Reduced nitrogen losses from drained temperate agricultural peatland after mineral soil coverage

Draining peatlands for agriculture induces peat decomposition, subsidence, and carbon (C) and nitrogen (N) losses, thereby contributing to soil degradation and climate change. To sustain the agricultural productivity of these organic soils, coverage with mineral soil material has increasingly been used. To evaluate the effect of this practice on the N flows within the plant–soil system, we conducted a 15N tracer experiment on a drained peatland that was managed as an intensive meadow. This peatland was divided into two parts, either without (reference “Ref”) or with ~ 40 cm mineral soil cover (coverage “Cov”). We applied 15NH415NO3 on field plots to follow the fate of 15N in plant–soil system over 11 months. In addition, N mineralization was determined by laboratory incubation. The field experiment showed that Cov lost less 15N (p < 0.05) than Ref, even though plant 15N uptake was similar at both sites. The lower net N loss from the Cov site was accompanied by higher soil 15N retention. The laboratory incubation revealed a ~ 3 times lower N mineralization at Cov than at Ref, whereas the N release per unit soil N was around two times higher at Cov than at Ref, suggesting a faster SOM turnover rate at Cov. Overall, the mineral soil cover increased the retention of fertilizer-N in the soil, thus reducing the system N losses. Our result indicates that agricultural production on drained peatland is less harmful to the environment with mineral soil coverage than using drained peatland directly.


Introduction
Although peatlands only cover approximately 3% of the terrestrial surface area, they are an essential soil organic matter (SOM) pool and can store 8-14 Gt N globally (Yu et al. 2010;Loisel et al. 2014;Leifeld and Menichetti 2018). However, long-term drainage for agricultural production has already resulted in ~ 51 Mha degraded peatlands worldwide, with the highest share occurring in tropical and temperate regions, where around half of the initial peatland surface has been disturbed due to agricultural production, forestry, or peat extraction (Kasimir et al. 2018;Leifeld and Menichetti 2018). Peatland degradation is typically associated with peat decomposition, which result in C and, to a smaller extent, in N losses, as well as strong soil subsidence. As a consequence, soil C to N ratios decrease (Klemedtsson et al. 2005;Leifeld 2018). The decomposition of peat is a substantial contributor to the N supply for agricultural production in drained peatlands. Therefore, the soil N supply and plant N uptake from drained peatlands might be higher than in mineral soil.
Around 30% of the agriculturally used peatland is managed as grassland globally (Leifeld and Menichetti 2018;Evans et al. 2021). For the temperate zone, plant N uptake in grasslands has been widely explored in both mineral soils and organic soils. It has been reported from grasslands on mineral soil that plant biomass accumulated up to ~ 130 kg N ha −1 year −1 without fertilization in Germany (Bessler et al. 2012). Müller et al. (2011) found that, based on a 38-year field observation in Germany, the aboveground grass N uptake ranges from 50 to 200 kg N ha −1 year −1 with N application of ~ 200 kg N ha −1 . In a study on grasslands on organic soil, Sonneveld and Lantinga (2011) reported an aboveground grass N uptake of 342 kg N ha −1 based on a 3-year field experiment in drained peatland without fertilization in the Netherlands. Schothorst (1977) even reported an aboveground grass N uptake of ~ 400 kg N ha −1 from a non-fertilized drained peatland in the Netherlands. These data tentatively suggest that plant N uptake in drained organic soil might be generally higher than in mineral soil, which might be related to the higher soil N supply in drained peatland through organic matter decomposition. Higher soil N mineralization often leads to a supply of N exceeding grass uptake, which consequently results in greater N losses to the environment of grass produced on drained organic soil compared to production on mineral soil (Pijlman et al. 2020). It has been estimated that with the ongoing agricultural use of degraded peatland, 9.7 Mt N year −1 will be released to the environment annually, and c. 2.3 Gt N will be released cumulatively with the full degradation of all currently managed peatland (Leifeld and Menichetti 2018). Therefore, it is vitally important to evaluate how the N losses from drained peatland can be reduced.
In order to compensate for continued soil subsidence of drained organic soils and thereby to maintain agricultural productivity, adding mineral soil as a cover fill with a thicknesses of 0.2-0.5 m on the surface of organic soil has increasingly been adopted by farmers working in Switzerland and other European countries (Schindler and Müller 1999;Ferré et al. 2019). With mineral soil cover, the soil N balance of drained peatlands may change due to various factors. First, the smaller surface soil C and N content in the mineral soil cover material supports smaller microbial biomass and microbial activity (Wardle 1998). This may result in lower SOM mineralization rates with mineral soil coverage compared with the surface soil from non-covered drained organic soil. Second, mineral soil cover might increase fertilizer N retention in drained peatland owing to its overall smaller N content. Third, a cover fill may also change other physical-chemical soil properties (e.g., clay content, soil pore volume, and soil cation exchange capacity) that feedback into soil N dynamics (Barrett and Burke 2002). Finally, for the peat layer underneath the mineral soil coverage, the addition of mineral soil material may compress the peat layer and push it deeper into zones with lower oxygen availability, thereby reducing the mineralization of easily degradable N in those peat layers. A prior study conducted at the same site as studied here proved that mineral soil cover induced a substantial reduction of N 2 O emissions , underpinning a strong influence of mineral soil coverage on the N balance in the soil-plant system of the drained peatland. However, a mechanistic understanding of the impact of mineral soil cover on the N cycling in the plant-soil system of drained organic soils is still missing.
In this study, we examined the N dynamics and N loss in plant-soil system in a drained peatland under grassland use both with and without mineral soil coverage. We did so by using isotopically labeled 15 N fertilizer in combination with measurements of the corresponding N pools in soil, roots, and harvests. The application of 15 N-enriched fertilizer is considered a useful and targeted tool for tracing the fate of applied N in plant-soil systems (Rahman and Parsons 1999;Wesselsperelo et al. 2006;Sebilo et al. 2013;Rowlings et al. 2016;Kalu et al. 2021). The specific objectives of this study were to (1) determine the fertilizer N recovery and fertilizer allocation in the plant-soil system in drained peatland with and without mineral soil coverage; (2) assess the soil mineral N (N and 15 N) release in drained peatland with and without mineral soil coverage; and (3) quantify the impact of mineral soil cover on the total plant-soil system N loss from drained peatland.

