The biodiversity - N cycle relationship: a 15N tracer experiment with soil from plant mixtures of varying diversity to model N pool sizes and transformation rates

Abstract

We conducted a 15N tracer experiment in laboratory microcosms with field-fresh soil samples from a biodiversity experiment to evaluate the relationship between grassland biodiversity and N cycling. To embrace the complexity of the N cycle, we determined N exchange between five soil N pools (labile and recalcitrant organic N, dissolved NH4+ and NO3 in soil solution, and exchangeable NH4+) and eight N transformations (gross N mineralization from labile and recalcitrant organic N, NH4+ immobilization into labile and recalcitrant organic N, autotrophic nitrification, heterotrophic nitrification, NO3 immobilization, adsorption of NH4+) expected in aerobic soils with the help of the N-cycle model Ntrace. We used grassland soil of the Jena Experiment, which includes plant mixtures with 1 to 60 species and 1 to 4 functional groups (legumes, grasses, tall herbs, small herbs). The 19 soil samples of one block of the Jena Experiment were labeled with either 15NH4+ or 15NO3- or both. In the presence of legumes, gross N mineralization and autotrophic nitrification increased significantly because of higher soil N concentrations in legume-containing plots and high microbial activity. Similarly, the presence of grasses significantly increased the soil NH4+ pool, gross N mineralization, and NH4+ immobilization, likely because of enhanced microbial biomass and activity by providing large amounts of rhizodeposits through their dense root systems. In our experiment, previously reported plant species richness effects on the N cycle, observed in a larger-scale field experiment within the Jena Experiment, were not seen. However, specific plant functional groups had a significant positive impact on the N cycling in the incubated soil samples.

Introduction

Anthropogenic activities have resulted in the loss of biodiversity, which can alter ecosystem functions including biomass productivity, organic matter decomposition rates, and nutrient cycling (Loreau et al. 2001; Hooper et al. 2005; Weisser et al. 2017). Nitrogen (N) is the quantitatively most important nutrient limiting primary productivity in many ecosystems (Elser et al. 2007; Fay et al. 2015). Therefore, knowledge of biodiversity-N cycle relationships is necessary to understand the consequences of biodiversity loss for the N supply of plants and N leaking into the atmosphere and surface and groundwaters. Nitrogen undergoes complex microbially mediated transformations in soil that are related to the quantity and quality of soil organic matter (Wedin and Pastor 1993; Benbi and Richter 2002; Booth et al. 2005; Fornara et al. 2011; Lang et al. 2015). The quality and quantity of organic matter in grassland soils largely depend on the plant species and functional group richness responsible for differences in litterfall, root turnover, and root exudates (Allan et al. 2013; Solly et al. 2013).

The N transformation processes that are most important for plants and microorganisms are those associated with the depolymerization of organic N into amino acids and mineralization-immobilization turnover of ammonium (NH4+) and nitrate (NO3), because these N species represent the major forms of bioavailable N taken up by plants and microorganisms (Davidson et al. 1990; Corre et al. 2002; Schimel and Bennett 2004; Zhang et al. 2016). Depolymerization of organic matter is the process by which proteins in organic matter are broken down into smaller, N-containing fragments, the amino acids which thereby become accessible for plants and microorganisms (Schimel and Bennett 2004; Wild et al. 2015). Gross N mineralization includes the release of amino groups as NH4+ which can also serve as a substrate for nitrification. By the mechanism of N immobilization, the mineral N is assimilated by microorganisms, which compete with plants for N. Nitrogen mineralization and nitrification rates are primarily controlled by soil microbial activity, as well as environmental factors, such as the availability, quality, and quantity of the microbial C source and mineral nutrients, soil moisture, and temperature (Booth et al. 2005). For grasslands, previous work has suggested that the nitrification to microbial immobilization ratio is an important factor controlling NO3 leaching (Stockdale et al. 2002). With regard to controls of the availability of NH4+ and subsequent nitrification in an ecosystem, immobilization of NH4+, and fixation and release of NH4+ by specific clay minerals (illites and interlayer minerals containing illite layers) may also play an important role (Brady and Weil 2002).

Most studies on the biodiversity-mineralization relationship have focused on net N mineralization and/or nitrification rates (Accoe et al. 2004; Fornara and Tilman 2009; Fornara et al. 2011; Rosenkranz et al. 2012; Mueller et al. 2013). However, net rates alone do not provide a process-based understanding of the N cycle (Hart et al. 1994; Verchot et al. 2002; Cheng et al. 2013), which requires the assessment of simultaneously occurring gross N transformations (Hatch et al. 2000; Paterson 2003; Bedard-Haughn et al. 2006; Müller et al. 2007; Cheng et al. 2014). Previous studies reported that increasing species richness increased net N mineralization rates (Rosenkranz et al. 2012; Mueller et al. 2013), as well as net nitrification rates (Scherer-Lorenzen et al. 2003; Mueller et al. 2013). However, there are currently only few studies that have evaluated the relationship between biodiversity and gross N transformation rates, with contrasting results. Zak et al. (2003) and West et al. (2006) reported for nutrient-poor, sandy soils from Minnesota, USA, a positive biodiversity-gross N mineralization relationship in laboratory incubations, whereas Lama et al. (2020) found the opposite relationship in the Jena Experiment based on a 24-h 15N pool dilution approach where the 0–5-cm surface soil layer was labeled with 15NH4Cl in the field to determine the rates of gross N mineralization, microbial assimilation of NH4+, and gross inorganic N immobilization at 76 plots with varying plant mixtures. Lama et al. (2020) attributed their finding to the mechanisms that increase the N-use efficiencies of plants with increasing plant species richness, which slowed down the N cycle, mainly because of increased C/N ratios of the roots.

One possible approach to simultaneously assess co-occurring transformation rates in soil involves the use of 15N-labeled substrates. Müller et al. (2007) developed a 15N tracing model (Ntrace) to quantify gross N transformations in soils. The model integrates pathways of N mineralization and immobilization of NH4+ and NO3 into labile and recalcitrant organic pools, nitrification of NH4+ to NO3 and from organic N to NO3, dissimilatory nitrate reduction to ammonium (DNRA) (under anaerobic conditions), and cation exchange (i.e., ad- and desorption) of NH4+ on clay minerals (Müller et al. 2007). Moreover, the model simulates the pool sizes of labile and recalcitrant organic N, NH4+, and NO3, and adsorbed NH4+ (Müller et al. 2007). The objective of this study was to apply the Ntrace model to data obtained from laboratory incubations of field-fresh soil from the Jena Experiment without plants to evaluate the legacy effects of plant community composition (species richness, functional group richness, presence and absence of four functional groups—legumes, grasses, tall herbs, and small herbs) on the N pool size and gross N transformation rates in grassland soils. In line with previous applications of Ntrace, we incubated soil without plants so that the plant diversity effect originates from the previous plant effects on the microbial community. In the Jena Experiment, it has been shown that the different mixtures of root deposits into the soil released by the differently diverse plant communities and the effects of the plant community composition on abiotic conditions including soil moisture and nutrient availability shape the microbial community composition (Lange et al. 2014; Dassen et al. 2017; Weisser et al. 2017). A better understanding of the relationship between biodiversity and the complex N cycle will improve our prediction of possible biogeochemical consequences arising from the expected loss of biodiversity and changing plant community composition. This includes possible changes in the N availability for plant growth and increasing N leaking in gaseous form to the atmosphere or as NO3- to surface and groundwaters with their known detrimental effects on climate and water quality (Sutton et al. 2011).

Materials and methods

Study site

Our study contributed to the Jena Experiment (www.the-jena-experiment.de), a long-term grassland biodiversity experiment established in 2002 (Roscher et al. 2004; Weisser et al. 2017). The site had been used as arable land for at least 40 years before the initiation of the Jena Experiment. The experimental site is located on the floodplain of the river Saale in Jena, Germany (50° 55′ N, 11° 35′ E; 130 m above sea level). The mean annual air temperature at the site is 9.9 °C, and the mean annual precipitation amounts to 610 mm (Hoffmann et al. 2014). The soil at the site is classified as Eutric Fluvisol developed from 2-m-thick loamy fluvial sediments (IUSS Working Group WRB 2014). The soil texture ranges from sandy loam close to the river to silty loam with increasing distance from the river. The experimental site is mown twice mimicking the locally common land use as a low-intensity hay meadow and weeded three times per year to maintain the designed diversity levels. The major aim of its establishment was to explore the effect of biodiversity on nutrient cycling and trophic interactions (Roscher et al. 2004).

