To what extend can soil moisture and soil contamination stresses affect greenhouse gas emissions? An attempt to calibrate a nitrification/denitrification model

Abstract. Continental biogeochemical models are commonly used to prospect the effect of land use, exogenous organic matter input or climate change on soil greenhouse gas emission. However, they can still not be used to investigate the effect of soil contamination while it is known to affect several soil processes and to concern a large fraction of land surface. We implemented a commonly used model estimating soil nitrogen (N) emission, the DeNitrification DeCompostion (DNDC) model, with a function taking into account soil copper (Cu) contamination in nitrate production modulation. Then, we aimed at using this model to predict N-N2O, N-NO2 NOx and N-NH4 emissions in the presence of contamination and in the context of changes in precipitations. For that, incubations of soils were performed at different soil moistures in order to mimic expected rainfall patterns during the next decades and in particular drought and excess of water. The effect of this double stress on soil nitrate production was studied using a bio-assay. Then, data of nitrate production obtained under each moisture treatment were used to parameterize the DNDC model and estimate soil N emission considering the various effect of Cu. Whatever the moisture preincubation, experimental results showed a N-NO3 decreasing production when Cu was added but with different sharpness depending on soil moisture. The DNDC-Cu version we proposed was able to reproduce these observed Cu effects on soil nitrate concentration with r2 > 0.99 and RMSE < 10 % for all treatments in the DNDC-Cu calibration range (> 40 % of the water holding capacity) but showed poor performances for the dry treatments. We modelled a Cu-effect inducing an increase in N-NH4 soil concentration and emissions due to a reduced nitrification activity, and therefore a decrease in N-NO3, N-N2O and N-NOx concentrations and emissions. The effect of added Cu was larger on N-N2 and N-N2O emissions than on the other N species and larger for the soils incubated under constant than variable moisture.



Introduction
The increase in atmospheric greenhouse gases [GHG] like CO2, CH4, or N2O is expected to induce a global climate change with e.g. higher mean temperature or changes in precipitation patterns with projections of increased precipitations or droughts depending on regions (Knutti and Sedláček 2012). These modifications in rainfall patterns may impact soil moisture which is one of the main drivers of soil microbial activity (Moyano et al. 2013).

45
Microbial communities ensure key activities supporting numerous ecosystem functions, such as those involved in nitrogen (N) cycle influencing N2O emissions (Jones et al. 2014) and are at the origin of more than 80% of N2O fluxes (IPCC 2019). In particular, nitrification/denitrification processes are largely controlled by the local (an-)oxic treatments and therefore by soil moisture, denitrification being the main source of soil N2O emission for moist soils whereas dry soil N2O emissions are mainly due to nitrification (Bateman and Baggs 2005). This strong 50 dependency to local soil O2 availability (Khalil et al. 2004), by playing on the realization of nitrification/denitrification reactions and N species diffusion (Conrad 1996;Schurgers et al. 2006), makes N soil fluxes dynamics particularly difficult to predict at larger scales. Despite this, some continental biogeochemical have shown improved predictions when N cycle is explicitly represented (Kesik et al. 2005;Butterbach-Bahl et al. 2009; Vuichard et al. 2018).

55
In addition to climate change, human activities introduce significant quantities of contaminants into the environment, such as trace elements (TE) which are persistent and can be toxic for soil biota (Bech et al. 1997; Giller et al. 2009). Indeed, the contamination of soils by TE has become a major concern at global scale (De Vleeschouwer et al. 2007; Khan et al. 2008) coming from atmospheric sources (Steinnes et al. 1997) or through the use of pesticides (Nicholson et al. 2003). In particular, TE contaminations are known to largely affect soil 60 microorganisms (Giller et al. 2009) and their activities, such as nitrification/denitrification processes (Broos et al. 2007; Mertens et al. 2010). Therefore, the combined effect of climate change and of soil contamination may largely impact the emissions of NOx and N2O from soils (Holtan-Hartwig et al., 2002;Vásquez-Murrieta et al. 2006).
However, the effect of the interactions between climate change and soil contamination on the GHG emissions is still poorly documented (Rillig et al. 2019;Zandalinas et al. 2021).

