Nitrous Oxide Emissions and Methane Uptake from Organic and Conventionally Managed Arable Crop Rotations on Farms in Northwest Germany

Land-use extensification by shifting from conventional to organic arable farming is often discussed as a measure for reducing greenhouse gas (GHG) emissions from agricultural land. Doubts about the benefits arise when emissions are calculated per product unit, particularly where high yields are possible under conventional management. Among the non-CO2 GHG emissions, nitrous oxide (N2O) is the main contributor from arable land and is controlled by soil type, environmental conditions and management. In order to investigate how land-use change from conventional to organic farming would perform under highly productive site conditions in northwest Germany, and how this would affect the important greenhouse gases N2O and methane (CH4), an on-farm field research was conducted over two experimental years. Two site-specific organic crop rotations, (i) with 25% legumes (grass + clover–winter wheat–winter rye–oats) and (ii) with 40% legumes (grass + clover–winter wheat–winter rye–spring field peas–winter rye), were compared with (iii) a conventional arable rotation (winter oilseed rape–winter wheat–winter wheat–sugar beet–winter wheat) and two reference systems, (iv) extensive grassland and (v) a beech forest), which were chosen as the baseline. The results showed that organic farming had lower N2O emissions of 0.7 N2O–N ha−1 year−1 than the conventional rotation, with 2.1 kg N2O–N ha−1 year−1 (p < 0.05), but higher emissions than the extensive grassland (0.3 kg N2O ha−1 year−1) and beech forest (0.4 kg N2O ha−1 year−1). CH4 emissions were a negligible part of total GHG emissions (as CO2 equivalents) in the two arable systems, and considerable uptake of CH4 from the forest soils showed this was a GHG sink in the first experimental year. Organic systems produced up to 40% lower crop yields, but the emissions per product unit in rotation (iii) was not superior to (ii) during the two experimental years. Thus, arable organic farming showed the ability to produce agricultural commodities with low N2O emissions per unit area, and no differences in product-related emissions compared with conventional farming. Conventional and organic systems both showed potential for further mitigation of N2O emissions by controlling the field level nitrogen surplus to a minimum, and by the optimized timing of the removal of the grass–clover ley phase.


Introduction
Nitrous oxide (N 2 O) is an important greenhouse gas (GHG) in agriculture and has a global warming potential (GWP) 265 times that of carbon dioxide (CO 2 ) over a 100-year time horizon [1]. Agricultural N 2 O emissions are therefore considered to contribute significantly to climate change [2] from arable conventional and organic systems, under common and current agricultural management, to examine whether GHG mitigation targets are fulfilled. In this context, a more detailed understanding of N losses is also crucial for further lifecycle impact assessments, as inaccurate prediction of N cycles when comparing organic and conventional production can result in large errors, which often show disadvantages for organic systems [36]. Moreover, typical regional production systems have to be chosen, rather than simple comparisons of identical crop rotations. Organic farming has different requirements for crops, such as greater reliance on N-fixing legumes compared with conventional systems that use external synthetic N, leading to different agricultural commodities and environmental impacts [39].
Accordingly, in this paper, we present results from an on-farm study to address these important questions: (1) Can stockless arable organic farming reduce direct GHG (N 2 O and CH 4 ) emissions from soils per ha and per product unit relative to a conventional stockless arable system, on highly productive sites in northern Germany? (2) What are the main drivers for the important field-level N 2 O emissions in the investigated systems, and how they can be improved? (3) To what extent do the examined arable systems behave in terms of GHG emissions compared with other referenced land-use systems (extensive grassland and beech forest)?

Research Area
The region where the study was conducted in Schleswig-Holstein, northern Germany, has a hilly terminal-moraine landscape and supports approximately 400,000 ha of agricultural land that is mainly used for intensive arable cropping. The weather is characterized by a temperate oceanic climate. Mean (1981-2010) annual temperature is 8.9 • C and annual precipitation is 737 mm. The region, together with similar areas along the Baltic Sea coast of Denmark, southern Sweden and north-east Germany, is one of the most suitable regions in Europe for cereals, oilseed rape and winter wheat; e.g., yields for winter wheat on commercial farms typically exceed 9.0 t ha −1 (based on 6-year average 2008-2013). Organic farming is currently practiced on only 3.7% of the utilized agricultural area in this region. In view of current policy aims to increase land under organic management, we considered on-farm investigations under these favorable climate and soil conditions to be appropriate to answering the question of to what extent a shift from conventional to organic arable farming would affect productivity and GHG emissions from sandy loam soils used for arable cropping.

