SOIL MICROBIOLOGY

Linking Microbial Community Structure and Functionto Seasonal Differences in Soil Moisture and Temperaturein a Chihuahuan Desert Grassland

Colin W. Bell & Veronica Acosta-Martinez &

Nancy E. McIntyre & Stephen Cox & David T. Tissue &

John C. Zak

Received: 15 January 2009 /Accepted: 25 April 2009 /Published online: 23 May 2009# Springer Science + Business Media, LLC 2009

Abstract Global and regional climate models predicthigher air temperature and less frequent, but largerprecipitation events in arid regions within the next century.While many studies have addressed the impact of variableclimate in arid ecosystems on plant growth and physiolog-ical responses, fewer studies have addressed soil microbialcommunity responses to seasonal shifts in precipitation andtemperature in arid ecosystems. This study examined theimpact of a wet (2004), average (2005), and dry (2006) yearon subsequent responses of soil microbial communitystructure, function, and linkages, as well as soil edaphicand nutrient characteristics in a mid-elevation desertgrassland in the Chihuahuan Desert. Microbial communitystructure was classified as bacterial (Gram-negative, Gram-positive, and actinomycetes) and fungal (saprophytic fungiand arbuscular mycorrhiza) categories using (fatty acidmethyl ester) techniques. Carbon substrate use and enzymicactivity was used to characterize microbial community

function annually and seasonally (summer and winter). Therelationship between saprophytic fungal community struc-ture and function remained consistent across seasonindependent of the magnitude or frequency of precipitationwithin any given year. Carbon utilization by fungi in thecooler winter exceeded use in the warmer summer eachyear suggesting that soil temperature, rather than soilmoisture, strongly influenced fungal carbon use andstructure and function dynamics. The structure/functionrelationship for AM fungi and soil bacteria notably changedacross season. Moreover, the abundance of Gram-positivebacteria was lower in the winter compared to Gram-negative bacteria. Bacterial carbon use, however, washighest in the summer and lower during the winter. Enzymeactivities did not respond to either annual or seasonaldifferences in the magnitude or timing of precipitation.Specific structural components of the soil microbiotacommunity became uncoupled from total microbial func-tion during different seasons. This change in the microbialstructure/function relationship suggests that different com-ponents of the soil microbial community may providesimilar ecosystem function, but differ in response toseasonal temperature and precipitation. As soil microbesencounter increased soil temperatures and altered precipi-tation amounts and timing that are predicted for this region,the ability of the soil microbial community to maintainfunctional resilience across the year may be reduced in thisChihuahuan Desert ecosystem.

Introduction

Climate models suggest that arid ecosystems will bestrongly affected by changes in temperature and precipita-tion within the next century [1]. Deserts are particularly

Microb Ecol (2009) 58:827–842DOI 10.1007/s00248-009-9529-5

C. W. Bell (*) :N. E. McIntyre :D. T. Tissue : J. C. ZakDepartment of Biological Sciences, Texas Tech University,Lubbock, TX 79409-3131, USAe-mail: colin.bell@ttu.edu

V. Acosta-MartinezWind Erosion and Water Conservation Unit, USDA,3810 4th Street,Lubbock, TX 79415, USA

S. CoxThe Institute for Environment and Human Health,Texas Tech University,Lubbock, TX 79409-3131, USA

D. T. TissueCentre for Plant and Food Science, University of Western Sydney,Richmond, NSW 2753, Australia

susceptible to changes in climate [2–11], as higher airtemperature and reduced soil moisture may greatly affect soilchemical processes, soil microbial activity, and plant produc-tivity as well as higher trophic dynamics [5, 8, 12, 13].

In all terrestrial ecosystems, soil microbes maintain soilstructure and facilitate the degradation of organic matter,nutrient cycling and carbon sequestration [14]. Zak et al.[15] suggested that the high environmental variability andresource heterogeneity associated with arid ecosystems mayincrease microbial functional diversity. While there havebeen many studies conducted to assess relationships betweenabove-ground biodiversity and ecosystem function [16–19],fewer efforts have focused on linkages between soilmicrobial heterogeneity and subsequent activity in responseto temporal shifts in temperature and soil moisture in aridecosystems [20–22]. Consequently, our understanding of thefactors that determine relationships between soil microbialheterogeneity and functional responses to temporal dynamicsin desert ecosystems, and how these temporal responsesdetermine ecosystem function, remains unclear [23–25].

Despite decades of research on soil microbial activity inarid systems [7, 26–32], understanding ecosystem links tobelow-ground biodiversity dynamics in desert systems isrelatively unknown. Although the capability to enumeratethe diversity and composition of microbial communities insoil has improved, quantifying microbial communitystructure relative to microbial functional contributions tothe ecosystem remains complex [23, 25, 33, 34]. Methodssuch as Biolog [25] and Fungilog [35] have been developedto provide insights into the role of soil microbial functionaldiversity in regulating ecosystem function. These proce-dures have been recognized as an effective tool forestimating microbial response to disturbance and environ-mental stress (e.g., [36–38]. Soil enzyme activity can alsoelucidate soil microbial processes as a function of environ-mental conditions [39]. Soil enzymes facilitate organicmatter decomposition and C, N, P, and S nutrient trans-formations in soil. Soil enzyme activity is often correlatedwith edaphic properties and may be predictive of soilmicrobial functional dynamics [39–42]. In arid ecosystems,

soil enzyme activity is highly dependent on the quality andquantity of soil organic matter as well as seasonal moistureinputs [43, 44]. The analysis of fatty acid profiles may beused to assess microbial temporal responses and changes insoil fungal and bacterial community structure with depth inmany semiarid and arid ecosystems [30, 45, 46].

Global climate change predictions of higher soil tem-perature and increased variability in precipitation magni-tude and frequency within arid ecosystems [1] indicate theimportance of assessing soil biological responses toenvironmental change in order to understand the magnitudeof climate-related disturbance to these landscapes. In thisresearch, we examined relationships between soil microbialcommunity structure (bacterial and fungal abundance) andsoil microbial function (carbon substrate use and enzymeactivity) as they were influenced by annual and seasonalvariation in soil moisture, temperature, and nutrientavailability in this Chihuahuan Desert sotol grassland atBig Bend National Park. We hypothesized that therelationship between fungal functional activity and com-munity structure would not be influenced by seasonaltemperature and moisture variability. However, given theimportance of soil moisture to bacterial survival, therelationship between bacterial functional activity andcommunity structure for the bacterial component of thesoil microflora should change in response to seasonalmoisture and temperature variability between years. Thelarger goal for this research is to provide insight as to howglobal climate change predicted for this region mayinfluence soil microbial community structure and functionand subsequent interactions with above-ground plantspecies richness and productivity.

Materials and Methods

Site Description

This research was conducted in a Chihuahuan Desert sotolgrassland in the Pine Canyon Watershed [47], located in

Jan Feb Mar Apr May JuneJuly Aug Sep Oct Nov Dec

Ave

rage

Mon

thly

Pre

cipi

tatio

n (m

m)

0

20

40

60

80

100

120

1402004 (567 mm)2005 (329 mm)2006 (273 mm)

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

1976-2006 (360 mm)(a) (b)Figure 1 Historical monthlyprecipitation. Monthly precipi-tation averages (mm) for Pan-ther Junction Visitor Center(elevation 1,143 m) in Big BendNational Park for (a) 2004-2006and b historical 19-year average

828 C. W. Bell et al.

Big Bend National Park (BBNP), in west Texas. Theelevation of this desert grassland ranges from 1,500 to1,678 m. Mean daily air temperatures at the research siterange from 10°C in the winter months to 32°C in thesummer. Based upon data collected by BBNP researchstaff, the 19-year rainfall average for the park is 365 mm.On a seasonal scale, 10% of the annual rainfall occursduring winter (December–February; 38 mm), and 46% ofthe annual rainfall occurs during the summer (June–August;168 mm; Fig. 1). Over this 3-year study, annual precipita-tion steadily declined from a wet year in 2004 (567 mm;155% of the 19-year average), to average precipitation in2005 (329 mm; 90% of the 19-year average), and dryconditions in most of 2006 (273 mm; 74% of the 19-yearaverage). A more detailed description of the historicalclimate conditions at this site can be found in Robertson etal. [48]. Soil texture throughout this site is sandy loam(62% sand, 30% silt, 8% clay), as determined by WatersAgricultural Laboratories, Inc. (Owensboro, KY, USA). Amore detailed description of the soil characteristics at thisresearch site can be found in Bell et al. [38].

Experimental Design

In 2002, 12 3×3 m plots were established to study long-term soil microbial and plant responses to seasonal patternsin precipitation. Each plot contained the three dominantperennial plant species (Dasylirion leiophyllum, Opuntiaphaecantha, Bouteloua curtipendula) of the sotol grasslandsite. A more detailed description of the plant characteristicsat this research site can be found in Patrick et al. [49]. Soilcollections consisted of two composite soil samplescollected from each community plot (12 treatment plots,24 composite samples collected during each sample periodacross the study). One composite soil sample consisted of atleast four sub-samples within the plot to provide the bestpossible representation of soil nutrient and chemicalproperties for the plot as a whole. In every compositesample, soils were collected equally from under eachdominant plant (Dasylirion, Opuntia, and Bouteloua) aswell as from interplant spaces. Although interplant spacescan be characterized as bare area between plants (up toapproximately 1 m away from the base of above-groundplant structures), the expansive and shallow root systems ofthe dominant plants within each 3×3 m plot at this researchsite would suggest that the soil nutrient and chemicalparameters collected within each plot are in all likelihoodaccessible to the plants within each plot. Twenty-four soilsamples (two per plot) were collected biannually in March(end of winter) and September (end of summer) for threeconsecutive years (2004–2006). The rationale for thissampling strategy was to compare the responses betweenthe two seasons that were predicted to experience the

greatest shift in precipitation patterns due to global climatechange in this region [4, 5]. Moreover, the multi-yearsampling design allowed us to capture intra-seasonalenvironmental variability (March versus September)along with broader patterns of inter-season responses.Soil samples were collected from 0 to 15 cm depths(sieved through a 2.0-mm sieve in the field) and stored at4°C, and all soils were analyzed within 2 weeks ofcollection.

