalled at each site Individual collars were approximately6 m apart and the array staggered along two parallel transectsThere was a sim 12 m buffer zone between the two sites Dur-ing the period when the sites were inundated the same cham-ber was used for sampling but was inserted into a styrofoamsupport enabling it to be balanced on top of the collar whilefloating on the water surface The collars were also extendedto a height of 20 cm by fitting another similar 15 cm longtube just before the onset of the flooding period To preventthe chamber from being displaced by wind during samplingfour thin rods were fixed into the ground around each sam-pling point to maintain the chamber in position Sampling ofgases was generally undertaken two to four times each monthbut less frequently in the winter months

Measurements of the CO2 and N2O concentration insidethe chamber were made using a photoacoustic gas analyser(PAS) (Innova 1412 Denmark) connected to the chamberusing Teflon tubing The tubes were 6 m long with a 4 mminner diameter and the inlet and outlet of the PAS connectedto two ports on the top of the chamber For sampling thechamber was placed over the collar for between 5 and 6 minduring which time the gas concentration was analysed five toseven times to complete one sample Fluxes of CO2 and N2O(mg mminus2 hminus1) were calculated using

F = (1C1t)(VA)

where 1C1t is the rate of change in gas concentrationinside the chamber during the chamber placement periodwhich was calculated by fitting a best-fit linear regressionline to the flux data versus time V is the chamber volume(4069times 10minus3 m3) and A is the area bounded by the cham-ber (0016 m2) Fluxes of CO2 and N2O were computed iflinear regressions produced r2 gt 090 (P lt 005) for CO2and r2 gt 070 (P lt 005) for N2O

Annual CO2 and N2O emissions for 18 February 2014ndash17 February 2015 and 1 May 2014ndash30 April 2015 were com-puted for the SFS and LFS by linear interpolation of fluxesfor each sampling date The area under the curve was cal-culated using the trapezoid rule by integrating the area foreach 12-month period To estimate and compare the contribu-tion of CO2 and N2O fluxes to the global warming potential(GWP) N2O was converted to CO2 equivalents by multiply-ing it by 298 (IPCC 2007)

Values of Q10 for CO2 emissions were computed forthe SFS and LFS using the equation Q10= exp (slope middot 10)where the slope was derived from the regression coefficient

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2614 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 1 Aerial photograph of part of East Coast Nature Reserve at Blackditch Wood County Wicklow showing the study sites and thedirection in which water flows during flooding Inset shows map of Ireland with site location indicated

of the exponential equation fitted to the CO2 flux and tem-perature data

23 Environmental measurements

In addition to the flux measurements other environmen-tal variables that could potentially influence the GHGfluxes were also measured A weather station located aboutsim 100 m from the locality where measurements were madecomprised sensors for air temperature (RHT3nl-CA) humid-ity (RHT3nl-CA) solar radiation (PYRPA-03) and rainfall(RG2+WS-CA) Average air temperature and cumulativerainfall were recorded at 2 m height every 5 and 60 minrespectively Soil moisture content and temperature weremeasured adjacent to the collarschambers using a handheldThetaProbe (Delta-T Devices Ltd Cambridge UK) eachtime gas sampling was performed During the flooding pe-riod the depth of standing water (WD) was measured adja-cent to the gas sampling points using a graduated woodenruler Redox potential was measured from each collar usinga portable Hanna redox meter (HI9125 Hanna Instruments)with a 10 cm redox electrode Redox potential was measuredfrom the soil surface except when the LFS was flooded abovea height of 10 cm in which case the measurements were ac-quired from the surface of water Complete insertion of theelectrode to the top soil was avoided to prevent the uncertainimpact of the intrusion of water into the electrode throughthe rim at the top during sampling

24 Biomass determinations

In the summer of 2014 the aboveground plant biomass perunit area was estimated from eight quadrats (45times 45 cm) ateach site by clipping the vegetation to approximately ground

level Dry biomass was determined after oven drying thefresh samples at 80 C for 72 h

25 Soil sampling for physical and chemical analysis

Sampling of soil from the two sites (n= 6 for each site) foranalysis of its physicochemical properties was carried out inJuly 2015 and the samples were then air dried and sieved(2 mm) before analysis Soil texture was measured using thepipette method (Gee and Bauder 1986) by first removing theorganic content using hydrogen peroxide and then dispersingthe samples with sodium hexametaphosphate Particles wereclassified as sand (0063ndash2 mm) silt (0002ndash0063 mm) andclay (lt 0002 mm) Bulk density was determined by dryingeach soil sample in a soil core (each 5 cm diametertimes 7 cmheight) at 105 C for 48 h The density was calculated bydividing the dried soil mass by the core volume Soil pHwas determined on 10 g soil dispersed in 20 mL of deionisedwater after 10 min equilibration a reading was taken usinga pH probemeter (Thermo Fisher Scientific Inc WalthamMichigan USA) while stirring the suspension Total car-bon (TC) and nitrogen (TN) were determined through com-bustion of finely ground (0105 mm sieve size) soil using aLECO TruSpec carbonndashnitrogen analyser (TruSpecreg LECOCorporation Michigan USA) Analysis of TC TN and pHwere performed for samples taken every 5 cm from a 25 cmprofile (five samples per sampling point)

Sampling for analysis of NO3ndashN and NH4ndashN (n= 4ndash6 foreach site) was carried out on six occasions from 5 cm depthand an extraction performed by mixing 10 g fresh soil with2 M KCl After shaking for 1 h the KCl extracts were fil-tered and stored in the freezer until analysis NO3ndashN andNH4ndashN were determined from the extracts using an ion anal-

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2615

yser (Lachat QuikChemreg 5600 Lindburgh Drive LovelandColorado USA)

26 Microbial biomass

Microbial biomass C (MBC) and N (MBN) were determinedusing the chloroform fumigation extraction method (Vance etal 1987) 10 g samples of fresh soil (n= 4ndash6 for each site)were fumigated in a desiccator with 20 mL ethanol-free chlo-roform for 72 h and then extracted with 05 M K2SO4 Iden-tical numbers of subsamples were extracted with the samesolution but without fumigation the day after sampling Su-pernatants from both fumigated and non-fumigated sampleswere filtered through Whatman no 1 filter paper and storedin the freezer until analysis Organic carbon and total nitro-gen in the filtrate were analysed using a TOCTN analyser(Shimadzu Japan) Estimates of MBC and MBN were de-rived by calculating the difference between the results ofthe corresponding fumigated and non-fumigated analysisdivided by the extraction efficiency factor Factors of 045(Vance et al 1987) and 054 (Brookes et al 1985) were usedfor MBC and MBN respectively to account for uncompletedextraction of C and N in the microbial cell walls (Jonasson etal 1996)

27 Microbial activity

Soil enzymemicrobial activities were measured from sam-ples (n= 4ndash6 for each site) collected on six sampling occa-sions (March August and October 2014 March May andJuly 2015) All soil samples were sieved through a 2 mmsieve and analysed in triplicate

Beta-glucosidase (BG) was determined using the methoddescribed by Eivazi and Tabatabai (1988) After placing 1 gof soil in a 50 mL flask 025 mL toluene 4 mL of modi-fied universal buffer (pH 60) and 1 mL β-D-glucoside and p-nitrophenyl-α solutions (PNG) were added sequentially andmixed by swirling Samples were then incubated at 37 Cfor 1 h following which 1 mL of 05 M CaCl2 and 4 mL of01 M of tris(hydroxymethyl)aminomethane (pH 12) wereadded to halt further reactions The supernatants were fil-tered and the absorbance of the filtrate measured at 410 nmusing a spectrophotometer (Beckman Coulter DU 530 UVndashvis spectrophotometer) For control samples the same pro-cedure was followed except that PNG was added just beforefiltering the soil suspension instead of adding it at the begin-ning

Protease activity (PRO) was determined as described byKandeler et al (1999) Sodium caseinate (5 mL) solution wasadded to 1 g soil and the samples were incubated at 50 C for2 h and then filtered after adding 5 mL of trichloroacetic acidsolution Alkali and FolinndashCiocalteursquos reagents were addedto the filtrates before protease activity was determined col-orimetrically at 700 nm

To determine nitrate reductase (NR) activity 4 mL of 2 4-dinitrophenol solution and 1 mL of KNO3 were added to 5 gsamples of soil After incubation at 25 C for 24 h 10 mL 4 MKCl was added and filtered To 5 mL of the filtrate NH4Clbuffer (pH 85) sulfanilamide reagent was added and the ac-tivity of the enzyme NR was measured colorimetrically at520 nm

Total microbial activity was assayed using fluorescein di-acetate (FDA) based on the method described by Schnuumlrerand Rosswall (1982) later modified by Green et al (2006)Sodium phosphate buffer (pH 76) and FDA lipase substratewere added to flasks containing 1 g samples of soil and incu-bated for 3 h at 37 C The fluorescein content in the filteredsample was measured at 490 nm

28 Statistical analysis

All statistical analyses were performed using Minitab 16 Allthe values reported are means of three to six replicates andstandard errors were included when required To investigatethe effects of flooding we tested for significant differencesbetween the two sites (ie the LFS and SFS) for the peri-odspart periods (A B C D and E) defined on the basis ofthe state of each site in terms of waterlogging over the studyperiod This was carried out by applying analysis of vari-ance (ANOVA) for each flux soil enzymatic activity TCTN mineral N and MB Functional relationships between po-tential environmental drivers and the fluxes of CO2 and N2Owere performed assessed using linear or exponential regres-sion models Multiple regression analysis was used to deter-mine the relative contribution of more than one independentenvironmental driver to CO2 and N2O fluxes Normality andhomogeneity of the variance of all the models were checkedvisually from residual versus fitted plots and when neces-sary either square-root or log transformations applied Dif-ferences were considered statistically significant if P lt 005unless otherwise mentioned

3 Results

31 Soil characteristics

The relative proportion of clay is higher at the LFS but bothsites have sandy loam soil textures Soil pH was higher at theSFS than the LFS (Table 1) Porosity at the LFS was greaterthan at the SFS The soil C and N concentrations were sig-nificantly higher (P lt 0001) at the LFS site with the great-est difference in the upper soil layers At both sites C andN decreased with soil depth but more gradually at the SFS(Fig 2a b)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2616 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 2 Profile of (a) total carbon and (b) total nitrogen in the LFS and SFS Mean values (n= 6) are for each level within a profile andthe horizontal bars represent the standard errors

Table 1 Soil properties (0ndash10 cm) in the LFS and SFS Values aremeanplusmnSE

Soil properties LFS SFS

Sand () 6440plusmn 072 6900plusmn 095Clay () 1600plusmn 044 650plusmn 067Bulk density (g cmminus3) 073plusmn 008 102plusmn 004Porosity 073plusmn 008 061plusmn 004pH 479plusmn 011 528plusmn 021Total carbon (g kgminus1) 20470plusmn 1190 12000plusmn 650Total nitrogen (g kgminus1) 3850plusmn 652 1070plusmn 070C-to-N ratio 679plusmn 076 1095plusmn 015

32 Rainfall water depth redox potential and airtemperature

Depending on the timing and duration of flooding the fol-lowing periods could be identified as indicated in Fig 3 pe-riod A was characterized by LFS and SFS flooding period Bby LFS flooding period C by neither LFS nor SFS flood-ing period D by LFS flooding and period E by neither LFSnor SFS flooding Thus there was no equivalent of period Aduring 20142015 (ie the SFS was not flooded)

Air temperature followed the seasonal pattern typical forthis latitude (Fig 3a) with the highest values during the sum-mer months (JunendashAugust) and the lowest during the win-ter months (DecemberndashFebruary) The values ranged from ahigh of 23 C to a low of minus39 C with an average of 786 Cover the 2-year study period Rainfall was very variable overthe study period (Fig 3b) with some of the highest rainfallamounts occurring during the warmer months However the

number of days with rainfall was higher in the cooler pe-riod The mean annual rainfall (866 mm) at our site duringthe study period was considerably above the average valuefor the past 30 years (756 mm) measured at the Dublin Air-port station

Standing water was only present at the SFS site forsim 1 month (in the winter of 2014) and reached 5 cmdepth (Fig 3c) In contrast standing water was present forsim 6 months at the LFS site in 2015 and reached a depth ofsim 25 cm in 2014 Whilst standing water at the LFS site waslargely associated with the winter periods it also extendedinto the spring and autumn periods in 20142015 (Fig 3c)

Variations in soil redox potential broadly correlated withthe periods of standing water with the lowest values occur-ring in the LFS site in April 2014 and March 2015 (Fig 3d)reflecting the reducing conditions associated with longer pe-riods of inundation In contrast oxidising conditions werealways found at the SFS site and these were consistentlyhigher (more positive) than those at the LFS (Fig 3d) Mea-surements made from the water column when the LFS wasflooded to gt 10 cm depth indicated that this was always oxi-dising and had a low but positive redox potential (Fig 3d)

33 Seasonal variations in CO2 and N2O fluxes

The results are presented with reference to the periods AndashE which represent the state of the sites in relation to wa-ter availability The CO2 (Fig 4a) emissions showed markedseasonal variation with the highest CO2 values during thesummer months (C E) at both sites which correlated withthe lowest soil moisture values (Fig 4c) and the highest soiltemperatures (Fig 4d) The highest CO2 emissions were ob-served from the LFS site during each period (C E) in each

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2617

Figure 3 Variations in (a) daily air temperature (b) daily rainfall(c) average water depth above the soil surface and (d) redox poten-tial at the study sites Open circles represent redox values from theground surface of the SFS black filled circles represent values fromthe soil surface of the LFS and grey filled circles show values fromthe water surface while the LFS was flooded to greater than 10 cmdepth Period A LFS and SFS flooding period B LFS floodingperiod C neither LFS nor SFS flooding period D LFS floodingperiod E neither LFS nor SFS flooding Thus there was no equiv-alent of period A during late 2014ndashearly 2015

year and values were higher in 2015 compared to 2014 par-ticularly for the SFS site (Fig 4a) The lowest emissions wereobserved during the winter months of both years (A B D)when there were no significant differences (P = 0768) be-tween the LFS and SFS sites even though there was standingwater at the LFS site (Fig 3c)

In period A and the first part of period B CO2 emissionsare similar for the SFS and LFS Towards the end of pe-riod B the emissions are higher from the LFS corresponding

to the time when the water depth at the LFS was decreas-ing (Fig 3c) At the LFS CO2 emissions during period Cranged from 1694 to 49818 mg CO2ndashC mminus2 hminus1 whereasat the SFS the values ranged from 1668 to 429 mg CO2ndashC mminus2 hminus1 The initial part of period C is characterised bylow and similar values of CO2 emissions for each site thatare not appreciably different (P = 0956) from those of thepreceding period B Towards the middle of period C CO2emissions decline and then increase again corresponding to adecrease and then increase in soil moisture For the JunendashSeptember 2014 interval within period C CO2 emissionsfrom the two sites were significantly different (P lt 0034)

In period D similar trends in CO2 emissions with time(Fig 4a) were observed for the two sites despite differencesin soil moisture (Fig 4c) and water depth (Fig 3c) In pe-riod E CO2 emissions ranged from 5332 to 47726 mg CO2ndashC mminus2 hminus1 at the LFS and from 5568 to 34344 mg CO2ndashC mminus2 hminus1 at the SFS As for the corresponding of theprevious year (period C) the difference between the twosites was significant (P = 0013) for part of period E (MayndashAugust 2015)

Based on the relationships between soil respiration andtemperature estimated values for Q10 were 249 and 208 atthe LFS and SFS respectively

Unlike the CO2 fluxes the N2O emissions generallyshowed no discernible pattern between seasons duringthe study period (Fig 4b) and there were no system-atic changes over time No significant differences (P gt005) in N2O fluxes were observed between the LFS andSFS in any of the five periods In the first wetter peri-ods (periods A and B combined) N2O fluxes ranged fromminus0210 to 0319 mg N2OndashN mminus2 hminus1 and from minus0503 to0364 mg N2OndashN mminus2 hminus1 at the LFS and SFS respectivelyIn period D the N2O fluxes from the LFS showed highervariability than those from the SFS with values betweenminus0203 and 0695 mg N2OndashN mminus2 hminus1 for the LFS and be-tween minus0307 and 0206 mg N2OndashN mminus2 hminus1 for the SFSOverall N2O fluxes were generally higher from the LFSthan from the SFS during period D In the first growingseason (period C) the SFS showed consistently higher andmore positive N2O fluxes the maximum emission recorded(0651 mg N2OndashN mminus2 hminus1) was however from the LFS atthe end of July N2O fluxes less than zero suggest uptake ofN2O which was more common at the SFS site

Annual CO2 emissions were 818 and 876 (mean 847)and 1124 and 115 (mean 1137) Mg CO2ndashC haminus1 yrminus1 fromthe SFS and LFS respectively with values for the LFS 14times higher than that from the SFS For the same yearsthe annual N2O emissions were 13 times higher (709 and549 kg N2OndashN haminus1 yrminus1) from the LFS (mean 629) com-pared to the SFS (547 and 362 kg N2OndashN haminus1 yrminus1 mean454)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2618 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 4 Seasonal variations in (a) CO2 fluxes (b) N2O fluxes (c) soil moisture and (d) soil temperature in each site Periods as defined inFig 3

34 Relationship between CO2 and N2O fluxes andenvironmental parameters

Independent testing of the various environmental variablesshowed that soil temperature soil water content redox po-tential and for the LFS water depth were significantly cor-related with the CO2 emissions Significant positive corre-lations were found between CO2 emissions and soil tem-perature for both the SFS (R2

= 044 P lt 0001) and LFS(R2= 056 P lt 0001) with a stronger relationship at the

LFS (Fig 5a) A significant negative linear relationship be-tween soil moisture and CO2 emission was found at the