On-Farm Research Design
The two-year field study (1st year: October 2012-September 2013; 2nd year: October 2013-September 2014) was conducted on 400 ha land comprised of arable land, beech forest and permanent grassland (53 • 39' N, 10 • 34' E). All land-use types were arranged completely randomized within this area. The arable land was managed under the common agricultural practice for that area. The study area included two organically managed arable crop rotations and a typical conventional stockless cropping system with a history of more than ten years of continuous management. The crop rotations were established randomly on fields of 5-23 ha, and every crop in each crop rotation was present during both experimental years. There were four replicates for investigations established in each field to provide a representative average for each field. Soils were classified as sandy loam (10.6-11.4% clay, 26.3-29.4% silt and 59.4-63.0% sand in the 0-30 cm topsoil), with dominant soil types being Luvisols, Cambisols and Haplic Stagnosols.
Total precipitation during the experimental years (1st year: 645 mm; 2nd year: 615 mm) was less than the long-term average (737 mm). Compared with the long-term average of 8.9 • C, air temperature during the study period was similar in the first experimental year (8.4 • C) but higher in the second year (10.4 • C). The differences between the two experimental years are also shown by numbers of days with temperatures below freezing point. In the first winter period, there were 62 days with daily average temperatures below 0 • C, whereas only 16 days with sub-zero temperatures were recorded in the second year ( Figure 1).
In order to account for the N inputs in the organic systems, the determination of BNF by forage legumes was based on the empirical model of Høgh-Jensen et al. [40]. This approach is calculated from measures of dry matter (DM) yield of legumes and their N%. The relevant default parameters were set according to the assumptions of a 1-2-year-old grass-red clover sward given by Høgh-Jensen et al. [40]. To determinate the amount of N2 fixation of grain legumes, we used the extended difference method (reference crop oat) according to Wichmann et al. [41]. Based on these models the calculated average BNF for the crop rotation elements was 50 and 65 kg N ha -1 year -1 for organic-low-BNF and organic-semi-BNF, respectively, with no further N amendments. In the conventional system, however, N inputs were primarily as mineral fertilizers with small amendments of digestates and pig slurry, with total N rates of 224 kg N ha -1 year -1 . The N rates were consistent with the agricultural practice recommended for each specific crop (Table 1).

Figure 1.
Monthly sum of precipitation in mm month −1 (grey bars) compared with the long-term averages (black dots), and daily mean air temperature (grey line) compared with the long-term average (black line) during the experimental years (October 2012-October 2014).
In order to account for the N inputs in the organic systems, the determination of BNF by forage legumes was based on the empirical model of Høgh-Jensen et al. [40]. This approach is calculated from measures of dry matter (DM) yield of legumes and their N%. The relevant default parameters were set according to the assumptions of a 1-2-year-old grass-red clover sward given by Høgh-Jensen et al. [40]. To determinate the amount of N 2 fixation of grain legumes, we used the extended difference method (reference crop oat) according to Wichmann et al. [41]. Based on these models the calculated average BNF for the crop rotation elements was 50 and 65 kg N ha −1 year −1 for organic-low-BNF and organic-semi-BNF, respectively, with no further N amendments. In the conventional system, however, N inputs were primarily as mineral fertilizers with small amendments of digestates and pig slurry, with total N rates of 224 kg N ha −1 year −1 . The N rates were consistent with the agricultural practice recommended for each specific crop (Table 1).
Soil cultivation was adapted to the specific weed infestations and therefore varied between systems, years and crops. In the organic systems the primary soil tillage was carried out by ploughing, whereas a cultivator was used in the conventional system. In the organic systems the arable land for summer crops was managed mechanically in autumn and was not sown or covered with any plant residues during winter, with bare ground until seeding in spring (see Supplementary Information Table S1). Grass-clover mixtures in the first year were established by under-sowing in spring. The grass-clover mixtures in the second year were sown in August, shortly after harvesting. Grass-clover swards were cut three times a year in organic-low-BNF and twice a year in organic-semi-BNF. In addition, the study included two permanent land use types as reference systems. These were permanent extensive grassland and beech forest, each of which occupied more than 5 ha. The permanent grassland has been cut twice each year since its establishment in 2004. The sward composition showed typical patterns for extensively used permanent grassland, with 49% L. perenne, 13% T. repens, 12% P. trivialis, 8% T. sect. Ruderalia, 6% D. glomerata, 5% P. pratensis and 7% other species. The beech forest has existed in its current form for more than 50 years. Only beech trees were present in the study area. Both treatments were included in the research area to provide additional information for either extensively managed agricultural land (grassland) or reserve area (beech forest) as a baseline scenario for comparison with land-use for arable agriculture. Both reference systems were on land with the same soil conditions as described above.

