ECOSYSTEM ECOLOGY

Sean M. Schaeffer Æ R. D. Evans

Pulse additions of soil carbon and nitrogen affect soil nitrogendynamics in an arid Colorado Plateau shrubland

Received: 22 November 2004 / Accepted: 19 April 2005 / Published online: 7 July 2005� Springer-Verlag 2005

Abstract Biogeochemical cycles in arid and semi-aridecosystems depend upon the ability of soil microbes touse pulses of resources. Brief periods of high activitygenerally occur after precipitation events that provideaccess to energy and nutrients (carbon and nitrogen) forsoil organisms. To better understand pulse-drivendynamics of microbial soil nitrogen (N) cycling in anarid Colorado Plateau ecosystem, we simulated a pulsedaddition of labile carbon (C) and N in the field under thecanopies of the major plant species in plant interspaces.Soil microbial activity and N cycling responded posi-tively to added C while NH4

+–N additions resulted in anaccumulation of soil NO3

�. Increases in microbialactivity were reflected in higher rates of respiration andN immobilization with C addition. When both C and Nwere added to soils, N losses via NH3 volatilizationdecreased. There was no effect of soil C or N availabilityon microbial biomass N suggesting that the level ofmicrobial activity (respiration) may be more importantthan population size (biomass) in controlling short-termdynamics of inorganic and labile organic N. The effectsof C and N pulses on soil microbial function and poolsof NH4

+–N and labile organic N were observed to lastonly for the duration of the moisture pulse created bytreatment addition, while the effect on the NO3

�–N poolpersisted after soils dried to pre-pulse moisture levels.We observed that increases in available C lead to greaterecosystem immobilization and retention of N in soilmicrobial biomass and also lowered rates of gaseous Nloss. With the exception of trace gas N losses, the lack of

interaction between available C and N on controlling Ndynamics, and the subsequent reduction in plant avail-able N with C addition has implications for the com-petitive relationships between plants species, plants andmicrobes, or both.

Keywords Ammonia volatilization Æ Biogeochemicalcycles Æ Denitrification Æ Mineralization Æ Nitrification

Introduction

There is a growing scientific awareness of how changesin precipitation regimes may affect terrestrial ecologicalsystems (Weltzin and McPherson 2003). Low rainfall isthe primary constraint on biological activity in arid andsemi-arid ecosystems, so changes in the intensity, fre-quency, and seasonality of precipitation may affect pri-mary production, decomposition, trace gas flux, andother aspects of biogeochemical cycling. The ability ofsoil microbes to use resource pulses coinciding withinfrequent rain events is critical for arid and semi-aridecosystem biogeochemical cycles where rates of nutrientcycling can be high in the presence of adequate water(Noy-Meir 1973; West and Skujins 1977; Bolton et al.1993; Schaeffer et al. 2003). Brief periods of high activitygenerally occur following pulses of precipitation (Cuiand Caldwell 1997) that provide access to energy andnutrients (carbon and nitrogen) for soil organisms(Zaady et al. 1996; Bassirirad et al. 1999; Gebauer andEhleringer 2000). The dynamics of microbial soil nitro-gen (N) cycling in these systems, however, are not wellunderstood in relation to the pulse-dependent nature ofwater, energy, and nutrient availabilities necessary forbiologically driven soil N cycling (Zaady et al. 1996;Austin et al. 2004).

Nitrogen cycles in arid ecosystems have historicallybeen characterized by relatively low N availability, lowrates of inputs via atmospheric deposition andN2-fixation, and relatively high rates of gaseous loss of

Communicated by Jim Ehleringer

S. M. SchaefferUniversity of Arkansas Stable Isotope Laboratory,University of Arkansas, Fayetteville Arkansas, 72701, USA

R. D. Evans Æ S. M. Schaeffer (&)School of Biological Sciences, Washington State University,Pullman Washington, 99164-4236, USAE-mail: smschae@uark.eduTel.: +1-509-3356154Fax: +1-509-3353184

Oecologia (2005) 145: 425–433DOI 10.1007/s00442-005-0140-2

NOx, N2O, N2 and NH3 (Peterjohn and Schlesinger1990, 1991). Arid ecosystems also possess a large de-gree of spatial heterogeneity in resource availability(Schlesinger et al. 1996; Schlesinger and Pilmanis 1998;Cross and Schlesinger 1999) that is reflected in highlyvariable N dynamics across a landscape (Billings et al.2002, 2003; Schaeffer et al. 2003). Nitrogen dynamicsin these ecosystems may also be particularly affectedby soil C resources. Westerman and Tucker (1978),Virginia et al. (1982), Smith et al. (1994), and Gallardoand Schlesinger (1995) have demonstrated that C andN availability affect various N transformations. Be-cause C provides the energy driving heterotrophic Ntransformations, such as mineralization and immobili-zation of inorganic N, increases in soil C availabilitycan increase microbial activity (Gallardo and Schle-singer 1992; Smith et al. 1994; Groffman 1999) andmicrobial N cycling (Vitousek and Matson 1985;Schimel et al. 1989; Davidson et al. 1990; Chen andStark 2000).

