Changes in nitrogen cycling and retention processes in soils under spruce forests along a nitrogen enrichment gradient in Germany

Site description

We selected spruce forest sites from the European Level II program (Intensive Monitoring of Forest Ecosystems) that have similar previous land use, humus and soil types, climatic conditions, and stand characteristics (Table 1). These site factors were considered analogous in selecting our sites, important to minimizing their confounding effects on responses of forest ecosystems to chronic N deposition (Aber et al., 1998; Lovett & Rueth, 1999; Goodale & Aber, 2001). The major difference among our study sites was the N leaching to throughfall N deposition ratios (Table 1).

For the analyses of changes in gross rates of microbial N cycling in the organic horizon, we included three of the NITREX sites (Klosterhede, Gårdsjön, and Solling) and a second Solling site (referred to as the ambient no-roof plot in Corre & Lamersdorf, 2004) that have similar tree species and previous land use as our study sites (Table 2), in order to cover a much wider range of N enrichment gradient. The detailed site characteristics of the Klosterhede- and Gårdsjön-NITREX sites were reported by Gundersen et al. (1998), and the site description of the Solling-NITREX site was taken from Meesenburg et al. (1995) and the Lower Saxony Forest Research Station (2003). The gross rates of soil N cycling for the NITREX sites were measured by Tietema (1998) in 1994 and for the second Solling site by Corre & Lamersdorf (2004) in 2001. There are two main differences in measurement methods employed by Tietema (1998) and by us. First, Tietema (1998) conducted the measurements from disturbed organic horizon samples, while we measured from intact cores of organic horizon. Second, Tietema (1998) calculated the gross transformation rates using a model to fit ¹⁵N enrichments of mineral N pools and the net rates measured periodically within 7-day incubation period, while we used the ¹⁵N pool dilution technique. The Solling-NITREX site and the second Solling site (Table 2) are located adjacent to each other. The throughfall N deposition in Solling had decreased beginning in the 1990s (Corre et al., 2003; Corre & Lamersdorf, 2004), and the throughfall N deposition rates measured from the Solling-NITREX site and the second Solling site were the annual averages from 1986 to 1994 and from 1995 to 2001, respectively (Table 2). Hence, the soil N cycling rates measured from the Solling-NITREX site in 1994 reflect the influence of the high N deposition period (1986–1994), whereas the N cycling rates reported for the second Solling site in 2001 reflect the declining N deposition phase (1995–2001).

Sampling design

We measured gross rates of microbial N cycling, microbial biomass, and other supporting soil parameters in June 2003 and June 2004. Measurements were carried out separately for the organic horizon (combined Oi, Oe and Oa layers) and 0–5 cm mineral soil in June 2003, while only the organic horizon was sampled in June 2004. In each site, an area of 0.5–1 ha has been delineated for long term, N input and output monitoring. Within this area, we selected five sampling points with a distance from 10 to 50 m from each other. At each sampling point and horizon, four undisturbed core samples were taken using stainless-steel cores of 8 cm diameter and 5 cm length. During transport to the laboratory, the cores were placed in wooden cases with built-in fittings to maintain their integrity. For the organic horizon, all measurements were carried out on intact soil cores. For the mineral soil, however, the samples contained 16–27% stone (site averages), making it difficult to inject ¹⁵N solution. The field-moist soil samples were first sieved using a 5 mm sieve before ¹⁵N addition. Samples were incubated at 15° C, which is the average summer soil temperature of the sites.

¹⁵ N pool dilution for the measurement of gross rates of N mineralization and nitrification

We followed the ¹⁵N pool dilution procedures described by Davidson et al. (1991) and Hart et al. (1994b) for ¹⁵N injection into intact cores and subsequent extraction. From the four samples at each sampling point and horizon, two were added with (¹⁵NH₄)₂SO₄ solution (for gross N mineralization and NH₄⁺ immobilization) and the other two with K¹⁵NO₃ solution (for gross nitrification and NO₃⁻ immobilization). Each intact core from the organic horizon received five 1 mL injections, containing 30 μg N mL⁻¹ with 98%¹⁵N enrichment. The same amount was applied evenly on the sieved, mineral soil samples. This was equivalent to a rate of 2.5 (± 0.1) μg N g⁻¹ for the organic horizon samples and 0.9 (± 0.1) μg N g⁻¹ for the mineral soil samples. One sample of each labeled pair was immediately extracted with 0.5 mol L⁻¹ K₂SO₄ (approximately a 5 : 1 ratio of solution to dry mass soil). The time elapsed between ¹⁵N addition and extraction was 10 min (T₀). The T₀ samples were used to correct for the reactions that occur immediately after the addition of ¹⁵NH₄⁺ and ¹⁵NO₃⁻. The other sample of the labeled pair was incubated for 1 day for the ¹⁵NH₄⁺-labeled cores and for 2 days for the ¹⁵NO₃⁻-labeled cores (T₁), and was extracted with K₂SO₄. Extraction was done by shaking the samples for 1 h and filtering the extracts through K₂SO₄-rinsed filter papers. Gross N mineralization and nitrification rates were estimated using the modified calculation procedure of Davidson et al. (1991) from the Kirkham & Bartholomew (1954) model.

