Arbuscular Mycorhizal Fungi Associated with the Olive Crop across the Andalusian Landscape

12 Pages·2014·3.33 MB·English

Arbuscular Mycorhizal Fungi Associated with the Olive
Crop across the Andalusian Landscape: Factors Driving
Community Differentiation
Miguel Montes-Borrego1, Madis Metsis2, Blanca B. Landa1*
1DepartmentofCropProtection,InstituteforSustainableAgriculture(IAS-CSIC),Cordoba,Spain,2TallinnUniversity,InstituteofMathematicsandNaturalSciences,
Tallinn,Estonia
Abstract
Background:In the last years, many olive plantations in southern Spain have been mediated by the use of self-rooted
planting stocks, which have incorporated commercial AMF during the nursery period to facilitate their establishment.
However, this was practised without enough knowledge on the effect of cropping practices and environment on the
biodiversity ofAMFinoliveorchards inSpain.
Methodology/PrincipalFindings:Twoculture-independentmolecularmethodswereusedtostudytheAMFcommunities
associated with olive in a wide-region analysis in southern Spain including 96 olive locations. The use of T-RFLP and
pyrosequencing analysis of rDNA sequences provided the first evidence of an effect of agronomic and climatic
characteristics,andsoilphysicochemicalpropertiesonAMFcommunitycompositionassociatedwitholive.Thus,thefactors
moststronglyassociatedtoAMFdistributionvariedaccordingtothetechniquebutincludedamongthestudiedagronomic
characteristics the cultivar genotype and age of plantation and the irrigation regimen but not the orchard management
systemorpresenceofacovercroptopreventsoilerosion.Soilphysicochemicalpropertiesandclimaticcharacteristicsmost
strongly associated to the AMF community composition included pH, textural components and nutrient contents of soil,
and average evapotranspiration, rainfall and minimum temperature of the sampled locations. Pyrosequencing analysis
revealed33AMFOTUsbelongingtofivefamilies,withArchaeosporaspp.,Diversisporaspp.andParaglomusspp.,beingfirst
records in olive. Interestingly, two of the most frequent OTUs included a diverse group of Claroideoglomeraceae and
Glomeraceae sequences, not assigned to any known AMF species commonly used as inoculants in olive during nursery
propagation.
Conclusions/Significance:Our data suggests that AMF can exert higher host specificity in olive than previously thought,
which may have important implications for redirecting the olive nursery process in the future as well as to take into
considerationthe specific soils and environmentswherethe mycorrhized olivetreeswill beestablished.
Citation:Montes-BorregoM,MetsisM,LandaBB(2014)ArbuscularMycorhizalFungiAssociatedwiththeOliveCropacrosstheAndalusianLandscape:Factors
DrivingCommunityDifferentiation.PLoSONE9(5):e96397.doi:10.1371/journal.pone.0096397
Editor:RaffaellaBalestrini,InstituteforPlantProtection(IPP),CNR,Italy
ReceivedNovember22,2013;AcceptedApril7,2014;PublishedMay5,2014
Copyright:(cid:2)2014Montes-Borregoetal.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermits
unrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited.
Funding:ThisresearchwassupportedbygrantsfromProjectsAGL2008-00344/AGRfrom‘MinisteriodeCienciaeInnovacio´n’,ProjectP10-AGR-5908from
‘Consejer´ıadeEconom´ıa,Innovacio´nyCiencia’ofJuntadeAndaluc´ıa,ProjectAGL-2012-37521from‘MinisteriodeEconom´ıayCompetitividad’ofSpain,and
FondoEuropeodedesarrolloregional(FEDER)"UnamaneradehacerEuropa"fromtheEuropeanUnion.Thegrant219262ArimNET_ERANETFP72012–2015
ProjectPESTOLIVEfromInstitutoNacionaldeInvestigacio´nyTecnolog´ıaAgrariayAlimentaria(INIA),alsoprovidedpartialfinancialsupport.M.Montes-Borrego
enjoyedacontractfrom‘Consejer´ıadeEconom´ıa,Innovacio´nyCiencia’ofJuntadeAndaluc´ıa,Spain.Thefundershadnoroleinstudydesign,datacollectionand
analysis,decisiontopublish,orpreparationofthemanuscript.
