Greenhouse gas production from an intermittently dosed cold-climate wastewater treatment wetland S.H. Ayotte, C.R. Allen, A. Parker, O.R. Stein, E.G. Lauchnor © This manuscript version is made available under the CC-BY-NC-ND 4.0 license https:// creativecommons.org/licenses/by-nc-nd/4.0/ Made available through Montana State University’s ScholarWorks 1 GREENHOUSE GAS PRODUCTION FROM AN INTERMITTENTLY DOSED COLD-1 CLIMATE WASTEWATER TREATMENT WETLAND 2 S. H. Ayotte1,2,3, C. R. Allen1,2, A. Parker1,4, O.R. Stein1,2, E. G. Lauchnor1,2,3,*3 1 Center for Biofilm Engineering, Montana State University, Bozeman, MT 59717, USA 4 2 Department of Civil Engineering, Montana State University, Bozeman, MT 59717, USA 5 3 Thermal Biology Institute, Montana State University, Bozeman, MT 59717, USA 6 4 Department of Mathematical Sciences, Montana State University, Bozeman, MT 59717, USA 7 8 This study explores the greenhouse gas (GHG) fluxes of nitrous oxide (N2O), methane (CH4) and carbon 9 dioxide (CO2) from a two-stage, cold-climate vertical-flow treatment wetland (TW) treating ski area 10 wastewater at 3 °C average water temperature. The system is designed like a modified Ludzack-Ettinger 11 process with the first stage a partially saturated, denitrifying TW followed by an unsaturated nitrifying 12 TW and recycle of nitrified effluent. An intermittent wastewater dosing scheme was established for 13 both stages, with alternating carbon-rich wastewater and nitrate-rich recycle to the first stage. The 14 system has demonstrated effective chemical oxygen demand (COD) and total inorganic nitrogen (TIN) 15 removal in high-strength wastewater over seven years of winter operation. Following two closed-loop, 16 intensive GHG winter sampling campaigns at the TW, the magnitude of N2O flux was 2.2 times higher 17 for denitrification than nitrification. CH4 and N2O emissions were strongly correlated with hydraulic 18 loading, whereas CO2 was correlated with surface temperature. GHG fluxes from each stage were 19 related to both microbial activity and off-gassing of dissolved species during wastewater dosing, thus 20 the time of sampling relative to dosing strongly influenced observed fluxes. These results suggest that 21 estimates of GHG fluxes from TWs may be biased if mass transfer and mechanisms of wastewater 22 application are not considered. Emission factors for N2O and CH4 were 0.27% as kg-N2O-N/kg-23 TINremoved and 0.04% kg-CH4-C/kg-CODremoved, respectively. The system had observed seasonal 24 emissions of 600.5 kg CO2 equivalent of GHGs estimated over 130-days of operation. These results 25 indicate a need for wastewater treatment processes to mitigate GHGs. 26 27 Key Words: Nitrogen Removal, Constructed Wetland, GHG, N2O, CH4, CO2 28 *Corresponding Author. Email address: ellen.lauchnor@montana.edu (E. G. Lauchnor)29 30 31 32 33 34 mailto:ellen.lauchnor@montana.edu 2 HIGHLIGHTS 35 • Emissions varied between the two stages with higher CH4 and N2O in the first stage 36 • Fluxes of CH4, N2O and CO2 significantly increased during wastewater doses 37 • Mass transfer effects may cause flux overestimates in intermittently dosed systems 38 • High frequency of gas measurements is necessary to capture temporal variance 39 1 INTRODUCTION 40 Treatment Wetlands (TWs), a type of constructed wetland, are an alternative to traditional 41 mechanical wastewater treatment plants (WWTPs), combining mineral substrate, macrophytes, and 42 microorganisms to improve water quality (Dotro et al., 2017). TWs rely on microbially mediated 43 mechanisms similarly present in conventional WWTP for water quality improvements, and thus have 44 the potential to emit greenhouse gases (GHGs) including nitrous oxide (N2O), methane (CH4) and 45 carbon dioxide (CO2). Several recent studies have assessed microbially generated GHG emissions from 46 WWTP with emissions ranging from less than one to over several hundred (mg m-2 hr -1; Chen et al., 47 2020a; Fernández-Baca et al., 2018; Ma et al., 2011; Rodriguez-Caballero et al., 2015; Wang et al., 48 2019; Zhou et al., 2019). However, if TWs are to be considered a nature-based alternative to domestic 49 wastewater treatment, GHG emissions from TWs must be compared to those from traditional and 50 alternative nature-based treatment systems to assess their overall environmental impact. 51 Wastewater treatment in the United States (USA) is the second largest contributor of N2O, 52 accounting for 5.5% of total USA N2O emissions (EPA, 2022). The biotic and abiotic reactions that 53 occur during the nitrification and denitrification processes are the primary mechanisms of N2O 54 production during wastewater treatment (Faulwetter et al., 2013, 2009). These processes have 55 contributed to an increase in atmospheric N2O concentration, from a stable 275 parts per billion (ppb) 56 in the 18th century to 334 ppb by 2021, an 18% increase (EPA, 2022; Nakazawa and Matsuno, 2020; 57 Reay et al., 2012). Both conventional and nature-based wastewater systems can physically separate 58 microbial processes with different oxidation-reduction requirements, such as nitrification and 59 denitrification. This separation of processes allows for a more direct comparison of emissions, like N2O, 60 that can be generated by multiple distinct metabolic pathways. 61 3 Nitrification is a two-stage chemolithoautotrophic process mediated by two microbial symbionts 62 that oxidize ammonia (NH3) to nitrate (NO3 -) via the intermediary nitrite (NO2 -) (Shammas, 1986). N2O 63 production occurs during the first stage of this process when ammonia-oxidizing bacteria (AOB) oxidize 64 NH3 to hydroxylamine (NH2OH). The intermediate oxidation of hydroxylamine by AOB forms the 65 unstable intermediate nitroxyl (HNO) (Law et al., 2012) which can react with available NH2OH to form 66 hyponitrous acid (H2N2O2). Hyponitrous acid then chemically decomposes to N2O (H. Chen et al., 2020; 67 Duan et al., 2018; Wunderlin et al., 2012; Zhu-Barker et al., 2015). N2O production may increase 68 exponentially with NH3 oxidation, as seen in a previous study (Law et al., 2012). One study reported 69 that 86-96% of the N2O produced during nitrification occurred during NH2OH oxidation (Tumendelger 70 et al., 2019). Production of N2O by hydroxylamine oxidation is likely to occur under conditions of high 71 NH3 and low NO3 - which reflects influent domestic wastewater concentrations, indicating an 72 opportunity for N2O efflux during nitrification. 73 Denitrification is a four-step dissimilatory reduction of nitrate (NO3 -) to nitrogen gas (N2) by 74 facultative anoxic heterotrophic bacteria. Different metabolic enzymes (nitrate reductase, nitrite 75 reductase, nitric oxide reductase, and nitrous oxide reductase) catalyze each reduction step; N2O is 76 produced from nitric oxide (NO) reduction and is an obligatory intermediate of denitrification (H. Chen 77 et al., 2020; Faulwetter et al., 2009; Wunderlin et al., 2012). Per Law et al. (2012), it was determined 78 that the enzymatic reduction of N2O to N2 is almost four times faster than the rates of NO3 - or NO2 - 79 reduction, suggesting that N2O should not accumulate during heterotrophic denitrification. However, 80 environmental factors and slower release of N2O reductase may result in transient accumulation of N2O 81 (Law et al., 2012). Increased emissions of N2O in wastewater tend to result due to incomplete 82 denitrification, which can be further exacerbated by high levels of NO3 -, high temperatures, inadequate 83 carbon to nitrogen (C:N) ratios, sudden changes in dissolved oxygen conditions and short retention 84 times in the treatment process (Kaushal et al., 2014; Mander et al., 2021, 2011). 85 Methane is also produced during wastewater treatment processes which account for 2.8% of total 86 US CH4 emissions (EPA, 2022). In wastewater treatment, the production of CH4 is linked to biogenic 87 sources in treatment stages under highly reduced redox conditions and can account for an estimated 1% 88 4 of the influent COD load (Campos et al., 2016). CH4 production likely occurs in TWs due to less 89 stringent control of redox conditions in TW systems. Since CH4 production prefers highly reduced 90 conditions (-200mV), anaerobic environments must be well established in TWs; intermittent loading 91 and passive aeration may inhibit methanogenesis by increasing oxygenation of the system, resulting in 92 CH4 oxidation (Dotro et al., 2017). During wastewater treatment, CO2 is the most emitted GHG (by 93 volume) through heterotrophic respiration at both microbial and macrophyte levels (Oertel et al., 2016). 94 However, equilibrium with the aqueous carbonate cycle may influence observed emissions and be 95 attributed to the mineralization of organic carbon (Sharifian et al., 2021). 96 Many factors are correlated with TW GHG production, and findings are often contradictory. 