Soil Environmental Parameters (Moisture, Temperature,and pH)

Gravimetric soil moisture was measured at each samplingtime by drying soils in a drying oven at 60°C for 48 h [50].Soil temperatures (15 cm depth) were measured using OnsetComputer Corporation HOBO-H8 Pro Series data loggers at36-min intervals [51]. For every sample period, average soiltemperatures were calculated for the 4 weeks prior to soilsampling, which represented the critical time period formicrobial response to varying environmental conditions. SoilpH was measured using a 2:1 soil–DI H2O paste extract withan Orion pH/ISE model 710A ion-specific probe [52].

Soil Organic Matter and Soil Nitrogen Dynamics

Soil organic matter (SOM) was estimated via loss-on-ignition method [53]. Soils were air-dried for 24 h, placed inaluminum pans, and oven-dried for an additional 24 h in a 60°Cdrying oven. After drying, approximately 10 g of soil wasplaced in a weighed porcelain crucible and then incinerated at450°C for 24 h. After incineration, samples were removed andre-weighed to determine remaining ash mass. Exchangeablesoil ammonium (NH4

+-N) was determined via colorimet-ric assay using a 50-ml 2-M KCl solution from 5 g field-moist (dry weight equivalent) soil samples [54]. Levelsof extractable NO3-N were determined by A&L SoilLaboratories (Lubbock, TX, USA) using ion-specificprobes.

Microbial Biomass Carbon

Microbial biomass carbon was determined using thechloroform fumigation and extraction method [55]. In brief,two replicate 5.0-g dry weight equivalent sub-samples fromeach composite sample were fumigated with chloroform for48 h and extracted using a 50-ml (0.5 M) K2SO4 solution.The soil solutions were filtered using Fisherbrand-P2 fineporosity filter paper and measured spectrophotometricallyat 280 nm [56]. A non-fumigated set of samples weretreated in the same manner to serve as controls. Thedifference between the extracted C in the fumigated andnon-fumigated soils represented the microbial biomass C.

Linking Structure and Function to Seasonal Differences 829

Microbial Carbon Use

Microbial carbon guild use in response to environmentalconditions was assessed as described by the Biolog methodfor bacteria [25, 57] and the Fungilog method for fungi[35]. In brief, bacterial responses were quantified using soilsample dilutions (10−4), of which 150 µl of the soil dilutionwas inoculated into each well of the Biolog GN2 96-wellmicrotiter plates. Fungal functional responses were assessedby inoculating Biolog SFN2 96-well microtiter plates with100 µl of a 20-ml inoculation mixture containing 50 mg SOMparticles, in which soil organic matter particle size wascontrolled for and separated from the mineral soil fractionfollowing the soil Fungilog procedures [35]. The carbonsubstrates in the Biolog GN2 and SFN2 96-well microtiterplates (95 differing substrates) were categorized into sevendifferent carbon guilds: simple carbohydrates, carboxylicacids, amino acids, complex carbohydrates, polymers,amines–amides, and nucleotides [35, 58]. Functional responsesfor fungi and bacteria were quantified for total substrate guildactivity after a 72-h incubation period at 25°C [25].

Soil Enzyme Assays

The activities of β-glucosidase, β-glucosaminidase, andphosphodiesterase were assayed as described by [59, 60]

using 1 g of soil (<2 mm, air-dried) and incubated for 1 h attheir optimal pH. The product released (p-nitrophenol) wasdetermined colorimetrically at 400 nm. Enzyme activitieswere assayed in duplicate with one control to which thesubstrate was added after stopping the reaction. β-Glucosidaseis the key enzyme for the last limiting step of cellulosedegradation (C cycle), cleaving cellobiose β1–4 bonds linkingtwo glucose molecules [61]. β-glucosaminidase activity [59]provides information on chitin degradation, which is a keyenzyme involved in the hydrolysis of N-acetyl-β-D-glucos-amine. This hydrolysis is important in C and N cycling insoils because it participates in the processes whereby chitin isconverted to amino sugars, which is a major source ofmineralizable N in soil [62]. Phosphatases are crucial inorganic and inorganic P transformation, but are also signifi-cantly affected by soil pH, which controls P availabilityindependent of organic matter content or levels of disturbance.

Microbial Community Structure (FAME Analysis)

Fatty acid methyl ester (FAME) analysis was conducted onfield-moist soils following procedures developed for pureculture isolates and subsequently soil applications byAcosta-Martinez et al. [63]. Fungal and bacterial fatty acidswere saponified and methylated as outlined by SherlockMicrobial Identification System (MIDI, Inc., Newark, DE,

Table 1 Pearson correlation matrix for key seasonal soil microbial and environmental parameters measured in March and September in a sotolgrassland in Big Bend National Park

Sap AM G (+) G (–) Actino MBC FSA BSA β-gluc β-glsm Phos

March

SM% -0.30 −0.26 −0.47 −0.10 −0.60 −0.05 0.55 0.49 −0.10 −0.35 0.16

Temp −0.33 −0.19 0.08 −0.11 −0.27 −0.16 0.61 0.32 −0.26 −0.40 −0.06SOM −0.01 0.01 −0.02 −0.05 0.11 0.38 0.04 0.11 0.34 0.50 0.57

NO3−N −0.18 −0.18 −0.21 0.05 −0.19 0.26 0.34 0.35 −0.09 −0.11 0.24

NH4-N 0.11 0.05 0.27 0.06 0.38 0.19 −0.13 0.02 −0.11 0.34 0.09

pH 0.06 0.28 0.08 0.13 0.14 0.24 0.14 −0.09 0.29 −0.04 0.64

Sept

SM% −0.03 −0.28 0.21 0.38 0.45 0.45 0.67 0.61 0.27 0.32 0.38

Temp 0.37 −0.08 −0.01 0.40 0.25 0.50 0.31 0.05 0.18 −0.03 0.26

SOM 0.01 0.04 0.08 0.06 −0.01 0.00 −0.01 0.33 0.16 0.15 0.16

NO3-N −0.05 −0.32 0.07 −0.09 −0.03 −0.13 0.09 −0.09 −0.08 0.26 −0.13NH4-N 0.29 −0.48 0.22 0.48 0.53 0.47 0.70 0.51 0.21 0.29 0.20

pH −0.05 −0.42 0.21 −0.25 −0.07 −0.28 −0.08 0.23 −0.25 −0.10 −0.50

Pearson correlation matrix indicates microbial community structural and functional correlations to environmental variables for a March and bSeptember. Values in this matrix can range from −1.0 to 1.0, with 1.0 indicating perfectly correlated variables, and −1.0 indicating perfectlynegative correlations. Significant correlations (P≤0.05) are indicated in bold. The sample size for each variable used for each microbial andenvironmental parameter in the Pearson correlation matrix listed above is N=72

Sap saprophytic fungi, AM arbuscular mycorrhiza, G (+) Gram-positive bacteria, G (−) Gram-negative bacteria, Actino actinomycetes bacteria,MBC microbial biomass carbon, FSA potential carbon usage by fungi, BSA potential carbon usage by bacteria, β-gluc potential activity of β-glucosidase, β-glsm potential activity for β-glucosaminidase, Phos potential activity for phosphodiesterase, SM% soil moisture %, Temp soiltemperature, SOM soil organic matter, NO3-N extractable soil nitrate, NH4-N extractable soil NH4-N, pH soil pH

830 C. W. Bell et al.

USA). Microbial community structure was assessed usingmethyl-ester derivatives from the extracted lipids, in whichknown microbial markers were recorded via output peaksduring gas chromatography. The microbial output-peakprofiles were identified using the MIDI Sherlock peakidentification software, which is a fatty acid recognitionsignature program used to fingerprint and associate similaritiesof fatty acid markers to different microbial functional groups[64]. The relative abundance of soil microbial functionalgroups was assessed by using specific fatty acid markers.

Fungal structural groups were categorized into two sub-groups: saprophytic fungi and arbuscular mycorrhiza. Thespecific lipid indicators used to assess the abundance of thesaprophytic fungal components were 18:1w9c, 18:2w6,9,18:3w3c, 18:3w6c, and 20:5w3 [65–71]. The indicatorsspecific to arbuscular mycorrhiza were 16:1w5c, 20:1w9c,20:2w6c, and 22:1w9c [70–72].

