Opposing patterns of carbon and nitrogen stability in soil organic matter fractions compared to whole soil

Nitrogen (N) released from soil organic matter (SOM) is quantitatively important for crop uptake, even when adequate fertiliser N is supplied. Understanding of SOM has shifted to recognise distinct fractions that correlate with properties such as turnover time, carbon (C) and N content, and chemical composition. Yet, how these fractions relate to N supply from SOM is poorly understood. This study aimed to link N mobilisation and C stability in coarse (≥50 μm) and fine (≤50 μm) fractions, and evaluate the roles of these fractions in supplying N in cropping soil. Soils from long‐term continuous cotton, cotton‐vetch, and cotton‐wheat rotations and a nearby uncleared site (NV) were separated into coarse and fine fractions, left as whole soil, or dispersed and freeze‐dried as a fractionation control. Initial C chemistry in fractions and whole soils was measured by solid state 13C NMR spectroscopy. N mobilisation and CO2 loss were measured over a 14‐day incubation experiment. In the cropping soils, net immobilisation of N was measured in the separate fractions, while net mobilisation was found in the whole soils. In the NV soil, N mobilisation was greater in the fine fraction. C mineralisation followed the order fine fraction > fractionation control = coarse fraction > whole soil. C stability was best explained by physical protection within whole soil structure rather than chemical recalcitrance or mineral stabilisation. The results revealed an unexpected contrast between C and N mineralisation from SOM fractions and demonstrated the importance of soil aggregates for SOM stability. We show a cautionary impact of fractionation on C and N dynamic, highlighting the need for further research to understand the synergistic behaviour of SOM fractions in whole soils.


| INTRODUCTION
Soil organic matter (SOM) provides an essential store of carbon (C) and supply of nitrogen (N) in soils.These functions are inherently linked, with net storage of C implying storage of N, and mineralisation of N likely associated with mineralisation of C.This presents a challenge in cropping systems, where increasing soil C and ensuring an adequate supply of N to a crop are seemingly contradictory goals (Janzen, 2006).For N, this is circumvented in agricultural systems by providing mineral N as fertiliser.While this may help to reduce N mining and loss of SOM (Alvarez, 2005;Angus & Grace, 2017), there is often poor capture of N leading to large pools of inorganic N prone to loss (Macdonald et al., 2017).Furthermore, there is increasing evidence that despite an adequate supply of N from fertilisers, the soil can provide a significant portion of the N taken up by a crop (Gardner & Drinkwater, 2009;Yan et al., 2020).For example, in Australian cotton it has been shown that up to 75% of crop N can be derived from the soil, likely through mineralisation of SOM and carry over of N from the previous season (Macdonald et al., 2017;Rochester & Bange, 2016).This prescribes a need to improve our understanding of N release from SOM stores and integrate this with fertiliser management.
Despite the inextricable relationship between C and N in soil, their behaviour is often studied in isolation, leading to a disconnected understanding of the processes controlling storage and loss from SOM.Much of the research on SOM has focussed on the processes controlling the storage of organic C in soils with the aim of increasing soil C. Through this body of work, it has been established that SOM can be conceptually divided into fractions with distinct origins, stability, chemistry, residence time, C and N contents and association with soil minerals (Heckman et al., 2022;Lavallee et al., 2019;von Lützow et al., 2007).Many methodologies for separating SOM fractions exist (Poeplau et al., 2018), but at the highest level typically involve a distinction between particulate organic matter (POM) and mineral-associated organic matter (MAOM; Cotrufo et al., 2019;Lavallee et al., 2019).The MAOM fraction often contains the largest pool of C and N in soils (Baldock et al., 2013;Jilling et al., 2020).Relative to POM, it is considered most important for C storage due to a longer residence time and slower turnover attributed to stabilisation by association with clay and silt minerals and physical protection inside microaggregate structures (Haddix et al., 2020;von Lützow et al., 2007).However, other attributes, including a lower C:N ratio, simpler biochemical structures and the overall large store of N suggest that MAOM could account for a significant portion of N mineralised from SOM (Amorim et al., 2022;Daly et al., 2021;Jilling et al., 2018).This suggests an inequality in the processes controlling C and N availability in SOM fractions that requires further research.
Most research on N mineralisation has investigated the SOM pool as a whole, with few studies on SOM fractions.Some studies have investigated N mineralisation from fine or mineral-associated fractions, often finding that it is a large contributor to total soil N mineralisation (Bimüller et al., 2014;Parfitt & Salt, 2001;Turner et al., 2017).These studies also show lower C mineralisation from fine fractions and conclude C and N cycling is decoupled in these fractions.However, it remains unclear which factors influence this.For example, Parfitt and Salt (2001) found C mineralisation was related to O-alkyl C content, which is representative of labile carbohydrate C (Baldock & Preston, 1996), and N mineralisation inversely correlated with C/N ratio of soil fractions, whereas Bimüller et al. (2014) concluded neither was related to C and N mineralisation.Most studies speculate on mineral interactions with C and N compounds likely influencing microbial N immobilisation or mineralisation, but none adequately address this.It is further likely that the disruption caused by fractionation procedures affects C and N mineralisation, but not all studies account for this in experimental design.There is a particular lack of experiments with cropping soils which typically have lower C levels.These points highlight gaps in our understanding of C and N cycling in SOM fractions that may be hindering progress towards improved N management and C storage in agricultural soils.
In this study, we aimed to investigate the links between C stability and N mobilisation in coarse and fine SOM fractions and measured the contribution of each to overall soil N availability.We refer to 'N mobilisation' rather than 'N mineralisation' because we considered measurements of dissolved organic N (DON) in the available pool.In addition, we evaluated the effects of physical fractionation and SOM chemistry on these processes.The experiment included soil from three cotton crop rotation systems and a nearby native vegetation (NV) site to evaluate if differing soil C levels and inputs affect fraction N mobilisation.Based on the current understanding of SOM dynamics, we predicted that losses of C and N availability from the fine fraction would be lower than the coarse fraction due stabilisation via mineral association and lower labile C content.We further predicted that disruption of aggregates due to fractionation would lead to greater destabilisation of C and N in both isolated fractions and whole soils.