65
Despite recent progress, the Earth system models (ESMs) used to predict future climate change still don't take into account soil contamination effect on GHG emissions (Anav et al. 2013) whereas a large fraction of the soils are impacted by contaminants (Lado et al. 2008). Furthermore, soil biogeochemical models are often used to estimate loss or accumulation of N species (ammoniac NH4 volatilization, nitrate NO3 leaching - Giltrap et al. 2010) or they respective concentrations under scenarii of organic fertilizer amendments, but do not take into 70 account the contamination which often occurs simultaneously (Wuana and Okieimen 2011). Thus, there is a growing need to provide continentals models combining ecotoxicological/contamination and climate change concerns. Among biogeochemical models DeNitrification DeCompostion (DNDC, Changsheng Li et al. 1992) is a relatively simple model handling both biogeochemistry of denitrification and microbial growth (Li et al. 2000), to question the use of soil contamination data in climate change scenarios.
Aliquots of this sieved soil were used to measure the initial water content in addition to the maximum water holding capacity (WHC) for the further microcosm experiments. This site is located at Feucherolles near Paris,

95
France, and had been designed to evaluate urban compost fertility together with the monitoring of contaminant inputs (Cambier et al. 2019). Soil is a luvisol with 15% clay, 78% silt and 7% sand, a pH of 6.9, organic carbon (Corg) and total N contents at 10.5 ± 0.2 and 1.00 ± 0.03 g kg -1 soil, respectively, and with a CEC of 7.9 ± 0.8 cmol + kg -1 soil. This soil is not contaminated with Cu, and basal [Cu] measured by ICP-AES after HF-HClO4 extraction was of 12 mg Cu.kg -1 soil.

Experimental setup
In order to evaluate the impact of soil moisture on the sensitivity of nitrification to Cu toxicity, we carried out a two-step experiment. The first step consisted in 5 different WHC incubation during 45 days, and the second to a 3-day bioassay with spiked Cu gradient ( Fig. 1).
Five microcosms were built up with about 5g of sampled soil. Three of them were set up with a constant 105 moisture corresponding to 30%, 60% and 90% of their WHC in order to span respectively limiting, optimal, and saturating conditions for the microbial activities. These three samples will be called thereafter "30%, 60% and 90%", respectively. Their water contents were verified by weighting every two days and water added if necessary.
The two other microcosms were incubated in order to simulate two kinds of drought and dry-rewetting cycles.
One, thereafter called "Drought" (or DO), started with one week at 60% WHC and then the soil was left for 3 110 weeks without added water to mimic a dry period until 10% of the WHC before rewetting at the initial 60% WHC.
The other, thereafter called "Dry-rewetting" (or DR) encountered alternatives cycles of one-week dry period (10% of the WHC) followed by one-week near-saturation period (90% WHC). The moisture states of microcosms were performed by air-drying and controlled by weighting. analyses of NO3and NO2by colorimetric determinations, following the reduction of NO3in NO2by vanadium(III) and then the detection of NO2by the acidic Griess reaction (Miranda et al. 2001 T: Time of incubation.

135
Nitrification and denitrification processes are represented following the DNDC model proposed by Changsheng Li et al. (1992) and Li et al. (2000). In this study, we used a simplified version adapted by Zaehle and Friend (2010) initially calibrated for soil WHC >40%, that we intended here to test for 30% of WHC. This simplified version needs less boundary data but keeps a mechanistic description of the main processes.  The N-NH4 nitrified is transformed into N-N2O, N-NO or N-NO3 due to bacterial processes and chemonitrification following Eq. (6), (7) and (8): with and 2 two fixed rates (d -1 ), ftv a rate modifier controlled by temperature and given in Eq. (9) and R the ideal gas constant.

165
Then, the N-NO3 produced during the nitrification process enters the denitrification module where it is reduced sequentially into N-NOx, N-N2O or N-N2 following Eq. (10) to (12): The anaerobic fraction anvf is described following Eq. (13): with _ 2 , _ 2 being the partial pressure in the air and in the soil respectively. _ 2 is calculated following Eq. (14) (14) with SOC being the soil organic carbon stock (gC m -2 ), k the decomposition rate, _ 2 the O2 partial pressure related to the respiration, and fCu the effect of Cu on CO2 emissions as define in Eq.(15), following (Sereni et al. 2021;Eq. (5): The relative growth rate of N-NO3, N-NOx and N-N2O denitrifiers are described respectively by 3 , , 2 following Eq. (16), (17) and (18).
_ 2 ( ) being rates modifiers depending on air temperature and soil pH described in Eq. (19) to (22). We used measures of N species at the end of preincubation period as initial values of N species for DNDC 210 (Table 1a and Fig. 2.). To estimate the anaerobic volume fraction during the 3 days bio-assay, we used a C mineralization rate k (eq 14) determined on the basis of measurements performed on the same soil (Annabi et al. 2007) and ran DNDC for a 45 days equilibrium period. We then extracted the initial anaerobic volume fraction and partial O2 pressure. To estimate the effect of [Cu] and soil moisture on the different variables measured, we used nonparametric Kruskall-Wallis test. The fits between the model and the data were measured using root mean square error (RMSE,

225
Eq. (24)): where i is the number of observations (1 to N), X is the predicted value and is the observed value. RMSE was decomposed in standard bias (Eq. (25)), non-unity slope (Eq. (26)) and lack of correlation (Eq. (27)) component following Gauch et al. (2003), with ̅ and ̅ the mean modelled and observed values, b the slope of the least 230 squares regression of Y on X and r² the square of the correlation: All the analysis were done with R 3.2.3 (R Core Team, 2015).