The detailed description of the experimental design can be found in Roscher et al. (2004), and major results are reviewed in Weisser et al. (2017). The main field experiment consists of 82 plots in four blocks to account for the systematic change in soil texture perpendicular to the river, with a factorial design of different levels of plant species richness (SR 1, 2, 4, 8, 16, and 60) and 1 to 4 functional groups (grasses, legumes, small herbs, and tall herbs). The mixtures were randomly drawn from a pool of 60 species representing a typical Central European mesophilic grassland. Each level of species richness was replicated on 16 plots, except for the 16 and 60 species richness levels, which are replicated only on 14 and 4 plots, respectively. Only block 2 (n = 19 plots) was considered for this study, which included all the levels of plant species richness from 1–16 species in fourfold replication, except for the 16-species mixture, for which only three replicates existed.

Because the characterization of the soil microbial community composition might help in the interpretation of our results but was beyond the scope of our study, we refer to two studies from the same soils of the Jena Experiment (Lange et al. 2014; Dassen et al. 2017). Lange et al. (2014) reported for the year 2007 based on phospholipid fatty acid patterns that the fungal-to-bacterial biomass ratio was positively affected by plant functional group richness and negatively by the presence of legumes. Bacteria were more closely related to abiotic differences caused by plant diversity such as soil moisture, while fungi were more affected by plant-derived organic matter inputs defined by the composition of functional groups. Dassen et al. (2017) determined the composition of the fungi, bacteria, archaea, and protists community in the year 2010 based on 454-pyrosequencing. They found 4025 bacterial, 23 archaeal, and 826 unclassified OTUs based on the amplification 16S rRNA gene fragments and 431 fungi, 174 protists, 9 plants, and 374 unclassified OTUs based on the amplification of eukaryotic 18S rRNA fragments. The most dominant taxonomic group of bacteria was the Chloroflexi. The most diverse bacterial groups were Proteobacteria and Planctomycetes. A total of 19 putative rhizobial OTUs were recovered across the experimental fields. The most dominant taxonomic group of eukaryotes was Ascomycota, which was also the most diverse fungal group. In total, 19 arbuscular mycorrhiza fungi (AMF) OTUs (phylum Glomeromycota) were recovered across all plant communities. Of the main protist supergroups, Rhizaria were well represented. Although protists represent a relatively small proportion (< 2%) of the total eukaryotic community, their diversity was considerable, with 174 detected OTUs. The main findings with respect to the relationship between plant community composition and soil organisms were that plant and functional group richness had little influence on the soil microbial community composition, which was more driven by the presence of legumes and by the small-scale abiotic variation at the field site (Dassen et al. 2017).

15N tracing experiment and sample analysis

To assess the importance of NH4+ fixation by clay minerals such as illites in the study soils, we conducted a sorption experiment in the context of our field 15N tracer experiment reported in Lama et al. (2020). We added 25 μg N (98 at% 15N) as NH4Cl to a 100-cm3 stainless steel core inserted in the 0–5-cm soil layer and determined the recovery of the applied NH4+ by extraction with 1 M KCl 15 min after the application. Our mean recovery (± standard deviation) was 98 ± 1.4%, from which we infer that NH4+ fixation is negligible in our study soils.

Soil samples were collected from Block 2 of the experimental site in October 2014, i.e., 12 years after the establishment of the vegetation mixtures. Approximately 400 g of field-fresh soil was sampled from each plot by combining 15 soil cores (Ø = 1 cm, depth = 15 cm). The soil samples were sieved (< 2 mm) in the field-fresh state, and from each soil sample, three replicates of 100 g of soil were produced. These field-fresh soil sample replicates were amended with 15N-NH4+ ammonium, 15N-NO3-, or both (98 at%), applied as 0.5 μg 15NH4Cl-N and 0.25 μg K15NO3-N (g dry soil)-1. After the 15N-label addition, samples were mixed thoroughly to ensure a homogeneous 15N distribution and placed in incubation vessels with a ceramic filter (pore diameter of 0.4 μm). Above and below the soil samples, glass wool was inserted to prevent dispersion during rinsing. Finally, all the incubation vessels containing the soil samples were sealed with rubber stoppers and incubated for 16 days in the dark at a constant temperature of 20 ± 1 °C. To maintain aerobic conditions inside the incubation vessels, the soil samples were aerated by removing the rubber stoppers for 1 h each day. Soil samples were extracted by percolation with 100 mL of a N-free nutrient solution (4 mM CaCl2, 2 mM KH2PO4, 1 mM MgCl2, 1 mM K2SO4, 1 mM MgSO4, 25 μM H3BO3, 2 μM MnSO4, 2 μM ZnSO4, 0.5 μM CuSO4, and 0.5 μM Na2MoO4; Nadelhoffer 1990) 12 h and 2, 4, 9, and 16 days after the 15N application. The nutrient concentrations were adjusted to optimize conditions for soil microorganisms (Nadelhoffer 1990). Percolation leaching was facilitated by applying a vacuum of 20 kPa for half an hour. We conducted the experiment with field-fresh soils and determined the initial water content, which ranged from 26.7 to 31.4 vol%, by drying subsamples at 105 °C. We maintained the initial water content by always carefully extracting the full volume of the added nutrient solution with the help of a vacuum. Because the microcosms were closed except during the short openings for aeration, we assumed that the evaporation loss was negligible. Consequently, the water content should have remained stable. To reduce the analytical load, samples of days 4 and 9 were combined to yield a single composite sample.

The pH of the soil solutions was measured with a glass electrode (Orion U402-S7, Thermo Fisher Scientific, Waltham, MA, USA). Concentrations of NH4-N and NO3-N in the extracts were measured colorimetrically with a continuous flow analyzer (SAN++, Skalar, Breda, The Netherlands). The 15N isotopic composition of NO3 was determined with the bacterial denitrification method, in which Pseudomonas aureofaciens is used to convert NO3 to N2O, followed by isotope analysis (Sigman et al. 2001; McIlvin and Casciotti 2011). The isotope ratios of the N2O gas were analyzed with a Gas-Bench II pre-concentration unit interfaced with the Delta V Plus isotope ratio mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) at the Basel Stable Isotope and Biogeochemistry Laboratory, University of Basel. The N isotope ratios in NH4+ were determined using the hypobromite–azide method, in which NH4+ is first converted to NO2, and further to N2O by reduction with azide (Zhang et al. 2007). The N2O is then purified and analyzed as described above for NO3-derived N2O.

To determine the concentrations of total N (TN), aliquots of the soil samples were dried and sieved (2-mm mesh), and the dried samples were then ground using a ball mill. TN concentrations were determined with an elemental analyzer (Elementaranalysator vario Max CN, Elementar Analysensysteme GmbH, Hanau, Germany).

Microbial respiration was measured using an electrolytic O2 micro-compensation apparatus (Scheu 1992). O2 consumption of soil microorganisms in 5 g of fresh soil was measured at 22 °C over a period of 24 h. Basal respiration (μl O2 [g dry soil]-1 h−1) was calculated as the mean of the O2 consumption rates determined between 14 and 24 h after the start of the measurements. The measurement only started after 14 h, because initially, the O2 consumption showed strong variations which are caused by the soil disturbance and only after 14 h, the respiration rates stabilized.

The microbial C/N ratio was determined from the data of microbial biomass C and N, which was measured using chloroform fumigation extraction (Brookes and Landman 1985). Two samples of 7 g soil were taken from each plot, one was fumigated with chloroform vapor for 24 h, and the other was not fumigated. Both the fumigated and non-fumigated samples were extracted with 40 mL 0.5-M K2SO4 by shaking for 30 min. Total C and N concentrations in the extracts were analyzed by dry combustion in a DIMA-TOC 100 Analyzer (Dimatec, Essen, Germany). Microbial biomass C was calculated as (total C in fumigated soil – total C in non-fumigated soil)/0.45 (Wu et al. 1990). Likewise, microbial biomass N was calculated as (total N in fumigated soil – total N in non-fumigated soil)/0.54 (Brookes and Landman 1985). The microbial C/N data were only available for the year 2008. However, Strecker et al. (2016) showed that both, the basal respiration and the microbial biomass C had similar sizes and similar significant relationships with plant species richness in 2008 and 2014. Thus, the size and activity of the microbial community were similar in 2008 and 2014, from which we inferred that it is likely that this is also true for the microbial C/N ratio.