Flux-Measurements
Fluxes of N 2 O and CH 4 were measured with the static closed chamber method [42]. The minimum sampling frequency was once a week in all rotations and crop, taken between 10:00 a.m. and 12:00 p.m. during the two experimental years (01 October 2012-14 October 2014). In a pre-treatment, four collars (d = 60 cm, h = 15 cm) per crop (n = 4) made from polyvinyl chloride (PVC) were installed in the 10 cm soil depth at an angle of 45 • with uniform distances in order to capture a representative area across each field. The collars were removed briefly during harvesting and tillage operations. During the flux measurements, the collars were closed gas tight with white PVC chambers (d = 60 cm, h = 35 cm). The area between the basal ring and chamber was tightened with a taut butyl rubber band. The chambers were closed for 60 min on each measurement day, when a fan within each chamber allowed homogenized air conditions before sampling. Gas samples were taken at 0, 20, 40 and 60 min after closure through a gas-tight septum on the top of the chamber using a 30 ml syringe. Samples were directly transferred into 12 ml pre-evacuated septum-capped vials (Labco, High Wycombe, UK). After fertilizer application events, measurements were conducted more frequently over two weeks at irregular intervals. The size of the collars and chambers allowed for undisturbed plant growth within the collars. In the later growth stages, when plant growth exceeded chamber height, extensions (h:35 cm) were used. Gas samples were analyzed for N 2 O and CH 4 through a gas chromatograph (SCION 456-GC, Bruker, Leiderdorp, Netherlands). Samples were injected using an autosampler (model 271 LH, Gilson Inc., Middleton, WI, USA). Data were processed using the software Compass CDS (Version 3.0.1). The change of gas concentration in the chamber headspace during the measurement was calculated by linear regression. In order to calculate the CO 2-equivalents (CO 2eq ), the GWP values of 265 for N 2 O and 28 for CH 4 were used [1]. In order to calculate the product carbon footprint (PCF), derived from the two non-CO 2 trace gases, namely N 2 O and CH 4 , the accumulated annual CO 2eq fluxes were divided by the functional unit grain equivalents (see Section 2.5.) and is referred to as PCF NON-CO2 in the remainder of the paper.

Determination of Yield and N-Balance
The fresh-matter (FM) yields of crops were recorded at a field scale by weighing the commercial machinery with the harvested crops. To calculate the dry matter yields, additional subsamples of aboveground biomass were taken with shears at random in the field, with five replicates prior to each harvest. In the permanent grassland and grass-clover swards, subsamples were taken at a stubble height of 5 cm from an area of 0.25 m 2 . In crops, the biomass samples were taken on 1.0 m 2 directly above the soil surface, although for sugar beet, samples were subdivided into primary and secondary products from yield samples. The dry matter contents of the subsamples were estimated after oven drying for 24 h at 58 • C. After drying, the samples were milled to pass a 2 mm sieve (Cyclotec mill, Foss, Hillerød, Denmark). Subsequently, the N and C contents of all samples were measured using a C/N analyzer (Vario Max CN, Elementar, Hanau, Germany). The values for the N balance (kg N ha −1 year −1 ) were calculated after deduction of the N yields (N output) from the amounts of applied N and BNF (N-input): For product unit comparisons, the functional unit "grain equivalents" (GE) was chosen, using values obtained from the German Federal Ministry of Food and Agriculture [43], with GE = 1.04 for wheat, GE = 1.01 for rye, GE = 0.84 for oats, GE = 1.04 for peas, GE = 1.3 for oilseed rape, GE = 0.23 for sugar beets and GE = 0.58 for grass-clover. One GE was defined as the feeding value of 100 kg barley. The GE factors were multiplied by the DM yields of the different crops.

