Published Ahead of Print 25 April 2014 101128AEM03939-13

2014 80(13)4021 DOIAppl Environ Microbiol Snape and Belinda C FerrariJosie van Dorst Steven D Siciliano Tristrom Winsley Ian Diesel Fuel Toxicity in Subantarctic SoilsBacterial Targets as Potential Indicators of

httpaemasmorgcontent80134021Updated information and services can be found at

These include

SUPPLEMENTAL MATERIAL Supplemental material

REFERENCEShttpaemasmorgcontent80134021ref-list-1at

This article cites 58 articles 15 of which can be accessed free

CONTENT ALERTS moreraquoarticles cite this article)

Receive RSS Feeds eTOCs free email alerts (when new

httpjournalsasmorgsitemiscreprintsxhtmlInformation about commercial reprint orders httpjournalsasmorgsitesubscriptionsTo subscribe to to another ASM Journal go to

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

Bacterial Targets as Potential Indicators of Diesel Fuel Toxicity inSubantarctic Soils

Josie van Dorsta Steven D Sicilianob Tristrom Winsleyabc Ian Snapec Belinda C Ferraria

School of Biotechnology and Biomolecular Sciences UNSW Australia Randwick New South Wales Australiaa Department of Soil Science University of SaskatchewanSaskatoon Saskatchewan Canadab Terrestrial and Nearshore Ecosystems Program Australian Government Antarctic Division Kingston Tasmania Australiac

Appropriate remediation targets or universal guidelines for polar regions do not currently exist and a comprehensive under-standing of the effects of diesel fuel on the natural microbial populations in polar and subpolar soils is lacking Our aim was toinvestigate the response of the bacterial community to diesel fuel and to evaluate if these responses have the potential to be usedas indicators of soil toxicity thresholds We set up short- and long-exposure tests across a soil organic carbon gradient Utilizingbroad and targeted community indices as well as functional genes involved in the nitrogen cycle we investigated the bacterialcommunity structure and its potential functioning in response to special Antarctic blend (SAB) diesel fuel We found the pri-mary effect of diesel fuel toxicity was a reduction in species richness evenness and phylogenetic diversity with the resultingcommunity heavily dominated by a few species principally Pseudomonas The decline in richness and phylogenetic diversity waslinked to disruption of the nitrogen cycle with species and functional genes involved in nitrification significantly reduced Of the11 targets we evaluated we found the bacterial amoA gene indicative of potential ammonium oxidation the most suitable indi-cator of toxicity Dose-response modeling for this target generated an average effective concentration responsible for 20change (EC20) of 155 mg kg1 which is consistent with previous Macquarie Island ecotoxicology assays Unlike traditional sin-gle-species tolerance testing bacterial targets allowed us to simultaneously evaluate more than 1700 species from 39 phyla in-clusive of rare sensitive and functionally relevant portions of the community

Petroleum hydrocarbon contamination in polar regions occurscommonly as a result of resource extraction human habita-

tion and subsequent activities (1) Toxicity information relatingto the effects of petroleum hydrocarbon contamination on terres-trial ecosystems in polar and subpolar regions is limited (2) yetavailable evidence suggests that the effects of oil spills are moredamaging than in temperate regions due to low temperatures andslower ecosystem recovery (1) Currently available remediationtrigger concentrations in polar and subpolar soils are largely basedupon countriesrsquo domestic guidelines for diesel fuel The subse-quent values are highly variable ranging between 100 mg kg1

and 2000 mg kg1 with no evidence-based guidance for site-specific modifications (3) It has been recommended that aweight-of-evidence approach integrating site-specific ecotoxico-logical chemical and ecological assessments is needed to developuniversally accepted guidelines for petroleum hydrocarbon con-tamination in cold regions (4)

Single-species tolerance testing used commonly for ecotoxico-logical assessments is resource- and time-intensive The confi-dence surrounding five to 10 model organisms accurately repre-senting the sensitivity of an ecosystem is also questionable (5 6)Further the selection criterion of model organisms has resulted ina bias toward temperate and Northern Hemisphere species to theexclusion of rare and often sensitive species (6) In the Antarcticand subantarctic regions traditional toxicology indicator speciessuch as earthworms and large invertebrates are sparse or nonex-istent further compounding the uncertainty surrounding the ap-plication of traditional tolerance testing ldquomodelrdquo organisms In2008 Hickey et al suggested that new rapid tolerance testingapproaches that target a suite of organisms across many taxa wererequired to increase the confidence in the resulting toxicity esti-mates Rapid tolerance testing has been broadly defined as toxicitytesting that aims to reduce the time and resources per species by

reducing replication the number of treatments and testing spe-cies concurrently Rather than small numbers of precise estimatesrapid tests are expected to promote greater numbers of speciesassessed (more approximately) with increased relevance to theecosystem thereby delivering more accurate estimates of commu-nity sensitivity overall

Microorganisms are important to soil health due to their inte-gral role in biogeochemical cycles and ecosystem sustainability(7ndash9) With developments in sequencing technologies the phylo-genetic and functional understanding of environmental microbialcommunities is rapidly expanding and in turn promoting the de-velopment of molecular microbial indicators as integrated mea-sures of ecosystem health (10 11) To date the majority of studiesassessing the impact of petroleum hydrocarbons on polar soil mi-crobial populations have focused on the stimulated portion of thebacterial community with the aim of establishing the natural at-tenuation capacity of a soil or the effects of active bioremediationThese studies have identified a loss of diversity combined with aselection toward heterotrophic species with known petroleum hy-drocarbon-degrading potential (12ndash15) By comparison little in-formation exists on the species sensitive to petroleum hydrocar-bon toxicity Of those studies available assessments are confined

Received 27 November 2013 Accepted 17 April 2014

Published ahead of print 25 April 2014

Editor J E Kostka

Address correspondence to Belinda C Ferrari bferrariunsweduau

Supplemental material for this article may be found at httpdxdoiorg101128AEM03939-13

Copyright copy 2014 American Society for Microbiology All Rights Reserved

doi101128AEM03939-13

July 2014 Volume 80 Number 13 Applied and Environmental Microbiology p 4021ndash 4033 aemasmorg 4021

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

to monitoring predetermined microbial functional groups non-inclusive of species with unknown sensitivities or potential eco-system functional services (16 17)

A key challenge for developing microbial indicators is the highvariability of microbial communities observed naturally acrosstemporal and spatial scales Many studies have linked this micro-bial variation to changes in the physical and chemical properties ofsoil with factors such as carbon content and pH particularly im-portant (18 19) For the subantarctic region bacterial and alkane-degrading diversity has already been correlated with organic car-bon and total nitrogen availability (20) However it is unclear howthe variability in soil types and resulting microbial communitycomposition will affect the resilience of soil bacterial communitieswhen exposed to petroleum hydrocarbons Another key challengefor implementing microbial monitoring specifically for petro-leum hydrocarbons is the concurrent stimulation and inhibitioneffect of petroleum hydrocarbons on bacteria which complicatestoxicity assessments To counter this key nutrient-cycling param-eters can be targeted that are sensitive to petroleum hydrocarbontoxicity but not stimulated by increases in available carbon (9)For example ammonia oxidation as a chemolithic process is notaffected by increases in organic carbon yet it is inhibited by hy-drocarbons through the general mechanism of biochemical toxic-ity (nonpolar narcosis) as well as specifically inhibiting ammo-nium monooxygenase (21 22) As bacteria and archaea areprimarily responsible for the conversion of ammonia to nitritetoxicity or inhibition suggests overall ecosystem impairmentSimilarly functional genes including nifH and nosZ that are re-sponsible for key enzymatic steps within the nitrogen cycle havebeen targeted to assess functional aspects of soil microbial com-munities (23 24)

Macquarie Island a world heritage-listed Australian subant-arctic territory is located approximately 1500 km southeast ofTasmania (54deg37=53==S 158deg52=15==E) The island has been thelocation of a permanently occupied research station since 1949Historical and recent activity at the research station has led tothree contaminated sites exhibiting medium to high levels of pe-troleum hydrocarbons (25 26) All three sites are contaminatedwith hydrocarbons in the diesel fuel range (C9 to C28) withheavier hydrocarbons also present originating from relic oildumping and burn pits in the C29 to C36 range However themajority of the contamination originates from the widespread useof special Antarctic blend (SAB) diesel fuel (C9 to C18) a lighteraromatic diesel fuel suitable for use in cold regions Following asite-specific bioassessment of the site bioremediation combiningnutrient addition and aeration in situ began in 2009 (25ndash27) Asyet no remediation trigger values or site-specific remediation tar-get concentrations exist for this site