Quantification of N pools and gross transformation rates

The initial pool size of the exchangeable (= adsorbed) NH4+ pool, which represents the NH4+ retention by the soil shortly after the addition of the 15NH4+, was calculated as the difference between applied NH4+ and initial dissolved NH4+ (on day 0). Because the first measurement of dissolved NH4+ only occurred after 12 h, we inferred the initial dissolved NH4+ concentration by back-extrapolation of those measured on days 1 and 2 (Müller et al. 2004). The start values of the exchangeable (= adsorbed) NH4+ pool ranged 0.174–0.180 μg N (g soil)−1 (mean, 0.177 ± standard error 0.0002 μg N (g soil)−1). The initial pool size of soil organic N was calculated from the difference between the concentrations of TN and the sum of 1 M KCl-extractable N (NH4+-N and NO3--N, see Oelmann et al. 2011 for a detailed description of the 1 M KCl extract). Soil organic N was divided into two pools, labile organic N (Nlab) and recalcitrant organic N (Nrec). In the absence of measured start values of labile and recalcitrant organic N concentrations, we used the model default values of 1% labile and 99% recalcitrant N as start values in line with previous studies in which the same model (Ntrace) was applied (Müller et al. 2004, 2007; Huygens et al. 2007). The estimate of 1% labile organic N is based on a study of Causarano et al. (2008). The start values of Nlab ranged 21.0–31.0 μg N (g soil)−1 (26.4 ± 0.71 μg N (g soil)−1) and of Nrec 2080–3065 μg N (g soil)−1 (2610 ± 70.7 μg N (g soil)−1). The changes of the pool sizes of exchangeable (= adsorbed) NH4+ and the two organic N pools during our 16-day incubation experiment were minor, and therefore, we only evaluated the influence of plant community composition on the start values of these pools.

We determined eight gross N transformation rates by integrating the experimental data (i.e., pool sizes and 15N enrichment in various N pools with time) in the Ntrace model (Müller et al. 2007; Fig. 1). The measured NH4+ and NO3 concentrations and 15N enrichment values were supplied to the model and gross N transformation rates were calculated using zero-order or first-order kinetics. The best fit between modeled and observed data was determined based on the Akaike information criterion (AIC) by stepwise modification of the parameters included in the optimization routine and their respective kinetic settings (Table 1). Based on the kinetic settings and the final parameters, gross N transformation rates were calculated by integrating the rates over the 16-day period divided by the total time. The Ntrace model was programmed in the software MatLab 7.9 (The MathWorks Inc., Natick, MA, USA) and the 15N tracing model, that was separately set up, in Simulink 7.4 (The MathWorks Inc.).

Fig. 1
figure1

Schematic representation of the 15N tracing model (Müller et al. 2007). It includes five N pools: dissolved ammonium (NH4+), dissolved nitrate (NO3), labile soil organic N (Nlab), recalcitrant organic N (Nrec), and exchangeable (termed “adsorbed” in previous work; Müller et al. 2004, 2007; Huygens et al. 2007) NH4+ (NH4+ads) and ten gross N transformation rates: mineralization of recalcitrant organic N (MNrec), mineralization of labile organic N (MNlab), immobilization of NH4+ to recalcitrant organic N (INH4-Nrec), immobilization of NH4+ to labile organic N (INH4-Nlab), oxidation of recalcitrant organic N to NO3 (ONrec), oxidation of NH4+ to NO3 (ONH4), immobilization of NO3 to recalcitrant organic N (INO3), dissimilatory NO3 reduced to NH4+ (DNO3, which was negligible in our study), adsorption of NH4+ into the exchangeable NH4+ pool (ANH4), release of exchangeable (adsorbed) NH4+ to NH4+ (RNH4a)

Table 1 Description of model parameters and average gross N transformation rates (mean and standard deviation) estimated with the Ntrace model reported for each plant species richness (SR) level: 1, 2, 4, 8, and 16 (SR1, SR2, SR4, SR8, and SR16) in a grassland soil

Total mineralization rates were calculated by summing up mineralization rates from both, the labile and recalcitrant organic N pools (MNlab + MNrec). Total NH4+ immobilization rates were calculated by summing up NH4+ immobilization rates from both NH4+ immobilization rates (INH4-Nlab + INH4-Nrec). Total nitrification rates were calculated by summing up the rate of NH4+ oxidation and organic N oxidation (ONH4 + ONrec). Since the dissimilatory nitrate reduction to ammonium (DNRA, DNO3) and the desorption of NH4+ (RNH4a) were negligible in our experiment at the given conditions, we excluded these two transformation rates from further data analysis. All N transformation rates and N pools were additionally normalized to the TN concentration of the soil solid phase.

Statistical analyses

We used repeated measures and sequential ANOVA (type I sum of squares) to inspect the effects of plant species richness, functional group richness, and presence/absence of each functional group on the two dissolved mineral N pools of different days, the initial exchangeable (= adsorbed) NH4+, and the total organic N pools and for the eight different gross N transformations. Lilliefors normality test and histograms were used to check for the normal distribution of residuals. The residuals vs. fitted and Q-Q plots were also used to check the assumption of homoscedasticity and normality of the residuals. NH4+ and NO3 pools were log-transformed; MNlab and ONH4 were square root-transformed; and INH4-Nlab was log-transformed to improve the normal distribution of the residuals. The ANOVA was performed with plant species richness and presence/absence of each functional group as explanatory variables to analyze the effect of plant species richness and presence/absence of each functional group on mineral N pools and gross N transformations. The functional groups were fitted in the following order: legumes, grasses, tall herbs, and small herbs. Because we assumed that legumes have the strongest effect on the N cycle as a consequence of their N2-fixing ability, we fitted legumes before other functional groups. Grasses also significantly impacted N transformations, while small herbs were shown to have the least or no effect (Oelmann et al. 2007; Eisenhauer et al. 2010). The interactions between plant species richness and presence/absence of functional groups were not significant and therefore were not considered in the final models. To avoid the collinearity between functional group richness and each functional group, a separate model was set up to test the effect of functional group richness on N pools and gross N transformation rates. All the statistical analyses were conducted in R studio (R Studio, Version 1.1.456, R Studio Inc., Boston, MA USA) with the free statistical software R 3.5.1 (R Core Team 2016). The type I error rate for all statistical analyses was p < 0.05.

Results

Pool-size changes of dissolved mineral N during the incubation

The N amendments, which contributed less than 6% of the existing mineral N pool at the time of the experiment, did not markedly affect the total amount of mineral N in the soil samples (as extracted with 1 M KCl). About 97–99% of the added 15N enrichments were recovered until the end of the experiment (day 16) in the solutions indicating that there were no or negligible gaseous losses by denitrification and/or ammonia volatilization. The pH of the soil solutions ranged from 7.6 to 8.2.

Both the dissolved NH4+-N and NO3-N concentrations showed parallel temporal courses irrespective of the kind of labeling and the species richness (Fig. 2). Across 15N treatments and plots, the average concentrations of NH4+ declined from 0.30 ± 0.03 μg N (g soil−1) measured on the first day of incubation to 0.07 ± 0.01 μg N (g soil−1) on day 16. In contrast, NO3 concentrations changed only from 1.28 ± 0.14 μg N (g soil−1) on day 1 to 1.18 ± 0.11 μg N (g soil−1) on day 16 of the incubation experiment, respectively. Differences in the pool sizes of both dissolved NH4-N and NO3-N at least between some incubation days were significant as reflected by the significant effect of time (day) on these pools (Table 2).