Soil microbial populations very often control theavailability of inorganic N (NH4

+ and NO3�), and their

biomass can act as a reservoir of relatively labile C andN compared to soil organic matter (Stevenson and Cole1999; Barrett and Burke 2000). Available inorganic Ncan be incorporated by plants and microbes, or trans-formed by chemoautotrophic and heterotrophic organ-isms via nitrification and denitrification. Intermediatesteps in both nitrification and denitrification result ingaseous losses of NO, N2O, and NOx. Other chemicallydriven gaseous N losses to the atmosphere can be causedby volatilization of ammonia due to alkaline soil pH(Stevenson and Cole 1999). While it is clear that soilmicrobes mediate the availability of N in most arid soils,it is not clear whether microbial population size oractivity level is more important in mediating soil Ntransformations in these pulse-dependent ecosystems. Itis also not well understood how microbial populations,soil C and N resources, and soil moisture interact toaffect ecosystem-level N dynamics over relatively shorttemporal scales (e.g., as soils dry down after a precipi-tation event).

Our objective in this study was to characterize and toidentify the controls of available soil C and N on soil Ncycling in response to a pulse of soil C, N, or both. Weadded labile C and N in the field along with a simulatedprecipitation pulse to soils in interspaces and under thecanopies of the major plant species of an arid ColoradoPlateau ecosystem. We hypothesized that the addition ofeither soil C, N, or the synergistic interaction of C and Ntogether would increase both microbial uptake rates ofinorganic N and rates of gaseous N losses via increasesin nitrification and denitrification. We also predictedthat there would be differences in the magnitude andpattern of the above responses across a landscape in thatmicrobial N cycling in the interspaces between plantcanopies would be relatively more limited by both C andN availability than those areas underneath plant cano-pies.

Materials and methods

Study site

The study site is located in the Needles district of Can-yonlands National Park, Utah, USA (38�07¢N,109�45¢W) approximately 1.5 km inside the park en-trance at an elevation of 1,475 m. The site has been freefrom recreation and grazing disturbances for approxi-mately 30 years. The site is an intact Sarcobatus vermi-culatus (Hook) Torr. and Atriplex canescens (Pursh)Nutt. desert scrub community located on a secondarystream terrace above Salt Creek. The interspaces be-tween shrubs are free of herbaceous annuals andperennials and are dominated by a well-developedbiological soil crust consisting of lichens, mosses, andfree-living cyanobacteria. Soils at the site are ustictorrifluvents derived from sandstone with a texture ofloamy very fine sand in the shallow A1 horizon(0–0.3 m). Soils are well drained and pH ranges from 8.0to 8.5 (USDA, Soil Conservation Service).

Treatment additions

In May of 2000, C and N additions were applied totwenty replicate soils under three distinct plant covertypes: (1) under the canopies of S. vermiculatus (ever-green shrub), (2) under A. canescens (deciduous shrub);and (3) in the interspace between plants for a total of 60experimental units across all treatments and cover types.With the exception of interspace soils, each soil replicatehad treatment locations at a point on the northeast sideof the plant to minimize diurnal variations in soil tem-perature. Soil treatments of C and N were applied ran-domly for each cover type in a full factorial designconsisting of control (no C or N additions), +C, +N,and C+N treatment combinations. For trace gas fluxmeasurements, treatments were applied inside PVCpipes measuring 25-cm diameter by 20-cm high that wereinserted in the soil to a depth of 10 cm following themethods outlined in Matson and Goldstein (2000) formeasuring soil gas fluxes at LTER sites. For measure-ments requiring destructive soil sampling, 25 cm by25 cm patches of soil were located next to each PVCpipe and surrounded by aluminum flashing that wasinserted in the soil to a depth of 10 cm. Both PVC andflashing were inserted 1 month prior to the start of theexperiment to decrease the effect of soil disturbance.Treatments were applied to soils as a 30-mm precipita-tion event, with control treatments receiving water only.Nitrogen treatments were 2.6 g (NH4)2SO4–N m�2,while carbon treatments consisted of 52 g dextrose-C m�2. For the 4 weeks prior to treatment addition,daily maximum air temperatures were relatively high(30– 40�C) with no measurable precipitation. The alu-minum flashing was removed after treatment applicationwhile PVC collars remained in the ground in order to

426

facilitate gas flux measurements. Soil samples were re-moved from flashing bordered areas for measurementsof soil moisture, inorganic N, extractable organic N,microbial biomass N. Samples were collected prior totreatment addition, as well as 1, 3, and 7 day afteraddition. Flux rates of N2O, NH3, and CO2 were taken1, 3, and 7 day after treatment addition. Samples ofuntreated bulk soil were collected for laboratory analysisof organic C and total N content.

Internal N cycling

Soil samples consisting of one 10-cm deep by 2.5- cmdiameter soil core were collected just outside flashingbordered areas before treatment and within those areasafter treatment addition for analysis of soil moisture,bulk density, inorganic N (NH4

+ and NO3�), extractable

organic N, microbial biomass N, organic C, and total Ncontent. Samples were immediately sealed in water-tightcontainers after collection and transported to a nearbyfacility for preparation the same day. Soil moisture wasdetermined gravimetrically by drying an 8 g subsampleat 60�C for 48 h. This was used to correct estimates ofinorganic N, extractable organic N, and microbial bio-mass N for differences in water content. Inorganic N wasestimated by extracting a 8-g subsample in 40 ml of 2 MKCl for 24 h, filtering extracts through filter paper(Whatman #4), and storing at 4�C until analysis. Ex-tracts were then transported back to the laboratory andNH4

+ and NO3� concentrations were determined colori-

metrically using an autoanalyzer (Alpkem FS3000, OIAnalytical, College Station, TX, USA).

