Copper availability governs nitrous oxide accumulation in wetland soils and stream sediments

Copper availability governs nitrous oxide accumulation in wetland soils and stream sediments Neha Sharma, Elaine D. Flynn, Jeffrey G. Catalano and Daniel E. Giammar Department of Energy, Environmental and Chemical Engineering, Washington University in St. Louis, St. Louis, Missouri 63130, United States Department of Earth and Planetary Sciences, Washington University in St. Louis, St. Louis, Missouri 63130, United States *Corresponding Author: Address: Department of Energy, Environmental and Chemical Engineering, Washington University in St. Louis, St. Louis, MO 63130, USA Phone: (314) 935-6849 Email: giammar@wustl.edu


22
Nitrous oxide (N2O) is a potent greenhouse gas whose global warming potential per unit mass is 265-298 23 times that of carbon dioxide (CO2) for a 100-year timescale (IPCC, 2014; Sovacool et al., 2021). Global 24 N2O emissions in the decade between 2007-2016 averaged 17 Tg N yr -1 , of which 57% (9.7 Tg N yr -1 ) were 25 contributed by natural soils and oceans (Tian et al., 2020). Denitrification, an anoxic process in which 26 nitrate (NO3 -) is reduced to N2, is a key biogeochemical process that regulates the amount of N2O released 27 from both terrestrial and aquatic ecosystems into the atmosphere (Makowski, 2019;Tian et al., 2020;28 Martinez-Espinosa et al., 2021). Natural aquatic systems, especially those that display vertical redox 29 gradients, such as wetlands and hyporheic zones in streams, are active sites for denitrification (Merill and  regions below redox transition zones promote denitrification with NO3being reduced to nitrite (NO2 -), 33 nitric oxide (NO), N2O, and N2. The incomplete conversion of NO3and NO2to N2 causes N2O to be 34 released from the aquatic systems to the atmosphere. 35 An array of metalloenzymes that contain Fe, Cu, and Mo are involved in reducing nitrate and 36 intermediate species to N2 during denitrification (Godden et al., 1991;Bertero et al., 2003;Nojiri et al., 37 2007). The transformation of NO3to NO2is catalyzed by respiratory nitrate reductase (Nar), which requires 38 Fe and Mo for complete conversion (Bertero et al., 2003;Jormakka et al., 2004). Depending on the type of 39 microorganism, reduction of NO2to NO is catalyzed by either an iron-containing nitrite reductase (NirS) 40 or a Cu-containing nitrite reductase (NirK) (Godden et al., 1991;Nojiri et al., 2007). NO is rapidly 41 transformed to N2O with an Fe-bearing nitric oxide reductase (cNOR), and in the final step, N2O is reduced 42 to N2 by a Cu-rich nitrous oxide reductase (Nos) (Brown et al., 2000). 43 A scarcity of available Cu can limit the conversion of N2O to N2. Laboratory studies of pure cultures 44 have demonstrated that Cu limitation resulted in N2O accumulation (Iwasaki et al., 1980;Granger and Ward, 45 2003; Black et al., 2016). Granger and Ward (2003) conducted a study with Pseudomonas stutzeri and 46 47 They observed that Cu concentrations of approximately 3 nM caused N2O to accumulate, whereas 10 nM 48 Cu resulted in increased growth rates and complete conversion of N2O to N2. The growth of two denitrifying 49 microorganisms, Alcaligenes sp. NGIB 11015 and Alcaligenes faecalis lAM 1015, was also stimulated by 50 the addition of Cu in the range of 0.5 to 40 µM (Iwasaki et al., 1980). Another study on P. stutzeri revealed 51 that when dissolved Cu concentrations were varied between 0-80 µM, the maximum conversion of N2O to 52 N2 was achieved at a concentration of 0.80 µM (Black et al., 2016). 53 In soils and sediments, high concentrations of Cu can inhibit denitrification, whereas low 54 availability of Cu can limit microbial activity causing accumulation of intermediate nitrogen species. In 55 estuarine sediments, the addition of 79 µg g -1 Cu inhibited microbial activity by 85% and specifically 56 increased the accumulation of NO2and N2O (Magalhaes et al., 2007). Similarly, the addition of Cu at high 57 loadings of 250-1000 µg g -1 increased N2O emissions from soils and wetland sediments (Sakadevan et al., 58 1999;Shaaban et al., 2019). In these three studies, the associated dissolved Cu concentrations were not 59 measured. While the three studies just noted had increased accumulation of N2O associated with high Cu, 60 two other studies found that the addition of Cu to systems with initially low Cu decreased N2O accumulation. 