LFS (R2= 052 P lt 0001) and the SFS (R2

= 054 P lt0001) (Fig 5b) Correlations between CO2 emissions andredox potential were low with R2 values of 025 (LFS) and016 (SFS) respectively but significant at P lt 005 CO2emissions at the LFS were exponentially correlated with wa-ter depth (R2

= 045 P lt 0001) (Fig 5c) Combining soiltemperature and soil water content in multiple regressionanalyses with the CO2 fluxes only resulted in a small increasein explanatory power to sim 58 and 66 at the SFS and LFSrespectively but including redox potential had no impact onthe explanatory power

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

2612 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

sibility of these ecosystems being rapidly inundated withfreshwater following extreme precipitation events or ex-tended periods of rainfall Episodic flooding is a commonfeature of these ecosystems with potential sources of wa-ter from inland or marine incursions although inundationfrom freshwater sources is probably the more common(Doornkamp 1998) Many assessments of GHG emissionsand an understanding of the factors that control them arebased on information from ecosystems less prone to floodingepisodes These data may not be directly applicable to coastalareas or floodplains where both the volumes involved and thefrequency and duration of water incursions can be high Forexample soil water availability can affect the emissions ofCO2 and N2O by influencing the rates of C and N mineral-isation (Alongi et al 1999 Noe et al 2013) but it is lessclear how variations in standing water levels influence theseprocesses although this likely depends on the extent and theduration of inundation (Lewis et al 2014) Previous studiesprovide contrasting evidence as to whether soil inundationenhances or impedes organic matter mineralisation Wilsonet al (2011) and Kim et al (2015) showed that the rate of or-ganic matter mineralisation can be enhanced after a short hy-droperiod (the period of time the soil area was waterlogged)whereas longer inundation periods suppressed decomposi-tion by limiting oxygen supply (Lewis et al 2014) Flood-ing generally promotes anaerobic conditions by excludingair from the pore spaces in the soil that would reduce min-eralisation of organic matter (Altor and Mitsch 2008) De-spite lower mineralisation rates under these conditions CO2emissions are still possible from anoxic soils (Glatzel et al2004) A number of factors such as the quality and quantityof substrates available for microbial processes temperaturesoil microbial activity and oxidation-reduction potential po-tentially impact on the magnitude of CO2 or N2O exchangein intermittently flooded ecosystems High soil organic mat-ter concentrations in combination with warmer conditions inflooded soil can enhance CO2 production (Oelbermann andSchiff 2008 Kim et al 2015) When associated with suf-ficient oxygen CO2 emissions in wetlands increase as a re-sult of accelerated organic matter decomposition whilst CH4production decreases because of aerobic methane oxidation(Smith et al 2003) Although CH4 is often considered to bethe more significant GHG in wet or flooded ecosystems thismay not be the case where there are only temporary hydrope-riods (Audet et al 2013 Batson et al 2015) where CO2fluxes are still likely to dominate the annual budget Finallythe effect of hydroperiod duration on N2O fluxes has receivedless attention We therefore focussed on N2O and CO2 fluxeswhich are likely to be the more important in many terrestrialecosystems

Denitrification and nitrification are often considered to bethe main mechanisms by which N2O is produced in soil al-though denitrification may be the dominant process for therelease of N2O both under aerobic and anaerobic soil con-ditions (Bateman and Baggs 2005) Denitrification and in-

creased N2O production is enhanced by high soil water con-tents (Machefert and Dise 2004 Maag and Vinther 1996)although the frequency and duration of flooding may alsobe important In riparian systems a shorter flooding dura-tion correlated with the highest N2O emissions which di-minished the longer the period of inundation (Jacinthe et al2012) In addition to water availability temperature is alsorecognised as an important variable in determining the pro-duction and consumption of N2O and CO2 via its effects onthe metabolic activity of microorganisms and plants (David-son and Janssens 2006 Butterbach-Bahl et al 2013 Kirwanet al 2014 Kim et al 2015)

Flooding stimulates changes in the structure of soil mi-crobial communities (Bossio and Scow 1998 Unger et al2009 Wilson et al 2011) and this in turn affects the rate ofdecomposition of organic material (Van Der Heijden et al2008) The extent of any change in microbial biomass (MB)andor microbial populations could also vary with floodingduration (Rinklebe and Langer 2006) For example Wilsonet al (2011) showed significant changes in microbial struc-ture and increases in soil enzymatic activity after short-term(24-day) inundation of floodplain soil Nevertheless an in-crease in microbial activity and GHG emissions as a responseto short-term flooding (SFS) is not always found (Unger etal 2009 Jacinthe 2015) and the timing as well as the fre-quency of flooding events may be important in determin-ing both the microbial community change and the associ-ated GHG emissions Organic substrate availability (via thebreakdown of plant litter or the introduction of organic com-pounds in flowing water) is a key factor regulating the activ-ities of soil microorganisms and the mineralisation and im-mobilisation of carbon and nitrogen (Badiou et al 2011 Liet al 2015) The availability of organic substrates influencesthe activity of carbon-cycling extracellular enzymes and asa result can impact on the rate of CO2 emissions and poten-tially other GHGs (Li et al 2015)

Owing to the episodic nature of flooding events captur-ing their impact on GHG emissions is quite challenging anda possible explanation for the lack of data from field studiesGiven the projected increase in flooding events an evaluationof their impact on GHG emissions is clearly required Theidentification of sites with flooding potential before floodingevents and the establishment of the appropriate measurementprotocols are crucial steps required to capture the dynam-ics of GHG fluxes in response to real flooding events Thiswould enable the capture of the full pattern of GHG fluxesprior to during and after a flooding event In this paperwe assess the impact of freshwater inundation with differ-ent hydroperiods on CO2 and N2O fluxes before during andafter real flooding events in a coastal grassland ecosystemover sim 2 years The main objective was to assess the effectsof short- and long-term flooding on the dynamics of CO2and N2O fluxes and how this impacts on the annual budgetsfor each gas We also evaluate the significance of a number

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2613

of soil physical chemical and biological parameters whichmay influence the fluxes of CO2 and N2O

2 Materials and methods

21 Site description

The study site (53

05prime87primeprime N 604prime07primeprimeW) was situated on alow lying area (sim 0 m asl) of coastal grassland that formspart of the East Coast Nature Reserve (ECNR) a portion ofthe larger coastal wetland complex called the Murrough wet-lands on the east coast of Ireland near Newcastle CountyWicklow (Fig 1) The site is owned and managed by Bird-Watch Ireland predominantly as a habitat for a wide vari-ety of birds The grassland area lies between a long drainageditch that runs in a northndashsouth direction and a shingle beachbordering the Irish Sea In the past 12 years restoration meth-ods including water management low-intensity grazing andcrop planting have been undertaken to maintain the site Thestanding water level varies spatially across the site and fluctu-ates seasonally in response to rainfall and the water-retainingcapacity of the drainage ditch The site experiences near tocomplete saturation with localised flooding in autumn andwinter but drains and dries out in spring and summer Despitebeing located in a coastal area the ditch water is generallycharacterized as being freshwater with the majority sourcedfrom the landward side

Differences in elevation (sim 20ndash25 cm vertical height) overthe site result in variations in hydrological connectivity be-tween different portions of the grassland and the ditch watersystem Initial monitoring of the general area indicated thatthis resulted in spatial hydroperiodic differences The sitewas therefore divided into two areas based on these hydrope-riod characteristics The site on the more elevated groundis characterized by less-frequent and shorter-term floodingwhilst the site located at lower elevations becomes inun-dated seasonally for an extended period during the winter(LFS) The plant community of the SFS is mainly dominatedby creeping bent (Agrostis stolonifera) (48 ) velvet grass(Holcus lanatus) (27 ) hair grass (Eleocharis acicularis)(11 ) meadow grass (Poa) (10 ) and couch grass (Elymusrepens) (7 ) The LFS is dominated by grasses and rushesincluding common couch grass (Elytrigia repens) (80 )sharp-flowered rush (Juncus acutiflorus) (12 ) and curleddock (Rumex crispus) (4 ) The mean annual rainfall of thestudy area is 756 mm (based on data from the Dublin Air-port station 50 km north of our study site data source MetEacuteireann)

22 Greenhouse gas measurements

The fluxes of the greenhouse gases CO2 and N2O were mea-sured using closed chambers The chambers were made from16 cm diameter acrylic tubing cut to a length of 23 cm witha flat cap of similar material fitted over the top with an open

base In order to increase the opacity and reflectivity of thechambers they were painted dark grey inside and outsideand additionally their outer surfaces were wrapped in sil-ver duct tape The chambers were placed on top of collars(sim 16 cm diameter) with rubber lips around the top whichwere inserted into the soil to 5 cm depth The purpose of therubber lip was to create a secure seal between the chamberand the collar during measurements After establishment ofthe LFS and SFS sites in September 2013 six collars wereinstalled at each site Individual collars were approximately6 m apart and the array staggered along two parallel transectsThere was a sim 12 m buffer zone between the two sites Dur-ing the period when the sites were inundated the same cham-ber was used for sampling but was inserted into a styrofoamsupport enabling it to be balanced on top of the collar whilefloating on the water surface The collars were also extendedto a height of 20 cm by fitting another similar 15 cm longtube just before the onset of the flooding period To preventthe chamber from being displaced by wind during samplingfour thin rods were fixed into the ground around each sam-pling point to maintain the chamber in position Sampling ofgases was generally undertaken two to four times each monthbut less frequently in the winter months

Measurements of the CO2 and N2O concentration insidethe chamber were made using a photoacoustic gas analyser(PAS) (Innova 1412 Denmark) connected to the chamberusing Teflon tubing The tubes were 6 m long with a 4 mminner diameter and the inlet and outlet of the PAS connectedto two ports on the top of the chamber For sampling thechamber was placed over the collar for between 5 and 6 minduring which time the gas concentration was analysed five toseven times to complete one sample Fluxes of CO2 and N2O(mg mminus2 hminus1) were calculated using

F = (1C1t)(VA)

where 1C1t is the rate of change in gas concentrationinside the chamber during the chamber placement periodwhich was calculated by fitting a best-fit linear regressionline to the flux data versus time V is the chamber volume(4069times 10minus3 m3) and A is the area bounded by the cham-ber (0016 m2) Fluxes of CO2 and N2O were computed iflinear regressions produced r2 gt 090 (P lt 005) for CO2and r2 gt 070 (P lt 005) for N2O

Annual CO2 and N2O emissions for 18 February 2014ndash17 February 2015 and 1 May 2014ndash30 April 2015 were com-puted for the SFS and LFS by linear interpolation of fluxesfor each sampling date The area under the curve was cal-culated using the trapezoid rule by integrating the area foreach 12-month period To estimate and compare the contribu-tion of CO2 and N2O fluxes to the global warming potential(GWP) N2O was converted to CO2 equivalents by multiply-ing it by 298 (IPCC 2007)

Values of Q10 for CO2 emissions were computed forthe SFS and LFS using the equation Q10= exp (slope middot 10)where the slope was derived from the regression coefficient

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2614 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 1 Aerial photograph of part of East Coast Nature Reserve at Blackditch Wood County Wicklow showing the study sites and thedirection in which water flows during flooding Inset shows map of Ireland with site location indicated

of the exponential equation fitted to the CO2 flux and tem-perature data

23 Environmental measurements

In addition to the flux measurements other environmen-tal variables that could potentially influence the GHGfluxes were also measured A weather station located aboutsim 100 m from the locality where measurements were madecomprised sensors for air temperature (RHT3nl-CA) humid-ity (RHT3nl-CA) solar radiation (PYRPA-03) and rainfall(RG2+WS-CA) Average air temperature and cumulativerainfall were recorded at 2 m height every 5 and 60 minrespectively Soil moisture content and temperature weremeasured adjacent to the collarschambers using a handheldThetaProbe (Delta-T Devices Ltd Cambridge UK) eachtime gas sampling was performed During the flooding pe-riod the depth of standing water (WD) was measured adja-cent to the gas sampling points using a graduated woodenruler Redox potential was measured from each collar usinga portable Hanna redox meter (HI9125 Hanna Instruments)with a 10 cm redox electrode Redox potential was measuredfrom the soil surface except when the LFS was flooded abovea height of 10 cm in which case the measurements were ac-quired from the surface of water Complete insertion of theelectrode to the top soil was avoided to prevent the uncertainimpact of the intrusion of water into the electrode throughthe rim at the top during sampling

24 Biomass determinations

In the summer of 2014 the aboveground plant biomass perunit area was estimated from eight quadrats (45times 45 cm) ateach site by clipping the vegetation to approximately ground

level Dry biomass was determined after oven drying thefresh samples at 80 C for 72 h

25 Soil sampling for physical and chemical analysis

Sampling of soil from the two sites (n= 6 for each site) foranalysis of its physicochemical properties was carried out inJuly 2015 and the samples were then air dried and sieved(2 mm) before analysis Soil texture was measured using thepipette method (Gee and Bauder 1986) by first removing theorganic content using hydrogen peroxide and then dispersingthe samples with sodium hexametaphosphate Particles wereclassified as sand (0063ndash2 mm) silt (0002ndash0063 mm) andclay (lt 0002 mm) Bulk density was determined by dryingeach soil sample in a soil core (each 5 cm diametertimes 7 cmheight) at 105 C for 48 h The density was calculated bydividing the dried soil mass by the core volume Soil pHwas determined on 10 g soil dispersed in 20 mL of deionisedwater after 10 min equilibration a reading was taken usinga pH probemeter (Thermo Fisher Scientific Inc WalthamMichigan USA) while stirring the suspension Total car-bon (TC) and nitrogen (TN) were determined through com-bustion of finely ground (0105 mm sieve size) soil using aLECO TruSpec carbonndashnitrogen analyser (TruSpecreg LECOCorporation Michigan USA) Analysis of TC TN and pHwere performed for samples taken every 5 cm from a 25 cmprofile (five samples per sampling point)

Sampling for analysis of NO3ndashN and NH4ndashN (n= 4ndash6 foreach site) was carried out on six occasions from 5 cm depthand an extraction performed by mixing 10 g fresh soil with2 M KCl After shaking for 1 h the KCl extracts were fil-tered and stored in the freezer until analysis NO3ndashN andNH4ndashN were determined from the extracts using an ion anal-

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2615

yser (Lachat QuikChemreg 5600 Lindburgh Drive LovelandColorado USA)

26 Microbial biomass

Microbial biomass C (MBC) and N (MBN) were determinedusing the chloroform fumigation extraction method (Vance etal 1987) 10 g samples of fresh soil (n= 4ndash6 for each site)were fumigated in a desiccator with 20 mL ethanol-free chlo-roform for 72 h and then extracted with 05 M K2SO4 Iden-tical numbers of subsamples were extracted with the samesolution but without fumigation the day after sampling Su-pernatants from both fumigated and non-fumigated sampleswere filtered through Whatman no 1 filter paper and storedin the freezer until analysis Organic carbon and total nitro-gen in the filtrate were analysed using a TOCTN analyser(Shimadzu Japan) Estimates of MBC and MBN were de-rived by calculating the difference between the results ofthe corresponding fumigated and non-fumigated analysisdivided by the extraction efficiency factor Factors of 045(Vance et al 1987) and 054 (Brookes et al 1985) were usedfor MBC and MBN respectively to account for uncompletedextraction of C and N in the microbial cell walls (Jonasson etal 1996)

27 Microbial activity

Soil enzymemicrobial activities were measured from sam-ples (n= 4ndash6 for each site) collected on six sampling occa-sions (March August and October 2014 March May andJuly 2015) All soil samples were sieved through a 2 mmsieve and analysed in triplicate

Beta-glucosidase (BG) was determined using the methoddescribed by Eivazi and Tabatabai (1988) After placing 1 gof soil in a 50 mL flask 025 mL toluene 4 mL of modi-fied universal buffer (pH 60) and 1 mL β-D-glucoside and p-nitrophenyl-α solutions (PNG) were added sequentially andmixed by swirling Samples were then incubated at 37 Cfor 1 h following which 1 mL of 05 M CaCl2 and 4 mL of01 M of tris(hydroxymethyl)aminomethane (pH 12) wereadded to halt further reactions The supernatants were fil-tered and the absorbance of the filtrate measured at 410 nmusing a spectrophotometer (Beckman Coulter DU 530 UVndashvis spectrophotometer) For control samples the same pro-cedure was followed except that PNG was added just beforefiltering the soil suspension instead of adding it at the begin-ning

Protease activity (PRO) was determined as described byKandeler et al (1999) Sodium caseinate (5 mL) solution wasadded to 1 g soil and the samples were incubated at 50 C for2 h and then filtered after adding 5 mL of trichloroacetic acidsolution Alkali and FolinndashCiocalteursquos reagents were addedto the filtrates before protease activity was determined col-orimetrically at 700 nm

To determine nitrate reductase (NR) activity 4 mL of 2 4-dinitrophenol solution and 1 mL of KNO3 were added to 5 gsamples of soil After incubation at 25 C for 24 h 10 mL 4 MKCl was added and filtered To 5 mL of the filtrate NH4Clbuffer (pH 85) sulfanilamide reagent was added and the ac-tivity of the enzyme NR was measured colorimetrically at520 nm

Total microbial activity was assayed using fluorescein di-acetate (FDA) based on the method described by Schnuumlrerand Rosswall (1982) later modified by Green et al (2006)Sodium phosphate buffer (pH 76) and FDA lipase substratewere added to flasks containing 1 g samples of soil and incu-bated for 3 h at 37 C The fluorescein content in the filteredsample was measured at 490 nm

28 Statistical analysis

All statistical analyses were performed using Minitab 16 Allthe values reported are means of three to six replicates andstandard errors were included when required To investigatethe effects of flooding we tested for significant differencesbetween the two sites (ie the LFS and SFS) for the peri-odspart periods (A B C D and E) defined on the basis ofthe state of each site in terms of waterlogging over the studyperiod This was carried out by applying analysis of vari-ance (ANOVA) for each flux soil enzymatic activity TCTN mineral N and MB Functional relationships between po-tential environmental drivers and the fluxes of CO2 and N2Owere performed assessed using linear or exponential regres-sion models Multiple regression analysis was used to deter-mine the relative contribution of more than one independentenvironmental driver to CO2 and N2O fluxes Normality andhomogeneity of the variance of all the models were checkedvisually from residual versus fitted plots and when neces-sary either square-root or log transformations applied Dif-ferences were considered statistically significant if P lt 005unless otherwise mentioned

