Fungal and bacterial growth responses to N fertilization and pH in the 150-year ‘Park Grass’ UK grassland experiment

Authors


  • Editor: Christoph Tebbe

Correspondence: Johannes Rousk, Environment Centre Wales, Bangor University, LL57 2UW Bangor, UK. Tel.: +44 756 390 6829; fax: +44 1248 354997; e-mail: j.rousk@bangor.ac.uk

Abstract

The effects of nitrogen (N) fertilization (0–150 kg N ha−1 year−1 since 1865) and pH (3.3–7.4) on fungal and bacterial growth, biomass and phospholipid fatty acid (PLFA) composition were investigated in grassland soils from the ‘Park Grass Experiment’, Rothamsted Research, UK. Bacterial growth decreased and fungal growth increased with lower pH, resulting in a 50-fold increase in the relative importance of fungi between pH 7.4 and 3.3. The PLFA-based fungal : bacterial biomass ratio was unchanged between pH 4.5 and 7.4, and decreased only below pH 4.5. Respiration and substrate-induced respiration biomass both decreased three- to fourfold with lower pH, but biomass concentrations estimated using PLFAs were unaffected by pH. N fertilization did not affect bacterial growth and marginally affected fungal growth while PLFA biomass marker concentrations were all reduced by higher N additions. Respiration decreased with higher N application, suggesting a reduced quality of the soil organic carbon. The PLFA composition was strongly affected by both pH and N. A comparison with a pH gradient in arable soil allowed us to generalize the pH effect between systems. There are 30–50-fold increases in the relative importance of fungi between high (7.4–8.3) and low (3.3–4.5) pH with concomitant reductions of respiration by 30–70%.

Introduction

Increasing nitrogen (N) deposition loads have been recognized as an important factor directly influencing belowground carbon (C) cycling, including the soil microorganism activity (Knorr et al., 2005; Liu & Greaver, 2010). Moreover, the net primary productivity of most temperate terrestrial ecosystems is N-limited (Vitousek & Howarth, 1991), and consequently increased N loading will lead to increased C inputs into the systems in the form of plant biomass and plant-derived C efflux (Liu & Greaver, 2010). Application of N will also have effects on the plant community composition (Bobbink et al., 1998; Clark et al., 2007) and diversity (Stevens et al., 2004; Suding et al., 2005). Because both altered input of C to the soil and an altered plant community will affect the often C-limited (Demoling et al., 2007) saprotrophic microorganisms in soil systems, soil microbial activity may also be indirectly affected by N fertilization.

Although there have been several studies on the effects of N fertilizers on the microbial community, these have mostly been restricted to the outcome of heterotrophic microbial activity measured as respiration (e.g. Carreiro et al., 2000; Saiya-Cork et al., 2002; Ramirez et al., 2010; Rifai et al., 2010) or as indicated by the composition of the microbial biomass or biomass components (including gene frequencies and potential enzyme activities) in soil (DeForest et al., 2004; Gallo et al., 2004; Waldrop et al., 2004; Sinsabaugh et al., 2005; Nemergut et al., 2008). However, evidence is accumulating to indicate that the connections between the microbial biomass concentrations and their active contribution to ecosystem processes are at best loosely connected (Rousk & Bååth, 2007; Kemmitt et al., 2008; Rousk et al., 2009b) and that respiration and microbial growth can be uncoupled (Rousk et al., 2008, 2009b, 2010c) in soil. Consequently, the effect of N fertilization on the active heterotrophic microorganisms and their growth remains largely unassessed.