Field site
The field experiment was carried out in the Swiss Rhine Valley, at the site Rüthi (47° 17′ N, 9° 32′ E), a drained fen with a peat thickness of ~ 10 m. The site has a cool temperate-moist climate with a mean annual precipitation of 1297 mm and a mean annual temperature of 10.1 °C (1981-2010, https:// www. meteo swiss. admin. ch; for precipitation and temperature during the experimental period please see Fig. S1). The site was drained with ditches before 1890 (https:// map. geo. admin. ch). In 1973, an intensive drainage system with pumps and pipes was built. The site was used as pasture, and since 2013 as an intensively managed meadow. From 2006 to 2007, one part of the field (~ 2 ha) was covered with mineral soil material (without mixing with the peat underneath) to improve the agricultural usability. We established the field experiment at this mineral soil coverage site (Cov, with mineral soil coverage thickness ~ 40 cm) and used the adjacent drained organic soil (~ 9 ha) without mineral soil coverage as the reference (Ref, see Fig. S2 A). The basic soil properties for both sites are provided in Table 1. Both sites have similar vegetation and identical farming practices with 5-6 cuts per year and ~ 230 kg N ha −1 fertilizer application, both as slurry (applied with a splash plate) and as mineral fertilizer (ammonium nitrate or ammonium sulfate, applied with fertilizer sprayer). The atmospheric N deposition at the study site for 2015 is 20-30 kg N ha −1 year −1 (Rihm and Künzle 2019). Dominant grass species are Lolium perenne, Alopecurus pratensis, Festuca arundinacea, Trifolium spec., and Festuca pratensis.