97 Temperature has been found to significantly impact N2O and CO2 emissions in constructed wetlands, 98 with substrate and atmospheric temperatures positively correlated with emissions (Bahram et al., 2022; 99 Jiang et al., 2020; Kaushal et al., 2014; Mander et al., 2021). However, the relationship between 100 temperature and CH4 flux is more complex. Only some of the studies conducted on gas fluxes in TWs 101 have found a positive correlation between wetland substrate temperatures and CH4 flux in wetlands 102 treating domestic wastewater (Teiter and Mander, 2005). Several studies indicated that intermittent 103 loading of TWs significantly increased N2O (Hernandez and Mitsch, 2007, 2006) and CH4 flux relative 104 to continuous loading but had limited to no effect on CO2 emissions (Kaushal et al., 2014; Mander et 105 al., 2011). Other studies suggest pulse loading decreased CH4 flux (Altor and Mitsch, 2008). It is crucial 106 to understand the impacts of environmental and operational factors on GHG emissions in TWs not only 107 to compare to alternative systems, but also to both quantify GHG production and to discover potential 108 operational controls that might minimize GHG emissions from TWs. Several previous studies have 109 investigated GHG emissions from treatment wetlands, and to the best of our knowledge, no studies have 110 considered the combined effects of pulse loading and low temperature operation on emissions (Kasak 111 et al., 2022b; Mander et al., 2011, 2008; Teiter and Mander, 2005). Additionally, no other studies have 112 monitored GHGs from TWs with high frequency measurements capable of discerning temporal 113 variation under rapidly changing environmental or operational factors. 114 5 This study investigates GHG emissions from an intermittently loaded two-stage vertical flow TW 115 during winter when water temperature was approximately 3°C. The system was designed for total 116 nitrogen removal, with varying degrees of saturation to optimize for nitrification and denitrification 117 processes. The primary objective of this research is to quantify low-temperature GHG production with 118 a focus on N2O emissions by nitrogen-removal processes associated with fixed biofilms in the TW. Our 119 team predicts the production of GHGs follows patterns related to nutrient availability, controlled by 120 intermittent wastewater application. For this study, two hypotheses are investigated: (1) Nitrifying 121 bacteria produce similar N2O emissions to denitrifying bacteria during low-temperature TW operation, 122 and (2) the observed fluxes of N2O, CH4, and CO2 correlate with time of wastewater application. 123 2 MATERIAL AND METHODS 124 2.1 SITE DESCRIPTION AND SYSTEM OPERATION. 125 Gas emissions and water quality analyses were performed at a pilot-scale TW operating in the 126 winter season. The TW (Fig. 1) treats high-strength, domestic wastewater generated from toilets and 127 kitchens at the Bridger Bowl Ski Area near Bozeman Montana USA (45o49’01.0” N 110o54’21.8” W). 128 Mean seasonal air temperature and annual snowfall at the studied TW were –4.7 oC and 6.2 m, 129 respectively (Bridger Bowl Inc., 2023). The TW design is similar to a modified Ludzack-Ettinger 130 process for advanced nitrogen removal (Metcalf & Eddy, Inc. et al., 2014). It was designed as a two-131 stage, sub-surface vertical flow system where the first stage removes influent organic carbon (measured 132 as chemical oxygen demand; COD) and denitrifies NO3 - produced and recycled from the second stage. 133 Each stage is comprised of two cells in parallel (area per cell, AC = 23.8 m2; nominal depth, do = 0.9 134 m). The first stage is saturated to a depth of 0.71 m which contains a treatment layer (nominal thickness 135 = 0.9 m) comprised of crushed gravel media (d50 = 5.3 mm). The second stage is completely unsaturated 136 containing a washed concrete-sand treatment layer (nominal thickness = 0.9 m; d50 = 0.53 mm). In both 137 stages, the treatment layer sits over a 0.15 m drainage layer of coarser material and is covered with 0.1 138 m of crushed gravel media to provide insulation. 139 Influent wastewater first undergoes primary treatment in a series of sedimentation tanks. The 140 primary treated influent is dosed to the first stage independently from doses of water recycled from the 141 6 second stage. During the gas sampling campaign, the TW was treating 3.6 m3/d of influent and was 142 operating at a 2:1 (recycle:influent) v:v ratio. The first stage received influent wastewater doses every 143 eight hours, resulting in an areal dose depth of 2.5 cm (volume per dose, Vdose =1200 L). The effluent 144 of the second stage was recycled approximately every 90 minutes resulting in an areal dose depth of 0.8 145 cm (Vdose = 400 L). The second stage received the mixed influent and recycled effluent from the first 146 stage every four hours, resulting in a 5.1 cm (Vdose = 2400 L) areal dose depth. Carex utriculata (sedge) 147 and Scheneoplectus acutus (bulrush) were planted in 2013 and currently grow in the TW. Due to the 148 lack of unplanted controls and the inability to install the gas sampling collars and caps over plants, 149 offgassing from plants was not directly measured. Plants and their root exudates have been shown to 150 alter subsurface microbial communities (Faulwetter et al., 2013), which may influence GHG 151 production. The pore-space air in the non-saturated porous media was assumed to represent a bulk 152 average of all processes occurring in the treatment media and provide representative flux measurements. 153 2.2 GAS FLUX MEASUREMENTS. 154 Substrate gas fluxes were measured using the closed-loop dynamic chamber method with a 155 Picarro G2508 (Picarro Inc., Santa Clara, CA, USA) gas analyzer. Eight locations in the TW (four each 156 in the first and second stages; Fig. 1) were randomly selected to test spatial variability of gas fluxes 157 from the stages assumed to perform predominantly denitrification (first stage) or nitrification (second 158 stage), respectively. Any plant detritus was removed from the surface of the TW before inserting 20.3 159 cm diameter and 15.2 cm length polyvinyl-chloride (PVC) rings approximately 10 cm into the substrate 160 a minimum of 24-hrs before sampling. The interior space of the ring was insulated with a 5 cm thick 161 foam cap to maintain ground temperature and prevent snow accumulation on the exposed wetland 162 substrate surface prior to sampling. Approximately 60 cm of snow was removed from the area 163 surrounding the PVC collars prior to the start of each sampling campaign. 164 2.2.1 MANUAL SAMPLING. 165 In March 2022, gas efflux from the first (denitrification) and second (nitrification) stages in the 166 system was monitored over three days by manually placing a chamber over the inserted rings (Fig. 1.). 167 The chamber was made from a flat-bottomed polyvinyl chloride (PVC) endcap fitted with a rubber 168 7 gasket along the bottom lip creating a tight seal with the ring (Fig. A.1-A.4). Two air-tight Swagelok® 169 (Swagelok Co., Cleveland, OH, USA) sampling ports were located at the top of the chamber headspace 170 and a pressure vent was located on the side of the chamber. Each chamber (volume of 0.0042 m3) 171 covered a 0.032 m2 substrate surface area. A closed loop was created by connecting the chamber to the 172 Picarro G2508 gas analyzer with two 15 m lengths of 3.18 mm polyethylene-lined Bev A-line IV® 173 Tubing (USP®, Lime, OH, USA) attached to the sampling ports. The Picarro G2508 gas analyzer, which 174 uses cavity ring-down spectroscopy, monitored N2O, CH4, and CO2 gas concentrations at 1.17 Hz. Air 175 in the chamber was circulated through the closed loop at a rate of 1.8 L min-1 using a diaphragm pump 176 to prevent stratification and to establish continual mixing (Fig. A.5). 177 Gas concentrations were measured by placing the chamber on a sample ring over an eight-minute 178 sampling interval. The system was flushed with ambient air between each measurement. Sampling 179 rotated between four locations in each stage (Fig. 1) in a recurrent pattern over a two-day (first stage) 180 or one-day (second stage) period. A total of 40 sampling intervals (10 per location) in the first stage and 181 36 sampling intervals (9 per location) in the second stage were collected. The slopes of the gas 182 concentrations over each eight-minute sampling interval were used to calculate fluxes as described in 183 section 2.4. 184 2.2.2 AUTOMATED SAMPLING. 185 In April 2022, a 48-hr continuous monitoring at one location in the first (denitrification) stage 186 assessed gas efflux patterns associated with intermittent dosing schedules. A Li-COR® 8100-103 187 Survey Chamber (volume of 0.0068 m3) replaced the manual chamber and was connected to the Picarro 188 with two 30 m lengths of the Bev A-line IV tubing. The chamber closed automatically for a seven-189 minute read time. Flushing time with ambient air between measurements was three minutes, resulting 190 in a frequency of six sampling intervals per hour. All other factors remained the same as described for 191 manual sampling. The gas concentrations over time were collected to calculate gas flux, resulting in a 192 total of 309 samples in April. 193 2.3 ANCILLARY MEASUREMENTS. 