Microbial biomass N was estimated using a modifiedchloroform (CHCl3) fumigation extraction techniqueoutlined by Brookes et al. (1985) and used by Gallardoand Schlesinger (1992) in a Chihuahuan Desert ecosys-tem. Two 8 g aliquots of soil were placed in separate50 ml centrifuge tubes. One tube was immediately ex-tracted in 40 ml 0.5 M K2SO4 for 24 h. Two large cot-ton balls dosed with 3 ml of hydrocarbon-free CHCl3were inserted into the headspace of the second tube thenincubated for 5 days at 30�C before extraction. Theextracts were then filtered through filter paper (What-man#4), and the extractant stored at 4�C. Total N in theextractant was digested to convert all organic N to NO3

�

using a persulfate digest (D’Elia et al. 1977). Nitrateconcentrations were measured colorimetrically using anautoanalyzer. The difference between fumigated andunfumigated subsamples, when divided by the extractionefficiency (Kn) gives the microbial biomass N. Microbialbiomass for the samples was calculated using a Kn of0.69 (Brookes et al. 1985, Gallardo and Schlesinger1992). The extractable organic N was determined fromunfumigated microbial biomass N samples and definedas the fraction of the organic N pool that was soluble in0.5 M K2SO4.

Soil organic C (SOC) and total N contents weremeasured on 1 g subsamples, which were returned to the

laboratory where they were air dried, ground to a finepowder, and washed three times with 3 N H3PO4 anddeionized water to remove carbonates. Samples werethen analyzed for %C and %N on a Carlo Erba ele-mental analyzer (NA1500 CHN Combustion Analyzer,Carlo Erba Strumentazione, Milan).

N outputs

Closed chamber measurements of CO2 and N2O fluxwere made 1, 3, and 7 day after treatment addition byplacing fitted tops to the PVC collars inserted into thesoil. Chamber tops were made of a 10-cm length of PVCpipe, an upper lid sealed with epoxy, and a closed-cellfoam gasket on the bottom. Collar tops had a male ca-jon ultra-torr fitting (Swagelok, Solon, OH, USA) with ablack butyl rubber septa installed (Bellco Glass Inc.,Vineland, NJ, USA) for gas sampling. Additional pro-tection from gas leaks was provided by placing a 5-cmwide band of tightly fitting rubber around the chambertop and bottom seal. Samples from chamber headspaceswere collected 0, 1, and 2 h after chamber tops wereplaced on collars. Gas samples were transported back tothe laboratory and analyzed for CO2 and N2O concen-tration using a gas chromatograph (GC-14A, ShimadzuScientific Instruments Inc., Dallas, TX, USA) equippedwith electron capture (for N2O) and thermal conduc-tivity (for CO2) detectors. Because regression analysisrevealed a linear increase in gas concentrations over the2-h chamber tops were in place, fluxes were calculatedbased on the slope of the linear regression of gas con-centration over time.

Ammonia volatilization was measured immediatelyafter CO2/N2O fluxes following the methods outlined bySchlesinger and Peterjohn (1991). An aliquot of 20 ml of2% (v/v) H2SO4 was placed in an open container insideeach collar. Collars were then covered with aluminumfoil for 24 h. Acid samples were collected after 24 h,brought back to the laboratory, and NH4

+ concentrationdetermined colorimetrically using the autoanalyzer.Differences between samples and blanks were used tocalculate flux rates.

Total N2 fluxes from soils resulting from denitrifica-tion of soil NO3

� were estimated using C2H2, to blocknitrous oxide reductase activity and force denitrifiedNO3

� to be released as N2O. A single 10-cm deep by 2.5-cm diameter soil core was collected from flashingsquares 1, 3, and 7 days after treatment addition. Soilcores were immediately sealed in air-tight mason jars(435-ml headspace) fitted with septa ports. Enough C2H2

was injected to create a 10% C2H2 atmosphere to blocknitrous oxide reductase activity. Headspace air sampleswere collected just after C2H2 injection and after 24-hincubation. Headspace gas samples were transportedback to the laboratory and analyzed for N2O concen-tration on the gas chromatograph. The N2 flux wascalculated as the difference in N2O production betweenthe beginning and end of the incubation.

427

Statistical analyses

A repeated measures, mixed random and fixed effectsanalysis (PROC MIXED, SAS 8.01) was used to deter-mine effect of C and N treatment, date of sampling,cover type, and their interaction on a given parameter.This test allowed us to model the covariance structure ofthe dataset to account for unevenly spaced samplingdates (Littell et al. 1996). A three-factor analysis ofvariance (PROC GLM, SAS 8.01) was used for deter-mining differences in SOC with C and N treatment,plant cover type, and their interaction. When necessary,data were log transformed for normality. All analyseswere performed using SAS statistical software (Cary,NC, USA). Statistical significance was determined ata=0.05. Errors are presented as one standard error ofthe mean.

Results

Soils from under plant canopies and the interspaces weresimilar in terms of carbonate content, bulk density, andorganic C/total N ratios (Table 1), but there were sig-nificant differences in the amount total N (F=5.83,P=0.0056) and organic C (F=3.65, P=0.0338). Soilsfrom under Sarcobatus contained the greatest amountsof both total N and organic C while soils from plantinterspaces contained the least. However, these differ-ences did not result in differences in the C:N ratio be-tween soils from the three cover types (P>0.9)(Table 2).