Statistical Analysis
The statistical software R (2015) was used to evaluate the data. The data evaluation started with the definition of an appropriate statistical model. The data were normally distributed and heteroscedastic due to the experimental year and single crop rotation elements within the system. For comparison of crops within a single system the statistical model included the experimental year (first and second) and crops, as well as their interaction terms as fixed factors. For management system comparison, the experimental year and crops were considered as random factors. Based on these mixed effect models an analysis of variance (ANOVA) was conducted to answer the questions of the trial. Multiple contrast tests were then conducted in order to compare the several levels of the influence factors. In addition, simple linear regression models were developed to test whether N 2 O emissions can be estimated from other measured variables (e.g., N balance, CN ratio of crop residues). The significance of the tested factors, comparisons of means, and regression equations (intercepts and slopes) were declared when p < 0.05.

GHG-Fluxes
The experimental year showed an interaction with crop and the tested systems on GHG fluxes per ha −1 and PCF NON-CO2 (Table 2). Daily GHG fluxes measured during the two experimental years in the arable crop rotations and reference system are presented in  Figure S1). In the comparison of the tested arable crops, grass-clover showed the lowest N 2 O fluxes (Figures 2 and 3). In general, there were clearly detectable emission peaks for conventional crops when fertilizer dressings took place ( Figure 4). In all systems, there were additional N 2 O peaks after soil-tillage and frost-thaw events during winter. The organically managed crops showed low emissions on average, with the highest figures for cereals, of 1 kg N 2 O-N ha −1 year −1 or less. In comparison, the conventional system had the highest emissions, with annual emissions of > 1.5 kg N 2 O-N ha −1 (Table 3). Annual N 2 O-emissions in the two reference systems (grassland and beech forest) were lowest on average and were below 0.5 kg N 2 O-N ha −1 year −1 in both experimental years. 1 kg N2O-N ha -1 year -1 or less. In comparison, the conventional system had the highest emissions,

95
with annual emissions of > 1.5 kg N2O-N ha -1 (Table 3). Annual N2O-emissions in the two reference 96 systems (grassland and beech forest) were lowest on average and were below 0.5 kg N2O-N ha -1 year -97 1 in both experimental years.      The amounts of N input, N balance and the C/N ratio were statistically tested against amounts of released annual N 2 O emissions. In general, the conventional crop rotations showed higher N inputs and positive N balance in all crops and experimental years. With regard to the N balance levels in the organic systems, the values in the subsequent crop after cereals were negative. The highest surpluses were measured in the winter wheat after incorporation of the preceding grass-clover swards.
The regression analysis showed, on average, a positive linear relationship between N input and annual N 2 O emissions for the tested arable systems (intercept: 0.44, slope:0.007, R 2 = 0.6, p < 0.01). The relationship of N surplus on annual N 2 O emissions was also significant (intercept: 0.86, slope:0.01, R 2 = 0.6, p < 0.01). Comparing this N surplus to the N surplus of the preceding crop, the effect of the preceding crop was a more reliable indicator to predict annual N 2 O emissions (intercept: 0.4, slope:0.03, R 2 = 0.8, p < 0.01) ( Figure 5). This fact was also confirmed by a strong effect of the C/N ratio of plant residues on the measured N 2 O losses in the subsequent months after incorporation into soil. The measured emissions of N 2 O declined exponentially, with increasing C/N ratio from incorporated straw and stubble residues ( Figure 6). On average, for the different arable crop rotations, the C/N ratios of crop residues differed in the order of conventional (65 (SE 5.9)) < organic-semi-BNF (102 (SE 6.4)) < organic-low-BNF (125 (SE 5.9)) (p < 0.05).  According to the performed regression analysis, the emission factor for N inputs was 0.7%. However, taking the N balance of the main crop and the preceding crop into account, the emissions factors are 1% and 3% of each kg N in surplus, respectively.
For CH 4 , the tested arable systems showed only negligible exchange (values close to zero), or they acted as a sink. The measured sink of the beech forest, as a natural reference ecosystem, exceeded those of the agriculturally managed systems. Thus, annual accumulated emissions showed a notable CH 4 sink capacity of 3.9 and 2.5 kg CH 4 -C ha −1 in the first and second year, respectively.   sugar beet as a preceding crop (Table 3).