N concentration and ¹⁵N analyses

The NH₄⁺ and NO₃⁻ contents of the extracts were analyzed using the same procedures described by Corre & Lamersdorf (2004). Organic N content in the extracts was determined by persulfate digestion (Cabrera & Beare, 1993; Stark & Hart, 1996), which involves oxidation of NH₄⁺ and organic N to NO₃⁻ while NO₃⁻ remains unchanged. In short, 10 mL of the soil extract was combined with 10 mL of an oxidizing reagent (50 g K₂S₂O₈, 30 g H₃BO₃, and 100 mL of 3.75 mol L⁻¹ NaOH in a 1 L solution), autoclaved for 1 h at 120° C, and the digest was made to a final volume of 100 mL by adding deionized-distilled water. The N concentration of the persulfate digests was analyzed using continuous flow injection colorimetry (copper–cadmium reduction method; Cenco/Skalar Instruments, Breda, the Netherlands). Extractable organic N is the difference between persulfate-N and NH₄⁺+ NO₃⁻ concentrations.

To identify the fates of added ¹⁵N at T₀, part of the T₀ extracts was used for serial diffusion of NH₄⁺ and NO₃⁻ (Corre & Lamersdorf, 2004) and part was used for persulfate digestion for determination of ¹⁵N enrichment in the extractable organic N pool. For the extractable organic N pool, 50 mL of the persulfate digests was added with 2 mL of 10 mol L⁻¹ NaOH (raising the pH to > 13) and left open for 5 days to eliminate any residual NH₄⁺. Another 2 mL of 10 mol L⁻¹ NaOH (maintaining the pH at > 13) and Devarda's alloy were added, converting persulfate-N (in NO₃⁻ form) to NH₃ which is subsequently trapped in Teflon (polytetrafluoroethylene)-encased acidified disks (Stark & Hart, 1996). The same diffusion procedure and blank correction were followed as described in our previous works (Corre et al., 2003; Corre & Lamersdorf, 2004). We calculated the ¹⁵N enrichment of the extractable organic N pool based on isotope mixing equation using the difference in ¹⁵N enrichments and N concentrations between the persulfate-N and NH₄⁺+ NO₃⁻ pools. For the T₁ samples, only NH₄⁺ was diffused from the ¹⁵NH₄⁺-labeled samples (for gross N mineralization) and only NO₃⁻ from the ¹⁵NO₃⁻-labeled samples (for gross nitrification). Part of the T₁ extracts was reserved for the microbial N immobilization assay. ¹⁵N was analyzed using isotope ratio mass spectrometry (Finigan MAT, Bremen, Germany).

Estimation of NH₄⁺ and NO₃⁻ immobilization rates and microbial biomass carbon (C) and N by chloroform fumigation

We used the T₁¹⁵NH₄⁺- and ¹⁵NO₃⁻-labeled samples to assess NH₄⁺ and NO₃⁻ immobilization rates, respectively, as described in our previous works (Corre et al., 2003; Corre & Lamersdorf, 2004). About 25 g of the T₁¹⁵NH₄⁺- and ¹⁵NO₃⁻-labeled samples were fumigated with CHCl₃ for 5 days and extracted with 0.5 mol L⁻¹ K₂SO₄ (approximately a 5 : 1 ratio of solution to dry mass soil). From the fumigated T₁ extracts and the corresponding unfumigated T₁ extracts, extractable organic N and ¹⁵N enrichment were determined using persulfate digestion and the diffusion procedures described above. NH₄⁺ and NO₃⁻ immobilization rates were calculated using the nonlinear model described by Davidson et al. (1991).