61 A recent study of freshwater wetland sediments that initially had 37.8 µg g -1 Cu and were amended with 62 CuSO4 to have 26 µM dissolved Cu showed an increased abundance of nitrite and nitrous oxide reductase 63 genes that enhanced the conversion of N2O to N2 (Giannopoulos et al., 2020). Similarly, a study of 64 freshwater sediments collected from central Indiana also showed that N2O accumulation decreased when 65 the sediments were amended with 50-100 µg g -1 Cu (Jacinthe and Tedesco, 2009). On the other hand, the 66 addition of 100 µg L -1 Cu did not have any effect on N2O emissions release in freshwater wetland sediments 67 (Doroski et al., 2019). Copper concentrations in uncontaminated environments are typically low (Black et 68 al., 2011;Black et al., 2016), and hence limited Cu bioavailability in such settings may significantly affect 69 N2O conversion via denitrification. 70 In natural aquatic systems, the relationship between the total Cu amount in the sediments and the 71 bioavailable concentration of dissolved Cu is controlled by the presence of solid phases, such as sulfide 72 mm) was maintained at 50 °C for 7.5 min, after which it was ramped to 130 °C using a rate of 20 °C/min 200 and then kept at this temperature for 2 min. Helium was used as the carrier gas at a flow rate of 30 mL/min. 201 A pulse discharge detector (PDD) at 150 °C was used for the analysis of N2O. The concentration of the 202 standards varied from 10 ppmv to 0.1% N2O and were prepared using a certified gas standard from Airgas. 203 The concentration of N2O dissolved in the fluid was determined using the ideal gas law and Henry's gas 204 solubility law. The value of Henry's constant at 25 °C used for determining dissolved N2O was 2.4*10 -4 205 mol m -3 Pa -1 (Sander, 2015). The total N2O present in the microcosm at the time of sampling was the sum 206 of the gas in the headspace and the gas dissolved in the water. 207 To measure the dissolved phase constituents, 1 mL of slurry from each serum bottle was sampled 208 and centrifuged (Spectrafuge 16M) for 5 min at 5000 rpm. To determine the dissolved metal (Cu, Fe, and 209 Mn) concentrations, 300 µL of the supernatant was acidified to 1% HNO3 and analyzed by ICP-MS. The 210 remaining supernatant was divided to estimate the pH using Whatman pH indicator strips and to determine 211 the nutrient concentrations (NO3 -, NO2and NH4 + ) spectrophotometrically using a discrete multi-chemistry 212 analyzer (Seal Analytical AQ300). The samples used for nutrient analysis were frozen and then thawed 213 overnight at 4°C before analysis. Nitrite was measured by the reaction of the sample with sulfanilamide in 214 dilute phosphoric acid to form a diazonium compound which binds to N-(1-naphthyl)-ethylenediamine 215 dihydrochloride to form an azo dye detected at 520 nm (Huffman and Barbarick, 1981). To determine the 216 nitrate concentration, NO3was first reduced to NO2by cadmium and then the NO2was measured. 217 Ammonium present in the samples was determined by reacting the samples with hypochlorite liberated 218 from dichloroisocyanurate in an alkaline solution followed by a reaction with salicylate in the presence of 219 nitroferricyanide to form a blue indophenol dye, which was measured at 660 nm (Krom, 1980). At the end 220 of incubation experiments, the final pH value was recorded using a pH electrode. The water was filtered 221 and stored at 4°C to determine the dissolved organic carbon (DOC) concentration using a total organic 222 carbon analyzer (Shimadzu TOC-L) at the end of incubation experiments. 223

224
To estimate the speciation of Cu in the presence of organic carbon in the incubation experiments, the 225 nonideal competitive adsorption-Donnan (NICA-Donnan) model in Visual MINTEQ 3.1 (Yan and Korshin,226 2014) was used. The model is a combination of NICA which enables simulation of metal complexation to 227 humic substances, and a Donnan model describing nonspecific electrostatic interactions between ions and 228 humic substances Ren et al., 2015). Although humic 229 substances might not truly represent the dissolved organic matter present in aquatic systems (Kleber and 230 Lehmann, 2019; Myneni, 2019), we used the NICA-Donnan model to estimate the aqueous speciation of 231 Cu because this model has previously provided accurate predictions of metal speciation and availability in 232 natural systems (Han et al., 2014;Ponthieu et al., 2016). NICA considers competitive binding between 233 protons and metals to humic