1The effects of global change on below-ground processes of the nitrogen (N) cycle have repercussions for plant communities, productivity and trace gas effluxes. However, the interacting effects of different components of global change on nitrification or denitrification have rarely been studied in situ.
2We measured responses of nitrifying enzyme activity (NEA) and denitrifying enzyme activity (DEA) to over 4 years of exposure to several components of global change and their interaction (increased atmospheric CO2 concentration, temperature, precipitation and N addition) at peak biomass period in an annual grassland ecosystem. In order to provide insight into the mechanisms controlling the response of NEA and DEA to global change, we examined the relationships between these activities and soil moisture, microbial biomass C and N, and soil extractable N.
3Across all treatment combinations, NEA was decreased by elevated CO2 and increased by N addition. While elevated CO2 had no effect on NEA when not combined with other treatments, it suppressed the positive effect of N addition on NEA in all the treatments that included N addition. We found a significant CO2–N interaction for DEA, with a positive effect of elevated CO2 on DEA only in the treatments that included N addition, suggesting that N limitation of denitrifiers may have occurred in our system. Soil water content, extractable N concentrations and their interaction explained 74% of the variation in DEA.
4Our results show that the potentially large and interacting effects of different components of global change should be considered in predicting below-ground N responses of Mediterranean grasslands to future climate changes.
The components of global change cited above have the potential, directly or indirectly, to affect nitrification and denitrification. Elevated CO2 has generally been found to increase soil water content and the availability of below-ground labile C through its effect on plants (Hungate 1999). These CO2-induced changes are expected to have direct positive effects on the environmental conditions that control denitrification and negative effects on those controlling nitrification, and also have indirect effects on both of these processes, e.g. through their impact of soil inorganic N availability (Hungate 1999). Nitrification and denitrification are affected by changes in temperature in most soil incubation studies (Grundmann et al. 1995; Maag & Vinther 1996). However, the few in situ studies of the effect of warming on nitrification and denitrification do not show consistent results across experiments, and the underlying mechanisms remain poorly understood (Verburg, Van Loon & Lükewille 1999; Shaw & Harte 2001). Increased precipitation should increase soil water content and soil heterotrophic respiration, thereby reducing soil [O2] and, under moderately moist soil conditions, is expected to favour denitrifiers while tightening environmental constraints on nitrifiers (Tiedje 1988; Grundmann et al. 1995; Maag & Vinther 1996). N addition is expected to increase substrate availability for nitrifiers and denitrifiers, which should result in higher rates of nitrification and denitrification. Finally, little is known of the mechanisms of interaction of the components of global change on nitrification or denitrification, since only a few studies have measured significant effects of treatment interactions (see review in Barnard, Leadley & Hungate 2005).
Measurements of nitrifying and denitrifying enzyme activity (NEA and DEA, respectively) provide a measure of the amount of functionally active enzyme associated with these processes in the soil (e.g. Smith & Tiedje 1979). These assays reflect the environmental constraints on nitrifying and denitrifying communities in the soil, such as N substrate availability, soil aeration, labile C availability, temperature and pH (Tiedje 1988; Paul & Clark 1989). It is not the goal of this study to estimate in situ N transformation rates, but to understand the environmental constraints on the microbial enzymatic activities. In the present study, we measured the effect of four interacting components of global change – increased elevated atmospheric CO2 concentrations, temperature, precipitation and N addition – on NEA and DEA in an annual grassland, together with measurements of soil moisture, microbial biomass C and N, and soil extractable N.
The objectives of this study were to: (i) quantify the amplitude of changes in NEA and DEA in response to components of global changes with particular reference to interaction effects; and (ii) identify the mechanisms that underlie the response of NEA and DEA to these components of global climate change during peak biomass period, a key period in the growing season of these Mediterranean grasslands.