Microbial biomass C and N were determined from undisturbed core samples taken from the same sampling point and layer. We used the fumigation–extraction method (Brookes et al., 1985; Davidson et al., 1989) and following the same procedure described in our previous works (Corre et al., 2003; Corre & Lamersdorf, 2004). Organic C from the extracts was analyzed by ultraviolet (UV)-enhanced persulfate oxidation using a Dohrmann DC-80 Carbon Analyzer with an infrared detector (Rosemount Analytical Division, CA, USA). Organic N was determined using persulfate digestion described earlier. Microbial biomass C and N are calculated as the difference in extractable organic C and persulfate-N between fumigated and unfumigated soils divided by k_C= 0.45 (Joergensen 1996) and k_N= 0.68 for 5-day fumigated samples (Shen et al., 1984; Brookes et al., 1985).

Calculation of mean residence time (MRT)

The MRT indicates the average length of time an N atom resides in a given pool; a lower MRT indicates a faster pool turnover rate and hence a more dynamic pool. The calculation of MRT (N pool÷flux rate; e.g. microbial N_MRT= microbial N pool÷total N immobilization rate) assumed that the NH₄⁺, NO₃⁻, and microbial biomass N pools were at steady state and that the fluxes were equal to gross rates of N mineralization, nitrification and total N (NH₄⁺+ NO₃⁻) immobilization, respectively.

Other supporting parameters

NO fluxes (Table 1) were measured in the field in June 2003, at the same time when soil samples were taken for the first measurement of soil N cycling rates. We used a similar dynamic chamber and analytical methods employed by Veldkamp et al. (1998). NO was analyzed with a Scintrex LMA-3 chemiluminescence detector (Scintrex, Ontario, Canada) after conversion to NO₂ by a CrO₃ catalyst. N₂O fluxes (Table 1) and CO₂ fluxes were measured from undisturbed soil cores taken from the organic layer in June 2004, at the same time when samples were taken for the second measurement of soil N cycling rates. The soil cores were incubated for 30 min in 1 L glass incubation vessels with a gas sampling port fitting on the lid. Gas samples were analyzed for N₂O and CO₂ using a gas chromatograph (GC 14A; Shimadzu, Duisburg, Germany) equipped with an electron capture detector (Loftfield et al., 1997). Fluxes were calculated from the increase in N₂O and CO₂ concentrations during the incubation period minus the background N₂O and CO₂ concentrations (incubation vessel without a soil core). The CO₂ evolution was used for the calculation of total C utilized by microbes ([microbial C : N ratio × total N immobilization rate] + CO₂-C evolution rate), which we used as an index of available C similar to that of Schimel (1988) and Hart et al. (1994a). Applying the formula used by the same authors, we estimated microbial growth efficiency as the amount of C assimilated into microbial biomass [microbial C : N ratio × total N immobilization rate] divided by total C utilized.

Soil characteristics (Table 3) were determined at the beginning of the study (June 2003). Total soil organic C and N were measured from air-dried, ground samples using a CNS Elemental Analyzer (Elementar Vario EL, Hanau, Germany). In the organic layer, total Ca, Mg and Al contents were determined from air-dried, ground samples, digested under high pressure in concentrated HNO₃, and the digests were analyzed for element contents using an Inductively Coupled Plasma-Atomic Emission Spectrometer (ICP-AES) (Spectro Analytical Instruments, Kleve, Germany). In the mineral soil, exchangeable Ca, Mg and Al contents and cation exchange capacity (from which base saturation was calculated) were determined from air-dried, 2 mm sieved samples, percolated with 1 m NH₄Cl, and the percolates were analyzed for element contents using a Flame-Atomic Absorption Spectrometer (Varian, Darmstadt, Germany).

Statistical analyses

We first tested the spatial independence of our sampling points using the rank version of von Neumann's ratio test (Bartels, 1982). This test can be carried out when the sampling points have uniform distance. In the June 2003 sampling, we selected one of our study sites that have sampling points regularly spaced at 10 m and tested their spatial independence using the data on gross N mineralization rates. We found that these sampling points were spatially independent. The sampling points from the rest of our study sites and in the June 2004 sampling were spaced even farther, up to 50 m apart. Based on the results of the spatial independence test, we considered the sampling points at each site as replicates in the succeeding analyses. Tests for normality using Kolmogorov–Smirnov D statistic and for equality of variance using Levene statistic (Sokal & Rohlf, 1981) were first conducted for each parameter. When parameters were heterogeneous, they were log transformed before analysis. Analyses were carried out using one-way analysis of variance (anova) for soil characteristics (Table 3) and microbial N cycling in the 0–5 cm mineral soil and using two-way anova (with sites and sampling periods as factors) for microbial N cycling in the organic horizon. Multiple comparisons of treatment effects were conducted using the least significant difference test at P≤0.05. Correlation analyses were conducted on the mean values of each site (for which N is the number of sites) using Pearson's correlation test.