194 8 Temperature measurements were collected via a thermistor located at the top of the LI-COR® 195 survey chamber (8100-104 Thermistor, LI-COR®, Lincoln, NE, USA). Additional hourly temperature 196 and pressure measurements were collected from the Bridger Base Area weather station (Bridger Bowl 197 Inc., Bozeman, MT, USA). Water samples were collected from the influent, effluent of the first stage 198 and final effluent from the system as either grab samples or as averaged composite samples over three-199 day periods. Continuous monitoring of the wetland water quality has occurred during the winter season 200 for the past ten years (2013-present), and variability of nutrient parameters is consistently low once the 201 system reaches a steady state (by January). COD was measured using HACH® digestion vials (HACH® 202 Co., Loveland, CO, USA). A modified Berthelot reaction approach, as described by Rhine et al. (1998), 203 was scaled to be performed in 96-well plates and used to determine ammonium (NH4 +-N) concentrations 204 on a BioTek Synergy HTX Multimode Reader (Agilent Technologies, Inc., Santa Clara, CA, USA) at 205 a wavelength of 660nm. NO3 - and NO2 - were measured on a Metrohm Eco ion chromatograph with a 206 Metrosep A Supp 5 150/4.0 column, Metrohm Suppressor Module and 3.20/1.00 mM sodium 207 bicarbonate/sodium carbonate eluent at 0.7 mL/min (Metrohm USA, Riverview, FL, USA). Total 208 inorganic nitrogen (TIN) was defined as the summation of measured inorganic nitrogen species and was 209 externally validated by MSU’s Environmental Analytical Lab using a Shimadzu TOC-VSH (Shimadzu 210 Scientific Instruments, Inc., Columbia, MD, USA). 211 2.4 GAS FLUX CALCULATION AND EMISSION FACTOR (EF). 212 Increases in gas concentration in the chamber headspace over each sampling interval were used to 213 calculate gas fluxes by the linear approach described in Livingston et al. (2006). Quality control (QC) 214 checks were completed on both the raw concentration data as well as the regressions of concentration 215 over time. Raw data included 7- or 8-minute sampling intervals. From these data, the initial two minutes 216 were considered a stabilization period and removed from the determination of slope. Quality control 217 checks were additionally performed on each of the 385 slopes calculated from the raw data. This QA/QC 218 used the linear regression R2 values for CO2 measurements to assess goodness of fit. If the R2 value for 219 CO2 was ≥ 0.9, the slopes for all gases were accepted. Slopes for CO2 with R2 < 0.9 were removed from 220 the dataset (Fig. A.6). After quality control checks, a combined 368 slopes (95.56%) were used to 221 9 calculate gas flux in March and April. The measured change of gas concentration versus time were 222 combined with the measured chamber volume, ambient air pressure, chamber temperature and surface 223 area to estimate the flux using Eq. (1). 224 𝐽𝐽𝑐𝑐 = 𝑉𝑉∙𝑃𝑃𝑜𝑜 𝑅𝑅∙𝐴𝐴∙(𝑇𝑇+273.15) ∙ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 (1) 225 where Jc is the gas efflux (mg m-2 hr-1), V is the combined volume of the gas chamber and the tubing 226 (m3), Po is the ambient pressure (Pa), R is the gas rate constant (8.314 Pa m3 mol-1K-1), A is the surface 227 area of the chamber (m2), T is the initial chamber air temperature (oC), and 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 is the rate of gas 228 accumulation over time, expressed as a gas mole fraction (µmol mol-1 s-1). 229 The emission factor (EF) is a measure of the ratio of GHG produced relative to the daily mass of 230 nitrogen or carbon removed (kg N2O-N/kg TINremoved; kg CH4-C/kg CODremoved) and is reported as a 231 percentage or fraction (IPCC, 2019). The IPCC (2019) has not officially specified whether the total load 232 or mass removal should be used to determine the EF. For this paper, the treated or removed mass of 233 carbon or nitrogen was used to determine the EF. The EFs were calculated for CH4 and N2O using the 234 Tier I approach by IPCC (2019) as shown in Eqs. (2) and (3) and compared various wastewater 235 treatment facilities and constructed wetlands. 236 𝐸𝐸𝐸𝐸 = (𝐽𝐽𝑁𝑁2𝑂𝑂−𝑁𝑁∙𝐴𝐴) (𝑇𝑇𝑇𝑇𝑇𝑇𝑖𝑖𝑖𝑖−𝑇𝑇𝑇𝑇𝑇𝑇𝑒𝑒𝑒𝑒𝑒𝑒) (2) 237 𝐸𝐸𝐸𝐸 = (𝐽𝐽𝐶𝐶𝐶𝐶4−𝐶𝐶∙𝐴𝐴) (𝜕𝜕𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖−𝜕𝜕𝐶𝐶𝐶𝐶𝑒𝑒𝑒𝑒𝑒𝑒) (3) 238 where EF is emission factor of the gas of interest expressed as a mass ratio (kg / kg removed), JN2O-N is 239 the N2O gas flux from system (kg N2O-N m-2 day-1), JCH4-C is the CH4 gas flux from system (kg C m-2 240 day-1), A is the total surface area of the TW system (47.6 m2 per stage), TINin is the influent daily mass 241 flux of total inorganic nitrogen through the system (kg day-1), TINeff is the effluent daily mass flux of 242 total inorganic nitrogen (kg day-1), CODin is the influent daily mass flux of COD (kg day-1), and CODeff 243 is the effluent daily mass flux of COD (kg day-1). Daily mass flux of TIN and COD were calculated as 244 the product of aqueous concentrations (mg L-1) and the system’s daily volumetric flow rate (m3 day-1) 245 after unit conversions. 246 10 The EF was not determined for CO2 due to the IPCC (2019) guidance that determined organic 247 carbon found in domestic wastewater to be from modern organic matter found in human excrement and 248 is thus considered to be predominantly biogenic. Consequently, since this carbon is not related to the 249 transfer of ancient carbon stores from the lithosphere to the atmosphere, reporting CO2 emission factors 250 from domestic wastewater is currently excluded from GHG inventories (IPCC, 2019). Additionally, no 251 supplemental carbon was added to the system and this study considered only the measured direct 252 emissions from the TW surface, excluding any potential non-biogenic sources that would be used to 253 calculate a CO2 EF according to current reporting standards (IPCC, 2019). 254 The CO2-equivalent (CO2eq) was determined using each GHG's global warming potential 255 (GWP). A 100-year GWP of 27 and 273 were used for CH4 and N2O, respectively (IPCC, 2021). 256 Equation (4) shows how the CO2eq was determined: 257 𝐶𝐶𝐶𝐶2𝑒𝑒𝑒𝑒 = 𝐽𝐽𝑐𝑐 ∙ 𝑀𝑀𝑀𝑀 𝐴𝐴𝑀𝑀 ∙ 𝐺𝐺𝐺𝐺𝐺𝐺 (4) 258 where CO2eq is the carbon dioxide equivalent (mg CO2e m-2 h-1), MW is the molecular weight of the 259 compound of interest (CH4, N2O or CO2; mg/mmol), AW is the atomic weight of the element of 260 interest (C or N; mg/mmol), and GWP is the 100-year global warming potential of CH4 or N2O (mg 261 CO2/mg GHG). 262 2.5 DATA ANALYSIS 263 R version 4.2.1 (R Core Team, 2022) was used to analyse and visualize all data. Analysis of 264 covariance (ANCOVA) investigated spatial influences on gas flux, such as system stage, and sample 265 location. Interactions between time*stage and time*location were assessed in separate ANCOVA 266 models to account for temporal effects. 267 Generalized Additive Models (GAM) were used to model temporal trends associated with 268 intermittent dosing schedules. GAM uses smoothers to uncover nonlinear relationships with covariates 269 (Hastie, Trevor and Tibshirani, Robert, 1986) and was essential for determining patterns within our 270 dataset. The package “mgcv” (Wood, 2017, 2011) was used to fit GAMs with tensor product 271 interactions and P-splines, as proposed by Eilers & Marx (1996). The restricted maximum likelihood 272 (REML) approach was used to provide unbiased estimates by accounting for loss in degrees of freedom 273 11 from estimating fixed effects (Harville, 2023). A flux was calculated by Eq (1) every ten minutes after 274 each wastewater dose (primary-clarified influent or recycled effluent). Measurements were grouped 275 into eight-hour blocks determined by the occurrence of an influent dose. GAMs were fit to N2O, CO2 276 and CH4 fluxes with smoothers for absolute time, time after the influent dose and time after the recycle 277 dose. N2O and CO2 models additionally included chamber temperature; however, chamber temperature 278 did not strongly influence CH4 flux (p > 0.05) and was not included in that model. Imputation was used 279 to assure time points at 10-minute intervals which allowed an auto-regression and moving average 280 (ARMA) time series model to be fit. Gas flux was log-transformed for all analyses to meet Gaussian 281 and constant variance assumptions. All model assumptions were met and validated using diagnostic 282 plots of residuals, including Autocorrelation Function and Partial Autocorrelated Function plots to 283 assess ARMA model fit. For each GAM-ARMA fit, the Benjamini-Hochberg method was applied to 284 maintain a family-wise false discovery rate of 5% (α = 0.05). 285 3 RESULTS 286 During March and April of the 2021-22 operational season, the TW removed COD, TIN and 287 NH3 from the influent wastewater, with results averaged from 20 samples provided in Table 1. Influent 288 concentrations of COD (808 mg/L) and TIN (161 mg/L) were higher than typical domestic wastewater 289 due to the lack of laundry and shower facilities at the resort. The system removed an average of 96.6% 290 influent COD. The TW removed nearly 3/4 of the total inorganic nitrogen (NH4 +-N, NO3 --N, and NO2 -291 -N), removing 74.3%. Ammonium levels in the second-stage effluent were consistently below the 292 detection limit of < 0.1 mg/L NH4 +-N, resulting in greater than 98% removal of NH4 +-N. The first stage 293 also exhibited an average of 73.7% NO3 --N and 24.6% NH4 +-N removal. Nitrate was added to the first 294 (denitrification) stage via recycle from the second (nitrification) stage effluent at twice the influent flow 295 rate, resulting in an average of 26 mg/L NO3 --N applied to the first stage at a one-part influent to two-296 part recycle ratio. 