161
The annual emissions of CO2eq averaged over each of the whole crop rotations are given in Table   162 4. All the studied agricultural systems were sources of GHG emissions, whereas the beech forest

Crop Yields and Yield-Related GHG Emissions
The crop yields of the organic crop rotations were lower in both experimental years compared with the conventional farming system. On average, the GE output from organic farming was 0.32 of the yield of the conventional system. The total average yields of the two organic systems were similar for dry matter but slightly higher for energy expressed as GE in the organic-semi-BNF (38 vs. 36 GE year −1 ). Within the organically managed systems the grass-clover leys showed the lowest emissions per GE. Slightly higher emissions of 5 kg CO 2 e GE −1 were observed for winter rye. A higher PCF NON-CO2 was found in the organically managed winter wheat due to the effect of grass-clover removal as a preceding crop. Due to the relatively low yield level of spring oat and spring field peas, the product-related emissions of these crops exceeded all other crops in the organically managed systems. In the conventional system the highest PCF NON-CO2 values were found in oilseed rape and winter wheat with sugar beet as a preceding crop ( Table 3).
The annual emissions of CO 2eq averaged over each of the whole crop rotations are given in Table 4. All the studied agricultural systems were sources of GHG emissions, whereas the beech forest functioned as a sink for CO 2eq because of its negligible N 2 O losses and significant CH 4 consumption during the first year. The highest CO 2eq emissions were observed for the conventional system, whereas emissions from the two organically managed systems, on average over the two years, were 0.35 that of the conventional system. The differences in the PCF NON-CO2 values showed no clear trend between the different systems. Both the organic-low-BNF and conventional cropping systems showed differences between the experimental years, but these were not different between organic-semi-BNF and grassland. During the first experimental year, the PCF NON-CO2 values were lowest for the reference grassland and highest for conventional farming. However, during the second year PCF NON-CO2 was lowest for conventional farming and highest for organic-low-BNF, with no differences to organic-semi-BNF (Table 4). Thus, comparing the two organic systems, the organic-semi-BNF showed an advantage, with either lower emissions than the organic-low-BNF or no differences to the conventional system. Table 4. Average global warming potential (GWP) and product carbon footprint (PCF NON-CO2 ) per system and experimental year (first and second). Yields were estimated at farm level by weighing the harvested crops. Product carbon footprints were derived from GWP and grain equivalents (GE). Different uppercase letters show significant differences between the two experimental years within each system. Different lowercase letters show significant differences between the systems within each experimental year. (p < 0.05). Standard errors (SE) are shown in brackets.

