Correspondence: Ellen Kandeler, Institute of Soil Science and Land Evaluation, Soil Biology Section, University of Hohenheim, Emil-Wolff-Straße 27, D-70599 Stuttgart, Germany. Tel.: +49 711/459 24220; fax: +49 711/459 23117; e-mail: email@example.com
A field-scale manipulation experiment conducted for 16 years in a Norway spruce forest at Solling, Central Germany, was used to follow the long-term response of total soil bacteria, nitrate reducers and denitrifiers under conditions of reduced N deposition. N was experimentally removed from throughfall by a roof construction (‘clean rain plot’). We used substrate-induced respiration (SIR) to characterize the active fraction of soil microbial biomass and potential nitrate reduction to quantify the activity of nitrate reducers. The abundance of total bacteria, nitrate reducers and denitrifiers in different soil layers was analysed by quantitative PCR of 16S rRNA gene, nitrate reduction and denitrification genes. Reduced N deposition temporarily affected the active fraction of the total microbial community (SIR) as well as nitrate reductase activity. However, the size of the total, nitrate reducer and denitrifier communities did not respond to reduced N deposition. Soil depth and sampling date had a greater influence on the density and activity of soil microorganisms than reduced deposition. An increase in the nosZ/16S rRNA gene and nosZ/nirK ratios with soil depth suggests that the proportion of denitrifiers capable of reducing N2O into N2 is larger in the mineral soil layer than in the organic layer.
Nitrogen emissions and atmospheric deposition are globally significant in their potential to alter the nutrient balance in soils, triggering changes in the composition of plants and soil organisms (Gidman et al., 2006; McLauchlan et al., 2007). The current hypothesis suggests that increased N deposition promotes the rate of soil organic matter accumulation by either increasing leaf/needle biomass and litter production or by reducing the decomposition of organic matter (de Vries et al., 2006). Based on data collected at the monitoring plots of different European forest sites, the contribution of N deposition to net sequestration of C in trees and soil in the period 1960–2000 is c. 5.1 Mton year−1 in tree wood and 6.7 Mton year−1 in soil – about 10% of the total C sequestration during that period (de Vries et al., 2006). Several N-addition experiments (partly including 15N) in temperate forests revealed that soils, rather than plants, are a main long-term sink for the added nitrate and ammonium (Nadelhoffer et al., 1999; Providoli et al., 2006). Specific groups of soil microorganisms differed in their response to N deposition or N addition (Gundersen, 1998; Butterbach-Bahl et al., 2002; Hungate et al., 2007). For example, the composition of the ammonia-oxidizing community in acidic forest soil was not affected by nitrogen deposition (Jordan et al., 2005; Schmidt et al., 2007). In contrast, growth of ectomycorrhizal fungi in a Norway spruce forest soil was reduced under N deposition (Nilsson et al., 2007). In addition, the functioning of microorganisms was affected by N fertilization with a stimulation of the initial decomposition of cellulose and solubles and a suppression of the decomposition of older humus fractions (Hagedorn et al., 2003).
Whereas the mean chronic annual nitrogen deposition in Europe is still 17 kg N ha−1 year−1 (Stevens et al., 2004), a reduction in emissions of nitrogenous pollutants under the Gothenburg Protocol is presupposed (Power et al., 2006). As one of seven NITREX sites across Europe (see Tietema et al., 1998), the ‘Solling roof project’ was established in 1989 in a Norway spruce forest at Solling, Central Germany, to simulate preindustrial N deposition. Partial deionization of rainfall was established in this field-scale roof experiment in 1991 to investigate whether the effects of N saturation on ecosystem functioning are reversible by decreasing N input (Bredemeier et al., 1995a, b, 1998). Three years after the start of the experiment, microbial biomass (Cmic) within the soil profile was not affected (Raubuch et al., 1999); after 10 years, the mean annual soil respiration rate was 24% higher in the clean rain vs. control plot (Lamersdorf & Borken, 2004). The long-term reduction of nitrogen and proton inputs did not affect nitrous oxide emission, which ranged from 0.25 to 0.41 kg N2O-N ha−1 year−1 in the spruce forest after 10 years of the experiment (Borken et al., 2002). There are reasons to expect that lengthier reduction of N deposition alters the community composition of soil microorganisms in the acid forest ecosystem.