substances by accounting for binding site heterogeneity and ion-specific 234 nonideality . The generic parameters obtained in previous studies for Cu and proton 235 binding to humic material were used (Milne et al., 2001;Milne et al., 2003;. Water 236 chemistry parameters (pH, total dissolved elements (Table S1)), dissolved Cu, Fe, and Mn, and dissolved 237 organic carbon (DOC) were used as the input parameters for determining dissolved Cu speciation. Three 238 sets of conditions were used to account for Cu speciation at different total dissolved Cu concentrations 239 corresponding to the control, low Cu loading, and high Cu loading experiments (details of the methodology 240 in Section S1 in SM, Table S3). Under anoxic environments, as in our incubation experiments, Cu(II) can 241 be reduced to Cu(I) by microorganisms, inorganic reductants, and redox-active functional groups on   The surface water in the marsh wetlands (Marsh 1) and stream sediments (Stream 1 and Stream 2) had pH 279 values ranging from 7.5-8.1, and they contained high concentrations of calcium, magnesium, and sulfate. 280 However, the riparian wetland surface water samples (Riparian 1 and Riparian 2) were at pH 5.5-6.0 with 281 substantially lower concentrations of major elements. 282 The mineralogy at all of the studied sites is dominated by quartz, with variations in minor phases (Yan 283 et al., 2021). The total organic carbon content of aquatic systems varies with location: the marsh wetland 284 site (Marsh 1) contained the highest carbon content (9.0%), whereas Stream 2 exhibited the lowest carbon 285 content (0.5%) ( Table S2). The sulfur content was low at all sites (below 0.24%), following the trend Marsh 286 1 > Riparian 2 ≈ Stream 1 > Riparian 1 ≈ Stream 2. The concentrations of Cu were well below the crustal 287 abundance (428 ± 61 nmol/g) (Rudnick and Gao, 2003) at all the studied sites. The marsh wetland soil 288 (Marsh 1) and Riparian 2 location in the riparian wetlands contained higher concentrations of solid-phase 289 Cu than the other locations. The total Fe concentration in the solid phases was similar at all the sites (200 290 to 460 µmol/g), except for Riparian 1, which contained an order of magnitude less Fe than the other 291 soils/sediments. The concentration of Mn was two to three orders less than the Fe content, with the highest 292 values observed in stream sediments. Extractable ammonium in soils/sediments was greater than extractable 293 NO3and NO2at all the locations, which may indicate denitrification and/or ammonium retention via cation 294 exchange (Table S2). 295 296 The controls and the low-Cu amended sets showed similar NO3reduction profiles in all the systems 297 studied ( Figure 2). However, a small delay in NO3reduction after Cu addition was observed at high Cu 298 loading in Riparian 1, Riparian 2, and Stream 2 incubation experiments (Figure 2a, 2e, and 2m). In Riparian 299 1 and Stream 2 experiments, complete reduction of NO3did not occur, even after 27 days and 20 days of 300 incubation, respectively. 301

Evolution of nitrogen species concentrations during incubation
The presence of detectable NO2was transient and showed a brief appearance followed by a rapid 302 decline in concentration in Riparian 2, Stream 1, and Marsh 1 incubations. In Riparian 1 soils, the dissolved 303 NO2concentrations were below the detection limit (0.0005 mmol-N L -1 ) throughout the experiment, 304 suggesting rapid conversion of NO2in these soils ( Figure 2b). In the case of Riparian 2 and Stream 2 305 systems, Cu addition affected NO2formation/reduction because more NO2was detected in controls as 306 compared to Cu-amended sets. 307 For all the systems studied, less N2O accumulated in the sets amended with Cu. We did not observe 308 persistent accumulation of N2O in the case of Riparian 1 soils (Figure 2c), however, the maximum 309 concentration of N2O (control: 0.040 mmol-N L -1 at 14 days, low-Cu: 0.032 mmol-N L -1 at 10 days, and 310 high-Cu: 0.021 mmol-N L -1 at 12 days) decreased as the dissolved Cu concentration increased. For Riparian 311 1 controls and low-Cu amended microcosms, N2O started to accumulate after 3 days of incubation, whereas 312 in high-Cu amended sets, N2O accumulation was only observed after 6 days of incubation. The complete 313 reduction of N2O to N2 was fast in low-Cu amended Riparian 1 sets; we did not observe N2O after 14 days 314 in low-Cu added sets, whereas it took 29 and 23 days to completely reduce N2O in high-Cu added and 315 control sets, respectively. In the Riparian 2 control, N2O accumulated in the headspace and persisted until 316 the end of the experiment at 30 days, whereas the N2O concentration first increased and then decreased 317 after 10 days and 16 days in Riparian 2 sets amended with low Cu and high Cu, respectively ( Figure 2g). 