The soil used in laboratory incubation experiments was a sandy loam with 17% clay, 61% sand, 3% organic matter, and pH (1:5 H₂O) 6.7. It was collected from a field at SLU's experimental farm Götala close to Skara in western Sweden (58°22′N, 13°29′E). The crop rotation at the farm includes both 3–4 years grass/clover forage leys and annual crops, with frequent addition of slurry and farmyard manure from beef cattle. Top soil was collected from a plowed field after a cereal crop in late autumn, ~2 months before experiment start and stored in open containers at cool temperature (0–5°C) until use. Two separate incubation experiments, one on N turnover and one on C turnover, were performed. These started on different dates (February 8 and 22, 2022, respectively), so soil preparation was performed using two separate batches, one for each experiment. The procedure used was the same for both batches, and started 10 and 13 days before the start of the N and C turnover incubation, respectively. To reduce the water content, soil was air-dried at 16–18°C for 2 days, with gentle mixing twice a day. The soil was sieved through a coarse (7 mm) mesh sieve and water-holding capacity (WHC) was determined using the method described in Jansson (1958). Water content after drying was 56 and 51% of WHC, respectively, for the two batches. Both soil batches were pre-incubated at 15°C in a climate chamber (Termaks, KB 6395, Nordic Labtech AB, Fjärås, Sweden), for 8 and 11 days, respectively.

Liquid (LF) and solid (SF) fractions from screw-press separated cattle slurry were collected from an organic dairy cattle farm (Otterslättens Lantbruk, Hova, Västra Götaland County, Sweden) 2 weeks before the start of the first experiment. Both fractions were collected directly from the slurry separator (CRI-MAN SM 260/75 FA DM, CRI-MAN, Correggio, Italy). The fractions were stored at cool temperature (0–5°C) until the start of the experiment.

Winter wheat straw was collected from a conventional farm. To obtain chopped straw, the straw was cut with scissors into ~0.5–1.0 cm pieces. To obtain ground straw, pieces of straw were milled in a grain mill to maximum particle size 0.8 mm. Sawdust (spruce and pine in unknown proportions) was collected from a commercial sawmill and sieved through a 2 mm mesh sieve.

Samples of LF from slurry separation and the bedding materials used in the experimental slurries were analyzed for dry matter content (DM), total C, total N, and ammonium-N at the Soil and Plant Laboratory, Swedish University of Agricultural Sciences (SLU), Uppsala, Sweden. Total C and organic N were analyzed by combustion of dried samples followed by elemental analysis [928 Series Macro Determinator (model CN928), LECO, St. Joseph, MI, USA]. Ammonium-N was extracted with 2M potassium chloride (KCl) and analyzed using a flow injection analyzer [FIAstar Analyzer (5000), FOSS, Hillerød, Denmark]. Total N was calculated as the sum of organic N and ammonium-N.

After the incubations, an additional analysis of fiber fractions [neutral detergent fiber (NDF), acid detergent fiber (ADF), and acid detergent lignin (ADL)] was performed on stored samples of the original bedding materials, at the laboratory of the Department of Applied Animal Science and Welfare, SLU, Uppsala, using the method described by Van Soest et al. (1991). Acid detergent lignin was analyzed by the Klason lignin method, using sulfuric acid (72%). Hemicellulose content was calculated as the difference between NDF and ADF, and cellulose as the difference between ADF and ADL. As shown in Table 1, the concentrations of fiber fractions (hemicellulose, cellulose, and lignin) were similar in solid fraction and straw, while sawdust had a higher content of both cellulose and lignin, and higher total NDF content.

Soil samples from the N incubation were analyzed for ammonium-N and nitrate-N at the Soil and Plant Laboratory, SLU, Uppsala. Milled, frozen soil samples were extracted with 2M KCl and then analyzed by colorimetry on a Seal AA3 auto-analyzer (SEAL Analytical Inc., Mequon, Wisconsin, USA).

In all treatments, the bedding materials were added to the LF from slurry separation, which was assumed to represent “pure” cattle manure without any bedding material added. This allowed differences in slurry properties related to the degree of slurry separation to be analyzed. Using LF (rather than unseparated slurry with higher DM content) as the basis for the treatments also resulted in greater variation in C/N ratio in the mixtures, without reaching unrealistically high C/N ratios in relation to that on real farms. The mixtures of LF and bedding materials are referred to hereafter as “slurry” or “slurries.”