Thus, changes in the amount and composition of needles and fine roots might indirectly affect the density and activity of soil microorganisms, whereas altered N availability and pH might directly affect soil microorganisms under reduced atmospheric nitrogen loads and reduced proton input.
The present study evaluates the possible responses of the total microbial and the nitrate reducer communities within profiles of a spruce forest soil to reduction of chronic N deposition after an experimental duration of 16 years. Nitrate-reducing prokaryotes constitute a wide taxonomic group sharing the ability to produce energy from dissimilatory reduction of nitrate into nitrite, the first step of two different processes: denitrification and dissimilatory reduction of nitrate to ammonium. We used substrate-induced respiration (SIR) to characterize the active fraction of soil microbial biomass and potential nitrate reduction to quantify the activity of nitrate reducers. The densities of total bacteria, nitrate reducers and denitrifiers in different soil layers were analysed by quantitative PCR (qPCR) of the 16S rRNA genes, the nitrate reduction genes, and denitrification genes (Henry et al., 2004, 2006; López-Gutiérrez et al., 2004; Kandeler et al., 2006; Bru et al., 2007). We used the narG and napA genes encoding the membrane-bound and the periplasmic nitrate reductase, respectively, as molecular markers of the nitrate reducer community (Bru et al., 2007). In addition, the nirK, nirS and nosZ genes encoding the copper and cytochrome cd1 nitrite reductase and the nitrous oxide reductase, respectively, were used as molecular markers of the denitrifier community (Philippot, 2005). We hypothesized that reduced N deposition into different horizons of the acid forest soil would modify the densities and activities of the total soil microbial community as well as those of functional microbial communities involved in N cycling.
Materials and methods
Experimental site and soil sampling
The Solling roof project was established in 1989 in a 57-year-old Norway spruce plantation growing on strongly acidic and weakly podzolized loam-silt at the Solling plateau in Central Germany (51°31′N, 9°34′E, elevation c. 500 m above sea level). The climate is dominated by Atlantic streams with evenly distributed precipitation (mean annual precipitation=1090 mm) and a moderate variation in temperature (mean annual temperature=6.4 °C) throughout the year (Lamersdorf & Borken, 2004). The control plot is covered by a translucent roof (300 m2). Throughfall water is permanently collected and immediately resprinkled without any chemical treatment in the control plot. The clean rain plot is also covered with an identical roof. There, throughfall water is partly deionized and resprinkled immediately on the plot since the start of the experiment. Controlling the efficiency of the deionization, Corre & Lamersdorf (2004) reported a reduction of the major elements in the clean rain plot compared with the control plot of 78% for protons, 53% for sulphate, 86% for ammonium and 49% for nitrate. The pH of the throughfall solution at the clean rain plot (pH=5.0) was also higher than that at the control plot (pH=4.4). The Ca and Mg input in the clean rain plot was twice as much as in the control plot during the first 4 years of the roof experiment (1992–1995), but thereafter, the levels were similar between plots (Corre & Lamersdorf, 2004). Soils of the roofed plots are separated from the surrounding area by a vertical plastic foil (Xu et al., 1998).
Four replicate soil samples (‘subplots’) were collected from the clean rain plot as well as from the control plot in late April and late October 2006. Because of the lack of roof replications, the results are based on pseudoreplications. Up to five soil cores (8 cm in diameter) per subplot were taken and mixed according to the horizons Oe, Oa, Ah and Bw. Samples were sieved through a 2-mm sieve (Oe: 5 mm) and stored at −20 °C before analysis.