Our aim was to identify the microbial indicators of fresh dieselfuel toxicity in Macquarie Island soils The effects of SAB on thebacterial community were evaluated with broad and targetedcommunity indices as well as the abundances of functional genesencoding key enzymes within the nitrogen cycle We utilized theresults in dose-response curves to further test the utility of micro-bial targets as toxicity indicators We hypothesized that the bacte-rial community would change with increasing fuel contamina-tion with a loss of diversity and a selection toward species capableof hydrocarbon degradation We further hypothesized that thetargeted portions of the bacterial community would provide moresensitive indicators than communitywide diversity estimates

MATERIALS AND METHODSSampling sites and diesel fuel spiking Macquarie Islandrsquos climate isheavily influenced by its oceanic position Average air temperatures rangefrom 33degC in the winter to 70degC in the summer Precipitation eventsoccur approximately 312 days per year and strong northwest to westerlywinds often gale force (556 km h1) predominate (Australian Bureauof Meteorology) For this experiment four uncontaminated bulk soilplots consisting of sandy to peaty soil were selected from the isthmusapproximately 100 to 150 m from the major contaminated site(54deg37=53==S 158deg52=15==E) Approximately 500 g of soil was collectedfrom each plot (a total of 2 kg) to a depth of 30 cm to target the most activemicrobial zone and to correspond with related invertebrate studies (28)Each of the four bulk soils were homogenized by mixing separated into 10individual subsamples of 50 g and placed in 100-ml amber jars Onecontrol sample for each of the four bulk soils was left unamended Theremaining soils were spiked with SAB diesel fuel to three target concen-tration ranges low (0 to 400 mg kg1 n 3) medium (401 to 5000 mgkg1 n 3) and high (5000 to 20000 mg kg1 n 3) (Table 1) Bulksoils were chosen to reduce the variability of soil properties within plotsand to have control over the range of concentrations All soils were incu-bated aerobically in the dark Three of the bulk soils (P1 to P3) wereincubated for 21 days (short exposure) while the remaining bulk soil wasincubated for 18 months (long exposure P4) While no comment can bemade on the short-term response of P4 the extended exposure was set upto determine if microbial community responses to the diesel fuel wereobserved between both short- and long-term samples (29) For the ex-tended exposure the lids of the amber jars were opened every 4 weeks tomaintain aerobic conditions Although not highly aerobic these condi-tions will provide aerobic and anaerobic pockets within the jars consis-tent with the soil matrix in situ

Nutrients and soil parameters were analyzed after the incubation pe-riod according to the Australasian Standard Soil protocols (30) (Table 1)Briefly total carbon was determined by the loss-on-ignition (LOI)method and expressed as a percentage The concentrations of anions andcations in the soil were measured on a water extract (1 g5 ml water) andexpressed as mg kg1 Conductivity and pH were also measured on thesame water extract A 10-g subsample of soil from each sample was ex-tracted with hexane and assessed by gas chromatography to determine thetotal petroleum hydrocarbon (TPH) concentration (9) The detectionlimit (DL) for TPH concentrations was 20 mg kg1 TPH concentrationsbelow the detection limit were estimated with a substitution methodbased on half the detection limit ie estimated TPH concentrations of 10mg kg1 (Table 1) Average measured TPH concentrations were calcu-lated for each of the fuel categories within the four plots (Fig 1A Table 1)Dry matter fraction was determined gravimetrically using the same sam-ples analyzed for TPH We have reported all nutrient and TPH concen-trations on a dry matter basis unless stated otherwise

Barcoded amplicon pyrosequencing targeting the 16S rRNA geneAfter the short (21 days) and long (18 months) incubation periods totalsoil genomic DNA was extracted from 50-mg subsamples of each soilsample in triplicate with the FastDNA SPIN kit for soil (MP BiomedicalsNSW Australia) In total 120 DNA extracts (40 samples in triplicate)were titrated to a standard working concentration range of 5 to 10 ngl1 The technical replicates were analyzed with automated ribosomalintergenic spacer analysis (ARISA) to evaluate interintrasample similar-ity according to van Dorst et al (31) Intersample replication was evalu-ated with analysis of similarity (ANOSIM) in PRIMER v6 After inter-sample similarity was confirmed (see Fig S1 and Table S1 in thesupplemental material) the extracted genomic DNA from a randomlyselected replicate was used as a template for barcoded tag pyrosequencingThe V1 V2 and V3 hypervariable regions of the small-subunit (SSU)ribosomal gene were targeted using the 27F and 519R universal primers(32) The PCR amplification and barcoded sequencing were performed atthe Research and Testing Laboratory (Lubbock TX USA) Although sys-tematic errors remain with barcoding sequencing technologies stochastic

van Dorst et al

4022 aemasmorg Applied and Environmental Microbiology

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

replication (3) was performed to limit some of the PCR bias Sampleswere run on the Roche 454 FLX Titanium platform Data were providedfrom the sequencing facility in the form of standard flowgram format (sff)files (33)

Processing amplicon pyrosequencing data The sequence data wereprocessed using the MOTHUR software package (34) This involved ex-traction of the sequence information and flowgrams from the sff filesdemultiplexing and removal of short reads (200 bp) truncation of longreads (500 bp) and sequences with homopolymers (8-bp repeats)and ambiguous bases Remaining reads were screened for quality by tak-ing an average length phred read score of 25 Reads were aligned againstthe SILVA 16S rRNA database file provided with the MOTHUR package(35) Chimera checking with the UCHIME algorithm (36) allowed re-moval of any chimeric artifacts and sequences were preclustered at 1 tonegate instrument error rate The reads were then clustered into opera-tional taxonomic units (OTUs) based on 96 sequence similarity forspecies-level OTUs (37) Taxonomic assignment was determined againstthe Greengenes database provided within the MOTHUR package (38) AnOTU abundance-by-sample matrix was generated from the resulting out-put and various diversity indices were calculated A neighbor-joining treewas created through MOTHUR (34) with the Clearcut program addition(39) The neighbor-joining tree was used to run the weighted UniFracalgorithm (40)

Data analysis Multivariate data analysis was conducted with the soft-ware packages PRIMER v6 and Permanova (41) To account for thelog-normal distribution of the data the abundance-by-sample matrixwas square root transformed After transformation the skewness andkurtosis was reduced closer to 0 For sample comparison the abun-dance-by-sample matrix was then standardized to express the OTUabundances as relative abundances Resemblance matrices were calcu-lated with both the Bray-Curtis similarity coefficient and the weightedUniFrac measurements To test the null hypothesis of no differencesbetween plots ANOSIM was performed on the resemblance matriceswith 999 permutations To test the null hypothesis of no differencesbetween diesel fuel categories a one-way analysis of variance(ANOVA) was used to test for significant differences in the microbialcommunities between fuel categories within each plot

Results from the chemical and physical analysis of the soils were usedto evaluate the influence of environmental variables The results werenormalized and a similarity matrix was calculated with the Euclideandistance metric The relative relationship of the environmental variablesto the biological distribution was analyzed using a distance-based linearmodel (DistLM) The selection criterion for the model was adjusted R2The selection procedure was stepwise with 999 permutations The Dis-tLM was based on the null hypothesis of no relationship between theenvironmental variables and the biological resemblance matrix All statis-tical tests were considered significant at a P value of 005

Individual OTUs were evaluated across the SAB spiking range to de-termine if they were significantly inhibited or stimulated with increasingTPH concentration After OTUs present in only one sample were re-moved the log-transformed abundance of individual OTUs was plottedagainst the log-transformed TPH data The resulting dose responses ofeach OTU were then fitted to a linear equation in R (httpwwwR-projectorg) The number of OTUs significantly stimulated or inhibited withTPH was determined based on the quality of fit (P 005) and slope of theline The percentage abundance of each OTU was aggregated into generato create a heatmap within R Representative OTUs that were unable to bereliably classified to the genus level were listed with the closest classifica-tion level The OTUs were then aggregated into phylum-level phylogenyAll phyla were analyzed for positive or negative correlations to increasingfuel concentrations The AcidobacteriaProteobacteria ratio consideredrepresentative of the oligotrophscopitrophs ratio in the environmentwas calculated and used as an additional community index

Targeting the nitrogen cycle with quantitative PCR QuantitativePCR (qPCR) was used to measure the abundance of the nifH amoA andnosZ genes present The same DNA extracts used in the barcoded pyrose-quencing analysis were utilized for the qPCR analysis and each of theselected genomic DNA extracts was analyzed by qPCR in triplicate Sam-ples were analyzed using the QuantiTect Fast SYBR green PCR master mixreal-time PCR kit (Qiagen Doncaster VIC Australia) and run on the ABI7500 real-time PCR machine (Applied Biosystems) Each 20-l reactionmixture contained 125 l of master mix 125 l of template DNA and 1M the forward and reverse primers The thermal cycling program con-sisted of 94degC for 5 min 45 cycles of 94degC for 20 s 54degC for 50 s followed