Bacterial structural groups were categorized into threesub-groups: Gram-positive (G+) bacteria, Gram-negative(G–) bacteria, or actinomycetes. The lipid indicators used to

assess the abundance of the G+ bacterial soil componentwere 14:0iso, 15:0, 15:0iso, 15:0anteiso, 16:0iso, 17:0iso,17:0anteiso [65–68, 73–75]. Gram-negative bacteria, char-acterized by shorter chained lipid structures (often eithercomprised of cyclo-propyl lipids or mono-unsaturatedlipids) were 16:1w9c, 16:1w7c, 16:1w7t, cyclo17:0,18:1w7c, 18:1w5c, and cyclo19:0 [65–68, 73–75]. Actino-mycetes were characterized by the lipid indicators10Me16:0, 10Me17:0, and 10Me18:0 [65–67, 73, 76, 77].

Statistical Analysis

To identify seasonal pair-wise relationships between soilmicrobial parameters and seasonal environmental parame-ters, a Pearson correlation matrix was calculated incorpo-rating all microbial and environmental parameters in March(representing post-winter conditions) and September (rep-resenting post-summer conditions) using the statisticalsoftware package SPSS 14.0 (Chicago, IL, USA; Table 1).Sample sizes for every variable in both Pearson correlationmatrices was N=72. Pearson correlations between speciesand environmental parameters were considered significantat (P≤0.05).

Differences across sampling periods for microbialcommunity abundance (FAME indicators) and microbialexoenzyme activity over the 3-year period were evaluatedusing repeated-measures ANOVA (rmANOVA) with con-secutive sample dates as the within-subject variable (Figs. 2

Rel

ativ

e F

unga

l Abu

ndan

ce (

%)

6.0

7.5

9.0

10.5

12.0

13.5

15.0

16.5

18.0Sap AM

Sample Date

Mar 04 Sept 04 Mar 05 Sept 05 Mar 06 Sept 06

Rel

ativ

e B

acte

rial A

bund

ance

(%

)

1.21.41.61.82.02.22.42.62.83.03.23.43.6

6.0

8.0

10.0

12.0

14.0 G+ G- (lower-case italics) Actino (Greek)

B B

B

AA

A

cab

bc

aab

c

B AB

C

AB

AAB

bc

c

bcb

aba

χ βχ χ β

α α

(a)

(b)

Figure 2 FAME analysis. Temporal response patterns of microbialcommunity structure expressed as relative abundance of a fungi and bbacteria from the sotol grassland in Pine Canyon, Big Bend N.P.,using FAME analysis procedures. Values are mean ± SE; N=24.Letters above error bars indicate significant differences at P≤0.05using Fisher’s LSD post hoc tests. Sap saprophytic fungi, AMarbuscular mycorrhiza, G+ Gram-positive bacteria, G− Gram-negative bacteria, Actino actinomycetes bacteria

Sample Date

Mar 04 Sept 04 Mar 05 Sept 05 Mar 06 Sept 06

mg

p-ni

trop

heno

l (kg

-1 s

oil h

-1)

40

80

120

160

200

240

400

500

600β-gluc Phosβ-glsm

C

BC

A

BC

ABAB

b b b bb

a

χ β β β

α α

Figure 3 Enzyme assays. Temporal response patterns of potentialenzyme activity of soil microbial communities in regards to β-glucosidase, β-glucosaminidase, and phosphodiesterase (expressed ascolorimetric absorbance from phenol oxidation after 1 h of incubation)for soil microbial communities from the sotol grassland in PineCanyon, Big Bend N.P. Values are mean ± SE; N=24. Letters aboveerror bars indicate significant difference at P≤0.05 using Fisher’sLSD post hoc tests. β-gluc potential activity of β-glucosidase, β-glsmpotential activity for β-glucosaminidase, Phos potential activity forphosphodiesterase

Linking Structure and Function to Seasonal Differences 831

and 3). To determine significant seasonal contrasts withineach microbial and environmental parameter, multivariateanalysis of variance (MANOVA) was performed using allmicrobial and environmental parameters collected in March(post-winter) and September (post-summer) 2004–2006.Confidence intervals (CI=0.95) were calculated andreported to better characterize the mean variability acrossyears. Sample sizes for mean MANOVA comparisons wasN=72, and seasonal differences in all microbial andenvironmental parameters were considered significant at(P≤0.05) using the Wilks’ Lambda test statistic. Post hocindependent mean comparisons were performed followingMANOVA on all environmental and microbial parametersusing the Bonferroni correction to adjust for the multipleindependent variable comparisons.

To determine microbial community relationships toseasonal environmental conditions, representational differ-ence analysis (RDA; constrained PCA or redundancyanalysis) was performed using soil microbial communityparameters (FAME indicators, soil enzyme activity, andcarbon substrate utilization) as the response variables to theenvironmental parameters (predictor variables) using Can-oco for Windows 4.5 (Ceske Budejovice, Czech Republic)[78]. The RDA analysis was chosen because thistechnique identifies specific responses of measured param-eters elucidated from environmental conditions in multi-dimensional space, thus allowing for species andenvironmental variation to be correlated within principalcomponents (ordination axes) that represent maximumvariation shared by microbial and environmental relation-ships simultaneously. Displaying RDA directional relation-ships and strength relationships (constrained correlations)for all microbial and environmental parameters in RDAordination plots was used to assess soil microbialseasonal-scale responses to the environmental conditions.In each RDA ordination graph, axes 1 and 2 were chosenbecause these axes accounted for the most variationreported by the eigenvalues.

Results

Pair-Wise Correlations

In March, significant correlations (P≤0.05) between micro-bial and environmental parameters were observed in 46%of the measured pair-wise correlations, with 10% exhibitingstrong (r=±0.5) correlations (Table 1). Soil moisture hadthe greatest impact on microbial parameters at this time ofyear. Community structure (FAME levels) displayed negativecorrelations with soil moisture for Gram (+) bacteria (r=−0.47)and actinomycetes (r=−0.60). Contrastingly, microbial func-tional parameters were positively correlated with soil mois-

ture for both fungal substrate activity (r=0.55), and bacterialsubstrate activity (r=0.49). Fungal substrate activity alsodisplayed a positive correlation with soil temperature (r=0.61). β-Glucosaminidase activity was positively correlat-ed with soil organic matter levels (r=0.50), whilephosphodiesterase activity was positively correlated withsoil organic matter levels (r=0.57) and soil pH (r=0.64)(Table 1).

In September, significant correlations between microbialand environmental parameters were observed in 48% of themeasured pair-wise correlations, with 18% exhibitingstrong (r=±0.5) correlations. Soil NH4-N levels had thegreatest impact on microbial parameters at this time of year.Negative correlations were observed between microbialparameters and soil NH4-N levels for AM fungi (r=−0.48);while positive correlations were observed between micro-bial parameters and soil NH4-N for Gram (−) bacteria (r=0.48), actinomycetes (r=0.53), microbial biomass (r=0.47),fungal substrate activity (r=0.70), and bacterial substrate

Table 2 Pearson correlation matrix for soil microbial community andfunctional parameters measured in March and September in a sotolgrassland in Big Bend National Park

FSA BSA β-gluc β-glsm Phos

March

Sap −0.45 −0.19 0.16 0.24 0.13

AM −0.23 −0.26 0.19 0.21 0.22

G (+) −0.06 −0.23 −0.09 0.08 −0.11G (−) −0.22 −0.07 0.06 0.11 0.12

Actino −0.33 −0.38 0.13 0.45 0.12

MBC −0.21 −0.13 0.23 0.44 0.43

Sept

Sap 0.19 0.04 0.08 −0.10 −0.01AM −0.41 −0.40 0.05 −0.17 0.36

G (+) 0.30 0.34 0.14 0.16 −0.08G (−) 0.49 0.35 0.32 0.20 0.39

Actino 0.43 0.39 0.23 0.29 0.20

MBC 0.31 0.17 0.34 0.14 0.44

Pearson correlation matrix indicates microbial community structuralcorrelations to microbial functional parameters for March andSeptember. Values in this matrix can range from −1.0 to 1.0, with1.0 indicating perfectly correlated variables, and −1.0 indicatingperfectly negative correlations. Significant correlations (P≤0.05) areindicated in bold. The sample size for each variable used for eachmicrobial and environmental parameter in the Pearson correlationmatrix listed above is N=72

Sap saprophytic fungi, AM arbuscular mycorrhiza, G (+) Gram-positive bacteria, G (−) Gram-negative bacteria, Actino actinomycetesbacteria, MBC microbial biomass carbon, FSA potential carbon usageby fungi, BSA potential carbon usage by bacteria, β-gluc potentialactivity of β-glucosidase, β-glsm potential activity for β-glucosaminidase, Phos potential activity for phosphodiesterase

832 C. W. Bell et al.

activity (r=0.51). Soil moisture had the next greatestimpact on microbial parameters in September, displayingpositive correlations with actinomycetes (r=0.45), microbi-al biomass (r=0.45), fungal substrate activity (r=0.67), andbacterial substrate activity (r=0.61). Microbial biomass wasalso positively correlated with soil temperature (r=0.50) atthis time of year. Activity of phosphodiesterase wasnegatively correlated with soil pH (r=−0.50) in September.Correlations of the other enzymes were not significantlycorrelated to any measured abiotic parameter in September(Table 1).