Biomass-based assessments on the influence of N fertilization on the soil microbial community have indicated a connection between a decrease in fungi and higher N additions in both grassland (e.g. Bardgett et al., 1999; de Vries et al., 2006, 2007) and, especially, forest systems (e.g. Högberg et al., 2003; Demoling et al., 2008). It should be noted, however, that the presence of ectomycorrhizal fungi in forest soils could confound the estimate of saprotrophic fungi using specific phospholipid fatty acids (PLFAs) or ergosterol as biomass proxies. Ectomycorrhizal fungi appear to be very sensitive to increased N conditions in soil (Nilsson et al., 2005, 2007; Gillet et al., 2010). The scarce assessments of the influence of N fertilization on the microbial community based on growth measurements also complicate this picture. While the monitoring of 13C-root exudate uptake into PLFA markers in a grassland soil revealed a more reduced fungal activity compared with bacterial activity following N application, the 13C incorporation did not coincide with a similar development of the concentration of the saprotrophic fungal biomass markers (PLFA concentrations) (Denef et al., 2009). In the study by de Vries et al. (2006), N fertilization reduced the fungal biomass more than the bacterial biomass, which was also mirrored by an increasing bacterial growth rate with higher N. However, the increasing N load in this study was confounded by an increasing soil pH, as clearly recognized by the authors, confounding the relationship between N application and microbial variables. Demoling et al. (2008) studied the influence of N deposition in three forest systems. Although the concentration of fungal biomass markers decreased more than bacterial ones with N addition, the bacterial growth rate was more adversely affected than the fungal growth rate.

Soil pH is one of the most powerful determinants of the microbial community composition (e.g. Blagodatskaya & Anderson, 1998; Bååth & Anderson, 2003; Fierer & Jackson, 2006; Lauber et al., 2009; Rousk et al., 2010a, b). Recently, it was also demonstrated that fungal growth was favoured at low pH, while bacterial growth was favoured at high pH, in a century-old experiment in an arable soil (Rousk et al., 2009b). The shift of the microbial groups between low and high pH resulted in a 30-fold decrease in the relative importance of fungi (fungal : bacterial growth ratio) between pH 4.5 and 8.3, while no change in the concentration of fungal and bacterial biomass markers could be detected over the same range.

In the Park Grass Experiment at Rothamsted Research, Harpenden, UK (henceforth Park Grass), the influence of N fertilization on a grassland soil has been monitored and managed since the 1850s (Rothamsted Research, 2006). Additionally, the soil pH has been regulated to include a pH range within each of the various N treatments. In this study, we contrasted the influence of N fertilization and soil pH on the fungal and bacterial growth as well as respiration and microbial biomass composition. We selected four different N fertilizer rates (0, 48, 96, 144 kg ha−1 year−1) that had each been further subdivided into four soil pH levels due to liming (unlimed pH, and limed to maintain pH 5, 6, and 7). We had four specific objectives. First, we wanted to test whether differences in soil pH in Park Grass would result in a similar shift (30-fold reduction between pH 7.0 and 4.5) in the fungal : bacterial growth ratio described for arable soil (Rousk et al., 2009b). Secondly, we wanted to test whether the lack of influence of pH on biomass markers previously described for the arable soil would be reproduced in Park Grass. Thirdly, we wanted to investigate the influence of N fertilization on both fungal and bacterial growth as well as on the biomass markers. Fourthly, we tested how the factors pH and N fertilization affected the microbial PLFA composition in Park Grass, contrasting the pH influence with that described previously for arable soils (Rousk et al., 2010b) and forest soils (Bååth & Anderson, 2003; Nilsson et al., 2007).

Materials and methods

Soils, sampling and preparation

The soil from the Park Grass Experiment is classified as a Typic Paleudalf (USDA, 1992) or Chromic Luvisol (FAO, 1989). Such soils were originally acidic, well-drained to moderately well-drained and developed in a relatively silty (loess-containing) superficial deposit overlaying, and mixed with, clay-with-flints. The top soil is a flinty, silty clay loam (18–27% clay).