Soil moisture (Fig. 1) was uniformly low for all covertypes prior to treatment addition (0.27±0.02,0.36±0.04, and 0.37±0.08% by weight for Sarcobatus,Atriplex, and interspace, respectively). Treatment addi-tions (control, +C, +N, or C+N) did not have a sig-nificant effect on soil moisture at any time (P>0.6).Mixed model analyses did reveal a day by cover typeinteraction (F=4.01, P=0.0008). One day after treat-ment addition soils from under Sarcobatus and Atriplexwere significantly (P=0.0330) wetter than those frominterspaces (7.73±0.34, 7.13±0.28, and 6.14±0.20%for Sarcobatus, Atriplex, and interspace, respectively).Three days after pulse addition, soils under Sarcobatuswere significantly wetter (P=0.0085) than those fromunder Atriplex and plant interspaces (3.80±0.19,2.87±0.15, and 2.70±0.15% for Sarcobatus, Atriplex,and interspace, respectively). By day 7 post-pulse, soilmoisture levels were not significantly different (P>0.05)than those prior to pulse addition indicating that theduration of the entire moisture pulse (for the top 10 cmof soil) was captured over the course of this study.

Gaseous N losses (N2O, N2 from denitrification, andNH3) did not vary with cover type (P>0.1), but did varysignificantly with day in the case of N2O–N (F=8.32,P=0.0004) and N2–N (F=5.01, P=0.0079), and with aday by treatment interaction for NH3–N volatilization(F=47.20, P<0.0001). Fluxes of N2O–N from both T

able

1Significance

values

(P)at

a=0.5

from

independentmixed

model

(PROC

MIX

ED,SAS)analysisofallthedependantvariablesin

thisstudy

SOC

TotalN

C:N

Soilmoisture

N2O

flux

Total

denitrification

NH

3

volatilization

SoilNH

4+–N

SoilNO

3�–N

MBN

Organic

NSoil

respiration

Day

––

–<

0.0001

0.0004

0.0079

0.0155

0.1355

0.0411

<0.0001

<0.0001

<0.0001

Cover

0.0388

0.0056

0.8463

<0.0001

0.1211

0.6136

0.7966

0.4148

0.3482

0.0091

0.0181

<0.0001

C–

––

0.2438

0.3326

0.3186

<0.0001

0.1538

0.0034

0.2989

<0.0001

<0.0001

N–

––

0.7239

0.8968

0.9750

<0.0001

<0.0001

<0.0001

0.1393

<0.0001

0.0013

Day

·cover

––

–0.0008

0.0624

0.1167

0.2483

0.1419

0.3701

0.0056

0.0484

0.0003

Day

·C

––

–0.9947

0.5875

0.0893

0.3069

0.0018

0.5447

0.0756

0.0012

<0.0001

Day

·N

––

–0.7689

0.2943

0.1279

0.2415

0.4023

0.0027

0.5961

0.2262

0.0016

Day

·cover

·C

––

–0.8364

0.3580

0.6725

0.6154

0.3042

0.4422

0.8447

0.2746

<0.0001

Day

·cover

·N

––

–0.2042

0.4159

0.8737

0.9288

0.9116

0.7722

0.9376

0.1857

0.0003

Day

·C

·N

––

–0.6805

0.6654

0.4780

<0.0001

0.0018

0.0028

0.9608

0.7645

<0.0001

Day

·cover

·C

·N

––

–0.5384

0.6864

0.2241

0.8475

0.3048

0.2233

0.9580

0.8686

0.0311

Values

shownare

forallmain

effects

andallinteractionscontainingdayasafactor.Values

inbold

denote

thehighestorder

significantinteraction

428

nitrification and denitrification processes were the lowestof the three fluxes measured and declined over the courseof the experiment. The mean across all treatments andcover types was 2.3±0.4 on day 1 to 0.4±0.4 ng N2O–N m�2 h�1 on day 7 (Fig. 2a). The N2O–N flux on day 7was not significantly different from zero (P>0.3). Fluxesof N2–N from denitrification measured as N2O–N fluxfrom C2H2 block cores was greater than N2O–N fluxesin field and increased over the course of the experimentwith mean values across all treatments and cover typesof 10.8±6.1 lg N2–N m�2 h�1 on day 1 to 21.7±4.5 lgN2–N m�2 h�1 on day 7 (Fig. 2b). Nitrogen losses fromNH3 volatilization were the greatest of all measured soilN effluxes, ranging from 31.1±7.0 lg N2–N m�2 h�1

from control soils on day 1 to 404.7±45.5 lg N2–N m�2 h�1 from +N treated soils on day 3 (Fig. 2c).Regardless of cover type, soils treated only with N hadconsistently higher fluxes of NH3–N over the course ofthe experiment compared to controls (P<0.0001). Incontrast, soils treated with C or C+N exhibited rates ofNH3–N volatilization that were not different from con-trols (P>0.1).