| Soil collection
Soil was collected from a long-term cotton crop rotation and N fertiliser rate field experiment located at the Australian Cotton Research Institute, Narrabri, New South Wales, Australia (30 12 0 09.1 00 S 149 35 0 44.8 00 E), described in Rochester and Bange (2016).The soil at this site is classified as a self-mulching Grey Vertosol using the Australian Soil Classification (Isbell, 2016) or Haplic Vertisol (IUSS Working Group WRB, 2022).During the field experiment, N fertiliser was applied as urea or anhydrous ammonia at rates between 0 and 320 kg N ha À1 .The trial had ended in 2018 after a cotton rotation; the site had since been managed with winter wheat/summer fallow in 2018 and 2019, and no additional fertiliser applied.Soil was collected in March 2020.At the time of sampling, the trial site was in fallow with wheat stubble retained from the previous season across all plots.
Soil samples were collected from the 0 N rate plots of the continuous cotton (CC; Gossypium hirsutum L.), cotton-vetch (CV; Vicia villosa Roth.) and cotton-wheat (CW; Triticum aestivum L.) rotations within the trial.Surface stubble was removed, then soil collected from 0 to 100 mm depth.Approximately 3 kg of soil was collected randomly from 4 replicate plots and combined to form 1 sample per treatment.Additionally, soil was collected from an area of nearby natural vegetation (NV), approximately 300 m from the field trial.Plants in this area included native and introduced grasses and herbs (Malva parviflora L., Tribulus sp.L., Urochloa panicoides P.Beauv., Seteria sp.(L.) P.Beauv., Chloris virgata Sw., Chloris truncata R.Br., Paspalidium jubiflorum (Trin.)Hughes) with a dominant canopy of Corymbia terminalis (F.Muell.)K.D.Hill & L.A.S.Johnson, and occurrences of Eucalyptus camaldulensis Dehnh.After collection, the soils were sieved to 2 mm and airdried at 40 C. Soil properties are provided in Table 1; whole soil Organic C and total N are reported in Table 2.

| Soil organic matter fractionation
Soils from the four management systems (CC, CV, CW and NV) were separated into coarse (≥50 μm) and fine (≤50 μm) fractions using a method adapted from Baldock et al. (2013).This method of particle size fractionation was chosen as the resulting two fractions are chemically distinct, thus likely differ in the rate of N and C mineralisation, and it aligns with recommendations for consistent definitions of SOM fractions (Baldock et al., 2013;Lavallee et al., 2019).Both fractions contain some amount of resistant organic C (ROC; Baldock et al., 2013); this was quantified by 13 C NMR (Figure S1), however its contribution to N or C cycling could not be distinguished.As such, we refrain from referring to the fine fraction as mineral-associated-and coarse as particulate organic C.
Portions of soil (10 g) were dispersed by mixing with 40 mL of deionised (MilliQ) water and shaking overnight at 180 rpm on an orbital shaker.Water was used rather than a chemical dispersant (e.g., sodium hexametaphosphate) to avoid large excesses of salts in the subsequent incubation experiment.A comparison of water and hexametaphosphate dispersion showed no significant difference in fraction C and N content (Table S1).Two portions at a time (20 g soil) were fractionated using an automated sieve-shaker (Vibratory Sieve Shaker Analysette 3 PRO; FRITSCH GmbH, Idar-Oberstein, Germany) fitted with a 50 μm sieve and lid containing two water spray nozzles delivering MilliQ water via a peristaltic pump (Masterflex L/S Model 7553-79 attached to Masterflex L/S Modular Controller; Cole-Parmer, Vernon Hills, IL, USA).The shaker ran at an amplitude of 2.5 mm and water flow rate of 150 mL min À1 for a minimum of 3 min, or until no more fine particles flowed through the sieve.Taking care to minimise loss of soil material, the ≥50 μm coarse material collected above the sieve was transferred to a 50 mL centrifuge tube, while the ≤50 μm fine material collected below the sieve was transferred into two 500 mL plastic bottles.Both fractions were frozen, then lyophilised (FD80; Cuddon Freeze Dry, Blenheim, New Zealand).The dried fractions were weighed to ensure >95% of the original soil subsample was recovered.
The above procedure was repeated until sufficient coarse and fine fraction material from each soil had been collected.The accumulated subsamples of coarse or fine material were combined and stored until further use.To serve as a control for the fractionation process, additional subsamples of each soil were treated with the same procedure except for sieving (dispersed by shaking with water, frozen and lyophilised), and are referred to as 'fractionated whole soil' (frac.whole soil).The C and N content of all whole soil and fraction materials are given in Table 2; analysis methods are described below.