3 Results

31 Soil characteristics

The relative proportion of clay is higher at the LFS but bothsites have sandy loam soil textures Soil pH was higher at theSFS than the LFS (Table 1) Porosity at the LFS was greaterthan at the SFS The soil C and N concentrations were sig-nificantly higher (P lt 0001) at the LFS site with the great-est difference in the upper soil layers At both sites C andN decreased with soil depth but more gradually at the SFS(Fig 2a b)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2616 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 2 Profile of (a) total carbon and (b) total nitrogen in the LFS and SFS Mean values (n= 6) are for each level within a profile andthe horizontal bars represent the standard errors

Table 1 Soil properties (0ndash10 cm) in the LFS and SFS Values aremeanplusmnSE

Soil properties LFS SFS

Sand () 6440plusmn 072 6900plusmn 095Clay () 1600plusmn 044 650plusmn 067Bulk density (g cmminus3) 073plusmn 008 102plusmn 004Porosity 073plusmn 008 061plusmn 004pH 479plusmn 011 528plusmn 021Total carbon (g kgminus1) 20470plusmn 1190 12000plusmn 650Total nitrogen (g kgminus1) 3850plusmn 652 1070plusmn 070C-to-N ratio 679plusmn 076 1095plusmn 015

32 Rainfall water depth redox potential and airtemperature

Depending on the timing and duration of flooding the fol-lowing periods could be identified as indicated in Fig 3 pe-riod A was characterized by LFS and SFS flooding period Bby LFS flooding period C by neither LFS nor SFS flood-ing period D by LFS flooding and period E by neither LFSnor SFS flooding Thus there was no equivalent of period Aduring 20142015 (ie the SFS was not flooded)

Air temperature followed the seasonal pattern typical forthis latitude (Fig 3a) with the highest values during the sum-mer months (JunendashAugust) and the lowest during the win-ter months (DecemberndashFebruary) The values ranged from ahigh of 23 C to a low of minus39 C with an average of 786 Cover the 2-year study period Rainfall was very variable overthe study period (Fig 3b) with some of the highest rainfallamounts occurring during the warmer months However the

number of days with rainfall was higher in the cooler pe-riod The mean annual rainfall (866 mm) at our site duringthe study period was considerably above the average valuefor the past 30 years (756 mm) measured at the Dublin Air-port station

Standing water was only present at the SFS site forsim 1 month (in the winter of 2014) and reached 5 cmdepth (Fig 3c) In contrast standing water was present forsim 6 months at the LFS site in 2015 and reached a depth ofsim 25 cm in 2014 Whilst standing water at the LFS site waslargely associated with the winter periods it also extendedinto the spring and autumn periods in 20142015 (Fig 3c)

Variations in soil redox potential broadly correlated withthe periods of standing water with the lowest values occur-ring in the LFS site in April 2014 and March 2015 (Fig 3d)reflecting the reducing conditions associated with longer pe-riods of inundation In contrast oxidising conditions werealways found at the SFS site and these were consistentlyhigher (more positive) than those at the LFS (Fig 3d) Mea-surements made from the water column when the LFS wasflooded to gt 10 cm depth indicated that this was always oxi-dising and had a low but positive redox potential (Fig 3d)

33 Seasonal variations in CO2 and N2O fluxes

The results are presented with reference to the periods AndashE which represent the state of the sites in relation to wa-ter availability The CO2 (Fig 4a) emissions showed markedseasonal variation with the highest CO2 values during thesummer months (C E) at both sites which correlated withthe lowest soil moisture values (Fig 4c) and the highest soiltemperatures (Fig 4d) The highest CO2 emissions were ob-served from the LFS site during each period (C E) in each

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2617

Figure 3 Variations in (a) daily air temperature (b) daily rainfall(c) average water depth above the soil surface and (d) redox poten-tial at the study sites Open circles represent redox values from theground surface of the SFS black filled circles represent values fromthe soil surface of the LFS and grey filled circles show values fromthe water surface while the LFS was flooded to greater than 10 cmdepth Period A LFS and SFS flooding period B LFS floodingperiod C neither LFS nor SFS flooding period D LFS floodingperiod E neither LFS nor SFS flooding Thus there was no equiv-alent of period A during late 2014ndashearly 2015

year and values were higher in 2015 compared to 2014 par-ticularly for the SFS site (Fig 4a) The lowest emissions wereobserved during the winter months of both years (A B D)when there were no significant differences (P = 0768) be-tween the LFS and SFS sites even though there was standingwater at the LFS site (Fig 3c)

In period A and the first part of period B CO2 emissionsare similar for the SFS and LFS Towards the end of pe-riod B the emissions are higher from the LFS corresponding

to the time when the water depth at the LFS was decreas-ing (Fig 3c) At the LFS CO2 emissions during period Cranged from 1694 to 49818 mg CO2ndashC mminus2 hminus1 whereasat the SFS the values ranged from 1668 to 429 mg CO2ndashC mminus2 hminus1 The initial part of period C is characterised bylow and similar values of CO2 emissions for each site thatare not appreciably different (P = 0956) from those of thepreceding period B Towards the middle of period C CO2emissions decline and then increase again corresponding to adecrease and then increase in soil moisture For the JunendashSeptember 2014 interval within period C CO2 emissionsfrom the two sites were significantly different (P lt 0034)

In period D similar trends in CO2 emissions with time(Fig 4a) were observed for the two sites despite differencesin soil moisture (Fig 4c) and water depth (Fig 3c) In pe-riod E CO2 emissions ranged from 5332 to 47726 mg CO2ndashC mminus2 hminus1 at the LFS and from 5568 to 34344 mg CO2ndashC mminus2 hminus1 at the SFS As for the corresponding of theprevious year (period C) the difference between the twosites was significant (P = 0013) for part of period E (MayndashAugust 2015)

Based on the relationships between soil respiration andtemperature estimated values for Q10 were 249 and 208 atthe LFS and SFS respectively

Unlike the CO2 fluxes the N2O emissions generallyshowed no discernible pattern between seasons duringthe study period (Fig 4b) and there were no system-atic changes over time No significant differences (P gt005) in N2O fluxes were observed between the LFS andSFS in any of the five periods In the first wetter peri-ods (periods A and B combined) N2O fluxes ranged fromminus0210 to 0319 mg N2OndashN mminus2 hminus1 and from minus0503 to0364 mg N2OndashN mminus2 hminus1 at the LFS and SFS respectivelyIn period D the N2O fluxes from the LFS showed highervariability than those from the SFS with values betweenminus0203 and 0695 mg N2OndashN mminus2 hminus1 for the LFS and be-tween minus0307 and 0206 mg N2OndashN mminus2 hminus1 for the SFSOverall N2O fluxes were generally higher from the LFSthan from the SFS during period D In the first growingseason (period C) the SFS showed consistently higher andmore positive N2O fluxes the maximum emission recorded(0651 mg N2OndashN mminus2 hminus1) was however from the LFS atthe end of July N2O fluxes less than zero suggest uptake ofN2O which was more common at the SFS site

Annual CO2 emissions were 818 and 876 (mean 847)and 1124 and 115 (mean 1137) Mg CO2ndashC haminus1 yrminus1 fromthe SFS and LFS respectively with values for the LFS 14times higher than that from the SFS For the same yearsthe annual N2O emissions were 13 times higher (709 and549 kg N2OndashN haminus1 yrminus1) from the LFS (mean 629) com-pared to the SFS (547 and 362 kg N2OndashN haminus1 yrminus1 mean454)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2618 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 4 Seasonal variations in (a) CO2 fluxes (b) N2O fluxes (c) soil moisture and (d) soil temperature in each site Periods as defined inFig 3

34 Relationship between CO2 and N2O fluxes andenvironmental parameters

Independent testing of the various environmental variablesshowed that soil temperature soil water content redox po-tential and for the LFS water depth were significantly cor-related with the CO2 emissions Significant positive corre-lations were found between CO2 emissions and soil tem-perature for both the SFS (R2

= 044 P lt 0001) and LFS(R2= 056 P lt 0001) with a stronger relationship at the

LFS (Fig 5a) A significant negative linear relationship be-tween soil moisture and CO2 emission was found at the

LFS (R2= 052 P lt 0001) and the SFS (R2

= 054 P lt0001) (Fig 5b) Correlations between CO2 emissions andredox potential were low with R2 values of 025 (LFS) and016 (SFS) respectively but significant at P lt 005 CO2emissions at the LFS were exponentially correlated with wa-ter depth (R2

= 045 P lt 0001) (Fig 5c) Combining soiltemperature and soil water content in multiple regressionanalyses with the CO2 fluxes only resulted in a small increasein explanatory power to sim 58 and 66 at the SFS and LFSrespectively but including redox potential had no impact onthe explanatory power

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2613

of soil physical chemical and biological parameters whichmay influence the fluxes of CO2 and N2O

2 Materials and methods

21 Site description

The study site (53

05prime87primeprime N 604prime07primeprimeW) was situated on alow lying area (sim 0 m asl) of coastal grassland that formspart of the East Coast Nature Reserve (ECNR) a portion ofthe larger coastal wetland complex called the Murrough wet-lands on the east coast of Ireland near Newcastle CountyWicklow (Fig 1) The site is owned and managed by Bird-Watch Ireland predominantly as a habitat for a wide vari-ety of birds The grassland area lies between a long drainageditch that runs in a northndashsouth direction and a shingle beachbordering the Irish Sea In the past 12 years restoration meth-ods including water management low-intensity grazing andcrop planting have been undertaken to maintain the site Thestanding water level varies spatially across the site and fluctu-ates seasonally in response to rainfall and the water-retainingcapacity of the drainage ditch The site experiences near tocomplete saturation with localised flooding in autumn andwinter but drains and dries out in spring and summer Despitebeing located in a coastal area the ditch water is generallycharacterized as being freshwater with the majority sourcedfrom the landward side

Differences in elevation (sim 20ndash25 cm vertical height) overthe site result in variations in hydrological connectivity be-tween different portions of the grassland and the ditch watersystem Initial monitoring of the general area indicated thatthis resulted in spatial hydroperiodic differences The sitewas therefore divided into two areas based on these hydrope-riod characteristics The site on the more elevated groundis characterized by less-frequent and shorter-term floodingwhilst the site located at lower elevations becomes inun-dated seasonally for an extended period during the winter(LFS) The plant community of the SFS is mainly dominatedby creeping bent (Agrostis stolonifera) (48 ) velvet grass(Holcus lanatus) (27 ) hair grass (Eleocharis acicularis)(11 ) meadow grass (Poa) (10 ) and couch grass (Elymusrepens) (7 ) The LFS is dominated by grasses and rushesincluding common couch grass (Elytrigia repens) (80 )sharp-flowered rush (Juncus acutiflorus) (12 ) and curleddock (Rumex crispus) (4 ) The mean annual rainfall of thestudy area is 756 mm (based on data from the Dublin Air-port station 50 km north of our study site data source MetEacuteireann)

22 Greenhouse gas measurements

The fluxes of the greenhouse gases CO2 and N2O were mea-sured using closed chambers The chambers were made from16 cm diameter acrylic tubing cut to a length of 23 cm witha flat cap of similar material fitted over the top with an open

base In order to increase the opacity and reflectivity of thechambers they were painted dark grey inside and outsideand additionally their outer surfaces were wrapped in sil-ver duct tape The chambers were placed on top of collars(sim 16 cm diameter) with rubber lips around the top whichwere inserted into the soil to 5 cm depth The purpose of therubber lip was to create a secure seal between the chamberand the collar during measurements After establishment ofthe LFS and SFS sites in September 2013 six collars wereinstalled at each site Individual collars were approximately6 m apart and the array staggered along two parallel transectsThere was a sim 12 m buffer zone between the two sites Dur-ing the period when the sites were inundated the same cham-ber was used for sampling but was inserted into a styrofoamsupport enabling it to be balanced on top of the collar whilefloating on the water surface The collars were also extendedto a height of 20 cm by fitting another similar 15 cm longtube just before the onset of the flooding period To preventthe chamber from being displaced by wind during samplingfour thin rods were fixed into the ground around each sam-pling point to maintain the chamber in position Sampling ofgases was generally undertaken two to four times each monthbut less frequently in the winter months

Measurements of the CO2 and N2O concentration insidethe chamber were made using a photoacoustic gas analyser(PAS) (Innova 1412 Denmark) connected to the chamberusing Teflon tubing The tubes were 6 m long with a 4 mminner diameter and the inlet and outlet of the PAS connectedto two ports on the top of the chamber For sampling thechamber was placed over the collar for between 5 and 6 minduring which time the gas concentration was analysed five toseven times to complete one sample Fluxes of CO2 and N2O(mg mminus2 hminus1) were calculated using

F = (1C1t)(VA)

where 1C1t is the rate of change in gas concentrationinside the chamber during the chamber placement periodwhich was calculated by fitting a best-fit linear regressionline to the flux data versus time V is the chamber volume(4069times 10minus3 m3) and A is the area bounded by the cham-ber (0016 m2) Fluxes of CO2 and N2O were computed iflinear regressions produced r2 gt 090 (P lt 005) for CO2and r2 gt 070 (P lt 005) for N2O

Annual CO2 and N2O emissions for 18 February 2014ndash17 February 2015 and 1 May 2014ndash30 April 2015 were com-puted for the SFS and LFS by linear interpolation of fluxesfor each sampling date The area under the curve was cal-culated using the trapezoid rule by integrating the area foreach 12-month period To estimate and compare the contribu-tion of CO2 and N2O fluxes to the global warming potential(GWP) N2O was converted to CO2 equivalents by multiply-ing it by 298 (IPCC 2007)

Values of Q10 for CO2 emissions were computed forthe SFS and LFS using the equation Q10= exp (slope middot 10)where the slope was derived from the regression coefficient

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2614 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 1 Aerial photograph of part of East Coast Nature Reserve at Blackditch Wood County Wicklow showing the study sites and thedirection in which water flows during flooding Inset shows map of Ireland with site location indicated

of the exponential equation fitted to the CO2 flux and tem-perature data

23 Environmental measurements

In addition to the flux measurements other environmen-tal variables that could potentially influence the GHGfluxes were also measured A weather station located aboutsim 100 m from the locality where measurements were madecomprised sensors for air temperature (RHT3nl-CA) humid-ity (RHT3nl-CA) solar radiation (PYRPA-03) and rainfall(RG2+WS-CA) Average air temperature and cumulativerainfall were recorded at 2 m height every 5 and 60 minrespectively Soil moisture content and temperature weremeasured adjacent to the collarschambers using a handheldThetaProbe (Delta-T Devices Ltd Cambridge UK) eachtime gas sampling was performed During the flooding pe-riod the depth of standing water (WD) was measured adja-cent to the gas sampling points using a graduated woodenruler Redox potential was measured from each collar usinga portable Hanna redox meter (HI9125 Hanna Instruments)with a 10 cm redox electrode Redox potential was measuredfrom the soil surface except when the LFS was flooded abovea height of 10 cm in which case the measurements were ac-quired from the surface of water Complete insertion of theelectrode to the top soil was avoided to prevent the uncertainimpact of the intrusion of water into the electrode throughthe rim at the top during sampling

24 Biomass determinations

In the summer of 2014 the aboveground plant biomass perunit area was estimated from eight quadrats (45times 45 cm) ateach site by clipping the vegetation to approximately ground

level Dry biomass was determined after oven drying thefresh samples at 80 C for 72 h

25 Soil sampling for physical and chemical analysis

Sampling of soil from the two sites (n= 6 for each site) foranalysis of its physicochemical properties was carried out inJuly 2015 and the samples were then air dried and sieved(2 mm) before analysis Soil texture was measured using thepipette method (Gee and Bauder 1986) by first removing theorganic content using hydrogen peroxide and then dispersingthe samples with sodium hexametaphosphate Particles wereclassified as sand (0063ndash2 mm) silt (0002ndash0063 mm) andclay (lt 0002 mm) Bulk density was determined by dryingeach soil sample in a soil core (each 5 cm diametertimes 7 cmheight) at 105 C for 48 h The density was calculated bydividing the dried soil mass by the core volume Soil pHwas determined on 10 g soil dispersed in 20 mL of deionisedwater after 10 min equilibration a reading was taken usinga pH probemeter (Thermo Fisher Scientific Inc WalthamMichigan USA) while stirring the suspension Total car-bon (TC) and nitrogen (TN) were determined through com-bustion of finely ground (0105 mm sieve size) soil using aLECO TruSpec carbonndashnitrogen analyser (TruSpecreg LECOCorporation Michigan USA) Analysis of TC TN and pHwere performed for samples taken every 5 cm from a 25 cmprofile (five samples per sampling point)

Sampling for analysis of NO3ndashN and NH4ndashN (n= 4ndash6 foreach site) was carried out on six occasions from 5 cm depthand an extraction performed by mixing 10 g fresh soil with2 M KCl After shaking for 1 h the KCl extracts were fil-tered and stored in the freezer until analysis NO3ndashN andNH4ndashN were determined from the extracts using an ion anal-

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2615

yser (Lachat QuikChemreg 5600 Lindburgh Drive LovelandColorado USA)