CompetingInterests:Theauthorshavedeclaredthatnocompetinginterestsexist.
*E-mail:[email protected]
Introduction a vegetative cover to prevent soilerosion [3,4,5]. Additionally, in
thelastyears,manynewoliveplantationshavebeenmediatedby
Spain is the world’s largest olive oil producer, accounting for the use of self-rooted planting stocks which have incorporated
more than one-third of global production [1,2]. In Andalusia commercial arbuscular mycorrhizal fungi (AMF) in the potting
(Southern Spain), olive orchards dominate the landscape in an mixture during the nursery period to facilitate establishment
impressive monoculture that covers approximately 17% of the [6,7,8] due to their beneficial effects against biotic and abiotic
total surface of the region (1.5 million ha) [1,3]. In this region, stresses [9,10,11,12,13].
different olive farming systems can be found including: i) One of the critical steps for applying AMF to improve crop
conventional farming with rain fed orchards of low plant density, health is the appropriate selection of effective and well adapted-
intensive tillage, and low inputs in fertilizer, as well as intensive isolates to be used as inoculants. Although it is well known that
drip-irrigatedorchards,grownwithhigherinputsofpesticidesand olivetreeisamycotrophicplant[7]thereisnotenoughknowledge
fertilizers in order to push up olive yields, and ii) organic farming concerning the effect of cropping practices, olive genotype and
usingnochemicalinputsandmainlynonorlight-tillageanduseof environment(soiltypeandclimate)onthebiodiversityofAMFin
PLOSONE | www.plosone.org 1 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
oliveorchardsintheMediterraneanRegion.Knowledgeofthose fromSigMapa,GeographicInformationSystemfromtheSpanish
factors may be essential to take advantage on the use of AMF in Ministryof‘‘MedioAmbienteyMedioRuralyMarino’’(http://
modern oliviculture. sig.mapa.es/geoportal/) using ArcGIS 10 (ESRI, Redlands,
In the present study we have examined the structure and California, EE.UU.)(TableS1).
diversityofAMFcommunitiesintherhizosphereofcultivatedand
wild olives in Andalusia, southern Spain, by using two culture- DNA Extraction from Rhizosphere Samples
independent molecular approaches: Fluorescent terminal restric- Pooled olive root samples were cut into 1-cm pieces with a
tion fragment length polymorphism (T-RFLP) analyses of ampli- sterile scissors to get a uniform sample per location. Rhizosphere
fied28S rDNAsequences andSSUrDNAampliconparallel 454 suspensions (including rhizosphere soil and rhizoplane) were
pyrosequencing. We also have determined which agronomic or obtained by vigorously shaking 2g of root segments (four
environmentalfactorsassociatedtotheoliveorchardssampledare
independent replications) suspended in 20ml of sterile distilled
themain driversoftheAMF structure.
waterinanorbitalshakerfor10minandsonicated(Ultrasons,JP
SelectaSA,Barcelona,Spain)for10minutes.Then,3 mlofthose
Materials and Methods rhizosphere suspensions were subjected to two consecutive
centrifugations at 11,000 rpm for 4 min and the pellet was kept
Ethics Statement
at220uCuntilprocessed.DNAfromeachofthefourrhizosphere
Nospecificpermitswererequiredforthedescribedfieldstudies.
soilpellets(approximately200mg;fourreplicationpereachofthe
Location of organic olive orchards was provided by the
96 olive orchards) was extracted using the PowerSoil DNA
Andalusian Committee of Organic Farming (CAAE, Junta de
Isolation Kit (MO BIO Laboratories, Inc., Carlsbad, USA) and
Andaluc´ıa).Permissiontosampletheoliveorchardsweregranted
the FastPrep-24 (MP Biomedicals, Inc., Illkirch, France) instru-
by the landowner. The samples from wild or feral forms of olive
ment run at 6.0m/s for40sas describedelsewhere [14].