To examine intra-seasonal pair-wise relationships be-tween microbial function and structure in the sotol grass-lands, Pearson correlations were calculated betweenmicrobial community structure (FAME indicators andmicrobial biomass) and microbial functional parameters(carbon activity and extra-cellular enzyme activity) inMarch and September (Table 2). In March, 33% of thepair-wise correlations were significant, but only 6% (twopair-wise correlations) exhibited strong (r=±0.5) correla-

tions, as saprophytic fungi were negatively correlated withfungal substrate activity (r=−0.45), and actinomycetesabundances were positively correlated with β-glucosaminidase (r=0.45; Table 2). Overall, correlationsobserved in March demonstrated that fungal substrateactivity was negatively correlated with microbial commu-nity structure, although significant correlations were onlyobserved with saprophytic fungi, AM fungi, and actino-mycetes. Likewise, bacterial substrate activity was nega-tively correlated with microbial community structure, withsignificant correlations between levels of AM fungi, Gram(+) bacteria and actinomycetes. Contrastingly, the majorityof the extra-cellular enzyme activities were positivelycorrelated with microbial community structure, and allenzyme activities were significantly correlated to levels ofmicrobial biomass in March.

In September, 53% of the pair-wise correlations weresignificant, but only 3% (one pair-wise correlation)exhibited strong (r=±0.5) correlations, as Gram (–) bacteriawas positively correlated with fungal substrate activity (r=

Table 3 Seasonal mean comparisons (at 95% confidence intervals) for microbial community structure, function, and environmental parameters

95% CI Mean 95% CI P value

Lower Upper March September Lower Upper

Soil parameters

SM 3.85 5.38 4.61 7.08 6.32 7.84 0.000

Temp 13.64 14.10 13.87 25.70 25.46 25.93 0.000

SOM 70.97 77.74 74.35 67.21 63.83 70.59 0.004

NO3-N 2.05 2.95 2.50 5.18 4.73 5.62 0.000

NH4-N 8.16 12.28 10.22 9.40 7.34 11.46 0.578

pH 6.15 6.41 6.28 6.15 6.02 6.28 0.144

Microbial community

Sap 12.24 14.23 13.24 13.85 12.86 14.84 0.389

AM 7.00 8.80 7.90 8.26 7.36 9.16 0.577

G (+) 7.92 8.96 8.44 9.49 8.97 10.01 0.006

G (−) 2.15 2.76 2.46 2.58 2.27 2.88 0.576

Actino 1.98 2.38 2.18 2.19 1.99 2.39 0.959

MBC 192.76 368.51 280.64 430.59 342.72 518.46 0.018

Microbial function

FSA 12.22 19.81 16.02 14.95 11.16 18.75 0.696

BSA 9.89 15.22 12.56 15.83 13.16 18.50 0.088

β-gluc 439.62 534.20 486.91 479.62 432.33 526.90 0.830

β-glsm 79.08 94.46 86.77 94.72 87.03 102.41 0.150

Phos 168.65 205.20 186.92 209.59 191.32 227.86 0.085

Seasonal comparisons for key environmental parameters, soil microbial community parameters, and soil microbial functional parameters,measured in a sotol grassland in Big Bend National Park. Significant seasonal differences (P values) were computed using post hoc one-wayANOVA. The sample size for each microbial and environmental parameters within each season listed above is N=72

Sap saprophytic fungi, AM arbuscular mycorrhiza, G (+) Gram-positive bacteria, G (−) Gram-negative bacteria, Actino Actinomycetes bacteria,MBC microbial biomass carbon, FSA potential carbon usage by fungi, BSA potential carbon usage by bacteria, β-gluc potential activity of β-glucosidase, β-glsm potential activity for β-glucosaminidase, Phos potential activity for phosphodiesterase, SM% soil moisture %, Temp soiltemperature, SOM soil organic matter

Linking Structure and Function to Seasonal Differences 833

0.49; Table 2). Overall, the majority of the Septembercorrelations between microbial function and structure werepositive. Like March observations, fungal and bacterialsubstrate activity was negatively (P≤0.05) correlated withAM fungal levels. Contrastingly, in September, fungal andbacterial substrate activity was positively correlated to theremaining FAME indicators. Extra-cellular enzyme activi-ties differed in their responses to microbial structureindicators. β-Glucosidase was positively correlated (P≤0.05) to Gram (−) bacteria, actinomycetes, and microbialbiomass. β-Glucosaminidase was positively correlated (P≤0.05) to actinomycetes, and phosphodiesterase was posi-tively correlated (P≤0.05) to AM fungi, Gram (−) bacteria,and microbial biomass. In September, Gram (−) bacteriaand actinomycetes were found to be positively correlatedwith more microbial functional parameters than any othermicrobial structural component (Table 2).

Temporal Contrasts

Soil Parameters

Significant differences in soil parameters were evidentwhen comparing values for March and September usingMANOVA (F6, 137=992.19; P<0.001; Table 3). Indepen-dent pair-wise mean comparisons determined that soilmoisture in September was significantly higher than inMarch (P<0.001; Table 3). Repeated measures ANOVAdemonstrated that the highest soil moisture occurred duringthe September 2006 sampling period (P<0.001; Fig. 1a).Across the study, mean soil temperature in September

(25.7°C) was significantly higher than mean soil temperaturein March (13.9°C; P<0.001; Table 3). Soil organic matterwas significantly higher in March compared to September(P=0.004; Table 2). In contrast, extractable soil NO3-N wassignificantly lower in March than in September (P<0.001;Table 3). Neither levels of extractable soil NH4-N nor soilpH differed significantly across season (Table 3).

Microbial Community Structure

Overall, microbial community structural parameters (FAMElevels) were significantly higher in September than inMarch (F6, 137=4.61; P<0.001; Table 2). However, G+bacteria abundance (F1, 142=7.86; P=0.006) and microbialbiomass carbon (F(1, 142)=5.69; P=0.018) were the onlytwo independent pair-wise microbial community structuralparameters that were significantly higher in September thanin March (Table 3).

The relative abundances of all microbial functionalgroups were dynamic across years and sampling periods(Fig. 2a, b). Repeated measures ANOVA demonstrated thatthe relative abundance of saprophytic fungi was signifi-cantly higher during the last three sample periods (P<0.001), during drought-like conditions in September 2005and March 2006, and immediately after large, late-summerrainfall events in September 2006 (Fig. 2a). Arbuscularmycorrhiza relative abundance was highly variable acrossyears and seasons. The lowest abundance was observedafter the 2004 summer that experienced high rainfall, and inMarch 2006, after the dry winter (Fig. 2a). Gram (+)bacterial abundance remained relatively stable over the

95% CI Mean 95% CI P value

Lower Upper March September Lower Upper

Fungal guild (FSA)

Simple carbohydrates 3.43 5.26 4.34 3.37 2.46 4.29 0.139

Carboxylic acids 1.85 3.79 2.82 3.81 2.84 4.78 0.155

Amino acids 1.86 3.66 2.76 3.45 2.55 4.34 0.289

Complex carbohydrates 3.43 4.85 4.14 2.02 1.31 2.73 <0.001

Polymers 0.83 1.08 0.99 0.88 0.75 1.00 0.367

Amines/amides 0.40 0.91 0.65 0.92 0.66 1.17 0.155

Nucleotides 0.14 0.33 0.23 0.26 0.16 0.36 0.733

Bacterial guild (BSA)

Simple carbohydrates 1.98 3.43 2.71 3.62 2.89 4.35 0.083

Carboxylic acids 3.90 5.82 4.86 6.06 5.10 7.02 0.082

Amino acids 1.90 3.14 2.52 2.61 1.99 3.23 0.841

Complex carbohydrates 0.48 1.23 0.85 1.63 1.26 2.01 0.004

Polymers 0.54 0.85 0.69 0.81 0.65 0.97 0.302

Amines/amides 0.32 0.72 0.52 0.77 0.57 0.97 0.087

Nucleotides 0.11 0.25 0.18 0.13 0.06 0.21 0.379

Table 4 Seasonal meancomparisons (w/ 95% confi-dence intervals) for fungal andbacterial potential substrateguild utilization

Seasonal comparisons for fungalcarbon substrate guild usage andbacterial carbon substrate guildusage measured in a sotolgrassland in Big Bend NationalPark. Significant seasonal dif-ferences (P values) were com-puted using post hoc one-wayANOVA. Significant seasonaldifferences (P≤0.05) for all po-tential substrate utilizationparameters for fungi and bacte-ria are indicated in bold. Thesample size for fungal and bac-terial guilds (substrate utiliza-tion) within each season listedabove is N=72

834 C. W. Bell et al.

study except in March 2005, when significantly lowerlevels were recorded relative to the other sampling periods(P≤0.009; Fig. 2b). The Gram (−) and actinomycetesfunctional groups, which had the lowest abundancethroughout the study, progressively increased after theSeptember 2004 sampling period, with the highest FAMElevels for both Gram (−) and actinomycetes bacterialabundance occurring during the 2006 sampling year(Fig. 2b).

Independent of season or year (as evaluated by one-wayANOVA), saprophytic fungi were relatively more abundant(12.8–14.2% of the total FAME levels (P≤0.001) thanGram (+) bacteria (8.5–9.3%) and AM fungi (7.4–8.5%;Fig. 2a). The abundance of both Gram (−) bacteria andactinomycetes did not significantly differ between oneanother and were the least abundant of the microbialcommunity functional groups throughout the study, rangingfrom 2.0% to 2.7%; (Fig. 2b).

Microbial Functional Dynamics

Overall microbial community function (evaluated byMANOVA) was significantly higher in September than inMarch (F5, 138=2.44; P<0.037; Table 3). Seasonal shifts inmicrobial use of complex carbohydrates were observed forboth fungi and bacteria (Table 4). Complex carbohydrateuse exhibited inverse seasonal patterns between bacteriaand fungi, as the fungal utilization of complex carbohy-drates was significantly higher in March (P<0.001)compared to September (Table 4), while soil bacterialpotential to utilize complex carbohydrates was signifi-cantly higher in September than in March (P=0.004;Table 4).

