Rainfall and crop production
Rainfall at the site Reinshof was slightly over average during the period from 1 April to 1 July in 2014 (201 mm) and considerably below average during the period from 1 April to 1 July in 2015 (100 mm). Therefore, the 2014 spring period was much wetter than spring 2015. Overall, 2014–2015 winter period was quite mild with no significant freeze-thaw event.
Soil mineral N
The time course of the soil mineral-N concentrations are shown in Fig. 2. In soils planted with arable crops, soil NH4
+ concentrations during the investigation period remained rather low (below 10 kg N ha−1 at 30 cm layer) in non-fertilized treatments. The application of mineral-N fertilizer in NWT treatment caused a slight increase in topsoil NH4
+ concentrations for a short time period which decreased rapidly to the background concentrations within a week. The low concentration of soil NH4
+ even in N-fertilized treatments (in the form of calcium-ammonium-nitrate) can be attributed to a rapid nitrification as soil texture and pH (6.7) serve ideal conditions for nitrification. In all treatments, there was a significant increase in topsoil NH4
+ concentrations in spring 2015 regardless from the N application. Early spring period in 2015 was reasonably dry compared to the same period in 2014. Therefore, the latter can be attributed to the processes related to the soil wetting after a long dry period (such as in spring 2015) which can accelerate N release from the mineralization of soil organic matter immediately after rewetting [32].
Overall treatments, soil NO3
− concentrations in 0–15 cm soil segment varied between 10 and 85 kg NO3
−–N ha−1. Here, NO3
− was generally the dominant soil N form and highly variable when sampled soon after additions of fertilizer N. In all soils, concentrations of NO3
− in the 0–15-cm layer decreased over time with the largest decrease found in the arable crops specifically in WT and NWT treatments. During the vegetation period in 2014, soil NO3
− content was generally higher in agro-forestry soils than in arable soils. Plant nutrient and water uptake was expected to be higher in cropland compared to the young agro-forestry treatments in 2014 due to small size and low growth rate of young trees. Thus, more rapid depletion of soil mineral N in arable crops than agro-forestry treatments can mainly be attributed to the differences in plant N uptake.
In arable land stand, soil NO3
− concentrations were clearly higher (significant in 2014, p < 0.01) in FB than in other non-fertilized treatments during the vegetation period (Fig. 2a, b). For legume crops, inputs of biologically fixed N largely supplement to the uptake of soil mineral N to meet crop N demand. Thus, the legume species also take up mineral N from soils for growth before fixing additional N. The preferential use of soil mineral N helps explain why there is also significant depletion of soil NO3
− in FB treatment [33]. In a review study, authors reported that average 41 % (for chickpea), 65 % (for faba bean), and 66 % (for field peas) of N that were present in legumes were derived from soil N [33]. However, slightly higher soil NO3
− concentration in FB treatments than non-legume soils suggests that there were still reasonably more NO3
− available for potential denitrification losses during the legume-growing season.
Seasonal N2O emissions
In both years, flux data indicate that N2O emissions were dominated by specific event periods (Fig. 3). Overall, maximum daily emissions of N2O in the early summer period in 2014 were 0.16 ± 0.07 and 0.04 ± 0.01 kg N2O–N ha−1 day−1 in agro-forestry and arable land treatments, respectively. In a 2-year field study, Lebender et al. [12] observed similar flux rates over a nearby site with similar soil conditions and agricultural practices (wheat and spring barley). Maximum N2O emissions measured in agro-forestry treatments (0.16 ± 0.07 kg N2O–N ha−1 day−1) have been usually observed in young agro-forest ecosystems [7, 27]. Almost all significant N2O fluxes occurred as daily peak N2O emissions and were measured only during the early summer period in 2014. The importance of these peak emissions in early summer period on the annual budget of N2O emissions highlights the necessity of continuous flux monitoring to accurately determine the N loss from agro-ecosystems specifically in spring and early summer seasons [7, 20].