Pools of inorganic N in the soil (NH4+–N and NO3

�–N) did not vary significantly with cover type and werecombined (P>0.5 for both NH4

+–N and NO3�–N), but

did vary with a day by treatment interaction (F=6.57,P=0.0018 for NH4

+–N and F=6.09, P=0.0028 forNO3

�–N). Prior to treatment addition (day 0), samplescollected from all locations showed no differences instanding NH4

+–N pools (mean of 1.0±0.3 g NH4+–

N m�2 across cover types and treatment locations,Fig 3a). One day after treatment addition, NH4

+–N poolsizes diverged with the greatest amount of N in +Ntreated soils (3.1±0.3 g NH4

+–N m�2) and the least in+C treated soils (0.2±0.1 g NH4

+–N m�2). By 7 daysafter treatment addition, there were no significant dif-ferences among treatments (P>0.1). Pools of NO3

�–Nwere also not different between treatment locations priorto treatment additions (mean of 0.22±0.07 g NO3

�–N m�2 across cover types and treatment locations,Fig. 3b). After treatment addition, there was signifi-cantly (F=6.09, P=0.0396) greater amounts of NO3

�–Nin +N treated soils, and significantly (F=8.86,P<0.0001) less NO3

�–N in +C and C+N treated soils,compared to control soils. This pattern persistedthroughout the entire experiment.

Microbial biomass N showed significant interactionwith cover type and day of experiment (F=5.31,P=0.0056) while organic N (extractable in 0.5 MK2SO4) showed a significant day by C treatment inter-action (F=5.48, P=0.0012). Microbial biomass N(Fig. 4a) was lowest before treatment addition (mean of0.57±0.18 g N m�2 across all cover types) with no dif-ferences between cover types (P>0.06). By one daypost-pulse for all treatment additions, there was greaterbiomass N in soils from under Sarcobatus (5.56±0.65 gN m�2) than in soils from plant interspaces(3.56±0.54 g N m�2). By day 3, and through to day 7,there were no significant differences between cover types(P>0.1). On day 7, biomass N was still greater than thatbefore C or N addition (mean of 1.24±0.25 g N m�2,P<0.0001). Organic N that was extractable in 0.5 MK2SO4 (Fig. 4b) was significantly different between daysonly when comparing treatments with no C added(control, +N) to those with C added (+C, C+N).Extractable organic N was not different between treat-ment locations before the treatments were added(P>0.5, mean of 5.2±1.0 g N m�2 across all treatmentlocations). Soils with C added exhibited a decrease inextractable N by day 1 and remained lower than soilswith no C added until day 7 (from 2.8±05 to 2.6±0.6and from 4.3±0.5 to 2.8±0.4 g N m�2 for soils with Cadded and no C added, respectively).

There was a significant interaction between day,treatment, and cover type for soil respiration (F=1.87,

Fig. 1 Mean gravimetric soil water content (%) in the top 10 cmprior to addition of water (day 0), and subsequent to water, C, andN addition (days 1, 3, and 7) to Colorado Plateau soils in 2000.Means are for soils from under the canopies of Sarcobatus andAtriplex, and the interspace between plants; with all treatmentscombined for each cover type. Asterisks (*) represent significantlydifferent (P<0.05) means within a given day. Error bars representone standard error (SE) of the mean

Table 2 Selected properties of soils, in the top 10 cm, collected in May 2000 from the study site in Canyonlands, N.P.

Total N (g m�2) Organic C (g m�2) C:N Bulk density (g/cm3)

A. canescens 25.1 (1.4) 226 (15)ab 9.0 (0.2)ab 1.3S. vermiculatus 29.8 (2.2) 272 (25)a 8.9 (0.2)a 1.3Interspace 22.3 (1.7) 201 (20)b 8.9 (0.3)b 1.3

Lowercase letters (a, b, c) indicate significantly different means (P<0.05) between cover types for a given parameter

429

P=0.0311). The highest CO2-C flux rates were observedcoming from soils under the canopies of Sarcobatus1 day after they had been treated with both C and N(Fig. 5, 360±48 mg CO2–C m�2 h�1). The lowest fluxrates were observed on day 7 from soils in the inter-spaces that were treated with N (4.5±4.5 mg CO2–C m�2 h�1). For all cover types soil respiration throughday 3 was consistently greater from soils treated with Cand C + N than from control and soils treated with Nonly (P<0.0041 for all treatments and cover types). Byday 7 of the experiment soil respiration was the lowest atany time for all cover types and there was no statisticaldifference in CO2–C fluxes between treatments or covertypes (P>0.1 for all treatments and cover types). Fluxesfrom control and N treated soils were not different withthe exception of soils from under Sarcobatus on day 1(P=0.0249).

Discussion

Our results suggest that pulsed additions of soil C and Nhave several key effects on soil N cycling. First, Soilmicrobial activity and N cycling responded positively toadded C suggesting a C-limitation to microbial N cy-cling. Increases in microbial activity were reflected inhigher rates of respiration and N immobilization with Caddition. Second, N availability has important implica-tions for the form of inorganic and organic N in the soil,and gaseous losses of N from the ecosystem. Nitrogenadditions resulted in an accumulation of soil NO3

�, andwhen both C and N were added to soils, N losses viaNH3 volatilization decreased. And third, there was noeffect of soil C or N availability on microbial biomasssuggesting that the level of microbial activity may be

Fig. 2 Mean soil nitrificationN2O flux (a), totaldenitrification N2O flux asmeasured via acetylene block(b), and NH3 volatilization (c)in the top 10 cm over the courseof a pulse of water and soilresources. Means for covertypes are combined for eachtreatment. Asterisks (*)represent significantly different(P<0.05) means within a givenday. Error bars represent onestandard error (SE) of the mean

Fig. 3 Mean soil NH4+–N (a)

and NO3�–N (b) pools in the top

10 cm prior to treatmentaddition (day 0) and over thecourse of a pulse of water, +C,+N, and +C+N. Means forcover types are combined foreach treatment. Asterisks (*)represent significantly different(P<0.05) means within a givenday. Error bars represent onestandard error (SE) of the mean

430

more important than population size (biomass) in con-trolling short-term dynamics of inorganic and labileorganic N. Also, the effects of C and N pulses on soilmicrobial function and pools of NH4

+–N and labile or-ganic N appear to last only for the duration of themoisture pulse while the effects on the soil NO3

� -N poolpersist after soils have dried. Spatial heterogeneity wasnot as large as we hypothesized because even though soilcover type affected microbial biomass and the potentialmaximum soil respiration rate when C was added, therewere no cover type-specific effects on the magnitude orpattern of responses of soil N transformations to pulsesof C or N availability.