Experimental design and field management
The study was conducted from July 2020 to July 2021. In July 2020, eight separate plots (four for Cov, four for Ref; size, 3.5 m × 1.5 m) were distributed on the experimental site. Each plot was divided into two subplots (1.5 m × 1.5 m, Fig. S2.B), and the distance between the two subplots was 0.5 m. At each plot, one subplot received 15 N double-labeled ammonium nitrate ( 15 NH 4 15 NO 3 ) as a treatment plot, and the other one received the same amount of non-labeled ammonium nitrate (NH 4 NO 3 ) as a control plot. For these treatment plots, 15 NH 4 15 NO 3 was dissolved in water, and the salt solution was applied in three campaigns at the same time the farmer fertilized the overall field. We dissolved 1.35 g, 1.35 g, and 0.8 g 98 atom % 15 N 15 NH 4 15 NO 3 in 2.25 L water for each application, which is equivalent to 1 mm precipitation per application. The control plot (i.e., the plot without a label) always received the same amount of NH 4 NO 3 solution. The additional N input rate was chosen based on the typical field fertilizer N application. A total of extra 15 N input of 0.57 g N m −2 for each plot were chosen; this was equivalent to ~ 2.5% of the regular field fertilizer N input, and it was assumed that this small extra dose would not cause a major disturbance in the N cycle of the ecosystem. The plot received the same regular fertilizer as the overall field. Extra 15 NH 4 15 NO 3 salt solution was sprayed directly onto the ground on 10 September 2020, 25 March 2021, and 13 May 2021. In order to spread the fertilizer solution homogenously, each subplot was divided into 15 units (0.3 m × 0.5 m, Fig. S2.C), and the same amount of the fertilizer solutions was applied to each unit.

Plant and soil sample collection and analysis
During the experimental period, soil and plant samples were taken 1 day before the regular field harvest events in October of year 2020, May, June, and July of year 2021. In addition, extra soil samples were taken on August 2020 and August 2021. Soil samples from the first sampling were used to determine the background 15 N signature over all experimental plots, and those from the last sampling were used to determine the soil bulk density from 0-5 cm to 5-15 cm. For each subplot, composite soil samples from 3 units were collected by using a 6.5-cm-diameter corer for 0-20 cm depth and a 2.6-cm-diameter corer for 20-60 cm depth. Those samples were divided into 4 layers: 0-5 cm, 5-15 cm, 15-30 cm, and 30-60 cm. The samples from Cov were additionally divided at the boundary of mineral soil cover and underlying peat by a stainless steel knife during field sampling. After sampling, soil samples were stored at 4 °C in a cooling room overnight, and visible root and stones were removed from composite soil samples the next day. Soil samples were then dried at 105 °C for 72 h, ground with mortar and pestle, milled in a ball mill (Retsch, MM 400, Germany) at 25 rotation s −1 for 3 min, and finally loaded in a tin capsule to determine the soil N and 15 N content via elemental analysis isotope ratio mass spectrometry (EA-IRMS) (vario PYRO cube, Elementar, Germany and isoprime precisION, Elementar, Germany).
For each subplot, aboveground biomass samples from three units were harvested by grass clippers to a height of 3 cm. At the same unit, root samples were collected by taking soil cores with a 6.5-cm-inner-diameter corer down to a depth of 20 cm. Composited grass samples were dried at 60 °C in the oven for 72 h to determine the dry biomass and to calculate the aboveground biomass based on the covered area of the three units (0.15 m 2 each). Dried plant samples were cut into small pieces, milled in a ball mill (Retsch, MM 400, Germany) at 25 rotation s −1 for 3 min, and then loaded into a tin capsule to determine the grass N and 15 N content with elemental analysis isotope ratio mass spectrometry (EA-IRMS) (vario PYRO cube, Elementar, Germany and isoprime precisION, Elementar, Germany). Roots were extracted from each soil core in the lab. To do so, soil material from the soil core was removed by hand, and the remaining root from the removed soil material was picked out. The left soil core and the roots that were picked out from the soil were submerged in distilled water for 2-3 h, and then put on a fine mesh screen to be washed with a gentle water shower until the residual soil material was removed. The bare roots were dried at 60 °C in the oven until the constant weight (~ 48 h) to determine the dry biomass and calculate the biomass of the root based on the covered area of the three soil cores (0.033 m 2 each) for each subplot, and then, the root N and 15 N content was determined by following the same procedure described above.

