Regulation of CO2 and N2O fluxes by coupled carbon and nitrogen availability

Carbon (C) and nitrogen (N) interactions contribute to uncertainty in current biogeochemical models that aim to estimate greenhouse gas (GHG, including CO2 and N2O) emissions from soil to atmosphere. In this study, we quantified CO2 and N2O flux patterns and their relationship along with increasing C additions only, N additions only, a C gradient combined with excess N, and an N gradient with excess C via laboratory incubations. Conventional trends, where labile C or N addition results in higher CO2 or N2O fluxes, were observed. However, at low levels of C availability, saturating N amendments reduced soil CO2 flux while with high C availability N amendments enhanced it. At saturating C conditions increasing N amendments first reduced and then increased CO2 fluxes. Similarly, N2O fluxes were initially reduced by adding labile C under N limited conditions, but additional C enhanced N2O fluxes by more than two orders of magnitude in the saturating N environment. Changes in C or N use efficiency could explain the altered gas flux patterns and imply a critical level in the interactions between N and C availability that regulate soil trace gas emissions and biogeochemical cycling. Compared to either N or C amendment alone, the interaction of N and C caused ∼60 and ∼5 times the total GHG emission, respectively. Our findings suggested that the response of CO2 and N2O fluxes along stoichiometric gradients in C and N availability should be accounted for interpreting or modeling the biogeochemistry of GHG emissions.


Introduction
Carbon dioxide (CO 2 ) and nitrous oxide (N 2 O) are major greenhouse gases (GHGs) that produce a strong positive radiative forcing in the atmosphere. Extensive work has been directed to understanding single substrate dependences of CO 2 on organic carbon (C) and N Burzaco et al 2013). However, the plasticity of C and N metabolism in microorganisms (Ter Schure 2000, Horák 1997) produces large uncertainties in coupling of either CO 2 or N 2 O trace gas emissions to single substrate availability.
Recent evidence suggests N availability can influence CO 2 production and in turn C availability may influence N 2 O emissions (Piao et al 2013, Jain et al 2013, Liu and Greaver 2009. Studies have generally not evaluated emissions of both trace gases simultaneously, although potential interactions between substrate availabilities may lead to important connections between the two fluxes through a coupling of the C and N biogeochemical cycles (Sokolov et al 2008, Thornton et al 2009, Bonan and Levis 2010, Zaehle and Dalmonech 2011, Lal 2008. The interactive influence of C and N substrate dependences on the biogeochemical processes mediating soil CO 2 and N 2 O fluxes remains a key uncertainty in understanding the regulation and magnitude of GHG emissions from soils. As directly measured byproducts of microbial C or N metabolism, CO 2 and N 2 O fluxes provide a window to inspect the energy (C) and nutrient (e.g. N) allocation of soil microorganisms through direct relationships with C and N use efficiency (CUE and NUE). In ecological stoichiometry, CUE or NUE is commonly applied to quantify the balance of C or N between biomass growth and consumption (Mooshammer et al 2014, Manzoni et al 2012, Roland and Cole 1999. In general, a high CUE or NUE means an increasing microbial biomass but slowed C or N mineralization rate, resulting in low soil CO 2 or N 2 O fluxes. In contrast, a low CUE or NUE indicates an inefficient conversion of C or N to biomass, a large return of C or N to the environment, and increased soil CO 2 or N 2 O fluxes. A limiting C substrate produces a relatively high CUE while a limiting N source can reduce the CUE, a consequence of coupling or uncoupling of microbial catabolism and anabolism (Sinsabaugh et al 2013). Microbial NUE is likely controlled and regulated similarly to CUE but directly coupled to the N cycle and associated emissions of N trace gases (Mooshammer et al 2014). Thus, variation in CO 2 or N 2 O flux patterns can be used as an assessment of CUE or NUE under different C or N levels (Eberwein et al revised). Because of the intrinsic linkage between microbial C and N metabolism (Richardson 2000, Robertson andGroffman 2007), how the overlap between microbial CUE and NUE simultaneously mediates CO 2 and N 2 O fluxes needs evaluation.
In this study, we conducted a series of soil incubations to identify the potential interactions between soil CO 2 and N 2 O emissions in response to variation in labile C and N amendments. We asked: 1) are soil CO 2 and N 2 O soil emissions dependent on the availability of both C and N, and 2) are emissions of the two trace gasses correlated in their flux rates? Answering these questions will test alternate hypotheses of trace gas emission regulation, 1) a single-substrate hypothesis currently used in most trace gas emissions models that predicts regulation by a single resource and 2) a dynamic efficiency hypothesis for C and N that predicts interactions between resources will regulate both CO 2 and N 2 O fluxes. The results from this study will improve understanding of how both C and N biogeochemical cycles are influenced by multiple limiting resources and demonstrate the potential coupling between these biogeochemical cycles with direct consequences for total GHG emissions.