The Corg and Nt contents of soil were characterized for soil sampled in April, nitrate, ammonium and water contents were analysed for April and October samples (Tables 1 and 2). In spring, the moisture content of soils was much higher (organic layers: +22%, mineral layers: +12%) than for soils sampled in October (Table 2), mainly due to wet periods in March and April 2006 as well as due to a drought period in September and low precipitation in October 2006 (data not shown). The water contents of the clean rain and control plot did not differ significantly in April and October.
Table 1. Chemical soil properties of the acid forest site at Solling, Central Germany, in April 2006 in the control and clean rain plots at different soil depths
OC (g 100 g−1)
NT (g 100 g−1)
Results are given as means ± SE. OC, organic carbon; NT, total nitrogen.
( ± 1.1)
( ± 1.2)
( ± 0.02)
( ± 0.07)
( ± 0.7)
( ± 2.0)
( ± 0.06)
( ± 0.12)
( ± 3.2)
( ± 1.2)
( ± 0.12)
( ± 0.06)
( ± 0.7)
( ± 0.6)
( ± 0.07)
( ± 0.03)
( ± 0.4)
( ± 0.3)
( ± 0.02)
( ± 0.02)
( ± 0.5)
( ± 0.6)
( ± 0.06)
( ± 0.04)
( ± 0.1)
( ± 0.2)
( ± 0.01)
( ± 0.01)
( ± 0.4)
( ± 0.9)
( ± 0.02)
( ± 0.05)
Table 2. NH4+, NO3− and percentage water content of the acid forest site at Solling, Central Germany, in April and October 2006 in the control and clean rain plots at different soil depths
NH4+ (μg N g−1)
NO3− (μg N g−1)
% water content
Results are given as means ± SE.
( ± 30.46)
( ± 29.93)
( ± 1.81)
( ± 35.74)
( ± 5.25)
( ± 0.74)
( ± 0.79)
( ± 0.13)
( ± 0.9)
( ± 1.2)
( ± 1.0)
( ± 3.3)
( ± 14.95)
( ± 26.93)
( ± 15.52)
( ± 37.51)
( ± 3.61)
( ± 0.38)
( ± 5.36)
( ± 1.56)
( ± 1.4)
( ± 0.7)
( ± 2.0)
( ± 4.9)
( ± 0.30)
( ± 1.38)
( ± 1.15)
( ± 7.42)
( ± 0.34)
( ± 0.19)
( ± 1.87)
( ± 0.55)
( ± 0.9)
( ± 1.3)
( ± 0.8)
( ± 1.2)
( ± 0.23)
( ± 0.13)
( ± 0.68)
( ± 1.27)
( ± 0.20)
( ± 0.19)
( ± 2.49)
( ± 2.51)
( ± 0.2)
( ± 0.6)
( ± 1.2)
( ± 0.8)
Soil chemical analyses, microbial biomass and nitrate reductase activity
Organic carbon (OC) and total nitrogen (NT) were measured with a CNS analyzer (Vario MAX, Elementar GmbH, Hanau, Germany) using 250 mg of the organic soils and 800 mg of the mineral soils. Soil pH was measured in a 0.01 M CaCl2 solution (soil to solution ratio 1 : 10 for the organic layers and 1 : 2.5 for the mineral soils). After extraction of inorganic N with 1 M KCl (soil to solution ratio of 1 : 10 for the organic layers and 1 : 5 for the mineral soils), nitrate and ammonium were measured at the SPINMAS [automated sample preparation unit for inorganic nitrogen (SPIN) species coupled to a quadrupole Mass Spectrometer (MAS)] according to Stange et al. (2007).
For the SIR measurement (determination in duplicate), substrate saturation and the maximum initial respiration response were obtained at an amendment rate of 8.0 mg glucose g−1. CO2 evolved was trapped in 50 mM NaOH for a 4-h incubation at 25 °C and measured by titration (Anderson & Domsch, 1978). The release of CO2 was linear over a period of 4 h and the SD of the analytical replicates was <15%.
