4 RESULTS & DISCUSSION
4.2 Power of stable isotopes to characterize N 2 O dynamics in soils
4.2.1 N2O production processes in soils – Natural abundance approach Analyses of the stable isotope composition of N2O using natural abundance approaches ( δ15Nbulk; δ18O and SP) were useful in identifying sources of N2O in sub-Arctic peatlands and tropical soils, but not without limitations. We found no difference between BP soil δ15Nbulk values of N2O emitted (-17 ‰ to -8 ‰) with that of BAF soils (-22 ‰ to -5 ‰) (Park 2005, Article I). Even though the values are scattered, δ15Nbulk values of N2O emitted by natural soils were comparable and within the range found for fungal and bacterial denitrification (-54 to -10‰) (Perez et al., 2006; Well et al., 2006; Koster et al., 2013). For the VSC soils, δ15Nbulk - N2O values (-49
‰ to 6 ‰) were on average more depleted in 15N and within the range reported for N2O from denitrification and nitrification (–90 to –40‰) (Baggs et al., 2008 and references therein). Hence, the δ15Nbulk results suggest that denitrication plays a major role in the N2O emissions from the natural soils here while nitrification has also importance in the N2O production in the agricultural soil. However, the possibility that N2O is produced also by other processes (e.g. nitrifier-denitrification) cannot be completely ruled out based on the δ15Nbulk values of N2O. Thus, the δ15Nbulk has limitations to differentiate the N2O sources.
Recently incubation experiments and mathematical models have been used to show that isotope exchange and aparent isotopes effects occur that control the δ18O values of N2O produced both in aerobic and anaerobic conditions (Snider et al., 2012;
Lewicka-Szczebak et al., 2014; Lewicka-Szczebak et al., 2016). The δ18O values of N2O produced in nitrification (aerobic environments) have been suggested to be between +13 ‰ and +35‰ while N2O from nitrifier-denitrification shows more depleted values but also more scatter ranging from -5.4 ‰ to +46.6‰ (Snider et al., 2012). For denitrification, δ18O-(N2O/H2O) values (related to soil water δ18O values) after correction for N2O reduction are expected to be around +17.5‰ and larger values (>
+19‰) might indicate the contribution of other processes (Lewicka-Szczebak et al., 2016). Thus, the power of using δ18O in separating sources of N2O is weak, which is expected since oxygen in N2O may exhanges with water during its production. In this work, the δ18O values of N2O from the studied soils were between 3 ‰ and 38
‰, with one exception, a low value of -6% from the BP soil, and they are thus within the range of previously reported values. We cannot derive any information from this signature, other than concluding that nitrification, nitrifier-denitrification and denitrification, might take place simultaneously in these soils. More studies on δ18O values of N2O are required in order to improve δ18O as a tool to differentiate the N2O
production pathways. Overall, our results show that source partitioning based on isotope bulk composition (δ15Nbulk, δ18O) of N2O emissions in situ remains difficult, mainly because the different pathways produce N2O with a wide and overlapping range of isotope values. This is particularly true if δ18O values of N2O are used.
The individual 15N site preference (SP) values of N2O emitted from the study sites varied from -85 ‰ to 58 ‰ for BP soils, -37 ‰ to 33 ‰ for the VSC soils and 1 ‰ to 25 ‰ for BAF soils (Park 2005, Article I). They were highly variable and partly overlap with the range of SP values of known processes reported from pure culture bacteria studies on denitrification (-11 ‰ to 0 ‰) (Toyoda et al., 2005; Sutka et al., 2006) and nitrification (27 ‰ to 36 ‰) (Sutka et al., 2006; Decock and Six, 2013 and references therein). Due to the fact that various nitrifying and all denitrifying processes (e.g., denitrification, nitrifier-denitrification, and fungal denitrification) can occur simultaneously in soil, it is more reasonable to compare in situ SP values with the values obtained in soil incubations and in situ field measurements. With the exception of one extremely low SP value found in BP (-85‰) at the beginning of the growing season, SP values in this study (SP= -36‰ to 58‰) are in the range of SP values reported from soil N2O fluxes and N2O concentrations from in situ measurements (chamber and profile) and laboratory incubations, summarize by Toyoda et al., (2017) (SP= -40‰ to 90‰). The extreme SP values reported from BP soils, -85‰ and 58‰, are from the peat surfaces with the lowest N2O fluxes, consequently the correction needed for mixing of soil emitted N2O and atmospheric N2O was large for these two samples. To avoid biases in the interpretations of results we excluded those cases where N2O concentration difference between the chamber headspace and the troposphere was less than 10 ppb. Nevertheless, this was not the case for these samples, were the N2O concentration after the correction were 41 ppb (SP=85‰) and 12.6 ppb (SP=58‰) and there was not any other indications that these results were unreliable.
A shortcoming when comapring in situ SP values with the values obtained in soil incubations, is that only limited soil incubation studies with total soil microbial communities exist, showing rather contrasting SP results, particularly for nitrification processes. The incubation studies available found SP values from 19 ‰ to 36 ‰ for nitrification (Well et al., 2008), from -25 ‰ to -8 ‰ for nitrification/nitrifier-denitrification (Perez et al., 2006; corrected values) and from 1‰
to 21 ‰ for denitrification (Perez et al., 2006; corrected values; Well et al., 2006; Well
& Flessa, 2009; Meijide el al., 2010; Lewicka-Szczebak et al., 2014). The individual SP values from N2O emitted from the BAF soils overlapped with the SP values for denitrification while SP values from the BP and VSC, with the exception of two very light value (-85 ‰ in BP and -37 ‰ in VSC) overlapped with the SP values for nitrification, nitrifier-denitrification and denitrification. At this point, SP values seem to be more valuable for identifying temporal shifts in microbial pathways, rather than partitioning overall sources of N2O. Such usefulness is strengthen when compared with other soil properties such as water content and dissolved oxygen.