Soil characterization
The soil used for our study was collected from an agricultural field (13 ha) located at the University of California Desert Research and Extension Center, El Centro, California (32°N 48′ 42.6′, 115°W 26′ 37.5″). The site is a high temperature, low elevation, desert environment with mean annual precipitation of 5.8 mm and monthly mean air temperatures between 13.9 and 33.9°C, and extremely high midday temperatures up to 50°C (www.weather.com). The site has deep alluvial soils (42% clay, 41% silt, 16% sand) with 2.34% C and 0.13% N, and a pH of 8.3 (Oikawa et al 2014). Prior to soil collection, the field was fallow for 8 months then planted with forage sorghum for two years. Soils were collected between 0-10 cm depths from 5 random locations in the field.

Laboratory incubations
Prior to incubations, the soil was air dried in the lab, sieved (2 mm mesh), and then homogenized. Soil water holding capacity (WHC) was determined by the gravimetric method (Pansu and Gautheyrou 2006). Three replicate samples (100 g dry weight) were placed in glass jars (∼473 ml) and maintained a 40% WHC by weighing the jar every two days and adding de-ionized water as necessary during the incubation period.
To investigate soil CO 2 and N 2 O flux responses to C and N amendments and their interaction, two series of laboratory incubations were conducted that included a control (de-ionized water only), dextrose (as a labile carbon source) only, N (ammonium nitrate, NH 4 NO 3 ) only, and both dextrose and N. The first series of incubations (Experiment 1, Exp1) were conducted to quantify soil CO 2 and N 2 O fluxes under six levels of N amendment with two levels of C amendment (with and without C). Six levels of N amendment as 0, 10, 50, 200, 700, or 1500 μg N g −1 soil were selected. Along the N gradient, a control and saturating C level (60 g L −1 dextrose, which is equivalent to 18 mg g −1 soil) were selected to investigate C and N interactions. Each treatment included three replicates with 36 samples in total. For the second series of incubations (Experiment 2, Exp2), another 36 samples were used to investigate CO 2 and N 2 O fluxes under different C level with saturating N supplement. A C amendment of 0, 1.5, 3, 7, 18, or 30 mg dextrose g −1 soil was set and combined with either no N or a saturating N level (700 μg N g −1 soil). All 72 jars were incubated at 25°C in the lab for 7 days and CO 2 and N 2 O fluxes were measured daily.