In 2014, N2O emissions gradually decreased to the background levels (below 10 g N2O–N ha−1 day−1) from the months of May to July and remained in background levels until March 2015 (Fig. 3). Interestingly, the latter was less than 0.01 kg N2O–N ha−1 day−1 during the early summer period in 2015 for all treatments (including N-fertilized treatments). As seen in Fig. 1, early summer period in 2014 was relatively wet (April–June, 201 mm rainfall) compared to the same period in 2015 (April–June, only 100 mm rainfall). Therefore, we may attribute higher daily N2O fluxes in 2014 than in 2015 (in early summer period) to the differences in mineral N and moisture content of the soil. Mineral N content of all soils in June 2014 was almost similar as compared to the same period in 2015, whereas N2O fluxes were still about 10-fold higher in June 2014 than in June 2015. In this context, we may conclude that soil moisture seems to be the major driving factor of higher N2O emissions in the 2014 summer period than in 2015.
All the abovementioned factors (e.g., high moisture, high soil temperature in early summer, and moderate or high NO3
− content of the soil) are known to trigger specifically the denitrification rate in soils. Thus, we may speculate that denitrification (fungal or bacterial) was the potential key source of measured large N2O fluxes in the 2014 early summer period. Our earlier report supports this hypothesis in which sandy loam soil was incubated under laboratory conditions, and similar to the field experiment, large N2O peak events were observed immediately after rewetting of the soil. Here, a stable isotope-labeling study clearly showed that denitrification was the major source (over 90 % of emitted N2O) of large N2O peaks that occurred in wet seasons [8, 19]. Furthermore, there is now a growing evidence that fungal denitrification may be the key process producing N2O in such situations rather than bacterial denitrification [17]. Surely, more research is needed to reveal (i) the dominant processes and (ii) key microbial or fungal strains producing N2O under these specific conditions (especially during early summer period).
Effect of plant species on N2O emissions
Mean cumulative N2O fluxes during the vegetation period in 2014 were 152 ± 58, 217 ± 29, and 441 ± 10 g N2O–N ha−1 in WT, WFB, and FB treatments, respectively. Among the non-fertilized arable crops (wheat, wheat mixed intercropped with bean, and bean), N2O emission over the 2014 growing seasons is highest in soils when faba bean (FB) was grown (Fig. 4). Introducing N-fixing legumes into cereal-based crop rotations may reduce synthetic mineral-N fertilizer use and thought to mitigate N2O fluxes. However, the present study clearly showed that when faba bean was grown as a mono-crop, N2O fluxes were about threefold higher compared to WT treatment. In contrast to the present study, authors reported that growing season N2O emissions from N2-fixing legumes are significantly lower than from non-legumes and are often comparable to unfertilized background emissions [26, 34]. In line with the present study, Rochette and Janzen [26] (in a review study) concluded that legumes can produce substantial N2O emissions. They speculated that the main source of N2O emissions from soils planted with N2-fixing legumes during the vegetation period may be attributable to the N release from root exudates and/or from the decomposition of dead root residues. An alternative process that may contribute to the latter would be the N2O emission during the N2-fixation process in the nodules where N2 is fixed. Authors reported that several Rhizobium species in the free-living forms or in legume roots can denitrify NO3
− and release N2O from active nodules most likely to prevent excess NO3
− that inhibits the activity of N2-fixing enzymes [35].
In agro-forestry treatments, both daily and cumulative N2O emissions did not differ among each other in the 2014 growing period. Here, mean seasonal N2O emissions (during the growing season of arable crops) were 1121 ± 161, 1102 ± 159, and 1052 ± 266 g N2O–N ha−1 in PL, PRB, and RB treatments (no significant difference), respectively. The cumulative mean N2O fluxes during the 2014 growth period were considerably higher in agro-forestry than in arable crop treatments. During the first year, young plantations in agro-forestry domains generally have limited N and water uptake, while wheat and faba bean as arable crops are at their most productive growth stage specifically during May and June (growth rates are almost at their maximum during this period). Here, soil conditions seem to be more favorable specifically for denitrification in agro-forestry treatments than in soils planted with arable crops that may explain large N2O emissions [27]. The cumulative mean N2O emissions in agro-forestry treatments were about fivefold higher than in both WT and WFB treatments, and the latter was still more than twofold higher when compared to FB plots. In line with the present data set, authors reported large N2O fluxes after conversion of pastures lands [7] or grasslands [36]. Here, authors attributed large N2O fluxes to the soil disturbance associated with tillage and cultivation that accelerate soil organic matter decomposition and microbial activity (nitrification and denitrification) leading to N2O emissions. In the present experiment, soil tillage has been done almost at the same time for all treatments. Thus, we may speculate that the difference in N2O emissions when comparing arable land to agro-forestry was mainly due to differences in water and nutrient uptake of plant species. Water and nutrient demand of young plantations during the first year are generally low which in return may cause more favorable conditions specifically for denitrification and N2O losses from denitrification. Overall, we can therefore summarize that direct plant effect seems to be one of the key variables that regulates N2O losses from soils. Zona et al. (2013) concluded that vegetation uptake of NO3
− together with water ultimately may reduce the anaerobic volume of soils and may lower both denitrification rate and product stoichiometry of denitrification (lower N2O/N2O + N2 ratio; meaning reduced N2O and enhanced N2 production) in agricultural soils [8, 17].