Space-for-time substitution approach: site characteristics

For our present study sites, throughfall N deposition and N leaching were not correlated with mean annual precipitation and temperature. NH₄⁺ and dissolved organic N (DON) were similar across sites but NO₃⁻ increased with increasing leaching losses (Table 1). N leaching was correlated with throughfall N deposition (R= 0.83, P= 0.00, N= 5). The sites have similar soil chemical characteristics: acidic soil pH (≤4) and very low base saturation (≤8%) in which Al dominated over Ca and Mg in the exchange complex (Table 3). With respect to the soil biochemical properties, although the sites differed in total C, total N and total C : N ratios, these were not correlated with the throughfall N deposition and leaching losses (Table 3). Regarding biochemical measures that may reflect the more active microbial processes (e.g. N oxide emissions), NO and N₂O fluxes corroborated our index of N enrichment (Table 1) – fluxes increased across the N enrichment gradient. Volumetric moisture contents at the time of measurement did not differ across sites: 36 ± 4% to 40 ± 4% for the organic horizon and 36 ± 3% to 43 ± 6% for the 0–5 cm depth mineral soil.

¹⁵ N recovery 10 min (T₀) after ¹⁵N addition

From the ¹⁵NH₄⁺-labeled samples of the organic horizon, 82 ± 2% (averaged across sites and sampling periods, N= 50) was recovered in the NH₄⁺ pool and there was no difference detected among sites and sampling periods (Fig. 1). ¹⁵N recoveries in the NO₃⁻ pool and extractable organic N pool (on average, 0.7 ± 0.3% and 1.0 ± 0.6%, respectively) were not significantly different from zero (one-sample t-test, P > 0.05). We did not measure total ¹⁵N recovery from the whole soil. Assuming that what we did not recover from the extractable N pools remains in the sample as nonextractable (or insoluble) N, this fraction constituted 18% of added ¹⁵NH₄⁺. For the mineral soil, ¹⁵NH₄⁺ recoveries differed among sites, but its trend did not follow the N enrichment gradient. No ¹⁵N above the background level was detected in the NO₃⁻ pool and extractable organic N pool (Fig. 1). The fraction that presumably remained in the mineral soil as insoluble N form ranged from 0% (Idar–Oberstien and Heidelberg) to 36% (Adenau) of the added ¹⁵NH₄⁺.

From the ¹⁵NO₃⁻-labeled samples, ¹⁵N recoveries in the NH₄⁺ pool at both depths were negligible. ¹⁵NO₃⁻ recoveries increased with the N enrichment gradient, while ¹⁵N recoveries in the extractable organic N pool decreased with the N enrichment gradient (Fig. 2). In all sites, ¹⁵N recoveries in the extractable organic N pool did not differ (paired-samples t-test at P > 0.05) between the organic horizon (15 ± 2% of injected ¹⁵NO₃⁻) and mineral soil (11 ± 2% of injected ¹⁵NO₃⁻). If we assume that what we did not recover in the extractable N pools remained in the samples as insoluble organic N, this contributed on average 35% of the added ¹⁵NO₃⁻ in the organic layer and 42% in the mineral soil.

Gross rates of NH₄⁺ transformation and MRT of NH₄⁺ and microbial N pools

For all the processes measured, there was no difference detected between sampling periods (June 2003 and June 2004) for the organic horizon. Gross N mineralization rates in the organic horizon (Fig. 3) increased up to the intermediate N enrichment level (i.e. an N leaching :throughfall ratio of 0.26/Heidelberg), followed by a downturn in the highly N-enriched sites (i.e. N leaching : throughfall ratio of≥0.42/Solling). Microbial NH₄⁺ immobilization rates showed a similar pattern, although we did not detect significant differences among sites (Fig. 3). For mineral soil, gross rates of N mineralization and NH₄⁺ immobilization were much lower and no difference was detected among sites (Table 4).

For our present study sites (low to intermediate N enrichment), gross N mineralization and NH₄⁺ immobilization were neither related to total C, N and C : N ratio nor to microbial biomass (Table 5). However, the increasing gross N mineralization rates in the organic layer across low to intermediate N enrichment were paralleled with increasing available C (Table 5). The increase in available C in the organic layer also corresponded to a decrease in MRT (fast turnover) of the NH₄⁺ pool (Table 5 and Fig. 4) and the microbial N pool (Fig. 5). In mineral soil, the MRT of NH₄⁺ pools (Fig. 4) and microbial N (Table 4) also declined slightly across our present study sites gradient.