arelocatedinpublicareasordegradedformationsandabandoned
groves. The specific location of all samples from the study is
T-RFLP Analysis
providedinTableS1.The96oliveorchardssampledinthisstudy
For T-RFLP analysis PCR amplification of partial LSU of
werealsoincludedinpreviousstudies[5,14]inwhichthebacterial
rDNA from mycorrhiza were performed using a nested-PCR
communitiesandfunctionaldiversityoftheoliverhizospherewas
assessed.Thesitesarenotprotectedinanyway.Theareasstudied approach, the first PCR round employing 20ng of template and
donot involve anyspecies endangered orprotected in Spain. the primer pair LR1/FLR2 [15] and the second one the primer
pair FLR3/FLR4 [16] following conditions described by Mum-
mey and Rillig [17]. Primer FLR3 was 59 end-labeled with the
Location of Olive Orchards and Rhizosphere Sampling
fluorescent dye FAM. T-RFLP analysis was performed for all
Soil and roots samples were collected from 90 commercial
samples using 5 ml of PCR products (about 1000ng) and TaqI
orchards differing in management system [conventional (49
restriction enzyme (Fast Digest, Fermentas, Germany) in a final
orchards) vs. organic (41 orchards)] located in the main olive-
volumeof10ml.TaqIrestrictionenzyme wasselectedfromthose
growing areas of Co´rdoba (41 orchards), Granada (3 orchards),
(AluI,MboIandTaqI)thatwereshowntodiscriminatemoreAMF
Jae´n (34 orchards), and Sevilla (12 orchards) provinces in
groups in a previous study [17] after preliminary testing with a
Andalusia,southernSpain[5;TableS1].Inaddition,sixsamples
subset of our rhizosphere samples (data not shown). Terminal
(LO, LOBA, BAETICA, MACO, LOMCO, EPCO) from three
restriction fragments (TRF) were loaded and separated on a
siteseachinCo´rdobaandCa´dizprovincescontainingwildorferal
formsofolive‘Acebuches’(i.e.,secondarysexualderivativesofthe 3130XLgeneticanalyzer(AppliedBiosystems,California,USA)at
cultivated clones or products of hybridization between cultivated the SCAI-University of Co´rdoba sequencing facilities. Size of
trees and nearby oleasters) were included in thestudy [14; Table fragments were determined using a ROX500 size standard, and
S1]. Orchards across all locations sampled differ in climate, soil matrices containing incidence as well as peak area data of
texture and physicochemical characteristics, soil management individualTRFsweregeneratedforallsampleswithGeneMapper
system (use of cover crops vs. bare soil), and irrigation regimen software(AppliedBiosystems).Peaksoflessthan100fluorescence
(rain-fed vs. drip-irrigated). When possible, we tried to sample a units(FU)andshorterthan50bpwerenotincludedintheanalysis
representative distribution of the above considered factors within to eliminate primer dimmers and other small charged molecules.
each orchard management system. Detailed description of soil Similarly, molecules that were not present in at least two of the
physicochemical properties, and agronomic and climatic charac- four replicate profiles were disregarded. Also, TRFs that differed
teristicsofthesampledorchardsisprovidedinTableS1.Someof by less than 1 bp were clustered, unless individual peaks were
the soil physicochemical and climatic characteristics of the detectedinareproduciblemanner.TRFsprofileswerestandard-
sampled locations were provided recently[5,14]. izedbasedonmethodsdescribedpreviouslybyDunbaretal.[18].
Root(onlyyoungandactive)andsoilsampleswerecollectedin TherelativeabundanceofeachTRFwascalculatedastheratioof
MaytoJuly2009asdescribedbyArandaetal.[14]andMontes- thepeakareaforthatTRFtothesumofpeakareasforallTRFs
Borrego et al. [5] in the area of the canopy projection from the intheprofileandwasexpressedasapercentage.Diversitystatistics
upper 5 to 30cm of soil from three different points around each werecalculatedfromstandardizedprofilesofrhizospheresamples
individualtree.Eighttreesperorchardweresampled,andallroots by using the number and area of peaks in each profile as
from all trees were thoroughly mixed to obtain a single representativeofthenumberandrelativeabundanceofOTUs,as
representative sample per orchard. Intact root systems were definedbyDunbaretal.[19].Phylotyperichnesswascalculatedas
shaken gently by hand to remove all but the soil close- and thetotalnumberofdistinctTRFsizes(withlengthbetween50and
naturally-adhering to the plant root and were kept at 5uC until 500bp)inaprofile andtheShannon-Wiener diversityindexwas
processing. calculated as described before [14]. Finally, a single standardized
Additionally the geographic location and altitude of the T-RFLPprofileforeachorchardwasproducedbyaveragingpeak
sampling sites were determined using a global positioning system area foreach TRFfromfour replicates.