Materials and methods
experimental set-up and soil sample collection
This study was conducted at Stanford University's Jasper Ridge Biological Preserve in central California (37°24′N, 122°14′W, elevation 150 m), in a moderately fertile, natural Mediterranean grassland dominated by annual grasses (Avena barbata Link and Bromus hordeaceus L.) and forbs (Geranium dissectum L. and Erodium botrys (Cav.) Bertol.). The Jasper Ridge Global Change Experiment provided full factorial combination of four, two-level global change treatments and was initiated in November 1998. A detailed description of the experimental set-up and management is given by Shaw et al. (2002). In brief, 32 experimental plots (2 m diameter) were organized as a split-plot design, each plot being divided into four 0·78 m2 quadrants. Treatments were atmospheric CO2 concentration (ambient and 680 µmol mol−1), temperature (ambient and ambient + 80 W m−2 thermal radiation, resulting in a soil-surface temperature increase of 0·8–1 °C), precipitation (ambient and ambient + 50%+ 3-week growing-season elongation) and N addition (ambient and ambient + 7 g N–Ca(NO3)2 m−2 year−1). These treatments and their combinations are referred to as CO2, T, W, N and their combinations, respectively. Treatments were applied using CO2 emitter rings, infrared heaters, spray/drip systems and slow-release fertiliser. The addition of N with slow-release fertiliser was intended to mimic in a crude way atmospheric N deposition. Each of the 16 combinations of CO2, temperature, N addition and precipitation was replicated eight times.
Two soil cores (2·2 cm diameter × 15 cm deep) were sampled in each plot quadrant on 25–26 April 2003, pooled and homogenized. Sampling date corresponds to the period of peak biomass in this Mediterranean annual grassland (Zavaleta et al. 2003a). A subsample of the soil was used to determine gravimetric soil water content. Soil was stored at 4 °C until the enzyme assays began (as observed by Luo et al. (1996) a few days of storage did not affect enzyme activities).
nitrifying and denitrifying enzyme activity measurements
We measured DEA (Smith & Tiedje 1979; modified by Patra et al. 2005) at soil pH over a short period by making all factors affecting denitrification non-limiting. Five grams equivalent dry soil were placed in a 150 ml plasma flask containing 1 mg C-glucose g−1 dry soil, 1 mg C–glutamic acid g−1 of dry soil, and 0·1 mg N– g−1 dry soil. The atmosphere of each tube was replaced by a 90:10 He–C2H2 mixture providing anaerobic conditions and inhibition of N2O-reductase activity. N2O efflux was measured in this flask after 0, 4 and 7 h, to make sure no de novo synthesis of denitrifying enzyme took place. N2O concentrations were analysed on a gas chromatograph equipped with an electron capture detector (Varian Star 3400CX, Varian Chromatography Group, Palo Alto, CA).
microbial biomass n and c, soil extractable n, soil and
Soil microbial N and C were measured by chloroform fumigation-extraction (Brookes et al. 1985). Soil samples (10–15 g) were fumigated for 5 days with chloroform vapour. Control samples were not fumigated. Control and fumigated samples were extracted in 0·5 m K2SO4 by shaking for 60 min. Samples were frozen until analysis for total carbon using a total organic C analyser (Shimadzu TOC-5000 A, Shimadzu Corp., Tokyo, Japan). Total N in the extracts was later obtained by persulphate digestion and analysis on a continuous flow autoanalyser (Astoria 2, Astoria Pacific, Clackamas OR). Microbial biomass N was calculated as [(total N in fumigated soil) – (total N in non-fumigated soil)]/0·54 (Brookes et al. 1985). Microbial biomass C was calculated in a similar way, with an extraction efficiency coefficient of 0·45 (Wu et al. 1990). Soil extractable N was measured in the unfumigated extracts.
Bags containing 1·5 g combination anion–cation ion-exchange resin (Bio-Rad™ AG-1-X8 CL− form and Baker™ HCR-W2 H+ form) were placed in the field from 1 April 2003 to 1 June 2003, at the end of a PVC tube inserted into the soil at a 45° angle such that the bag was at approximately 25 cm depth. When removed, the bags were rinsed in de-ionised water and each bag was extracted in 15 ml of a 2 m KCl solution by shaking for 30 min. The extracts were frozen until analysis in October 2004.