NH₄⁺ immobilization was highly correlated with gross N mineralization across the entire N enrichment gradient (Table 5). NH₄⁺ immobilization rates increased proportionately with increasing gross N mineralization rates (error bars overlap; Fig. 3), but when gross N mineralization rates declined in the highly N-enriched site NH₄⁺ immobilization rates tended to be lower than gross N mineralization rates (error bars do not overlap; Fig. 3). Data from the previous NITREX study (Tietema, 1998) did not contain an error term; therefore, interpreting NH₄⁺ immobilization rates across this N gradient was not determinable with a significant degree of surety.

For our present study sites, microbial C and N in the organic horizon slightly increased toward the site with a leaching : throughfall ratio of 0.16/Adenau (Table 6) followed by a decrease. In mineral soil, these markedly decreased as the sites become more N enriched (Table 4). Microbial C and N were not related to total C and N in both depths. The pattern for microbial C : N ratio in the organic horizon did not follow the N enrichment gradient (Table 6) but was related to pH (R=−0.88, P= 0.05, N= 5). For the organic horizon, sites with pH 3.8–4.0 (Table 3) showed lower microbial C : N ratios (Table 6) than sites with pH≤3.7. The microbial C : N ratios in the mineral soil, with uniformly low pH (pH≤3.7), clearly decreased across the N enrichment gradient (Table 4).

Gross rates of NO₃⁻ transformation and MRT of the NO₃⁻ pool

Gross rates of NO₃⁻ transformation were three times to an order of magnitude lower than the NH₄⁺ transformation rates (Table 4 and 3, 6). In the organic horizon, there were no differences detected in the gross rates of NO₃⁻ transformation between sampling periods (June 2003 and June 2004). For the mineral soil, the rates were much lower; however, no difference was detected among sites (Table 4). Gross nitrification and NO₃⁻ immobilization rates in the organic horizon (Fig. 6) increased up to the intermediate N enrichment level (i.e. an N leaching : throughfall ratio of 0.26/Heidelberg) and somewhat declined in the highly N-enriched sites (i.e. N leaching : throughfall ratio of ≥0.42/Solling). The negligible NO₃⁻ immobilization reported for the Solling-NITREX site caused a sharp decline in the curve fitting for NO₃⁻ immobilization (Fig. 6). Nitrate immobilization rates were higher than gross nitrification rates (paired-samples t-test at P < 0.05; Fig. 6) in the sites with N leaching : throughfall ratios of 0.06–0.16 (where nitrification was low) and 0.42 (where nitrification showed a decline).

Gross nitrification was positively correlated to gross N mineralization across the entire N enrichment gradient (Table 5). The increase in gross nitrification covaried with a decrease in NH₄⁺ pool MRT and an increase in available C across our present study sites (Table 5). NO₃⁻ immobilization was correlated with gross nitrification across the entire N enrichment gradient (Table 5). Because NO₃⁻ immobilization and gross nitrification covaried, they showed similar correlations with gross N mineralization and MRT of the NH₄⁺ pool. The correlation between NO₃⁻ immobilization and NH₄⁺ immobilization (Table 5) reflected their similar pattern across the N enrichment gradient (3, 6). NO₃⁻ immobilization and available C share a common parameter in their calculation and their correlation may be a methodological artifact.

The MRT of the NO₃⁻ pool increased across our present study sites (Fig. 7). Unlike that of NH₄⁺ pool MRT, we observed no correlation between NO₃⁻ pool MRT and available C (Table 5). Instead, we observed an inverse relationship in the organic horizon between NO₃⁻ pool MRT and the T₀¹⁵N recovery in extractable organic N from ¹⁵NO₃⁻-injected cores across our present study sites (R=−0.96, P= 0.01, N= 5).

Space-for-time substitution to represent N enrichment gradient

The extensive review of Aber et al. (1998) showed that the rates at which forest ecosystems progress along the N enrichment continuum are regulated by two main factors: inherent N status of the system (e.g. N supply and demand) and rate of N deposition or N addition. The inherent N status is in turn influenced by forest type (e.g. broadleaf vs. coniferous forests; Kristensen et al., 2004; Magill et al., 2004), land use history and management (Goodale & Aber, 2001; Compton & Boone, 2002), and soil and organic horizon characteristics (e.g. C : N ratio; Dise et al., 1998; Emmett et al., 1998; Macdonald et al., 2002; Borken & Matzner, 2004). We did not observe any correlation between leaching losses and C : N ratios of the organic horizon because our sites encompassed only a narrow range of organic horizon C : N ratios (between 22 and 26; Table 3). Our present sites have generally similar characteristics (stand, climatic, and soil chemical properties), and they mainly differed in throughfall N deposition and losses. These qualities support the implicit assumption of the space-for-time substitution approach that the sites should have similar characteristics such that changes in soil N processes are a consequence of their exposure to chronic and different rates of N deposition. Our index of N enrichment (N leaching : throughfall N ratio) indicates the degree to which sites are able to retain their respective N deposition, and sites should then reflect changes in soil N retention processes over the N enrichment gradient.