Laboratory incubation
To determine the net N and 15 N mineralization rate of the surface soil at both sites, the 0-5 cm and 5-15 cm soil samples, which were collected in October 2020, were incubated for 28 days. Five duplicated (n = 160) soil samples equivalent to 10 g dry soils were weighted into 50-ml PET containers with soil moisture adjusted to 60% of their water holding capacity. Water holding capacity was determined following Franzluebbers (2020). The PET containers were incubated at 25 °C, and soil moisture was adjusted every 2 days by adding distilled water. After 0, 7, 14, 21, and 28 days of incubation, the soil samples were suspended in 80 ml 0.01 M CaCl 2 salt solution to extract soil N and 15 N (Steffens et al. 1996), shaken at 160 cycles min −1 for 30 min, and filtered. Total N and 15 N from the soil extracts were determined by EA-IRMS (vario TOC cube, Elementar, Germany and iso TOC cube, Elementar, Germany). Daily net N and 15 N mineralization rates (N r_min , mg N kg −1 soil day −1 ; 15 N r_min , mg 15 N kg −1 soil day −1 ) from two sites and depth were calculated based on the regression slope of the five total dissolved N (mg N kg −1 soil) and 15 N (mg 15 N kg −1 soil) against their incubation times (day). The specific N and 15 N mineralization rate (specific N r_min , mg N g −1 soil N day −1 ; specific 15 N r_min , mg 15 N g −1 soil 15 N day −1 ) was calculated as the regression slope of the five specific dissolved N (mg N kg −1 soil N) and 15 N (mg 15 N kg −1 soil 15 N) against their incubation times (day). The specific extractable N was determined as the ratio of the total extractable N (mg N kg −1 soil) and the soil N content (%). Correspondingly, the specific extractable 15 N was determined as the ratio of the total dissolved 15 N (mg 15 N kg −1 soil) and the soil 15 N content (%).

Isotope calculation and statistics
The N isotope ratios of the samples are presented by using the δ notation (Fry 2006).
where R sample and R standard are the ratios between 15 N and 14 N of the sample and the standard, respectively. Here, atmospheric N 2 is used as a standard with R standard = 0.003665 (Mariotti 1983). The isotope enrichment in the sample from the treatment plot (δ 15 N sample ) is expressed as 15 N enrichment relative to that of the control plot (δ 15 N control ).
The recovery of the 15 N fertilizer in the labeled N pools is calculated as follows: where % 15 N sample is 15 N atom percent in the soil sample from the labeled plot; % 15 N control is 15 N atom percent in the corresponding control plot; M label is the amount of the 15 N applied to the treatment plot (g 15 N m −2 ); and % 15 N label is the 15 N atom percent in the labeled fertilizer, and M pool is the N amount of the labeled pool (g N m −2 ). In this study, there are three labeled pools, M pool,grass , M pool,root and M pool,soil . M pool,grass and M pool,root were determined based on the dry biomass for each plot. M pool,soil was calculated as follows: BD i is the soil bulk density (g cm −3 ) at four different soil depths (0-5 cm, 5-15 cm, 15-30 cm, and 30-60 cm). Soil bulk density from 15-30 cm to 30-60 cm was determined plot wise based on the correlation between soil bulk density and the soil organic carbon from Wang et al. (2021). L i is the thickness of each depth (cm), and N sample is the total N content of the soil sample from the labeled plot (%).
The mass balances of 15 N in the system were used to quantify the 15 N recovery in the system; any 15 N which was not retained in the plant and soil system was defined as losses. Plot-based 15 N losses (N losses ) were calculated as the difference between 15 N input through 15 N tracer application, as well as the N output from harvest and the 15 N retained in soil and roots. The 15 N input through regular fertilizer and atmospheric 15 N deposition is accounted for by the 15 N abundance from the associated control plot. The cumulative N losses at each harvest event are calculated as follows: where N losses, i is the N losses after the i th harvest event, n = 1, 2, 3, 4; 15 N fer is the cumulative 15 N input through fertilization; 15 N grass is the cumulative 15 N uptake through harvest; and N root, i and N soil, i are the 15 N retained in roots and soil at the i th harvest event, respectively.
Statistical analysis and data visualization were performed using the open source software R (version 4.1.3). Significant differences between the two sites for soil and plant N content, δ 15 N content, 15 N enrichment, net N mineralization rate, 15 N recovery, and 15 N losses were determined using a t-test. Significant differences in the ratio of the specific 15 N r_min : N r_min , N and 15 N release rate, and specific N and 15 N mineralization rate in soil layers 0-5 cm and 5-15 cm between the two sites were determined through ANOVA.
In case of a significant effect, a Tukey's HSD test was performed for multiple pairwise comparisons between different sampling dates. The error probability was set as p < 0.05. The results were always reported as mean ± 1 standard error (se).