We analysed our data with a split-plot, randomised complete block analysis of variance (n = 8 for all measurements), using SAS 8·02 (SAS Institute, Cary, NC). To correct non-equal variance, NEA data were square-root transformed; DEA and microbial biomass C and N data were log-transformed. A least significant difference test (LSD) was performed to determine differences of means between groups. Relative treatment effects were calculated as follows: % effect = 100 ×[treatment – control]/control. We also checked for correlations and performed stepwise regressions for NEA and DEA on treatment means for the 16 treatment combinations. The explanatory variables in the stepwise regressions were gravimetric soil water content, soil extractable N, and microbial biomass N and C.
nitrifying and denitrifying enzyme activities
Across all treatments, elevated CO2 depressed NEA by 46% (Fig. 1, Table 1). Under ambient CO2, N addition increased NEA, but this effect was not apparent at elevated CO2 (significant CO2 × N interaction). We found no significant overall effects of the precipitation and temperature treatments.
Table 1. Effect and interaction of the different treatments (CO2: elevated CO2, T: elevated temperature, N: increased N addition, W: increased precipitation) on soil water content (SWC), nitrifying enzyme activity (NEA), denitrifying enzyme activity (DEA), microbial biomass N and C, soil extractable N, and nitrate and ammonium trapped on resins. % effect = 100 ×[elevated − ambient]/ambient. Significant responses are indicated in bold.
Microbial biomass N
Microbial biomass C
Soil extractable N
Main plot effects
CO2 × T
CO2 × N
CO2 × W
T × N
T × W
W × N
CO2 × T × N
CO2 × T × W
CO2 × N × W
T × W × N
Increased precipitation and N addition significantly increased DEA (respectively +45 and +68%, Fig. 2, Table 1). Elevated CO2 increased DEA at high N addition and ambient temperature (CO2 × N interaction), but at high N addition and warming, elevated CO2 had no effect on DEA (CO2 × T × N interaction). Pairwise comparisons showed that at elevated CO2, the N, WN and TWN treatments significantly increased DEA, respectively, by 77, 34 and 163%. We found no significant overall effects of CO2 or temperature treatments.
microbial biomass n and c, soil extractable n, soil and
Microbial biomass N was significantly increased by elevated N addition (+11%), an effect that was attenuated in combination with increased precipitation (Table 1). Microbial biomass C was not significantly affected by any treatment (Table 1). Increased N addition significantly increased soil extractable N (+28%, Table 1). Increased N addition resulted in a significant increase of and a large significant increase of trapped in soil resin bags (Table 1). We also measured a significant T × N interaction for in resin extracts, the positive effect of elevated temperature at ambient N becoming negative with N addition. The values of microbial biomass N and C, and soil extractable N in the main treatment plots are given in Table 2.
Table 2. Microbial biomass N and C, and soil extractable N in the main treatment plots (T: temperature, N: N addition, W: precipitation). Values indicate mean ± SE. Refer to Table 1 for the statistical analysis of the complete data set.
Microbial biomass N (µg N g−1 dry soil)
Microbial biomass C (µg C g−1 dry soil)
Soil extractable N (µg N g−1 dry soil)
38·8 ± 1·5
394·6 ± 19·8
9·0 ± 0·4
36·6 ± 1·2
395·2 ± 18·6
8·8 ± 0·4
38·2 ± 1·4
401·6 ± 19·7
8·9 ± 0·4
37·2 ± 1·4
387·7 ± 18·6
8·9 ± 0·4
35·8 ± 1·4
388·2 ± 19·6
7·8 ± 0·2
39·5 ± 1·3
401·6 ± 18·8
10·0 ± 0·5
38·9 ± 1·3
397·4 ± 16·7
9·0 ± 0·4
36·6 ± 1·4
392·5 ± 21·2
8·8 ± 0·3
soil water content
The precipitation and N addition treatments both caused a slight but significant increase in soil water content in our samples (+5%, P = 0·003 and +4%, P = 0·02, respectively, Table 1), but other treatments did not alter soil water content. Across all treatments, average soil water content was 18·7%.