Park Grass is the oldest permanent grassland experiment in the world, established in 1856 to investigate ways of improving the yield of hay by the application of inorganic fertilizers and organic manure. The experiment is described fully elsewhere (Rothamsted Research, 2006). Briefly, Park Grass probably never received the large applications of lime that were often applied to arable fields in this part of England. The soil (0–23 cm) on Park Grass had a pH (in water) of about 5.5 when the experiment began (pH 5.3–5.5 in CaCl2; Blake et al., 1999). The fertilizer treatments decreased pH further, and to counteract this acidification, a liming treatment was initiated in 1965. The fertilizer plots were divided into four subplots. These receive different amounts of lime, when necessary, to achieve and/or maintain the soil (0–23 cm) at pH 5, 6 and 7, respectively. The fourth subplot receives no lime and its pH reflects inputs from the various treatments and the atmosphere. In the present study, we sampled four fertilizer applications: Nil (Park Grass plot 3), henceforth N0, receiving no fertilizers; N1 (Park Grass plot 1), henceforth N1, receiving N as (NH4)2SO4 at a rate of 48 kg N ha−1 year−1; N2 P K Na Mg (Park Grass plot 9/2), henceforth N2, receiving N as (NH4)2SO4 at a rate of 96 kg N ha−1 year−1 as well as 35 kg P, 225 kg K, 15 kg Na and 10 kg Mg (as sulphates) ha−1 year−1; and finally N3 P K Na Mg (Park Grass plot 11/1), henceforth N3, receiving N as (NH4)2SO4 at a rate of 144 kg N ha−1 year−1 and the other indicated mineral nutrients as in N2. The N fertilization treatments are applied annually in April as single dressings. We used all four subplots of these, including pHs 7 (subplot a), 6 (subplot b), 5 (subplot c) and unlimed (subplot d) (Rothamsted Research, 2006) for the four different fertilizer treatments, totaling 16 treatments.

One composite sample of 10 cores was collected from each treatment with a soil corer (0–10 cm depth, 2 cm diameter) in August 2009. The 16 resulting soil samples were sieved moist (<2 mm) and dry matter contents were determined (105 °C; 24 h). The field moist soils were at about 50% of the water-holding capacity, and so further moisture adjustment was not required. They were incubated at 22 °C for 1 week before the microbial analyses were performed. Subsamples of the sieved soil were frozen until PLFA analysis 4 weeks later. Other sieved samples were air-dried and ground (<180 μm) for chemical analyses. Three replicates from each soil sample were used for the growth measurements while the PLFA, respiration and substrate-induced respiration (SIR) biomass measurements as well as the soil chemical analyses were performed in duplicate (n=2). The technical replicates were averaged before statistical analyses.

Microbial growth measurements

The bacterial growth was estimated using leucine (Leu; Kirchman et al., 1985) incorporation in bacteria extracted from soil using the homogenization/centrifugation technique (Bååth, 1994) with modifications (Bååth et al., 2001). We added 2 μL radiolabelled Leu ([3H]Leu, 37 MBq mL−1, 5.74 TBq mmol−1, Amersham) combined with nonlabelled Leu to each tube, resulting in 275 nM Leu in the bacterial suspensions. The amount of Leu incorporated into extracted bacteria per hour and gram soil was used as a measure of bacterial growth.

Fungal growth was assessed using the acetate incorporation into ergosterol method (Newell & Fallon, 1991) adapted for soil (Pennanen et al., 1998; Bååth, 2001) with modifications (Rousk et al., 2009b), adding 1-[14C]acetic acid (sodium salt, 7.4 MBq mL−1, 2.04 GBq mmol−1, Amersham) combined with unlabelled sodium acetate resulting in a final acetate concentration of 220 μM in a soil slurry and having a 4-h incubation at 22 °C without light. Ergosterol was extracted, separated and quantified using HPLC equipped with a UV detector (282 nm). The ergosterol peak was collected and the amount of incorporated radioactivity was determined. The amount of acetate incorporated into fungal ergosterol was used as a measure of fungal growth.

Respiration and SIR biomass

Respiration was determined from CO2 evolved from 3 g soil in 20-mL glass vials closed with crimp caps and incubated under dark conditions at 22 °C for 22–24 h. The CO2 was determined using GC. Microbial biomass was estimated using the SIR method (Anderson & Domsch, 1978). Glucose : talcum (4 : 1; 6 mg g−1) was added to the soil samples following the GC analyses. After 20 min, the atmosphere was purged of CO2 with pressurized air, after which the vials were again closed with crimp caps, and incubated for 2–3 h at 22 °C. The CO2 evolved was then determined. SIR respiration was converted to biomass using the relationship 1 mg CO2 h−1 at 22 °C corresponds to 20 mg biomass C (recalculated from Anderson & Domsch, 1978), and that microbial biomass contains 45% C.