Effect of mineral N supply on N2O emissions
Although it was not the main goal of the present experiment to study the effect of mineral-N addition on N2O fluxes, we added fertilizer N to mono-crop wheat (80 kg N ha−1; calcium-ammonium-nitrate) in parallel plots to be able to compare N2O fluxes from soils planted with N2-fixing plants (faba bean mono-culture or faba bean intercropped with wheat) with fertilized and non-fertilized wheat soils. N fertilization during the first year of new agro-forestry plantations is also not a common practice. However, N doses similar to the arable treatments were applied at the same date in order to be able to gain better scientific knowledge about the dominant factors regulating N2O fluxes in agro-forest ecosystems. Expectedly, in all N-fertilizer treatments, N2O fluxes increased immediately after fertilizer application, however, only in 2014 (wet early summer) but no response observed in 2015 (dry early summer). The latter clearly suggests that environmental factors specifically soil moisture was the most dominant factor in 2014 that leads to relatively high N2O fluxes and without sufficient moisture or rainfall, fertilizer application alone does not affect N2O fluxes significantly, e.g., in 2015. In line with the present study, authors also reported that the application of organic or inorganic fertilizers affects N2O fluxes only in wet seasons but not in dry years [10, 17, 19].
Overall, the application of nitrogenous fertilizer affected N2O fluxes predominantly in cropland soils and had limited impact in agro-forestry soils. Here, cumulative N2O fluxes (from May to September 2014) were 46 and 121 % higher in fertilized than in non-fertilized agro-forestry (non-fertilized poplar vs. fertilized poplar) or in cropland treatments (non-fertilized wheat vs. fertilized wheat), respectively. The latter clearly suggests that N2O fluxes were more dependent on soil mineral-N content in arable crops than in agro-forestry most likely due to greater competition between plants and N2O-producing soil microorganisms in cropland than in agro-forestry soils. Fertilizer-derived cumulative N2O emissions (emission factor) during the period from April to December 2014 were 0.21 and 0.45 % of applied N in wheat and poplar soils, respectively. Measured emission factors were significantly lower than what IPCC predicts (1 %; [4]). However, the latter was similar to what we reported in our previous study for central and northern Germany ([12, 19]). For operational reasons in the present IPCC protocol, the N2O emission factor was set to 1 % for all fertilizer N regardless of crop or soil type. Large variations in N2O emissions from different agricultural systems due to differences in management, climate, and soil type are very well known. Low N2O emission factors in the present experiment and in our previous reports suggest that in the future, different emission factors should be considered at least for different crops or regions that account for their different risks of N2O emissions.
Cumulative mean N2O emission during the growing season was still 31 % higher (p < 0.05) in FB than in NWT treatment. In a review study, Rochette and Jansen [26] summarized that legumes can increase N2O emissions during growth compared to evenly fertilized arable crops most likely due to the N release from the root exudates and decomposition of crop residues. Our study clearly agrees with Rochette and Jansen [26] and many others (e.g., [7, 27]) that growing legumes as mono-crop can increase N2O fluxes compared to N-fertilized arable crops. On the other hand, seasonal N2O fluxes were 35 % lower in WFB (wheat mixed intercropped with faba bean) than in NWT (wheat as mono-crop) treatment. The latter suggests that using legume crops as intercrop or mixed crop in wheat may significantly mitigate fertilizer-derived N2O fluxes. However, surely more research is needed to upscale current findings due to the complexity and variability of N2O fluxes in complex agricultural systems, e.g., mixed cropping systems.