Effect of mineral soil coverage on plant biomass, N uptake, and plant 15 N enrichment
During the experimental period, the cumulated grass yield was not different between sites ( For roots, the differences in biomass, N uptake, and 15 N enrichment were not constant between the two sites. In June 2021, root biomass and 15 N enrichment were significantly higher (p < 0.05) at Cov than at Ref, and a significantly higher root N content was found at Ref in July 2021.

Effect of mineral soil coverage on soil 15 N enrichment
Applications of 15 N labeled fertilizer induced an increase in the soil 15 N signature. At both sites, the highest soil 15 N signature was found in July 2021 after the three labeling events were finished, although no 15 N tracer was applied directly before that sampling event. At 0-5 cm soil depth, the 15 N enrichment was 146.0 ± 13.3‰ at Cov and 49.4 ± 13.7‰ at Ref (Fig. 1D). At 5-15 cm soil depth, the 15 N enrichment was 32.7 ± 8.8‰ at Cov and 7.4 ± 1.4‰ at Ref (Fig. 1D). The higher 15 N signature was only found at the surface 0-30 cm, below 30 cm depth, and the soil 15 N enrichment was similar to the value prior the 15 N tracer application, which was near zero (Fig. 1). The surface (0-5 cm, 5-15 cm) soil 15 N enrichment was higher (p < 0.05) at Cov than at Ref, whereas below 15 cm, the difference in 15 N enrichment between the sites was less pronounced at any sampling date (Fig. 1). Table 2 Plant biomass, N uptake, and 15 N enrichment of a drained peatland with (Cov) and without mineral soil coverage (Ref) Significant differences between the two sites over the experimental period are indicated with asterisks ("**", p < 0.01, "*", p < 0.05, "ns", no significant difference). At Cov, the 15 N signal moved from the surface soil to the deeper (15-30 cm) layer during the growing season and resulted in a slightly higher 15 N enrichment (49.4 ± 13.7‰) at 15-30 cm depth compared with the upper layer (32.7 ± 8.8‰) at the last sampling date (July 2021). However, no such trend was found for Ref (Fig. 1D).

The effect of mineral soil coverage on 15 N recovery from drained organic soil
During the experimental period, the recovery of 15 N in plants (aboveground biomass and roots) was not different between Cov and Ref. The cumulative tracer exports through aboveground biomass harvest accounted for 32.2 ± 2.2% and 30.0 ± 0.3% of the applied 15 N for Cov and Ref, respectively ( Fig. 2A). Roots took up 2.5 ± 0.3% and 3.9 ± 0.5% of the applied 15 N from Cov and Ref, respectively, after the three labeling events were finished ( Fig. 2A). Hence, a significant part of the applied 15 N was not used by the plants. A share of 10-20% was incorporated into the soil N pool. At site Cov, 19.8 ± 2.0% of the tracer remained in the soil N pool, more (p < 0.05) than at Ref (9.8 ± 3.2% see Fig. 2B). Overall, site Cov showed smaller N losses (p < 0.05) compared to Ref. At Cov, 45.4 ± 3.0% of the applied labeled mineral fertilizer was lost outside the plant-soil system boundary of the study, whereas at Ref, the loss accounted for 56.2 ± 3.1% (Fig. 2C).

Specific N mineralization
The specific soil N mineralization (specific soil N sr_min ) was significantly higher with mineral soil For each sampling date, significant differences between two sites at different soil depths are indicated with asterisks ("**" p < 0.01, "*"p < 0.05, "ns" no significant difference) coverage (Table 3) for both soil layers (0-5 cm and 5-15 cm). Throughout 28 days of incubation, the specific soil N sr_min at site Cov was 0.60 ± 0.07 mg N g −1 N day −1 and 0.35 ± 0.03 mg N g −1 N day −1 at Ref at a soil depth of 0-5 cm. A similar difference was also found at 5-15 cm soil depth, where Cov released 0.58 ± 0.03 mg N g −1 N day −1 , significantly more than Ref (0.36 ± 0.03 mg N g −1 N day −1 ). In addition, specific soil 15 N sr_min was also significantly higher with mineral soil coverage (Table 3) for both soil layers (0.68 ± 0.10 mg 15 N g −1 15 N day −1 at 0-5 cm depth and 0.71 ± 0.14 mg 15 N g −1 15 N day −1 at 5-15 cm depth) than at Ref (0.39 ± 0.02 mg 15 N g −1 15 N day −1 at 0-5 cm depth and 0.37 ± 0.02 mg 15 N g −1 15 N day −1 at 5-15 cm depth).
The ratio of the specific soil 15 N sr_min release to the N r_min release was above one for both layers and sites (Fig. 4). No difference was found between the two sites; however, the ratio of specific 15 N sr_min and N sr_min from the 5-15 cm soil layer was lower than from the 0-5 cm soil layer. Significant differences among the two soil layers and the two sites are indicated with lowercase letters (ANOVA and Tukey's honest significant differences)