In addition to biological processes, N2O can be produced also by abiotic pathways under ambient conditions, e.g. through, chemical and photochemical reactions (Rubasinghege et al., 2011, Heil et al., 2015) or reduction of nitrate (NO3-) and nitirte (NO2-) on active surfaces of Fe(II)-containing mineral, procees know as chemodenitrification (Peters et al., 2014 ). These studies suggested that N2O produce by abiotic pathways would have an isotopic signature similar to that produced by biotic mechanism and indistinguishable from that expected from a mixture of several microbial processes. However, to occur they would require high NH4+ concentrations (Rubasinghege et al., 2011) and high pH (Heil et al., 2015), conditions that were not present in the BP soils during the sampling period in 2011. Also high Fe levels and high NO3- concetrations favour abotic production of N2O in soils (Peters et al., 2014 ).
In BP soils NO3- comcetrations were high, but the levels of Fe are unknow, and the contribution of abiotic patways to N2O producction could not be determined from isotopic composition alone.
4.2.2 N2O production processes in soils - 15N tracer approach
The 15N tracer study was only conducted in the sub-Arctic BP. There, evidently the
15N approach in situ did not disturb the BP soil system. The N2O fluxes from the labeled plots (~10 mg N2O m-2 d-1) were within the range of those measured in previous years from the BP soil (1.9 – 31 mg N2O m-2 d-1) (Repo et al., 2009;
Marushchack et al., 2011). Furthermore, the concentrations of NO3- or NH4+ never exceeded those observed before. Some difference between the N2O emissions from the BP labeled and non-labeled plots could be attributed to the natural spatial variability within the BP surfaces. During the field campaigns in 2010 and 2011, the coefficient of variation (CV %) in the N2O emissions using three chambers at BP (<
1m distance) was large (41 to 89%) but high spatial variation in N2O emissions is typical for soils in general (14 to 160%) (Butterbach-Bahl et al., 2002; Savage et al., 2014; Cowan et al., 2015).
We found typical recovery of the 15N label in N2O, which gave us confidence on the label technique. However, there were significant differences in recovery of 15N in the soil inorganic N forms. One possible explanation for this could be that the pro-duction of N2O occurred from NO2- rather that NO3- and the isotopic enrichment of N2O would be much more similar to that of NO2- than of NO3- (Mulvaney et al., 1997).
However, the presence or accumulation of NO2- was never detected in the soils before or during the 15N tracer experiment. This is expected because in soils, NO2- typically is rapidly consumed and; therefore, we could not detect it in the soils extracts. An-other feasible explanation is the dilution of the label in inorganic N sources during the preparatory steps of the microdiffusion. During microdiffusion the errors associ-ated with the breakdown of labile soil organic N and alteration of target 15N abdance by N contamination of reagents, are common causing the main source of un-certainties in the estimation of 15N in the mineral N in soil extracts (Robinson, 2001).
This is true also for soils extracts from soils with complex organic matrix such as peat soils (Mulvaney & Khan, 1999). Methodological experiments conducted after this field labeling experiment, suggest significant underestimation of label in mineral N, particularly in NH4+, by microdiffusion, if amino acids are present (unpublished re-sults; manuscript in preparation). Therefore, it is possible that during the microdiffu-sion of the BP soil extracts, labile organic matter (e.g amino acids) was hydrolyzed to NH4+ causing the dilution of the label 15N in the NH4+ extracted from the soil. Incom-plete recovery of the mineral N could also result in 15N-depleted values of NH4+ and NO3- during microdiffusion (Fry, 2006). In this study, higher amounts of N-NH4+ in BP soil extracts where measured by colorimetry (UV-vis spectrophotometry) com-pared to those measured by mass spectrometry (EA-IRMS) while the opposite was found for N-NO3-, suggesting incomplete recovery of N-NH4+ but not of N-NO3-. On the other hand, determination of N-NH4+ by colorimetry (Berthelot reaction) is not entirely specific to N-NH4+ and interferences from organic N compounds (e.g amino acids) are known to occur resulting in overestimation of the amounts of N-NH4+ in the solution (Herrmann et al., 2005). Consequently is it possible that labile organic-N present in BP soils extracts contributed to the dilution of the 15N label of NH4+ and NO3 during microdiffusion, but also to the overestimation of the N-NH4+ concentra-tions from soil extracts by colorimetric analysis. Since the native concentraconcentra-tions of mineral N in VP soils are very low, it is also possible that the label added was imme-diately immobilize by microbes and assimilated by mosses. In BP soils, also physical interactions and chemical reactions with the organic matter might have affected the recovery of the label.
The 15N-enrichment approach may be desirable since fractionation can be ignored, but is quite labor intensive. The unresolved processes causing the lack of 15N mass balance in the organic soils, might also be occurring with natural abundance studies.
This could represent a problem since simple mixing model cannot be applied because the enrichment on 15N2O might be higher than the enrichment in the inorganic N forms, violating the basic assumption of an isotope mixing model. The basic assump-tion of an isotope mixing model to separate sources is that the isotope composiassump-tion of the mixture (here N2O) lies between the one of the contribution sources (here NO 3-and/or NH4+). However, the methodology used in this study for N2O source parti-tioning was largely independent of the 15N in inorganic N pools, thus the results were achieved with high confidence. On the other hand, high uncertainty remains in the gross mineralization and nitrification results, because of the problems in determina-tion of nitrogen isotope composidetermina-tion in soils with high content of organic matter.