297 3.1 FIRST STAGE VS. SECOND STAGE GAS FLUX – MARCH SAMPLE CAMPAIGN 298 12 The median GHG fluxes (Table 1) from the first stage were 0.66 mg N2O-N m-2 h-1, 0.85 mg 299 CH4-C m-2 h-1, and 221 mg CO2-C m-2 h-1. Median second stage emissions were 0.34 mg N2O-N m-2 h-300 1, 0.2 mg CH4-C m-2 h-1, and 606 mg CO2-C m-2 h-1. Figure 2 illustrates the variability of N2O, CH4, and 301 CO2 gaseous efflux at each sampling location in March 2022. The ranges of fluxes in March were 0.025 302 - 4.66 mg N2O-N m-2 h-1, 0.029 - 28.07 mg CH4-C m-2 h-1, and 81.10 - 1062.49 mg CO2-C m-2 h-1. Since 303 samples were collected over the course of several days, a two-way ANCOVA with a time dependent 304 interaction was used to model the data. Sample locations within each stage had no statistically 305 significant difference in emissions (p ≥ 0.16). Although there was not a significant difference, the 306 proximity of a sample location to a dosing orifice may explain increased variability of emissions at that 307 location. 308 Emissions of N2O, CH4 and CO2 were found to be different between stages (p ≤ 0.05). 309 The first stage of the system, where denitrification and potentially methanogenesis occur, produced 310 greater emissions of the more potent greenhouse gases, N2O and CH4 (Fig. 2). Conversely, the second 311 stage, designed primarily for nitrification, emitted more CO2, despite lower influent COD levels. The 312 median N2O flux was 2.2 times greater and the median CH4 flux was 4.8 times greater in the anoxic 313 first stage than the aerobic second stage, while CO2 efflux was 2.46 times greater in the second stage. 314 Higher emissions in the first stage were generally observed immediately after active dosing, 315 suggesting that the temporal variability in gas fluxes was related to the time of the sampling interval 316 relative to the dosing schedule. Other studies have noted high variability of gas emissions in 317 intermittently dosed wetlands (Ji et al., 2021; Mander et al., 2014, 2011), but have not considered 318 instantaneous shifts in emissions relative to dosing. 319 3.2 TEMPORAL FLUCTUATIONS WITH HYDRAULIC LOADING -APRIL SAMPLE CAMPAIGN 320 To further understand temporal dependence, a 48-hour continuous monitoring of one location in 321 the denitrifying stage was conducted in the first week of April 2022. Frequent flux measurements at a 322 single location and over a longer duration provided more clarity to observe trends. For reference, the 323 first stage received primary-treated influent doses every 8-hours, with recycle doses loaded six-times 324 13 over each 8-hour cycle, resulting in a periodic introduction of organic carbon 3-times per day and NO3 - 325 18-times per day. The second stage received NH3 rich doses every 4-hours (6 times per day). Emissions 326 from the second stage in March indicated no statistically significant differences (p ≥ 0.06) with time; 327 therefore, temporal trends of emissions were considered only in the first stage. 328 Calculated gas flux and temperature exhibited trends with high variability over a 48-hour period 329 at one location in the first stage (Fig. 3). Temperature in the air and the chamber varied over the 48-330 hour sampling period from a low of -9 oC to a high of 13 oC. As expected, large peaks of N2O and CH4 331 flux corresponded to influent doses, other peaks in N2O and CO2 generally align with the smaller recycle 332 doses. Gas fluxes during the April sampling campaign ranged from 0.067 - 6.03 mg N2O-N m-2 h-1, 333 0.065 - 61.71 mg CH4-C m-2 h-1, and 27.38 - 927.27 mg CO2-C m-2 h-1. The median values were 0.66 334 mg N2O-N m-2 h-1, 0.85 mg CH4-C m-2 h-1, and 221 mg CO2-C m-2 h-1. The median and variance of N2O 335 and CH4 emission were higher than observed from the first stage in March, perhaps due to the tighter 336 sampling frequency picking up higher emissions around the timing of a dose. 337 The calculated fluxes were modeled by a GAM-ARMA that shows the overall smoothed trends 338 over the dose cycles and corresponds to the intermittent loading of wastewater (Fig. 4). The GAM 339 highlights the temporal variability due to the dose schedule. 340 3.2.1 NITROUS OXIDE EMISSIONS 341 N2O exhibited strong trends with dosing schedules (Fig. 4a). The median N2O flux across all 342 cycles for the first hour after an influent dose high in COD was 0.78 mg N2O-N m-2 hr-1 (95% CI: 0.66-343 0.92). The flux then decreased by 76% (p < 0.001) during hour two across all cycles to a minimum 344 value of 0.19 mg N2O-N m-2 hr-1 (95% CI: 0.16-0.22). A steady linear increase was observed from hours 345 two to five (p < 0.05), followed by a stable period of flux (p > 0.05) from hours five to eight with a 346 median flux of 0.99 mg N2O-N m-2 hr-1 (CI: 0.94-1.03). Secondary peaks were noted throughout the 347 eight hours and corresponded to the onset of each recycle dose, which contained high concentrations of 348 NO3 - (doses intervals not shown). The differences in median N2O flux between cycles were strongly 349 correlated to temperature effects (Fig. 3). Cycles with similar average temperatures resulted in N2O 350 fluxes that were not greatly different (p > 0.05). 351 14 3.2.2 METHANE EMISSIONS 352 The gas efflux of CH4 was at a maximum at the start of each influent dose cycle and displayed 353 a general exponential decrease thereafter (Fig. 4b). In the first hour of each cycle, the median flux was 354 9.24 mg CH4-C m-2 hr-1 (95% CI: 7.5-11.3) which rapidly decreased to 5.4 mg CH4-C m-2 hr-1 (95% CI: 355 4.4-6.5) by the second hour, a decline of 41.5% (p < 0.001). By hour seven, the flux reached a steady 356 state of 0.18 mg CH4-C m-2 hr-1 (95% CI: 0.15-0.21), which was 2% of the initial flux. A comparison of 357 the median flux between each hour within a cycle showed that all hours were significantly different (p 358 < 0.001), except for hours seven and eight, which corresponded to the steady state flux at the end of the 359 cycle. The high initial rates of CH4 emissions were linked to the large influent wastewater doses. Median 360 fluxes over entire dose cycles ranged from 0.70 to 1.12 mg CH4-C m-2 hr-1. 361 3.2.3 CARBON DIOXIDE EMISSIONS 362 Spikes in CO2 emissions corresponding to both the three influent doses and the eighteen recycle 363 doses per day are apparent, but repeating patterns of CO2 emissions within the 8-hour influent dose 364 cycle were not as obvious as those observed for N2O and CH4 (Fig. 4c). The six eight-hour cycles were 365 compared using a paired-mean comparison approach and data were grouped into cycles numbered 366 chronologically one to six (Fig. 4c). A moderate decrease with time within the cycle was observed 367 across all cycles except cycle three, the only cycle in which the surface temperature steadily increased 368 throughout the 8-hour period (Fig. 3). Median CO2 fluxes for cycles one, three, five, and six were not 369 different (p > 0.05), however, the median values for cycles two and four were significantly different 370 from the other four (p < 0.05), in a pattern similar to the findings for N2O. Cycle 2 had both the lowest 371 emissions and lowest average temperature, cycle 4 had the highest emissions and highest temperature 372 and median values over a cycle strongly correlated to surface temperature (p < 0.001). The initial 373 median CO2 flux at the start of a cycle was calculated to be 275.1 mg CO2-C m-2 hr-1 (95% CI: 246.7-374 306.7), which then decreased by 35.1% (p < 0.001) to a median of 178.5 mg CO2-C m-2 hr-1 (95% CI: 375 161.8-197.0) by the eighth hour. 376 3.3 GLOBAL WARMING POTENTIAL AND EMISSION FACTORS 377 15 For estimates of seasonal emissions, the Global Warming Potentials (GWP) as CO2 equivalents 378 (IPCC, 2021) were determined for the GHGs across the entire TW system using Eq (4). Daily, the TW 379 system contributed 1.0 kg of CO2eq of N2O and a combined CH4 and CO2 of 3.5 kg CO2eq. Over the 380 2021-2022 ski season, the total estimated GWP of the TW was 600.4 kg of CO2eq assuming 130-days 381 of operation, of which CH4 and N2O emissions accounted for 0.9% and 24.2%, respectively. 382 Emission factors of GHG are typically determined by the ratio of mass emitted per mass removed 383 as nitrogen or carbon (for example: kg N2O-N/ kg TIN-N removed) (Foley et al., 2010). However, EF 384 is not a standardized metric and is often provided in terms of influent daily mass, which can influence 385 the EF value (Tumendelger et al., 2019). Emission factors used in this study were standardized to carbon 386 and nitrogen daily mass removal. The studied TW had a system EF of 0.04% and 0.27% for CH4 and 387 N2O, respectively. 