The amount of slurry added to the soil was based on supplying 50 kg total N ha⁻¹ in all treatments, i.e., 26 tons ha⁻¹ with LF, in this experimental setting with a soil depth of 50 mm corresponding to 102 mg kg⁻¹ dry weight (DW) of soil. To obtain the different desired C/N ratios, the relative proportions of LF and bedding material were varied between the treatments. Straw and sawdust contributed very little N to the slurries, and in these treatments the proportion of bedding material was low and the amount of LF was more or less constant between C/N ratios. For SF, with higher ammonium N concentration, the amount of LF was reduced by at most 23%. Ammonium N content relative to total N in the slurries varied slightly between treatments and was on average 35% (Table 2). Due to the differences in N concentration, the amount of bedding material that needed to be added varied widely, between 12 and 127 g kg⁻¹ LF, and there was also variation in the amount of total C applied to the soil (Table 2).

Small plastic flowerpots (height 50 mm, volume ~98 cm³) were used as soil containers. On the day before experiment start, a filter paper was placed at the bottom of each pot to prevent soil loss through the drainage holes, and 70 g of pre-incubated soil were added and lightly packed down. The pots were placed in plastic boxes (36 x 24 x 13 cm), each containing pots for three replicates of one treatment and all sampling times, and placed in the same climate chamber where the soil was pre-incubated. The temperature was uniform in all parts of the chamber and was kept at 15 ± 0.3°C throughout the experiment. At the start of the experiment, one box at a time was removed from the climate chamber. The bedding material was added first to each pot and evenly distributed over the soil surface, and then LF was added. The separate weighing and addition of bedding material and LF was done to ensure that accurate C/N ratios were obtained. For ground straw and sawdust, the manure layer was stirred gently, taking care not to disturb the soil beneath, in order to create a slurry from the hydrophobic bedding material and the LF. Different amounts of water was added to obtain the same soil water content (70% of WHC) in all treatments and then another 45 g of pre-incubated soil were added to each pot, on top of the slurry, and gently packed down. The pots in their plastic box were then returned to the climate chamber. The reason for adding the experimental slurry as a layer in the middle, rather than thoroughly mixing it with the entire soil volume, was to mimic a real-life situation with animal slurry incorporated by plowing or injection, resulting in a more uneven distribution of slurry within the soil.

The lids of the plastic boxes were kept closed to prevent the soil from drying out. To ensure enough oxygen supply and avoid anaerobic conditions, the boxes were aerated at intervals of 3–4 days by opening the lid for a few minutes. On days 3, 7, 14, and 28, the three replicates of each treatment were destructively sampled for mineral N analysis. The entire soil volume from each replicate was placed in a plastic bag and frozen within 30 min for later analysis. To ensure that the water content remained constant during incubation, the boxes were weighed on each sampling occasion and, when necessary, deionized water was sprayed over the pots to replace the weight loss.

The procedure for preparing the soil and adding LF and bedding materials in incubation for analysis of CO₂ production was similar to that used in incubation for N analysis. However, the water content was kept slightly lower, at 65% of WHC, since the 70% used in the N incubation proved to be slightly too high, causing mold in some pots. Three replicates of each treatment were prepared. Each replicate was placed in a glass jar (outer diameter 120 mm, height 185 mm, volume 1,660 cm³) with metal screw-lid, together with a CO₂ trap. The CO₂ trap consisted of a 50 mL plastic scintillation vial with 30 mL of 0.3 M KOH for the first measurements and 0.2 M KOH from day 10, absorbing the CO₂ produced in the jar. To ensure that all CO₂ produced was captured, the jar lids were kept closed, except when the traps were changed and CO₂ production measured. The KOH concentration was changed in order to permit the desired sampling frequency throughout the experiment, since the activity was expected to be lower during the later stage of incubation and a sufficiently high saturation level in the traps was needed to allow precise measurements. On days 2, 5, 10, 15, 21, and 28, the CO₂ traps were changed and electrical conductivity (EC) in the KOH solution was measured using a conductivity meter (ProfiLine Cond 3310 Portable Conductivity Meter, WTW, Weilheim, Germany). At each sampling occasion, the jars were aerated for a few minutes, thereafter a new CO₂ trap was placed in the jar, the lid was immediately closed and the jar was returned to the climate chamber. To correct for background CO₂ in the air and for basic soil respiration, the set-up included one blank treatment with only a CO₂ trap and one control treatment without any slurry added to the soil.

For calibration of the conductivity meter, solutions with different ratios of KOH and potassium carbonate (K₂CO₃) were prepared in order to imitate absorption of different amounts of CO₂, following the procedure described by Smirnova et al. (2014). Constant A, describing the theoretical maximum amount of CO₂ absorbed, was calculated for the 0.3 M and 0.2 M KOH solutions as:

where C(t₀) is conductance before any absorption of CO₂, C(t₁) is conductance at time t₁ and A is empirically determined.