Biomass composition of the community

The PLFA pattern was determined using 1 g of frozen soil according to Frostegård et al. (1993) with modifications described by Nilsson et al. (2007). An internal standard (methyl nonadecanoate fatty acid, 19:0) was added before the methylation step. The PLFAs chosen to indicate bacterial biomass were i15:0, a15:0, i16:0, 16:1ω9, 16:1ω7c, 10Me16:0, cy17:0, i17:0, a17:0, 18:1ω7 and cy19:0, while PLFA 18:2ω6,9 was used to indicate fungal biomass (Frostegård & Bååth, 1996). Ergosterol, determined in the fungal growth estimations, was used as an alternative fungal biomass indicator.

Soil chemical analyses

Soil pH was measured at a soil : water ratio of 1 : 2.5 (w/w). Air-dry soil (10 g, <2 mm) and 25 mL distilled water were shaken together for 2 min, left to settle for 30 min, repeating the procedure once more, and then pH was determined with a pH electrode. Total N and organic C measurements were performed by dry combustion using a Leco CNS-2000 autoanalyser.

Statistical analyses

We contrasted treatments on a per soil organic carbon (SOC) basis. The basic statistical design included two main factors, each with four unreplicated levels, and where the pH-factor differed between different N treatments. The effect of the two factors N fertilization treatment and soil pH was analysed using analyses of covariance (ancova) with N fertilizer treatment (N0, N1, N2 and N3) as a categorical factor with four levels and the measured soil pH (a continuous factor) used as the covariate (n=16; the technical replication was used only to calculate mean and variation in the graphs). A deviation in the pH relationship for the fungal response variables was observed at the lowest pH and thus statistical tests from both the whole pH interval as well as >4.5 are presented (see Discussion). The PLFA composition (mol% of the 30 most abundant PLFAs) was analysed with a principal component analysis (PCA) after standardizing to unit variance.

Statistical tests were performed using jmp 7.0 for Mac (SAS Institute Inc., Cary, NC), and the PCA was performed with multivariate statistical package (mvsp, Kovach Computing Services, Anglesey, Wales). The principal driver of most of the studied variables was pH, and thus they were also regressed against the measured soil pH in the figures. Curves were fitted in the regression analyses with kaleidagraph 4.1 for Mac (Synergy Software, Reading, PA).

Results

Soil pH matched the levels intended to be maintained, at pH levels 7 (7.3±0.1; average±SE), 6 (6.2±0.1) and 5 (4.9±0.1). The pH of the subplots without lime amendment depended strongly on the associated N treatment, however. The unfertilized treatment (N0) had a pH of 5.2 while the background pH in the N-treated soils (N1, N2 and N3) ranged between pH 3.3 and 3.6.

The organic C and total N were both unaffected by soil pH, while the N treatment affected both. The concentration of organic C increased with higher N applications (P<0.01, Fig. 1a), and so did total N (P<0.01, Fig. 1b), the former increasing by about 60% and the latter by 30% between N0 and N3.

Figure 1.

 The organic C (a) and total N (b) in relation to the soil pH in the Park Grass soils. Symbols are coded according to the N fertilization treatments. The curves are fitted to the average of the four N treatments, and are regressed against soil pH.

Fungal growth was positively influenced by decreasing soil pH at pHs>4.5. Below this pH, the pattern for fungal growth reversed, decreasing with lower pH. Starting high at pH 4.5, fungal growth decreased exponentially by more than fourfold with higher pH (P<0.0001, R2=0.75, Fig. 2a). Bacterial growth, in contrast, was positively influenced by soil pH (P<0.0001), increasing exponentially between the low and high pH ends by about 20-fold (P<0.0001, R2=0.65, Fig. 2b). The dynamics of fungal and bacterial growth over the whole pH range covered in Park Grass resulted in a fungal : bacterial growth ratio that decreased about 50-fold from the low-pH end to the high-pH end (P<0.0001, R2=0.77, Fig. 2c). There were no significant effects of N regime on either fungal or bacterial growth, although the fungal : bacterial growth ratio tended to be lower with higher N additions (Fig. 2c).

Figure 2.