26 Microbial biomass

Microbial biomass C (MBC) and N (MBN) were determinedusing the chloroform fumigation extraction method (Vance etal 1987) 10 g samples of fresh soil (n= 4ndash6 for each site)were fumigated in a desiccator with 20 mL ethanol-free chlo-roform for 72 h and then extracted with 05 M K2SO4 Iden-tical numbers of subsamples were extracted with the samesolution but without fumigation the day after sampling Su-pernatants from both fumigated and non-fumigated sampleswere filtered through Whatman no 1 filter paper and storedin the freezer until analysis Organic carbon and total nitro-gen in the filtrate were analysed using a TOCTN analyser(Shimadzu Japan) Estimates of MBC and MBN were de-rived by calculating the difference between the results ofthe corresponding fumigated and non-fumigated analysisdivided by the extraction efficiency factor Factors of 045(Vance et al 1987) and 054 (Brookes et al 1985) were usedfor MBC and MBN respectively to account for uncompletedextraction of C and N in the microbial cell walls (Jonasson etal 1996)

27 Microbial activity

Soil enzymemicrobial activities were measured from sam-ples (n= 4ndash6 for each site) collected on six sampling occa-sions (March August and October 2014 March May andJuly 2015) All soil samples were sieved through a 2 mmsieve and analysed in triplicate

Beta-glucosidase (BG) was determined using the methoddescribed by Eivazi and Tabatabai (1988) After placing 1 gof soil in a 50 mL flask 025 mL toluene 4 mL of modi-fied universal buffer (pH 60) and 1 mL β-D-glucoside and p-nitrophenyl-α solutions (PNG) were added sequentially andmixed by swirling Samples were then incubated at 37 Cfor 1 h following which 1 mL of 05 M CaCl2 and 4 mL of01 M of tris(hydroxymethyl)aminomethane (pH 12) wereadded to halt further reactions The supernatants were fil-tered and the absorbance of the filtrate measured at 410 nmusing a spectrophotometer (Beckman Coulter DU 530 UVndashvis spectrophotometer) For control samples the same pro-cedure was followed except that PNG was added just beforefiltering the soil suspension instead of adding it at the begin-ning

Protease activity (PRO) was determined as described byKandeler et al (1999) Sodium caseinate (5 mL) solution wasadded to 1 g soil and the samples were incubated at 50 C for2 h and then filtered after adding 5 mL of trichloroacetic acidsolution Alkali and FolinndashCiocalteursquos reagents were addedto the filtrates before protease activity was determined col-orimetrically at 700 nm

To determine nitrate reductase (NR) activity 4 mL of 2 4-dinitrophenol solution and 1 mL of KNO3 were added to 5 gsamples of soil After incubation at 25 C for 24 h 10 mL 4 MKCl was added and filtered To 5 mL of the filtrate NH4Clbuffer (pH 85) sulfanilamide reagent was added and the ac-tivity of the enzyme NR was measured colorimetrically at520 nm

Total microbial activity was assayed using fluorescein di-acetate (FDA) based on the method described by Schnuumlrerand Rosswall (1982) later modified by Green et al (2006)Sodium phosphate buffer (pH 76) and FDA lipase substratewere added to flasks containing 1 g samples of soil and incu-bated for 3 h at 37 C The fluorescein content in the filteredsample was measured at 490 nm

28 Statistical analysis

All statistical analyses were performed using Minitab 16 Allthe values reported are means of three to six replicates andstandard errors were included when required To investigatethe effects of flooding we tested for significant differencesbetween the two sites (ie the LFS and SFS) for the peri-odspart periods (A B C D and E) defined on the basis ofthe state of each site in terms of waterlogging over the studyperiod This was carried out by applying analysis of vari-ance (ANOVA) for each flux soil enzymatic activity TCTN mineral N and MB Functional relationships between po-tential environmental drivers and the fluxes of CO2 and N2Owere performed assessed using linear or exponential regres-sion models Multiple regression analysis was used to deter-mine the relative contribution of more than one independentenvironmental driver to CO2 and N2O fluxes Normality andhomogeneity of the variance of all the models were checkedvisually from residual versus fitted plots and when neces-sary either square-root or log transformations applied Dif-ferences were considered statistically significant if P lt 005unless otherwise mentioned

3 Results

31 Soil characteristics

The relative proportion of clay is higher at the LFS but bothsites have sandy loam soil textures Soil pH was higher at theSFS than the LFS (Table 1) Porosity at the LFS was greaterthan at the SFS The soil C and N concentrations were sig-nificantly higher (P lt 0001) at the LFS site with the great-est difference in the upper soil layers At both sites C andN decreased with soil depth but more gradually at the SFS(Fig 2a b)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2616 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 2 Profile of (a) total carbon and (b) total nitrogen in the LFS and SFS Mean values (n= 6) are for each level within a profile andthe horizontal bars represent the standard errors

Table 1 Soil properties (0ndash10 cm) in the LFS and SFS Values aremeanplusmnSE

Soil properties LFS SFS

Sand () 6440plusmn 072 6900plusmn 095Clay () 1600plusmn 044 650plusmn 067Bulk density (g cmminus3) 073plusmn 008 102plusmn 004Porosity 073plusmn 008 061plusmn 004pH 479plusmn 011 528plusmn 021Total carbon (g kgminus1) 20470plusmn 1190 12000plusmn 650Total nitrogen (g kgminus1) 3850plusmn 652 1070plusmn 070C-to-N ratio 679plusmn 076 1095plusmn 015

32 Rainfall water depth redox potential and airtemperature

Depending on the timing and duration of flooding the fol-lowing periods could be identified as indicated in Fig 3 pe-riod A was characterized by LFS and SFS flooding period Bby LFS flooding period C by neither LFS nor SFS flood-ing period D by LFS flooding and period E by neither LFSnor SFS flooding Thus there was no equivalent of period Aduring 20142015 (ie the SFS was not flooded)

Air temperature followed the seasonal pattern typical forthis latitude (Fig 3a) with the highest values during the sum-mer months (JunendashAugust) and the lowest during the win-ter months (DecemberndashFebruary) The values ranged from ahigh of 23 C to a low of minus39 C with an average of 786 Cover the 2-year study period Rainfall was very variable overthe study period (Fig 3b) with some of the highest rainfallamounts occurring during the warmer months However the

number of days with rainfall was higher in the cooler pe-riod The mean annual rainfall (866 mm) at our site duringthe study period was considerably above the average valuefor the past 30 years (756 mm) measured at the Dublin Air-port station

Standing water was only present at the SFS site forsim 1 month (in the winter of 2014) and reached 5 cmdepth (Fig 3c) In contrast standing water was present forsim 6 months at the LFS site in 2015 and reached a depth ofsim 25 cm in 2014 Whilst standing water at the LFS site waslargely associated with the winter periods it also extendedinto the spring and autumn periods in 20142015 (Fig 3c)

Variations in soil redox potential broadly correlated withthe periods of standing water with the lowest values occur-ring in the LFS site in April 2014 and March 2015 (Fig 3d)reflecting the reducing conditions associated with longer pe-riods of inundation In contrast oxidising conditions werealways found at the SFS site and these were consistentlyhigher (more positive) than those at the LFS (Fig 3d) Mea-surements made from the water column when the LFS wasflooded to gt 10 cm depth indicated that this was always oxi-dising and had a low but positive redox potential (Fig 3d)

33 Seasonal variations in CO2 and N2O fluxes

The results are presented with reference to the periods AndashE which represent the state of the sites in relation to wa-ter availability The CO2 (Fig 4a) emissions showed markedseasonal variation with the highest CO2 values during thesummer months (C E) at both sites which correlated withthe lowest soil moisture values (Fig 4c) and the highest soiltemperatures (Fig 4d) The highest CO2 emissions were ob-served from the LFS site during each period (C E) in each

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2617

Figure 3 Variations in (a) daily air temperature (b) daily rainfall(c) average water depth above the soil surface and (d) redox poten-tial at the study sites Open circles represent redox values from theground surface of the SFS black filled circles represent values fromthe soil surface of the LFS and grey filled circles show values fromthe water surface while the LFS was flooded to greater than 10 cmdepth Period A LFS and SFS flooding period B LFS floodingperiod C neither LFS nor SFS flooding period D LFS floodingperiod E neither LFS nor SFS flooding Thus there was no equiv-alent of period A during late 2014ndashearly 2015

year and values were higher in 2015 compared to 2014 par-ticularly for the SFS site (Fig 4a) The lowest emissions wereobserved during the winter months of both years (A B D)when there were no significant differences (P = 0768) be-tween the LFS and SFS sites even though there was standingwater at the LFS site (Fig 3c)

In period A and the first part of period B CO2 emissionsare similar for the SFS and LFS Towards the end of pe-riod B the emissions are higher from the LFS corresponding

to the time when the water depth at the LFS was decreas-ing (Fig 3c) At the LFS CO2 emissions during period Cranged from 1694 to 49818 mg CO2ndashC mminus2 hminus1 whereasat the SFS the values ranged from 1668 to 429 mg CO2ndashC mminus2 hminus1 The initial part of period C is characterised bylow and similar values of CO2 emissions for each site thatare not appreciably different (P = 0956) from those of thepreceding period B Towards the middle of period C CO2emissions decline and then increase again corresponding to adecrease and then increase in soil moisture For the JunendashSeptember 2014 interval within period C CO2 emissionsfrom the two sites were significantly different (P lt 0034)

In period D similar trends in CO2 emissions with time(Fig 4a) were observed for the two sites despite differencesin soil moisture (Fig 4c) and water depth (Fig 3c) In pe-riod E CO2 emissions ranged from 5332 to 47726 mg CO2ndashC mminus2 hminus1 at the LFS and from 5568 to 34344 mg CO2ndashC mminus2 hminus1 at the SFS As for the corresponding of theprevious year (period C) the difference between the twosites was significant (P = 0013) for part of period E (MayndashAugust 2015)

Based on the relationships between soil respiration andtemperature estimated values for Q10 were 249 and 208 atthe LFS and SFS respectively

Unlike the CO2 fluxes the N2O emissions generallyshowed no discernible pattern between seasons duringthe study period (Fig 4b) and there were no system-atic changes over time No significant differences (P gt005) in N2O fluxes were observed between the LFS andSFS in any of the five periods In the first wetter peri-ods (periods A and B combined) N2O fluxes ranged fromminus0210 to 0319 mg N2OndashN mminus2 hminus1 and from minus0503 to0364 mg N2OndashN mminus2 hminus1 at the LFS and SFS respectivelyIn period D the N2O fluxes from the LFS showed highervariability than those from the SFS with values betweenminus0203 and 0695 mg N2OndashN mminus2 hminus1 for the LFS and be-tween minus0307 and 0206 mg N2OndashN mminus2 hminus1 for the SFSOverall N2O fluxes were generally higher from the LFSthan from the SFS during period D In the first growingseason (period C) the SFS showed consistently higher andmore positive N2O fluxes the maximum emission recorded(0651 mg N2OndashN mminus2 hminus1) was however from the LFS atthe end of July N2O fluxes less than zero suggest uptake ofN2O which was more common at the SFS site

Annual CO2 emissions were 818 and 876 (mean 847)and 1124 and 115 (mean 1137) Mg CO2ndashC haminus1 yrminus1 fromthe SFS and LFS respectively with values for the LFS 14times higher than that from the SFS For the same yearsthe annual N2O emissions were 13 times higher (709 and549 kg N2OndashN haminus1 yrminus1) from the LFS (mean 629) com-pared to the SFS (547 and 362 kg N2OndashN haminus1 yrminus1 mean454)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2618 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 4 Seasonal variations in (a) CO2 fluxes (b) N2O fluxes (c) soil moisture and (d) soil temperature in each site Periods as defined inFig 3

34 Relationship between CO2 and N2O fluxes andenvironmental parameters

Independent testing of the various environmental variablesshowed that soil temperature soil water content redox po-tential and for the LFS water depth were significantly cor-related with the CO2 emissions Significant positive corre-lations were found between CO2 emissions and soil tem-perature for both the SFS (R2

= 044 P lt 0001) and LFS(R2= 056 P lt 0001) with a stronger relationship at the

LFS (Fig 5a) A significant negative linear relationship be-tween soil moisture and CO2 emission was found at the

LFS (R2= 052 P lt 0001) and the SFS (R2

= 054 P lt0001) (Fig 5b) Correlations between CO2 emissions andredox potential were low with R2 values of 025 (LFS) and016 (SFS) respectively but significant at P lt 005 CO2emissions at the LFS were exponentially correlated with wa-ter depth (R2

= 045 P lt 0001) (Fig 5c) Combining soiltemperature and soil water content in multiple regressionanalyses with the CO2 fluxes only resulted in a small increasein explanatory power to sim 58 and 66 at the SFS and LFSrespectively but including redox potential had no impact onthe explanatory power

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

2614 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 1 Aerial photograph of part of East Coast Nature Reserve at Blackditch Wood County Wicklow showing the study sites and thedirection in which water flows during flooding Inset shows map of Ireland with site location indicated

of the exponential equation fitted to the CO2 flux and tem-perature data

23 Environmental measurements

In addition to the flux measurements other environmen-tal variables that could potentially influence the GHGfluxes were also measured A weather station located aboutsim 100 m from the locality where measurements were madecomprised sensors for air temperature (RHT3nl-CA) humid-ity (RHT3nl-CA) solar radiation (PYRPA-03) and rainfall(RG2+WS-CA) Average air temperature and cumulativerainfall were recorded at 2 m height every 5 and 60 minrespectively Soil moisture content and temperature weremeasured adjacent to the collarschambers using a handheldThetaProbe (Delta-T Devices Ltd Cambridge UK) eachtime gas sampling was performed During the flooding pe-riod the depth of standing water (WD) was measured adja-cent to the gas sampling points using a graduated woodenruler Redox potential was measured from each collar usinga portable Hanna redox meter (HI9125 Hanna Instruments)with a 10 cm redox electrode Redox potential was measuredfrom the soil surface except when the LFS was flooded abovea height of 10 cm in which case the measurements were ac-quired from the surface of water Complete insertion of theelectrode to the top soil was avoided to prevent the uncertainimpact of the intrusion of water into the electrode throughthe rim at the top during sampling

24 Biomass determinations

In the summer of 2014 the aboveground plant biomass perunit area was estimated from eight quadrats (45times 45 cm) ateach site by clipping the vegetation to approximately ground

level Dry biomass was determined after oven drying thefresh samples at 80 C for 72 h

25 Soil sampling for physical and chemical analysis

Sampling of soil from the two sites (n= 6 for each site) foranalysis of its physicochemical properties was carried out inJuly 2015 and the samples were then air dried and sieved(2 mm) before analysis Soil texture was measured using thepipette method (Gee and Bauder 1986) by first removing theorganic content using hydrogen peroxide and then dispersingthe samples with sodium hexametaphosphate Particles wereclassified as sand (0063ndash2 mm) silt (0002ndash0063 mm) andclay (lt 0002 mm) Bulk density was determined by dryingeach soil sample in a soil core (each 5 cm diametertimes 7 cmheight) at 105 C for 48 h The density was calculated bydividing the dried soil mass by the core volume Soil pHwas determined on 10 g soil dispersed in 20 mL of deionisedwater after 10 min equilibration a reading was taken usinga pH probemeter (Thermo Fisher Scientific Inc WalthamMichigan USA) while stirring the suspension Total car-bon (TC) and nitrogen (TN) were determined through com-bustion of finely ground (0105 mm sieve size) soil using aLECO TruSpec carbonndashnitrogen analyser (TruSpecreg LECOCorporation Michigan USA) Analysis of TC TN and pHwere performed for samples taken every 5 cm from a 25 cmprofile (five samples per sampling point)

Sampling for analysis of NO3ndashN and NH4ndashN (n= 4ndash6 foreach site) was carried out on six occasions from 5 cm depthand an extraction performed by mixing 10 g fresh soil with2 M KCl After shaking for 1 h the KCl extracts were fil-tered and stored in the freezer until analysis NO3ndashN andNH4ndashN were determined from the extracts using an ion anal-

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2615

yser (Lachat QuikChemreg 5600 Lindburgh Drive LovelandColorado USA)

26 Microbial biomass

Microbial biomass C (MBC) and N (MBN) were determinedusing the chloroform fumigation extraction method (Vance etal 1987) 10 g samples of fresh soil (n= 4ndash6 for each site)were fumigated in a desiccator with 20 mL ethanol-free chlo-roform for 72 h and then extracted with 05 M K2SO4 Iden-tical numbers of subsamples were extracted with the samesolution but without fumigation the day after sampling Su-pernatants from both fumigated and non-fumigated sampleswere filtered through Whatman no 1 filter paper and storedin the freezer until analysis Organic carbon and total nitro-gen in the filtrate were analysed using a TOCTN analyser(Shimadzu Japan) Estimates of MBC and MBN were de-rived by calculating the difference between the results ofthe corresponding fumigated and non-fumigated analysisdivided by the extraction efficiency factor Factors of 045(Vance et al 1987) and 054 (Brookes et al 1985) were usedfor MBC and MBN respectively to account for uncompletedextraction of C and N in the microbial cell walls (Jonasson etal 1996)

27 Microbial activity

Soil enzymemicrobial activities were measured from sam-ples (n= 4ndash6 for each site) collected on six sampling occa-sions (March August and October 2014 March May andJuly 2015) All soil samples were sieved through a 2 mmsieve and analysed in triplicate

Beta-glucosidase (BG) was determined using the methoddescribed by Eivazi and Tabatabai (1988) After placing 1 gof soil in a 50 mL flask 025 mL toluene 4 mL of modi-fied universal buffer (pH 60) and 1 mL β-D-glucoside and p-nitrophenyl-α solutions (PNG) were added sequentially andmixed by swirling Samples were then incubated at 37 Cfor 1 h following which 1 mL of 05 M CaCl2 and 4 mL of01 M of tris(hydroxymethyl)aminomethane (pH 12) wereadded to halt further reactions The supernatants were fil-tered and the absorbance of the filtrate measured at 410 nmusing a spectrophotometer (Beckman Coulter DU 530 UVndashvis spectrophotometer) For control samples the same pro-cedure was followed except that PNG was added just beforefiltering the soil suspension instead of adding it at the begin-ning