Fig. 2
figure2

Relationship between plant species richness and inorganic N pools (NH4+, NO3) differentiated according to the labeling of 15N-NH4Cl (a, b), 15N-KNO3 (c, d) and combination of 15N-NH4Cl + 15N-KNO3 (e, f). Data are presented as mean ± standard error (SE). Note the differences in y-axis scaling

Table 2 Results of repeated measure ANOVA showing the effects of plant species richness, functional group richness, and presence (+)/absence (−) of each functional group on ammonium (NH4+) and nitrate (NO3) pools measured on different days of incubation

The interaction between day and species richness had a marginally significant influence on the dissolved NO3 pool (Table 2). Functional group richness did not show significant effects on the dissolved NH4+ and NO3 pool sizes (Table 2). The relationship between the presence of grasses and the dissolved NH4+ pool size was significantly positive (Table 2; Fig. 3). The effect of legumes on the dissolved NH4+ pool changed over time as reflected by the significant interaction between day and presence of legumes (Table 2). The concentration of 15N in dissolved NH4+ and NO3- remained constant (Fig. 4a) or decreased during the incubation (Fig. 4b-f). Interestingly, 15N was deteced in NH4+, when only 15N-NO3- was applied (Fig. 4c).

Fig. 3
figure3

Effect of grasses on the NH4+ pool over different days of incubation. Data are presented as mean ± standard error (SE). Grey and white bars represent presence (+) and absence (−) of grasses respectively. Significance code: double asterisk indicates p < 0.01

The presence of grasses had a marginally significant negative effect on the initial exchangeable (= adsorbed) NH4+ pool (Table S1). Plant species richness increased both, the initial labile and recalcitrant organic N pools (Table S2; Fig. S1).

Gross NH4+ production

Gross N mineralization from labile organic N ranged between 0.01 and 1.94 μg N (g soil)−1 day−1 and from recalcitrant organic N between 0.006 and 1.35 μg N (g soil)−1 day−1 (means and standard deviations [SD] are shown in Table 1). We did not find any significant relationship between plant diversity and N mineralization from recalcitrant organic N. The positive effect of functional group richness on N mineralization from the labile organic N pool was only marginally significant (Table 3; Fig. 5a). The presence of legumes or grasses had a significant positive effect on the N mineralization rate from labile organic N (Table 3; Fig. 5b, c).

Table 3 Sequential ANOVA results showing the effects of plant species richness, functional group richness and presence (+) /absence (-) of each functional group on gross N mineralization from the labile organic N pool (MNlab) and on immobilization of NH4+ into the labile organic N pool (INH4-Nlab)

Gross NO 3 production

Gross heterotrophic and autotrophic nitrification rates ranged from 0.05 to 3.66 μg N (g soil)−1 day−1 and from 0.20 to 3.62 μg N (g soil)−1 day−1, respectively (means and SD in Table 1). Neither plant species richness nor functional group richness (Table 4) significantly affected autotrophic nitrification. The presence of legumes significantly increased autotrophic nitrification rates (Table 4; Fig. 6a). We did not detect any significant effects of plant community composition on heterotrophic nitrification from the recalcitrant organic N pool.

Table 4 Sequential ANOVA results showing the effects of plant species richness, functional group richness, and presence (+)/absence (−) of each functional group on autotrophic nitrification (ONH4)

Gross NH 4 + and NO 3 immobilization

The NH4+ immobilization rates into the labile and recalcitrant organic N pools ranged from 0.05 to 0.55 and from 0.003 to 0.04 μg N (g soil)−1 day−1, respectively and the NO3 immobilization rates ranged from 0.94 to 6.97 μg N (g soil)−1 day−1 (means and SD in Table 1). Neither plant species richness nor functional group richness significantly affected the NH4+ immobilization into the labile (Table 3) and the recalcitrant organic N pools. Grasses significantly increased the immobilization of NH4+ into the labile organic N pool (Table 3; Fig. 6b). There were no significant effects of plant community composition on the immobilization of NO3 into the organic N pool. The normalization of N transformation rates and pool sizes to the total N concentrations did not change the overall results (Tables S3–S7).

Microbial properties versus gross N transformation rates

Given their known role as drivers of N transformations (Fornara et al. 2011), microbial activity likely is a principal factor that needs to be considered when trying to explain the observed differences in gross N transformation rates (Booth et al. 2005). Microbial C/N ratios showed a marginally significant negative correlation with N mineralization from labile organic N, and a significant negative correlation with autotrophic nitrification (Fig. 7a, b). Furthermore, we found a marginally significant positive relationship between basal respiration and immobilization of NH4+ into the labile organic N (Fig. 7c).

Discussion

Pool sizes and gross N transformation rates

The drastic dilution of applied 15N-NH4+ in the soil extracts during the incubation (Figs. 2a and 4a) indicated that there was a rapid release of unlabeled NH4+ from the organic matter into the 15N-labeled NH4+ pool. Huygens et al. (2007) suggested that the rapid disappearance of labeled NH4+ might be attributable to the exchange of the labeled NH4+ by adsorbed NH4+ on clay minerals or other cation-exchanger sites. The increase in the 15N enrichment of the NH4+ pool in the 15NO3 labeled treatments (Fig. 4c) can be attributed to the re-mineralization of recently immobilized 15NO3. The gradual decline of 15NO3 concentrations in the soil extracts during the incubation (Fig. 4d) demonstrated that NO3 at natural abundance entered into the 15N-labeled NO3 pool via autotrophic or heterotrophic nitrification. Throughout the incubation, the pool sizes of dissolved mineral N (i.e., the sum of the NH4-N and NO3-N concentrations) remained nearly constant (Fig. 2), which showed that the net N transformation rates were similar for all plots. The study conducted by Huygens et al. (2007) in unpolluted South Chilean forests also found almost constant pool sizes at low net mineralization and nitrification rates.

Fig. 4
figure4

Relationship between plant species richness and 15N enrichment (15NH4+, 15NO3) differentiated according to the labeling of 15N-NH4Cl (a, b), 15N-KNO3 (c, d) and combination of 15N-NH4Cl + 15N-KNO3 (e, f). Data are presented as mean ± standard error (SE). Note the differences in y-axis scaling

The rate of gross N mineralization (MNrec + MNlab) in our experiment fell within the range of 0.40–4.07 μg N (g soil)−1 day−1 reported in the literature for grasslands (Jamieson et al. 1999; Accoe et al. 2004; Müller et al. 2004; McKinley et al. 2008; Müller et al. 2014). The measured total NH4+ immobilization rates (INH4-Nrec + INH4-Nlab) were also in the range of 0.10–0.88 μg N (g soil)−1 day−1 reported by other grassland studies (Hungate et al. 1997; Verchot et al. 2002; Müller et al. 2011). The measured rates of heterotrophic nitrification in this study were similar to or higher than the range of 0.07–1.41 μg N (g soil)−1 day−1 reported in other studies in grassland soils (Müller et al. 2004, 2009; Laughlin et al. 2009). The autotrophic nitrification rates determined in this study are in the range of 0.10–2.88 μg N (g soil)−1 day−1 reported for other grassland studies (Zaman et al. 1999; Accoe et al. 2004; Müller et al. 2009; Demey et al. 2014).

The NO3 immobilization rates were similar to or higher than the range of 0.81–3.84 μg N (g soil)−1 day−1 reported in the literature for grasslands (Davidson et al. 1990; Watson et al. 2000; Corre et al. 2002). The NO3 immobilization rates in our study were comparable to the total nitrification rates, which showed that the NO3 produced via nitrification was completely assimilated by microorganisms, leaving little space for NO3 leaching or denitrification. Aber et al. (1989) and Huygens et al. (2007) suggested that N losses via leaching or denitrification may not occur if N inputs do not exceed plant or microbial N demand. The occurrence of high NO3 immobilization is also attributable to the insufficient availability of NH4+ in soil (Fig. 2) to meet the microbial demand for N (Rice and Tiedje 1989; Corre et al. 2002). However, both nitrification and NO3 immobilization rates were higher than under field conditions, because our microcosm experiment did not include plants and thus excluded plant uptake of NH4+. This is in line with the suggestion of Kammann et al. (1998) that the increased NO3 concentrations observed in laboratory experiments are not likely to occur in the field, because plant uptake and leaching would decrease the NO3 concentration in soil.