Temperature calibrations for the 0.3 M and 0.2 M KOH solutions at temperatures between 15 and 30°C revealed a negative linear correlation between conductivity and temperature. The resulting equations, obtained by fitting a straight line to the data points in Excel, were used for temperature correction in the conductivity calculations.

The high initial CO₂ production rate in all treatments was most likely an effect of breakdown of short-chain volatile fatty acids (VFA) from LF, as they are an easily available C source that is quickly consumed during the 1st days of incubation (Kirchmann and Lundvall, 1993). The higher initial CO₂ production from slurries with lower C/N ratios (Supplementary Table 5, Figure 3, day 2) can be explained by the smaller amount of “extra” C added with bedding material in the slurry mixes at low C/N ratios (Table 2). Hence, those slurries contained a higher proportion of easily degradable C components, including VFAs.

Slurry with solid fraction had the highest initial CO₂ production of the different bedding materials at C/N ratio 12 (Supplementary Table 6, Figure 5A), probably as a result of an already established micro fauna, due to its previous contact with animal feces. Chopped straw had the lowest initial CO₂ production, probably owing to its larger particle size and thereby smaller specific surface area compared with sawdust and ground straw. Sawdust had higher initial CO₂ production rate than chopped straw despite lower chemical C availability (more lignin), likely as a result of the smaller particle size making the most easily degradable C fractions accessible from the start (Sørensen et al., 1996).

Rapid initial N immobilization, similar to that in this study (Figures 1, 5), has been observed in a previous study which found a strong correlation between N immobilization and initial VFA content of slurries (Kirchmann and Lundvall, 1993). All treatments in the present study contained similar amounts of LF (the main source of VFAs) and showed similar initial N immobilization. This indicates that mineralization/immobilization processes during the 1st days after soil application were mainly driven by the properties of LF, while the added bedding materials had little influence.

From day 5, the CO₂ production rate followed a slowly declining pattern that continued during the rest of the incubation. During this phase, the most easily available C sources had been depleted and the processes controlling C turnover seemed to be dominated by the properties of the bedding materials added. This was reflected e.g., in sawdust giving the lowest CO₂ production through having the highest lignin content, and in higher cumulative CO₂ production from ground straw than chopped straw. Higher CO₂ production from ground wheat straw compared with chopped straw during the 1st weeks after soil application was also observed by Angers and Recous (1997), and was hypothesized to be a result of higher N accessibility with finer particles. In studies with low C/N ratio amendments such as rye residue (Angers and Recous, 1997) and pea residue (Jensen, 1994), where N content in the residues was less limiting for decomposition, the opposite relationship was seen. This was attributed to more intimate association with soil mineral particles for the ground residues, leading to physical protection of C in microbial biomass and by-products. In the present study, CO₂ production rate (averaged over C/N ratio 10–14) was consistently higher from SF compared with chopped straw, despite similar fiber composition, which was likely a combined effect of differences in particle size and degradation by microorganisms, as discussed in the section above for initial CO₂ production. For nitrogen, the decline in net mineral N release at C/N ratios 8, 10, and 12 from day 14 to day 28 (Figure 1A) could potentially be explained by re-immobilization of mineralized N and/or gaseous N losses through denitrification (not measured).

The negative linear relationship between C/N ratio and net mineral N release (Figure 2) is in line with hypothesis 3. The slope of the relationship was anticipated to be steeper with chopped straw than SF, based on the assumption that the straw contained more un-degraded and easily available C, resulting in greater N immobilization. This was not the case, and analysis of the fiber fraction revealed very similar content of hemicellulose, cellulose, and lignin in SF and straw (Table 1). The tendency for a steeper slope of the relationship for SF at day 14 (Figure 2A) is likely because microorganisms were already integrated with that material, whereas the straw had to be colonized, slowing down the straw degradation process and hence the N demand and immobilization. In addition, chopped straw had larger particle size, making the C less easy to access, which likely also delayed the degradation process (Sørensen et al., 1996).

A negative relationship between C/N ratio and N availability has been reported previously, for different organic fertilizers by Delin et al. (2012) and for cattle slurry by Sørensen et al. (2003). Sørensen et al. (2003) found that N availability was positively correlated with ammonium N content of the slurry, but in the present study the ammonium N content was within a narrow range (Table 2) and the relationship could not be evaluated. Norris et al. (2019) found that prediction of N supply from slurries could be improved by considering the ratio of total C to water-extractable N (rather than total N), but that considering more labile C fractions rather than total C weakened the predictions. This suggests that the amount of bioavailable N in organic material has a stronger influence on N mineralization/immobilization processes than amount of bioavailable C.