 The fungal growth (Ac-in-Erg) (a), bacterial growth (Leu) (b) and the ratio between the two (c) in relation to soil pH in the Park Grass soils. All values are given per unit of soil organic C (SOC). The curves are fitted to the average of the four N treatments, and are regressed against soil pH. The regression in (a) includes only data points above pH 4.5. Symbols are coded according to the N fertilization treatments and denote the mean (n=3) with error bars denoting ±1 SE. Ac, acetate.

The respiration was affected by both N addition and soil pH (Fig. 3a), decreasing with higher N additions (P<0.001) and lower pH (P<0.001), although there was little effect of N regime at low pH (P=0.07 for the interaction term). The respiration decreased by about three times between the high- and low-pH ends, from about 90 μg C h−1 g−1 SOC to about 30 μg C h−1 g−1 SOC (P<0.01, R2=0.47, Fig. 3a), while higher N additions decreased the respiration less consistently, varying between 7% and 67% reduction, at the different pH levels. Respiration is a direct estimate of the C available for the microbial community, as evident by instant increases in respiration following additions of labile C to soil (e.g. Anderson & Domsch, 1978). Consequently, respiration on a per SOC basis should present a rough index for the quality of the SOC as experienced by the heterotrophic soil microbial community.

Figure 3.

 The respiration (a), the SIR biomass (b) and ergosterol concentration (c) in relation to soil pH in the Park Grass soils. All values are given per soil organic C (SOC). The curves are fitted to the average of the four N treatments, and are regressed against soil pH. Symbols are coded according to the N fertilization treatments and denote the mean [(n=2 for (a) and (b), and n=3 for (c)] with error bars denoting ±1 SE.

The SIR biomass followed a similar pattern (Fig. 3b), reduced by both higher N additions (P<0.01) and lower soil pH (P<0.001). The SIR biomass decreased linearly by about 80%, from about 20 mg C g−1 SOC at the high pH end and to about 4 mg C g−1 SOC at the low pH-end (P<0.001, R2=0.56, Fig. 3b). The fungal biomass, as indicated by the ergosterol concentration, was reduced by higher N additions (P<0.01), but was unaffected by pH (Fig. 3c).

The first principal component (PC1) of the PLFA pattern accounted for 47.4% of the variation in the data, while PC2 accounted for 16.1% (Fig. 4a). The soil pH effect on the microbial PLFA composition was directed on a trajectory about 45° offset from the x-axis, with increasing pHs toward the upper right corner of the plot of scores of the two first PCs (Fig. 4a). The influence of N application on the microbial PLFA composition was directed along a trajectory about 135° off the x-axis, and very nearly orthogonal (90°) to the trajectory of the pH effect. Both factors, pH and N fertilizer, were thus components of the two first axes of the PCA. The loadings of the individual PLFAs (Fig. 4b) showed that the main effect of soil pH was high relative abundances of PLFA markers including monounsaturated PLFAs 16:1ω7c, 17:1ω8, 16:1ω5, 16:1ω9 and 18:1ω7 at high pH, while a high relative abundance of PLFA markers including cy19:0, 10Me16:0, i16:0, and i15:0 was associated with low pH. Lower N rates were associated with high relative abundances of PLFA markers including the fungal indicators PLFA 18:2ω6,9 and 18:1ω9 (Fig. 4b).

Figure 4.

 The scores of the two first components from a PCA on the microbial PLFA composition (a) and a loading plot of the two first components (b). The samples (a) are coded with the soil pH (3.3–7.5) followed by the N treatment (N0–N3) and denote the mean (n=2) with error bars denoting ±1 SE. The vectors ‘Soil pH’ and ‘N-fert.’ denote the positive orientation of the trajectory in which samples have assembled themselves, with higher soil pH and N fertilization addition in the direction the vectors are pointing.

There was no significant effect of pH on the concentration of PLFA markers indicating bacterial biomass (Fig. 5b) or on the total PLFA (Fig. 5d). This was also the case for the concentration of the fungal PLFA marker 18:2ω6,9 above pH 4.5 (Fig. 5a). However, at the lowest pH, low concentrations of 18:2ω6,9 were found, resulting in a low fungal : bacterial PLFA ratio at pHs below 4.5, but that was stable at pH>4.5 (Fig. 5c).

Figure 5.