Flux measurements
We used a flux measurement system that allowed simultaneous measurements of both CO 2 and N 2 O trace gases within a total sampling period of less than five minutes. Soil N 2 O emissions have typically relied on syringe extraction over a thirty minute to one hour sampling period at minimum and subsequent analysis on a gas chromatograph (Alves et al 2012, Dobbie andSmith 2003). Our system provides the capability to measure the instantaneous fluxes of N 2 O and CO 2 and allows investigation of the potential relationships between CO 2 and N 2 O fluxes.
We built a dynamic closed system (figure 1) to measure CO 2 and N 2 O fluxes simultaneously with a Li-7000 infrared gas analyzer for CO 2 (Licor Biosciences, Lincoln, Nebraska, USA) connected to a N 2 O gas analyzer (913-EP, Los Gatos Research, Mountain View, California, USA). The N 2 O analyzer uses offaxis integrated cavity output spectroscopy (Off-Axis ICOS) to provide a real-time accurate N 2 O concentration measurement with a precision of 0.05 ppb at 1 hz sampling frequency. The CO 2 and N 2 O fluxes were determined by the linear regression fit between the CO 2 or N 2 O concentration change and the measured time. For CO 2 flux calculation, the original model from Licor (Licor 8100 Manual) was adapted for our jar measurements as follows: where F c is CO 2 flux (μmol CO 2 g −1 soil s −1 ) from the soil in the jar. V (cm 3 ) is the volume difference between the jar plus the tubing and the soil (calculated using a bulk density value 1.15 g cm −3 of our soil). P 0 is the initial pressure (kPa). w 0 is initial water vapor in mole fraction (mmol mol −1 ). R is the ideal gas constant (8.314 × 10 3 kPa cm 3 K −1 mol −1 ). M s is the mass of soil (g) and T 0 is the initial air temperature ⎦ is the changing rate of CO 2 concentration along time (μmol mol −1 s −1 ). The N 2 O flux was calculated using the same method but the dry N 2 O concentration reported from N 2 O analyzer during the measured intervals was used and thus the water correction term in equation (1) was not needed. Both trace gas measurements were completed within 3-10 mins depending on flux rate. Seven day cumulative CO 2 and N 2 O fluxes were calculated by interpolating the measurements from each day and then integrating.

Statistics
We performed two-way fixed-model ANOVA to test the response of the seven day cumulative CO 2 and N 2 O fluxes to carbon and nitrogen addition. Prior to conducting ANOVA, the normality of the data and the homogeneity of variances were tested using the Shapiro-Wilk test (Royston 1982, Shapiro andWilk 1965) and the Levene's test (Brown andForsythe 1974, Levene 1960), respectively. The Tukey's honestly significant difference (HSD) test (Tukey 1949) was used to examine intra-group differences. When necessary, Box-Cox transformations (Box and Cox 1964) were applied to meet the assumptions of ANOVA. For N 2 O fluxes, we added a constant positive value to meet the logarithmic transformation because of some negative values observed during the incubation in association with low rates of net uptake (Majumdar 2013). All statistical analyses and data processing were performed using MATLAB R2011b (The MathWorks Inc., Natick, MA, USA) and the R package (R Core Team 2013).

Results
3.1. CO 2 flux Cumulative CO 2 flux was significantly affected by C (p < 0.0001 in Exp1 and Exp2) and N (p < 0.0001 in Exp1 and p = 0.0006 in Exp2) amendments and their interaction effects (p < 0.0001 in Exp1 and Exp2) based on a two-way ANOVA. C effects on CO 2 flux were positive, indicating a higher dextrose concentration produced higher CO 2 flux (figure 2(a)). However, the N effects on CO 2 flux were diverse. Under saturating N conditions, the CO 2 flux was reduced 39% and 36% at low dextrose levels (1.5 and 3 mg g −1 soil, respectively) compared to treatments without any N amendment (figures 2(a)-(b) without N). But high dextrose levels (18 and 30 mg g −1 soil) resulted in significantly higher CO 2 fluxes with N addition compared to without N (figures 2(a)-(b) with N). Although no significant decrease in CO 2 flux was found at lower dextrose levels (1.5 and 3 mg g −1 soil) with N addition, there was a significant decrease in CO 2 flux at lower N addition when dextrose was saturating (figures 2(c)-(d) with dextrose). CO 2 flux increased at a high but subtoxic N level (700 μg g −1 soil) and decreased at an inhibitory N level (1500 μg g −1 soil). However, there were no significant effects of N addition on CO 2 flux when dextrose was not added (figure 2(c) no dextrose). . Soil CO 2 flux response to dextrose and nitrogen gradient. Panel a illustrates the CO 2 flux response to increasing C amendments with and without saturating N additions (at 700 μg g −1 soil). Panel c shows CO 2 response to increasing N amendments with and without saturating dextrose (at 18 mg g −1 soil). Panel b shows the CO 2 flux difference with N amendment (with N minus no N) at increasing C levels and d shows the difference with C amendment (with dextrose minus no dextrose) at increasing N levels. The significance level is set to p = 0.05. additional C increased CO 2 fluxes (figure 2(a) no nitrogen) but reduced N 2 O fluxes (figure 3(a) no nitrogen), resulting in a negative relationship between them. In contrast, when N availability was saturating, CO 2 and N 2 O fluxes increased with additional C amendments from low to high (figures 2(a) and 2(a) with nitrogen), resulting in a positive relationship. In an unlimited C environment, a negative N effect on CO 2 flux and a positive N effect on N 2 O flux resulted in a positive relationship between CO 2 and N 2 O fluxes. When N crossed a critical level (between 50-200 μgN g −1 soil in our study), the negative N effect on CO 2 flux switched and resulted in a positive relationship between CO 2 and N 2 O fluxes.