Some of our observations suggest that water may bean overriding factor in the response of ecosystem Ncycling to pulse additions of resources. Microbial bio-mass, microbial activity, and N2O fluxes, measuredparameters increased, but the observed means were notsignificantly different from controls. This may mean thatthese parameters were more sensitive to the wateraddition than any C or N additions. While other studieshave shown that adding water and nutrients to soil may

increase (Agarwal et al. 1979; Alon and Steinberger1999; Van Gestel et al. 1993) or decrease microbialbiomass (Bottner et al. 1985; Groffman and Tiedje1988), microbial activity has been observed to increasefollowing water addition (Davidson et al. 1993; Schaef-fer et al. 2003). Thus, when sufficient moisture is presentfor microbial metabolism to occur, soil C and N avail-ability can have significant impacts on N dynamics. Inmost arid and semi-arid systems, soil C and N accu-mulate in dry periods between precipitation events(Davidson et al. 1993; Austin et al. 2004). Addition ofwater to these dry environments results in a flush ofavailable soil resources (Fierer and Schimel 2002) fromboth accumulated available C and N as well as that re-leased by the lysing of microbial cells caused by suddenchanges in soil water potential (Halvorson et al. 2000).

We observed the stimulation of microbial activitywhen water and C were added to soils. Increased activitylevels were reflected in greater net immobilization ofmineral N (independent of N additions) and lower ratesof gaseous N losses. Increased SOC has been shown tocontribute directly to increased microbial immobiliza-

Fig. 4 Mean microbial biomass N and organic N extractable in0.5 M K2SO4 pools in the top 10 cm prior to treatment addition

(day 0) and over the course of a pulse of water, C, and N. Formicrobial biomass N, means for treatments are combined for eachcover type. For extractable organic N, means for cover types arecombined for each treatment. Asterisks (*) represent significantly

Fig. 5 Mean soil respiration in the top 10 cm over the course of a pulse of water and soil resources. Asterisks (*) represent significantlydifferent (P<0.05) means within a given day. Error bars represent one standard error (SE) of the mean

431

tion of N (Smith et al. 1994; Gallardo and Schlesinger1995; Vance and Chapin 2001; Schaeffer et al. 2003) thatlowers N availability for nitrification and denitrification(Westerman and Tucker 1978; Mummy et al. 1994;Gallardo and Schlesinger 1995) and NH3 volatilization(Schlesinger and Peterjohn 1991). Denitrification rates inthis study, as measured by the acetylene block method(21.7 lg N m�2 h�1), were well below the mean ratesobserved by Peterjohn and Schlesinger (1991) for Chi-huahuan desert soils (329 lg N m�2 h�1), but closer tomean rates observed by Billings et al. (2003) for Mojavedesert soils (6.7 lg N m�2 h�1). In contrast to thefindings of Peterjohn and Schlesinger (1991), N2O fluxesfrom soils in the field and N2 fluxes via acetylene blockwere not affected by C or N addition, suggesting that therelative rates of nitrification and denitrification are notsensitive to changes in soil C or N. Fluxes of N2O andN2 were not measured until 24 h after treatment addi-tion, however, and a brief pulse of N2O from these soilsmay not be detected with our methods (Billings et al.2003; Schaeffer et al. 2003). Addition of C and N alsoaffected losses of N via NH3 volatilization. Ammoniavolatilization rates from control soils averaged 31 lgN m�2 h�1 and were greater than values reported in theliterature for the Chihuahuan (Schlesinger and Peter-john 1991) and Mojave (Billings et al. 2003) deserts.Similar to the findings of Billings et al. (2003), NH3

volatilization appears to be a more important pathwayfor gaseous N loss than has been previously thought forother arid systems (Peterjohn and Schlesinger 1990).This may also have important implications for control-ling nitrification rates and NO3

� concentrations (Praveenand Aggarwal 1998).

While not a primary limitation to soil N dynamics, Navailability has important implications for gaseous Nlosses and the balance of inorganic N pools in the soil.We observed that the pulse of NH4

+ added to the soilwas no longer in the soil NH4

+ pool after 3 days whilesoil NO3

� accumulated over the 7-day experiment. Theadded N may have moved into other soil N poolsthrough several means: (1) via nitrification to NO3

�, (2)direct loss via NH3 volatilization, and (3) microbialuptake and incorporation into organic N forms andpossibly the subsequent mineralization to NO3

�. Fluxesof NH3, while high with N addition, do not account forall the added NH4

+ leaving the soil NH4+ pool. The

maximum possible amount of N lost via NH3 by takingthe maximum flux of NH3 observed (405 lg N m�2 h�1)for 7 days would be 68 mg N m�2, which is far less thanthe pulse of N added to the soil (approximately 2–3% of2.6 g N m�2 that was added). Fluxes of N2O–N and N2

from total denitrification were even lower (0.000017 and0.02%, respectively). We cannot calculate the fluxes ofadded to and from inorganic and organic N pools as wemeasured pool size, which is the net result of severalfluxes. Even so, the increase in extractable organic Nwith N addition is consistent with the hypothesis thatmicrobial populations are taking up inorganic N, whichmay then be released as organic or inorganic N with

metabolism or upon cell death (Halvorson et al. 2000).An important aspect of N dynamics not addressed byour study are gross flux rates of inorganic N, whichwould provide a more accurate estimate of the residencetimes of soil N in inorganic pools.