Effect of Cu on potential nitrification activity (PNA): statistical model selection
The soil N species measured at the end of the soil pre-incubation in each soil moisture treatment were used to initialise the DNDC model (Table 1). Two anomalous points leading to anomalous calculated N-NO2 values were 240 excluded from the experimental results because of technical problems during measurements (the C replicates in the DR and DO cases).
The bioassay experiments performed at the end of the soil pre-incubation allowed us to determine the rate of nitrate production as a function of soil [Cu] for each soil moisture (Fig. 1)

3.2.a. Set up of the DNDC-Cu model
The DNDC model was originally constructed to model both C and N soil cycles. The relative proportion of 270 nitrification and denitrification processes thus depends on soil aerobic fraction determined both by soil C respiration and soil moisture (Eqs. (13) and (14)). Before any addition of Cu function in DNDC, we estimated this soil aerobic fraction, arising from C mineralisation and the aerobic fraction. Therefore, we used previous data from 366 days incubations made on the same uncontaminated soil (Annabi et al. 2007

280
constant rate of C mineralisation, k, was adjusted to take into account Cu through the eq 13 and the eq 28-31 adjusted N-NO3 production rate ( Fig. 1)

295
However, for the 30% WHC, RMSE was 139.9 thus 3.7 times more than for the other treatments (Suppl. Fig. 1).
Despite the reduction of nitrate production rate from 0.2000 to 0.1753 gN.hour -1 (see material and methods), soil nitrate stock was still slightly overestimated in the 90% WHC as shown by the largest lack of correlation in this case compared to the 60% WHC treatment (Fig. 3a, suppl. Fig. 1

3.3.a. Effect of soil [Cu]on soil N stocks.
The soil Cu function we included in the DNDC-Cu model modified specifically the default nitrification equation in complement to pH, soil moisture and O2 availability (Eq (2.)). In the presence of low [Cu] (12-512 mgCu.kg soil -1 ), the predicted N-NO3 soil stocks were found equivalent between 60 and DO and, to a less extend, DR 310 treatments (Suppl. Fig. 2

3.3.b. Estimation of soil N emissions under various moistures
Large differences are predicted in the N-NH4, N-NOx and N-N2O fluxes between the 90% WHC soil and the 3 other soil moisture treatments (Fig. 5). For instance, we modelled a decrease comparable in N-NOx emissions between DR/DO and 60-90% WHC for soil [Cu] about 112 mgCu.kg soil -1 (2-3% respectively - Table 2a and 2b.) but with the increase in soil [Cu], the variation of emissions between soil moisture became larger. Hence, around 340 2012 mgCu.kg soil -1 we modelled more than 50% decrease in N-NOx and 62% decrease in N-N2O emission fluxes for soils at 60% WHC against only 40% decrease in N-NOx and 54% in N-N2O emission for soils previously exposed to DR (Tables 2a and b.). Thus, intensities of fluxes between two moisture treatments reversed with an increase in soil Cu contamination.
The smallest fluxes were predicted for the wetter treatment despite higher modelled N-N2O stocks at 90% 345 WHC whatever [Cu]. N-NH4 fluxes were modelled higher for the DR soils than in the 60% WHC incubated for soil Cu higher than 1774 mgCu.kg soil -1 and smallest below. The emissions of N-NH4 in the DO treatment were predicted to be higher than those of the DR treatment for soil Cu higher than 1290 mgCu.kg soil -1 and smallest below (Fig. 5a). In the studied range of added Cu, N-NOx fluxes predicted by the model are largest from 60%WHC to DO, DR and 90% WHC (Fig. 5(b.)) for moderate Cu input (~ below 1380 mgCu.kg soil -1 ). The decrease in N-

350
NOx emission with the increase in soil [Cu] was however steeper for soils incubated at 60%WHC. Hence, at 2012 mgCu.kg soil -1 N-NOx fluxes in soil incubated at 60%WHC were similar to those in the soils incubated under drought treatment (Fig. 5 (b.)). N-N2O fluxes had similar trends than N-NOx for moderate Cu inputs but fluxes were still largest from 60% WHC to DO, DR and 90% WHC (Fig. 5 (c.)).
The N-N2O emissions fluxes in the presence of Cu were predicted to be 4 times smallest in the 90%WHC 355 treatment compared to the others, whereas N-N2 emissions were larger at this wettest treatment (Fig. 5 (d.)). ratio of emitted N-N2O per denitrification products (e.g N-N2O/ N-N2O+ N-N2) was hence smallest in the moistest soils, which means that the largest soils N-N2O stocks in the case of 90%WHC had more chance to be transformed rather than emitted (Fig. 6).