Discussion
Through a series of laboratory experiments we found important connections between C and N biogeochemical cycles with both resources important for CO 2 and N 2 O emissions. Additional C or N substrates caused an increasing soil CO 2 or N 2 O flux because of more C or N resources available for decomposition. The straightforward prediction of the relationship between trace gas fluxes and its primary substrate is true, although it masks substantial contributions from coupled C and N interactions on microbial activity. Our results show that N availability can substantially influence the effect of C availability on CO 2 emissions,  and C availability can alter N 2 O flux sensitivity to N addition by more than two orders of magnitude.
Notably depending on the stoichiometry of resource amendments, the effects of the secondary resource could both enhance or inhibit emissions of gases. CO 2 emissions were inhibited by N at non-saturating conditions and similarly N 2 O was inhibited by C at non-saturating conditions. However, saturating levels of both C and N accelerated trace gas emissions of both CO 2 and N 2 O. These divergent effects of altered resources with both enhancement and inhibition suggest complex interactions between C and N biogeochemical cycles, with substantial implications for predicting emissions of GHGs.

Regulating CO 2 flux by carbon use efficiency (CUE)
High CUE is commonly observed in response to C limitation (Sinsabaugh et al 2013), which results in a lower respiration rate. At saturating N and subsaturated, an increasing CUE could mobilize more C into microbes, leading to relatively lower CO 2 fluxes compared to those without N application (figures 2(a) and (b)). Owing to more N availability, C becomes limiting and soil microorganisms with a relatively fixed organismal stoichiometry require relatively more carbon for growth, which results in a higher CUE and reduced CO 2 fluxes. This dynamic CUE could explain why N addition triggered a decreased CO 2 flux. After C availability increases to a critical level, more C substrate leads to a decreasing CUE and increasing CO 2 fluxes (figure 2(a)). During N limited conditions (figure 2(c)), even when C substrate is saturating, a decreasing CO 2 flux occurs in response to relatively low levels of N addition. Exposed to an excess C source and restricted in growth by N, the microorganisms may adjust their metabolism, i.e., uncoupling catabolism and anabolism via energy spilling pathways associated with decreased CUE (Sinsabaugh et al 2013, Gallmetzer and Burgstaller 2002, Vrabl et al 2009, Larsson et al 1995. However, N amendment will alleviate N limitation and increase CUE. When N is not limiting, the microbes coupled catabolism and anabolism again and the excess C source will introduce a higher CO 2 flux associated with lower CUE (figure 2(c) with dextrose). Thus, our results support a hypothesis of dynamic CUE that can explain CO 2 flux response to C and N additions. However, the mechanism for soil microbial changes in metabolic pathway under different resource environments that allow adjustment of CUE is unclear and should be targeted for future research. Nevertheless, process models that incorporate a dynamic CUE to estimate CO 2 flux seem warranted.