We found no significant correlation between NEA and DEA measurements and other measurements performed at the same time (i.e. above-ground biomass, laboratory-incubated soil N2O and CH4 effluxes, data not shown). We found significant positive correlations for treatment means between DEA and soil water content (R2 = 0·60; P = 0·0004, Fig. 3).
The variation of NEA was not significantly explained by the explanatory variables that we measured. Stepwise regression showed that soil water content, soil extractable N and their interaction explained 74% of the variation of DEA, with soil water content explaining 60% (P < 0·001), soil extractable N an additional 9% (P = 0·02) and their interaction 7% (P = 0·04).
nitrifying enzyme activity
At ambient CO2, N addition showed a large positive effect on NEA across all treatments. Although information on the effect of moderate N addition on NEA in natural grassland ecosystems is lacking, measures of gross and net nitrification fluxes generally indicate a positive response to N addition (Barnard et al. 2005). The positive effect of N addition on the amount of trapped on resins suggests that long-term N addition, even in the form of , enhanced substrate availability for nitrification. We did not, however, find any significant correlations of NEA with measured soil N availability.
The absence of a response of NEA to warming in our study is consistent with other field experiments that have addressed the effect of warming on nitrification in herbaceous ecosystems. These studies have shown no significant response of net nitrification (Verburg et al. 1999; Shaw & Harte 2001) or gross nitrification (Shaw & Harte 2001). The scarcity of experiments calls attention to the need for studying the response of nitrification with in situ warming experiments.
When not combined with other treatments, elevated CO2 had no effect on NEA. These results are consistent with those of other studies in grassland ecosystems, in which NEA has been found to decrease or be left unchanged at elevated CO2 (Niklaus et al. 2001; Barnard et al. 2004a,b). Across all treatments, however, elevated CO2 suppressed the positive effect of N addition on NEA in all the treatments that included N addition. The response of nitrification to elevated CO2 and N addition in multifactorial experiments has so far been addressed in only a few studies. Zak et al. (2000a) found a significant stimulation of gross and net nitrification at high N availability, but no significant effect of elevated CO2 and no significant CO2 × N availability interaction. However, in grassland mesocosms using soil from Jasper Ridge, Hungate et al. (1997) found a pattern of response of gross nitrification that was similar to that measured for NEA in our study. In their experiment, elevated CO2 had no effect on gross nitrification in unfertilised conditions, but decreased gross nitrification under nutrient enrichment.
This strong CO2–N interaction in Jasper Ridge experiments is puzzling, especially since we cannot explain the patterns in NEA using soil properties that we hypothesised would be driving responses to global change factors (e.g. soil concentrations and soil water content). We have several possible explanations for this interaction, but none can be clearly demonstrated with our data. First, Hungate et al. (1997) suggested that fertilisation with slow-release fertilizer pellets, which were also used in our experiment, might cause a more homogeneous distribution of N in the soil, resulting in a tighter coupling of nitrification and gross microbial immobilization of . This could accentuate the effects of elevated CO2 on microbial competition for . The marginally significant CO2 × N interaction for microbial biomass N – elevated CO2 consistently increased soil microbial N only in the high N treatments – lends some credence to this explanation, but we did not measure rates of microbial N immobilisation. Second, elevated CO2 and N addition can have strong synergistic effects on microbial respiration (Niklaus 1998) which would reduce soil [O2]. There is a wide range of other possible explanations, including decreased mineralisation rates (Zak et al. 2000b), increased plant uptake (Bassirirad, Gutschick & Lussenhop 2001) or increased root density (Raynaud & Leadley 2004; see also Niklaus et al. 2001 for additional hypotheses).