Plant diversity effects on N pool sizes

The presence of grasses significantly increased the dissolved NH4+ pool, probably because of their dense rooting system (Oelmann et al. 2007; Bessler et al. 2009; Ravenek et al. 2014). The dead roots along with their exudates, which remained in our sample, likely increased microbial activity (Van der Krift et al. 2001; Lange et al. 2015; Eisenhauer et al. 2017). The increased microbial activity accelerated the decomposition of soil organic matter, which is also reflected by the positive effect of grasses on the gross mineralization rate (Table 3; Fig. 5c). At the same time, the initial exchangeable (= adsorbed) NH4+ pool was marginally significantly lower in the presence of grasses, possibly because of the exhaustive N exploitation of the dense grass roots prior to our experiment without plants (Table S1). This exhaustive N exploitation is also supported by the fact that the presence of grasses reduced NO3-N and total dissolved N leaching in the Jena Experiment (Leimer et al. 2016). The increase in the pool sizes of the initial labile and recalcitrant organic matter with increasing species richness (Table S2; Fig. S1) can be attributed to the positive effect of increasing species richness on organic matter and total N accumulation at the study sites of the Jena Experiment as a consequence of the positive species richness-biomass production relationship (Weisser et al. 2017).

Fig. 5
figure5

Relationship between a functional group richness and the gross N mineralization from the labile organic N pool (MNlab) and effects of b presence (+)/absence (−) of legumes and (c) presence/absence of grasses on the gross N mineralization from the labile organic N pool (MNlab). The whiskers in Fig. 3b, c represent standard errors. Significance codes: asterisk indicates p < 0.05 and solid bullet indicates p < 0.1

Plant diversity effects on NH4+ production and immobilization processes

We observed that functional group richness had a marginally significant positive effect on gross N mineralization from the labile organic N pool (Table 3; Fig. 5a). A similar positive effect of plant species richness on gross N mineralization was reported by Zak et al. (2003) and West et al. (2006) in laboratory incubation experiments with soils of the Cedar Creek biodiversity experiments in Minnesota, USA, where sandy, nutrient-poor soils prevail. In both studies, the range of species richness was the same as in our study (i.e., 1 to 16 species, but the plant community composition was different). However, both studies did not distinguish between mineralization from the labile and recalcitrant organic N pools. Wedin and Pastor (1993) have previously reported that labile organic N is important for the N supply of plants in grassland, while the recalcitrant organic N is responsible for longer-term N storage. Zak et al. (2003) and West et al. (2006) attributed the significant positive relationship between plant species richness and gross N mineralization to the high plant productivity resulting in high organic inputs to soil, which would have remained in the incubated samples of our experiment. Furthermore, we found a marginally significant positive effect of the microbial C/N ratio on gross N mineralization rates (Fig. 7a). The microbial C/N ratio is also considered one of the potential variables influencing the rate of N mineralization, because inorganic N production increases when microbial activity increases (Booth et al. 2005).

Most of the studies on the biodiversity-N cycle relationship reported a positive effect of legumes on N pools and transformations. To test if the functional group richness on gross N mineralization was mostly driven by legumes, we ran a separate ANOVA by fitting “presence of legumes” before “functional group richness.” We found that functional group richness explained 19.3% of the total variance, of which 19.2% was explained by the presence of legumes alone. This suggests that the presence of legumes indeed explained the functional group richness effect. This is in line with earlier findings of Hooper and Vitousek (1998) that nutrient cycling is more dependent on certain functional groups rather than on species richness. However, later studies have shown that plant species richness significantly influences the N cycle irrespective of the functional group composition of the community (Weisser et al. 2017). We cannot rule out that the failure to see a species richness effect in our experiment is attributable to the comparatively low statistical power of our experiment, which only included soil samples from one out of four blocks of the Jena Experiment, and also to the fact that the incubation experiment did not include living plants.

In a field experiment at the same study site, Lama et al. (2020) observed a significant negative relationship between species richness and gross N mineralization, which was mainly driven by the increasing root C/N ratios with increasing species richness. Higher species richness increased root C/N ratios via the dilution of plant nutrient concentrations, because of the greater height of plants in species-rich mixtures as a consequence of the competition for light. We can only speculate that under the optimum decomposition conditions of our incubation experiment, and in the absence of active plants, the negative effect of the increasingly smaller C/N ratios in roots with increasing species richness was overprinted.

The positive influence of legumes on gross N mineralization rates from the labile organic N pool (Table 3; Fig. 5b) is likely related to the fact that legumes generally increase N concentrations in soils (Oelmann et al. 2007; Fornara and Tilman 2008). This results from atmospheric N2-fixation (Ledgard 2001; Spehn et al. 2002) or the generally higher N concentrations in legumes (Marschner 2012) which will also result in the return of more N to the soil. The legume-derived more readily degradable organic matter is introduced into the soil via rhizodeposition and aboveground litterfall (Read 1996). The N accumulation in soil in the presence of legumes resulted in a higher aboveground biomass in the legume-containing plots of the Jena Experiment compared with that in the legume-free plots (Marquard et al. 2009) further increasing the available C pool in soil. Moreover, soil microbial biomass C increases in the presence of legumes (Eisenhauer et al. 2010; but see Strecker et al. 2016 for changing legume effects over time), and this might have further enhanced gross N mineralization.

The positive relationship between NH4+ immobilized from labile organic N and the presence of grasses (Fig. 6b) might be attributable to an enhanced microbial activity (Fig. 7c; Eisenhauer et al. 2010). Grasses are characterized by dense fibrous roots with a high length (Weigelt et al. 2008). Therefore, grasses likely enhanced microbial biomass and activity by providing large amounts of root exudates (Van der Krift et al. 2001; Eisenhauer et al. 2010), and this grass effect might have persisted in our experiment without plants.

Fig. 6
figure6

Effects of a the presence (+)/absence (−) of legumes on autotrophic nitrification (ONH4) and b of presence (+)/absence (−) of grasses on the immobilization of NH4+ in the labile organic N pool (INH4-Nlab). Whiskers represent standard errors. Significance code: one asterisk indicates p < 0.05, two asterisks indicate p < 0.01

Fig. 7
figure7

Relationships between a microbial C:N ratio and N mineralization into labile organic N (MNlab), b microbial C:N ratio and autotrophic nitrification (ONH4), and c basal respiration and ammonium immobilization from labile organic N (INH4-Nlab). p and r values refer to the results from Pearson’s correlation tests. Solid line shows significance at p < 0.05 and dotted lines show significance at p < 0.1. Note the differences in y-axis scaling

Plant diversity effects on NO3 production and immobilization processes

Our study indicates that heterotrophic nitrification of organic N is an important process of NO3 production in the studied grassland soils, because heterotrophic nitrification rates were similar to the rates of autotrophic nitrification and also because heterotrophic nitrification is the other direct way of producing mineral N from organic N. The study by Müller et al. (2004) regarded heterotrophic nitrification as the predominant pathway for NO3 production in soils at high recalcitrant organic C in a grassland ecosystem. The NH4+ produced by mineralization, which is not taken up by plants or immobilized by microbes, is oxidized by nitrifiers and results in elevated soil NO3 concentrations. This assumption is corroborated by the positive correlation between mineralization and nitrification rates reported in the review of Booth et al. (2005). Our incubation experiment did not include plants, and therefore, the produced NH4+ was not taken up by plants offering more substrate for the nitrification to NO3 than under field conditions with plants. Furthermore, we observed an increasing rate of autotrophic nitrification in the presence of legumes (Table 4; Fig. 6a), because of the higher N concentrations in the legume-containing plots of the Jena Experiment (Oelmann et al. 2007). In addition, autotrophic nitrification exhibited a significant negative relationship with the microbial C/N ratio (Fig. 7b). Lower microbial C/N ratios which are associated with a substrate of high quality (Hart et al. 1994) increase microbial activity, thereby enhancing autotrophic nitrification (Booth et al. 2005; Inselsbacher et al. 2013). Previous studies in the Jena Experiment have shown elevated net nitrification (Scherer-Lorenzen et al. 2003) and increased KCl-extractable soil NO3 concentrations (Oelmann et al. 2011; Leimer et al. 2014) in the presence of legumes. However, Hooper and Vitousek (1997) and Niklaus et al. (2006) found no effects of plant diversity on nitrification.

Conclusions

Our study demonstrated that in the absence of plant uptake, almost all the produced NH4+ was converted into NO3. We observed a strong legacy effect of legumes for gross N transformations. Legumes particularly had a positive effect on gross N mineralization and autotrophic nitrification. Grasses also increased the dissolved NH4+ pool, gross N mineralization, and NH4+ immobilization. Heterotrophic nitrification was found to play a vital role in soil N cycling. Consequently, future studies should focus on identifying the controlling factors of heterotrophic nitrification in grassland soils.