Protease activity (PRO) was determined as described byKandeler et al (1999) Sodium caseinate (5 mL) solution wasadded to 1 g soil and the samples were incubated at 50 C for2 h and then filtered after adding 5 mL of trichloroacetic acidsolution Alkali and FolinndashCiocalteursquos reagents were addedto the filtrates before protease activity was determined col-orimetrically at 700 nm

To determine nitrate reductase (NR) activity 4 mL of 2 4-dinitrophenol solution and 1 mL of KNO3 were added to 5 gsamples of soil After incubation at 25 C for 24 h 10 mL 4 MKCl was added and filtered To 5 mL of the filtrate NH4Clbuffer (pH 85) sulfanilamide reagent was added and the ac-tivity of the enzyme NR was measured colorimetrically at520 nm

Total microbial activity was assayed using fluorescein di-acetate (FDA) based on the method described by Schnuumlrerand Rosswall (1982) later modified by Green et al (2006)Sodium phosphate buffer (pH 76) and FDA lipase substratewere added to flasks containing 1 g samples of soil and incu-bated for 3 h at 37 C The fluorescein content in the filteredsample was measured at 490 nm

28 Statistical analysis

All statistical analyses were performed using Minitab 16 Allthe values reported are means of three to six replicates andstandard errors were included when required To investigatethe effects of flooding we tested for significant differencesbetween the two sites (ie the LFS and SFS) for the peri-odspart periods (A B C D and E) defined on the basis ofthe state of each site in terms of waterlogging over the studyperiod This was carried out by applying analysis of vari-ance (ANOVA) for each flux soil enzymatic activity TCTN mineral N and MB Functional relationships between po-tential environmental drivers and the fluxes of CO2 and N2Owere performed assessed using linear or exponential regres-sion models Multiple regression analysis was used to deter-mine the relative contribution of more than one independentenvironmental driver to CO2 and N2O fluxes Normality andhomogeneity of the variance of all the models were checkedvisually from residual versus fitted plots and when neces-sary either square-root or log transformations applied Dif-ferences were considered statistically significant if P lt 005unless otherwise mentioned

3 Results

31 Soil characteristics

The relative proportion of clay is higher at the LFS but bothsites have sandy loam soil textures Soil pH was higher at theSFS than the LFS (Table 1) Porosity at the LFS was greaterthan at the SFS The soil C and N concentrations were sig-nificantly higher (P lt 0001) at the LFS site with the great-est difference in the upper soil layers At both sites C andN decreased with soil depth but more gradually at the SFS(Fig 2a b)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2616 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 2 Profile of (a) total carbon and (b) total nitrogen in the LFS and SFS Mean values (n= 6) are for each level within a profile andthe horizontal bars represent the standard errors

Table 1 Soil properties (0ndash10 cm) in the LFS and SFS Values aremeanplusmnSE

Soil properties LFS SFS

Sand () 6440plusmn 072 6900plusmn 095Clay () 1600plusmn 044 650plusmn 067Bulk density (g cmminus3) 073plusmn 008 102plusmn 004Porosity 073plusmn 008 061plusmn 004pH 479plusmn 011 528plusmn 021Total carbon (g kgminus1) 20470plusmn 1190 12000plusmn 650Total nitrogen (g kgminus1) 3850plusmn 652 1070plusmn 070C-to-N ratio 679plusmn 076 1095plusmn 015

32 Rainfall water depth redox potential and airtemperature

Depending on the timing and duration of flooding the fol-lowing periods could be identified as indicated in Fig 3 pe-riod A was characterized by LFS and SFS flooding period Bby LFS flooding period C by neither LFS nor SFS flood-ing period D by LFS flooding and period E by neither LFSnor SFS flooding Thus there was no equivalent of period Aduring 20142015 (ie the SFS was not flooded)

Air temperature followed the seasonal pattern typical forthis latitude (Fig 3a) with the highest values during the sum-mer months (JunendashAugust) and the lowest during the win-ter months (DecemberndashFebruary) The values ranged from ahigh of 23 C to a low of minus39 C with an average of 786 Cover the 2-year study period Rainfall was very variable overthe study period (Fig 3b) with some of the highest rainfallamounts occurring during the warmer months However the

number of days with rainfall was higher in the cooler pe-riod The mean annual rainfall (866 mm) at our site duringthe study period was considerably above the average valuefor the past 30 years (756 mm) measured at the Dublin Air-port station

Standing water was only present at the SFS site forsim 1 month (in the winter of 2014) and reached 5 cmdepth (Fig 3c) In contrast standing water was present forsim 6 months at the LFS site in 2015 and reached a depth ofsim 25 cm in 2014 Whilst standing water at the LFS site waslargely associated with the winter periods it also extendedinto the spring and autumn periods in 20142015 (Fig 3c)

Variations in soil redox potential broadly correlated withthe periods of standing water with the lowest values occur-ring in the LFS site in April 2014 and March 2015 (Fig 3d)reflecting the reducing conditions associated with longer pe-riods of inundation In contrast oxidising conditions werealways found at the SFS site and these were consistentlyhigher (more positive) than those at the LFS (Fig 3d) Mea-surements made from the water column when the LFS wasflooded to gt 10 cm depth indicated that this was always oxi-dising and had a low but positive redox potential (Fig 3d)

33 Seasonal variations in CO2 and N2O fluxes

The results are presented with reference to the periods AndashE which represent the state of the sites in relation to wa-ter availability The CO2 (Fig 4a) emissions showed markedseasonal variation with the highest CO2 values during thesummer months (C E) at both sites which correlated withthe lowest soil moisture values (Fig 4c) and the highest soiltemperatures (Fig 4d) The highest CO2 emissions were ob-served from the LFS site during each period (C E) in each

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2617

Figure 3 Variations in (a) daily air temperature (b) daily rainfall(c) average water depth above the soil surface and (d) redox poten-tial at the study sites Open circles represent redox values from theground surface of the SFS black filled circles represent values fromthe soil surface of the LFS and grey filled circles show values fromthe water surface while the LFS was flooded to greater than 10 cmdepth Period A LFS and SFS flooding period B LFS floodingperiod C neither LFS nor SFS flooding period D LFS floodingperiod E neither LFS nor SFS flooding Thus there was no equiv-alent of period A during late 2014ndashearly 2015

year and values were higher in 2015 compared to 2014 par-ticularly for the SFS site (Fig 4a) The lowest emissions wereobserved during the winter months of both years (A B D)when there were no significant differences (P = 0768) be-tween the LFS and SFS sites even though there was standingwater at the LFS site (Fig 3c)

In period A and the first part of period B CO2 emissionsare similar for the SFS and LFS Towards the end of pe-riod B the emissions are higher from the LFS corresponding

to the time when the water depth at the LFS was decreas-ing (Fig 3c) At the LFS CO2 emissions during period Cranged from 1694 to 49818 mg CO2ndashC mminus2 hminus1 whereasat the SFS the values ranged from 1668 to 429 mg CO2ndashC mminus2 hminus1 The initial part of period C is characterised bylow and similar values of CO2 emissions for each site thatare not appreciably different (P = 0956) from those of thepreceding period B Towards the middle of period C CO2emissions decline and then increase again corresponding to adecrease and then increase in soil moisture For the JunendashSeptember 2014 interval within period C CO2 emissionsfrom the two sites were significantly different (P lt 0034)

In period D similar trends in CO2 emissions with time(Fig 4a) were observed for the two sites despite differencesin soil moisture (Fig 4c) and water depth (Fig 3c) In pe-riod E CO2 emissions ranged from 5332 to 47726 mg CO2ndashC mminus2 hminus1 at the LFS and from 5568 to 34344 mg CO2ndashC mminus2 hminus1 at the SFS As for the corresponding of theprevious year (period C) the difference between the twosites was significant (P = 0013) for part of period E (MayndashAugust 2015)

Based on the relationships between soil respiration andtemperature estimated values for Q10 were 249 and 208 atthe LFS and SFS respectively

Unlike the CO2 fluxes the N2O emissions generallyshowed no discernible pattern between seasons duringthe study period (Fig 4b) and there were no system-atic changes over time No significant differences (P gt005) in N2O fluxes were observed between the LFS andSFS in any of the five periods In the first wetter peri-ods (periods A and B combined) N2O fluxes ranged fromminus0210 to 0319 mg N2OndashN mminus2 hminus1 and from minus0503 to0364 mg N2OndashN mminus2 hminus1 at the LFS and SFS respectivelyIn period D the N2O fluxes from the LFS showed highervariability than those from the SFS with values betweenminus0203 and 0695 mg N2OndashN mminus2 hminus1 for the LFS and be-tween minus0307 and 0206 mg N2OndashN mminus2 hminus1 for the SFSOverall N2O fluxes were generally higher from the LFSthan from the SFS during period D In the first growingseason (period C) the SFS showed consistently higher andmore positive N2O fluxes the maximum emission recorded(0651 mg N2OndashN mminus2 hminus1) was however from the LFS atthe end of July N2O fluxes less than zero suggest uptake ofN2O which was more common at the SFS site

Annual CO2 emissions were 818 and 876 (mean 847)and 1124 and 115 (mean 1137) Mg CO2ndashC haminus1 yrminus1 fromthe SFS and LFS respectively with values for the LFS 14times higher than that from the SFS For the same yearsthe annual N2O emissions were 13 times higher (709 and549 kg N2OndashN haminus1 yrminus1) from the LFS (mean 629) com-pared to the SFS (547 and 362 kg N2OndashN haminus1 yrminus1 mean454)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2618 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 4 Seasonal variations in (a) CO2 fluxes (b) N2O fluxes (c) soil moisture and (d) soil temperature in each site Periods as defined inFig 3

34 Relationship between CO2 and N2O fluxes andenvironmental parameters

Independent testing of the various environmental variablesshowed that soil temperature soil water content redox po-tential and for the LFS water depth were significantly cor-related with the CO2 emissions Significant positive corre-lations were found between CO2 emissions and soil tem-perature for both the SFS (R2

= 044 P lt 0001) and LFS(R2= 056 P lt 0001) with a stronger relationship at the

LFS (Fig 5a) A significant negative linear relationship be-tween soil moisture and CO2 emission was found at the

LFS (R2= 052 P lt 0001) and the SFS (R2

= 054 P lt0001) (Fig 5b) Correlations between CO2 emissions andredox potential were low with R2 values of 025 (LFS) and016 (SFS) respectively but significant at P lt 005 CO2emissions at the LFS were exponentially correlated with wa-ter depth (R2

= 045 P lt 0001) (Fig 5c) Combining soiltemperature and soil water content in multiple regressionanalyses with the CO2 fluxes only resulted in a small increasein explanatory power to sim 58 and 66 at the SFS and LFSrespectively but including redox potential had no impact onthe explanatory power

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2615

yser (Lachat QuikChemreg 5600 Lindburgh Drive LovelandColorado USA)

26 Microbial biomass

Microbial biomass C (MBC) and N (MBN) were determinedusing the chloroform fumigation extraction method (Vance etal 1987) 10 g samples of fresh soil (n= 4ndash6 for each site)were fumigated in a desiccator with 20 mL ethanol-free chlo-roform for 72 h and then extracted with 05 M K2SO4 Iden-tical numbers of subsamples were extracted with the samesolution but without fumigation the day after sampling Su-pernatants from both fumigated and non-fumigated sampleswere filtered through Whatman no 1 filter paper and storedin the freezer until analysis Organic carbon and total nitro-gen in the filtrate were analysed using a TOCTN analyser(Shimadzu Japan) Estimates of MBC and MBN were de-rived by calculating the difference between the results ofthe corresponding fumigated and non-fumigated analysisdivided by the extraction efficiency factor Factors of 045(Vance et al 1987) and 054 (Brookes et al 1985) were usedfor MBC and MBN respectively to account for uncompletedextraction of C and N in the microbial cell walls (Jonasson etal 1996)

27 Microbial activity

Soil enzymemicrobial activities were measured from sam-ples (n= 4ndash6 for each site) collected on six sampling occa-sions (March August and October 2014 March May andJuly 2015) All soil samples were sieved through a 2 mmsieve and analysed in triplicate

Beta-glucosidase (BG) was determined using the methoddescribed by Eivazi and Tabatabai (1988) After placing 1 gof soil in a 50 mL flask 025 mL toluene 4 mL of modi-fied universal buffer (pH 60) and 1 mL β-D-glucoside and p-nitrophenyl-α solutions (PNG) were added sequentially andmixed by swirling Samples were then incubated at 37 Cfor 1 h following which 1 mL of 05 M CaCl2 and 4 mL of01 M of tris(hydroxymethyl)aminomethane (pH 12) wereadded to halt further reactions The supernatants were fil-tered and the absorbance of the filtrate measured at 410 nmusing a spectrophotometer (Beckman Coulter DU 530 UVndashvis spectrophotometer) For control samples the same pro-cedure was followed except that PNG was added just beforefiltering the soil suspension instead of adding it at the begin-ning

Protease activity (PRO) was determined as described byKandeler et al (1999) Sodium caseinate (5 mL) solution wasadded to 1 g soil and the samples were incubated at 50 C for2 h and then filtered after adding 5 mL of trichloroacetic acidsolution Alkali and FolinndashCiocalteursquos reagents were addedto the filtrates before protease activity was determined col-orimetrically at 700 nm

To determine nitrate reductase (NR) activity 4 mL of 2 4-dinitrophenol solution and 1 mL of KNO3 were added to 5 gsamples of soil After incubation at 25 C for 24 h 10 mL 4 MKCl was added and filtered To 5 mL of the filtrate NH4Clbuffer (pH 85) sulfanilamide reagent was added and the ac-tivity of the enzyme NR was measured colorimetrically at520 nm

Total microbial activity was assayed using fluorescein di-acetate (FDA) based on the method described by Schnuumlrerand Rosswall (1982) later modified by Green et al (2006)Sodium phosphate buffer (pH 76) and FDA lipase substratewere added to flasks containing 1 g samples of soil and incu-bated for 3 h at 37 C The fluorescein content in the filteredsample was measured at 490 nm

28 Statistical analysis

All statistical analyses were performed using Minitab 16 Allthe values reported are means of three to six replicates andstandard errors were included when required To investigatethe effects of flooding we tested for significant differencesbetween the two sites (ie the LFS and SFS) for the peri-odspart periods (A B C D and E) defined on the basis ofthe state of each site in terms of waterlogging over the studyperiod This was carried out by applying analysis of vari-ance (ANOVA) for each flux soil enzymatic activity TCTN mineral N and MB Functional relationships between po-tential environmental drivers and the fluxes of CO2 and N2Owere performed assessed using linear or exponential regres-sion models Multiple regression analysis was used to deter-mine the relative contribution of more than one independentenvironmental driver to CO2 and N2O fluxes Normality andhomogeneity of the variance of all the models were checkedvisually from residual versus fitted plots and when neces-sary either square-root or log transformations applied Dif-ferences were considered statistically significant if P lt 005unless otherwise mentioned

3 Results

31 Soil characteristics

The relative proportion of clay is higher at the LFS but bothsites have sandy loam soil textures Soil pH was higher at theSFS than the LFS (Table 1) Porosity at the LFS was greaterthan at the SFS The soil C and N concentrations were sig-nificantly higher (P lt 0001) at the LFS site with the great-est difference in the upper soil layers At both sites C andN decreased with soil depth but more gradually at the SFS(Fig 2a b)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2616 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 2 Profile of (a) total carbon and (b) total nitrogen in the LFS and SFS Mean values (n= 6) are for each level within a profile andthe horizontal bars represent the standard errors

Table 1 Soil properties (0ndash10 cm) in the LFS and SFS Values aremeanplusmnSE

Soil properties LFS SFS

Sand () 6440plusmn 072 6900plusmn 095Clay () 1600plusmn 044 650plusmn 067Bulk density (g cmminus3) 073plusmn 008 102plusmn 004Porosity 073plusmn 008 061plusmn 004pH 479plusmn 011 528plusmn 021Total carbon (g kgminus1) 20470plusmn 1190 12000plusmn 650Total nitrogen (g kgminus1) 3850plusmn 652 1070plusmn 070C-to-N ratio 679plusmn 076 1095plusmn 015

32 Rainfall water depth redox potential and airtemperature

Depending on the timing and duration of flooding the fol-lowing periods could be identified as indicated in Fig 3 pe-riod A was characterized by LFS and SFS flooding period Bby LFS flooding period C by neither LFS nor SFS flood-ing period D by LFS flooding and period E by neither LFSnor SFS flooding Thus there was no equivalent of period Aduring 20142015 (ie the SFS was not flooded)

Air temperature followed the seasonal pattern typical forthis latitude (Fig 3a) with the highest values during the sum-mer months (JunendashAugust) and the lowest during the win-ter months (DecemberndashFebruary) The values ranged from ahigh of 23 C to a low of minus39 C with an average of 786 Cover the 2-year study period Rainfall was very variable overthe study period (Fig 3b) with some of the highest rainfallamounts occurring during the warmer months However the

number of days with rainfall was higher in the cooler pe-riod The mean annual rainfall (866 mm) at our site duringthe study period was considerably above the average valuefor the past 30 years (756 mm) measured at the Dublin Air-port station

Standing water was only present at the SFS site forsim 1 month (in the winter of 2014) and reached 5 cmdepth (Fig 3c) In contrast standing water was present forsim 6 months at the LFS site in 2015 and reached a depth ofsim 25 cm in 2014 Whilst standing water at the LFS site waslargely associated with the winter periods it also extendedinto the spring and autumn periods in 20142015 (Fig 3c)

Variations in soil redox potential broadly correlated withthe periods of standing water with the lowest values occur-ring in the LFS site in April 2014 and March 2015 (Fig 3d)reflecting the reducing conditions associated with longer pe-riods of inundation In contrast oxidising conditions werealways found at the SFS site and these were consistentlyhigher (more positive) than those at the LFS (Fig 3d) Mea-surements made from the water column when the LFS wasflooded to gt 10 cm depth indicated that this was always oxi-dising and had a low but positive redox potential (Fig 3d)