Unlike DEA (see below), our values of NEA were not correlated with a variety of potential explanatory variables that we measured. There are two possible explanations for this lack of correlation: (i) we did not measure key explanatory variables (e.g. consumption rates, soil [O2]) or (ii) our measurements of explanatory variables did not integrate over a long enough time period. Concerning this second point, nitrifying bacteria are known to have low growth rates as compared to denitrifying bacteria (Prosser 1989). Thus, our point measurements of explanatory variables, e.g. extractable N and soil water content, may provide an adequate picture of environmental constraints on DEA, but not NEA. For example, elevated CO2 has been shown to substantially increase soil water content in the Jasper Ridge experiment (Zavaleta et al. 2003b), but did not have an effect on soil water content at our sampling date, presumably because of abundant rainfall prior to sampling.
denitrifying enzyme activity
Increased soil water content, extractable N concentrations and their interaction explained much of the variation in DEA in this experiment. This corresponds to our hypotheses that DEA should be tightly coupled to N availability and to soil water content (the latter affecting DEA through its effects on soil [O2], Tiedje 1988). These factors clearly explain the large positive responses of DEA to the N addition (where N was added as ) and increased precipitation treatments. In particular, soil water content explained 60% of the variance in DEA across all treatments, and appears to be a major factor controlling DEA across a wide array of components of global change and their interactions in this experiment.
Increased temperature showed no significant effect on DEA in our experiment. The significant three-way interaction between CO2, temperature and N treatments is challenging to interpret. Elevated CO2 and temperature tended to decrease DEA, unless in combination with N addition, while N addition tended to increase DEA only when in combination with elevated temperature or CO2. Other experimental results suggest interactive effects of CO2 and temperature on C availability in the soil that would affect DEA (Tscherko, Kandeler & Jones 2001), but the underlying mechanisms are still unknown in field conditions.
The unresponsiveness of DEA to elevated CO2 is consistent with other undisturbed grassland studies that have found little effect of elevated CO2 on DEA (Kammann 2001; Barnard et al. 2004a). The response of DEA showed a significant CO2 × N interaction in our study, with a positive effect of elevated CO2 on DEA only in the treatments that included N addition. This suggests that the absence of an elevated CO2 effect on DEA in treatments without N addition might be explained by N-limitation of denitrifiers. The few other experiments that measured the response of denitrifiers to both elevated CO2 and N addition found no significant interaction between these treatments, but the systems studied differed from our grassland field study (Ambus & Robertson 1999: poplar mesocosms; Martín-Olmedo, Rees & Grace 2002: barley microcosms).
Our study shows that different components of global changes can have large and interacting effects on N cycling in grasslands. In particular, N addition drove the response of NEA at ambient CO2, but elevated CO2 suppressed this effect; the determinism of this response still remains to be identified. The mechanisms that drove the response of DEA across a broad range of treatments are the factors that determine soil [O2] and N substrate availability. Further investigations are in progress to determine to what extent the patterns observed in the present study may show intra- and interannual variability. This study stresses the importance of multifactorial experimental designs in revealing such interactions that are necessary to predict the overall effects of global changes on N dynamics.
We would like to thank Nona Chiariello and Chris Field for their help in making this work possible, Jeff Dukes, Lisa Moore and the people who were involved with the soil sampling, Nadine Guillaumaud for her support with the enzymatic measurements, Hans-Rudolf Roth for precious help with the statistical model, and Yuka Otsuki Estrada for handling the JRGCE results database. The Jasper Ridge Global Change Experiment was supported by the National Science Foundation, the David and Lucile Packard Foundation, the Morgan Family Foundation, and the Jasper Ridge Biological Preserve. This work was funded by CNRS support of the Laboratoire d’Ecologie, Systématique et Evolution (UMR CNRS 8079), CNRS and INRA supports of the Laboratoire d’Ecologie Microbienne (UMR CNRS 5557; USC INRA 1193) and the US National Science Foundation (DEB-0092642). E.E.C. was supported by a US Department of Energy Graduate Research in Environment Fellowship. R.B. was supported in part by the Schneider-Forest Grant of the Chancellerie des Universités de Paris.