Extra-cellular enzyme activities (β-glucosidase, β-glucosaminidase, and phosphodiesterase) did not differsignificantly across seasons (Table 3). However, β-glucosidase activity was the most variable across seasons.Phosphodiesterase activity was highest after large rainfallevents in September 2006 (P≤ 0.042), while β-glucosaminidase displayed the highest activity levels in bothMarch 2006 (dry conditions) and September 2006 (largerainfall events; P≤0.038; Fig. 3). Independent of season oryear, differences in relative potential enzyme activities wereobserved. Throughout the study, β-glucosidase activity wasthe highest (P<0.001; Fig. 3), followed by phosphodiester-ase (P<0.001), with β-glucosaminidase activity the lowest(P<0.001; Fig. 3).

Comparisons Between Fungal and Bacterial Activity

Overall, soil fungi were more functionally dominant withrespect to complex carbohydrate use than bacteria. InMarch, fungal activity (FSA) was significantly higher than

bacterial activity (BSA; P≤0.001; Fig. 4a). In September,although bacteria displayed higher activity potential than inMarch, there were no significant differences betweenbacterial and fungal utilization (P=0.34; Fig. 4b). Furtherseasonal carbon use comparisons revealed that fungalactivity exceeded bacterial activity in March with respectto simple carbohydrates (4.343–2.706 nm; P=0.006),complex carbohydrates (4.140–0.851 nm; P<0.001), andpolymers (0.958–0.693 nm; P=0.009; Fig. 4a), whilebacterial activity exceeded fungal activity in the utilizationof carboxylic acids (4.858–2.815 nm; P=0.001; Fig. 4a). InSeptember, differences between the microbial groups wereobserved for only two carbon substrate guilds; bacterialutilization of carboxylic acids exceeded fungi (6.059–3.806 nm; P=0.001), and fungi displayed higher activityon nucleotide utilization over bacterial (0.258–0.133 nm; P=0.047; Fig. 4b).

Tot

al S

ubst

rate

Act

ivity

(M

arch

)

0

2

4

6

FungiBacteria

(P = 0.006)

(P = 0.003)

(P < 0.001)

(P = 0.009)

Carbon Guild

Simple

Car

b

Carbo

xylic

Acid

s

Amino

Acid

s

Comple

x Car

b

Polym

ers

Amine

s/Am

ides

Nucleo

tides

Tot

al S

ubst

rate

Act

ivity

(S

epte

mbe

r)

0

2

4

6

(P = 0.047)

(P = 0.006)

(a)

(b)

Figure 4 Fungal versus bacterial seasonal substrate (carbon guild)utilization. Seasonal response patterns of potential carbon guild usagefor soil fungi and bacteria measured as substrate activity for a Marchand b September from soils collected in the sotol grassland in PineCanyon, Big Bend N.P. using Biolog and Fungilog procedures after72 h of incubations. Values are mean ± SE; N=72/carbon guildsubstrate

Linking Structure and Function to Seasonal Differences 835

Carbon guild activity taken as a whole (evaluated byrmANOVA) was highly variable across sample periods forsoil fungi and bacteria. Across all sampling dates, signif-icant differences were observed in total carbon activity forsoil fungi (P≤0.03; Fig. 5a). During the first 2 years of thestudy, potential fungal carbon use (FSA) exhibited in-creased activity in March (associated with decreasedtemperature) versus September. However, this patterndramatically shifted in September 2006 (after large late-summer rainfall events within 30 days prior to sampling) asfungal activity was significantly higher than previouslyobserved (P<0.001; Fig. 5a). Bacterial carbon use (BSA)was also highly variable across sample dates, and similarto fungal activity, the highest bacterial activity wasrecorded in the September 2006 sampling period (P<0.001; Fig. 5b).

Soil Microbial Functional and Structural Relationships

Bacterial Structural and Functional Relationshipsin Response to Season

During March 2004–2006, RDA analysis distinguishedbacterial FAME indicators, extra-cellular enzyme activity,and carbon substrate utilization as the three distinctcategories in response to yearly variation of precipitationand soil edaphic and nutrient levels (Fig. 6). The variationin FAME indicators (Gram-positive, Gram-negative, andactinomycetes) closely corresponded to levels of availablesoil NH4-N (Fig. 6), whereas enzyme activities (β-glucosidase, phosphodiesterase, and β-glucosaminidase)corresponded to soil organic matter and soil pH (mostlyassociated with the second axis). Bacterial potential carbon

-1.0 1.0

-1.0

1.0

G-

G+

Actino

BSA

B-gluc

Phos

B-glsm

SM%

SOM

NO3-N

NH4-N

pH

Axis 1: 0.39

Axi

s 2:

0.2

7

Figure 6 March 2004–2006 bacterial structural and functionalparameters as a function of environmental parameters. RDA graphdisplaying significant post-winter season response of bacterialstructural and functional parameters (response variables) as a functionof environmental parameters (predictor variables) for all soil samplescollected from the sotol grassland in Pine Canyon, Big Bend N.P. inMarch (2004–2006). Labels for axes 1 and 2 are the eigenvaluesassociated with each axis; this model accounts for 66% of totalvariance. The mean soil temperature (at the sampling site) in Marchover the 3-year period was ~13.9°C. The sample size for species andenvironmental variables in this model is N=72. G+ Gram-positivebacteria, G− Gram-negative bacteria, Actino actinomycetes bacteria,BSA potential carbon usage by bacteria (72-h incubation), B-glucpotential activity of β-glucosidase, B-glsm potential activity for β-glucosaminidase, Phos potential activity for phosphodiesterase, SM%soil moisture %, SOM soil organic matter, NO3-N extractable soilnitrate, NH4-N extractable soil NH4-N, pH soil pH

Fun

gal U

tiliz

atio

n S

ubst

rate

Gui

lds

369

1215182124273033363942454851 Simple Carb

Carboxylic Acids Amino Acids Complex CarbPolymers Amines/Amides Nucleotides

Sample DateMar 04 Sept 04 Mar 05 Sept 05 Mar 06 Sept 06

Bac

teria

l Util

izat

ion

Sub

stra

te G

uild

s

369

1215182124273033363942454851

(a)

(b)

B

E

C

B

BCCD

DEE

A

E

D

A

Figure 5 Temporal substrate (carbon guild) utilization. Temporalresponse patterns of potential carbon guild usage measured assubstrate activity for a total fungal activity (FSA) over all sampleperiods and b total bacterial activity (BSA) over all sample periods,from soils collected in the sotol grassland in Pine Canyon, Big BendN.P. using Biolog and Fungilog procedures after 72-h incubations.Values are mean ± SE; N=24/sample period

836 C. W. Bell et al.

guild activity (BSA) corresponded closely with soilmoisture and available soil NO3-N. The cumulativeeigenvalue for the first two axes (the total percent variationrepresented) was 0.66 (Fig. 6).

In September, bacterial responses to environmentalparameters dramatically shifted from those observed inMarch, as RDA analysis portrayed the FAME indicators,extra-cellular enzyme activity, and carbon substrate utiliza-tion as one general group (Fig. 7). Relationships betweenbacterial and environmental parameters indicated thatbacterial FAME indicators, exoenzyme activity, and poten-tial carbon guild activity at the end of summer allcorresponded to yearly variation in soil moisture, soilorganic matter, and available soil NH4-N. Phosphodiester-ase activity in September followed similar patterns ob-served in March and corresponded to yearly variation insoil pH. The cumulative eigenvalue for the first two axes inthe RDA analysis for soil bacterial responses to environ-

mental conditions in September (the total percent-variationrepresented) was 0.44 (Fig. 7).

Fungal Structural and Functional Relationshipsin Response to Season

Across all post-winter sampling dates (March 2004–2006),RDA analysis indicated that fungal FAME, β-glucosidase,and β-glucosaminidase levels responded similarly to yearlyvariation in precipitation and soil edaphic and nutrientparameters (Fig. 8). Phosphodiesterase activity and fungalcarbon substrate utilization (FSA), however, were influ-enced by a different suite of environmental factors thanFAME levels or enzyme activities (Fig. 8). Fungal responseto environmental conditions suggested that variationexhibited in the fungal FAME indicators (saprophytic andarbuscular mycorrhiza), β-glucosidase activity, and β-glucosaminidase activity closely corresponded to yearly

-1.0 1.0

-1.0

1.0

AM

Sapro

FSA

B-gluc

Phos

B-glsm

SM%

SOM

NO3-N

NH4-N

pH

Axis 1: 0.29

Axi

s 2:

0.0

9

Figure 8 March 2004–2006 fungal structural and functional param-eters as a function of environmental parameter. RDA graph displayingsignificant post-winter season response of fungal structural andfunctional parameters (response variables) as a function of environ-mental parameters (predictor variables) for all soil samples collectedfrom the sotol grassland in Pine Canyon, Big Bend N.P. in March(2004–2006). Labels for axes 1 and 2 are the eigenvalues associatedwith each axis; this model accounts for 37% of total variance. Theaverage soil temperature (at the sampling site) in March over the3-year period was ~13.9°C. The sample size for species andenvironmental variables in this model is N=72. Sap saprophyticfungi, AM arbuscular mycorrhiza, FSA potential carbon usage by fungi(72-h incubation), B-gluc potential activity of β-glucosidase, B-glsmpotential activity for β-glucosaminidase, Phos potential activity forphosphodiesterase, SM% soil moisture %, SOM soil organic matter,NO3-N extractable soil nitrate, NH4-N extractable soil NH4-N, pH soil pH