First Experimental Year
Second

GHG-Fluxes
In general, there were differences between the investigated organic and conventional systems in crop rotations, type of grown crops, and the intensity of N input. Even though the conventional system was managed in accordance with the German fertilizer ordinance (2017), it had a substantially higher N input than the organic systems and reference land-use systems. It is widely accepted that the N input from synthetic fertilizers and manure is the main source of N 2 O emissions from agricultural soils [28,29]. The daily N 2 O flux rates and the measured emission peaks generally followed the fertilizer applications, which is consistent with similar previous findings [10,[44][45][46]. This also accords with GHG inventories using N input as the main variable to calculate direct N 2 O emissions from agricultural soils. However, although the standard emission factor used was 1% for each kg of N-applied, we found emission factors of 0.2 and 1.8% in the conventional crop rotation. This may be explained partly by differences between different types of N fertilizer [47,48], as greater N 2 O peaks after application of manure, in comparison with applications of mineral N fertilizer, are likely [46,49] due to the higher availability of easily accessible carbon triggering high N 2 O peaks from soil denitrifiers, when oxygen is limiting. Other authors have also concluded that calculations of N 2 O emissions based on N-fertilizer input are not reliable, as better correlations exist between N surplus and N 2 O emissions [28]. Hence, to reduce N 2 O emissions, the aim at the farm scale should be to increase N use efficiency [32]. This aim was also supported by the results of our study, with a significant correlation between N 2 O emissions and the calculated N balance of the preceding crop showing a better relationship than only the N input. In addition to the fertilizer application and its management, the potential mineralizable N from crop residues is a further main driver of N 2 O release in soil. This was particularly evident in the conventional systems for winter oilseed rape, as the N removed in the harvested seeds is low and the crop residues are within a narrow C/N ratio. This situation is also true after sugar beet harvest, due to the narrow C/N ratio of the leaves and tops with high moisture content compared with cereal straw and rape straw [50]. In the case of cereal straw, low N 2 O emissions can be further explained by temporary immobilization of soil N due to the high C/N ratios. This may have occurred in the case of incorporated cereal straw material, potentially explaining the comparatively low emissions of N 2 O.
In the organic systems, grass-clover swards are utilized as the main source of N [32,51], but with different findings of the effect on N 2 O emissions. According to the results of Benoit et al. [52], intact legume swards were shown to be the crops with the lowest N 2 O emissions in agricultural systems because of a high potential for effective N cycling [53]. However, in the year of grassland removal, high N surpluses and annual N 2 O emissions were measured in the following crop of winter wheat.
Agriculture has also been identified as a major anthropogenic source of CH 4 due to microbial activity in ruminant digestion and in livestock slurry, while wetlands are considered as a major natural source of emission [54]. However, under aerobic conditions, soils under agricultural management act as a sink for CH 4 [55,56]. According to Boeckx and Van Cleemput [55], arable soils have a lower mean CH 4 sink function (−1.5 kg CH 4 ha −1 year −1 ) than grassland soils (−2.5 kg CH 4 ha −1 year −1 ). Mineral N fertilizer application can also have a negative effect on the CH 4 oxidation capacity of arable soils [57]. The highest CH 4 oxidation rates have been reported for near-natural forest soils [25,58,59]. In support of this, in our study the beech-forest showed the highest average CH 4 sink capacity. However, our findings did not confirm higher CH 4 uptake of grassland soils compared with the arable systems, and only a minimal higher CH 4 oxidation capacity of the unfertilized organic rotations.

Crop Yields and Yield-Related GHG Emissions
In the conventional system the relatively high but site-typical application rate of mineral fertilizers, combined with appropriate plant protection measures, led to three-fold higher crop yields compared with the organic systems. Such distinct differences in yields have generally not been documented in other studies [35,60]. Lower yields from organic systems are due partly to lower supplies of nutrients, including possible deficits of phosphorus, potassium, magnesium and micronutrients, and the comparatively lower soil pH values of the organically managed fields (see Supplementary  Table S1). Nutrient deficits are a particular problem in pure arable organic systems without animal husbandry and external N supplies, leading to negative N balance levels and decreasing soil N stocks.
In the context of increasing demand for organic farming products and addressing the environmental problems associated with agriculture, including the challenges of climate change, there is a need to also consider emissions per unit of product when conventional and organic farming systems are compared [61]. This, and the concept of sustainable intensification in agriculture, are today often considered with the twin aims of maximizing the production of agricultural goods and minimizing the negative impacts on the environment [62]. In the present study, we considered the product unit (grain equivalents) as an appropriate measure relevant at a global environmental scale, and here we found that the three systems showed comparable values. Similar results were reported by Flessa et al. [63], where organic farming did not result in different emissions per unit of crop yield. Controversially, in a meta-analysis, Skinner et al. [23] found 29% higher yield-related N 2 O emissions for arable organic systems. Also, Chirindaa et al. [64] and Knudsen et al. [65] showed higher product-related N 2 O emissions for organic arable systems on average for all investigated crop rotations. However, comparisons in the literature between organic and conventional arable production depend strongly on the types of crops under investigation [29,31,52,64] with different crops affecting yield-related emissions differently. Further, in the present study, we also included non-food crops in the analysis, in this case red clover as fodder legumes, and, as they were sold, they have a market value as a product. Beyond the measured field-level emissions, the different systems should be further evaluated in terms of a full lifecycle assessment (LCA) analysis that also includes any emissions produced from external resources [65]. With regard to our field measurement results and the expected (though unmeasured) higher energy use in the conventional system, we assume that the tested organic systems would show further net benefit, relative to intensive conventional systems, under the perspective of an LCA analysis. However, other impact categories like eutrophication, acidification and cumulative energy use should be taken into account in order to avoid negative implication on other environmental resources. In our study, associated research groups from the project group Hof Ritzerau identified significant benefits from the conversion to organic farming over a ten-year period, e.g., regarding soil biota and birds [66,67]. Thus, when modelling options for optimized land use systems in terms of sustainable intensification on high fertility arable soils, an optimal share of the area under organic farming has to be predicted in order to ensure long-term ecosystem functioning in landscapes. Finally, the economic consequences for producers, and for wider society and its welfare, also need to be considered in relation to any large-scale land-use change, either to organic farming or other extensive land-use systems like grassland and forestry.