388 4 DISCUSSION 389 4.1 WATER QUALITY 390 The TW effectively removed influent nitrogen and COD to similar or better levels than more 391 energy- and labor-intensive mechanical systems. Pollutant removal efficiencies in the studied system 392 met or exceeded discharge requirements set by the Montana Department of Environmental Quality for 393 Level II treatment systems, which require oxidation of NH4 + and 60% total nitrogen (TN) removal. 394 Removal of NH4 + in the pilot system corresponded to treatment efficiencies observed in full-scale 395 mechanical wastewater treatment facilities, such as a conventional activated sludge (CAS) system and 396 a modified Ludzack-Ettinger (MLE) system (Table 2) which removed 98.2%and 91.6% of NH3 397 (Tumendelger et al., 2019). The mean removal efficiencies of COD and TIN in the studied TW (Table 398 1) were similar to a hybrid-constructed wetland treating anaerobic digestate, which removed 72.5% and 399 94.6% of COD and TN, respectively (Zhou et al., 2020). Both TW systems treated concentrated influent, 400 with NH4 + levels of the studied TW between two and eight times higher than typical domestic 401 wastewater (Henze, 2008). 402 4.2 EMISSION FACTORS. 403 16 The emission factor for N2O, which was 0.27% (Table 2), was only minimally lower than 404 emissions from a similarly designed hybrid TW treating anaerobic digestate, with an EFN2O-N of 0.34% 405 (Zhou et al., 2020). If only TIN loading was accounted for, the EFN2O-N of the TW was 0.21%, which is 406 only slightly lower than the EF value determined based on removal. The EFN2O-N for the studied TW fell 407 within the reported range of various TWs; however, traditional full scale WWTPs emitted two to twelve 408 times more N2O than the studied system. EFN2O-N values calculated in terms of influent TN are indicated 409 with the letter ‘C’ in Table 2. Only a few studies reported the EF for CH4 (Table 2); however, the range 410 of CH4 emissions was heavily skewed by whether EF was calculated based on influent carbon or overall 411 removal. Based on these results, the studied TW appears to generate less N2O and CH4 than most 412 mechanical treatment plants. Additionally, some studies have reported that macrophytes in wetlands 413 can affect GHGs depending on plant species (X. Chen et al., 2020; Faulwetter et al., 2009; Hernández 414 et al., 2018; Jiang et al., 2020; Mander et al., 2021, 2014; Maucieri et al., 2019). One study estimated 415 that in optimal conditions, anywhere from 1.56 x10-2 to 2.08 x10-2 kg CO2-C m-2 day-1 may be 416 assimilated by plants (Mander et al., 2008). In Montana, the typical growing season is 135 days or less 417 (Western Agricultural Research Center, 2019), resulting in the potential sequestration of 181.8 to 243.4 418 kg CO2 per 135-days or 30-40% of the generated CO2eq during our winter operation (~91% plant 419 coverage from a 2022 survey of the pilot TW). However, assimilation is typically unstable and may be 420 easily mineralized when the water table is lowered (Mander et al., 2014). Research of TW meso-cosms 421 indicated that plants universally increased CO2 emissions, regardless of season (Allen, 2016). As a 422 result, the studied TW is more likely a net emitter of CO2 than a sink. Since this study was performed 423 during plant senescence, uptake and active transport of gas by plants was assumed to be negligible. The 424 effect of plants on passive movement of gases in TWs due to open stomates and broken stems has been 425 hypothesized (Armstrong, 2000; Verboven et al., 2014), but was beyond the scope of this investigation. 426 Future studies on systems operating over an annual plant growth cycle and with unplanted controls 427 would be necessary to accurately evaluate plant impact on the release of the GHGs studied here. 428 4.3 PRODUCTION VS EMISSIONS 429 17 Often GHG emissions are assumed to be produced at the location of measurements (Mander et 430 al., 2014). However, in TWs and wastewater treatment systems, convective transport and aqueous water 431 chemistry may play a bigger role than often considered. As shown in Figure 5, the hydraulic design of 432 TWs may result in physical mechanisms that increase emissions via mass transfer, including 1) the 433 volatilization of dissolved gases at the dosing orifices, 2) the displacement of gas in the unsaturated 434 porous media by large volume doses, and 3) the release of large concentrations of dissolved gases when 435 the gas-liquid surface area increases in unsaturated or aerated systems (Bruun et al., 2017; Doran, 1995; 436 Foley et al., 2010; Law et al., 2012; Sharifian et al., 2021; Tumendelger et al., 2019). Mass transfer 437 effects may cause incorrect assumptions of where the GHG is generated across wastewater systems, 438 leading to over or underestimations of production by different microbial processes. Interestingly, most 439 studies link the increased emissions from intermittent dosing to biological factors such as nitrification 440 and denitrification. However, physical processes associated with mass transfer, displacement, and off-441 gassing likely confound changes in emissions. The effect of mass transfer on GHGs is discussed in 442 Sections 4.3.1-4.3.3. 443 4.3.1 NITROUS OXIDE 444 The mostly anoxic first stage of the wetland effectively contributed 68.8% of the total N2O 445 emissions from the system corresponding to the common paradigm that N2O is predominantly generated 446 by anoxic denitrification (Meyer et al., 2005; Velthuis and Veraart, 2022; Vilain et al., 2014). While 447 emissions in the second stage were likely due only to nitrification, NH3 removal in the first stage 448 indicated that while denitrification dominated, NH3 removal also contributed to TIN removal and likely 449 N2O emissions. Traditional mechanical wastewater treatment plants comparatively observe the opposite 450 trend, where substantially higher emissions of N2O are measured in aerated nitrification reactors. For 451 example, one MLE in Ruelzhiem, Germany, found that 21.1% of emissions were associated with 452 nitrification, whereas only 7% were associated with denitrification (Tumendelger et al., 2019). 453 Additional studies have found that N2O emissions were approximately 2-3 times greater in reactors 454 undergoing active aeration (Law et al., 2012). Active aeration, paired with the high mass transfer 455 18 coefficient of N2O, may quickly strip any generated N2O to the atmosphere resulting in greater observed 456 emissions (Foley et al., 2010). 457 The temporal trends associated with N2O in the first stage may be linked to mass transfer rather 458 than changes in the biological production of the gas. During dosing of influent wastewater, in which 459 dissolved N2O gas is likely minimal, we noted relatively large peaks in the N2O flux. During these 460 instances, it is hypothesized that N2O accumulated in the unsaturated porous volume of the gravel media 461 (top 0.19 cm of first stage) and was displaced by the doses, as described in mechanism 2 (Fig. 5). 462 Influent doses (2.5 cm dose depth) displace approximately 30% of the unsaturated gravel pore space, 463 assuming a conservative 40% porosity, potentially resulting in a large flushing of N2O gas to the 464 atmosphere. The same mechanism could also explain the smaller peaks in flux associated with the 465 smaller (0.8 cm) recycle doses which displace approximately 10% of the pore volume. Other studies of 466 wetlands have noted significant increases in N2O emissions due to pulse loading hydrology (Mander et 467 al., 2011); while another study reported seeing no impact on the total net export of N2O (Bruun et al., 468 2017). Mander et al. (2011) suggested that the length and frequency of individual doses may be the 469 reason for contradictory findings. 470 4.3.2 METHANE 471 The average CH4 flux from the first stage was 4.9 times greater than that observed from the 472 second stage, which may be attributed to differences in organic carbon loading, degree of saturation in 473 the first stage that favors anaerobic processes such as methanogenesis or volatilization of dissolved 474 gases at the dosing orifices. The rapid increase and subsequent decrease of CH4 emissions from the first 475 stage (Fig. 4b) suggests that dissolved CH4, likely produced in the septic tank, was off-gassed during 476 doses as described in mechanism 1 (Fig. 5). One study reported approximately 1% of influent COD 477 may be converted to CH4 in the anaerobic sewer distribution system (Campos et al., 2016), and others 478 have reported high dissolved CH4 concentrations in influent wastewater suggesting that CH4 may be 479 transported and off-gassed from preceding stages (Tumendelger et al., 2019). Mander et al. (2011) 480 observed 7-12 times higher CH4 emissions from TWs that were intermittently loaded or had fluctuating 481 water tables compared to systems with constant water tables. In all these studies, large increases in CH4 482 19 emissions appear to be attributable to mass transport and ebullition. The low observed CH4 fluxes by 483 hour two of each cycle in the present study are likely more indicative of the inherent methanogenic 484 capacity of the first stage, but measurement of dissolved gases in future investigations will be needed 485 to confirm this hypothesis. Ultimately, release of CH4 from the influent water may overestimate 486 emissions actively produced in the wetland but are representative of the CH4 generation across the entire 487 wastewater treatment system. 