33 Seasonal variations in CO2 and N2O fluxes

The results are presented with reference to the periods AndashE which represent the state of the sites in relation to wa-ter availability The CO2 (Fig 4a) emissions showed markedseasonal variation with the highest CO2 values during thesummer months (C E) at both sites which correlated withthe lowest soil moisture values (Fig 4c) and the highest soiltemperatures (Fig 4d) The highest CO2 emissions were ob-served from the LFS site during each period (C E) in each

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2617

Figure 3 Variations in (a) daily air temperature (b) daily rainfall(c) average water depth above the soil surface and (d) redox poten-tial at the study sites Open circles represent redox values from theground surface of the SFS black filled circles represent values fromthe soil surface of the LFS and grey filled circles show values fromthe water surface while the LFS was flooded to greater than 10 cmdepth Period A LFS and SFS flooding period B LFS floodingperiod C neither LFS nor SFS flooding period D LFS floodingperiod E neither LFS nor SFS flooding Thus there was no equiv-alent of period A during late 2014ndashearly 2015

year and values were higher in 2015 compared to 2014 par-ticularly for the SFS site (Fig 4a) The lowest emissions wereobserved during the winter months of both years (A B D)when there were no significant differences (P = 0768) be-tween the LFS and SFS sites even though there was standingwater at the LFS site (Fig 3c)

In period A and the first part of period B CO2 emissionsare similar for the SFS and LFS Towards the end of pe-riod B the emissions are higher from the LFS corresponding

to the time when the water depth at the LFS was decreas-ing (Fig 3c) At the LFS CO2 emissions during period Cranged from 1694 to 49818 mg CO2ndashC mminus2 hminus1 whereasat the SFS the values ranged from 1668 to 429 mg CO2ndashC mminus2 hminus1 The initial part of period C is characterised bylow and similar values of CO2 emissions for each site thatare not appreciably different (P = 0956) from those of thepreceding period B Towards the middle of period C CO2emissions decline and then increase again corresponding to adecrease and then increase in soil moisture For the JunendashSeptember 2014 interval within period C CO2 emissionsfrom the two sites were significantly different (P lt 0034)

In period D similar trends in CO2 emissions with time(Fig 4a) were observed for the two sites despite differencesin soil moisture (Fig 4c) and water depth (Fig 3c) In pe-riod E CO2 emissions ranged from 5332 to 47726 mg CO2ndashC mminus2 hminus1 at the LFS and from 5568 to 34344 mg CO2ndashC mminus2 hminus1 at the SFS As for the corresponding of theprevious year (period C) the difference between the twosites was significant (P = 0013) for part of period E (MayndashAugust 2015)

Based on the relationships between soil respiration andtemperature estimated values for Q10 were 249 and 208 atthe LFS and SFS respectively

Unlike the CO2 fluxes the N2O emissions generallyshowed no discernible pattern between seasons duringthe study period (Fig 4b) and there were no system-atic changes over time No significant differences (P gt005) in N2O fluxes were observed between the LFS andSFS in any of the five periods In the first wetter peri-ods (periods A and B combined) N2O fluxes ranged fromminus0210 to 0319 mg N2OndashN mminus2 hminus1 and from minus0503 to0364 mg N2OndashN mminus2 hminus1 at the LFS and SFS respectivelyIn period D the N2O fluxes from the LFS showed highervariability than those from the SFS with values betweenminus0203 and 0695 mg N2OndashN mminus2 hminus1 for the LFS and be-tween minus0307 and 0206 mg N2OndashN mminus2 hminus1 for the SFSOverall N2O fluxes were generally higher from the LFSthan from the SFS during period D In the first growingseason (period C) the SFS showed consistently higher andmore positive N2O fluxes the maximum emission recorded(0651 mg N2OndashN mminus2 hminus1) was however from the LFS atthe end of July N2O fluxes less than zero suggest uptake ofN2O which was more common at the SFS site

Annual CO2 emissions were 818 and 876 (mean 847)and 1124 and 115 (mean 1137) Mg CO2ndashC haminus1 yrminus1 fromthe SFS and LFS respectively with values for the LFS 14times higher than that from the SFS For the same yearsthe annual N2O emissions were 13 times higher (709 and549 kg N2OndashN haminus1 yrminus1) from the LFS (mean 629) com-pared to the SFS (547 and 362 kg N2OndashN haminus1 yrminus1 mean454)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2618 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 4 Seasonal variations in (a) CO2 fluxes (b) N2O fluxes (c) soil moisture and (d) soil temperature in each site Periods as defined inFig 3

34 Relationship between CO2 and N2O fluxes andenvironmental parameters

Independent testing of the various environmental variablesshowed that soil temperature soil water content redox po-tential and for the LFS water depth were significantly cor-related with the CO2 emissions Significant positive corre-lations were found between CO2 emissions and soil tem-perature for both the SFS (R2

= 044 P lt 0001) and LFS(R2= 056 P lt 0001) with a stronger relationship at the

LFS (Fig 5a) A significant negative linear relationship be-tween soil moisture and CO2 emission was found at the

LFS (R2= 052 P lt 0001) and the SFS (R2

= 054 P lt0001) (Fig 5b) Correlations between CO2 emissions andredox potential were low with R2 values of 025 (LFS) and016 (SFS) respectively but significant at P lt 005 CO2emissions at the LFS were exponentially correlated with wa-ter depth (R2

= 045 P lt 0001) (Fig 5c) Combining soiltemperature and soil water content in multiple regressionanalyses with the CO2 fluxes only resulted in a small increasein explanatory power to sim 58 and 66 at the SFS and LFSrespectively but including redox potential had no impact onthe explanatory power

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

2616 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 2 Profile of (a) total carbon and (b) total nitrogen in the LFS and SFS Mean values (n= 6) are for each level within a profile andthe horizontal bars represent the standard errors

Table 1 Soil properties (0ndash10 cm) in the LFS and SFS Values aremeanplusmnSE

Soil properties LFS SFS

Sand () 6440plusmn 072 6900plusmn 095Clay () 1600plusmn 044 650plusmn 067Bulk density (g cmminus3) 073plusmn 008 102plusmn 004Porosity 073plusmn 008 061plusmn 004pH 479plusmn 011 528plusmn 021Total carbon (g kgminus1) 20470plusmn 1190 12000plusmn 650Total nitrogen (g kgminus1) 3850plusmn 652 1070plusmn 070C-to-N ratio 679plusmn 076 1095plusmn 015

32 Rainfall water depth redox potential and airtemperature

Depending on the timing and duration of flooding the fol-lowing periods could be identified as indicated in Fig 3 pe-riod A was characterized by LFS and SFS flooding period Bby LFS flooding period C by neither LFS nor SFS flood-ing period D by LFS flooding and period E by neither LFSnor SFS flooding Thus there was no equivalent of period Aduring 20142015 (ie the SFS was not flooded)

Air temperature followed the seasonal pattern typical forthis latitude (Fig 3a) with the highest values during the sum-mer months (JunendashAugust) and the lowest during the win-ter months (DecemberndashFebruary) The values ranged from ahigh of 23 C to a low of minus39 C with an average of 786 Cover the 2-year study period Rainfall was very variable overthe study period (Fig 3b) with some of the highest rainfallamounts occurring during the warmer months However the

number of days with rainfall was higher in the cooler pe-riod The mean annual rainfall (866 mm) at our site duringthe study period was considerably above the average valuefor the past 30 years (756 mm) measured at the Dublin Air-port station

Standing water was only present at the SFS site forsim 1 month (in the winter of 2014) and reached 5 cmdepth (Fig 3c) In contrast standing water was present forsim 6 months at the LFS site in 2015 and reached a depth ofsim 25 cm in 2014 Whilst standing water at the LFS site waslargely associated with the winter periods it also extendedinto the spring and autumn periods in 20142015 (Fig 3c)

Variations in soil redox potential broadly correlated withthe periods of standing water with the lowest values occur-ring in the LFS site in April 2014 and March 2015 (Fig 3d)reflecting the reducing conditions associated with longer pe-riods of inundation In contrast oxidising conditions werealways found at the SFS site and these were consistentlyhigher (more positive) than those at the LFS (Fig 3d) Mea-surements made from the water column when the LFS wasflooded to gt 10 cm depth indicated that this was always oxi-dising and had a low but positive redox potential (Fig 3d)

33 Seasonal variations in CO2 and N2O fluxes

The results are presented with reference to the periods AndashE which represent the state of the sites in relation to wa-ter availability The CO2 (Fig 4a) emissions showed markedseasonal variation with the highest CO2 values during thesummer months (C E) at both sites which correlated withthe lowest soil moisture values (Fig 4c) and the highest soiltemperatures (Fig 4d) The highest CO2 emissions were ob-served from the LFS site during each period (C E) in each

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2617

Figure 3 Variations in (a) daily air temperature (b) daily rainfall(c) average water depth above the soil surface and (d) redox poten-tial at the study sites Open circles represent redox values from theground surface of the SFS black filled circles represent values fromthe soil surface of the LFS and grey filled circles show values fromthe water surface while the LFS was flooded to greater than 10 cmdepth Period A LFS and SFS flooding period B LFS floodingperiod C neither LFS nor SFS flooding period D LFS floodingperiod E neither LFS nor SFS flooding Thus there was no equiv-alent of period A during late 2014ndashearly 2015

year and values were higher in 2015 compared to 2014 par-ticularly for the SFS site (Fig 4a) The lowest emissions wereobserved during the winter months of both years (A B D)when there were no significant differences (P = 0768) be-tween the LFS and SFS sites even though there was standingwater at the LFS site (Fig 3c)

In period A and the first part of period B CO2 emissionsare similar for the SFS and LFS Towards the end of pe-riod B the emissions are higher from the LFS corresponding

to the time when the water depth at the LFS was decreas-ing (Fig 3c) At the LFS CO2 emissions during period Cranged from 1694 to 49818 mg CO2ndashC mminus2 hminus1 whereasat the SFS the values ranged from 1668 to 429 mg CO2ndashC mminus2 hminus1 The initial part of period C is characterised bylow and similar values of CO2 emissions for each site thatare not appreciably different (P = 0956) from those of thepreceding period B Towards the middle of period C CO2emissions decline and then increase again corresponding to adecrease and then increase in soil moisture For the JunendashSeptember 2014 interval within period C CO2 emissionsfrom the two sites were significantly different (P lt 0034)

In period D similar trends in CO2 emissions with time(Fig 4a) were observed for the two sites despite differencesin soil moisture (Fig 4c) and water depth (Fig 3c) In pe-riod E CO2 emissions ranged from 5332 to 47726 mg CO2ndashC mminus2 hminus1 at the LFS and from 5568 to 34344 mg CO2ndashC mminus2 hminus1 at the SFS As for the corresponding of theprevious year (period C) the difference between the twosites was significant (P = 0013) for part of period E (MayndashAugust 2015)

Based on the relationships between soil respiration andtemperature estimated values for Q10 were 249 and 208 atthe LFS and SFS respectively

Unlike the CO2 fluxes the N2O emissions generallyshowed no discernible pattern between seasons duringthe study period (Fig 4b) and there were no system-atic changes over time No significant differences (P gt005) in N2O fluxes were observed between the LFS andSFS in any of the five periods In the first wetter peri-ods (periods A and B combined) N2O fluxes ranged fromminus0210 to 0319 mg N2OndashN mminus2 hminus1 and from minus0503 to0364 mg N2OndashN mminus2 hminus1 at the LFS and SFS respectivelyIn period D the N2O fluxes from the LFS showed highervariability than those from the SFS with values betweenminus0203 and 0695 mg N2OndashN mminus2 hminus1 for the LFS and be-tween minus0307 and 0206 mg N2OndashN mminus2 hminus1 for the SFSOverall N2O fluxes were generally higher from the LFSthan from the SFS during period D In the first growingseason (period C) the SFS showed consistently higher andmore positive N2O fluxes the maximum emission recorded(0651 mg N2OndashN mminus2 hminus1) was however from the LFS atthe end of July N2O fluxes less than zero suggest uptake ofN2O which was more common at the SFS site

Annual CO2 emissions were 818 and 876 (mean 847)and 1124 and 115 (mean 1137) Mg CO2ndashC haminus1 yrminus1 fromthe SFS and LFS respectively with values for the LFS 14times higher than that from the SFS For the same yearsthe annual N2O emissions were 13 times higher (709 and549 kg N2OndashN haminus1 yrminus1) from the LFS (mean 629) com-pared to the SFS (547 and 362 kg N2OndashN haminus1 yrminus1 mean454)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2618 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 4 Seasonal variations in (a) CO2 fluxes (b) N2O fluxes (c) soil moisture and (d) soil temperature in each site Periods as defined inFig 3

34 Relationship between CO2 and N2O fluxes andenvironmental parameters

Independent testing of the various environmental variablesshowed that soil temperature soil water content redox po-tential and for the LFS water depth were significantly cor-related with the CO2 emissions Significant positive corre-lations were found between CO2 emissions and soil tem-perature for both the SFS (R2

= 044 P lt 0001) and LFS(R2= 056 P lt 0001) with a stronger relationship at the

LFS (Fig 5a) A significant negative linear relationship be-tween soil moisture and CO2 emission was found at the

LFS (R2= 052 P lt 0001) and the SFS (R2

= 054 P lt0001) (Fig 5b) Correlations between CO2 emissions andredox potential were low with R2 values of 025 (LFS) and016 (SFS) respectively but significant at P lt 005 CO2emissions at the LFS were exponentially correlated with wa-ter depth (R2

= 045 P lt 0001) (Fig 5c) Combining soiltemperature and soil water content in multiple regressionanalyses with the CO2 fluxes only resulted in a small increasein explanatory power to sim 58 and 66 at the SFS and LFSrespectively but including redox potential had no impact onthe explanatory power

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2617

Figure 3 Variations in (a) daily air temperature (b) daily rainfall(c) average water depth above the soil surface and (d) redox poten-tial at the study sites Open circles represent redox values from theground surface of the SFS black filled circles represent values fromthe soil surface of the LFS and grey filled circles show values fromthe water surface while the LFS was flooded to greater than 10 cmdepth Period A LFS and SFS flooding period B LFS floodingperiod C neither LFS nor SFS flooding period D LFS floodingperiod E neither LFS nor SFS flooding Thus there was no equiv-alent of period A during late 2014ndashearly 2015

year and values were higher in 2015 compared to 2014 par-ticularly for the SFS site (Fig 4a) The lowest emissions wereobserved during the winter months of both years (A B D)when there were no significant differences (P = 0768) be-tween the LFS and SFS sites even though there was standingwater at the LFS site (Fig 3c)

In period A and the first part of period B CO2 emissionsare similar for the SFS and LFS Towards the end of pe-riod B the emissions are higher from the LFS corresponding

to the time when the water depth at the LFS was decreas-ing (Fig 3c) At the LFS CO2 emissions during period Cranged from 1694 to 49818 mg CO2ndashC mminus2 hminus1 whereasat the SFS the values ranged from 1668 to 429 mg CO2ndashC mminus2 hminus1 The initial part of period C is characterised bylow and similar values of CO2 emissions for each site thatare not appreciably different (P = 0956) from those of thepreceding period B Towards the middle of period C CO2emissions decline and then increase again corresponding to adecrease and then increase in soil moisture For the JunendashSeptember 2014 interval within period C CO2 emissionsfrom the two sites were significantly different (P lt 0034)

In period D similar trends in CO2 emissions with time(Fig 4a) were observed for the two sites despite differencesin soil moisture (Fig 4c) and water depth (Fig 3c) In pe-riod E CO2 emissions ranged from 5332 to 47726 mg CO2ndashC mminus2 hminus1 at the LFS and from 5568 to 34344 mg CO2ndashC mminus2 hminus1 at the SFS As for the corresponding of theprevious year (period C) the difference between the twosites was significant (P = 0013) for part of period E (MayndashAugust 2015)

Based on the relationships between soil respiration andtemperature estimated values for Q10 were 249 and 208 atthe LFS and SFS respectively

Unlike the CO2 fluxes the N2O emissions generallyshowed no discernible pattern between seasons duringthe study period (Fig 4b) and there were no system-atic changes over time No significant differences (P gt005) in N2O fluxes were observed between the LFS andSFS in any of the five periods In the first wetter peri-ods (periods A and B combined) N2O fluxes ranged fromminus0210 to 0319 mg N2OndashN mminus2 hminus1 and from minus0503 to0364 mg N2OndashN mminus2 hminus1 at the LFS and SFS respectivelyIn period D the N2O fluxes from the LFS showed highervariability than those from the SFS with values betweenminus0203 and 0695 mg N2OndashN mminus2 hminus1 for the LFS and be-tween minus0307 and 0206 mg N2OndashN mminus2 hminus1 for the SFSOverall N2O fluxes were generally higher from the LFSthan from the SFS during period D In the first growingseason (period C) the SFS showed consistently higher andmore positive N2O fluxes the maximum emission recorded(0651 mg N2OndashN mminus2 hminus1) was however from the LFS atthe end of July N2O fluxes less than zero suggest uptake ofN2O which was more common at the SFS site

Annual CO2 emissions were 818 and 876 (mean 847)and 1124 and 115 (mean 1137) Mg CO2ndashC haminus1 yrminus1 fromthe SFS and LFS respectively with values for the LFS 14times higher than that from the SFS For the same yearsthe annual N2O emissions were 13 times higher (709 and549 kg N2OndashN haminus1 yrminus1) from the LFS (mean 629) com-pared to the SFS (547 and 362 kg N2OndashN haminus1 yrminus1 mean454)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2618 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 4 Seasonal variations in (a) CO2 fluxes (b) N2O fluxes (c) soil moisture and (d) soil temperature in each site Periods as defined inFig 3