The potential activity of the nitrate reductase was determined by anaerobic incubation of soil following a modified protocol of Kandeler (1996). The method was based on the determination of the NO2−-N production after adding nitrate as a substrate and 2,4-dinitrophenol as an uncoupler of oxidative phosphorylation that interfered with electron transfer, but allowed nitrate reduction to continue. Substrate as well as inhibitor concentrations were optimized in pre-experiments. In detail, 0.2 g soil was weighed in five replicates into 2.0-mL reaction tubes. Two hundred micrograms of 2,4-dinitrophenol per gram soil (fresh weight) was added to inhibit the nitrite reductases. After a 24-h incubation in 1 mM KNO3 in a total volume of 1 mL at 25 °C in the dark, the soil mixture was extracted with 4 M KCl and centrifuged for 1 min at 1400 g. The accumulated nitrite in the supernatant was determined by a colorimetric reaction. All analytical results were calculated on the basis of the oven-dry (105 °C) weight of soil.
Quantification of 16S rRNA gene, nitrate reductase genes (narG and napA) and denitrification genes (nirK, nirS and nosZ)
DNA was extracted from 0.3 g of soil using the FastDNA Spin Kit for soil (BIO 101, Qbiogene, France), according to the protocol of the manufacturer. Because of the large amounts of PCR-inhibiting substances such as humid acids, an additional purification step with polyvinylpolypyrrolidone-loaded columns (Sigma Aldrich) was performed according to Martin-Laurent et al. (2001). DNA quantity was checked using a BioPhotometer (Eppendorf) at 260 nm. qPCR products were amplified with an ABI Prism 7900 (Applied Biosystems) using SYBR green as the detection system in a 25-μL reaction mixture containing 0.5 μM (each) primer, 12.5 μL of SYBR green PCR master mix (QuantiTect SYBR green PCR Kit; Qiagen, France), 1.25 μL of DNA-diluted template corresponding to 12.5 ng of total DNA and 500 ng of T4gp32 (Qbiogene). The thermal cycling conditions for the 16S rRNA gene and the nirS, nirK and nosZ genes were as described previously (Henry et al., 2004, 2006; López-Gutiérrez et al., 2004; Kandeler et al., 2006). The narG and napA qPCR was performed as described in Bru et al. (2007). Thermal cycling, fluorescent data collection and data analysis were carried out using the ABI Prism 7900 sequence detection system according to the manufacturer's instructions.
Standard curves were obtained with serial plasmid dilutions of a known amount of plasmid DNA containing a fragment of the 16S rRNA gene or the narG, napA, nirK, nirS or nosZ gene. Sequences of the primers and the thermal conditions used for the real-time PCR are given in Supporting Information. Purified soil DNAs were tested for inhibitory effects of coextracted substances by diluting soil DNA extracts and by quantifying by qPCR a known amount of plasmid DNA mixed to soil DNA extracts. In all cases, no inhibition was detected.
If necessary, data were Box-Cox, log or sin transformed before analysis. The influence of reduced N deposition on the chemical and physical soil properties and on SIR for each soil depth was determined using the paired t-test. Two- and three-way univariate anova was applied to test differences of the means of respiration, nitrate reductase activity and gene copy numbers of 16S rRNA gene, narG, napA, nirK and nosZ between depths, treatments and sampling dates. Homogeneity of variances was proved by a Levene test.
Multiple regression analysis was applied to evaluate the relationship between soil environmental factors (water content, OC, NT, NH4+, NO3− and pH) and respiration, nitrate reductase activity, the density of total bacteria and the nitrate-reducing/denitrifying community. Significance was accepted at the P<0.05 level of probability.