Soil 15 N retention
The field 15 N tracer experiment showed that 10% of the applied 15 N tracer resided in the soil pool in the Ref site, of which more than 90% was found in the top 30 cm of soil. This result was similar to the 15 N retention from the drained fen peatland reported by Augustin et al. (1997) who found that 10-20% of the applied labeled 15 N fertilizer were recovered in the soil pool in a 15 N tracer experiment from two drained peatlands in Germany, of which more than 90% was located at the 0-20 cm depth. To the best of our knowledge, soil 15 N recovery from drained peatland with mineral soil coverage has never been studied. Our results indicate that the soil 15 N recovery (~ 20% of the applied 15 N tracer) from the Cov site was generally significantly higher than at Ref at the end of the study period, suggesting a better retention of the fertilizer-N though mineral soil coverage. The recovery was at the lower end of the range of data reported from 15 N tracer (with 15 N enriched fertilizer or slurry) studies in grassland on mineral soil in Europe. This included 20-25% soil 15 N retention from a grassland in southern England (Jenkinson et al. 2004), ~ 15% from grassland in the Netherlands (De Vries et al. 2011), and 30-40% from grassland in Germany (Zistl-Schlingmann et al. 2020).
However, we observed a downward movement of the 15 N tracer to the deeper layer at Cov, whereas no such trend was found at Ref (Fig. 1D) over the course of the experiment. This indicates that, despite a higher overall recovery in the studied soil layers, fertilizer N might leach faster in the covered mineral soil material at Cov than in the drained peatland at Ref. This may, after longer periods, also change the overall recovery once the leachate leaves the investigated zone of 0-60 cm. The higher 15 N leaching from Cov may be attributed to the low absorption rate of the mineral cover material compared to the degraded peat, as the sand content of the mineral soil coverage sites over the experimental period are indicated with asterisks ("**" p < 0.01, "*" p < 0.05, "ns" no significant difference) in B and C. The dashed line in A separates the aboveground biomass 15 N recovery and the belowground root 15 N recovery is much higher than that of the peat at Ref (Table 1). Hence, the low adsorption potential of the sand for anions compared to organic materials with higher anion sorption capacity may have induced a higher N leaching at the mineral soil layer from Cov.
Higher soil 15 N recovery at Cov might be due to a higher microbial 15 N uptake. Microbial N use is positively correlated with soil substrate availability and soil pH (Elrys et al. 2022). Compared with Ref, the SOM pool from Cov is small (Table 1), but relatively young and labile as previous shown from a 14 CO 2 measurement on the same site (Wang et al. 2021). The higher availability of labile SOM stimulates soil microbial activity, which ultimately promotes microbial N uptake (Barrett and Burke 2000;Booth et al. 2005;Yang et al. 2022). Moreover, the relative old and stable SOM pool from Ref may exist in forms that the microorganism cannot easily use (Baldock and Skjemstad 2000;Fontaine et al. 2007). This may result in a The shaded area indicates the maximum and minimum rate of soil N r_min and 15 N r_min release from soil. The p values indicate significant differences (t-test) of soil N r_min and 15 N r_min release between the two sites lower microbial N uptake at Ref than at Cov. In addition, soil pH of the surface layer at Cov was higher than at Ref (Table 1), which may enhance soil microbial N uptake and further induce better soil 15 N retention at Cov than at Ref, due to the enhanced microbial activity under high soil pH (Zhang et al. 2017).