318 For Riparian 2, relative to controls, the maximum N2O concentration decreased by 38.6% in low Cu-added 319 sets and by 70.1% in high Cu-added sets. In the case of the stream sediments, Stream 1 showed a significant 320 effect of Cu addition on N2O reduction; with respect to controls, the peak N2O concentration decreased by 321 2.6 times and 7.8 times in sets with low and high Cu loading, respectively ( Figure 2k). In Stream 2 and 322 Marsh 1 systems, we were able to measure detectable N2O in the headspace of only the controls; in the Cu-323 amended sets, any N2O generated was rapidly reduced before the N2O reached detectable levels ( Figure 2n

347
The effect of Cu on denitrification was quantified with the help of the kinetic model. We obtained 348 Michaelis-Menten parameters (Table 2) and the abiotic rate constant for the set of differential equations 349 defined earlier (Eq 1-4). Here, K NO 3 −, K NO 2 − , and K N 2 O values reflect the ability of the microbial community 350 present in the soils and sediments to reduce NO3to NO2 -, NO2to NO, and N2O to N2, respectively under 351 the conditions studied. NO is rapidly transformed to N2O, hence the conversion of NO to N2O was assumed 352 to not be rate-limiting. The rate constant for abiotic reduction of NO2to N2 by inorganic donors in the 353 system is defined by k ab , and inclusion of this reaction helped us reproduce the major features of all the 354 experiments ( Figure 4). The abiotic reduction of NO2to N2O is also a possible pathway, but the 355 incorporation of this reaction into the kinetic model did not improve the fit to experimental data. Michaelis-356 Menten parameters have an inverse relationship with rates, unlike rate constants; the smaller the value of 357 K y , the faster the forward reaction; whereas the greater the value of k ab , the faster the rate of abiotic nitrite 358 reduction. The model was able to describe the major features for nitrogen species reduction at all the sites 359 except for Marsh 1. For the Marsh 1 site, we observed a lag in NO3reduction in all the incubation 360 experiments. Our model does not account for the acclimatization time of microorganisms after NO3an 361 addition which could have caused poor fitting of data from the Marsh 1 experiments. 362 The modeled K NO 3 − values show that NO3reduction is fastest in Stream 1 sediments followed by 363 Marsh 1, Riparian 2, Riparian 1, and Stream 2. The parameter K NO 3 − was similar for control and low Cu-364 loading in all the systems studied. However, the modeled value increased (Table 2)  substantially decreased the lability of Cu in the systems studied. Here, the labile Cu concentration is defined 392 as the sum of Cu 2+ , Cu(OH) + , and Cu(OH)2(aq). In riparian wetland controls, Riparian 1 and Riparian 2, in 393 the studied pH range (5-6), Cu is predominantly present as Cu-organic matter complexes (Figure 6), and 394 the labile Cu concentration (Table S4) (Table S4), and the higher 418 concentratons could have inhibited NO3reduction during the incubation experiments. 419 Incomplete reduction of NO3was observed in incubation experiments using Riparian 1 soils and 420 Stream 2 sediments even when Cu was not added and had low labile concentrations. This suggests that 421 denitrification was limited by the low total organic carbon content at these sites (Table S2) In Riparian 1 soils, both the background dissolved Cu concentration (29.3 nM) and the solid-phase-443 associated Cu (48.3 nmol/g) were less than the Cu concentration in Riparian 2 soils (dissolved: 41 nM; 444 solid-phase: 262.3 nmol/g), but N2O did not accumulate persistently in the headspace of Riparian 1 soils. 445 This observation suggests that bioavailable Cu for N2O to N2 conversion is more abundant in Riparian 1 446 soils. Riparian 1 soils have less dissolved organic carbon (23 mg C/L) than Riparian 2 soils, and thus are 447 less able to decrease the bioavailability of Cu by forming soluble complexes of organic matter with Cu. The 448 speciation results corroborated the hypothesis because in the pH range studied, the dissolved labile Cu in 449 Riparian 1 controls (2.8 ± 0.9 nM) was greater than Riparian 2 controls (1.4 ± 0.8 nM). Prior study on a 450 lake system indicated that the presence of 3 nM dissolved Cu decreased N2O accumulation during 451 denitrification relative to systems containing no Cu (Twining et al., 2007). Although the rate of N2O to N2 452 conversion was slow in Riparian 1 controls as compared to Cu-amended sets (Table 2)