TABLE 1 Summary of measured physicochemical parameters

Soil property

Concn (mg kg1)a

P1 (low C) P2 (medium C) P3 (high C) P4 (long exposure)

ControlLow(n 3)d

Medium(n 3)

High(n 3) Control

Low(n 3)

Medium(n 3)

High(n 3) Control

Low(n 3)

Medium(n 3)d

High(n 3) Control

Low(n 3)

Medium(n 3)

High(n 3)d

TPH (avg) DL 157 1146 10329 DL 41 1185 10894 DL 206 1442 17845 DL 47 279 11447SD 84 815 6583 54 1055 6582 173 1117 12407 18 39 6449

Carbonb 5 5 5 5 8 8 8 8 36 36 36 36 67 67 67 67Dry matter

fraction08 083 085 083 075 077 070 070 033 030 030 030 042 043 050 030

SD 003 005 003 003 006 000 000 000 000 003 000 000Conductivityc 102 103 95 101 271 290 308 275 670 633 631 633 776 1333 915 980

SD 9 3 2 4 6 27 8 1 3 286 404 343pH 57 57 57 56 66 66 65 67 53 53 54 52 48 47 52 51

SD 00 01 00 00 00 00 00 01 00 00 02 02Nitrite 015 07 06 11 53 43 44 22 969 1047 634 433 015 015 015 015

SD 06 04 05 02 07 11 81 41 17 0 0 0Nitrate 65 66 51 75 302 353 380 254 1415 1457 1329 1403 20 21 20 8

SD 5 7 3 8 27 50 34 16 22 11 4 6Ammonium 31 11 08 14 29 39 53 49 528 561 518 329 78 81 48 56

SD 03 02 01 03 04 08 17 35 28 07 10 18Phosphate 125 105 98 83 327 316 349 348 3798 4169 4426 4270 90 92 102 77

SD 07 08 04 06 13 05 189 25 85 27 14 07Sulfate 308 340 314 342 651 713 771 698 844 836 807 776 70 89 63 80

SD 07 18 05 08 17 64 12 13 08 44 31 19

a Controls are unamended Low low TPH concentration (range of detection limit [DL] to 400 mg kg1) medium medium TPH concentration (range of 401 to 5000 mg kg1)high high TPH concentration (5000 mg kg1) All values are in mg kg1 unless otherwise statedb Values are LOI (loss on ignition)c Values are in Scmd One sample in this group had substantially less sequencing reads than the other samples as such they were excluded from further phylogenetic analysis

Microbial Indicators of Toxicity in Subantarctic Soil

July 2014 Volume 80 Number 13 aemasmorg 4023

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

by a melt-curve step from 50degC to 95degC The quantitative fluorescencedata were collected during the 54degC step To test for the potential presenceof PCR inhibitors a four- to five-point curve (in duplicate) of differentDNA concentrations for each soil sample was analyzed according to Ma etal (42) At the DNA concentrations used no inhibition was detectedQuantification data were collected only when there was no detected PCRinhibition no amplification in the ldquono-template controlrdquo a single peak inthe melt curve consistent with specific amplification and a reaction effi-ciency of 100 10

Subunits of the nitrogenase enzyme are encoded by the genes nifHnifD and nifK The nifH gene is the most widely sequenced and utilizedmarker for nitrogen fixation Here we used the primers IGK3 and DVV(43) to target the nifH gene as a proxy for potential nitrogen fixationactivity (Table 2) A standard curve was generated with the IGK3DVV-amplified PCR product from a control subantarctic soil The number ofcopies was determined spectrophotometrically The standard curves werelinear over six orders of magnitude with amplification efficiencies of 938to 951 and an R2 value of 0997

Nitrification refers to the oxidation of ammonia into nitrite and thesubsequent oxidation of nitrite into nitrate As PCR primers targetingthe entire nitrite oxidation functional group are not currently available

the ammonium oxidizing step was used as a proxy for potential nitrifyingactivity The ammonium oxidizing step can be performed by ammoniumoxidizing bacteria (AOB) and ammonium oxidizing archaea (AOA) Weevaluated both groups with primer sets targeting the AOB (amoA1FamoA2R) (44) and AOA (ArchamoAFArchamoAR) (45) amoA genes(Table 2) Standard curves were generated with the amplified PCR prod-ucts from a control subantarctic soil with the number of copies deter-mined spectrophotometrically For bacterial amoA the standard curvewas linear over seven orders of magnitude with amplification efficienciesof 916 to 920 R2 0998 The archaeal amoA standard curve was linearover six orders of magnitude with amplification efficiencies of 906 to948 (R2 0974)

Classified genes known to be present in denitrifying microorganismsinclude nir (nitrate reductase) and nos (nitrous oxide reductase) Wechose to target the nosZ gene as a proxy for potential denitrification withthe primer set nosZ2-FnosZ2-R (46) These primers target primarily Pro-teobacteria excluding denitrifiers within Firmicutes Although not presentin all denitrifying species the nosZ gene has been widely evaluated andapplied in polar soils due to its relevance in nitrous oxide production (apotent greenhouse gas) Further in Siciliano et al the applicability ofquantifying nosZ abundance from mixed templates was specifically tested

FIG 1 The average measured total petroleum hydrocarbon (TPH) concentration and bacterial community similarity of samples across SAB spiking fuelconcentration ranges and soil plots (A) The average measured TPH log concentration for the spiking fuel ranges within each soil plot The dashed line indicatesthe TPH detection limit (DL) (B) The average community similarity within each soil plot based on the weighted UniFrac distance Error bars account for thestandard deviation between biological replicates significant decline in community similarity was found (P 005)

van Dorst et al

4024 aemasmorg Applied and Environmental Microbiology

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

with positive results (47) In another study nosZ was more sensitive toenvironmental changes than nirS and nirK (48) To determine the copynumbers of the nosZ gene a standard curve was generated using thenosZ2-FnosZ2-R (46)-amplified PCR product from Pseudomonasstutzeri a well-known denitrifier species The number of copies of thenosZ gene was determined spectrophotometrically (Table 2) The stan-dard curve was linear over five orders of magnitude the amplificationefficiency determined from the slope of the standard curve was 998(R2 096)

Dose-response modeling For dose-response modeling a large num-ber of data points are recommended over high replication to best capturepossible shifts in the curve (49 50) Hence we used individual samplesand their TPH concentrations instead of the concentration ranges to max-imize the confidence surrounding the dose-response curves The dose-response measurements were calculated as a percentage of the control toenable comparison between soil types (P1 to P4) The dose responsecurves were generated within the drc R package (51) This included plot-ting the data and selecting the most suitable model as evidenced by Akai-kersquos information criterion (AIC) and ANOVA comparisons of modelsEffective concentration values (ECx) (including standard error and con-fidence intervals) were then calculated from the curve generated by themost suitable model for each data set The resulting effective concentra-tion responsible for 20 change (EC20) values calculated from the dose-response curves were used to compare the relative sensitivities of soil typesand community measures

RESULTSBacterial community composition A total of 127053 quality-checked sequence reads were obtained averaging 3304 (1626)sequence reads per sample For further comparative analysis thesequences were subsampled at 1450 sequences resulting in ap-proximately 14000 per plot and 4350 sequences per SAB treat-ment Three samples (within P1-low P3-med and P4-high) hadexceptionally low numbers of sequence reads at 800 and wereexcluded from further analysis The distribution of the bacterialcommunities was used to create two similarity matrices based onthe Bray-Curtis correlation coefficient and UniFrac distance Uti-lizing the weighted UniFrac similarity matrix a one-way ANOVAconfirmed significant differences between bacterial communitiesaccording to their TPH concentration ranges control low (10 to400 mg kg1) medium (401 to 5000 mg kg1) and high (5000mg kg1) Significant dissimilarity between the control and thetreatments was observed in all four soil plots (Fig 1B Table 3)The soil plots P1 to P4 were also found to be significantly differ-ent in a one-way ANOSIM global R value 073 (P 0001)(Table 3)

Diesel fuel substantially reduced the similarity of communities

TABLE 2 Details of primers used in qPCR to target functional genes within the nitrogen cycle

Process Target gene Primers Reference Primer sequence Positive control

Nitrogen fixation nifH IGK3 43 GCIWTHTAYGGIAARGGIATHGGIA gDNA from soilDVV ATIGCRAAICCICCRCAIACIACRTC

Ammonium oxidation BamoAa amoA1F 44 GGGGTTTCTACTGGTGGT gDNA from soilamoA2R CCCCTCKGSAAAGCCTTCTTC