Soil N mineralization
The laboratory incubation results showed that soil N r_min and soil 15 N r_min at 0-5 cm and 5-15 cm depth were significantly higher (p < 0.05) at Ref than at Cov. We attribute this to the overall higher soil N content of the surface peat compared to the mineral soil cover material. In contrast, the specific soil N sr_min release was higher at Cov than at Ref (Table 3), which indicates that the surface soil organic matter (SOM) at Cov was more labile compared to the Ref site. The labile SOM pool at Cov coincided with the young carbon age at Cov from the former study on the same site (Wang et al. 2021). By definition, the labile SOM pool decomposes very quickly and is easily accessible to plants and microbes (Dungait et al. 2012;Liu et al. 2017).
At both sites, the soil 15 N r_min release was faster than the N r_min release (Fig. 4), implying that the added 15 N turnover rate was higher than the gross N turnover rate. This finding indicates that the newly applied mineral N ( 15 N and 14 N), after incorporation into SOM, was preferably stored in the labile soil N pool. This finding is consistent with former studies showing that the exogenous N input is mostly labile (Shevtsova et al. 2003;Mulvaney et al. 2009;Sebilo et al. 2013). This part of the soil organic N pool releases available N for plant uptake in the growing season, but likewise bears the risk of N losses to the environment.

The effect of mineral soil coverage on plant 15 N uptake
Over the experimental period, both sites had similar aboveground biomass; however, the aboveground plant N uptake was higher at Cov than at Ref (p < 0.05), suggesting that mineral soil coverage not only sustains the agricultural productivity of the drained peatland, but also increases the fertilizer N use efficiency. However, mineral soil coverage did not influence plant and root 15 N content. At both sites, plants took up ~ 30% of the applied 15 N fertilizer, similar to the results reported from a meta-analysis of 15 N tracer studies, which found that on average, 30% of the applied 15 N is taken up by plants in grassland (Templer et al. 2012). It is often assumed that plant N uptake tends to be higher with higher soil N availability (Stevens et al. 2005;Tateno and Takeda 2010). However, we found that the application of Table 3 Average soil N release rate and specific N mineralization rate in soil layers at depths of 0-5 cm and 5-15 cm from drained peatland with (Cov) and without mineral soil coverage (Ref) Significant differences among the two soil layers and the two sites are indicated with lowercase letters (ANOVA and Tukey's honest significant differences). 0.68 ± 0.08 a 0.71 ± 0.14 a 0.39 ± 0.02 b 0.37 ± 0.02 b Fig. 4 The ratio of specific 15 N mineralization rate to specific N mineralization rate in soil layers with depth of 0-5 cm and 5-15 cm from drained peatland with (Cov) and without mineral soil coverage (Ref). This ratio was calculated from the specific extractable 15 N (mg 15 N kg −1 soil 15 N) and specific extractable N (mg N kg −1 soil N) from five different measurement days. Significant differences among the two soil layers and the two sites are indicated with lowercase letters mineral soil material, which was relatively poor in N and also released an absolutely smaller amount of N in the incubation experiments, did not reduce the plant N uptake and the plant 15 N recovery. The similar aboveground biomass and 15 N recovery might be driven by the ample amount of N supplied at both sites and the lack of any N limitations. During the experimental period, ~ 230 kg N ha −1 was applied equally to both sites, and the soil N r_min results suggest that soil mineralization could supply further N, exceeding the demand for grass production at both sites. Therefore, as N was not limited in the system, the presence of relatively N-poor mineral soil did not impair aboveground yields.

Fertilizer N loss reduction
The two sites received ~ 230 kg N ha −1 year −1 fertilizer N input. Together, the fertilizer N input and the soil N supply largely exceed the plant N demand and consequently lead to N losses to the environment, i.e., release into the atmosphere via ammonia volatilization and denitrification as well as via leaching to the groundwater (Robertson and Vitousek 2009;Bowles et al. 2018). At Cov, less of the applied N was lost through the experimental period, considering the storage in the 0-60 cm soil layer. Thus, mineral soil coverage at this site may prevent ~ 25 kg N ha −1 year −1 fertilizer N from being lost to the environment compared with Ref if no substantial leaching below 60 cm will occur.