| N mobilisation incubation
N mobilisation from the fraction materials (coarse, fine, whole soil and fractionated whole soil) were compared by mixing with a sand matrix and incubating for 14 days.The sand matrix was prepared by mixing 2.7 kg of sand with 75 g each of CC, CV, CW and NV whole soils, and wetting to 60% water holding capacity (WHC).Combining the 4 soils intended to inoculate the sand and standardise any potential differences in the microbial communities between the sampling sites (Baldock et al., 2021).The sand matrix was pre-incubated for 10 weeks at 25 C with the water content adjusted by adding deionised water and thoroughly mixing every few days.
Coarse fraction, fine fraction, whole soil or fractionated whole soil equivalent to 85-99 mg C, and 20 g (dry equivalent) of the pre-incubated sand matrix were placed into tubes made of PVC pipe (21.5 mm internal diameter) with nylon mesh covering the base.For the whole soil and fractionated whole soil treatments, the original mass of soil to add to each microcosm was calculated using soil total C content that had not been corrected for air dry moisture.When accounting for this, the C addition rates varied between 88 and 99 mg C per tube.This calculation error was corrected for by scaling all results to the actual T A B L E 2 Carbon (C) and nitrogen (N) characteristics of whole soil, coarse fraction (>50 μm), fine fraction (<50 μm), and fractionated whole soil collected from long-term cotton crop rotation plots or native vegetation at Narrabri, Australia.C rate.Water was added to reach 60% WHC, then the soils were mixed until homogeneous with a spatula and the bulk density adjusted to 1000 kg m À3 by lightly tamping to the set volume with a manual press.Five replicates of each treatment (soil Â fraction) and a control containing only the pre-incubated sand matrix were prepared in the same way.The tubes were placed into individual 237 mL glass jars with a septum fitted to the lid along with a 10 mL vial of water to maintain humidity.The jars were arranged into five randomised blocks and incubated in the dark at 25 C.After 14 days of incubation, the soils were destructively harvested.Subsamples (5 g fresh weight) were extracted with 25 mL of 0.5 M K 2 SO 4 and analysed for N pools and C as described in the following section.Another subsample was dried at 105 C to determine the gravimetric water content, and the remaining incubation soil air-dried at 40 C.

Cropping
During the incubation period, headspace CO 2 concentrations were measured on days 1, 2, 6, 7, 8, 9, 12, 13 and 14 using a Servomex 1450 infra-red gas analyser (Servomex, UK).For each measurement interval, the jar's headspace was refreshed, the lid sealed, and an initial measurement taken.At the end of the interval, a final measurement was recorded, then the jars opened, and the headspace was refreshed before immediately resealing the jar and commencing the next sampling interval.The CO 2 accumulated during each interval was taken as the final minus initial measurement.As there were no gaps between sampling intervals, the cumulative CO 2 was determined by summing the CO 2 accumulated in successive sampling intervals.

| Soil and extract sample analyses
Soil total C and N in whole soils, coarse fractions, fine fractions and fractionated whole soils were measured via combustion using a Leco Trumac combustion analyser (LECO Corporation, USA).All samples returned a negative 1 M HCl fizz test for carbonates, thus total C was interpreted as total organic C. Subsamples of each fraction material were also extracted with 0.5 M K 2 SO 4 for determination of starting extractable N and C concentrations.
Soil extracts were performed by shaking 5 g soil and 25 mL of 0.5 M K 2 SO 4 for 1 h at 180 rpm on an orbital shaker.The extracts were then centrifuged and filtered through pre-rinsed ashless filter papers, and frozen at À20 C until analysis.Extract samples were analysed for N pools using colorimetric assays: nitrate via the Griess reaction using a method adapted from Miranda et al. (2001), ammonium via the Berthelot reaction (Rhine et al., 1998), and free amino acid-N (FAAN) using the method of Jones et al. (2002).Extract total C and N were measured using a Thermalox total organic C-total N analyser (Analytical Sciences Limited, Cambridge).DON was calculated as extract total N minus the sum of nitrate-and ammonium-N.

| 13 C NMR of fraction materials and whole soils
Solid-state 13 C NMR spectroscopy was used to characterise organic matter chemistry in coarse and fine fractions, whole soil and the fractionated whole soil, and to determine the amount of ROC in each fraction using the methods of Baldock et al. (2013).Additional 10 g subsamples of each soil were dispersed by shaking with 5 g L À1 sodium hexametaphosphate and then fractionated using the same wet sieving method described above.Organic matter in each fraction was concentrated for 13 C NMR analysis as described in Baldock et al. (2013): coarse organic matter was separated from sand minerals, and fine organic matter was demineralised with 2% hydrofluoric acid (HF).Whole soil and fractionated whole soil samples were similarly treated with HF.
13 C NMR spectra were acquired using a 200 Advance spectrometer (Bruker, USA) equipped with a 4.7 T widebore superconducting magnet operating at a resonance frequency of 50.33 MHz.Samples were packed into 7 mm da zirconia rotors with Kel-F end caps and spun at 5 kHz.Chemical shift values were calibrated to the methyl resonance of hexamethylbenzene at 17.36 ppm, and a Lorentzian line broadening of 50 Hz was applied to all spectra.Cross-polarisation experiments and spectra processing were performed as described in Baldock et al. (2013).The total signal was integrated in the following chemical shift regions, with correction for signal from spinning sidebands: 0-45 (alkyl), 45-60 (N-alkyl/methoxyl), 60-95 (O-alkyl), 95-110 (di-O-alkyl), 110-145 (aryl), 145-165 (O-aryl), 165-190 (amide/carboxyl), and 190-215 ppm (ketone).Resistant organic C (ROC) in the coarse and fine fractions was estimated using the proportion of 13 C NMR signal in the aryl and O-aryl regions.This was subtracted from the total organic C in the coarse and fine fractions to estimate the particulate organic C (POC) and mineral-associated organic C (MAOC), respectively.