Soil chemical properties
The results clearly showed that 16 years of reduced N deposition did not affect the organic C and N pools (Tables 1 and 3). Organic layers (Oe and Oa) were characterized by much higher OC and NT contents than both of the deeper layers (Ah and Bw). Because of high spatial heterogeneity, NO3− contents were not significantly different between the treatments. Long-term reduced N deposition showed a tendency towards higher NH4+ contents in all layers of the clean rain plots compared with the control plots (Table 2). As expected, the contents of inorganic N (NH4+, NO3−) depended on the soil depth (Table 3). Although the roof construction removed the nitrogen and proton input into the ecosystem, the higher pH values in the clean rain treatment were statistically significant only for the Oa horizon. The pH values varied within the soil profile, showing the highest values in the Bw horizon. The soil water content was not affected by N deposition at either sampling date (April and October, Table 2).
Table 3. Influence of reduction of N deposition on chemical soil properties
Effects of treatment and depths were estimated by univariate anova.
P-values ≤0.05 are indicated in bold.
Activity of microorganisms and nitrate reducers
Measurement of SIR yielded estimates of the CO2 released by the soil microbial community under optimal substrate availability. At both sampling dates, reduced N deposition did not significantly affect SIR in the organic layers. In the mineral layers, a significant interaction between treatment and sampling date was observed in the Ah horizon, but not in the Bw horizon (Fig. 1). Sampling date had a significant influence on SIR in the Oe and Bw horizons. The active microbial community strongly decreased with depth (Fig. 1).
Nitrate reductase activity provided an insight into the potential activity of the nitrate reducers. Activity ranged from 0.1 to 1.2 μg N g−1 day−1 and was modified by N deposition, soil depth and sampling time. Whereas reduced N deposition did not influence nitrate reductase in April, significant treatment effects – lower activity of the clean rain treatment – were detected in October (Fig. 2). Generally, organic layers were characterized by a higher activity than the mineral soil layers.
Densities of total bacteria, nitrate reducers and denitrifiers
Samples from the forest site collected at four different soil depths and at two different dates contained amounts of 16S rRNA gene target molecules ranging from 3.8 × 104 to 1.9 × 105 copies ng–1 DNA (Fig. 3). The narG, napA, nirK and nosZ gene copy numbers were two to three logs lower than the 16S rRNA gene. The gene copy number of narG was higher than the other functional genes, with densities ranging from 79 to 1.4 × 103 copies ng–1 DNA, while napA ranged from 23 to 3.3 × 102, nirK from 25 to 9.0 × 102 and nosZ from 18 to 1.9 × 102 copies ng–1 DNA (Figs 4 and 5). The copy numbers of nirS gene fragments were below the detection limit of the nirS qPCR assay (102 copies ng–1 DNA).
Reduction of N deposition did not affect the abundance of 16S rRNA gene copy numbers in the soil profile at any time (Fig. 3). The densities of 16S rRNA genes decreased within the soil profiles, but this effect was less obvious than that for overall soil microbial activity (Figs 1 and 3).
Similar to the 16S rRNA gene, the densities of the nitrate reduction genes (narG and napA) or of the nirK denitrification gene were not affected by reduced N deposition (Figs 4 and 5, Table 4). In contrast, a slight effect was observed for the nosZ gene (Table 4). All functional genes showed a significant depth effect (Table 4). Thus, the narG, napA and nirK gene copy numbers decreased within the soil profile, whereas nosZ showed the highest abundance in the Bw horizon (Figs 4 and 5). A significant sampling date effect was also observed for all genes, except for nosZ (Table 4).
Table 4. Influence of reduced N deposition, soil depth and sampling date on nitrate reductase activity (NRA), SIR and on the density of total bacterial community, nitrate reducers and denitrifiers
P-values for the density and activity of microorganisms analysed by anova are shown.
P-values ≤0.05 are indicated in bold.
Nitrogen × depth
Nitrogen × date
Depth × date
Nitrogen × depth × date
The ratios of the functional genes to 16S rRNA gene from total eubacteria revealed maximum proportions of 0.91% for narG, 0.28% for napA, 0.42% for nirK and 0.05% for nosZ (data not shown). The relative abundance of napA and nirK genes was constant within the soil profile, whereas narG was mainly enriched in the upper horizons and nosZ in deeper soil layers. Therefore, the ratio of nosZ/16S rRNA gene and nosZ/nirK copy numbers increased from about 0.02 and 0.15 in the organic horizons to about 0.16 and 0.51 in the mineral soils, respectively.