AamoAb AamoAF 45 STAATGGTCTGGCTTAGACG gDNA from soilAamoAR GCGGCCATCCATCTGTATGT

Denitrification nosZ nosZ2F 46 CGCRACGGCAASAAGGTSMSSGT Pseudomonas stutzerinosZ2R CAKRTGCAKSGCRTGGCAGAA

a Bacterial amoAb Archaeal amoA

TABLE 3 ANOVA and ANOSIM results testing the ampliconpyrosequencing community UniFrac and Bray-Curtis dissimilaritiesbetween fuel categories and soil plots

Differences and testb

Pairwise testresult tR statistic P valuec

P1 differences between fuelcategories (1-wayANOVA)

Global 174 0006Control high 80 0009Control med 55 0026Control low 42 0032Low high 55 0027Low med 19 0241Med high 38 0044

P2 differences between fuelcategories (1-wayANOVA)

Global 579 0001Control high 128 lt0001Control med 97 0001Control low 47 0016Low high 114 lt0001Low med 71 0003Med high 44 0021

P3 differences between fuelcategories (1-wayANOVA)

Global 2136 lt0001Control high 28 lt0001Control med 194 lt0001Control low 156 lt0001Low high 174 lt0001Low med 62 0007Med high 94 0001

P4 differences between fuelcategories (1-wayANOVA)

Global 2785 lt0001Control high 286 lt0001Control med 131 lt0001Control low 83 lt0001Low high 28 lt0001Low med 66 0005Med high 236 lt0001

Differences between plotsa

(1-way ANOSIM)Global 0729 lt0001P1 P2 0722 lt0001P1 P3 0807 lt0001P1 P4 0765 lt0001P2 P3 0781 lt0001P2 P4 0781 lt0001P3 P4 0767 0002

a Plots include P1 (low-carbon soil) P2 (medium-carbon soil) P3 (high-carbon soil)and P4 (aged soil)b Fuel categories refer to TPH concentrations Unamended control low TPHconcentration (range of DL to 400 mg kg1) medium TPH concentration (range of401 to 5000 mg kg1) and high TPH concentration (5000 mg kg1)c Significant differences are in bold P values were considered significant at 005

Microbial Indicators of Toxicity in Subantarctic Soil

July 2014 Volume 80 Number 13 aemasmorg 4025

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

from the controls largely by stimulating or reducing the relativeabundances of key lineages rather than removing entire lineages ofbacteria (Fig 2) The communities shifted from large numbers ofspecies in relatively low abundance to communities heavily dom-inated by only a few species We found that the OTUs consisting of

26 to 79 of the relative abundance were significantly inhibitedacross the SAB spiking range while only a small proportion ofOTUs 04 to 78 were stimulated (Table 4) The genus moststimulated was Pseudomonas contributing a maximum of 55relative abundance in the control soils compared to 60 relative

FIG 2 Relative abundances of genera present in soil samples from low to high special Antarctic blend (SAB) diesel fuel concentration Samples are sortedaccording to SAB fuel concentration within each soil plot and are labeled according to fuel category (control low medium and high) The log SAB fuelconcentration is on a scale of 0 to 45 with 45 31000 mg kg1 Representative operational taxonomic units (OTUs) that were unable to be reliably classifiedto the genus level were listed with the closest classification level

van Dorst et al

4026 aemasmorg Applied and Environmental Microbiology

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

abundance in the highest-concentration samples for the low (P1)-and medium (P2)-carbon soils and 20 in the long-exposurespiked soils (P4) By comparison the high-organic-carbon soil(P3) exhibited a lower relative abundance of Pseudomonas (1)and maintained a relative abundance between 006 and 4across the entire spiking range Instead the genus Parvibaculumwhich contributed 02 of the total abundance in the controlsoils increased substantially contributing to 43 of the final rel-ative abundance (Fig 2) Bacteria involved in the nitrification pro-cess specifically ammonium oxidation or the oxidation of nitriteto nitrate were significantly inhibited following the addition ofdiesel fuel A functional group for the nitrification species ldquoAllnitrification speciesrdquo was consolidated consisting of Nitrosospiraspecies Nitrosococcus species Nitrobacter species Nitrosomonasspecies Nitrospira species and unclassified species from the fam-ily Nitrosomonadaceae and the order Nitrosomonadales The groupwas significantly inhibited across the diesel fuel concentrationrange for each soil plot (P 005) (Fig 2)

Environmental predictors Carbon was a significant predictorvariable (F 513 P 0001) when analyzed with a marginalDistLM model (Table 5) However in sequential DistLM testsorganic carbon contributed only 41 to the total biological vari-ation observed between soil samples The environmental variablesof pH nitrate phosphate and TPH all had a greater influence onthe biological distribution than organic carbon The soil pH was

the greatest predictor value accounting for 18 of the total vari-ation between all samples as calculated with Pearsonrsquos correlation(P 0001) (Table 5)

Community indices With increasing diesel fuel concentra-tion the total species richness species diversity species evennesssimilarity indices and UniFrac measurements across most soilsdeclined from the controls (Fig 3) The control communitieswithin the low-carbon soils (P1) consisted of substantially lowerspecies numbers diversity and evenness than the higher-carbonsoils and as a result the effect of the diesel fuel on the communitieswas less severe The species richness within the medium-carbonsoils (P2) was sensitive to increasing diesel fuel with abundance-based coverage estimation (ACE) Chao1 and species observed(Sobs) values decreasing by up to 20 The Sobs value in the higher-carbon soils (P3) declined almost 50 from the control whileACE and Chao1 values were variable across all fuel concentra-tions In the long-exposure soils (P4) the loss of species richnessfrom the control occurred at higher concentrations than in theacute soils for Sobs ACE and Chao1 with a total decrease of 60observed

The Shannon diversity index (H=) Simpson diversity indexand Pieloursquos evenness index (J=) in the low-carbon soils (P1) werenot significantly impacted by SAB concentrations below 5000 mgkg1 In the medium carbon (P2) soils a gradual decline in theShannon diversity index was observed from 100 mg kg1 result-

TABLE 4 OTUs significantlya stimulated and inhibited across the SAB diesel fuel spiking range

Soil plotTotal no ofunique OTUs

Inhibited OTUs Stimulated OTUs

NoRelative abundancein control soils

Relative abundance inhigh soils (SE) No

Relative abundancein control soils

Relative abundance inhigh soils (SE)

P1 386 28 39 65 (195) 7 04 680 (104)P2 723 178 56 53 (08) 8 78 738 (77)P3 712 77 26 25 (31) 11 06 515 (13)P4 715 119 79 159 (03) 14 08 525 (14)a P 005

TABLE 5 DistLM results indicating strength of environmental variable as a predictor of the biological distribution and patterns of bacterialcommunities across all samples

Variable

Value (mg kg1)a

Marginal tests Sequential tests

Pseudo-F P Prop Pseudo-F P Prop Cumul

pH 728 lt0001 018 728 lt0001 018 018Nitrate 509 lt0001 013 573 lt0001 012 030Phosphate 430 lt0001 012 645 lt0001 012 042TPH 228 0019 006 399 lt0001 007 049Carbonb 513 lt0001 013 258 0006 004 053Nitrite 444 lt0001 012 222 0013 003 057Bromine 243 0008 006 152 0116 002 059Sulfate 277 0005 008 169 0079 002 062Conductivityc 406 0002 011 158 0110 002 064Chlorine 533 lt0001 014 146 0140 002 066Ammonium 130 0137 004 118 0289 002 068Dry matter fraction 540 lt0001 014 104 0377 002 069a Marginal test results are based on individual variables Sequential test results are based on the relative proportion of influence when all variables are considered Significantdifferences are in bold P values were considered significant at 005 Pseudo-F a multivariate version of Fisherrsquos F statistic utilized in PRIMER E Prop the proportion of variancepredicted with environmental variable Cumul the cumulative proportion of variance explained Values are in mg kg1 unless otherwise specifiedb Values are LOIc Values are in Scm

Microbial Indicators of Toxicity in Subantarctic Soil

July 2014 Volume 80 Number 13 aemasmorg 4027

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

ing in a 70 decrease from the control This decrease was lesssevere in the higher-carbon and aged soils (P3 and P4 respec-tively) with a 40 decrease from the control observed Pieloursquosevenness index in high-carbon soils (P3) decreased by approxi-mately 20 from the control with the decline detectable at con-centrations below 100 mg kg1 The P4 soil exposed to the con-taminant for 18 months exhibited different sensitivity comparedwith that of the short-exposure soils with the evenness index par-ticularly inhibited