-1.0 1.0

-1.0

1.0

G-

G+

Actino

BSA

B-gluc

Phos

B-glsm

SM%

SOM

NO3-N

NH4-N

pH

Axis 1: 0.37

Axi

s 2:

0.0

76

Figure 7 September 2004–2006 bacterial structural and functionalparameters as a function of environmental parameters. RDA graphdisplaying significant post-summer season response of bacterialstructural and functional parameters (response variables) as a functionof environmental parameters (predictor variables) for all soil samplescollected from the sotol grassland in Pine Canyon, Big Bend N.P. inSeptember (2004–2006). Labels for axes 1 and 2 are the eigenvaluesassociated with each axis; this model accounts for 44% of totalvariance. The average soil temperature (at the sampling site) inSeptember over the 3-year period was ~25.7°C. The sample size forspecies and environmental variables in this model is N=72. G+ Gram-positive bacteria, G− Gram-negative bacteria, Actino actinomycetesbacteria, BSA potential carbon usage by bacteria (72-h incubation), B-gluc potential activity of β-glucosidase, B-glsm potential activity forβ-glucosaminidase, Phos potential activity for phosphodiesterase, SM% soil moisture %, SOM soil organic matter, NO3-N extractable soilnitrate, NH4-N extractable soil NH4-N, pH soil pH

Linking Structure and Function to Seasonal Differences 837

differences in available soil NH4-N. Phosphodiesteraseactivity was influenced primarily by yearly differences insoil organic matter and soil pH. Fungal substrate activitywas most influenced by yearly differences in soil moistureand levels of available soil NO3-N following the winterperiod. The cumulative eigenvalue for the first two axes inthe RDA analysis for soil fungal responses to environmen-tal conditions in March (the total percent variationrepresented) was 0.37 (Fig. 8).

In September, fungal dynamics dramatically differedfrom March, as the RDA analysis associated saprophyticFAME indicators, β-glucosidase, β-glucosaminidase, andfungal carbon substrate activity as one general group(Fig. 9). Phosphodiesterase activity and arbuscular mycor-rhiza was characterized as a second group, exhibiting lowor negative correlations to the other fungal communitystructure components. Phosphodiesterase activity exhibitedslightly negative relationships with saprophytic fungi while

AM fungi displayed negative relationships with saprophyticfungal indicators. In September, saprophytic FAME indica-tors, β-glucosidase activity, β-glucosaminidase activity,and fungal carbon substrate activity closely correspondedto soil moisture and available soil NH4-N. Phosphodiester-ase activity exhibited strong positive correlations with soilorganic matter and soil pH. AM fungi and phosphodiester-ase activity demonstrated strong, negative relationshipswith levels of available NO3-N following summer. Thecumulative eigenvalue for the first two axes in the RDAanalysis for soil fungal responses to environmental con-ditions in September (the total percent variation repre-sented) was 0.59 (Fig. 9).

Discussion

In this desert grassland, the relationship between sapro-phytic fungal community structure and function remainedconsistent across season independent of the magnitude orfrequency of precipitation within a given year as distin-guished by the RDA analysis. This desert system histori-cally experiences wet summers and dry winters, and overthis 3-year study, annual precipitation varied across years(567 mm in 2004, 329 mm in 2005, and 273 mm in 2006).While summer soil moisture levels across the study variedannually, fungal function and structure was positivelycorrelated from year to year, suggesting that declines in thetotal precipitation received was not sufficient to uncouplethe relationship between structure and function. Further-more, fungal carbon use in the winter exceeded thatobserved in the summer, suggesting that factors other thansoil moisture (e.g., higher levels of soil organic matter,cooler temperatures in winter) strongly influenced themagnitude of fungal carbon use and structure/functiondynamics in the winter. These results suggest that themoderate winter temperatures in this desert grassland aremore favorable for fungal activity than in summer. Thehigher summer temperature extremes may cross a func-tional threshold for the soil fungi in this ecosystem, whichwould explain why fungal activity declines during thesummer season.

While the relationship between saprophytic fungal com-munity structure and function was consistent in response toyearly intra-seasonal amounts of precipitation, the structure/function relationship for AM fungi and soil bacteria notablychanged across season. The divergence in AM fungaldynamics from microbial functional parameters in latesummer (September) suggests that AM fungi do not respondto the same suite of environmental conditions that controlssaprophytic fungal activity throughout the summer. Otherstudies have also observed that AM fungal communitystructure and species richness respond to seasonal shifts in

-1.0 1.0

-1.0

1.0

AM

Sapro

FSA

B-Gluc

Phos

B-Glsm

SM%SOM

NO3-N

NH4-N

pH

Axis 1: 0.53

Axi

s 2:

0.0

6

Figure 9 September 2004–2006 fungal structural and functionalparameters as a function of environmental parameters. RDA graphdisplaying significant post-summer season response of fungal struc-tural and functional parameters (response variables) as a function ofenvironmental parameters (predictor variables) for all soil samplescollected from the sotol grassland in Pine Canyon, Big Bend N.P. inSeptember (2004–2006). Labels for axes 1 and 2 are the eigenvaluesassociated with each axis; this model accounts for 59% of totalvariance. The average soil temperature (at the sampling site) inSeptember over the 3-year period was ~25.7°C. The sample size forspecies and environmental variables in this model is N=72. Sapsaprophytic fungi, AM arbuscular mycorrhiza, FSA potential carbonusage by fungi (72-h incubation), B-gluc potential activity of β-glucosidase, B-glsm potential activity for β-glucosaminidase, Phospotential activity for phosphodiesterase, SM% soil moisture %, SOMsoil organic matter, NO3-N extractable soil nitrate, NH4-N extractablesoil NH4-N, pH soil pH

838 C. W. Bell et al.

host availability and patterns in fine root development [79].Similarly, negative relationships between mycorrhizal hostplant availability and increased soil organic matter have beenobserved in arid ecosystems [22, 80], which is consistentwith winter soil conditions in this desert grassland site, assoil organic matter levels were higher in the winter relative tothe end of the summer. The AM functional/structural patternsobserved in the winter months suggest that while above-ground activity and/or host root growth was reduced duringthis period, AM fungi may produce additional assimilativehyphae or enzymes, sporulate, or initiate other soil microbialinteractions [81–83]. The functional ability of AM fungi waslimited without a host for extended periods because AMfungi rely on plant root symbionts as their sole source ofcarbon [82]. Therefore, seasonal shifts in the relationshipbetween AM fungal community structure and functionalattributes suggests that root growth patterns and physiolog-ical demands for phosphorous by the perennial grasses inthis ecosystem uncouple AM fungal community structurefrom overall microbial functional parameters in the summermonths rather than abiotic constraints imposed by soiltemperature and moisture.

Bacterial functional characteristics in this Chihuahuandesert grassland were similar to those in other desertsystems in that soil bacterial communities were physiolog-ically responsive to rainfall and soil moisture pulses (andsubsequent soil nitrogen availability) during warmer peri-ods of the year [30, 84]. The decrease in G (+) bacterialabundances observed in the winter months (typical of lowersoil temperature and moisture, and less soil NO3-Navailability) suggests that seasonal climate strongly influ-enced the abundance of soil bacteria. For example, soilbacteria maintained higher activity levels associated withhigher temperature and soil moisture in the summer anddemonstrated positive relationships between function andstructure, as well as exhibiting higher carbon substrateutilization than soil fungi. In both summer and winter, soilbacteria exhibited the highest substrate activity utilizingcarboxylic acids, with somewhat lower utilization ofcarbohydrates and amino acids. Soil bacteria were func-tionally dominant in the utilization of carboxylic acids,surpassing fungi regardless of season. Given that G (+)bacteria constitute the majority of the overall bacterialabundance in this desert grassland, seasonal declines in G(+) abundance may substantially affect seasonal ecosystemcontributions of soil bacterial activity in this desertgrassland system and influence subsequent nutrient avail-ability. Likewise, if soil moisture was lower during summerbecause of increased climate variability [1, 85], bacterialfunctionality would likely decline generating reductions innitrogen mineralization and plant growth. However, regard-less of climate patterns, G (+) bacteria are likely to remainthe dominant bacterial functional group in these soils,

supported by in situ observations by Kieft et al. [45] whoreported lower stress responses associated with select G (+)bacteria than G (–) ones when exposed to varying climaticscenarios within the upper soil horizons. Kieft et al. [45]also found high long-term survivorship for both bacterialfunctional groups in natural porous substrates, which suggeststhat porous soils in the sotol grassland may provide a long-term refugia for bacterial during extreme climate events.

Microbial functional group dynamics within the soilmicrobial community determine the extent to which soilmicrobes contribute to overall ecosystem functionality [24].The biotic and abiotic factors that limit the activities of soilmicrobial functional groups determine seasonal contribu-tions of soil microbial biomass to ecosystem processes. Ourresearch has demonstrated that seasonal climate shifts elicitsignificant responses from specific components of the soilmicrobiota in this desert ecosystem. Sinsabaugh et al. [86]suggested that when seasonal biotic or abiotic factorslimited individual functional contributors within the soilmicrobial community, other microbial contributors provid-ed these ecosystem processes by replacing those seasonallyrepressed microbial components. Therefore, in order topredict soil functional responses to climate change, it iscritical to identify the primary seasonal controls on thefunctional groups within the microbial biomass pool.