Multiple regression analysis including microbiological data as independent factors (Table 5a) revealed no significant correlation between SIR and 16S rRNA gene but between nitrate reductase activity and the copy number of narG (P<0.05).
Table 5a. Multiple regression analysis of size of total and nitrate-dissimilating bacteria using the microbial parameters SIR and nitrate reductase activity (NRA) as independent factors
Standardized coefficient (β)
16S rRNA gene copies
Significant correlations with P-values ≤0.05 are indicated in bold.
Linking soil microbiological to environmental properties
Multiple regression analysis, including soil water content, OC, NT, NH4+, NO3− and pH as independent factors and copy numbers per nanogram DNA, respiration rates and activity rates as dependent factors, was used to relate changes in enzyme activities and gene densities to soil properties in April (Table 5b): SIR was positively related to OC content and negatively related to NT. Nitrate reductase activity depended on the soil water content. None of the soil properties analysed could explain the abundance of 16S rRNA gene, narG or nosZ genes, while the abundance of the napA gene was significantly influenced by the OC content. Multiple regressions also revealed that NT, NH4+ and pH could explain the copy numbers of nirK (Table 5b).
Table 5b. Multiple regression analysis of the size and activity of total* and nitrate-dissimilating bacteria using soil water content, NT, OC, NH4+, NO3− and pH as independent factors
Standardized coefficient (β)
Soil water content
Activity of all microorganisms.
Data from the April sampling date were used for the analysis.
Significant correlations with P-values ≤0.05 are indicated in bold.
NRA, nitrate reductase activity.
16S rRNA gene level
N deposition and soil acidification affect N cycling in many forest ecosystems (Lamersdorf & Borken, 2004). The Solling roof experiment, one of the rare long-term field experiments, was designed to provide an insight into future developments of acidified and nitrogen-saturated forest ecosystems in an envisaged environment of reduced N emissions. Our study, performed 16 years after the start of the experiment, investigated the effects of reduced N deposition on both the total microbial community and on the functional microbial guilds involved in N-cycling.
The total microbial community response to reduced N deposition and soil depth
The impact of reduced deposition was investigated by quantifying the activity and the size of the total microbial community using SIR and qPCR of 16S rRNA gene target molecules. Whereas community size did not respond, SIR temporarily increased in the Ah horizon of the clean rain plot. Because the Ah horizon was characterized by increased growth of fine roots (Lamersdorf & Borken, 2004), we suggest that rhizodeposition stimulated the heterotrophic soil microbial community here. The weak effect of reduced N deposition and proton input can also be explained by enhanced internal nitrogen turnover of soil microorganisms: because long-term reduction of N had a minor effect on inorganic N contents (NH4+, NO3−) and on the active fraction of the soil microbial community, enhanced internal nitrogen turnover might have partly compensated the lower N input by throughfall in the clean rain plot during recent years. This hypothesis is also supported by higher gross mineralization of soil organic nitrogen in that plot (Corre & Lamersdorf, 2004).
SIR showed values up to 40 mg CO2 100 g−1 h−1 in the Oe horizon, in the range of previous results for spruce forests (Kandeler et al., 1999), while the values were 2–4 mg CO2 100 g−1 h−1 in the Bw horizon in the mineral layer. 16S rRNA gene target molecules of the forest site ranged from 3.8 × 104 to 1.9 × 105 copies ng–1 DNA and are in agreement with previous reports (López-Gutiérrez et al., 2004; Kandeler et al., 2006). A significant decrease with soil depth was also observed for the 16S rRNA gene copy numbers (Table 4). However, comparison of SIR and 16S rRNA gene depth profiles revealed a stronger depth effect for the former than for soil microorganism density, suggesting that soil microorganisms were less active in deeper soil layers. Alternatively, this could be due to a higher fungal biomass in Oe because SIR covers the active fraction of both bacterial and fungal biomass while only bacteria are targeted using the 16S rRNA gene.