The fact that we conducted our experiment without plants as is commonly done to collect the data needed by the used N cycling model Ntrace limited the transferability of the results to the field. Therefore, future studies should include plants and be conducted in growth chambers or Ecotrons. Given the partly small effect sizes of plant community composition on several elements of the N cycle, it would also be desirable to increase the statistical power of such experiments by including more replicates of the various species mixtures than we were able to include.

Our results indicate that changing contributions of legumes and grasses in response to environmental and land-use change will markedly influence the N availability for the plant community and possibly also the N leaking into atmosphere and water. However, we could not confirm that species or functional group richness tighten the N cycle and deplete mineral N concentrations in soil, possibly because of a limited statistical power of our experiment.

Data availability

The data will be uploaded to the database of the Jena Experiment, which will become public after an embargo time.

References

  1. Aber J, Nadelhoffer K, Steudler P, Melillo J (1989) Nitrogen saturation in northern forest ecosystems. Bioscience 39:378–386

    Google Scholar 

  2. Accoe F, Boeckx P, Busschaert J, Hofman G, Van Cleemput O (2004) Gross N transformation rates and net N mineralisation rates related to the C and N contents of soil organic matter fractions in grassland soils of different age. Soil Biol Biochem 36:2075–2087

    CAS  Google Scholar 

  3. Allan E, Weisser W, Fischer M, Schulze E-D, Weigelt A, Roscher C (2013) A comparison of the strength of biodiversity effects across multiple functions. Oecologia 173:223–237

    PubMed  Google Scholar 

  4. Bedard-Haughn A, Matson A, Pennock D (2006) Land use effects on gross nitrogen mineralization, nitrification and N2O emissions in ephemeral wetlands. Soil Biol Biochem 38:3398–3406

    CAS  Google Scholar 

  5. Benbi D, Richter J (2002) A critical review of some approaches to modelling nitrogen mineralization. Biol Fertil Soils 35:139–157

    Google Scholar 

  6. Bessler H, Temperton V, Roscher C, Buchmann N, Schmid B, Schulze E-D, Weisser W, Engels C (2009) Aboveground overyielding in grassland mixtures is associated with reduced biomass partitioning to belowground organs. Ecology 90:1520–1530

    PubMed  Google Scholar 

  7. Booth M, Stark J, Rastetter E (2005) Controls on nitrogen cycling in terrestrial ecosystems: a synthetic analysis of literature data. Ecol Monogr 75:139–157

    Google Scholar 

  8. Brady N, Weil R (2002) The nature and properties of soils, 13th edn. Macmillan Publishing Company, New York

    Google Scholar 

  9. Brookes P, Landman A (1985) Chloroform fumigation and the release of soil nitrogen: a rapid direct extraction method to measure microbial biomass nitrogen in soil. Soil Biol Biochem 17:837–842

    CAS  Google Scholar 

  10. Causarano HJ, Franzluebbers AJ, Shaw JN, Reeves DW, Raper RL, Wood CW (2008) Soil organic carbon fractions and aggregation in the southern Piedmont and coastal plain. Soil Sci Soc Am J 72:227–230

    Google Scholar 

  11. Cheng Y, Wang J, Mary B, Zhang J, Cai Z, Chang S (2013) Soil pH has contrasting effects on gross and net nitrogen mineralizations in adjacent forest and grassland soils in central Alberta, Canada. Soil Biol Biochem 57:848–857

    CAS  Google Scholar 

  12. Cheng Y, Wang J, Wang S, Zhang J, Cai Z (2014) Effects of soil moisture on gross N transformations and N2O emissions in acid subtropical forest soils. Biol Fertil Soils 50:1099–1108

    CAS  Google Scholar 

  13. Core Team R (2016) A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria

    Google Scholar 

  14. Corre M, Schnabel R, Stout W (2002) Spatial and seasonal variation of gross nitrogen transformations and microbial biomass in a Northeastern US grassland. Soil Biol Biochem 34:445–457

    CAS  Google Scholar 

  15. Dassen S, Cortois R, Martens H, de Hollander M, Kowalchuk GA, van der Putten WH, De Deyn GB (2017) Differential responses of soil bacteria, fungi, archaea and protists to plant species richness and plant functional group identity. Mol Ecol 26:4085–4098

    PubMed  CAS  Google Scholar 

  16. Davidson E, Stark J, Firestone M (1990) Microbial production and consumption of nitrate in an annual grassland. Ecology 71:1968–1975

    Google Scholar 

  17. Demey A, Ruetting T, Huygens D, Staelens J, Hermy M, Verheyen K, Boeckx P (2014) Hemiparasitic litter additions alter gross nitrogen turnover in temperate semi-natural grassland soils. Soil Biol Biochem 68:419–428

    CAS  Google Scholar 

  18. Eisenhauer N, Bessler H, Engels C, Gleixner G, Habekost M, Milcu A, Partsch S, Sabais A, Scherber C, Steinbeiss C, Weigelt A, Weisser W, Scheu S (2010) Plant diversity effects on soil microorganisms support the singular hypothesis. Ecology 91:485–496

    PubMed  CAS  Google Scholar 

  19. Eisenhauer N, Lanoue A, Strecker T, Scheu S, Steinauer K, Thakur M, Mommer L (2017) Root biomass and exudates link plant diversity with soil bacterial and fungal biomass. Sci Rep 7:1–8

    CAS  Google Scholar 

  20. Elser J, Bracken M, Cleland E, Gruner D, Harpole W, Hillebrand H, Ngai J, Seabloom E, Shurin J, Smith J (2007) Global analysis of nitrogen and phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems. Ecol Lett 10:1–8

    Google Scholar 

  21. Fay P, Prober W, Harpole W, Knops J, Bakker J, Borer E, Lind E, MacDougall A, Seabloom E, Wragg P, Adler P, Blumenthal D, Buckley Y, Chu C, Cleland E, Collins S, Davies K, Du G, Feng X (2015) Grassland productivity limited by multiple nutrients. Nat Plant 1:1–5

    Google Scholar 

  22. Fornara D, Tilman D (2008) Plant functional composition influences rates of soil carbon and nitrogen accumulation. J Ecol 96:314–322

    CAS  Google Scholar 

  23. Fornara D, Tilman D (2009) Ecological mechanisms associated with the positive diversity-productivity relationship in an N-limited grassland. Ecology 90:408–418

    PubMed  CAS  Google Scholar 

  24. Fornara D, Bardgett R, Steinbeiss S, Zak D, Gleixner G, Tilman D (2011) Plant effects on soil N mineralization are mediated by the composition of multiple soil organic fractions. Ecol Res 26:201–208

    CAS  Google Scholar 

  25. Hart S, Nason G, Myrold D, Perry D (1994) Dynamics of gross nitrogen transformations in an old-growth forest: the carbon connection. Ecology 75:880–891

    Google Scholar 

  26. Hatch D, Jarvis S, Parkinson R, Lovell R (2000) Combining field incubation with nitrogen-15 labelling to examine nitrogen transformations in low to high intensity grassland management systems. Biol Fertil Soils 30:492–499

    CAS  Google Scholar 

  27. Hoffmann K, Bivour W, Früh B, Kossmann M, Voss PH (2014) Klimauntersuchungen in Jena für die Anpassung an den Klimawandel und seine erwarteten Folgen. Offenbach am Main: Berichte des Deutschen Wetterdienstes 243

  28. Hooper D, Vitousek P (1997) The effects of plant composition and diversity on change. Science 302:1512–1513

    Google Scholar 

  29. Hooper D, Vitousek P (1998) Effects of plant composition and diversity of nutrient cycling. Ecol Monogr 68:121–149

    Google Scholar 

  30. Hooper D, Chapin F, Ewel J, Hector A, Inchausti P, Lavorel S, Lawton J, Lodge D, Loreau M, Naeem S, Schmid B, Setälä H, Symstad A, Vandermeer J, Wardle D (2005) Effects of biodiversity on ecosystem functioning: a consensus of current knowledge. Ecol Monogr 75:3–35

    Google Scholar 

  31. Hungate B, Lund C, Pearson H, Chapin F (1997) Elevated CO2 and nutrient addition alter soil N cycling and N trace gas fluxes with early season wet-up in a California annual grassland. Biogeochemistry 37:89–109

    CAS  Google Scholar 

  32. Huygens D, Rütting T, Boeckx P, Cleemput O, Godoy R, Müller C (2007) Soil nitrogen conservation mechanisms in a pristine south Chilean Nothofagus forest ecosystem. Soil Biol Biochem 39:2448–2458

    CAS  Google Scholar 

  33. Inselsbacher E, Wanek W, Strauss J, Zechmeister-Boltenstern S, Müller C (2013) A novel 15N tracer models reveals: plant nitrate uptake governs nitrogen transformation rates in agricultural soils. Soil Biol Biochem 57:301–310

    CAS  Google Scholar 

  34. IUSS Working Group WRB (2014). World Reference Base for Soil Resources 2014. World Soil Resource Reports No. 106. FAO, Rome.