The UniFrac dissimilarity measurement provided a sensitivetarget for all soils (Fig 3) The phylogenetic distance measurementchanged between 40 and 60 from the control soils for all foursamples The long-exposure soils (P4) exhibited a higher level ofsimilarity to the control soil and to soils spiked with 1000 mgkg1 fuel suggesting a higher tolerance or recovery in the geneticpotential of the community than short-exposure soils The ratio ofAcidobacteria to Proteobacteria which is indicative of the oligotro-phiccopiotrophic species ratio was reduced suggesting a shift

FIG 3 The effect of increasing SAB fuel concentrations on richness diversity and phylogenetic indices and key functional genes within the nitrogen cycle (nifHnitrogen fixation amoA ammonium oxidation nosZ denitrification) Values are expressed as percentage change from the control The best-fitting dose responsecurve as determined by AIC was fitted to each data set in the drc R package The selected dose-response regression model was used to calculate EC20 values

van Dorst et al

4028 aemasmorg Applied and Environmental Microbiology

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

toward faster opportunistic species within gamma- and betapro-teobacteria The low- and medium-carbon soils and long-expo-sure soils were sensitive to this ratio with 75 decreases observedcompared to the control soils The high-carbon soils were not

sensitive to this measurement with very high EC20 values gener-ated (Table 6)

Abundance of functional genes Of the functional genes eval-uated bacterial amoA (indicative of nitrification) was the most

TABLE 6 Generated EC20s across a range of community indices

Community indexa Modelb Plot

Treatment EC20 Model parameterse

C Exposure Concnc SEd b d e

Sobs W23 P1 5 Short NAf NA 003 1710 19SE (2532) P2 8 Short 140 80 044 919 4034df 23 P3 36 Short 30 340 015 1248 8713

P4 67 Long 1300 1000 066 1114 27637

ACE W23 P1 5 Short NA NA 004 1540 48671SE (2617) P2 8 Short 20 220 026 949 1491df 23 P3 36 Short NA NA 003 1532 515

P4 67 Long 1300 2000 053 1032 31184

Chao1 W23 P1 5 Short NA NA 000 1573 000001SE (205) P2 8 Short 30 16 028 939 1466df 23 P3 36 Short 11000 6400 061 979 23398

P4 67 Long 1500 620 059 1091 3334

Shannon (H=) W23 P1 5 Short 8100 850 191 988 10384SE (714) P2 8 Short 250 110 040 999 8164df 23 P3 36 Short 2200 1900 028 985 11865

P4 67 Long 3700 5300 021 1023 36297

Pieloursquos evenness (J=) W23 P1 5 Short 10400 830 113 1005 15863SE (330) P2 8 Short 140 470 024 1020 99925df 23 P3 36 Short 50 390 009 1254 11809

P4 67 Long NA NA 50273 1456 8824

Weighted UniFrac W23 P1 5 Short 10 128 013 1594 642SE (741) P2 8 Short 60 110 026 919 3484df 23 P3 36 Short 10 016 012 1729 198

P4 67 Long 940 750 041 934 30279

AcidobacteriaProteobacteriaratio

LL3 P1 5 Short 10 56 038 13 1841SE (34) P2 8 Short 40 110 103 104 1595

P3 36 Short 4500 2700 105 17 17195P4 67 Long 10 014 024 206 27

nifH W13 P1 5 Short NA NA 002 2825 15863SE(2539) P2 8 Short NA NA 001 3204 99925df 90 P3 36 Short 10 01 013 2440 11809

P4 67 Long 10 01 009 219 8824

amoA W23 P1 5 Short 10 006 016 214 04SE (170) P2 8 Short 410 240 040 1042 13566df 90 P3 36 Short 200 490 025 1137 13376

P4 67 Long 10 006 020 2284 02

nosZ W23 P1 5 Short 130 110 054 12440 20662SE (1737) P2 8 Short 740 540 078 17022 51191df 90 P3 36 Short 520 830 584 5262 6715

P4 67 Long 30 41 066 6056 3068a Sobs species observed ACE abundance-based coverage estimation nifH amoA and nosZ the abundance of functional genes within the nitrogen cycle as determined throughqPCRb The best-fitting dose response curve for the data as determined by AIC W23 Weibull 2 curve3 parameters W13 Weibull 1 curve3 parameters LL3 log logistic curve3parametersc The concentration that corresponds to a 20 effective change on the communityd The associated standard error of the EC20 estimate onlye Model parameters are defined as b relative slope d upper limit and e point of inflectionf NA no reliable dose response curve could be fitted

Microbial Indicators of Toxicity in Subantarctic Soil

July 2014 Volume 80 Number 13 aemasmorg 4029

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

sensitive indicator of hydrocarbon toxicity in soil (Fig 3) Thelow-carbon and aged soils (P1 and P4 respectively) were mostsensitive to the bacterial amoA measurement with a significantdecline in copy numbers observed at very low concentrations Adecline of bacterial amoA copy numbers in the medium- andhigh-carbon soils (P2 and P3 respectively) was also observed forSAB concentrations 100 mg kg1 The archaeal amoA gene waspresent in low to nondetectable levels in all but one of the soils (P2see Fig S2 in the supplemental material) Within P2 the abun-dance of archaeal amoA was an order of magnitude lower than thebacterial amoA gene ranging from 502 102 to 407 101 withabundances declining with increasing SAB concentrations Assuch only the bacterial amoA was pursued as an indicator andarchaeal amoA was excluded from further analysis The abun-dance of the nifH gene (representative of nitrogen fixation) wasvariable across all soils and diesel fuel concentrations analyzedthus no significant or measurable impact was observed ThenosZ gene abundance (indicative of denitrification) was stim-ulated in all soils spiked with diesel fuel While the greatestoverall increases were observed in the low- and medium-car-bon soils similar increases in the nosZ gene copy numbersoccurred at low TPH concentrations for all soils and the low-carbon and long-term exposure soils generated similar sensi-tive dose-response curves between 100 and 1000 mg kg1

Dose-response modeling Within the low-carbon soil (P1)the EC20 values based on percentage change from the controlgenerated high values outside the experimental spiking range forthe Simpson index and no significant response was measurablefor ACE Chao1 and Sobs (Table 6) Overall the calculations ofEC20s across the traditional community diversity measures in-cluding Sobs ACE Chao1 Shannon Simpson and Pieloursquos even-ness (J=) generated results with high associated errors and largevariability between soil types (Fig 4 Table 6) In contrast theabundances of the bacterial amoA and nosZ genes along with thecommunity UniFrac measurements generated sensitive EC20 esti-mates with less associated error and a response that was consistentacross all soil types (Fig 4 Table 6) For bacterial amoA EC20

concentrations ranged from 10 mg kg1 to 400 mg kg1 with anaverage EC20 of 155 mg kg1 (standard deviation [SD] 195 mgkg1) For the nosZ gene and UniFrac measurements EC20 con-centrations ranged between 250 mg kg1 and 700 mg kg1 withan average of 355 mg kg1 (SD 330 mg kg1) for nosZ and 250 mgkg1 (SD 460 mg kg1) for UniFrac (Table 6) For bacterial amoAand nosZ and the AcidobacteriaProteobacteria ratio the soil plotswith the lowest carbon content (P1 and P4) generated the mostsensitive EC20 values For the UniFrac measurement the mostsensitive EC20 values were found in the low (P1)- and high (P3)-carbon soils (Table 6)

DISCUSSION

The primary observation of diesel fuel impacts on subantarcticsoil microbial communities was a reduction in evenness and rich-ness resulting in a bacterial community that was heavily domi-nated by a few specific genera Our results suggest that the declinein diversity and richness observed after diesel fuel contaminationwas linked to the disruption of the nitrogen cycle and affectedniche-specific species The sensitivity low associated estimate er-rors and sustained inhibition of the bacterial amoA gene acrossvariable soil types suggest inhibition of potential nitrification ac-tivity is likely to be the best microbial indicator of soil sensitivity todiesel fuel for subantarctic soils The nosZ gene abundance Aci-dobacteriaProteobacteria ratio and UniFrac similarity are also po-tentially valuable microbial targets The average EC20 value (155mg kg1 [standard error 95 mg kg1]) generated from the abun-dance of the bacterial amoA gene was within our range of lowdiesel fuel concentration (400 mg kg1) This concentration isconsistent with previous ecotoxicology investigations at Macqua-rie Island where an EC20 of 190 mg kg1 was generated from anacute potential nitrification enzyme assay and a protective con-centration between 50 and 200 mg kg1 was recommended basedon avoidance survival and reproduction tests of the endemicMacquarie Island earthworm (9 28) Within wetland assess-ments community indices that relate to ecosystem-specific func-tionality such as the AOBAOA ratio have likewise been identi-fied as promising microbial indicators (10)