Regulating N 2 O flux
From ecological stoichiometry, a higher NUE (related to a lower N 2 O flux) could be expected under N limitation (Mooshammer et al 2014, Sterner andElser 2002), which implies the limited N would be conserved primarily for growth. The decreasing N 2 O flux associated with low rates of extra C source (figure 3(a) no nitrogen; figure 3(b) with dextrose in lower N levels) suggests that more N has been used to build soil microbial biomass as the extra C is also distributed into growth, which results in a higher NUE and lower N 2 O flux. Alternatively, exogenous C source provides additional electrons (i.e., NADH) via carbon degradation pathways and the TCA cycle to reduce the N 2 O to N 2 by denitrifying enzymes (Giles et al 2012, Richardson 2000). Regardless of how N 2 O is generated from the N cycle pathways (either nitrification or denitrification) (Butterbach-Bahl et al 2013), NADH promotes reduction of N 2 O into N 2 via the electron transport chain. Such a reduction in the N 2 O:N 2 ratio in response to labile carbon substrates has been shown (Morley and Baggs 2010, Giles et al 2012, Lee and Jose 2003, Weier et al 1993, although the magnitude varies because of the divergence in C substrate quality, soil conditions and O 2 availability (Morley and Baggs 2010, Giles et al 2012, Lee and Jose 2003. N 2 O flux responses to additional C source under limiting N conditions may be regulated by NUE or the interaction between carbon and nitrogen metabolism through nitrification or denitrification pathways. As with CO 2 emissions, while the mechanism for variable NUE is unclear, these findings support the need for improvements in process models that account for resource stoichiometry and C and N interactions rather than N availability alone to estimate N 2 O emissions (Liu et al 2012, Jassal et al 2011).
4.3. Coupled CO 2 and N 2 O flux relationships Availabilities of C and N substrates simultaneously regulate CO 2 and N 2 O fluxes. From our results, the relationship between soil CO 2 and N 2 O fluxes can be switched from negative to positive (figure 4) based on the N supplement. The critical level for this switch might be a result of the switch in elemental requirement from C to N for microorganism growth. A threshold elemental ratio (TER), which is a parameter in quantifying when growth limitation switches from one element to another (Frost et al 2006, Sterner andHessen 1994), can control the metabolism of microorganisms (Mooshammer et al 2014). If the C:N ratio is above the TER, the metabolism of soil microbial communities is under N limitation and expresses a relatively higher NUE but lower CUE. The negative relationship between CO 2 and N 2 O fluxes occurs (figure 4 and figures 2(c) and 3(b) when N is low). In contrast, an expected lower NUE but higher CUE would occur when the C:N ratio is below the TER, which is a C limiting condition. The negative relationship still occurs between CO 2 and N 2 O when N is at the control level (figure 4 and figures 2(a) and 3(a) no nitrogen). When C and N availability are both available at high levels, elevated CO 2 and N 2 O fluxes will be produced and a positive relationship between them is seen (figure 4). Based on the contrary relationship between CO 2 and N 2 O fluxes, a critical level of C:N ratio could exist in regulating the response of soil microbial CUE and NUE to substrate availability and controlling the pattern of GHG emissions.

Implications for total GHG emissions
At global scale, about 80% of N 2 O emission is derived from agricultural ecosystems because of synthetic fertilizers used in agricultural soil management (Majumdar 2013, Davidson 2012. In high production agricultural ecosystem, a large amount of carbon substrate can be introduced into the soil via root exudation or residues (Oikawa et al 2014). These substrates have a large influence on CO 2 emissions and as suggested here may also influence N 2 O fluxes. Similarly, our results suggest patterns of fertilization may also have direct effects on CO 2 emissions. With N 2 O having a much higher warming potential (298 times that of CO 2 over 100 years) (IPCC 2007) than CO 2 , these interactions between C and N biogeochemical cycles may have important consequences for net emissions (figure 5). Without additional N, total GHG emissions increased linearly with C additions, while without additional C, total GHG emissions increased minimally with N additions. The largest increases occurred when both C and N were added and total GHG emissions were ∼70, ∼5 and ∼60 times higher than the control samples, C or N amendments, respectively. Extending these findings to the field is a clear research need for understanding how soil emissions of both CO 2 and N 2 O contribute to total warming potential in response to coupling between N and C cycles. The divergent effects from limited C or N on CO 2 and N 2 O fluxes result in an opposite relationship between them, suggesting the possibility to minimize total GHG emissions by optimizing fertilizer level and timing relative to growth in agricultural management. Moreover, these results highlight the importance of C and N interactions for the ability to understand and predict GHG emissions using biogeochemical models.