Multiple regression analyses were used to test whether the results of SIR and 16S rRNA genes depended on specific chemical and physical soil properties (Table 5b). Whereas the active fraction of the soil microbial community (SIR) was significantly related to OC, there was no relationship between 16S rRNA gene copy numbers and substrate pools (NT, OC and Nmin), pH or water content. This suggests that other factors such as temperature, redox potential or distribution of roots might be important.
Activity and size of the nitrate-reducing community within the soil profile
Quantification of potential nitrate reductase activity as well as genes encoding enzymes involved in nitrate reduction was used to assess the response of this specific functional community involved in nitrogen cycling to reduced N deposition. Nitrate reductase in the Solling spruce forest did not exceed 1.5 μg NO2−-N g−1 day−1. These values were about two to 20 times lower than the activities of alpine grasslands (Deiglmayr et al., 2004) and even more than one thousand times lower than the activities of agricultural soils (Philippot et al., 2006).
Lower N deposition significantly affected nitrate reductase activity in October. Thus, lower rates were observed in the clean rain vs. control plot at all depths, but this difference was only significant for the Oa layer (Fig. 2). Because nitrate reductase is an enzyme that is generally induced by nitrate and oxygen limitation, we expected that nitrate in the soil solution as well as moisture content would be important regulating factors. Accordingly, multiple regression analysis revealed that soil water content was related to nitrate reductase activity (Table 5b). However, the soil water content can only explain differences in nitrate reductase activity within the soil profile and not between the two treatments. Reduced N deposition induced a trend towards a lower nitrate concentration that was not significant between the treatments. Therefore, reduced nitrate reductase activity in the clean rain treatment cannot be attributed solely to differences in the nitrate contents of the soil solution, but also to other environmental factors.
The reduction of nitrate to nitrite is catalysed by two different types: a membrane-bound reductase (Nar) encoded by the narG gene and a periplasmic nitrate reductase (Nap) encoded by the napA gene. Nitrate reducers can carry either one or both nitrate reductase enzymes (Philippot & Hojberg, 1999). Whereas Nar has been purified from a large variety of microorganisms including Archaea, the periplasmic enzyme Nap is present only in Gram-negative bacteria. The narG and napA gene copy numbers estimated by qPCR were similar to those described for other ecosystems (López-Gutiérrez et al., 2004; Kandeler et al., 2006; Bru et al., 2007). Reduced N deposition did not modify the size of the nitrate reducer community, whatever the gene targeted (Figs 4 and 5). Therefore, the densities of both nitrate reducers and total bacteria seemed to be buffered against environmental changes resulting from the lower N deposition. This stability of the soil microbial community is also evident in the percentage of nitrate reducers to total bacteria, which was affected neither by N deposition nor by soil depth (data not shown). An alternative explanation is that the between-treatment variability in the chemical compositions of the soil solution was too high to yield differences in the relative abundance or size of the nitrate reducer community. Surprisingly, important regulating factors for the nitrate reducer community – such as organic and inorganic N pools, pH and soil water content – did not correlate with the density of nitrate reducers (Table 5b). On the other hand, univariate anova clearly showed that the densities of nitrate reducers were controlled by spatial and seasonal variation. Our plot-scale study did not account for small-scale heterogeneity of nitrate reducers and its physico-chemical controls. Further studies should clarify whether the size of the nitrate reducer community is driven by the micro-topography of the acid forest soil as described by Mohn et al. (2000) and Hafner & Groffman (2005) for the spatial variation of denitrifying activity and N transformation, respectively.