Implications of ¹⁵N recovery at T₀ on abiotic N immobilization

From the ¹⁵NH₄⁺-labeled samples, ¹⁵NH₄⁺ recoveries from both depths were comparable with our previous findings from similar forest soils in Solling, Germany (Corre et al., 2003; Corre & Lamersdorf, 2004). In the organic layer, a small fraction of injected ¹⁵NH₄⁺ was presumably retained as an insoluble N form. Such fast incorporation (within 10 min) of added ¹⁵NH₄⁺ into an insoluble N pool has been previously equated to abiotic N immobilization (Fitzhugh et al., 2003) and has been attributed to physical condensation with phenolic compounds in humus (Nömmik, 1970) and clay fixation (Davidson et al., 1991). All sites showed a similar reaction to the added ¹⁵NH₄⁺, which suggests that abiotic NH₄⁺ retention may not be affected by N status. A similar result was reported by Johnson et al. (2000) from a variety of forest types with differing N deposition rates (from 0.3 to 27 kg N ha⁻¹ yr⁻¹), whereby abiotic NH₄⁺ immobilization was unrelated to soil N status. In mineral soil, abiotic immobilization of added ¹⁵NH₄⁺ can be due to both fixation on clay minerals and reaction with organic compounds. The ¹⁵NH₄⁺ recoveries were generally comparable between the organic and mineral soil in each site (paired-samples t-test, P > 0.05), except in the Adenau site (Fig. 1). The low ¹⁵NH₄⁺ recovery in the mineral soil of the Adenau site may indicate higher NH₄⁺ fixation by the clay minerals. The differences in ¹⁵NH₄⁺ recoveries in mineral soil possibly reflect differences in clay mineralogy, and hence the pattern was not related to the N enrichment gradient.

On the other hand, rapid (within 10–15 min) disappearance of added ¹⁵NO₃⁻ from the NO₃⁻ pool has been equated and further verified to be due to abiotic NO₃⁻ immobilization (Berntson & Aber, 2000; Dail et al., 2001). We observed a decrease in the abiotic NO₃⁻ immobilization (rapid conversion of added ¹⁵NO₃⁻ to extractable organic N) with the N enrichment gradient. Berntson & Aber (2000) first reported that abiotic NO₃⁻ immobilization from a coniferous forest soil decreased after 11 years of high N addition. This raised an important issue of whether such a change in abiotic NO₃⁻ immobilization resulting from a high N fertilization rate is similar to the impact caused by relatively low, but chronic atmospheric N deposition to ecosystems. To our knowledge, our study is the first to affirm that such a decrease in abiotic NO₃⁻ immobilization can also occur under an ambient N enrichment gradient. Davidson et al. (2003) put forward a hypothesis explaining the abiotic immobilization of NO₃⁻ to organic N. They hypothesized that DOC drives this reaction: DOC reduces Fe(III), producing reactive Fe(II) species that reduce NO₃⁻ to NO₂⁻ in anaerobic microsites, which in turn reacts readily and abiotically with soil organic matter (Smith & Chalk, 1980; Azhar et al., 1986; Thorn & Mikita, 2000). The NO₂⁻ reactions with C-containing compounds may depend on the forms of reactive C, which may affect whether the ¹⁵N is recovered in soluble or insoluble organic N. For example, the abiotic NO₃⁻ reaction to soluble and insoluble organic N pools differed among temperate forests with differing N status: only 1% of added ¹⁵NO₃⁻ (from intact soil cores at 10 min) was detected in extractable organic N from an N-saturated spruce forest (Corre & Lamersdorf, 2004), 30% in extractable organic N and 5% (Dail et al., 2001) to 7–10% (Fitzhugh et al., 2003) in insoluble N (from disturbed samples at 15 min and 1-day incubation) from low N deposition (9–11 kg N ha⁻¹ yr⁻¹) deciduous forests, and 11% in extractable organic N and 37% in the insoluble N (at 0.1 day of ¹⁵NO₃⁻ addition in the field) from an unpolluted/N-limited evergreen mixed-angiosperm forest (Perakis & Hedin, 2001). There is a well-developed literature on the effects of increasing N inputs to changes in structure and decomposition in soil organic matter (Berg & Matzner, 1997; Aber et al., 1998; Neff et al., 2002), which may result to differing reactive C forms in forests with differing degrees of N enrichment. These studies and our present study foster an increasing awareness of the significance of abiotic NO₃⁻ immobilization in forest soils. The change in abiotic NO₃⁻ immobilization as forests progress along the N enrichment continuum may influence the NO₃⁻ retention capacity of forest soils.