From laboratory experiment to soil N emission modelling
Thanks to our laboratory experiments, we were able to define a function modulating the soil N-NO3 production rates in relation with soil [Cu] and depending on the soil moisture. Our results showed that soil nitrate decreases with an increase in soil [Cu].

365
After implementing these Cu modulating functions into the DNDC-Cu model, we were able to reproduce the observed soil nitrate stock particularly for the soils incubated at 60 and 90% of WHC. The variability around the mean due to the Cu effect was also reproduced by our DNDC-Cu version at 30% of WHC despite strong underestimation of mean soil nitrate stocks due to model moisture-limit (Changsheng Li et al. 1992). In the case of the DR and DO incubated soils, the so-called "Cu function" also accounted for the effect of drought stress. In 370 fact, our Cu functions were defined on the basis of soil nitrate production against the whole gradient of Cu thus also considering the control without Cu. However, the (double) stress effect was also well reproduced in nitrate production.

375
In the present study, we predicted highest soil N-N2, N-N2O and N-NOx stocks in the moistest treatments as they are produced by the denitrifying bacteria expected to behave optimally at 90% WHC or after DR cycles (Changsheng Li et al. 1992;Homyak et al. 2017  control (Khalil et al. 2004) with a dominating nitrification for aggregates up to 0.25cm (Kremen et al. 2005). Pore size distribution under dry/rewet events is controlled by cracking, (des)aggregation (Denef et al. 2001;Cosentino et al. 2006) or gas displacement (Kemper et al. 1985) that we weren't able to take into account in the present study.
In DNDC, the calculation of denitrification rate and diffusion was based on a rough description of anaerobic zone 395 with approximation of soil pore space distribution (Li et al. 2000;Blagodatsky et al. 2011). The soil pore space distribution approach has been demonstrated to be more generally applicable (Arah and Vinten 1995;Schurgers et al. 2006) whereas soil aggregates have been shown to control the extend of nitrification and denitrification (Kremen et al. 2005;Schlüter et al. 2018). However, if models have been proposed to take O2 availability at the aggregate size into account in the nitrous oxide production (Leffelaar 1988;Kremen et al. 2005), they also point 400 out the difficulty in parametrization which need a large panel of soil measurements. Moreover, they are rarely transposable at the meso-and regional scale due to high spatial variations in soil structure (Butterbach-Bahl et al. 2013). The DNDC-Cu version we used here particularly pointed out the difficulty in dealing on biogeochemistry model with physical processes, with large discrepancies between modelled soils stocks and emissions. The validation we performed focused on soil nitrates stocks and a second step to go further on would be the measure 405 of gaseous species to ensure that emissions were also impacted by soils treatment. Moreover, we assumed here that soil [Cu] affected the C mineralisation with a decrease in soil O2 production leading to an increase in denitrification and N-N2O, N-NOx. Nevertheless, the present DNDC-Cu version didn't take into account the retroaction between C and N cycles. Further research would thus be required to include Cu contamination into C and N interacting cycles.

Climate change could substantially modify contaminated soil N emission
It is well known that climate change and rainfall patterns could substantially modify the soil N balance and its GHG emissions (Galloway et al. 2003(Galloway et al. , 2008Butterbach-Bahl et al. 2013  . We (Sereni et al. in press) previously/also 440 showed that soil Cu contamination differently affect soils nitrification depending of primary soil moisture stress.
Here we showed that the N-N2O and N-NOx emission variations are significantly more sensitive to the combined effect of Cu and precipitation regime than the nitrate stock. Based on these results, soil Cu inputs on moistest soils would lead to a largest decrease in soil N-N2O and N-NOx emission compared to that on driest soils, and even more than on soils submitted to abrupt and intense shifts in rainfall patterns as the DR and DO soils.

CONCLUSION
In the present study, we aimed at combining ecotoxicological experiments and biogeochemical modelling focusing in biogeochemical modelling: i) the difficulty to take into account hydrological dynamics (produced N-NO3 and N-NH4 could be expected to leach) and soil structures at different spatial scale (denitrification is estimated based on rough estimation on anaerobic soil volume which also controlled emissions rates through diffusion processes) and ii) linking soil function to microbial dynamics, in particular in this case were only the N-NO3 stock was measured (without dealing between production and consumption for instance). Despite these two main points of 470 uncertainty, the combination of incubations and of modelisation we conducted here emphasize the need to account for soil contamination when dealing with soil GHG emission modelling and climate change, as both contamination and rainfall patterns affect in a different way the soil N-NOx and N-N2O emissions.