34 Relationship between CO2 and N2O fluxes andenvironmental parameters

Independent testing of the various environmental variablesshowed that soil temperature soil water content redox po-tential and for the LFS water depth were significantly cor-related with the CO2 emissions Significant positive corre-lations were found between CO2 emissions and soil tem-perature for both the SFS (R2

= 044 P lt 0001) and LFS(R2= 056 P lt 0001) with a stronger relationship at the

LFS (Fig 5a) A significant negative linear relationship be-tween soil moisture and CO2 emission was found at the

LFS (R2= 052 P lt 0001) and the SFS (R2

= 054 P lt0001) (Fig 5b) Correlations between CO2 emissions andredox potential were low with R2 values of 025 (LFS) and016 (SFS) respectively but significant at P lt 005 CO2emissions at the LFS were exponentially correlated with wa-ter depth (R2

= 045 P lt 0001) (Fig 5c) Combining soiltemperature and soil water content in multiple regressionanalyses with the CO2 fluxes only resulted in a small increasein explanatory power to sim 58 and 66 at the SFS and LFSrespectively but including redox potential had no impact onthe explanatory power

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

2618 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 4 Seasonal variations in (a) CO2 fluxes (b) N2O fluxes (c) soil moisture and (d) soil temperature in each site Periods as defined inFig 3

34 Relationship between CO2 and N2O fluxes andenvironmental parameters

Independent testing of the various environmental variablesshowed that soil temperature soil water content redox po-tential and for the LFS water depth were significantly cor-related with the CO2 emissions Significant positive corre-lations were found between CO2 emissions and soil tem-perature for both the SFS (R2

= 044 P lt 0001) and LFS(R2= 056 P lt 0001) with a stronger relationship at the

LFS (Fig 5a) A significant negative linear relationship be-tween soil moisture and CO2 emission was found at the

LFS (R2= 052 P lt 0001) and the SFS (R2

= 054 P lt0001) (Fig 5b) Correlations between CO2 emissions andredox potential were low with R2 values of 025 (LFS) and016 (SFS) respectively but significant at P lt 005 CO2emissions at the LFS were exponentially correlated with wa-ter depth (R2

= 045 P lt 0001) (Fig 5c) Combining soiltemperature and soil water content in multiple regressionanalyses with the CO2 fluxes only resulted in a small increasein explanatory power to sim 58 and 66 at the SFS and LFSrespectively but including redox potential had no impact onthe explanatory power

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2619

Figure 5 Relationships between CO2 fluxes and (a) soil temperature (b) soil moisture and (c) water depth at the LFS The lines in(a) represent the best-fit regression y = 5069(exp009 middot T ) R2

= 056 P lt 0001 and y = 5362(exp007 middot T ) R2= 044 P lt 0001

for the LFS and SFS respectively The lines in (b) represent the best-fit regression y = 44626minus 582 middot (SWC) R2= 052 P lt 0001

and y = 35195minus 592 middot (SWC) R2= 054 P lt 0001 for the LFS and SFS respectively The relationship in (c) is represented by

y = 10965(exp(minus004 middotWD)) R2= 045 P lt 0001 Values in (a b) are means for n= 3ndash4

No significant relationship was observed between N2Ofluxes and any of soil temperature soil moisture redox po-tential or water depth at the LFS At the SFS soil N2Ofluxes did correlate positively with soil moisture (R2

= 013P lt 001) and soil temperature (R2

= 013 P lt 001) butwith a low explanatory power The N2O flux was not corre-lated with redox potential at the SFS

35 Soil enzymaticmicrobial activity

While there are some significant differences between differ-ent sampling dates for BG FDA and PRO they overall showsimilar values for and variation at both sites (Fig 6) Forperiod B BG activity was significantly lower (P = 0017) atthe LFS however in the second flooding period (first part ofperiod D) BG was significantly higher (P = 0001) at theLFS than at the SFS In period E FDA activity was sig-nificantly higher (P = 0001) at the LFS than the SFS Incontrast NR activities were consistently and significantlylower (P lt 0001) at the SFS and independent of water sta-tus (standing water availability)

36 Microbial biomass and soil NOminus3 and NH+4

Seasonal variations in MB appear to differ between the LFSand SFS (Fig 7) Total MBC was generally higher at theLFS than at the SFS at each sampling period but significant(P lt 001) for periods B late D and early E (Fig 7a) MBNvalues were higher at the LFS than at the SFS (Fig 7b) forall except two sampling dates For MBC MBN ratios weresignificantly higher (P lt 001) at the LFS during periods ofstanding water (Fig 7c)

The concentrations of NH+4 and NOminus3 were generallyhigher at the LFS than at the SFS (Table 2) The highest con-centrations of NH+4 were generally associated with periodsof standing water (LFS March 20142015) or immediatelyafter (SFS March 2014) the disappearance of standing wa-ter (Table 2 and Fig 3c) For NOminus3 the highest concentrationswere generally found at the LFS but there was no discerniblepattern over the sim 2 years of the study at either the LFS orSFS

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

2620 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

Figure 6 Soil enzymatic activities of (a) protease (b) beta-glucosidase (c) fluorescein diacetate and (d) nitrate reductase measured atdifferent sampling dates from the LFS and SFS Periods corresponding to different sampling dates are indicated by different letters (seeFig 3) Values are meanplusmnSE (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date at(P lt 005)

37 Plant biomass

The values for aboveground plant biomass were approxi-mately 6 times higher at the SFS (3551 Mg haminus1) comparedto the LFS (602 Mg haminus1)

4 Discussion

The mean annual CO2 emissions were 1137 and847 Mg CO2ndashC haminus1 yrminus1 from the LFS and SFS re-spectively Longer-term flooding therefore increased ratherthan reduced the annual emissions by approximately 40 and suggests that any increase in freshwater flooding inresponse to climate change could result in a significantincrease in carbon dioxide emissions from these systemsThe annual emissions of CO2 found in this study are inline with those previously reported for floodplain wetlands(1091plusmn 054 Mg CO2ndashC haminus1 yrminus1) (Batson et al 2015)coastal plain wetlands (1129 Mg CO2ndashC haminus1 yrminus1) (Morseet al 2012) and occasionally (97 Mg CO2ndashC haminus1 yrminus1)

or frequently flooded (13 Mg CO2ndashC haminus1 yrminus1) riparianforests (Jacinthe 2015) Our results are also in part agree-ment with other investigations on comparable ecosystemsthat examined the impact of flooding on GHGs (Morse et al2012 Jacinthe 2015 Kim et al 2015 Mariacuten-Muntildeiz et al2015) Jacinthe (2015) reported that CO2 fluxes during sum-mer were larger from a riparian forest affected by floods inwinter and spring than from a flood protected area Similarly

Morse et al (2012) found higher rates of CO2 emission inthe dry period from short and intermittently flooded restoredwetland habitats compared to both permanently flooded andunflooded sites

41 Relationships between CO2 emissions and waterand nutrient availability

The overall negative relationship between CO2 emissionsand soil moisture (Fig 5b) suggests that flooding or high soilwater availability impedes the decomposition processes thatlead to CO2 production via the creation of low-oxygen con-ditions Standing water would also act as an additional con-straint on annual emissions by acting as a physical barrier togaseous diffusion However somewhat paradoxically largerannual CO2 emissions were associated with the LFS site iethe site inundated for longer However the highest CO2 emis-sions and the period when the differences in CO2 emissionsbetween the two sites were greatest occurred in the summerseason after the disappearance of standing water when thesoil was better oxygenated (Fig 3d) Potentially the presenceof standing water during the autumnwinter months could in-hibit CO2 emissions at the LFS by acting as a gaseous barrierHowever the values found for the SFS for the same periodare similar indicating that this is unlikely to have a signif-icant impact on the annual emissions Reductions in miner-alisation caused by low temperatures may be the more sig-nificant factor at these times of the year consistent with thestrong correlations between CO2 emissions and temperature

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2621

Figure 7 Seasonal variations in (a) MBC (b) MBN and (c) MBC MBN from the LFS and SFS measured at different sampling datesPeriods corresponding to different sampling dates are indicated by different letters (see Fig 3) Vertical bars represent the standard errors ofthe mean (n= 4ndash6) Asterisks indicate a significant difference between the LFS and SFS for the same sampling date (P lt 005)

Table 2 MeanplusmnSE of soil NH+4 and NOminus3 concentrations at 0ndash5 cm depths (mg N kgminus1) for different sampling dates in 2014 and 2015

Date (period) LFS SFS

NH+4 NOminus3 NH+4 NOminus3

4 March 2014 (B) 3049plusmn 379 099plusmn 008 2563plusmn 185 052plusmn 01530 April 2014 (C) 1293plusmn 436 030plusmn 006 437plusmn 080 000plusmn 00019 August 2014 (C) 1247plusmn 112 028plusmn 011 1685plusmn 167 029plusmn 0178 March 2015 (D) 2580plusmn 237 000plusmn 000 1766plusmn 084 000plusmn 00028 May 2015 (E) 1824plusmn 228 228plusmn 079 1588plusmn 136 012plusmn 01117 July 2015 (E) 1308plusmn 041 051plusmn 013 1211plusmn 036 024plusmn 015

that were observed in this study This observation impliesthat the impact of flooding on annual CO2 emissions couldtherefore be strongly sensitive to the intra-annual timing offlooding events

Higher CO2 emissions were obtained at the LFS during thedrier parts of the year when there were similar values for soilmoisturesoil temperature at both sites This may be related tohigher organic matter content and nutrient status and a gener-ally higher microbial biomass at the LFS Compared to soilssupplied with no or little organic matternutrients soils thathave received more organic matter are likely to emit substan-tially larger amounts of CO2 (Winton and Richardson 2015)The availability of organic matter is one of the most impor-tant factors controlling the production of GHGs in wetlands(Badiou et al 2011) The source of the organic matter willbe dependent on the hydrological connectivity of the wetlandto the upstream land use system in addition to plant litterderived from wetland plants (Hernandez and Mitsch 2007)

Based on the plant biomass estimates (Sect 37) similar oreven higher carbon contents would have been expected forthe SFS had the carbon been contributed mainly from theplant community The difference in total C and N values be-tween the LFS and SFS sites and specifically the rapid de-cline in these nutrients down through the soil profile indicatethese are derived largely from external sources rather thanproduced in situ from plant-related material These nutrientswere presumably introduced with the drainage water Manystudies have reported an increase in the release of CO2 viasoil respiration as a result of the increased input of organicmatter particularly in wetlands (Samaritani et al 2011 Win-ton and Richardson 2015) Given that the C fluxes observedin this work are determined largely by external nutrient in-puts this adds to the growing body of evidence that bio-geochemical processes including greenhouse gas emissionsin floodplain wetlands are predominantly determined by off-sitecatchment-related events (Batson et al 2015)

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

2622 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

42 Relationship between CO2 fluxes and soiltemperature

There was a positive relationship between temperature andCO2 production at the two sites Temperature was the sin-gle most important variable overall explaining variations inCO2 emissions Kim et al (2015) also showed that CO2production increased with rising temperature from incu-bated flooded and non-flooded boreal soils Variations inCO2 emissions frequently matched changes in temperatureat times when other variables such as soil moisture wereinvariant most obviously over the duration of period D(Fig 4a c and d) As soil temperature was essentially identi-cal between the sites at any given time the different relation-ships between soil temperature and CO2 emissions (Fig 5a)imply that at least one other variable is involved The greaterabundance of soil carbon and nitrogen accompanied by agenerally higher microbial biomass suggests substrate avail-ability at the LFS was greater for soil microbial processes(Wang et al 2003) The higher Q10 values at the LFS (249)compared to the SFS (208) also suggest differences in themicrobial populations at the two sites The higher Q10 val-ues for CO2 emissions at the LFS indicates that longer-term flooded sites could be more sensitive to climate-change-related warming (Zhou et al 2014)

43 Effect of redox potential on soil CO2and N2O fluxes

The weak correlation between redox potential and CO2 emis-sions as well as the absence of a relationship with N2Ofluxes (Sect 34) for both sites suggests soil redox potentialhad minimal influence on the emissions of either gas Thiscontrasts with the finding of Mariacuten-Muntildeiz et al (2015) whoidentified redox potential and water level as the main factorscontrolling GHG emissions in coastal freshwater wetlands

The redox potential of the two sites was different and thehighest CO2 emissions occurred at 220ndash362 and 145ndash259 mvat the SFS and LFS respectively These data do not implythat larger CO2 fluxes reflect more oxidised soil conditionsThe redox level of the soil is not critical as the highest CO2emissions at the LFS actually occurred in the lower end of theabove range of redox values The redox potential at the LFSincreases markedly and then remains constant soon after thedisappearance of standing water whilst there was no imme-diate response in terms of CO2 emissions Even at times (pe-riods C and E) when soil moisture contents are identical be-tween the sites CO2 emissions were higher at the LFS andredox values lower which could be due to the higher or-ganic matter content at the LFS Gardiner and James (2012)showed a marked decrease in redox potential (more nega-tive) as a result of organic matter addition in their wet soilmicrocosm study However the redox status of soil is alsocontrolled by the availability of electron acceptors and mi-

crobial activity (Oktyabrskii and Smirnova 2012 Huntingand Kampfraath 2013)

No particular redox range was confidently identified as be-ing associated with greater N2O emissions The fluxes weresomewhat higher when the redox value was above 249 mv atthe SFS and betweenminus232 and 228 mv at the LFS Similarlymany previous studies have reported higher N2O emissionsin flood-affected wetlands across a wide range of redox po-tentials (minus100 to 430 mv) (Yu et al 2001 Wlodarczyk et al2003 Morse et al 2012) Mariacuten-Muntildeiz et al (2015) foundan optimum range of 100ndash360 mv for the reduction of nitrateto N2O from a coastal wetland whereas Morse et al (2012)reported values of 89 and 53 mv from rarely and intermit-tently flooded areas respectively as the conditions conducivefor N2O production

44 Relationship between water depth and CO2 andN2O fluxes

The greater water depths at the LFS during flooding were ac-companied by a decrease in the rate of CO2 emissions per-haps due to a combination of diffusional constraints and de-creased oxygen availability as a result of higher standing wa-ter levels Water depth has been shown to be more significantthan temperature in determining the variation in CO2 fluxesduring inundation (Dixon et al 2014) Multiple regressionanalysis of CO2 fluxes with water depth and soil tempera-ture combined showed a significant paraboloid relationship(R2= 062 P lt 0001) (Fig 8) but most of this variation

was explained by changes in water depth alone (R2= 045

P lt 0001) (Fig 5c) Peak CO2 fluxes (above 75 mg CO2ndashC mminus2 hminus2) were recorded during periods B and D when wa-ter depths less than 9 cm were coupled with soil tempera-tures above 11 C No major variations in CO2 fluxes wereobserved when water depths exceeded 12 cm Other studieshave shown a significant relationship between N2O fluxesand water depth in riparian woodlands (Mander et al 2015Mariacuten-Muntildeiz et al 2015) but no such correlation was foundin this study (Sect 34) Audet et al (2013) also found nosignificant impact of water depth on N2O emission from atemperate riparian wetland

45 N2O fluxes and their controlling factors

No relationship was found between N2O fluxes and soilmoisture content or temperature at the LFS and only a weakrelationship at the SFS suggesting the influence of thesevariables on nitrification and denitrification processes lead-ing to N2O production from these sites is limited Severalother environmental factors such as pH oxygen concentra-tion and nitrogen availability are known to affect the produc-tion of N2O (Ullah and Zinati 2006 Van den Heuvel et al2011 Burgin and Groffman 2012) In this study the lowerpH as well as a higher nitrogen availability and higher MBCat the LFS would be expected to favour N2O emissions (Liu

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2623

Figure 8 Dependence of CO2 fluxes on water depth and soil tem-perature during the intervals in which the LFS is flooded A signifi-cant (R2

= 062 P lt 0001) 3-D paraboloid regression is shown inthe mesh plot Individual values shown are means (n= 3ndash4)

et al 2010 Van den Heuvel et al 2011) However apartfrom the constantly positive and higher N2O fluxes observedat the LFS for many of the sampling dates in period D noappreciable differences in N2O emission were detected be-tween the two sites The generally higher N2O emissionsfrom the LFS in period D might be associated with the per-sistent anaerobic conditions that would have favoured a tran-sient reduction of NOminus3 to N2O Higher soil NH+4 concentra-tions under largely anoxic conditions (periods A B and D)and its progressive decrease through the aerobic period (peri-ods C and E) in both sites (Table 2) indicate increased nitrifi-cation and ammonia oxidation that may lead to N2O produc-tion as aeration of soil and gas diffusion improves (Firestoneand Davidson 1989) This is a potential explanation for thelarger N2O emissions observed at the SFS during the largerpart of the first and on a few occasions during the secondsummer period (periods C and E) Continuous and relativelylow or negative N2O emissions were observed for both sitesduring the early part of periods C and E (ie after drainingof the flood water at the LFS) This is most likely due to in-creased uptake of N by growing vegetation that reduces theamount of inorganic N available for conversion to N2O andoffsetting any tendency for the increase in soil aeration topromote N2O formation This potentially also explains theabsence of N2O fluxes of a similar magnitude at the LFSduring the summer where unlike at the SFS the immediatedevelopment and recovery of grasses was likely impeded bythe preceding episode of prolonged waterlogging (Steffens etal 2005)

The annual N2O emissions estimated in this work arewithin the range of emissions obtained in restored temper-

ate wetlands in Denmark and the Netherlands (Hefting etal 2003 Audet et al 2013) Fisher et al (2014) reportedhigher annual N2O emissions (432 kg N2OndashN haminus1) fromflood prone riparian buffer zones in Indiana USA comparedto unflooded buffer zones (103 kg N2OndashN haminus1)