| Data analysis
Data analysis was performed in RStudio using base R (version 4.2.2), the tidyverse packages, and additional packages where indicated (R Core Team, 2022; Wickham et al., 2019).All extract N and CO 2 -C measurements are expressed per g of total organic C in each fraction to standardise evaluation across the variable C content between the soils.Principal components analysis (PCA) was completed for the NMR spectra regions as a proportion of the total signal for all fractions and soils.Data were standardised before PCA.
The change in each extract N pool (NO 3 À , NH 4 + , FAAN, and DON) was calculated by subtracting the amount initially present in the fraction materials from the amount measured at the end of the 14-day incubation.Extracted N attributed to the sand matrix was determined using the mean concentration of each pool measured in the sand matrix-only controls and subtracted from each sample.The remaining concentrations were considered the end concentration for each fraction material.Statistical analysis was performed on the change in extract N pools over the 14 days.
The change in extract N concentration (mg-N g C À1 ) data were analysed using either two-way ANOVA or generalised least squares regression when the assumptions of ANOVA were not met and could not be achieved using data transformations.The chosen models included 'soil' and 'fraction' as main effects with an interaction term.Plots of homogeneity of variances and normality of residuals were used to evaluate model assumptions.The twoway ANOVA was deemed suitable for NO 3 À and FAAN without transformation.Data for NH 4 + and DON showed significant heteroscedasticity and were refit with generalised least squares regression using the gls() function from the 'nlme' R package with weights assigned to the 'soil' variable using the varIdent() function (Pinheiro et al., 2020).Comparing plots of standardised residuals versus fitted values indicated improvement in model variance.
Comparisons between fractions (within soils) and soils (within fractions) were performed using the emmeans package (Lenth et al., 2020).Analysis for the CO 2 emission data additionally used the 'aomisc' (Onofri, 2020) and 'broom' (Robinson et al., 2022) R packages.A single-pool first-order reaction model (Equation 1) was fit to the cumulative CO 2 emission data using least-squares convergence and starting values of model coefficients estimated using the NLS.negExp() function from the aomisc package (Onofri, 2020). where ), k = rate constant (day À1 ), and t = time (days).Each replicate was fit individually.Estimates of C f and k were analysed for effects of 'soil' and 'fraction' using two-way ANOVA.Verification of adherence to model assumptions and contrasts between treatments were performed as above.

| Soil and fraction C and N characteristics
The OC and N content of the CC, CV and CW crop rotation whole soils were similar, averaging 12.7 ± 0.8 mg C g soil À1 (mean ± SE, n = 3), while in contrast, the NV soil had five times the OC and 4.5 times N of the cropping soils (Table 2).The fractionation control (fractionated whole soil), in which soil was dispersed and freezedried, did not significantly affect the OC and N content compared to the untreated whole soil (p = 0.9, n = 4, two-tailed t-test).In the separated fractions, the cropping soils were similar, with the coarse having a lower OC content than the fine fraction.However, the difference between the fractions was only slight, being at most 2.3 mg C g soil À1 (Table 2).This was reversed in the NV soil where the coarse fraction had a substantially greater C and N content than the fine.The C/N ratio was similar in the whole soil, fractionated whole soil and fine fractions across all four soils, averaging 11.5 ± 0.1 and ranging from 10.6 in the NV fine fraction, to 12.6 in the NV whole soil (Table 2).The coarse fractions overall had higher C/N ratios, ranging from 13.9 in the NV to 17.6 in the CW soil.
Solid-state 13 C NMR spectroscopy showed distinct chemistry of the organic C found in the coarse and fine fractions (Table 2 and Figure 1).PCA of the 13 C NMR spectra regions revealed separation of coarse and fine fractions along PC1 and separation between the cropping soil and NV along PC2 (Figure 1).Little difference was seen between the whole soil and fractionated whole soil.These treatments grouped closest to the fine fraction in the cropping soils, and intermediate to the coarse and fine in the NV.Principal component 1 accounted for 94% of the variance and had the greatest contribution from O-alkyl, aryl, and amide/carboxyl regions.Principal component 2 accounted for an additional 5.3% of variance, and most strongly related to alkyl, O-alkyl, and aryl regions.The alkyl/O-alkyl C ratio was lower in the coarse and higher in the fine fraction, indicating relatively less O-alkyl C and a greater degree of decomposition in the fine (Table 2).The CC, CV and CW soils were similar, while the alkyl/O-alkyl C ratio was higher in the NV coarse fraction, indicating relatively greater O-alkyl C than the cropping soils.The aromatic/aliphatic C ratio had similar patterns to the alkyl/O-alkyl C, being lower in the coarse than fine fraction, and similar between the cropping soils.In the NV soil, the fine fraction was lower than the other soils, and similar to the coarse fraction.Both the alkyl/O-alkyl-and aromatic/aliphatic C ratios in the whole soils were similar to the fine fraction, likely due to the dominance of this fraction (Figure S1).No variation was detected between the whole soil and fractionated whole soil.