Quantification of the genes encoding the nitrate reductase and of the nitrate reductase activity in relation to N deposition, soil depth and season allows us to study the possible linkage between the abundance and the activity of the nitrate reducer community in the acid forest soil. We hypothesized that the potential nitrate reductase activity is controlled by the abundance of nitrate reducers under suboptimal environmental conditions. Indeed, Pearson correlation analysis indicated that the density of the narG-carrying community – the more abundant group of nitrate reducers in this forest soil – is positively correlated to nitrate reductase activity (r2=0.487, P<0.001). Linkages between the size and the activity of functional communities involved in the N cycle have been reported previously for nitrifiers or denitrifiers (Patra et al., 2005).
Size of the denitrifier community within the soil profile
Quantification of nirK, nirS and nosZ genes should yield information on the effect of N deposition on the density of denitrifiers capable of reducing the nitrite produced by nitrate reducers into gaseous nitrogen in the acid forest soil. Nitrite is reduced to nitric oxide by microorganisms having either a Cu-containing nitrite reductase enzyme encoded by the nirK gene or a cd1 nitrite reductase encoded by the nirS gene (Zumft, 1997). Nitrous oxide is reduced to N2 by the nitrous oxide reductase encoded by nosZ. We were unable to detect any nirS gene encoding the cytochrome cd1 nitrite reductase, due to the lower sensitivity of the nirS assay compared with the assays of other denitrification genes. Lower N deposition did not affect the copy numbers of nirK and only weakly impacted the copy numbers of nosZ (Fig. 5, Table 4). This is probably due to the relatively constant conditions of soil solution chemistry since more than a decade (e.g. almost no detectable nitrate in the aboveground soil solution since 1995; see Lamersdorf & Borken, 2004). Similarly, no significant intertreatment differences were recorded in the relative abundance of denitrifiers to total bacteria. Mergel et al. (2001b) reported that nitrogen fertilization in an acid forest soil also had no impact on the relative abundance of denitrifiers.
The densities of nirK and nosZ genes were differently influenced by soil depth: whereas copy numbers of the nirK gene decreased within the soil profile, values of nosZ in the Bw layer were higher than those in the organic layers (Fig. 5). Applying the most probable number (MPN) method and colony hybridization using denitrification genes as probes, Mergel et al. (2001a) also showed a depth effect on denitrifier density with decreasing bacterial and denitrifier numbers with soil depth. Although the amount of organic substances is the most important factor determining the size of the denitrifier community in other ecosystems (Tiedje, 1988; Kandeler et al., 2006), we detected no correlation between OC and the number of nirK and nosZ functional genes (Table 5b). The higher nosZ/16S rRNA gene and nosZ/nirK ratios in the Ah and Bw horizons suggest that the proportion of denitrifiers capable of reducing the greenhouse gas N2O into N2 is higher in the mineral soil layer than that in the organic layer.
In conclusion, our study showed that reducing the N deposition had a smaller effect on the abundance and function of soil microorganisms, nitrate reducers and denitrifiers than soil depth and sampling date. The negligible effects of lower N deposition on nitrate reducers and denitrifiers in this acid forest ecosystem are likely due to the low level of nitrogen oxides respiration and the dominance of microbial NH4+ turnover in the internal N cycling. Variations of the nosZ/16S rRNA gene and nosZ/nirK ratios in the different soil horizons suggest that the relative abundance of microorganisms capable of performing complete denitrification is unequally distributed within the acid soil profile. This study provides first evidence that denitrifiers that can reduce nitrous oxide might be enriched in deeper soil layers. Studies on expression of denitrification genes and on N2O fluxes in relation to soil depth are necessary to confirm whether the risk of N2O release from mineral soil layers can be neglected.
We are grateful to Dirk Böttger, David Bru, Elke Feiertag and Sabine Rudolph for technical assistance. We thank our project partner François Buscot for intensive discussions. We thank COST 856 ‘Denitrification in agriculture, air and water pollution’ for funding the short-term scientific mission of Thomas Brune in Dijon as well as the DFG (KA 1590/4-1; PAK 12) for funding our project.