Soils under plant canopies had significantly greaterpools of C and N than interspace soils, and these dif-ferences in C and N resource availability between covertypes may have allowed microbial populations underplant canopies to respond to a relatively greater degreeto resource pulses than those in the interspaces. Thesedifferences however, were not great enough to lead todifferences in soil N availability (inorganic and extract-able organic) and gaseous N losses. Microbial biomass isknown to be positively correlated to available soil Ccontent (Gallardo and Schlesinger 1995), and activitylevels are also positively correlated with substrateavailability (Barrett and Burke 2000). The pattern weobserved may be due to differences in microbial com-munity composition, or in the timing of microbialactivity between locations.

Our study has important implications for predictingthe response of ecosystem N dynamics to processes thatmay alter established patterns of pulses of soil C or Navailability such as changes in either the frequency orintensity of precipitation events (Fierer and Schimel2002). The timing and intensity of precipitation eventswill determine the pattern and magnitude of microbialactivity responses and hence soil N dynamics in thesearid Colorado Plateau ecosystems (Austin et al. 2004).Changes in N cycling similar to those we observed in thisstudy (a greater proportion of available N beingimmobilized and residing in the ecosystem as microbialbiomass N and lower rates of gaseous N loss), thesubsequent reduction in plant available N may altercompetitive relationships leading to changes in plantdominance (Hooper and Vitousek 1998; Billings et al.2002). Even though the single wetting/drying cycle usedin this study showed few longer-term effects on soil Ndynamics, the effects of the frequency or intensity ofwet/dry cycles on soil C and N availability remain to bedetermined.

Acknowledgements This project was funded by NSF EcosystemStudies Program and Ecological and Evolutionary PhysiologyProgram (grants 98-14358 and 98-14510). We gratefully acknowl-edge the National Park service and the USGS-BRD. Thanks go toLynda Sperry, Sharon Billings, Steve Trimble, Hannah Schrum,Nicole Hardiman, Lynette Duncan, Sue Phillips, and Jayne Belnap.

References

Agarwal AS, Singh BR, Kanehiro Y (1979) Soil nitrogen andcarbon mineralization as affected by drying—rewetting cycles.Soil Sci Soc Am J 35:96–100

Alon A, Steinberger Y (1999) Response of the soil microbial bio-mass and nematode population to a wetting event in nitrogen-amended Negev desert plots. Biol Fertil Soils 30:147–152

Austin AT et al (2004) Water pulses and biogeochemical cycles inarid and semiarid ecosystems. Oecologia 141:221–235

432

Barrett JE, Burke IC (2000) Potential nitrogen immobilization ingrassland soils across a soil organic matter gradient. Soil BiolBiochem 32:1707–1716

BassiriRad H, Tremmel DC, Virginia RA, Reynolds JF, de SoyzaAG, Brunell MH (1999) Short-term patterns in water andnitrogen acquisition by two desert shrubs following a simulatedsummer rain. Plant Ecol 145:27–36

Billings SA, Schaeffer SM, Zitzer S, Charlet TD, Smith SD, EvansRD (2002) Alterations of nitrogen dynamics under elevatedcarbon dioxide in an intact Mojave Desert ecosystem: evidencefrom nitrogen-15 natural abundance. Oecologia 131:463–467

Billings SA, Schaeffer SM, Evans RD (2003) Trace N gas losses andN mineralization in an intact Mojave Desert ecosystem withelevated CO2. Soil Biol Biochem 34:1777–1784

Bolton H, Jr., Smith JL, Link SO (1993) Soil microbial biomassand activity of a disturbed and undisturbed shrub-steppe eco-system. Soil Biol Biochem 25:545–552

Bottner P (1985) Response of microbial biomass to alternate moistand dry conditions in a soil incubated with 14 C and 15 N-labelled plant material. Soil Biol Biochem 17:328–337

Brookes PC, Landman A, Pruden G, Jenkison DS (1985) Chlo-roform fumigation and the release of nitrogen: a rapid directextraction method to measure microbial biomass nitrogen insoil. Soil Biol Biochem 17:837–842

Chen J, Stark JM (2000) Plant species effects and carbon andnitrogen cycling in a sagebrush—crested wheatgrass soil. Soilbiology and biochemistry 32:47

Cross AF, Schlesinger WH (1999) Plant regulation of soil nutrientdistribution in the northern Chihuahuan Desert. Plant Ecol145:11–25

Cui C, Caldwell M (1997) A large ephemeral release of nitrogenupon wetting of dry soil and corresponding root reponses in thefield. Plant Soil 191:291–299

Davidson EA et al (1993) Processes regulating soil emissions of NOand N2O in a seasonally dry tropical forest. Ecology 74:130–139