| Change in total extracted N in fractions and whole soil
The change in total extracted N, represented by the slope of the lines in Figure 2, showed differences in direction and magnitude between the four fraction treatments, and varying patterns between the cotton rotation and NV soils.Two-way ANOVA returned significant main effects of Soil ( p < 0.001) and Fraction ( p < 0.001), and a significant interaction (p < 0.001).In the CC, CV and CW soils, the whole soil and fractionated whole soil had a net increase in extracted N over the incubation, while the coarse and fine fractions showed a net decrease in extract N (Figure 2).In the NV soil, all four fractions had a net increase in extracted N.
The total extractable N measured at Day 14 in the whole soil had increased by 62 ± 2% (NV) to 71 ± 7% (CC) from the starting concentrations (Figure 2).This was significantly different from the decrease in coarse fraction in all soils (p < 0.001), the decrease in the  fine fractions in the CC, CV and CW soils ( p < 0.001).
There was no significant difference between the whole soil and fractionated whole soil treatments in the CC, CW and NV soils ( p > 0.05).In the CV soil, the fractionated whole soil had 44% greater increase in total extracted N than the untreated whole soil (p = 0.02).
The fine fraction generally had the highest amount of extract N per gram of C at Day 0 across all four soils and showed the largest decrease of the four fractions in the CC, CV and CW soils ( p < 0.001), being 58 ± 4, 33 ± 10 and 71 ± 3% lower at Day 14, respectively (Figure 2).In the NV soil, no significant differences were found for the change in extract N between the fine fraction, whole soil and fractionated whole soil ( p > 0.05), despite the varying starting amounts.
Unlike the fine fraction, the coarse fraction started with the lowest amount of extractable N and had the smallest change over the 14 days (Figure 2).In all soils, the change in extractable N in the coarse fraction was significantly different from all other fractions ( p < 0.001).The largest decrease was found in the CW soil (76 ± 7%), followed by CC (51 ± 9%).The CV soil had close to no change at Day 14 (À0.08 ± 0.06 mg N g C À1 ).In the NV soil, the coarse fraction had the smallest increase in extract N, being 57% lower than the other fractions.
The greatest effect of soil management on the total extract N measured in each fraction was found between the cotton rotation and NV soils, with smaller differences between the cropping soils.There were no statistically significant differences between the CC, CV and CW soils for the increase in extract N measured in the untreated whole soil ( p > 0.05; Figure 2).Contrarily, the fractionated whole soil was similar between the CC and CW soils (p > 0.05), and significantly greater in the CV and NV soils (p < 0.001).The coarse and fine fractions showed greater variation between the soils.The decrease in extracted N in the fine fraction was significantly different between the CC, CV and CW soils (p < 0.05).In the coarse fraction, the CV was significantly different from the CW soil ( p = 0.001), while the CC was not significantly different to either (p > 0.05).In the NV soil, the four fractions had significantly greater change in extract N than the same fraction in the other soils (p < 0.01), except for CV fractionated whole soil (p = 0.44).

| Distribution of N pools in fractions and whole soil
Extractable N measured in the fraction materials before starting the incubation was distributed between DON, FAAN, NO 3 À , and NH 4 + (Figure 3).The fine fraction and whole soil were similar in the initial distribution of N, with NO 3 À and NH 4 + accounting for 90 to 97% of the total extract N in the CC, CV and CW soils, and 49 (fine) to 61% (whole soil) in the NV (Figure S2).The cropping soils had very low amounts of FAAN and DON, whereas up to half the extract N in the NV soil was in organic forms.The fractionated whole soil had higher initial concentrations of FAAN and DON than the untreated whole soil and was similar to the coarse fraction (Figure 3).The whole soil had the lowest starting concentration of FAAN, while the coarse and fine fractions, and fractionated whole soil were similar.Extract N in the coarse fraction was predominantly DON and FAAN, with a smaller proportion of NH 4 + ; no NO 3 À was initially present.Statistical analysis of the changes in N pools over the 14-day incubation showed a significant interaction between the soil and fraction variables for DON, FAAN, NH 4 + and NO 3 À (p < 0.05).After 14 days, DON, FAAN and NH 4 + were depleted in most fractions across all soils (Figure 3).Nitrate was the dominant N form in all cases except for the coarse fraction in the CC and CW soils, where DON was the largest pool.The increase in NO 3 À concentration observed in the whole soil and fractionated whole soil exceeded the loss of N from other forms and accounted for the net increase in extract N shown in Figure 2. The fine fraction typically showed a loss of extractable N forms, except in the NV soil where it had the greatest increase in NO 3 À of the fractions ( p < 0.001).
The coarse fraction increased in NO 3 À in the CV and NV soils but was significantly lower than other fractions (p < 0.001).The fractionated whole soil had similar starting amounts of DON as the coarse fraction in the CC, CV and CW soils, but showed a significantly greater decrease over the incubation ( p < 0.001).