In terms of the contribution of each gas to the GWPexpressed as CO2 equivalents N2O contributed only 16 to GWP which is low compared to the CO2 contribution(84 ) We have not assessed the contribution of CH4 emis-sions which are often a major GHG in permanently inun-dated soils Whilst other studies of floodplains and fresh-water wetlands do report a high contribution of methane tothe GWP (Altor and Mitsch 2006 Koh et al 2009) thismay not always be the case For instance Jacinthe (2015)showed that some terrestrial riparian ecosystems whichwere exposed to variable flooding frequencies routinelyacted as a strong sink for CH4 except for a small con-tribution in emissions from permanently flooded sites Inmany ecosystems where flooding is intermittent or periodicand of a shallow depth CO2 is the dominant gaseous flux(eg Altor and Mitsch 2006 Jerman et al 2009 Morseet al 2012 Batson et al 2015 Jacinth 2015 Wintonand Richardson 2015) For example Morse et al (2012)showed that CO2 fluxes comprised 60 to 100 of the con-tribution (8000ndash64 800 kg CO2 haminus1 yrminus1) to the total GHGemissions from an intermittently and permanently floodedcoastal plain in contrast CH4 fluxes ranged from minus687 to197 kg CH4 haminus1 yrminus1 The highest emissions of CH4 werefrom the permanently flooded site Broadly similar findingswere reported by Batson et al (2015) where CH4 made nocontribution to the total GHG emissions from floodplain ar-eas with contrasting hydroperiods As the major period offlooding also occurred during the cooler period of the yearthe lower temperatures would have inhibited CH4 emissions

5 Conclusions

This study provides evidence that the interaction between agrassland ecosystem and its hydrologic regime impacts onthe annual emissions of greenhouse gases Flooding durationaffected the dynamics of CO2 and to a lesser extent N2Ofluxes The emissions of CO2 were higher with longer-term(sim 6 months) compared to shorter-term (sim 2ndash4 weeks) flood-ing Temperature and soil water content are identified as themost important factors controlling the seasonal pattern of theCO2 fluxes especially for the longer-term flooded site Thehigher emissions from the longer-term flooded site are likelylinked to the higher inputs of organic materialsnutrients andassociated increases in microbial biomass and possibly tochanges in the microbial populations In contrast no individ-ual environmental parameter or any combination of themwas found to have a major influence on the emissions of N2OHowever flooding enhanced N2O production during the pe-riod in which water was standing on the longer-term flooded

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

2624 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

site likely due to enhanced denitrification The controllingmechanisms underpinning the observed N2O fluxes are notclear However based on the information obtained from thecurrent study the contribution of N2O emissions to globalwarming in these ecosystems would be minimal A moreextensive study of the effect of specific hydrologic patterns(flooding frequency timing and duration) on the dynamicsof GHGs including CH4 would be required to better assessthe global warming potential of flood-affected ecosystems

Data availability Data are available on request by contacting thecorresponding author

Competing interests The authors declare that they have no conflictof interest

Acknowledgements This project is part of the Earth and NaturalSciences Doctoral Studies Programme which is funded by theHigher Education Authority (HEA) through the Programme forResearch at Third Level Institutions Cycle 5 (PRTLI-5) and isco-funded by the European Regional Development Fund (ERDF)We wish to thank BirdWatch Ireland in Wicklow for providing thestudy site for this research and their support during the process ofsite formation We thank the Associate Editor and reviewers fortheir constructive comments

Edited by A ItoReviewed by T K Yoon and one anonymous referee

References

Alongi D M Tirendi F Dixon P Trott L A and BrunskillG J Mineralization of Organic Matter in Intertidal Sedimentsof a Tropical Semi-enclosed Delta Estuar Coast Shelf S 48451ndash467 1999

Altor A E and Mitsch W J Methane flux from created riparianmarshes Relationship to intermittent versus continuous inunda-tion and emergent macrophytes Ecol Eng 28 224ndash234 2006

Altor A E and Mitsch W J Pulsing hydrology methane emis-sions and carbon dioxide fluxes in created marshes A 2-yearecosystem study Wetlands 28 423ndash438 doi10167207-9812008

Audet J Elsgaard L Kjaergaard C Larsen S E and HoffmannC C Greenhouse gas emissions from a Danish riparian wetlandbefore and after restoration Ecol Eng 57 170ndash182 2013

Badiou P McDougal R Pennock D and Clark B Greenhousegas emissions and carbon sequestration potential in restored wet-lands of the Canadian prairie pothole region Wetlands EcolManag 19 237ndash256 doi101007s11273-011-9214-6 2011

Bateman E J and Baggs E M Contributions of nitrifica-tion and denitrification to N2O emissions from soils at dif-ferent water-filled pore space Biol Fertil Soils 41 379ndash388doi101007s00374-005-0858-3 2005

Batson J Noe G B Hupp C R Krauss K W Rybicki NB and Schenk E R Soil greenhouse gas emissions and carbonbudgeting in a short-hydroperiod floodplain wetland J GeophysRes-Biogeo 120 77ndash95 doi1010022014JG002817 2015

Beniston M Stephenson D B Christensen O B Ferro C AT Frei C Goyette S Halsnaes K Holt T Jylhauml K KoffiB Palutikof J Schoumlll R Semmler T and Woth K Fu-ture extreme events in European climate an exploration of re-gional climate model projections Climatic Change 81 71ndash95doi101007s10584-006-9226-z 2007

Bossio D A and Scow K M Impacts of Carbon and Flooding onSoil Microbial Communities Phospholipid Fatty Acid Profilesand Substrate Utilization Patterns Microb Ecol 35 265ndash278doi101007s002489900082 1998

Brookes P C Landman A Pruden G and Jenkinson D Chlo-roform fumigation and the release of soil nitrogen a rapid directextraction method to measure microbial biomass nitrogen in soilSoil Biol Biochem 17 837ndash842 1985

Burgin A J and Groffman P M Soil O2 controls denitrificationrates and N2O yield in a riparian wetland J Geophys Res-Biogeo 117 G01010 doi1010292011JG001799 2012

Butterbach-Bahl K Baggs E M Dannenmann M KieseR and Zechmeister-Boltenstern S Nitrous oxide emis-sions from soils how well do we understand the processesand their controls Philos T R Soc B 368 20130122doi101098rstb20130122 2013

Christensen J and Christensen O A summary of the PRUDENCEmodel projections of changes in European climate by the endof this century Climatic Change 81 7ndash30 doi101007s10584-006-9210-7 2007

Davidson E A and Janssens I A Temperature sensitivity of soilcarbon decomposition and feedbacks to climate change Nature440 165ndash173 doi101038nature04514 2006

Dixon S D Qassim S M Rowson J G Worrall F Evans MG Boothroyd I M and Bonn A Restoration effects on watertable depths and CO2 fluxes from climatically marginal blanketbog Biogeochemistry 118 159ndash176 doi101007s10533-013-9915-4 2014

Doornkamp J C Coastal flooding global warming and environ-mental management J Environ Manage 52 327ndash333 1998

Eivazi F and Tabatabai M A Glucosidases and galactosidases insoils Soil Biol Biochem 20 601ndash606 1988

Firestone M K and Davidson E A Microbiological basis of NOand N2O production and consumption in soil in Exchange oftrace gases between terrestrial ecosystems and the atmosphereWiley New York NY 47 7ndash21 1989

Fisher K Jacinthe P Vidon P Liu X and Baker M Ni-trous oxide emission from cropland and adjacent riparian buffersin contrasting hydrogeomorphic settings J Environ Qual 43338ndash348 doi102134jeq2013060223 2014

Gardiner D T and James S Wet Soil Redox Chemistry as Af-fected by Organic Matter and Nitrate Am J Climate Change 1205ndash209 104236ajcc201214017 2012

Gee G W and Bauder J W Particle-size analysis1 in Meth-ods of Soil Analysis Part 1-Physical and Mineralogical Meth-ods edited by Klute A SSSA ASA Madison WI 383ndash4111986

Glatzel S Basiliko N and Moore T Carbon diox-ide and methane production potentials of peats from

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017

A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions 2625

natural harvested and restored sites eastern QueacutebecCanada Wetlands 24 261ndash267 doi1016720277-5212(2004)024[0261CDAMPP]20CO2 2004

Green V S Stott D E and Diack M Assay for fluorescein di-acetate hydrolytic activity Optimization for soil samples SoilBiol Biochem 38 693ndash701 2006

Hansen M Clough T J and Elberling B Flooding-induced N2Oemission bursts controlled by pH and nitrate in agricultural soilsSoil Biol Biochem 69 17ndash24 2014

Hefting M M Bobbink R and de Caluwe H Nitrous ox-ide emission and denitrification in chronically nitrate-loadedriparian buffer zones J Environ Qual 32 1194ndash1203doi102134jeq20031194 2003

Hernandez M E and Mitsch W J Denitrification Potential andOrganic Matter as Affected by Vegetation Community WetlandAge and Plant Introduction in Created Wetlands J EnvironQual 36 333ndash342 doi102134jeq20060139 2007

Hunting E R and Kampfraath A A Contribution of bacteria toredox potential (Eh)measurements in sediments Int J EnvironSci Technol 10 55ndash62 doi101007s13762-012-0080-4 2013

IPCC Summary for Policymakers in Climate Change 2007 ThePhysical Science Basis Contribution of Working Group I to theFourth Assessment Report of the Intergovernmental Panel onClimate Change edited by Solomon S D Qin D ManningM R Chen Z Marquis M Averyt K B Tignor M andMiller H L Cambridge University Press Cambridge UK andNew York NY USA 996 pp 2007

Jacinthe P Bills J Tedesco L and Barr R Nitrous ox-ide emission from riparian buffers in relation to vegeta-tion and flood frequency J Environ Qual 41 95ndash105doi102134jeq20110308 2012

Jacinthe P A Carbon dioxide and methane fluxes in variably-flooded riparian forests Geoderma 241ndash242 41ndash50 2015

Jerman V Metje M Mandic-Mulec I and Frenzel P Wet-land restoration and methanogenesis the activity of microbialpopulations and competition for substrates at different temper-atures Biogeosciences 6 1127ndash1138 doi105194bg-6-1127-2009 2009

Jonasson S Michelsen A Schmidt I K Nielsen E V andCallaghan T V Microbial biomass C N and P in two arcticsoils and responses to addition of NPK fertilizer and sugar im-plications for plant nutrient uptake Oecologia 106 507ndash515doi101007BF003297091996

Juutinen S Alm J Martikainen P and Silvola J Effects ofspring flood and water level draw-down on methane dynamics inthe littoral zone of boreal lakes Freshwater Biol 46 855-869doi101046j1365-2427200100721x 2001

Kandeler E Luxhoslashi J Tscherko D and Magid J Xylanaseinvertase and protease at the soilndashlitter interface of a loamy sandSoil Biol Biochem 31 1171ndash1179 1999

Kim Y Ullah S Roulet N T and Moore T R Effect of inunda-tion oxygen and temperature on carbon mineralization in borealecosystems Sci Total Environ 511 381ndash392 2015

Kirwan M L Guntenspergen G R and Langley J A Tempera-ture sensitivity of organic-matter decay in tidal marshes Biogeo-sciences 11 4801ndash4808 doi105194bg-11-4801-2014 2014

Koh H-S Ochs C A and Yu K Hydrologic gradient and veg-etation controls on CH4 and CO2 fluxes in a spring-fed forested

wetland Hydrobiologia 630 271ndash286 doi101007s10750-009-9821-x 2009

Lewis D B Brown J A and Jimenez K L Effects of flood-ing and warming on soil organic matter mineralization in Avi-cennia germinans mangrove forests and Juncus roemerianus saltmarshes Estuar Coast Shelf S 139 11ndash19 2014

Li X Hou L Liu M Lin X Li Y and Li S Primary ef-fects of extracellular enzyme activity and microbial commu-nity on carbon and nitrogen mineralization in estuarine andtidal wetlands Appl Microbiol Biotechnol 99 2895ndash2909doi101007s00253-014-6187-4 2015

Liu B Moslashrkved P T Frostegaringrd Aring and Bakken L R Denitri-fication gene pools transcription and kinetics of NO N2O andN2 production as affected by soil pH FEMS Microbiol Ecol72 407ndash417 doi101111j1574-6941201000856x 2010

Maag M and Vinther F P Nitrous oxide emission by nitrifica-tion and denitrification in different soil types and at different soilmoisture contents and temperatures Appl Soil Ecol 4 5ndash141996

Machefert S E and Dise N B Hydrological controls on den-itrification in riparian ecosystems Hydrol Earth Syst Sci 8686ndash694 doi105194hess-8-686-2004 2004

Mander Uuml Maddison M Soosaar K Teemusk A Kanal AUri V and Truu J The impact of a pulsing groundwater ta-ble on greenhouse gas emissions in riparian grey alder standsEnviron Sci Pollut Res 22 2360ndash2371 doi101007s11356-014-3427-1 2015

Mariacuten-Muntildeiz J L Hernaacutendez M E and Moreno-Casasola PGreenhouse gas emissions from coastal freshwater wetlands inVeracruz Mexico Effect of plant community and seasonal dy-namics Atmos Environ 107 107ndash117 2015

McNicol G and Silver W L Separate effects of flooding andanaerobiosis on soil greenhouse gas emissions and redox sen-sitive biogeochemistry J Geophys Res-Biogeo 119 557ndash566doi1010022013JG002433 2014

Morse J L Ardoacuten M and Bernhardt E S Greenhouse gasfluxes in southeastern US coastal plain wetlands under contrast-ing land uses Ecol Appl 22 264ndash280 doi10189011-052712012

Noe G Hupp C and Rybicki N Hydrogeomorphology Influ-ences Soil Nitrogen and Phosphorus Mineralization in Flood-plain Wetlands Ecosystems 16 75-94 doi101007s10021-012-9597-0 2013

Oelbermann M and Schiff S L Quantifying Carbon Diox-ide and Methane Emissions and Carbon Dynamics fromFlooded Boreal Forest Soil J Environ Qual 37 2037ndash2047doi102134jeq20080027 2008

Oktyabrskii O N and Smirnova G V Redox potential changesin bacterial cultures under stress conditions Microbiology+ 81131-142 doi101134S0026261712020099 2012

Peralta A L Ludmer S and Kent A D Hydrologic historyinfluences microbial community composition and nitrogen cy-cling under experimental dryingwetting treatments Soil BiolBiochem 66 29ndash37 2013

Rinklebe J and Langer U Microbial diversity in three floodplainsoils at the Elbe River (Germany) Soil Biol Biochem 38 2144ndash2151 2006

Samaritani E Shrestha J Fournier B Frossard E Gillet FGuenat C Niklaus P A Pasquale N Tockner K Mitchell

wwwbiogeosciencesnet1426112017 Biogeosciences 14 2611ndash2626 2017

2626 A W Gebremichael et al Flooding-related increases in CO2 and N2O emissions

E A D and Luster J Heterogeneity of soil carbon pools andfluxes in a channelized and a restored floodplain section (ThurRiver Switzerland) Hydrol Earth Syst Sci 15 1757ndash1769doi105194hess-15-1757-2011 2011

Schnuumlrer J and Rosswall T Fluorescein diacetate hydrolysis as ameasure of total microbial activity in soil and litter Appl Envi-ron Microbiol 43 1256ndash1261 1982

Smith K Ball T Conen F Dobbie K Massheder J and ReyA Exchange of greenhouse gases between soil and atmosphereinteractions of soil physical factors and biological processes EurJ Soil Sci 54 779ndash791 doi101046j1351-075420030567x2003

Steffens D Huumltsch B Eschholz T Lošaacutek T and Schubert SWater logging may inhibit plant growth primarily by nutrientdeficiency rather than nutrient toxicity Plant Soil Environ 51545ndash552 2005

Ullah S and Zinati G Denitrification and nitrous oxide emis-sions from riparian forests soils exposed to prolonged nitrogenrunoff Biogeochemistry 81 253ndash267 doi101007s10533-006-9040-8 2006

Unger I M Kennedy A C and Muzika R-M Flooding effectson soil microbial communities Appl Soil Ecol 42 1ndash8 2009

Vance E Brookes P and Jenkinson D An extraction methodfor measuring soil microbial biomass C Soil Biol Biochem 19703ndash707 1987

Van den Heuvel R Bakker S Jetten M and Hefting MDecreased N2O reduction by low soil pH causes high N2Oemissions in a riparian ecosystem Geobiology 9 294ndash300doi101111j1472-4669201100276x 2011

Van Der Heijden M G A Bardgett R D and Van Straalen NM The unseen majority soil microbes as drivers of plant di-versity and productivity in terrestrial ecosystems Ecol Lett 11296ndash310 doi101111j1461-0248200701139x 2008

Wang W J Dalal R C Moody P W and Smith C J Relation-ships of soil respiration to microbial biomass substrate availabil-ity and clay content Soil Biol Biochem 35 273ndash284 2003

Wilson J S Baldwin D S Rees G N and Wilson B P Theeffects of short-term inundation on carbon dynamics microbialcommunity structure and microbial activity in floodplain soilRiver Res Appl 27 213ndash225 doi101002rra1352 2011

Winton R S and Richardson C The Effects of Organic Mat-ter Amendments on Greenhouse Gas Emissions from a Mitiga-tion Wetland in Virginiarsquos Coastal Plain Wetlands 35 969ndash979doi101007s13157-015-0674-y 2015

Wlodarczyk T Stecircpniewska Z and Brzezinska M Denitrifica-tion organic matter and redox potential transformations in Cam-bisols Int Agrophysics 17 219ndash227 2003

Yu K Wang Z Vermoesen A Patrick Jr W and Van CleemputO Nitrous oxide and methane emissions from different soil sus-pensions effect of soil redox status Biol Fert Soils 34 25ndash30doi101007s003740100350 2001

Zhou W Hui D and Shen W Effects of soil moistureon the temperature sensitivity of soil heterotrophic respira-tion a laboratory incubation study PLoS ONE 9 e92531doi101371journalpone0092531 2014

Biogeosciences 14 2611ndash2626 2017 wwwbiogeosciencesnet1426112017