Davidson EA, Stark JM, Firestone MK (1990) Microbial pro-duction and consumption of nitrate in an annual grassland.Ecology 71:1968–1975

D’Elia CF, Stueder PA, Corwin N (1977) Determination of totalnitrogen in aqueous samples using persulfate digestion. LimnolOceanogr 22:760–764

Fierer N, Schimel JP (2002) Effects of drying-rewetting frequencyon soil carbon and nitrogen transformations. Soil Biol Biochem34:777–787

Gallardo A, Schlesinger WH (1992) Carbon and nitrogen limita-tions of soil microbial biomass in desert ecosystems. Biogeo-chemistry 18:1–17

Gallardo A, Schlesinger WH (1995) Factors determining soilmicrobial biomass and nutrient immobilization in desert soils.Biogeochemistry 28:55–68

Gebauer RLE, Ehleringer JR (2000) Water and nitrogen uptakepatterns following moisture pulses in a cold desert community.Ecology 81:1415–1424

Groffman PM (1999) Carbon additions increase nitrogen avail-ability in northern hardwood forest soils. Biol Fertil Soils29:430–433

Groffman PM, Tiedje JM (1988) Denitrification hysteresis duringwetting and drying cycles in soil. Soil Sci Soc Am J 52:1626–1629

Halverson LJ, Jones TM, Firestone MK (2000) Release of intra-cellular solutes by four soil bacteria exposed to dilution stress.Soil Sci Soc Am J 64:1630–1637

Hooper DU, Vitousek PM (1998) Effects of plant composition anddiversity on nutrient cycling. Ecol Monogr 68:121–149

Littel RC, Milikin GA, Stroup WW, Wolfinger RD (1996) SASsystem for mixed models. SAS Institute, Cary

Matson PA, GoldsteinAH (2000) Biogenic trace gas exchanges. In:Sala OE, Jackson RB, Mooney HA, Howarth RW (eds)Methods in ecosystem science. Springer, New York, pp 235–248

Mummy DL, Smith JL, Bolton H Jr (1994) Nitrous oxide flux froma shrub-steppe ecosystem: sources and regulation. Soil BiolBiochem 26:279–286

Noy-Meir I (1973) Desert ecosystems: environment and producers.Annu Rev Ecol Syst 4:25–51

Peterjohn WT, Schlesinger WH (1990) Nitrogen loss from desertsin the southwestern United States. Biogeochemistry 10:67–79

Peterjohn WT, Schlesinger WH (1991) Factors controlling deni-trification in a Chihuahuan Desert ecosystem. Soil Sci Soc Am J55:1694–1701

Praveen K, Aggarwal RK (1998) Interdependence of ammoniavolatilization and nitrification in arid soils. Nutr Cycling Ag-roecosyst 51:201–207

Schaeffer SM, Billings SA, Evans RD (2003) Response of nitrogendynamics in an intact Mojave Desert ecosystem to manipula-tions in soil carbon and nitrogen availability. Oecologia134:547–553

Schimel JP, Jackson LE, Firestone MK (1989) Spatial and tem-poral effects on plant-microbe competition for inorganicnitrogen in a California annual grassland. Soil Biol Biochem21:1059–1066

Schlesinger WH, Peterjohn WT (1991) Processes controllingammonia volatilization from Chihuahuan Desert soils. Soil SciSoc Am J 52:54–58

Schlesinger WH, Pilmanis AM (1998) Plant-soil interactions indeserts. Biogeochemistry 42:169–187

Schlesinger WH, Raikes JA, Hartley AE, Cross AF (1996) On thespatial pattern of soil nutrients in desert ecosystems. Ecology77:364–374

Smith JL, Halvorson JJ, Bolton H Jr (1994) Spatial relationshipsof soil microbial biomass and C and N mineralization in asemi-arid shrub-steppe ecosystem. Soil Biol Biochem 26:1151–1159

Stevenson FJ, Cole MA (1999) Cycles of soil: carbon, nitrogen,phosphorous, sulfur, micronutrients, 2nd edn. Wiley, NewYork

Van Gestel M, Merck R, Vlassak K (1993) Microbial biomass re-sponses to soil drying and rewetting: the fate of fast- and slow-growing microorganisms in soils from different climates. SoilBiol Biochem 25:109–123

Vance ED, Chapin FS III (2001) Substrate limitations to microbialactivity in taiga forest floors. Soil Biol Biochem 33:173–188

Virginia RA, Jarell WM, Franco-Vizcaino E (1982) Direct mea-surements of denitrification in a Prosopis (Mesquite) dominatedSonoran Desert ecosystem. Oecologia 53:120–122

Vitousek PM, Matson PA (1985) Disturbance, nitrogen availabil-ity, and nitrogen losses in an intensively managed loblolly pineplantation. Ecology 66:1360–1376

Weltzin JF, McPherson GR (2003) Changing precipitation regimesand terrestrial ecosystems: a North American perspective, 1stedn. The University of Arizona Press, Tucson

West NE, Skujins J (1977) The nitrogen cycle in North Americancold-winter semi-desert ecosystems. Oecologia Plantarum12:45–53

Westerman RL, Tucker TC (1978) Factors affecting denitrificationin a Sonoran Desert soil. Soil Sci Soc Am J 42:596–600

Zaady E, Groffman PM, Shachaak M (1996) Litter as a regulatorof N and C dynamics in macrophytic patches in Negev Desertsoils. Soil Biol Biochem 28:39–46

433