488 4.3.3 CARBON DIOXIDE 489 The CO2 flux from the first stage was 59% lower than the second stage flux, even though 31 490 times more COD was removed in the first stage (combined daily load of influent and recycle doses) 491 than the second stage (Table 1). Emissions from the first stage are smaller and emissions from the 492 second stage are greater than the COD removed in each stage on a mass carbon basis (assuming 0.33 g 493 C per g COD, Dubber and Gray, 2010). This strongly suggests that CO2 flux measured from each stage 494 is not representative of the overall heterotrophic activity in that stage. The removal of COD in the first 495 stage of the system confirms mineralization of organic carbon present in the influent wastewater by 496 heterotrophic processes, including denitrification. However, the resulting inorganic carbon may be 497 dissolved as carbonate species and not released as CO2. Additionally, since CO2 is highly soluble, 498 significant quantities may have been transported with water to the second stage and off-gassed in the 499 unsaturated second stage by mechanism 1 or mechanism 3 (Fig. 5). The second stage offers greater 500 unsaturated surface area than the first stage, allowing for the volatilization of dissolved inorganic carbon 501 due to interphase mass transfer (Doran, 1995). Additionally, nitrification in the second stage consumes 502 alkalinity, resulting in a potentially lower pH and an increase of CO2 off-gassing (Sharifian et al., 2021). 503 Although pH and dissolved gases were not measured during the time of gas measurements in this study, 504 future investigations should monitor pH and dissolved inorganic carbon to better understand carbonate 505 system dynamics that contribute to emissions. 506 4.4 TEMPORAL INFLUENCE OF HYDRAULIC LOADING ON FIRST STAGE 507 The discrepancy in emissions due to pulse loading has the potential to skew observed fluxes 508 depending on sampling strategy. The majority of studies utilize the static chamber method (Bruun et 509 20 al., 2017; Johansson et al., 2004; Mander et al., 2014, 2011; Teiter and Mander, 2005; Tumendelger et 510 al., 2019; Zhou et al., 2020), where singular concentration measurements are taken periodically over 511 the course of an hour as grab samples. The static chamber method of sampling is infrequent and is more 512 suitable for long-term temporal observations over seasons and years, resulting in more smoothed 513 emission trends. However, to observe the instantaneous influence of pulse or intermittent loading, a 514 higher sampling frequency, as was completed in this study, is required to encompass the overall 515 fluctuations of emissions. If the static chamber method is employed immediately after a dose, emissions 516 may be falsely elevated, whereas samples taken hours later may be significantly lower. Both changes 517 in microbial production and in mass transfer due to intermittent dosing can create variability in the 518 observed emissions over a dosing schedule. However, further studies are required to separate the effect 519 of each factor, and in systems such as TWs, these factors have complex and interrelated effects on 520 emissions. 521 5 CONCLUSIONS 522 1. Emissions of CH4 and N2O were significantly higher in the first stage compared to the second 523 stage of the TW (Table 1 & Fig. 2). As a result, our initial hypothesis that emissions between 524 processes would be similar is not supported, nor can we effectively attribute differences in 525 emissions directly to changes in nitrification or denitrification. Chamber temperature was 526 correlated with N2O and CO2 emissions, but not CH4 emissions. 527 2. CO2 emissions were nearly 2.5 times higher in the second stage despite low organic carbon 528 loading (Table 1 & Fig. 2). The discrepancies in the carbon balance are hypothesized to be the 529 result of dissolved inorganic carbon produced in the first stage that are then off-gassing in the 530 second stage. Future research should more thoroughly track the carbon balance of the system. 531 3. We hypothesized a dynamic behavior of N2O, CH4 and CO2 emissions due to intermittent 532 hydraulic loading. Data confirmed statistically significant patterns in emissions that emerged 533 with hydraulic loading (Fig. 4). However, these patterns pointed more strongly to the impacts 534 of convective transport and mass transfer rather than microbial production. Median emissions 535 were within the range of other studied systems; however, the findings justify the need for more 536 21 frequent sampling to avoid over- or underestimating overall emissions due to pulse loading of 537 TW systems. Accurate estimations of emissions and GWP are essential for inventory tracking 538 and future mitigation of GHGs. 539 4. Without being able to effectively track biogenic and abiotic sources of emissions, it is difficult 540 to effectively develop methods for GHG mitigation. Future research should more thoroughly 541 account for dissolved concentrations and system mass balances of carbon and nitrogen to 542 determine whether flux rates can be attributed to microbial production within the system or to 543 mass transfer. 544 5. The emission factors for N2O and CH4 were similar to similarly operated TW systems and on 545 the low end of most other treatment processes, suggesting that TWs have a lower environmental 546 impact than most mechanical WWTPs. Some of these differences may be due to the method in 547 which EF was calculated. 548 Funding 549 This work was supported by the Bridger Bowl Foundation; the National Science Foundation 550 Research Traineeship [award no. 2125748]; the Benjamin Fellowship through Montana State 551 University’s Norm Asbjornson College of Engineering (MSU NACOE); and the Montana Department 552 of Environmental Quality under the State Revolving Fund (SRF) [contract no. 221031]. 553 Acknowledgement 554 We thank the following entities and individuals for their financial and personal support of this 555 project: the Montana Department of Environmental Quality, especially Paul Lavigne; Bridger Bowl 556 Inc., especially Erik Pidgeon; and graduate students Kristen Brush, Justin Gay and Bryce Currey. We 557 give special thanks to Drs Anthony Hartshorn and Elan (Jack) Brookshire who both so kindly provided 558 the equipment that allowed us to analyze GHG emissions from our treatment wetland. 559 CRediT authorship contribution statement: 560 Stephanie H. Ayotte – Conceptualization, Data curation, formal & statistical analysis, 561 Methodology, Investigation, Validation, Writing-original draft, Writing-review & editing; Christopher 562 R. Allen - Conceptualization, Data curation, Methodology, Investigation, Funding acquisition, Writing-563 22 review & editing; Albert Parker – Formal & statistical analysis, Writing-review & editing; Otto R. 564 Stein – Conceptualization, Funding acquisition, Methodology, Resources, Supervision, Writing-review 565 review & editing; Ellen G. Lauchnor – Conceptualization, Funding acquisition, Methodology, 566 Resources, Supervision, Writing-review & editing. 567 6 REFERENCES Allen, C.R., 2016. Nitrogen Removal and Associated Greenhouse Gas Production in Laboratory- Scale Treatment Wetlands. Altor, A.E., Mitsch, W.J., 2008. Methane and carbon dioxide dynamics in wetland mesocosms: effects of hydrology and soils. Ecol. Appl. 18, 1307–1320. https://doi.org/10.1890/07-0009.1 Armstrong, W., 2000. Oxygen Distribution in Wetland Plant Roots and Permeability Barriers to Gas- exchange with the Rhizosphere: a Microelectrode and Modelling Study with Phragmites australis. Ann. Bot. 86, 687–703. https://doi.org/10.1006/anbo.2000.1236 Bahram, M., Espenberg, M., Pärn, J., Lehtovirta-Morley, L., Anslan, S., Kasak, K., Kõljalg, U., Liira, J., Maddison, M., Moora, M., Niinemets, Ü., Öpik, M., Pärtel, M., Soosaar, K., Zobel, M., Hildebrand, F., Tedersoo, L., Mander, Ü., 2022. Structure and function of the soil microbiome underlying N2O emissions from global wetlands. Nat. Commun. 13, 1–10. https://doi.org/10.1038/s41467-022-29161-3 Bridger Bowl Inc., 2023. Weather History, Alpine Weather Station, Bridger Bowl, MT. Bruun, J., Hoffmann, C.C., Kjaergaard, C., 2017. Convective transport of dissolved gases determines the fate of the greenhouse gases produced in reactive drainage filters. Ecol. Eng. 98, 1–10. https://doi.org/10.1016/j.ecoleng.2016.10.027 Campos, J.L., Valenzuela-Heredia, D., Pedrouso, A., Val del Río, A., Belmonte, M., Mosquera- Corral, A., 2016. Greenhouse Gases Emissions from Wastewater Treatment Plants: Minimization, Treatment, and Prevention. J. Chem. 2016, 1–12. https://doi.org/10.1155/2016/3796352 Chen, H., Zeng, L., Wang, D., Zhou, Y., Yang, X., 2020. Recent advances in nitrous oxide production and mitigation in wastewater treatment. Water Res. 184, 116168. https://doi.org/10.1016/j.watres.2020.116168 Chen, X., Zhu, H., Yan, B., Shutes, B., Tian, L., Wen, H., 2020. Optimal influent COD/N ratio for obtaining low GHG emissions and high pollutant removal efficiency in constructed wetlands. J. Clean. Prod. 267, 122003. https://doi.org/10.1016/j.jclepro.2020.122003 Doran, P.M., 1995. Mass Transfer, in: Bioprocess Engineering Principles. Elselvier Science & Technology, pp. 190–217. Dotro, G., Langergraber, G., Molle, P., Nivala, J., Puigagut, J., Stein, O., von Sperling, M., 2017. Treatment Wetlands. IWA Publishing. https://doi.org/10.2166/9781780408774 Duan, H., Wang, Q., Erler, D.V., Ye, L., Yuan, Z., 2018. Effects of free nitrous acid treatment conditions on the nitrite pathway performance in mainstream wastewater treatment. Sci. Total Environ. 