| C stability
The cumulative C mineralisation, measured as emitted CO 2 -C, differed strongly between the fractions with similar patterns in all four soils.The fine fraction had the highest C mineralisation, followed by the coarse fraction and fractionated whole soil in the cropping soils, and the untreated whole soil lowest (Figure 4).Two-way ANOVA of the labile C pools (C f ; estimated by fitting first-order reaction models) showed a significant main effect of fraction ( p < 0.001) and soil by fraction interaction (p = 0.004).In all soils, the fine fraction had the highest C f (45.0 ± 5.6 in CV to 55.1 ± 0.1 mg C g À1 fraction C in CW) and was 1.9-3.2times higher than the whole soil (p < 0.001; Table 3).The labile C content of the coarse fraction was highest in the CC and lowest in the NV soil.
The coarse fraction C f was significantly lower than the fine fraction in the CW and NV soils (p < 0.001) and significantly greater than the whole soil in the CC, CV and CW soils (p < 0.001).The fractionated whole soil C f was 1.4 (NV) to 2.2 (CC) times greater than the whole soil and was significantly different in all soils (p < 0.002), except the NV (p = 0.08).
Estimates of CO 2 mineralisation rate constants (k) differed significantly between fractions ( p = 0.002) and soils (p < 0.001; Table 3).Values of k were largest in the fractionated whole soil and smallest in the coarse fraction.The whole soil was not significantly different from the coarse or fine fractions ( p > 0.05) but was significantly lower than the fractionated whole soil ( p < 0.001).Averaged across soils, the fine fraction had a significantly higher estimate of k than the coarse fraction ( p = 0.01).Between the soils, estimates of k in the CC and CW were significantly different to the NV (p < 0.01), while the CV was not ( p = 0.56).No differences were found between cropping soils (p > 0.05).

| DISCUSSION
This study set out to build an understanding of how N becomes available from coarse and fine SOM fractions and link this to the characteristics and stability of the corresponding organic C. Using soils from contrasting crop rotations and uncleared vegetation, short-term changes in extractable N (DON, FAAN, NH 4 + and NO 3 À ) and C mineralisation were compared between coarse and fine fractions and whole soil, and an additional control treatment which subjected whole soil to the same disruptive dispersal and freeze drying as the separated fractions (fractionated whole soil).Overall, surprising dissimilarity was found between the separated fractions and whole soil, and between the patterns of C mineralisation and N mobilisation.These findings raise questions about the utility of studying isolated SOM fractions for understanding N availability via mobilisation pathways and demonstrates the importance of soil structure in both C stability and N dynamics.

| N mobilisation from SOM fractions
At the beginning of this experiment, it was expected that the behaviour of N in whole soils would sit neatly intermediate to the coarse and fine fractions, with the fine having greater N mobilisation.This was observed in the NV soil, whereas in the cropping soils the individual fractions had net immobilisation of N, while available N increased in the whole soil (Figures 2 and 3).We further show that the decrease in extractable N occurred across DON, ammonium and nitrate, while the net mobilisation was via an increase in nitrate only, surpassing the decreases in other N pools.This is a particularly unique finding, and contrasts to other studies reporting higher N mineralisation from fine fractions (Bimüller et al., 2014;Parfitt & Salt, 2001).Immobilisation in the separate coarse and fine fractions of the cropping soils is indicative of N limitation relative to C availability.Overall lower N availability, as shown by the lower N content and higher C/N ratio (Table 2), likely contributed to immobilisation in the coarse fraction.While the lower alkyl/O-alky C and aromatic/aliphatic ratios suggest the C in the coarse fraction T A B L E 3 Model coefficients estimated by fitting first-order kinetic models to CO 2 mineralisation measured over 14 days from soil organic matter fractions and whole soil from cotton cropping rotations and native vegetation.

Soil
Fraction was more labile (Baldock & Preston, 1996), the larger organic matter particle sizes likely also restrict physical accessibility of both C and N to microbes or extracellular enzymes (Schmidt et al., 2011;Young et al., 2008).In the fine fraction, sorption of organic N molecules and ammonium to mineral surfaces could have further reduced N availability (Jilling et al., 2018).Given that the cropping soils were $86% fine fraction, this would likely also be a significant factor in the whole soil, and potentially have caused lower N mobilisation in the fractionated whole soil due to increased availability of surfaces with the disintegration of aggregates.Measurement of CO 2 mineralisation showed an increase in available C and lability with separation of the coarse and fine fractions (Figure 4 and Table 3), suggesting that relative N limitation in the coarse and fine fractions may have been induced due to C destabilisation with fractionation.
The similarity of N behaviour between the fractionated whole soil and whole soil demonstrates that the contrasting N mobilisation in the separate fractions is not due to disruption from the fractionation process.Comparing to cumulative CO 2 , which more than doubled in the fractionated whole soil compared to the untreated whole soil, this demonstrates a separation between C mineralisation and N mobilisation processes.It is possible that an interaction between the coarse and fine fractions promoted mineralisation and nitrification in the lower C soils.C/N ratio alone does not correlate with N mobilisation, as the whole soils were similar to the fine fractions (Table 2).It appears that when overall C and N levels are low, such as in the cropping soils, the characteristics of SOM may have more control of relative C or N limitation.