We determined the seasonal patterns of activity in soilmicrobial functional groups in this desert grassland. Forexample, in winter and summer, saprophytic fungi were themost abundant soil microbial component and demonstratedpositive correlations between community structure andfunction, although soil fungi generally exhibited highercarbon substrate activity in the winter. Regardless ofseason, microbial use of simple and complex carbohydratesexhibited the highest activity levels, followed by carboxylicacids and amino acids. These three substrates are the mainconstituents of plant roots and root exudates [87, 88], whichsuggests that plant carbon inputs strongly influencedsaprophytic fungal community functional capability. Thesefungal substrate activity patterns agree with fungal substrateutilization patterns recently observed in the Negev Desert[89], indicating that saprophytic fungi across differing aridsystems may have similar functional requirements. Moreso, this finding may reveal a direct link between above-groundand below-ground ecosystem functionality by elucidating howplant inputs of soil organic matter enable fungal structural andfunctional relationships to endure across seasons. Subsequent-ly, fungi perform soil functions that ultimately benefit plantproduction and higher trophic dynamics.

Saprophytic fungi may be the keystone contributors tosoil ecosystem function in this desert grassland because oftheir ability to maintain strong structure–function linksregardless of varying climate conditions. Soil fungi indesert environments have adapted life history strategies that

Linking Structure and Function to Seasonal Differences 839

allow them to avoid extreme thermal conditions that exist inhot arid systems, yet are capable of responding quickly tomoisture pulses [15, 89]. However, above- and below-ground interactions seem to strongly influence saprophyticfungal activity and ecosystem functioning in this desertgrassland system. This research suggests that soil bacterialfunctional groups, especially G (+) bacteria, may act assecondary microbial contributors to buffer ecosystemprocesses by providing soil functions similar to thoseprovided by saprophytic fungi, especially during thesummer when fungal activity was lower. The analogouscarbon substrate utilization patterns observed between fungiand bacteria in this study imply that both fungal andbacterial functional groups are strongly influenced by plantactivity and soil nitrogen availability. This suggests thatAM fungi may be the keystone microbe in this desertgrassland ecosystem because of their symbiotic capabilitiesto facilitate plant growth through increased nutrient uptake.If plant carbon inputs are the major control of soil fungaland bacterial functional patterns, then AM could indirectlyalter saprophytic fungal functionality by controlling plantgrowth.

Given that components of the microbial communitystructure become seasonally uncoupled from microbialfunctionality in the sotol grassland, similar patterns maybe observed in other arid ecosystems. We have observedthat the soil microbial community is generally self-correcting with different components assuming similarfunctional attributes in response to seasonal and yearlyenvironmental conditions. The negative effects of climatechange for this system would be to reduce the self-correction ability of the soil microbial community bysurpassing some inherent threshold to environmental stressthat currently exists. Global climate change models predictincreasing temperatures (by 3–5°C) over the next 50 years[1, 4]. Increased temperatures will result in higher evapo-transpiration rates, increased climate variability, and in-creased intra-seasonal storm magnitudes (possibly alteringprecipitation regimes by as much as 25%). Our researchreveals a complicated interaction of below- and above-ground functional dynamics which facilitate ecosystemprocesses in this desert grassland. Will the future climatepredictions for this region change temperature and moisturedynamics severely enough to cross a soil–microbialfunctional threshold? Any declines in primary productioncould also negatively impact soil microbial communityfunction and structural dynamics in this system, causing acascade of soil resource limitations, further limitingsubsequent plant production.

Acknowledgments This research was funded by the USGS GlobalClimate Change Small Watershed Project (J. Zak) and US NationalPark Service (J. Zak, D. Tissue). Special thanks are extended to Joe

Sirotnak along with the rest of the research staff at BBNP forcontinued support and cooperation. The authors would lastly like toextend thanks to Natasja Van Gestel and Heath Grizzle from TexasTech University for their continued support as friends and peersthroughout this research endeavor both in the field and in the lab.

References

1. Bernstein L, et al (2007) An assessment of the intergovernmentalpanel on climate change. Valencia, Spain, p 52

2. Stanley EH, Valett HM (1992) Interactions between drying andthe hyporheic zone of a desert stream, in global climate changeand freshwater ecosystems. Springer, New York, NY, p 211-233

3. NAST, N.A.S.T., US Global Change Research Program (2000)Climate change impacts on the united states: the potentialconsequences of climate variability and change. CambridgeUniversity Press, New York, NY

4. Johns TC et al (1996) The second Hadley Centre coupled ocean-atmosphere GCM: model description, spinup and validation. ClimDyn 13:103–134

5. Seager R et al (2007) Model projections of an imminent transitionto a more arid climate in southwestern North America. Science316:1181–1184

6. Easterling DR et al (2000) Climate extremes: observations,modeling, and impacts. Science 289:2068–2074

7. Noy-Meir I (1973) Desert ecosystems: environment and pro-ducers. Ann Rev Ecolog Syst 4:25–51

8. Rapport DJ, Whitford WG (1999) How ecosystems respond tostress. BioScience 49(3):193–203

9. Schwinning S, Ehleringer JR (2001)Water use trade-offs and optimaladaptations to pulse-driven arid ecosystems. J Ecol 89:464–480

10. Cable JM, Huxman TE (2004) Precipitation pulse size effects onSonoran Desert soil microbial crusts. Oecologia 141:317–324

11. Belnap J et al (2005) Linkages between microbial and hydrologicprocesses in arid and semiarid watersheds. Ecology 86(2):298–307

12. Mooney HA, Canadell J, Chapin FS, Ehleringer JR, and KornerC (2000) Ecosystem physiology responses to global change.Pages 141-189 in B. H. Walker, W. Steffen, J. Canadell, and J.Ingram, editors. The Terrestrial Biosphere and Global Change.Cambridge University Press, Cambridge, MA.

13. Huxman TE et al (2004) Convergence across biomes to a commonrain-use efficiency. Nature 429:651–654

14. Groffman PM, Bohlen PJ (1999) Soil and sediment biodiversity:cross-system comparisons and large scale effects. BioScience49:139–148

15. Zak JC, Sinsabaugh R, MacKay W (1995) Windows ofopportunity in desert ecosystems: their implications to fungalcommunity development. Can J Bot 73:S1407–S1414

16. Tilman D (1994) Competition and biodiversity in spatiallystructured habitats. Ecology 75(1):2–16

17. Tilman D et al (2001) Diversity and productivity in a long-termgrassland experiment. Science 294:843–845

18. Symstad AJ et al (2003) Long-term and large-scale perspectiveson the relationship between biodiversity and ecosystem function-ing. Bioscience 53:89–98

19. Zak DR et al (2003) Plant diversity, soil microbial communities,and ecosystem function: are there any links? Ecology 84(8):2042–2050

20. Belnap J, Phillips SL, Miller ME (2004) Response of desertbiological soil crusts to alterations in precipitation frequency.Oecologia 141:306–316

21. Vishnevetsky S, Steinberger Y (1997) Bacterial and fungaldynamics and their contribution to microbial biomass in desertsoil. J Arid Environ 37:83–90

840 C. W. Bell et al.

22. Collins SL et al (2008) Pulse dynamics and microbial processes inaridland ecosystems. J Ecol 96:413–420

23. Behan-Pelleteir VM, Newton G (1999) Linking soil biodiversityand ecosystem function—the taxonomic dilemma. BioScience 49(2):149–153

24. Groffman PM, Zaady E, Shachak M (2005) Microbial contrib-utors to biodiversity in deserts. In: Perevolotsky A (ed) Biodiver-sity in drylands. Oxford University Press, New York, NY, pp 109–121

25. Zak JC et al (1994) Functional diversity of microbial commu-nities: a quantitative approach. Soil Biol Biochem 26(9):1101–1108

26. Parker LW et al (1984) Carbon and nitrogen dynamics during thedecomposition of litter and roots of a Chihuahuan desert annual,Lepidium lasiocarpum. Ecol Monogr 54:339–360

27. Schlesinger WH, Fonteyn PJ, Marion GM (1987) Soil moisturecontent and plant transpiration in the Chihuahuan Desert of NewMexico. J Arid Environ 12:119–126

28. Peterjohn WT (1991) Denitrification: enzyme content and activityin desert soils. Soil Biol Biochem 23:845–855

29. Fliesbach A, Sarig S, Steinberger Y (1994) Effects of water pulsesand climatic conditions on microbial biomass kinetics andmicrobial activity in a Yermosol of the central Negev. Arid SoilRes Rehabil 8:353–362

30. Steinberger Y et al (1999) Phospholipid fatty acid profiles asindicators for microbial community structure in soils along aclimatic transect in the Judean Desert. Biol Fertil Soils 28:292–300

31. Zaady E, Groffman PM, Shachak M (1996) Release andconsumption of nitrogen from snail feces in Negev Desert soils.Biol Fertil Soils 23:399–405

32. Zaady E, Groffman PM, Shachak M (1996) Litter as a regulator ofnitrogen and carbon dynamics in macrophytic patches in NegevDesert soils. Soil Biol Biochem 28:39–46

33. Kennedy AC, Gewin VL (1997) Soil microbial diversity: presentand future. Soil Sci 162:607–617

34. Zak JC, Visser S (1996) An appraisal of soil fungal biodiversity:the crossroads between taxonomic and functional biodiversity.Biodivers Conserv 5:169–183