Effects on peat decomposition
The higher soil N release from Ref at 0-15 cm soil depth (Table 3) also implies a rapid peat decomposition and peatland degradation, as the soil N losses are closely linked to the C losses from the SOM mineralization (Leifeld et al. 2020;Klein et al. 2022). However, we only have evidence for this for the topsoil, whereas the peat underneath the mineral soil coverage was not used for determining soil N r_min in the incubation experiment due to the reasons below. Firstly, the soil 15 N signature underneath 15 cm soil layer was nearly natural abundance (Fig. 1B), which makes it impossible to determine the soil 15 N mineralization from the deeper soil layer. Second, the oxygen availability in the deeper soil layer is difficult to simulate in laboratory incubation, and the atmosphere oxygen availability may overestimate the soil N and 15 N mineralization rate from the deeper soil layer.
It may be suspected that at Cov, these subsoil organic layers may have a higher N r_min release than the surface organic soil from Ref site due to two possible mechanisms. First, the covered mineral soil material revealed some N leaching (Fig. 1). This leachate from the mineral soil material may stimulate the decomposition of the peat underneath the mineral soil cover via positive priming (Kuzyakov et al. 2000). Second, the mineral cover enhanced the soil pH of the peat layers underneath (Table 1). As SOM decomposition increases with soil pH (Sinsabaugh et al. 2008), a higher potential for peat decomposition underneath the mineral soil coverage may be possible.
On the other hand, oxygen availability is vital for peat mineralization (Blodau 2002;Tiemeyer et al. 2016). For the peat layer underneath the mineral soil coverage, the oxygen availability for SOM mineralization is reduced (Jørgensen et al. 2012), leading to a presumably lower N release. In addition, the absence of fresh plant residue input into the deeper layers of the organic soil underneath might limit SOM mineralization (Song et al. 2018;Zhang et al. 2021). Fresh plant inputs are the primary source of SOM formation, which could not only determine the chemical composition of SOM, but also impact soil microbial activities. The exclusion of fresh plant inputs may lead to a N limitation for microorganisms and further limit the N mineralization for peat underneath the mineral soil coverage (Mooshammer et al. 2014). Moreover, a former study conducted at the same site found that SOC from mineral soil material contributed greatly to heterotrophic soil respiration at the Cov site (Wang et al. 2021). The contribution of the peat layer underneath the mineral soil coverage was relatively small compared to the contribution of peat C at Ref (Wang et al. 2021). The lower contribution of subsoil peat to C loss from Cov suggests that mineral soil cover might be able to reduce the peat decomposition rate despite incoming N leachate and a higher pH. However, further in situ soil profile-based SOM mineralization experiments are still needed to support this interpretation.

Conclusion
Our findings suggest that mineral soil coverage has the potential to reduce N losses (due to higher soil N retention) from drained peatland and, hence, may make agricultural production on drained peatland less harmful to the environment compared to the continued direct use of these soils. However, we would like to point out that from a nature conservation standpoint as well as climate mitigation strategy, mineral soil coverage does not replace or substitute the mitigating effect that can be achieved by rewetting. Rather, we aim to encourage further research about mineral soil coverage as a peatland management measure in regions, where peatland rewetting is not supported, be it for reasons of national food and feed provision, economic incomes, or political strategies.
Our field 15 N trace experiment and laboratory incubation together provide the first insight into how mineral soil coverage influences the N balance of the plant-soil system in agriculturally managed drained peatland. Over the experimental period, mineral soil coverage of drained peatland significantly reduced the system fertilizer N loss. For the deeper peat layer, the effect of mineral soil coverage on peat decomposition and mineralization still needs to be further explored. In summary, the study suggests that mineral soil coverage, a measure used by farmers to counterbalance subsidence, provides an opportunity for reducing the environmental pollution induced by the agricultural use of drained peatland. Furthermore, the reduced N losses with mineral soil coverage from our study also highlighted the need for multiply field observation to evaluate the effectiveness of mineral soil coverage in general terms.
Acknowledgements We appreciate the help from Markus Jocher, Robin Giger, Steven Nagel, Martin Zuber, and Shiva Ghiasi at Agroscope during field sampling and lab analysis. We are thankful for many useful discussions with Christof Ammann and Chloé Wüst, Agroscope. We are grateful to Bernhard Schneider for the opportunity to collaborate on his farm.
Funding Open access funding provided by Agroscope The work was supported by funding received from the Swiss Federal Office for the Environment (contract number 06.0091.PZ/R261-2425) and the China Scholarship Council (NSCIS 201806350221).
Data Availability Data from this study are included in the article and supplementary material; further inquiries can be directed to the corresponding author.

Conflict of interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.