  35. Jamieson N, Manoghan R, Barraclough D (1999) Seasonal trends of gross N mineralization in a natural calcareous grassland. Glob Change Biol 5:423–431

    Google Scholar 

  36. Kammann C, Grünhage L, Müller C, Jacobi S, Jäger H-J (1998) Seasonal variability and mitigation options for N2O emission from differently managed grasslands. Environ Pollut 102:179–186

    CAS  Google Scholar 

  37. Lama S, Velescu A, Leimer S, Weigelt A, Chen H, Eisenhauer N, Scheu S, Oelmann Y, Wilcke W (2020) Plant diversity influenced gross nitrogen mineralization, microbial ammonium consumption and gross nitrogen immobilization in a grassland experiment. Oecologia [In review]

  38. Lang M, Li P, Han X, Qiao Y, Miao S (2015) Gross nitrogen transformations in black soil under different land uses and management systems. Biol Fertil Soils 52:233–241

    Google Scholar 

  39. Lange M, Habekost M, Eisenhauer N, Roscher C, Bessler H, Engels C, Oelmann Y, Scheu S, Wilcke W, Schulze E-D, Gleixner G (2014) Biotic and abiotic properties mediating plant diversity effects on soil microbial communities in an experimental grassland. PLOS ONE 9:e96182

    PubMed  PubMed Central  Google Scholar 

  40. Lange M, Eisenhauer N, Sierra C, Besseler H, Engels C, Griffiths R, Mellado-Vazquez P, Malik A, Roy J, Scheu S, Steinbeiss S, Thomson B, Trumbore S, Gleixner G (2015) Plant diversity increases soil microbial activity and soil carbon storage. Nat Commun 6:1–8

    Google Scholar 

  41. Laughlin R, Ruetting T, Mueller C, Watson C, Stevens R (2009) Effect of acetate on soil respiration, N2O emissions and gross N transformations related to fungi and bacterial in a grassland soil. Appl Soil Ecol 42:25–30

    Google Scholar 

  42. Ledgard S (2001) Nitrogen cycling in low input legume-based agriculture, with emphasis on legume/grass pastures. Plant Soil 228:43–59

    CAS  Google Scholar 

  43. Leimer S, Wirth C, Oelmann Y, Wilcke W (2014) Biodiversity effects on nitrate concentrations in soil solution: a Bayesian model. Biogeochemistry 118:141–157

    CAS  Google Scholar 

  44. Leimer S, Oelmann Y, Eisenhauer N, Milcu A, Roscher C, Scheu S, Weigelt A, Wirth C, Wilcke W (2016) Mechanisms behind plant diversity effects on inorganic and organic N leaching from temperate grassland. Biogeochemistry 131:339–353

    CAS  Google Scholar 

  45. Loreau M, Naeem S, Inchausti P, Bengtsson J, Grime J, Hector A, Hooper D, Huston M, Raffaelli D, Schmid B, Tilman D, Wardle D (2001) Biodiversity and ecosystem functioning: current knowledge and future challenges. Science 294:804–808

    PubMed  CAS  Google Scholar 

  46. Marquard E, Weigelt A, Temperton V, Roscher C, Schumacher J, Buchmann N et al (2009) Plant species richness and functional compostion drive overyielding in a six-year grassland experiment. Ecology 90:3290–3302

    PubMed  Google Scholar 

  47. Marschner P (2012) Marschner’s mineral nutrients of higher plants, 3rd edn. Academic Press, London

    Google Scholar 

  48. McIlvin M, Casciotti K (2011) Technical updates to the bacterial method for nitrate isotopic analyses. Anal Chem 83:1850–1856

    PubMed  CAS  Google Scholar 

  49. McKinley D, Rice C, Blair J (2008) Conversion of grassland to coniferous woodland has limited effects on soil nitrogen cycle processes. Soil Biol Biochem 40:2627–2633

    CAS  Google Scholar 

  50. Mueller K, Hobbie S, Tilman D, Reich P (2013) Effects of plant diversity, N fertilization, and elevated carbon dioxide on grassland soil N cycling in a long-term experiment. Glob Change Biol 19:1249–1261

    Google Scholar 

  51. Müller C, Stevens R, Laughlin R (2004) A 15N tracing model to analyse N transformations in old grassland soil. Soil Biol Biochem 36:619–632

    Google Scholar 

  52. Müller C, Rütting T, Kattge J, Laughlin R, Stevens R (2007) Estimation of parameters in complex 15N tracing models via Monte Carlo sampling. Soil Biol Biochem 39:715–726

    Google Scholar 

  53. Müller C, Rütting T, Abbasi M, Laughlin R, Kammann C, Clough T, Sherlock R, Kattge J, Jäger H-J, Watson C, Stevens R (2009) Effect of elevated CO2 on soil N dynamics in a temperate grassland soil. Soil Biol Biochem 41:1996–2001

    Google Scholar 

  54. Müller C, Laughlin R, Christie P, Watson C (2011) Effects of repeated fertilizer and cattle slurry applications over 38 years on N dynamics in a temperate grassland soil. Soil Biol Biochem 43:1362–1371

    Google Scholar 

  55. Müller C, Laughlin R, Spott O, Rütting T (2014) Quantification of N2O emission pathways via a 15N tracing model. Soil Biol Biochem 72:44–54

    Google Scholar 

  56. Nadelhoffer K (1990) Microlysimeter for measuring nitrogen mineralization and microbial respiration in aerobic soil incubations. Soil Sci Soc Am J 54:411–415

    CAS  Google Scholar 

  57. Niklaus P, Wardle D, Tate K (2006) Effects of plant species diversity and composition on nitrogen cycling and the trace gas balance of soils. Plant Soil 282:83–98

    CAS  Google Scholar 

  58. Oelmann Y, Wilcke W, Temperton V, Buchmann N, Roscher C, Schumacher J, Schulze E-D, Weisser W (2007) Soil and plant nitrogen pools as related to plant diversity in an experimental grassland. Soil Sci Soc Am J 71:720–729

    CAS  Google Scholar 

  59. Oelmann Y, Buchmann N, Gleixner G, Habekost M, Roscher C, Rosenkranz S, Schulze E-D, Steinbeiss S, Temperton V, Weigelt A, Weisser W, Wilcke W (2011) Plant diversity effects on aboveground and belowground N pools in temperate grassland ecosystems: development in the first 5 years after establishment. Global Biogeochem Cy 25:1–11

    Google Scholar 

  60. Paterson E (2003) Importance of rhizodeposition in the coupling of plant and microbial productivity. Eur J Soil Sci 54:741–750

    Google Scholar 

  61. Ravenek J, Bessler H, Engels C, Scherer-Lorenzen M, Gessler A, Gockele A, De Luca E, Temperton V, Ebeling A, Roscher C, Schmid B, Weisser W, Wirth C, De Kroon H, Weigelt A, Mommer L (2014) Long-term study of root biomass in a biodiversity experiment reveals shifts in diversity effects over time. Oikos 123:1528–1536

    Google Scholar 

  62. Read D (1996) The structure and function of the Ericoid mycorrhizal root. Ann Bot 77:365–374

    CAS  Google Scholar 

  63. Rice C, Tiedje J (1989) Regulation of nitrate assimilation by ammonium in soils and in isolated soil microorganisms. Soil Biol Biochem 21:597–602