Changes in and controls of NH₄⁺ transformation rates

For the organic horizon (Fig. 3), the low-N sites (i.e. an N leaching : throughfall ratio from 0.04 to 0.13) showed comparably low gross N mineralization and NH₄⁺ immobilization rates as has been observed for an N-limited coniferous forest in California (Davidson et al., 1992) and a pine forest at the Harvard Forest site (Venterea et al., 2004). At the intermediate N enrichment level (i.e. an N leaching : throughfall ratio of 0.26) with increased gross NH₄⁺ transformation rates, these rates were higher than those reported for California coniferous and Harvard pine forests. While it has been shown by long-term, high N fertilization that gross N mineralization tended to decrease (Corre et al., 2003; Venterea et al., 2004), our data were the first, to our knowledge, to show that a decrease in gross N mineralization rates can occur under ambient, highly N-enriched conditions (i.e. an N leaching : throughfall ratio of≥0.42). Our study provides compelling evidence of mechanistic changes in soil N cycling under an ambient N enrichment gradient.

For our present study sites (low to intermediate N enrichment), gross rates of NH₄⁺ transformation were not related to total C, N, C : N ratio and microbial biomass. Bulk soil C and N levels, which are a cumulative effect of the historic and recent N deposition, may not reflect the microbially labile C and N fractions. Also, the pattern of microbial biomass across sites (Tables 4 and 6) was not in step with the pattern of NH₄⁺ transformation rates (Fig. 3). The correlation between gross N mineralization rates and available C across low to intermediate N enrichment (Table 5) attested for improved substrate quality in response to increasing N enrichment. The increase in available C, in turn, drove faster turnover rates of the NH₄⁺ pool (Table 5 and Fig. 4) and microbial N pool (Fig. 5), as was demonstrated by Hart et al. (1994a). Our results demonstrated that across low to intermediate N enrichment the parallel increase in gross N mineralization and available C drove a corresponding increase in microbial immobilization of NH₄⁺; this combined with increased turnover rates of the microbial N pool would contribute to efficient retention of NH₄⁺. Furthermore, while abiotic retention of NH₄⁺ is of only minor importance (as shown above), microbial assimilation and turnover appeared to contribute largely to its retention. Such retention of N via microbial assimilation can be supported by the available C; our estimate of microbial growth efficiency in the organic horizon was on average 0.65 ± 0.18 (N= 25 for the June 2004 sampling of our present study sites), which was in the range of those reported for coniferous forests (Hart et al., 1994a; Stark & Hart, 1997).

For the highly N-enriched Solling sites, reduced gross N mineralization rates and much more reduced NH₄⁺ immobilization rates were attributed to a decrease in available C, microbial biomass (Corre et al., 2003), and DOC (Corre & Lamersdorf, 2004), which point to reduced microbial decomposition and altered soil organic matter structure (Berg & Matzner, 1997; Aber et al., 1998). In particular, the Solling-NITREX site (an N leaching : throughfall ratio of 0.45) showed build-up of organic matter stock from 49 ± 4 Mg ha⁻¹ (C/N = 25; Matzner, 1989) in 1968 (with throughfall N deposition of 43 kg N ha⁻¹ yr⁻¹ between 1969 and 1994) to 121 ± 8 Mg ha⁻¹ (C/N = 26) in 2001 (with throughfall N deposition of 35 kg N ha⁻¹ yr⁻¹ between 1995 and 2001; Meesenburg et al., 1999; Meiwes et al., 2002). This increase of organic matter stock was attributed to the retardation effect of excessive N levels on lignin degradation and on decomposition of recalcitrant organic matter (Matzner, 1989; Berg & Matzner, 1997; Berg & Meentemeyer, 2002). Other studies on chronic, high N fertilization consistently showed reduced microbial biomass (Compton et al., 2004; DeForest et al., 2004), decreased fungal : bacterial biomass and activity ratios (Frey et al., 2004; Wallenstein et al., 2006), and reduced β-glucosidase (a cellulose-degrading enzyme) and phenol oxidase activities (a lignin-degrading enzyme produced by white-rot fungi) (Carreiro et al., 2000; Saiya-Cork et al., 2002; Sinsabaugh et al., 2002; DeForest et al., 2004; Frey et al., 2004). We observed an apparent decrease in microbial C : N ratio in the mineral soil across our present study sites, which possibly indicated decreased fungal : bacterial biomass ratio. Our results and these aforementioned studies indicated that highly N-enriched conditions depress microbial decomposition, and ultimately reduce N mineralization, as a result of changes in microbial biomass, microbial community structure, and enzyme activity.