In animal and plant ecosystems a loss of biodiversity is oftenlinked to a decline in community functionality and a decreasedresilience to disturbance (52) In the past microbial ecosystemshave been thought to be more resilient to disturbance than plantand animal ecosystems due to their high functional redundancy(53) However uncertainty on the extent of this functional redun-dancy in microbial communities is growing For example Allisonand Martiny in 2008 reviewed over 110 studies investigating theeffects of heavy metals hydrocarbons and fertilizer amendmentsand found that microbial communities are functionally variable(54) and therefore not as resistant or resilient to disturbances aspreviously thought More recently Cravo-Laureau et al reportedthat a moderate loss of microbial diversity resulted in a significantdecline in functional phenanthrene degradation capacity high-lighting that a small reduction in diversity can result in significantimpacts on microbial ecosystem functionality (55)

High-throughput sequencing and qPCR targeting functionalgenes involved in the nitrogen cycle were used here to simultane-ously identify the stimulated and inhibited portions of the bacte-rial community This combined approach enabled the vulnerabletaxa within the bacterial community to be identified Across thefour subantarctic soil types the overall phylogenetic diversity the

FIG 4 Comparison of EC20 values derived from community measures acrossvariable carbon contents no reliable dose-response model could be fitted

van Dorst et al

4030 aemasmorg Applied and Environmental Microbiology

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

potential nitrification species and the Acidobacteria phylum werefound to be significantly inhibited in response to SAB diesel fuelThis was especially important in the low-carbon soils where thebroad community diversity estimates of Sobs ACE and Chao1were unable to detect a significant response to the SAB due topreexisting low species richness While the limitations of widelyused diversity estimates have been previously reported (56 57) itis important to note that without the targeted and functionallyimportant indices used here the establishment of sensitive specieswould have been missed due to the limited response in the low-carbon soils or masked by the dominance of the few stimulatedspecies

One of the novel potential microbial indicators found to besensitive to SAB contamination was the AcidobacteriaProteobac-teria ratio (Fig 3) In 2007 Fierer et al suggested ecological par-titioning within the total bacterial diversity based on the r and Kselection criteria (58) Although encompassing high levels of phy-logenetic and physiological diversity it was found certain phylaexhibit oligotrophic (slow-growing K selection criteria) or copi-otrophic (rapid growth r selection criteria) life strategies In Fi-erer et al high numbers of Acidobacteria were found to correlatewith low-nutrient environments with slow-growing K-selectedspecies occupying specific environmental niches Conversely Pro-teobacteria in particular Betaproteobacteria correlated with copi-otrophic conditions and species capable of rapid r-selected growth(58) The ratio of Acidobacteria to Proteobacteria observed herewas consistent with the ecological partitioning trend and gener-ated sensitive EC20 values for P1 P2 and P4 between 10 and 110mg kg1 all within our low fuel concentration range (400 mgkg1) The low EC20 values suggest that low-nutrient subantarcticsoils were particularly susceptible to reductions in niche-specificspecies The high nutrient soils with preexisting high ratios ofrapidly growing opportunistic species had a much higher sensi-tivity threshold at 4500 mg kg1

Pseudomonas the most stimulated genus found here has well-known hydrocarbon-degrading capabilities and its presence hasbeen widely reported in petroleum hydrocarbon-contaminatedsites in Antarctica and Macquarie Island (13 14 59) The othergenus most stimulated in response to high diesel fuel concentra-tions was Parvibaculum which has also been reported to have thegenetic potential for hydrocarbon degradation (60) While the rapidstimulation of potential hydrocarbon degraders that we observed hasbeen well documented the full extent of species inhibited with dieselfuel (contributing up to 80 of the total abundance in control com-munities) has not previously been highlighted (Table 4 Fig 2) Ofthose species most inhibited by diesel fuel nitrifying bacteria werenotably vulnerable with significant declines in both the relativeabundance of nitrification species and bacterial amoA gene copynumbers (Fig 2 3 and 4)

Ammonium oxidation to nitrite and the subsequent oxidationof nitrite into nitrate are crucial processes in soil The inhibition ofnitrification species and bacterial amoA copy numbers is consis-tent with previous reports identifying nitrification as a sensitivemeasurement for microbial communities following contamina-tion with petroleum hydrocarbons (9 61) heavy metals (62) andsolvents (63) The specific toxicity of petroleum hydrocarbons onnitrification potential has been attributed previously to competi-tive inhibition with the ammonia monooxygenase enzyme bysmall aliphatic compounds competing directly with ammonia forthe active binding site (22) as well as noncompetitive inhibition

from compounds with larger molecular weight through bindingto a hydrophobic region of the ammonia monooxygenase (AMO)enzyme and influencing enzyme turnover (22 64) Doubt hasbeen raised concerning the use of ammonia oxidizers to assesstoxicity because these organisms are thought to recover from hy-drocarbon pollution (65) We observed here that unlike the con-trol the soils chronically exposed to SAB in P4 for 18 months hadlow to nondetectable levels of the genera involved in nitrificationas well as the bacterial amoA copy numbers It is important to notethat in many soils the AOA are more abundant than AOB and therole of archaea in ammonium oxidation should be considered inthese soils Additionally the bacterial and archaea amoA gene andthe nifH and nosZ genes were targeted with only one primer seteach in this study Many other primers capable of targeting thesegenes exist and while we are confident the trends would be con-sistent different primers may generate slightly different results

The amount of organic carbon in soils is thought to mediatetoxicity in terrestrial systems by binding toxic substrates therebylimiting their bioavailability in the environment Carbon presentin a system may also provide an alternative carbon source formicroorganisms unable to utilize petroleum hydrocarbons Pre-viously on Macquarie Island organic carbon has been correlatedto the microbial distribution and hydrocarbon-degrading capac-ity of soils (20) Here we found the sensitivity to potential nitrifi-cation inhibition denitrification stimulation and decreases in theAcidobacteriaProteobacteria ratio were greatest in the soils withthe lowest organic carbon content (Fig 3 Table 6) However wefound the carbon content did not mediate sensitivity of soils to allindices for example the UniFrac distance measurement was leastsensitive in the chronically exposed soil despite relatively low soilcarbon content (Table 6)

While carbon was correlated with the biological distributionpH soil moisture and available nutrient levels had more signifi-cant effects on the resulting bacterial communities (Table 5) Fur-thermore the soil characteristics within each soil plot had moreinfluence on the bacterial distribution than the diesel fuel concen-tration alone In a previous study assessing the influence of soilproperties on alkane-degrading communities at Macquarie Is-land the soil parameters of 48 samples from two chronically con-taminated sites and nine reference sites were measured (20) Thesoil plots of P1 and P4 had carbon content similar to that of thetwo contaminated sites but also shared similar pH nitrate andphosphate levels The P2 and P4 soil plots were closer in resem-blance to the reference soils of the island with substantially higherlevels of carbon nitrate and phosphate and for P2 less acidic soilGiven the observed inhibition of potential nitrification and stim-ulation of potential denitrification the decline in nitrate observedin the highest fuel concentration samples from P4 was of particu-lar interest The extended exposure of high diesel fuel concentra-tions shared low nitrate levels similar to those of the chronicallycontaminated sites reported by Powell et al (20) It is difficult toinfer if the low levels of nitrite and nitrate were driven by nutrientlimitations in the low-carbon soils or were a result of higher tox-icity levels resulting from lower levels of organic matter availablefor binding We can conclude that the response of the bacterialpopulation was different across soil types with toxicity more likelyto be greatest in low-carbon soils The mode of action and extentto which other environmental characteristics mediate sensitivityor resilience especially pH moisture and nutrients need to befurther explored

Microbial Indicators of Toxicity in Subantarctic Soil

July 2014 Volume 80 Number 13 aemasmorg 4031

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

With increasing numbers of new and untested chemical com-pounds reaching the environment there is an urgent need formore rapid ecotoxicology approaches than the traditional single-species tolerance tests (5 6) The application of high-throughputsequencing and qPCR technologies used in this study evaluatedthe sensitivity of more than 1700 species across 266 familiesfrom 39 separate phyla Consistent with rapid tolerance testingapproaches we believe the species coverage and confidence in theecosystem relevance are high However it must be noted that thelack of replication in our control samples may have introducedbias into the final analysis and increased replication particularlyfor the control soils would improve confidence in the individualtoxicity values obtained here Given their integral roles in terres-trial ecosystem function and the capacity for high-throughputanalysis we believe microbial populations have a lot to offer asbiological indicators The large number of species in low abun-dance found in the control soils also provided an indication ofwhat a nonimpacted bacterial community should resemble Giventhe proposed reduction in functional capacity with disturbance tothis community structure an emerging challenge for remediationtechnologies is also highlighted that is how to restore a balancedmicrobial community capable of sustainable biogeochemical cy-cling