    CAS  Google Scholar 

  64. Roscher C, Schumacher J, Baade J, Wilcke W, Gleixner G, Weisser W, Schmid B, Schulze E-D (2004) The role of biodiversity for element cycling and trophic interactions: an experimental approach in a grassland community. Basic App Ecol 5:107–121

    Google Scholar 

  65. Rosenkranz S, Wilcke W, Eisenhauer N, Oelmann Y (2012) Net ammonification as influenced by plant diversity in experimental grasslands. Soil Biol Biochem 48:78–87

    CAS  Google Scholar 

  66. Scherer-Lorenzen M, Palmborg C, Prinz A, Schulze E-D (2003) The role of plant diversity and composition for nitrate leaching in grasslands. Ecology 84:1539–1552

    Google Scholar 

  67. Scheu S (1992) Automated measurement of the respiratory response of soil micro-compartments: active microbial biomass in earthworm faeces. Soil Biol Biochem 24:1113–1118

    Google Scholar 

  68. Schimel J, Bennett J (2004) Nitrogen mineralization: challenges of a changing paradigm. Ecology 85:591–602

    Google Scholar 

  69. Sigman D, Casciotti K, Andreani M, Barford C, Galanter M, Böhlke J (2001) A bacterial method for the nitrogen isotopic analysis of nitrate in seawater and freshwater. Anal Chem 73:4145–4153

    PubMed  CAS  Google Scholar 

  70. Solly E, Schöning I, Boch S, Müller J, Socher S, Trumbore S, Schrumpf M (2013) Mean age of carbon in fine roots from temperate forests and grasslands with different management. Biogeosciences 10:4833–4843

    Google Scholar 

  71. Spehn E, Scherer-Lorenzen M, Schmid B, Hector A, Caldeira M, Dimitrakopoulus P, Finn JA, Jumpponen A, O'Donovan G, Pereira JS, Schulze E-D, Troumbis AY, Körner C (2002) The role of legumes as a component of biodiversity in a cross-European study of grassland biomass nitrogen. Oikos 98:205–218

  72. Stockdale E, Hatch D, Murphy D, Ledgard S, Watson C (2002) Verifying the nitrification to immobilisation ratio (N/I) as a key determinant of potential nitrate loss in grassland and arable soils. Agronomie 22:831–838

    Google Scholar 

  73. Strecker T, Gonzalez O, Scheu S, Eisenhauer N (2016) Functional composition of plant communities determines the spatial and temporal stability of soil microbial properties in a long-term plant diversity experiment. Oikos 125:1743–1754

    Google Scholar 

  74. Sutton MA, Howard CM, Erisman JW, Billen G, Bleeker A, Grenfelt P, van Grinsven H, Grizetti B (2011) The European Nitrogen Assessment. Cambridge University Press, Cambridge, UK

    Google Scholar 

  75. Van der Krift T, Kuikman PM, Berendse F (2001) Plant species and nutritional-mediated control over rhizodeposition and root decomposition. Plant Soil 228:191–200

    Google Scholar 

  76. Verchot L, Groffman P, Frank D (2002) Landscape versus ungulate control of gross mineralization and gross nitrification in semi-arid grasslands of Yellowstone National Park. Soil Biol Biochem 34:1691–1699

    CAS  Google Scholar 

  77. Watson C, Travers G, Kilpatrick D, Laidlaw A, O’'Riordon E (2000) Overestimation of gross N transformation rates in grassland soils due to non-uniform exploitation of applied and native pools. Soil Biol Biochem 32:2019–2030

    CAS  Google Scholar 

  78. Wedin D, Pastor J (1993) Nitrogen mineralization dynamics in grass monocultures. Oecologia 96:186–192

    PubMed  Google Scholar 

  79. Weigelt A, Schumacher J, Roscher C, Schmid B (2008) Does biodiversity increase spatial stability in plant community biomass? Ecol Lett 11:338–347

    PubMed  Google Scholar 

  80. Weisser W, Roscher C, Meyer S, Ebeling A, Luo G, Allan E, Bessler H, Barnard R, Buchmann N, Buscot F, Engels C, Fischer C, Gessler A, Gleixner G, Halle S, Hildebrandt A, Hillebrand H, De Kroon H, Lange M, Leimer S, Le Roux X, Milcu A, Mommer L, Niklaus P, Oelmann Y, Proulx R, Roy J, Scherber C, Scherer-Lorenzen M, Scheu S, Tscharntke T, Wachendorf M, Wagg C, Weigelt A, Wilcke W, Wirth C, Schulze E-D, Schmid B, Eisenhauer N (2017) Biodiversity effects on ecosystem functioning in a 15-year grassland experiment: pPatterns, mechanisms, and open questions. Basic App Ecol 23:1–73

    Google Scholar 

  81. West J, Hobbie S, Reich P (2006) Effects of plant species diversity, atmospheric CO2, and N addition on gross rates of inorganic N release from soil organic matter. Glob Change Biol 12:1400–1408

    Google Scholar 

  82. Wild B, Schnecker J, Knoltsch A, Takriti M, Mooshammer M, Lashchinskiy M, Richter A (2015) Microbial nitrogen dynamics in organic and mineral soil horizons along a latitudinal transect in wester Siberia. Global Biogeochem Cy 29:567–582

    CAS  Google Scholar 

  83. Wu J, Joergensen R, Pommerening B, Chaussod R, Brookes P (1990) Measurement of soil microbial bio-mass by fumigation extraction—an automated procedure. Soil Biol Biochem 22:1167–1169

    CAS  Google Scholar 

  84. Zak D, Holmes D, White D, Peacock A, Tilman D (2003) Plant diversity, soil microbial communities, and ecosystem function: are there any links? Ecology 84:2042–2050

    Google Scholar 

  85. Zaman M, Di H, Cameron K, Frampton C (1999) Gross nitrogen mineralization and nitrificaiton rates and their relationships to enzyme activities and the soil microbial biomass in soils treated with dairy shed effluent and ammonium fertilizer at different water potentials. Biol Fertil Soils 29:178–186

    CAS  Google Scholar 

  86. Zhang L, Altabet M, Wu T, Hadas O (2007) Sensitive measurement of NH4+ 15N/14N (δ15NH4+) at natural abundance levels in fresh and saltwaters. Anal Chem 79:5297–5303

    PubMed  CAS  Google Scholar 

  87. Zhang J, Wang L, Zhao W, Hu H, Feng X, Mueller C, Cai Z (2016) Soil gross nitrogen transformations along the Northeast China Transect (NECT) and their response to simulated rainfall events. Sci Rep 6:1–8

    Google Scholar 

Download references

Acknowledgments

We thank everyone who helped with the management of the experiment and in particular the initiators, E.-D. Schulze, B. Schmid, and W.W. Weisser, as well as the scientific coordinators C. Roscher, A. Weigelt, and A. Ebeling. Thanks also to all those who assisted us during the weeding campaigns. The Jena Experiment was funded by the Deutsche Forschungsgemeinschaft (DFG, FOR 456 and 1451, Wi 1601/4) and the Swiss National Science Foundation (SNSF, 200021E-131195/1), with additional support from the Friedrich Schiller University Jena and the Max Planck Society. The 15N tracing analysis was carried out in close collaboration with the DFG-funded research unit DASIM (FOR2337). N. Eisenhauer acknowledges support by the German Center for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig funded by the DFG (FZT 118).

Funding

Open Access funding provided by Projekt DEAL. This study received financial support from Deutsche Forschungsgemeinschaft (DFG), Swiss National Science Foundation (SNSF), Friedrich Schiller-University Jena, Max Planck Society

Author information

Affiliations

Authors

Corresponding author

Correspondence to Wolfgang Wilcke.

Ethics declarations

Conflicts of interest

The authors declare that they have no conflict of interest.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

ESM 1

(DOCX 1981 kb)

Rights and permissions

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://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Lama, S., Kuhn, T., Lehmann, M.F. et al. The biodiversity - N cycle relationship: a 15N tracer experiment with soil from plant mixtures of varying diversity to model N pool sizes and transformation rates. Biol Fertil Soils 56, 1047–1061 (2020). https://doi.org/10.1007/s00374-020-01480-x

Download citation

Keywords

  • Ntrace model
  • Laboratory microcosms
  • Gross N transformation rates
  • Plant diversity
  • The Jena Experiment