 The concentration of fungal PLFA 18:2ω6,9 (a), bacterial PLFA (b), the ratio between them (c) and total PLFAs (d) in relation to soil pH in the Park Grass soils. All values are given per unit of soil organic C (SOC). The curves are fitted to the average of the four N treatments, and are regressed against soil pH. The regressions in (a) and (c) include only data points above pH 4.5. Symbols are coded according to the N fertilization treatments and denote the mean (n=2) with error bars denoting ±1 SE.

N application affected the concentrations of the fungal PLFA 18:2ω6,9 (P<0.0001; Fig. 5a), bacterial PLFA (P<0.001, Fig. 5b) and total PLFA (P<0.01; Fig. 5d) similarly, in that high N application resulted in lower concentrations. The negative influence of higher N application on the PLFA concentration was proportionally more pronounced for the fungal marker, resulting in 75% lower concentrations with high N applications compared with low N applications (Fig. 5a), as against smaller effects (around 25% lower concentrations in high N applications compared with low N applications) on the bacterial PLFA markers and total PLFA markers. This resulted in a fungal : bacterial PLFA ratio that decreased with higher N fertilization (P<0.01, Fig. 5c).

Discussion

Soil pH influence on microbial growth

The strongest influence of soil pH was on the fungal and bacterial growth. The relative dominance of fungi to bacteria, as indicated by the fungal : bacterial growth ratio, increased 50-fold between the high- (pH 7.4) and the low-pH ends (3.3) in Park Grass. These results closely mirrored those recently reported from a comparable pH gradient in an arable soil (‘Hoosfield acid strip’, Rousk et al., 2009b), where the fungal : bacterial growth ratio also increased 50-fold between pH 8.3 and 4.0. When these results are combined, they suggest that a one unit decrease in soil pH will increase the fungal : bacterial growth ratio 2.3–3.3 times. The negative effect of low pH on bacterial growth in soil was also earlier suggested from a study of a wide range of soils with different pHs (Bååth, 1998), in limed forest soil (Bååth & Arnebrant, 1994) and in a natural plant productivity gradient (Högberg et al., 2003). Although these reports include confounding effects, such as availability of C, and did not measure both fungal and bacterial growth, their data suggest that the pattern here identified for arable and grassland soils may also be extended to forest soils. Thus, a systematic pattern emerges with a considerably decreased bacterial growth with lower pH irrespective of soil type.

The confounding effect of reduced plant C input below pH 4.5 earlier identified for the Hoosfield acid strip (Rousk et al., 2009b, 2010a), which resulted in a break in the increasing fungal growth with lower pH, may also be relevant to the present study. There was a consistent reduction in plant production at pHs<4.5, but while the reduction compared with the average plant growth >pH 4.5 was about 60% for N1, it was only 30% and 20% for N2 and N3 (Rothamsted Research, 2006), respectively, and thus a much smaller difference than the increase in crop yield by higher N additions (see section on N effects below). This pattern of a proportionally smaller effect on plant productivity below pH 4.5 with higher N applications did not match the decrease in fungal growth, suggesting that the influence of unknown factors, potentially including plant community composition, complicate the fungal pH dependence below pH 4.5.

Soil pH influence on respiration

Respiration was no more affected by the drastic shifts in the actively growing microbial community, associated with soil pH, than by the small changes in it, associated with N application. Although lower pH decreased soil respiration by up to about 67%, this was small in relation to the 50-fold shift in the fungal : bacterial growth ratio. This resembles results of almost unaffected soil respiration despite significant shifts between fungal and bacterial growth reported previously during the decomposition of added plant material (Rousk & Bååth, 2007), and in an experiment where the bacterial contribution to decomposition was completely inhibited with specific antibiotics without any short-term effects on respiration (Rousk et al., 2008). The relatively small change in respiration despite the shift between the contributions of fungi and bacteria to the process could indicate that the two major decomposer groups involved in soil C mineralization are complementary, suggesting functional redundancy. Despite the intuitive connection between microbial growth and respiration, it should be noted that the processes are not directly connected (Schimel & Weintraub, 2003; Kemmitt et al., 2008), being offset by a variation in the partitioning of a substrate into growth and respiration, resulting in different growth efficiencies (Six et al., 2006; Thiet et al., 2006).