644, 360–370. https://doi.org/10.1016/j.scitotenv.2018.06.346 Dubber, D., Gray, N.F., 2010. Replacement of chemical oxygen demand (COD) with total organic carbon (TOC) for monitoring wastewater treatment performance to minimize disposal of toxic analytical waste. J. Environ. Sci. Health Part A 45, 1595–1600. https://doi.org/10.1080/10934529.2010.506116 Eilers, P.H.C., Marx, B.D., 1996. Flexible smoothing with B-splines and penalties. Stat. Sci. 11. https://doi.org/10.1214/ss/1038425655 23 EPA, 2022. Inventory of U.S. Greenhouse Gas Emissions and Sinks: 1990-2020. (No. EPA 430-R- 22-003). U.S. Environmental Protection Agency. Faulwetter, J.L., Burr, M.D., Parker, A.E., Stein, O.R., Camper, A.K., 2013. Influence of Season and Plant Species on the Abundance and Diversity of Sulfate Reducing Bacteria and Ammonia Oxidizing Bacteria in Constructed Wetland Microcosms. Microb. Ecol. 65, 111–127. https://doi.org/10.1007/s00248-012-0114-y Faulwetter, J.L., Gagnon, V., Sundberg, C., Chazarenc, F., Burr, M.D., Brisson, J., Camper, A.K., Stein, O.R., 2009. Microbial processes influencing performance of treatment wetlands: A review. Ecol. Eng. 35, 987–1004. https://doi.org/10.1016/j.ecoleng.2008.12.030 Fernández-Baca, C.P., Truhlar, A.M., Omar, A.E.H., Rahm, B.G., Walter, M.T., Richardson, R.E., 2018. Methane and nitrous oxide cycling microbial communities in soils above septic leach fields: Abundances with depth and correlations with net surface emissions. Sci. Total Environ. 640–641, 429–441. https://doi.org/10.1016/j.scitotenv.2018.05.303 Foley, J., de Haas, D., Yuan, Z., Lant, P., 2010. Nitrous oxide generation in full-scale biological nutrient removal wastewater treatment plants. Water Res. 44, 831–844. https://doi.org/10.1016/j.watres.2009.10.033 Harville, D.A., 2023. Maximum Likelihood Approaches to Variance Component Estimation and to Related Problems. Hastie, Trevor, Tibshirani, Robert, 1986. Generalized Additive Models. Stat. Sci. 1, 297–318. Henze, M. (Ed.), 2008. Biological wastewater treatment: principles, modeling and design. IWA Pub, London. Hernández, M.E., Galindo-Zetina, M., Juan Carlos, H.-H., 2018. Greenhouse gas emissions and pollutant removal in treatment wetlands with ornamental plants under subtropical conditions. Ecol. Eng. 114, 88–95. https://doi.org/10.1016/j.ecoleng.2017.06.001 Hernandez, M.E., Mitsch, W.J., 2007. Denitrification in created riverine wetlands: Influence of hydrology and season. Ecol. Eng. 30, 78–88. https://doi.org/10.1016/j.ecoleng.2007.01.015 Hernandez, M.E., Mitsch, W.J., 2006. Influence of hydrologic pulses, flooding frequency, and vegetation on nitrous oxide emissions from created riparian marshes. Wetlands 26, 862–877. https://doi.org/10.1672/0277-5212(2006)26[862:IOHPFF]2.0.CO;2 IPCC, 2021. Climate Change 2021 – The Physical Science Basis: Working Group I Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, 1st ed. Cambridge University Press, Cambridge, United Kingdom and New York, NY USA. https://doi.org/10.1017/9781009157896 IPCC, 2019. Climate Change and Land: an IPCC special report on climate change, desertification, land degradation, sustainable land management, food security, and greenhouse gas fluxes in terrestrial ecosystems [P.R. Shukla, J. Skea, E. Calvo Buendia, V. Masson-Delmotte, H.-O. Pörtner, D. C. Roberts, P. Zhai, R. Slade, S. Connors, R. van Diemen, M. Ferrat, E. Haughey, S. Luz, S. Neogi, M. Pathak, J. Petzold, J. Portugal Pereira, P. Vyas, E. Huntley, K. Kissick, M. Belkacemi, J. Malley, (eds.)].Refinement to the 2006 IPCC Guidelines for National Greenhouse Gas Inventories. Guidelines for National Greenhouse Gas Inventories. International Panel on Climate Change (IPCC). Ji, B., Chen, J., Li, W., Mei, J., Yang, Y., Chang, J., 2021. Greenhouse gas emissions from constructed wetlands are mitigated by biochar substrates and distinctly affected by tidal flow and intermittent aeration modes. Environ. Pollut. 271, 116328. https://doi.org/10.1016/j.envpol.2020.116328 Jiang, X., Tian, Y., Ji, X., Lu, C., Zhang, J., 2020. Influences of plant species and radial oxygen loss on nitrous oxide fluxes in constructed wetlands. Ecol. Eng. 142, 105644. https://doi.org/10.1016/j.ecoleng.2019.105644 24 Johansson, A.E., Gustavsson, A.-M., Öquist, M.G., Svensson, B.H., 2004. Methane emissions from a constructed wetland treating wastewater—seasonal and spatial distribution and dependence on edaphic factors. Water Res. 38, 3960–3970. https://doi.org/10.1016/j.watres.2004.07.008 Kasak, K., Kill, K., Uuemaa, E., Maddison, M., Aunap, R., Riibak, K., Okiti, I., Teemusk, A., Mander, Ü., 2022b. Low water level drives high nitrous oxide emissions from treatment wetland. J. Environ. Manage. 312, 114914. https://doi.org/10.1016/j.jenvman.2022.114914 Kaushal, S.S., Mayer, P.M., Vidon, P.G., Smith, R.M., Pennino, M.J., Newcomer, T.A., Duan, S., Welty, C., Belt, K.T., Sujay, S., Mayer, P.M., Vidon, P.G., Smith, R.M., Pennino, M.J., New- , T.A., Duan, S., Welty, C., Belt, K.T., Use, L., Amplify, C.V., 2014. LAND USE AND CLIMATE VARIABILITY AMPLIFY CARBON , NUTRIENT , AND CONTAMINANT PULSES : A REVIEW WITH MANAGEMENT IMPLICATIONS 1 50, 585–614. https://doi.org/10.1111/jawr.12204 Law, Y., Ye, L., Pan, Y., Yuan, Z., 2012. Nitrous oxide emissions from wastewater treatment processes. Philos. Trans. R. Soc. B Biol. Sci. 367, 1265–1277. https://doi.org/10.1098/rstb.2011.0317 Livingston, G.P., Hutchinson, G.L., Spartalian, K., 2006. Trace Gas Emission in Chambers: A Non‐ Steady‐State Diffusion Model. Soil Sci. Soc. Am. J. 70, 1459–1469. https://doi.org/10.2136/sssaj2005.0322 Ma, W.K., Farrell, R.E., Siciliano, S.D., 2011. Nitrous oxide emissions from ephemeral wetland soils are correlated with microbial community composition. Front. Microbiol. 2, 1–11. https://doi.org/10.3389/fmicb.2011.00110 Mander, Ü., Dotro, G., Ebie, Y., Towprayoon, S., Chiemchaisri, C., Furlan, S., Jamsranjav, B., Kasak, K., Truu, J., Tournebize, J., Mitsch, W.J., 2014. Greenhouse gas emission in constructed wetlands for wastewater treatment : A review. Ecol. Eng. 66, 19–35. https://doi.org/10.1016/j.ecoleng.2013.12.006 Mander, Ü., Lõhmus, K., Teiter, S., Mauring, T., Nurk, K., Augustin, J., 2008. Gaseous fluxes in the nitrogen and carbon budgets of subsurface flow constructed wetlands. Sci. Total Environ. 404, 343–353. https://doi.org/10.1016/j.scitotenv.2008.03.014 Mander, Ü., Maddison, M., Soosaar, K., 2011. The Impact of Pulsing Hydrology and Fluctuating Water Table on Greenhouse Gas Emissions from Constructed Wetlands 4, 1023–1032. https://doi.org/10.1007/s13157-011-0218-z Mander, Ü., Tournebize, J., Espenberg, M., Chaumont, C., Torga, R., Garnier, J., Muhel, M., Maddison, M., Lebrun, J.D., Uher, E., Remm, K., Pärn, J., Soosaar, K., 2021. High denitrification potential but low nitrous oxide emission in a constructed wetland treating nitrate-polluted agricultural run-off. Sci. Total Environ. 779, 146614. https://doi.org/10.1016/j.scitotenv.2021.146614 Maucieri, C., Borin, M., Milani, M., Cirelli, G.L., Barbera, A.C., 2019. Plant species effect on CO2 and CH4 emissions from pilot constructed wetlands in Mediterranean area. Ecol. Eng. 134, 112–117. https://doi.org/10.1016/j.ecoleng.2019.04.019 Metcalf & Eddy, Inc., Tchobanoglous, G., Stensel, H.D., Tsuchihashi, R., Burton, F.L., 2014. Wastewater Engineering: Treatment and Resource Recovery, 5th ed. McGraw Hill. Meyer, R.L., Zeng, R.J., Giugliano, V., Blackall, L.L., 2005. Challenges for simultaneous nitrification, denitrification, and phosphorus removal in microbial aggregates: Mass transfer limitation and nitrous oxide production. FEMS Microbiol. Ecol. 52, 329–338. https://doi.org/10.1016/j.femsec.2004.11.011 Nakazawa, T., Matsuno, T., 2020. Current understanding of the global cycling of carbon dioxide, methane, and nitrous oxide. Proc. Jpn. Acad. Ser. B Phys. Biol. Sci. 96, 394–419. https://doi.org/10.2183/pjab.96.030 25 Oertel, C., Matschullat, J., Zurba, K., Zimmermann, F., Erasmi, S., 2016. Greenhouse gas emissions from soils—A review. Geochemistry 76, 327–352. https://doi.org/10.1016/j.chemer.2016.04.002 R Core Team, 2022. R: A language and environment for statistical computing. Reay, D.S., Davidson, E.A., Smith, K.A., Smith, P., Melillo, J.M., Dentener, F., Crutzen, P.J., 2012. Global agriculture and nitrous oxide emissions. Nat. Clim. Change 2, 410–416. https://doi.org/10.1038/nclimate1458 Rhine, E.D., Mulvaney, R.L., Pratt, E.J., Sims, G.K., 1998. Improving the Berthelot Reaction for Determining Ammonium in Soil Extracts and Water. Soil Sci. Soc. Am. J. 62, 