| Disparate controls on C stability and N mobilisation
Measurements of C mineralisation produced unexpected results.Carbon dioxide evolution was lowest in the whole soil and higher in the individual fractions, suggesting C was most stable in the whole soil and least in the fine fraction (Figure 4 and Table 3).This, combined with the 13 C NMR analysis of SOM chemical composition (Table 2), is at odds with two major theories for stabilisation of C in soils, yet consistent with a third.
First is that association with clay minerals stabilises C in soils, reducing susceptibility to loss by microbial mineralisation (Kleber et al., 2021;Schweizer et al., 2021).This was not observed over the short-term incubation experiment, where a much greater portion of the total organic C in the fine fraction was attributed to the labile ('fast') pool (C f ) than in the coarse fraction or whole soil.Other studies report less C mineralisation from fine soil fractions, and often attribute this to mineral stabilisation providing resistance to microbial consumption (Bimüller et al., 2014;Mueller et al., 2014;Parfitt & Salt, 2001).The dominant clay minerals in the Vertosol cropping soil studied has been reported as Montmorillonite (Rochester, 2011).With large surface area and surface charge, this suggests large potential for organic matter stabilisation by mineral association (Sanderman et al., 2014), but this was not found in the present study.It is possible that mineral stabilisation would emerge over a longer incubation duration; we did not attempt to fit second-order mineralisation models or estimate the size of the 'slow' C pool due to the short time frame (Saidy et al., 2012).
Second is stability of SOM due to chemical recalcitrance of organic molecule structures.In this study, characterisation of coarse and fine SOM found the coarse fraction had a lower alkyl/O-alkyl C ratio, indicating relatively more carbohydrate and less lipid, while the fine had a higher aromatic/aliphatic ratio, likely due to depletion of more labile forms of C leaving aromatic structures (Baldock et al., 1997;Baldock & Preston, 1996).These differences were less pronounced in the NV soil, which receives greater fresh C inputs and less disturbance than the cropping soils.These factors indicate increased chemical recalcitrance of the fine fraction SOM (Baldock et al., 2004).However, C mineralisation data showed higher availability of C in the fine fraction and relatively little variation between soils (Figure 4 and Table 3).This suggests that the SOM chemistry present in these soils did not significantly influence C mineralisation over the short-term incubation.
The results from this study do however agree with a third major mechanism of C stabilisation in soil: physical protection in aggregate structures (Dungait et al., 2012;Six & Paustian, 2014).This is evident by the doubling of C mineralisation measured in the fractionated whole soil than the whole soil (Figure 4), and the increased mineralisation rate constant (k; Table 3), indicating that the destruction of soil aggregates during the dispersal and freezedrying steps of fractionation significantly destabilised C. Importantly, N mobilisation was not similarly affected, highlighting that the mechanisms providing protection of C do not necessarily apply to N release from SOM.Other studies found both no effect of fractionation (Bimüller et al., 2014), or similarly an overestimation of C mineralisation from separated fractions (Mueller et al., 2014).Our experiment provides a key distinction in using soils dominated by fine fraction C, demonstrating that this fraction can significantly contribute to C loss from soil and that physical protection far outweighed mineral association or chemical recalcitrance for short-term stability.
Comparing CO 2 emission (Figure 4) and N mobilisation (Figure 2) data, it is evident that C mineralisation did not directly correlate with a release of available N from SOM.On the contrary, the reverse was observed for the cropping soils in the fine fraction and the whole soil, where CO 2 emissions and N mobilisation were inverse.Parfitt and Salt (2001) similarly found lower C and higher N mineralisation in whole soil compared to size fractions but did not address why.The combined lower C mineralisation with higher N mobilisation in the whole soil is consistent with soil structure providing physical protection of SOM and inducing a relative C limitation in whole soil, and this in turn leads to the release of N excess to microbial requirement.However, this C limitation would have been alleviated in the fractionated whole soil, so cannot explain the N mobilisation.Alternatively, labile C released during fractionation could result in priming leading to higher CO 2 emission without concurrent N mobilisation (Jilling et al., 2021;Wild et al., 2019).The higher CO 2 emitted from the fine fraction could stem from an overall greater microbial biomass or lower microbial C use efficiency (CUE), hence N limitation (Mooshammer et al., 2014).The apparent decoupling of C and N shown in this study highlights that the dynamics of N cycling between SOM and available pools cannot necessarily be predicted by the movements or loss of C from SOM fractions.Discrete consideration of the mechanisms controlling N availability and microbial N demand are essential to understanding the functions of SOM fractions in providing N.

| Effects of fractionation on C and N dynamics
The fractionation process undoubtedly influenced the C and N cycling and limits the translation of results for separate fractions to whole soils.Our use of a fractionation control shows that this arises partly from disruption during fractionation, and partly from the reductionist approach of studying fractions in isolation.The fractionation method used here is comparatively mild to other studies, involving no sonication or harsh salts (sodium polytungstate or hexametaphosphate; Poeplau et al., 2018).Using water only may have ineffectively dispersed microaggregates, meaning some fine or mineralassociated organic matter could have been retained in the coarse fraction or physically protected (Table S1).Yet, we still see a clear overestimation of C mineralisation in both coarse and fine fractions.This may be partially attributed to labile C released from lysed microbial biomass during fractionation; however, the cumulative C mineralisation far exceeded the loss of extracted organic C (Figure S3).Priming of SOM mineralisation has been observed with glucose-C inputs similar to the initial concentration of DOC, hence this is a likely contributor to higher C mineralisation in separated fractions (Jilling et al., 2021).Other previous studies have arithmetically recombined C mineralisation from soil fractions, finding either no difference to whole soil, or an overestimation (Bimüller et al., 2014;Mueller et al., 2014).Overall N mobilisation was less directly affected by fractionation and instead highlights a pitfall of studying isolated fractions.There was a clear increase in extract DON and FAAN with fractionation (Figure 3 and Figure S2), likely also from disrupted microbes, but this did not appear to contribute to N mobilisation.Fractionation approaches can determine some mechanisms controlling C and N release from SOM, however alternative methods which avoid disruption of soil structure, such as isotope tracing, are needed to understand the behaviour of fraction C and N in intact soils.