ACKNOWLEDGMENTS

This work was supported by UNSW Australia internal grant fundingschemes and the Australian Antarctic Division

We thank the researchers and technical officers Tom Mooney SharynGaskin Susan Ferguson Tim Spedding Ben Raymond Jim WalworthClaire Wooldridge Brett Quinton Dan Wilkins Greg Hince AnnePalmer Lauren Wise and Jane Wasley These people took part in the 2009and 2010 Macquarie Island summer field seasons or provided supportfrom Kingston Tasmania Australia

REFERENCES1 AMAP 1998 AMAP assessment report Arctic pollution issues p 859

Arctic Monitoring and Assessment Programme Oslo Norway2 Wall DH Virginia RA 1999 Controls on soil biodiversity insights from

extreme environments Appl Soil Ecol 13137ndash150 httpdxdoiorg101016S0929-1393(99)00029-3

3 Snape I Acomb L Barnes DL Bainbridge S Eno R Filler DM Plato NJS P Raymond T Rayner J Riddle M Rike AG Rutter A Schafer ASiciliano SD Walworth J 2008 Contamination regulation and reme-diation an introduction to bioremediation of petroleum hydrocarbons incold regions p 1ndash37 Cambridge University Press Cambridge UnitedKingdom

4 Dagnino A Sforzini S Dondero F Fenoglio S Bona E Jensen JViarengo A 2008 A weight-of-evidence approach for the integration ofenvironmental ldquotriadrdquo data to assess ecological risk and biological vulner-ability Integr Environ Assess Manag 4314 ndash326 httpdxdoiorg101897IEAM_2007-0671

5 Hickey GL Kefford BJ Dunlop JE Craig PS 2008 Making speciessalinity sensitivity distributions reflective of naturally occurring commu-nities using rapid testing and Bayesian statistics Environ Toxicol Chem272403ndash2411 httpdxdoiorg10189708-0791

6 Kefford BJ Palmer CG Jooste S Warne MSJ Nugegoda D 2005 Whatis meant by ldquo95 of speciesrdquo An argument for the inclusion of rapidtolerance testing Human Ecol Risk Assess 111025ndash1046 httpdxdoiorg10108010807030500257770

7 Neilson MN Winding A 2002 Microorganisms as indicators of soilhealth p 14 ndash16 In NERI (ed) NERI technical report no 338 InstituteNER Roskilde Denmark

8 Avidano L Gamalero E Cossa G Carraro E 2005 Characterization ofsoil health in an Italian polluted site by using microorganisms as bioindi-cators Appl Soil Ecol 3021ndash33 httpdxdoiorg101016japsoil200501003

9 Schafer AN Snape I Siciliano SD 2007 Soil biogeochemical toxicity endpoints for sub-Antarctic islands contaminated with petroleum hydrocar-bons Environ Toxicol Chem 26890 ndash 897 httpdxdoiorg10189706-420R1

10 Sims A Zhang Y Gajaraj S Brown PB Hu Z 2013 Toward thedevelopment of microbial indicators for wetland assessment Water Res471711ndash1725 httpdxdoiorg101016jwatres201301023

11 Nazaries L Pan Y Bodrossy L Baggs EM Millard P Murrell JC SinghBK 2013 Evidence of microbial regulation of biogeochemical cycles froma study on methane flux and land use change Appl Environ Microbiol794031ndash 4040 httpdxdoiorg101128AEM00095-13

12 Ruberto L Dias R Lo Balbo A Vazquez SC Hernandez EA MacCor-mack WP 2009 Influence of nutrients addition and bioaugmentation onthe hydrocarbon biodegradation of a chronically contaminated Antarcticsoil J Appl Microbiol 1061101ndash1110 httpdxdoiorg101111j1365-2672200804073x

13 Aislabie JM Balks MR Foght JM Waterhouse EJ 2004 Hydrocarbonspills on Antarctic soils effects and management Environ Sci Technol381265ndash1274 httpdxdoiorg101021es0305149

14 Saul DJ Aislabie JM Brown CE Harris L Foght JM 2005 Hydrocar-bon contamination changes the bacterial diversity of soil from aroundScott Base Antarctica FEMS Microbiol Ecol 53141ndash155 httpdxdoiorg101016jfemsec200411007

15 Hamamura N Olson SH Ward DM Inskeep WP 2006 Microbialpopulation dynamics associated with crude-oil biodegradation in diversesoils Appl Environ Microbiol 726316 ndash 6324 httpdxdoiorg101128AEM01015-06

16 Guo GX Deng H Qiao M Mu YJ Zhu YG 2011 Effect of pyrene ondenitrification activity and abundance and composition of denitrifyingcommunity in an agricultural soil Environ Poll 1591886 ndash1895 httpdxdoiorg101016jenvpol201103035

17 Deng H Guo GX Zhu YG 2011 Pyrene effects on methanotrophcommunity and methane oxidation rate tested by dose-response experi-ment and resistance and resilience experiment J Soils Sedim 11312ndash321httpdxdoiorg101007s11368-010-0306-3

18 Lauber CL Hamady M Knight R Fierer N 2009 Pyrosequencing-based assessment of soil pH as a predictor of soil bacterial communitystructure at the continental scale Appl Environ Microbiol 755111ndash5120 httpdxdoiorg101128AEM00335-09

19 Fierer N Jackson RB 2006 The diversity and biogeography of soil bac-terial communities Proc Natl Acad Sci U S A 103626 ndash 631 httpdxdoiorg101073pnas0507535103

20 Powell SM Bowman JP Ferguson SH Snape I 2010 The importance ofsoil characteristics to the structure of alkane-degrading bacterial commu-nities on sub-Antarctic Macquarie Island Soil Biol Biochem 422012ndash2021 httpdxdoiorg101016jsoilbio201007027

21 Hyman MR Murton IB Arp DJ 1988 Interaction of ammonia mono-oxygenase from Nitrosomonas europaea with alkanes alkenes and alkynesAppl Environ Microbiol 543187ndash3190

22 Keener WK Arp DJ 1993 Kinetic studies of ammonia moonooxygenaseinhibition in Nitrosomonas europaea by hydrocarbons and halogenatedhydrocarbons in an optimized whole-cell assay Appl Environ Microbiol592501ndash2510

23 Bissett A Richardson AE Baker G Thrall PH 2011 Long-term land useeffects on soil microbial community structure and function Appl SoilEcol 5166 ndash78 httpdxdoiorg101016japsoil201108010

24 Colloff MJ Wakelin SA Gomez D Rogers SL 2008 Detection ofnitrogen cycle genes in soils for measuring the effects of changes in landuse and management Soil Biol Biochem 401637ndash1645 httpdxdoiorg101016jsoilbio200801019

25 Deprez P Arens M Locher H 1994 Identification and preliminaryassessment of contaminated sites in the Australian Antarctic TerritoryAustralian Antarctic Division Hobart Australia

26 Rayner JL Snape I Walworth JL Harvey PM Ferguson SH 2007Petroleum-hydrocarbon contamination and remediation by microbio-venting at sub-Antarctic Macquarie Island Cold Reg Sci Technol 48139 ndash153 httpdxdoiorg101016jcoldregions200611001

27 Walworth J Pond A Snape I Rayner J Ferguson S Harvey P 2007Nitrogen requirements for maximizing petroleum bioremediation in asub-Antarctic soil Cold Reg Sci Technol 4884 ndash91 httpdxdoiorg101016jcoldregions200607001

28 Mooney TJ King CK Wasley J Andrew NR 2013 Toxicity of dieselcontaminated soils to the subantarctic earthworm Microscolex macquar-

van Dorst et al

4032 aemasmorg Applied and Environmental Microbiology

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from

iensis Environ Toxicol Chem 32370 ndash377 httpdxdoiorg101002etc2060

29 Li H Zhang Y Kravchenko I Xu H Zhang C-G 2007 Dynamic changesin microbial activity and community structure during biodegradation ofpetroleum compounds a laboratory experiment J Environ Sci 191003ndash1013 httpdxdoiorg101016S1001-0742(07)60163-6

30 Rayment G Lyons D 2011 Soil chemical methodsmdashAustralasia CSIROPublishing Collingwood VIC Australia

31 van Dorst J Bissett A Palmer AS Brown M Snape I Stark JSRaymond B McKinlay J Ji M Winsley T Ferrari BC 21 March 2014Community fingerprinting in a sequencing world FEMS Microbiol Ecolhttpdxdoiorg1011111574-694112308

32 Lane DJ 1991 16S23S rRNA sequencing p 115ndash175 In Stackebrandt EGoodfellow M (ed) Nucleic acid techniques in bacterial systematics JohnWiley amp Sons Chichester United Kingdom