486
During the initial 2-3 days of incubation, the pH increased from 5.0 to ~ 6.5 for Riparian 1 and Riparian 487 2 soils, from 7.6 to ~ 8.9 for Stream 1 and Stream 2 sediments, and from 7.0 to ~ 8.2 for Marsh 1 soils, and 488 then remained relatively constant. The increase in pH values can be attributed to NO3and NO2reduction 489 during denitrification. Previous study on riparian soils indicated that the pH increased from 5 to 7 and 5 to 490 9 in unbuffered NO3reduction experiments with low (111 µmol N g -1 ) and high (500 µmol N g -1 ) nitrogen 491 loadings, respectively (Clement et al., 2005). In contrast, pH variation was limited (± 0.5) when 492 denitrification occurred in carbonate-buffered lowland soils from Northern Italy (Castaldelli et al., 2019). 493 Our previous study on the studied natural aquatic systems indicated that these soils and sediments lacked 494 carbonate minerals (Yan et al., 2021); hence, the buffering capacity of the soils/sediments was likely 495 inadequate to prevent the increase in pH upon NO3and NO2reduction. to competitive adsorption (Kurdi and Donner, 1983;Traina and Doner, 1985). In Stream 1 and Marsh 1 534 studies, the concentration of dissolved Mn decreased with time (Figure 3i and 3o). Mn adsorption has been 535 found to increase with pH on clay minerals, iron oxides/hydroxides and aluminum oxides. In reducing NO3 -, 536 Mn 2+ can also serve as an electron donor (Eq 7) to autotrophic denitrifiers belonging to the genera 537 and Riparian 2) showed substantial N2O accumulation despite having much higher dissolved background 547 Cu concentrations of 30-50 nM. This disparity suggests that the speciation of Cu, and hence its 548 bioavailability, plays a more important role than the total Cu content in controlling N2O to N2 conversion 549 in the studied environmental systems. 550 Even after Cu addition, the accumulated concentrations of N2O in the riparian wetland samples 551 were higher than in the other locations. One possible explanation is that the riparian wetland incubation 552 experiments were conducted at pH 5, whereas the incubations for marsh wetland soils (Marsh 1) and stream 553 sediments (Stream 1 and Stream 2) were performed at neutral pH conditions. Acidic soils decrease the 554 activity of the nitrous oxide reductase enzyme, leading to N2O accumulation (Knowles, 1982

632
Data associated with this article can be accessed at https://data.mendeley.com//datasets/t359pdpcxy/1. 633        (Table 1). Here, Cu-S1 shows Cu bound to carboxylic acids of organic carbon and Cu-S2 is the Cu bound to phenolic groups on organic carbon. Shaded areas indicate the pH range over the course of the experiment.

Number Details Page (s)
Table S1 Concentration of major elements and species in the simulated site water used for uptake studies and microcosm experiments 3 Table S2 Characterization of soils and sediments collected from different aquatic systems 3 Section S1 Dissolved Cu speciation in microcosms 4   Site Cu (nmol/g) Section S1: Dissolved Cu speciation in microcosms 9 Two different binding sites are considered in the NICA model, type 1 and type 2, corresponding 10 to carboxylic (low affinity) and phenolic (high affinity) sites respectively . Here, I is ionic strength and b is an empirical parameter describing the variation of Donnan volume with 24 ionic strength . The values of parameters used in estimating Cu speciation are listed 25 in Table S3. 26 Humic substances are normally assumed to be the main binding substances (Ren et al., 2015), and 27 their concentrations were determined from the dissolved organic carbon (DOC) concentrations. Humic 28 substances account for 60% of DOC in natural water systems (Zhang and Davison, 2000;Gueguen et al., 29 2011), and they comprise 50% carbon, so concentration of HS was assumed to be 1.2 times the DOC 30 concentration. The pH, temperature, total dissolved elements (Na, Mg, K, Ca, Cl, NO3, SO4, and PO4) 31 (Table EA1), and dissolved Cu, Fe, and Mn were used as the input parameters for determining Cu speciation. 32