473. https://doi.org/10.2136/sssaj1998.03615995006200020026x Rodriguez-Caballero, A., Aymerich, I., Marques, R., Poch, M., Pijuan, M., 2015. Minimizing N2O emissions and carbon footprint on a full-scale activated sludge sequencing batch reactor. Water Res. 71, 1–10. https://doi.org/10.1016/j.watres.2014.12.032 Shammas, N.K., 1986. Interactions of Temperature, pH, and Biomass on the Nitrification Process. Wiley 58, 52–59. Sharifian, R., Wagterveld, R.M., Digdaya, I.A., Xiang, C., Vermaas, D.A., 2021. Electrochemical carbon dioxide capture to close the carbon cycle. Energy Environ. Sci. 14, 781–814. https://doi.org/10.1039/D0EE03382K Soosaar, K., Mander, Ü., Maddison, M., Kanal, A., Kull, A., Lõhmus, K., Truu, J., Augustin, J., 2011. Dynamics of gaseous nitrogen and carbon fluxes in riparian alder forests. Ecol. Eng. 37, 40– 53. https://doi.org/10.1016/j.ecoleng.2010.07.025 Teiter, S., Mander, Ü., 2005. Emission of N2O, N2, CH4, and CO2 from constructed wetlands for wastewater treatment and from riparian buffer zones. Ecol. Eng. 25, 528–541. https://doi.org/10.1016/j.ecoleng.2005.07.011 Tumendelger, A., Alshboul, Z., Lorke, A., 2019. Methane and nitrous oxide emission from different treatment units of municipal wastewater treatment plants in Southwest Germany. PLOS ONE 14, e0209763. https://doi.org/10.1371/journal.pone.0209763 Velthuis, M., Veraart, A.J., 2022. Temperature Sensitivity of Freshwater Denitrification and N 2 O Emission—A Meta‐Analysis. Glob. Biogeochem. Cycles 36, 1–14. https://doi.org/10.1029/2022gb007339 Verboven, P., Pedersen, O., Ho, Q.T., Nicolai, B.M., Colmer, T.D., 2014. The mechanism of improved aeration due to gas films on leaves of submerged rice: Leaf gas films on submerged rice. Plant Cell Environ. n/a-n/a. https://doi.org/10.1111/pce.12300 Vilain, G., Garnier, J., Decuq, C., Lugnot, M., 2014. Nitrous oxide production from soil experiments: denitrification prevails over nitrification. Nutr. Cycl. Agroecosystems 98, 169–186. https://doi.org/10.1007/s10705-014-9604-2 Wang, H., Sun, Y., Wu, G., Guan, Y., 2019. Effect of anoxic to aerobic duration ratios on nitrogen removal and nitrous oxide emission in the multiple anoxic/aerobic process. Environ. Technol. U. K. 40, 1676–1685. https://doi.org/10.1080/09593330.2018.1427801 Western Agricultural Research Center, 2019. Summary of Montana Growing Conditions [WWW Document]. Summ. Mont. Grow. Condtions - West. Agric. Res. Cent. Mont. State Univ. URL https://agresearch.montana.edu/warc/guides/summary_of_Montana_growing_conditions.ht ml#:~:text=Montana%20has%20a%20short%20growing,frost%2Dfree%20days%20on%20 average. (accessed 1.31.24). Wood, S.N., 2017. Generalized Additive Models: An introduction with R, 2nd ed. CRC Press. Wood, S.N., 2011. Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. J. R. Stat. Soc. B 73, 3–36. 26 Wunderlin, P., Mohn, J., Joss, A., Emmenegger, L., Siegrist, H., 2012. Mechanisms of N2O production in biological wastewater treatment under nitrifying and denitrifying conditions. Water Res. 46, 1027–1037. https://doi.org/10.1016/j.watres.2011.11.080 Zhou, N., Dang, C., Zhao, Z., He, S., Zheng, M., Liu, W., Wang, X., 2019. Role of sludge retention time in mitigation of nitrous oxide emission from a pilot-scale oxidation ditch. Bioresour. Technol. 292, 121961. https://doi.org/10.1016/j.biortech.2019.121961 Zhou, S., Wang, C., Liu, C., Sun, H., Zhang, J., Zhang, X., Xin, L., 2020. Nutrient removal, methane and nitrous oxide emissions in a hybrid constructed wetland treating anaerobic digestate. Sci. Total Environ. 733, 138338. https://doi.org/10.1016/j.scitotenv.2020.138338 Zhu-Barker, X., Cavazos, A.R., Ostrom, N.E., Horwath, W.R., Glass, J.B., 2015. The importance of abiotic reactions for nitrous oxide production. Biogeochemistry 126, 251–267. https://doi.org/10.1007/s10533-015-0166-4 27 Fig 1. Pilot TW System Schematic. Schematic of the TW system showing physical characteristics and randomly selected gas sampling locations. Cells in each stage are physically separated. Primary-clarified influent and recycle water are independently dosed onto the first stage at specified time intervals. For every volume of influent there were two volumes of recycled water over a 24h period. Fig 2. Flux distributions for N2O, CH4 and CO2 by stage and sample location. The bar in the center indicates the median flux. The upper and lower reaches of the boxes indicate the upper (75th) and lower (25th) quartiles of the fluxes (interquartile range – IQR), and the whiskers represent the minimum and maximum up to ±1.5*IQR. Dots indicate calculated flux values, and points that extend beyond the whiskers are potential outliers. There was no statistically significant difference in median emissions between locations within a stage (p ≥ 0.05). Fig 3. Flux and Temperature data in the system’s first stage. N2O, CO2 and CH4 gas flux (mg-N or C m-2·hr-1) in the denitrification stage of the pilot TW plotted with surface temperature. Measurements were recorded over a 48-hour period, at a frequency of 6 measurements per hour. Dashed vertical lines indicate the occurrence of a primary-clarified influent wastewater dose onto the wetland (~750 mg L-1 COD). Lines connecting points represent piecewise linear trends visualized using the R ggplot() function. Fig 4. GAM fit of gas fluxes. GAM fit of N2O (a), CH4 (b), and CO2 (c) soil efflux over six continuous 8- hour cycles defined by an influent dose. Numbers 1-6 for CO2 indicate cycle number. Dashed vertical lines indicate occurrence of an influent wastewater dose. Six recycle doses (not plotted) generally correspond with smaller intermediate peaks. Fig 5. Emissions due to physical mechanisms. Physical mechanisms that increase emissions via mass transfer, including 1) the volatilization of dissolved gases at the dosing orifices, 2) the displacement of gas in the unsaturated porous media by large volume doses and 3) the release of large concentrations of dissolved gases when the gas-liquid surface area increases in unsaturated or aerated systems. 28 29 Table 1 Mean pollutant concentrations (w/ standard deviation) in the influent and effluent of the system stages and median flux (w/ 95% confidence intervals) for March - April 2022. Stage COD NH4 +-N NO3 --N TIN CO2 flux CH4 flux N2O flux (mg L-1) (mg L-1) (mg L-1) (mg L-1) (mg C m-2 h-1) (mg C m-2 h-1) (mg N m-2 h-1) Influent 808 (59.3) 161 (26.6) 0.4 (0.3) 161 (26.9) Effluent First 52 (10.8) 45 (11.3) 5.0 (4.0) 45 (11.3) 221 (211-230) 0.85 (0.77-0.93) 0.66 (0.63-0.69) Effluent Second 28 (3.8) 0.707b 39.0 (4.0) 39 (4.0) 606 (359-853) 0.2 (0.02-0.39) 0.34 (0.17-0.51) Total Removal (%)a 97% 74% a calculated based on daily mass loading b measured values below detection limit, estimated using [MDL/ (√2)] (Croghan & Egegy, 2003) 30 31 32 33 Table 2 Mean influent water quality (sd) and greenhouse gas emission factors from systems treating domestic wastewater. System Type Location Wastewater Description Q COD TOC TN EFCH4-C EFN2O-N Reference (m3 d-1) (g m-3) (g m-3) (g-N m-3) (%) (%) Constructed Wetlands Two stage VSSF Montana, US pre-settled municipal 3.6 808.3 (59.3) 268 (17.4) 161 (26.9) 0.04% 0.27% This Studya,b Hybrid VSSF/HSSF China swine anaerobic digestate 3.6 2725 (1264) 909 (420.3) 379 (58) n.a 0.34% [1]b,c Hybrid VSSF, HSSF and two FWS Estonia raw municipal 11.3 n.a. 16.1 50.9 0.88% 0.021% [2] d HSSF, planted sand filter Estonia hospital n.a. n.a. 69.0 109.0 9.9% 0.45% [2]d FWS Sweden secondary municipal 264 n.a. n.a. 237.0 0.25% n.a. [3] Mechanical Wastewater Treatment CAS Germany municipal 5000 746.6 n.a. 52.8 0.01% 0.001% [4]c MLE Germany municipal 14600 202.2 n.a. 83.0 0.004% 0.008% [4]c MBR Washington, US municipal 62000 283 (40) n.a. 52 (7.9) n.a. 0.60% [5] CAS QLD, Australia municipal 5000 499 (104) n.a. 64.0 (6.5) n.a. 1.90% [6]c MLE Western Australia municipal 63000 730 n.a. 50 n.a. 2.70% [7] SBR Western Australia municipal 137000 550 n.a. 47-58 n.a. 3.27% [7] "A2/O" New South Wales municipal 25000 850 n.a. 55-85 n.a. 1.40% [7] MLE South Australia municipal 49000 700 n.a. 69-103 n.a. 3.57% [7] a Total inorganic N b TOC estimated from COD c EF reported per kilogram of influent N and C d EF reported per kilogram of TOC removed 1 (Zhou et al., 2020) 2 (Teiter & Mander, 2005) 3 (Johansson et al., 2004) 4 (Tumendelger et al., 2019) 5 (Cavanaugh, 2021) 6 (Pan et al., 2016) 7 (Foley et al., 2010) 34 1 Introduction 2 Material and Methods 2.1 Site description and system operation. 2.2 Gas flux measurements. 2.2.1 Manual sampling. 2.2.2 Automated sampling. 2.3 Ancillary measurements. 2.4 Gas Flux Calculation and Emission Factor (EF). 2.5 Data Analysis 3 Results 3.1 First stage vs. second stage gas flux – March Sample Campaign 3.2 Temporal Fluctuations with hydraulic loading -April Sample Campaign 3.2.1 Nitrous oxide emissions 3.2.2 Methane emissions 3.2.3 Carbon dioxide emissions 3.3 Global Warming Potential and Emission Factors 4 Discussion 4.1 Water Quality 4.2 Emission Factors. 4.3 Production vs Emissions 4.3.1 Nitrous oxide 4.3.2 Methane 4.3.3 Carbon Dioxide 4.4 Temporal influence of hydraulic loading on first stage 5 Conclusions 6 REFERENCES Copy Cover Page.pdf Blank Page Blank Page