Further Mitigation Potential
An important finding of this study was the increased N 2 O emissions after soil tillage in arable systems. This was mainly controlled by the chemical composition of the crop residues, particularly when narrow C/N ratios were present, and was clearly shown over a wide range of arable crops and use intensities. A serious weakness in the conventional as well as the two organic systems was the inefficient use of nutrients in crop residues, as well as the failure not to take mineralizable N into account for fertilizer planning and to ensure optimal timing of ploughing in the grass-legume ley phase. Thus, further mitigation could be achieved through reduced input of mineral N fertilizer in the conventional systems. If organic fertilizers are applied, as manure or digestates, their N contents should be also taken into account for optimal N management strategies to avoid highly positive N surpluses. With regards to the replacement of the grass-clover ley in the rotation, one efficient measure would be to postpone ploughing until spring to reduce emissions, as high N 2 O fluxes occur during winter at locations where freeze-thaw cycles are expected [30]. Generally, there is a need to reduce N 2 O emissions in organic farming systems after the incorporation of N rich crop residues [31] such as those which occur after the replacement of a grass-clover ley. Organic arable farms especially should seek opportunities to enter into cooperation agreements with livestock farms to exchange herbage harvested from grass-clover crops (as silage bales) for livestock manure. This would not only support increased yields but also help maintain soil fertility. Another option is to remove the mown grass-clover biomass for biogas production and apply the digestate to the soil the following spring. Möller et al. [68] concluded that this option leads to additional energy yields, a lower risk of nitrate leaching, and lower N 2 O emissions compared with swards that are just mown and mulched. Manure application has been shown to be one of the main factors in arable organic farming needed to generate competitive crop yields. Nevertheless, the use of organic fertilizer can cause additional N 2 O emissions [65], and, thus, has to be further evaluated.
Finally, the aim of all arable systems should be to maintain or increase crop yields without causing additional environmental problems [69,70]. In this context, the key factor in organic farming is an efficient method of converting the grass-clover ley phase for a subsequent crop, without significant N losses, and efficiently managing the use of legumes in the crop rotation. In our study, the use of 25% of legumes in the arable crop rotation was not enough to achieve adequate yield to compete with the arable conventional systems in terms of emissions per product unit. This was only given with a share of 40% legumes in the crop rotation.

Conclusions
Organically managed arable rotations showed advantages in comparison with intensive conventional arable cropping through reduced GHG emissions and N surplus expressed on a per-ha basis. Hence, an increase in the area of organic farming on farmland with arable rotation systems is a suitable measure for reducing GHG emissions and nutrient loads at a national or landscape level. On the global scale, however, emissions have to be considered also on a per-unit of product basis, as climate is a global environmental good. Taking that into account, and the strongly limited productivity of organic all arable farming as well, it becomes evident that improved strategies beyond specialized conventional and organic approaches focusing on mixed farming systems with enhanced eco-efficiency are needed to ensure both global food demand and ecosystem services.