N fertilizer influence on microbial growth and respiration

The application of N fertilizers increased the concentration of organic C and total N in the soil. This is due to the increased plant growth in fertilized plots (Rothamsted Research, 2006), with a subsequent increased input of C into the soil. Omitting results from plots with pH<4.5, the crop yield increased from 3±0.3 tonnes ha−1 year−1 in N0 to 3.6±0.4, 8.0±0.2 and 9.4±0.5 tonnes ha−1 year−1 for N1, N2 and N3, respectively (Rothamsted Research, 2006). The unlimed plots with fertilizers (pH<4.5) consistently yielded about 2 tonnes ha−1 year−1 less than the average for the same N fertilizer level. However, although an increased plant residue input due to N fertilization would have a positive effect on microbial activity in the short term, prolonged N application appeared to result in a decreased availability of SOC, as indicated by lower respiration per SOC (i.e. lower biologically available SOC). Calculating respiration per unit SOC resulted in the respiration rate decreasing by up to 67% with higher N applications.

This negative effect of N application on microbial respiration has been noted frequently in N fertilization studies (Fog, 1988; Berg & McClaugherty, 2003; Burton et al., 2004; DeForest et al., 2004; Knorr et al., 2005; Olsson et al., 2005; Pregitzer et al., 2008; Ramirez et al., 2010; also see overviews by Treseder, 2008; Liu & Greaver, 2010) and has been attributed to the suppression of the activity of enzymes required for the mineralization of recalcitrant litter fractions (Fog, 1988; Berg & McClaugherty, 2003), resulting in an enrichment of the proportion of low-quality C in the SOC. A lower quality SOC has previously been associated with a promotion of fungal decomposers (van der Heijden et al., 2008). However, both the fungal : bacterial growth ratio (Fig. 2c) and the fungal : bacterial PLFA ratio (Fig. 5c) tended to decrease in N-fertilized soils, that is, with decreasing SOC availability.

The difference of up to 67% between the respiration in unfertilized soil and the highest N application was largely unrelated to changes in the actively growing microbial community. This emphasizes that it is C availability, rather than the size, structure or growth of different groups within microbial community, that directly influences soil respiration (Kemmitt et al., 2008).

Soil pH and N fertilizer effects on the microbial PLFA composition

Based on data from both forest (Bååth & Anderson, 2003; Nilsson et al., 2007) and arable soils (Aciego Pietri & Brookes, 2009; Rousk et al., 2010b), Rousk et al. (2010b) proposed that the combination of increases in relative concentration of the PLFAs 18:1ω7, 16:1ω7c and 16:1ω5 with a concomitant decrease of PLFAs cy19:0 and i16:0 could be used as an indicator of a direct pH effect. Based on this proposal, we should expect the soil samples to align on a trajectory between the low-pH samples, associated with high abundances of cy19:0 and i16:0, and high-pH samples, associated with higher abundances of 18:1ω7, 16:1ω7c and 16:1ω5. We found the high-pH PLFA markers to be oriented about 45° offset the positive x-axis in the top right corner and low pH PLFA makers to be oriented in the bottom left corner (Fig. 4b). The soil pHs of the samples from Park Grass aligned along this trajectory (Fig. 4a), supporting the suggested pH effects on the microbial PLFA composition (Rousk et al., 2010b).

The soil samples of different N additions aligned orthogonal to the pH variation in the PLFA composition (Fig. 4a). The fact that the difference was orthogonal to the pH effect suggests that there was no clear interaction between the factors, and that they exposed the microbial community composition in categorically different ways. There was an association between high relative abundances of fungal PLFA markers and low N applications. It is unclear whether the changes in PLFA composition associated with N fertilizer additions are related to the decreasing SOC quality, as indicated by lower respiration per SOC, associated with higher N applications, or whether it is a direct effect of higher N applications. Clearly, both studied factors, soil pH and N fertilization, strongly affected the microbial PLFA composition (the combined factors explained 63.5% of the PLFA variation).