| Implications for cropping soil management
The incubation experiment provided limited insights for the roles of SOM fractions in providing available N in cropping soil.Few differences were found between the crop rotation soils, implying minimal lasting effects of differing cropping management on N supply via SOM.Yield data indicated the legume rotations decreased fertiliser N requirements for cotton (Rochester & Bange, 2016), thus the incubation experiment may not have captured influential variables such as the freshness of particulate OM inputs, soil depth (Parajuli et al., 2021), or role of plant roots in stimulating N mobilisation (Jilling et al., 2018).The contrasting patterns in the NV soil, and in literature, suggests that when SOM levels are higher, the fine fraction is an important contributor to soil N availability.Disturbance of soil structure had a far greater impact on C stability, without a concomitant increase in available N. This suggests that disturbances such as tillage could disproportionately lose C from soils, without providing the benefit of enhanced N availability (Grandy & Robertson, 2007).

| CONCLUSIONS
Here, we presented an integrated characterisation of short-term C and N dynamics in coarse and fine SOM fractions and comparison to whole soil.In a low C cropping soil, we found disparate patterns of C and N mineralisation in separated fractions as opposed to whole soil, emphasising a shortcoming of a reductionist approach to studying isolated SOM fractions.Overall N mobilisation did not correlate with C loss, which showed greater variation between fractions than between crop rotation soils.Stabilisation of C was largely provided by physical protection within soil aggregates.Contrary to common thinking, fine fraction SOM was stabilised by neither mineral association nor chemical recalcitrance and can potentially be a large source of C loss from soil.Fully understanding the contribution of SOM fractions to N supply through mineralisation requires alternative techniques that avoid disruption of the soil system.

Highlights•
Carbon (C) and nitrogen (N) dynamics in soil organic matter (SOM) fractions were investigated in cropping soils.•Mineralisation is a key N source for crops, but the contribution of SOM fractions is unclear.• Whole soil showed the most stable C but greatest N mobilisation compared to fractions.• C and N stability in SOM fractions and whole soils soilF I G U R E 2 (a) Total extracted N (sum of dissolved organic nitrogen (DON), free amino acid-, ammonium-(NH 4 + ) and nitrate-(NO 3 À ) N) per gram of carbon measured in coarse (≥50 μm; red) and fine (≤50 μm; orange) soil fractions, whole soil (light blue) and whole soil that was dispersed and freeze-dried (Frac.Whole soil; dark blue) over a 14-day incubation.Soils were collected from continuous cotton (CC), cotton-vetch (CV) and cotton-wheat (CW) crop rotations, and a nearby native vegetation (NV) site (horizontal panels).(b) The control panel shows background N subtracted from treatments.Values at Day 14 are mean, error bars show standard error (n = 5).Letters show significant differences (α = 0.05) between Fractions within each panel (Soil) for the change in N concentration.
U R E 3 (a) Change in concentration per gram of carbon of dissolved organic nitrogen (DON), free amino acid-(FAAN), ammonium-(NH 4 + ) and nitrate-(NO 3 À ) nitrogen (N; vertical panels) in coarse (≥50 μm; red) and fine (≤50 μm; orange) soil fractions, whole soil (light blue) and whole soil that was dispersed and freeze-dried (Frac.Whole soil; dark blue).Soils were collected from continuous cotton (CC), cotton-vetch (CV) and cotton-wheat (CW) crop rotations, and a nearby native vegetation (NV) site (horizontal panels).(b) The control panel shows N concentrations subtracted from treatments.Values at Day 14 are mean, error bars show standard error (n = 5).U R E 4 Cumulative C mineralisation measured as CO 2 emitted from coarse (≥50 μm; red) and fine (≤50 μm; orange) soil fractions, whole soil (light blue) and whole soil that was dispersed and freeze-dried (Frac.Whole soil; dark blue).Soils were collected from continuous cotton (CC), cotton-vetch (CV) and cotton-wheat (CW) crop rotations, and a nearby native vegetation (NV) site (horizontal panels).Curves show first-order kinetic models fitted to five replicates per treatment.Dots show mean of measured cumulative CO 2 -C, error bars show standard error (n = 5).
T A B L E 1 Soil properties for samples collected from 0N rate plots of a cotton cropping systems experiment, and a nearby native vegetation site at ACRI, Narrabri, Australia.Cropping system pH 1:5 H2O Colwell P (mg kg À1 ) Extractable S (mg kg À1 ) Effective CEC (cmol c kg À1 ) Particle size distribution (%) Note: Uppercase letters show significant differences (α = 0.05) between Soils; lowercase letters show differences between fractions within each soil.Values are mean (SE), n = 5.Abbreviations: C f , labile C pool size per gram fraction total organic C; k, rate constant (d À1 ), n.s., not significant.