Changes in and controls of NO₃⁻ transformation rates

For the organic horizon (Fig. 6), the N-limited NITREX sites had similarly low NO₃⁻ transformation rates as the N-limited Harvard pine forest (Venterea et al., 2004), and the sites with N leaching : throughfall ratios from 0.06 to 0.16 and 0.42 showed comparable intermediate rates with the 13-year N fertilized Harvard pine forest (Venterea et al., 2004) and the coniferous forests in New Mexico (Stark & Hart, 1997). In these sites NO₃⁻ immobilization rates were higher than the gross nitrification rates, which was the same case for these cited studies. Our data indicated that in sites with extremely low to intermediate nitrification rates (which also had low to intermediate N mineralization rates), NO₃⁻ (and NH₄⁺) produced in the soil may be insufficient to meet microbial demand for N that microbial assimilation was stimulated by the injected ¹⁵NO₃⁻. The mechanisms resulting in substantial NO₃⁻ assimilation by microbes have been related to the spatial compartmentalization of NH₄⁺, NO₃⁻ and C availability. NH₄⁺ depletion may occur at microsites that have high C availability, and the greater diffusion rate of NO₃⁻ relative to NH₄⁺ may then lead to significant NO₃⁻ assimilation (Davidson et al., 1992; Stark & Hart, 1997). This is supported by our observations where stimulated NO₃⁻ immobilization occurred only from the undisturbed cores of the organic horizon, presumably having high microsite heterogeneity, and not from the mixed samples of mineral soil (Table 4).

On the other hand, in sites with the highest gross nitrification rates (i.e. N leaching : throughfall ratios of 0.23 and 0.26, also with the highest N mineralization rates), NO₃⁻ immobilization rates did not exceed gross nitrification rates (Fig. 6). Our results further implied that while acid spruce soils are poised to immobilize NO₃⁻, chronic high NO₃⁻ deposition on already highly nitrifying sites may possibly result in NO₃⁻ excess relative to microbial demand for N. This was depicted by an increasing MRT of the NO₃⁻ pool (or NO₃⁻ accumulation) across the N enrichment gradient (Fig. 7). The increasing NO₃⁻ deposition and MRT of the NO₃⁻ pool, which signal high potential for N losses in times of high soil water content, possibly contributed to increasing NO₃⁻ leaching and N trace gases emissions across the N enrichment gradient (Table 1). This was the case for the Solling site with an N leaching : throughfall ratio of 0.42; gross nitrification rates (and gross N mineralization rates) had somewhat declined in this site, while microbial NO₃⁻ immobilization rates remained high (Fig. 6). However, the extremely high N deposition, combined with low abiotic NO₃⁻ retention, in this site resulted in high NO₃⁻ pool MRT and high NO₃⁻ leaching (Corre & Lamersdorf, 2004).

The correlations of gross nitrification with gross N mineralization and NH₄⁺ pool MRT might indicate that nitrification in these sites depended on availability of NH₄⁺ as substrate (autotrophic nitrification). The correlation between gross nitrification and available C also implied a possible conversion of organic substrate to NO₃⁻ with simultaneous release of available C (heterotrophic nitrification). De Boer & Kowalchuk (2001) have discussed in detail that acid-tolerant autotrophic nitrifiers and heterotrophic fungal nitrifiers can operate concurrently in acid forest soils. On the other hand, the similar pattern of NO₃⁻ and NH₄⁺ immobilization across the N enrichment gradient precluded the assumption of microbial preference for NH₄⁺, but showed that NH₄⁺ and NO₃⁻ are immobilized simultaneously to meet microbial demand for N. Furthermore, the empirical relationship between NO₃⁻ pool MRT and abiotic conversion of NO₃⁻ to extractable organic N may indicate ecological importance, that the decrease in abiotic NO₃⁻ immobilization across the N enrichment gradient (Fig. 2) has in part contributed to the increase in NO₃⁻ pool MRT (Fig. 7) and NO₃⁻ losses (Table 1).