Microbial biomass responses to soil pH and N fertilization

Estimates of fungal and bacterial biomass in Park Grass suggested only minor differences along the pH gradient, with unchanging bacterial PLFAs and fungal biomass (ergosterol), while fungal PLFA 18:2ω6,9 decreased only at the lowest pH. This lack of pH effect on fungal and bacterial biomass is consistent with previous assessments using PLFA markers (Bååth & Anderson, 2003; Han et al., 2008; Rousk et al., 2009b), ergosterol compared with total biomass estimates (Marstorp et al., 2000; Gong et al., 2001; Rousk et al., 2009b) and quantitative PCR-based gene copy numbers (Rousk et al., 2010a), where variation in factors potentially confounding the soil pH effect, including organic matter concentration or plant community (and associated ectomycorrhizal fungi) differences, were minimized. This contrasts with the view of increasing dominance of fungal biomass associated with acid forest soils (Fierer et al., 2009). However, in many cases, assessments of pH effects on the fungal biomass ratio have not taken ectomycorrhizal fungi into account (Joergensen & Wichern, 2008; Strickland & Rousk, 2010). Acid forest soils usually have a vegetation with ectomycorrhiza, and thus a high proportion of fungal biomass markers is derived from these organisms and does not represent only saprotrophic fungal biomass (Högberg et al., 2003, 2007). In a recent study, it was also shown that N fertilization especially affected ectomycorrhizal fruit body production in forest soil, emphasizing that ectomycorrhizal fungi may be especially sensitive to increasing N availability (Gillet et al., 2010). When direct pH effects are studied in isolation, it appears that the fungal : bacterial biomass ratio estimated with PLFAs is merely influenced to a minor degree and does not reflect the drastic changes in the fungal : bacterial growth ratios. However, support for higher fungal dominance with biomass estimates has been found using the selective respiratory inhibition technique (Anderson & Domsch, 1973, but also see Rousk et al., 2009a). The application of this method has suggested four- to sixfold increases in the fungal : bacterial biomass ratio between neutral (pH 6–8) and acid (pH 3–4) soil conditions (Blagodatskaya & Anderson, 1998; Bååth & Anderson, 2003).

Higher rates of N fertilization consistently reduced the concentration of microbial biomass markers (SIR biomass, total PLFA, bacterial PLFA, fungal PLFA and ergosterol). Microbial biomass reductions due to N addition have previously been documented in several ecosystems including grasslands (Treseder, 2008; Liu & Greaver, 2010). The reduction in fungal markers appeared to be more pronounced than for total and bacterial markers. This is consistent with previous work in both grassland soils (Bardgett et al., 1999; de Vries et al., 2006, 2007) and forest soils (Högberg et al., 2003; Demoling et al., 2008). It is interesting to note the discrepancy between growth estimates and biomass estimates of fungi and bacteria using PLFA in soil. Although N affected the growth of fungi and bacteria only marginally, the concentration of biomass markers was reduced significantly. This contrasted sharply with the pronounced effects on fungal and bacterial growth, but only small differences in biomass markers, induced by soil pH differences.

Conclusion

Our results emphasize the need to carefully select the microbial response variable to study for the question one intends to address, as noted by Strickland & Rousk (2010). For example, if we want to contrast the functional contributions of the decomposer groups, we should use growth-related measurements; if C sequestration is the ecosystem process in question and if a build-up in the amount of fungal cell wall components is important for this process then measurements of fungal residues would be the most applicable; and if C sequestration is thought to be related to C use efficiency, substrate used for growth should be compared with the amount respired. Moreover, we observe that methods that use different proxies to estimate the same microbial variable varied depending on the environmental factor affecting them: N fertilization reduced microbial biomass, as estimated by both SIR and total PLFA, while low pH decreased only SIR biomass, without any effect on the total PLFA. Also, fungal biomass estimated with ergosterol and the PLFA 18:2ω6,9 is similarly affected by N fertilization, while only one of them, 18:2ω6,9, decreases at low pH. Thus, not only is it important to carefully select a suitable microbial response variable, but it may also be necessary to include more than one method to estimate it.

Acknowledgements

This study was supported by the Swedish Research Council (VR) grants to J.R. (Project No. 623-2009-7343) and E.B. (Project No. 2009-4503). Rothamsted Research receives grant-aided support from the Biotechnological and Biological Sciences Research Council (BBSRC).

Ancillary