The hypothesis that higher C availability due to smaller particle size would increase N immobilization was neither supported nor refuted. Ground straw showed higher cumulative CO₂ production than chopped straw (Figure 5) and a tendency for greater net N immobilization (Figure 4), but the difference in net mineral N release was not significant. However, on comparing ground and chopped barley residues added to soil, Ambus and Jensen (1997) found greater N immobilization with finer particle size after 2 weeks, and the difference remained throughout their 60-day incubation. Microbial growth was similar regardless of residue particle size in that study, leading to the conclusion that despite grinding, some C in the residues was still physically protected by lignin. In the present study, there were two different interacting sources of added C (LF and bedding material), which makes the results more difficult to interpret.

After 28 days, net mineral N release was similar for SF and chopped straw, despite differences in CO₂ production. In a study by Jensen et al. (2005), net N mineralization from a wide variety of mostly low-lignified plant materials was most closely correlated with concentrations of total plant N and neutral-detergent soluble N. In the same study, C mineralization was most closely correlated with holocellulose (hemicellulose + cellulose) content, which is in line with our observation that sawdust had both the lowest cumulative CO₂ production and the lowest holocellulose content. According to Chen et al. (2018), cellulose is rapidly degraded after soil incorporation of straw, while lignin is not broken down, but accumulates in the soil.

With regard to our three starting hypotheses, the results showed that: (1) C availability (as indicated by cumulative CO₂ production) depended on a combination of bedding material particle size, fiber composition, and probably previous degradation by microorganisms; (2) for bedding materials with a relatively low degree of lignification (i.e., straw and solid fraction) there was no relationship between C and N availability, while for sawdust, with higher lignin content, high net mineral N release was associated with low CO₂ production; and (3) for both SF and chopped straw, there was a negative linear relationship between slurry C/N ratio and net mineral N release, with similar slope for both bedding materials. The contrasting results for sawdust, with substantially higher net mineral N release than from the other bedding materials, show that in addition to slurry C/N ratio, bedding material fiber composition is an important factor affecting slurry N availability.

The results indicate that bedding materials with similar fiber composition (i.e., SF and straw) give slurries with similar short term fertilizer value, when applied in amounts resulting in the same slurry C/N ratio. Applied at farm scale, this means that since SF has a lower C/N ratio than straw (Table 1), addition of similar amounts of SF as with straw would result in a lower slurry C/N ratio and hence potentially in higher N availability. The results also indicate that sawdust as bedding material could result in higher N availability, provided that the amount used gives the same slurry C/N ratio as with other bedding materials. This study confirms the relationship between C/N ratio and N availability, suggesting that slurry C/N ratio can be used as a predictor for N use efficiency in crop production (Delin et al., 2012). The results indicate that a higher N use efficiency can be expected from slurry containing sawdust, compared with SF and straw.

The environmental effects of slurry separation are more complex and difficult to predict. For the liquid fraction, the lower DM content increases the infiltration rate and hence reduces ammonia emissions after slurry application (Pedersen et al., 2022). When applied in spring or summer, the lower C/N ratio and thereby increased N availability can also improve the synchronization between N supply and crop N uptake. This potentially reduces the risk of mineral N surplus in autumn and subsequent losses during winter, when crop N uptake is low or no crop is present to take up the nitrogen (Delin and Stenberg, 2021; Sørensen et al., 2023). Simultaneous application of easily available carbon and nitrogen, as with all animal slurries, may stimulate denitrification and increase the risk of nitrous oxide emissions (Senbayram et al., 2012). For soil application of the liquid fraction from separated slurries and digestates, studies have shown both similar (Holly et al., 2017) and higher (Fangueiro et al., 2008b; Meng et al., 2022) nitrous oxide emissions, compared with raw slurries.

With covered storage of the liquid fraction, reducing the ammonia emissions, the overall environmental effect of slurry separation is largely dependent on how the solid fraction is treated during storage. With composting of the solid fraction, the total ammonia emissions are increased, while greenhouse gas emissions are reduced (Amon et al., 2006; Fangueiro et al., 2008a). Storage of the solid fraction without adding oxygen by turning, could either increase (Holly et al., 2017) or reduce (Hansen et al., 2006) the emissions of nitrous oxide. A system where the solid fraction is used as bedding material is probably most comparable with composting, with good oxygen supply and hence increased ammonia emissions. However, an evaluation of that system would need to include a comparison of the in-house ammonia emissions with other types of bedding materials.