33 Dowd SE Sun Y Wolcott RD Domingo A Carroll JA 2008 Bacterialtag-encoded FLX amplicon pyrosequencing (bTEFAP) for microbiomestudies bacterial diversity in the ileum of newly weaned Salmonella-infected pigs Foodborne Pathog Dis 5459 ndash 472 httpdxdoiorg101089fpd20080107

34 Schloss PD Westcott SL Ryabin T Hall JR Hartmann M Hollister EBLesniewski RA Oakley BB Parks DH Robinson CJ Sahl JW Stres BThallinger GG Van Horn DJ Weber CF 2009 Introducing mothuropen-source platform-independent community-supported software fordescribing and comparing microbial communities Appl Environ Micro-biol 757537ndash7541 httpdxdoiorg101128AEM01541-09

35 Pruesse E Quast C Knittel K Fuchs BM Ludwig W Peplies J Gloumlck-ner FO 2007 SILVA a comprehensive online resource for qualitychecked and aligned ribosomal RNA sequence data compatible with ARBNucleic Acids Res 357188 ndash7196 httpdxdoiorg101093nargkm864

36 Edgar RC Haas BJ Clemente JC Quince C Knight R 2011 UCHIMEimproves sensitivity and speed of chimera detection Bioinformatics 272194 ndash2200 httpdxdoiorg101093bioinformaticsbtr381

37 Kim M Morrison M Yu Z 2011 Evaluation of different partial 16SrRNA gene sequence regions for phylogenetic analysis of microbiomes JMicrobiol Meth 8481ndash 87 httpdxdoiorg101016jmimet201010020

38 Labbeacute D Margesin R Schinner F Whyte LG Greer CW 2007 Com-parative phylogenetic analysis of microbial communities in pristine andhydrocarbon-contaminated Alpine soils FEMS Microbiol Ecol 59466 ndash475 httpdxdoiorg101111j1574-6941200600250x

39 Sheneman L Evans J Foster JA 2006 Clearcut a fast implementation ofrelaxed neighbor joining Bioinformatics 222823ndash2824 httpdxdoiorg101093bioinformaticsbtl478

40 Lozupone C Hamady M Knight R 2006 UniFracmdashan online tool forcomparing microbial community diversity in a phylogenetic contextBMC Bioinformatics 7371 httpdxdoiorg1011861471-2105-7-371

41 Clarke KR Warwick RM 2001 Changes in marine communities anapproach to statistical analysis and interpretation PRIMER-E PlymouthUnited Kingdom

42 Ma WK Bedard-Haughn A Siciliano SD Farrell RE 2008 Relationshipbetween nitrifier and denitrifier community composition and abundancein predicting nitrous oxide emissions from ephemeral wetland soils SoilBiol Biochem 401114 ndash1123 httpdxdoiorg101016jsoilbio200712004

43 Gaby JC Buckley DH 2012 A comprehensive evaluation of PCR primersto amplify the nifH gene of nitrogenase PLoS One 7e42149 httpdxdoiorg101371journalpone0042149

44 Rotthauwe JH Witzel KP Liesack W 1997 The ammonia monooxy-genase structural gene amoA as a functional marker molecular fine-scaleanalysis of natural ammonia-oxidizing populations Appl Environ Mi-crobiol 634704 ndash 4712

45 Francis CA Roberts KJ Beman JM Santoro AE Oakley BB 2005Ubiquity and diversity of ammonia-oxidizing archaea in water columnsand sediments of the ocean Proc Natl Acad Sci U S A 10214683ndash14688 httpdxdoiorg101073pnas0506625102

46 Henry S Bru D Stres B Hallet S Philippot L 2006 Quantitativedetection of the nosZ gene encoding nitrous oxide reductase and com-parison of the abundances of 16S rRNA narG nirK and nosZ genes in

soils Appl Environ Microbiol 725181ndash5189 httpdxdoiorg101128AEM00231-06

47 Siciliano SD Ma W Powell S 2007 Evaluation of quantitative polymer-ase chain reaction to assess nosZ gene prevalence in mixed microbial com-munities Can J Microbiol 53636 ndash 642 httpdxdoiorg101139W07-014

48 Zhang X Liu W Schloter M Zhang G Chen Q Huang J Li L Elser JJHan X 2013 Response of the abundance of key soil microbial nitrogen-cycling genes to multi-factorial global changes PLoS One 8(10)e76500httpdxdoiorg101371journalpone0076500

49 Cottingham KL Lennon JT Brown BL 2005 Knowing when to draw theline designing more informative ecological experiments Front EcolEnviron3145ndash152httpdxdoiorg1018901540-9295(2005)003[0145KWTDTL]20CO2

50 Knezevic SZ Streibig JC Ritz C 2007 Utilizing R software package fordose-response studies the concept and data analysis Weed Technol 21840 ndash 848 httpdxdoiorg101614WT-06-1611

51 Ritz C Streibig JC 2005 Bioassay analysis using R J Stat Software121ndash22 httpwwwjstatsoftorgv12i05paper

52 Tilman D Wedin D Knops J 1996 Productivity and sustainabilityinfluenced by biodiversity in grassland ecosystems Nature 379718 ndash720httpdxdoiorg101038379718a0

53 Yachi S Loreau M 1999 Biodiversity and ecosystem productivity in afluctuating environment the insurance hypothesis Proc Natl Acad SciU S A 961463ndash1468 httpdxdoiorg101073pnas9641463

54 Allison SD Martiny JBH 2008 Resistance resilience and redundancy inmicrobial communities Proc Natl Acad Sci U S A 10511512ndash11519httpdxdoiorg101073pnas0801925105

55 Cravo-Laureau C Hernandez-Raquet G Vitte I Jeacutezeacutequel R Bellet VGodon J-J Caumette P Balaguer P Duran R 2011 Role of environ-mental fluctuations and microbial diversity in degradation of hydrocar-bons in contaminated sludge Res Microbiol 162888 ndash 895 httpdxdoiorg101016jresmic201104011

56 Curtis TP Sloan WT 2004 Prokaryotic diversity and its limits microbialcommunity structure in nature and implications for microbial ecologyCurr Opin Microbiol 7221ndash226 httpdxdoiorg101016jmib200404010

57 Bent SJ Forney LJ 2008 The tragedy of the uncommon understandinglimitations in the analysis of microbial diversity ISME J 2689 ndash 695 httpdxdoiorg101038ismej200844

58 Fierer N Bradford MA Jackson RB 2007 Toward an ecological classi-fication of soil bacteria Ecology 881354 ndash1364 httpdxdoiorg10189005-1839

59 Powell SM Ma WK Siciliano SD 2006 Isolation of denitrifying bacteriafrom hydrocarbon-contaminated Antarctic soil Polar Biol 3069 ndash74httpdxdoiorg101007s00300-006-0161-2

60 Flocco CG Gomes NCM Mac Cormack W Smalla K 2009 Occurrenceand diversity of naphthalene dioxygenase genes in soil microbial commu-nities from the maritime Antarctic Environ Microbiol 11700 ndash714 httpdxdoiorg101111j1462-2920200801858x

61 Sverdrup LE Ekelund F Krogh PH Nielsen T Johnsen K 2002 Soilmicrobial toxicity of eight polycyclic aromatic compounds effects on ni-trification the genetic diversity of bacteria and the total number of pro-tozoans Environ Toxicol Chem 211644 ndash1650 httpdxdoiorg101002etc5620210815

62 Pereira R Sousa J Ribeiro R Gonccedilalves F 2006 Microbial indicators inmine soils (S Domingos mine Portugal) Soil Sedim Contam 15147ndash167 httpdxdoiorg10108015320380500506813

63 Miller JL Sardo MA Thompson TL Miller RM 1997 Effect of appli-cation solvents on heterotrophic and nitrifying populations in soil micro-cosms Environ Toxicol Chem 16447ndash 451 httpdxdoiorg101002etc5620160309

64 Chang SW Hyman MR Williamson KJ 2002 Cooxidation of naptha-lene and other polycyclic aromatic hydrocarbons by the nitrifying bacte-rium Nitrosomonas europaea Biodegradation 13373ndash381 httpdxdoiorg101023A1022811430030

65 Deni J Penninckx MJ 2004 Influence of long-term diesel fuel pollutionon nitrite-oxidising activity and population size of nitrobacter spp in soilMicrobiol Res 159323ndash329 httpdxdoiorg101016jmicres200406004

Microbial Indicators of Toxicity in Subantarctic Soil

July 2014 Volume 80 Number 13 aemasmorg 4033

on June 9 2014 by UN

SW

Libraryhttpaem

asmorg

Dow

nloaded from