35. Sobek E, Zak JC (2003) The soil FungiLog procedure: methodand analytical approaches toward understanding fungal functionaldiversity. Mycologia 95:590–602

36. Konopka A, Oliver L, Turco RF (1998) The use of carbonsubstrate utilization patterns in environmental and ecologicalmicrobiology. Microb Ecol 35:103–115

37. Smalla K et al (1998) Analysis of Biolog GN substrate utilizationpatterns by microbial communities. Appl Environ Microbiol64:1220–1225

38. Bell CW et al (2008) Soil microbial responses to temporalvariations of moisture and temperature in a Chihuahuan Desertgrassland. Microb Ecol 56(1):153–167

39. Trasar-Cepeda C, Leiros MC, Gil-Sotres F (2000) Biochemicalproperties of acid soils under climax vegetation (Atlanticoakwood) in an area of the European temperate-humid zone(Galicia, NW Spain): specific parameters. Soil Biol Biochem32:747–755

40. Ndiaye EL et al (2000) Integrative biological indicators fordetecting change in soil quality. Am J Altern Agric 15:26–36

41. Klose S, Moore JM, Tabatabai MA (1999) Arylsulfatase activityof microbial biomass in soils as affected by cropping systems.Biol Fertil Soils 29:46–54

42. Moore JM, Klose S, Tabatabai MA (2000) Soil Microbial biomasscarbon and nitrogen as affected by cropping systems. Biol FertilSoils 31:200–210

43. Pavel R, Doyle J, Steinberger Y (2003) Seasonal patterns of celluloseconcentration in desert soil. Soil Biol Biochem 36:549–554

44. Moorhead DL, Sinsabaugh R (1999) Simulated patterns of littledecay predict patterns of extracellular enzyme activities. Appl SoilEcol 14:71–79

45. Kieft TL et al (1997) Survival and phospholipids fatty acidprofiles of surface and subsurface bacteria in nature sedimentmicrocosms. Appl Environ Microbiol 63:1531–1542

46. Fierer N, Schimel JP, Holden PA (2003) Variations in microbialcommunity composition through two soil depth profiles. Soil BiolBiochem 35:167–176

47. Zak JC et al (1997) Big Bend National Park watershed programmonitoring microbial activity and diversity along an elevationalgradient. In 9th Conference on Research and Resource Manage-ment in Parks and on Public Lands. Albuquerque, NM

48. Robertson TR et al (2008) Precipitation timing and magnitudedifferentially affect aboveground annual net primary productivityin three perennial species in a Chihuahuan Desert grassland. NewPhytol 181(1):230–242

49. Patrick L et al (2009) Physiological responses of two contrastingdesert plant species to precipitation variability are differentiallyregulated by soil moisture and nitrogen dynamics. Glob ChangBiol 15(5):1214–1229

50. Jarrell WM et al (1999) Soil water and temperature status. In:Sollins P (ed) Standard soil methods for long-term ecologicalresearch. Oxford University Press, New York, NY, pp 55–73

51. Onset CC (2004) Hobo-H8 pro-series user’s manual. OnsetComputer Corporation, Bourne, MA

52. Robertson GP et al (1999) Soil carbon and nitrogen availability:nitrogen mineralization, nitrification, and soil respiration poten-tials. In: Sollins P (ed) Standard soil methods for long-termecological research. Oxford University Press, New York, NY,pp 258–271

53. Sollins P et al (1999) Soil carbon and nitrogen. In: Sollins P (ed)Standard soil methods for long-term ecological research. OxfordUniversity Press, New York, pp 89–105

54. Miller RH, Keeney DR (1982) Methods of soil analysis, 2nd edn.Academic, Madison, WI

55. Vance ED, Brookes PC, Jenkinson DS (1987) An extractionmethod for measuring soil microbial biomass C. Soil BiolBiochem 19(6):703–707

56. Nunan N, Morgan MA, Herlihy M (1997) Ultraviolet absorbance(280 nm) of compounds released from soil during chloroformfumigation as an estimate of the microbial biomass. Soil BiolBiochem 30(12):1599–1603

57. Garland JL, Millis AL (1991) Classification and characterizationof heterotrophic microbial communities on the basis of patterns ofcommunity-level sole-carbon-source utilization. Appl EnvironMicrobiol 57:2351–2359

58. Dobranic JK, Zak JC (1999) A microtiter plate procedure forevaluating fungal functional diversity. Mycologia 91:756–765

59. Parham JA, Deng SP (2000) Detection, quantification andcharacterization of B-glucosaminidase activity in soil. Soil BiolBiochem 32:1183–1190

60. Tabatabai MA (1994) Soil enzymes. In: Keeney DR (ed) Methodsof soil analysis. American Society of Agronomy, Inc., Madison,WI, pp 903–947

61. Deacon J (2006) Fungal nutrition. In: Deacon J (ed) Fungalbiology, 4th edn. Blackwell, Oxford, UK

62. Ekenler M, Tabatabai MA (2002) β-Glucosaminidase activity ofsoils: effect of cropping systems and its relationship to nitrogenmineralization. Biol Fertil Soils 35:1081–1094

63. Acosta-Martinez V, Zobek TM, Gill TE (2003) Enzyme activitiesand microbial community structure in semiarid agricultural soils.Biol Fertil Soils 38:216–227

64. Sasser M (2001) Identification of bacteria by gas chromatographyof cellular fatty acids. MIDI, Inc. Automated Microbial Identifi-cation Solutions, Newark, DE

Linking Structure and Function to Seasonal Differences 841

65. Zak DR et al (1996) Soil microbial communities beneath Populusgrandidentata grown under elevated atmospheric CO2. Ecol Appl6:257–294

66. Zogg GP et al (1997) Compositional and functional shifts inmicrobial communities due to soil warming. Soil Sci Soc Am61:475–481

67. Ringelberg DB et al (1997) Consequences of rising atmosphericcarbon dioxide levels for the belowground microbiota associatedwith white oak. J Environ Qual 26:495–503

68. Pinkart HC et al (2002) Biochemical approaches to biomassmeasurements and community structure analysis. In: Hurst CJ (ed)Manual of environmental microbiology. ASM, Washington, DC,pp 101–113

69. Zelles L (1997) Phospholipid fatty acid profiles in selectedmembers of soil microbial communities. Chemosphere 35:275–294

70. Madan R et al (2002) Use of fatty acids for identification of AMfungi and estimation of the biomass of AM spores in soil. SoilBiol Biochem 34:125–128

71. Olsson PA (1999) Signature fatty acids provide tools fordetermination of the distribution and interactions of mycorrhizalfungi in soil. FEMS Microbiol Ecol 29:303–310

72. Ruess L et al (2002) Fatty acids of fungi and nematodes-possiblebiomarkers in the soil food chain? Soil Biol Biochem 34:745–756

73. Rateledge C, Wilkinson SG (1998) Microbial lipids. Academic,London, UK

74. Bardgett RD, Hobbs PJ, Frostegard A (1996) Changes in soilfungal:bacterial biomass ratios following reductions in theintensity of management of an upland grassland. Biol Fertil Soils22:261–264

75. Frostegard A, Baath E (1996) The use of phospholipid fatty acidanalysis to estimate bacterial and fungal biomass in soil. BiolFertil Soils 22:59–65

76. Koppenstedt RM. The prokaryotes. Springer, New York, NY,p 1139–1156

77. White DC, Stair JO, Ringelberg DB (1996) Quantitative compar-isons or in situ microbial biodiversity by signature biomarkeranalysis. J Ind Microbiol 17:185–196

78. Leps J, Smilauer P (2003) Mulitvariate analysis of ecological datausing CANOCO. University of South Bohemia, Ceske Budejo-vice, Czech Republic

79. Vandenkoornhuyse P et al (2002) Arbuscular mycorrhizal com-munity composition associated with two plant species in agrassland ecosystem. Mol Ecol 11:1555–1564

80. Treseder KK, Cross A (2006) Global distributions of arbuscularmycorrhizal fungi. Ecosystems 9:305–316

81. Deacon J (2006) Fungal symbiosis. In: Deacon J (ed) Fungalbiology, 4th edn. Blackwell, Oxford, UK, pp 279–308

82. Rillig MC (2004) Arbuscular mycorrhizae and terrestrial ecosys-tem processes. Ecol Lett 7:740–754

83. Freise CF, Allen MF (1991) The spread of VA mycorrhizal fungalhyphae in the soil: inoculum types and external hyphal architec-ture. Mycologia 83:409–418

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

85. Brown AR, Webster S (2004) Orographic flow-blocking schemecharacteristics. Q J R Meteorol Soc 130:3015–3028

86. Sinsabaugh R et al (1999) Characterizing soil microbial communities.In: Sollins P (ed) Standard soil methods for long term ecologicalresearch. Oxford University Press, New York, NY

87. Jones DL (1998) Organic acids in the rhizosphere—a criticalreview. Plant Soil 205:25–44

88. Bolton H Jr, Fredrickson JK (1993) Microbial ecology in therhizosphere. In: Metting FB Jr (ed) Soil microbial ecology. NY,Marcel Decker, New York

89. Oren A, Steinberger Y (2008) Catabolic profiles of soil fungalcommunities along a geographic climate gradient in Israel. SoilBiol Biochem 40:2578–2587

842 C. W. Bell et al.