(GPS),andclimaticvariablesofeachsamplingsitewereobtained
PLOSONE | www.plosone.org 2 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
Pyrosequencing Analysis Taxonomy was assigned to OTUs by using the Basic Local
For454-pyrosequencingSSUrRNAGlomeromycotasequences AlignmentSearchTool(BLAST)foreachrepresentativesequence
were amplified from a DNA mixture obtained from the four against the Silva 108 database (http://www.arb-silva.de/
independentrhizosphereDNAextractionsperoliveorchard.This documentation/release-108/)aswellasbyBLASTsearchagainst
approach was taken to ideally cover as much biodiversity as the MaarjAM database (http://maarjam.botany.ut.ee/, [23]).
possibleandtoensurethatrepresentativeAMFcommunitiesfrom Sequences from the representative set of AMF OTUs obtained
each olive location were sampled [20]. The pyrosequencing was in this study, the reference AMF database from Redecker and
performed as described in Davison et al. [21] using a two-step Raab [28], the blast hits from Silva 108 and the MaarjAM
PCRprotocolwiththeprimersNS31andAML2,whichtargeta databases, and those AMF sequences reported in olive from
ca. 560-bp central fragment of the SSU rRNA gene in Calvente et al [7] and present in the GenBank database were
Glomeromycota [22], the most widely used marker in AMF aligned using ClustalW software [29] with default parameters.
surveys to date [23,24]. These primers were linked to partial SequencealignmentsweremanuallyeditedusingBioEditsoftware
sequencing primers A and B, respectively. Bar-code sequences, 8 [30]. Phylogenetic analysis of the sequence data sets was
bp in length, were inserted between the A primer and NS31 performed based on maximum likelihood (ML) and Bayesian
primer sequences. Thus, the composite forward primer was: 59- inference(BI)usingMrBayesversion3.1.2software[31].Thebest
GTCTCCGACTCAG(NNNNNNNN) TTGGAGGGCAAGTCTG- fitted modelofDNAevolutionwasobtainedusingjModelTest v.
GTGCC-39; and the reverse primer was 59-TTGGCAGTCT- 2.1.1 [32] with the Akaike information criterion (AIC). The
CAGGAACCCAAACACTTTGGTTTCC-3,wherepartialsequences Akaike-supported model, the base frequency, the proportion of
ofAandBprimersareunderlined,barcodeisindicatedbyN-sin invariablesites,andthegammadistributionshapeparametersand
parentheses and specific primers NS31 and AML2 are shown in substitution rates in the AIC were then used in phylogenetic
italic.Then,a10xdilutionofthefirstPCRproductwasusedina analyses.BIanalysisunderageneraltimereversibleofinvariable
secondPCRwherefullsequencingadapterswereadded(PrimerA sitesandagamma-shapeddistribution(TIM2+I+G)modelforthe
59-CCATCTCATCCCTGCGTGTCTCCGAC-39andPrimerB SSU rRNA,were runwith fourchains for1.06106generations.
59-CCTATCCCCTGTGTGCCTTGGCAGT -39). The reac-
tionscontained5 mlofSmart-TaqHotRed2xPCRMix(Tartu, Statistical Analysis
Naxo Ltd, Estonia), 1 ml of extracted DNA, and 0.2mM of each The rank-based Kruskall-Wallis test was used to determine
primer in a final volume of 10ml. The reactions were performed differences in the Richness and Shannon diversity indexes in
using a Thermal cycler 2720 (Applied Biosystems) under the relation to the different agronomic factors of the olive orchard
followingconditions:95uCfor15min;fivecyclesof42uCfor30s, evaluated using the Statistical Analysis System software package
72uC for 90 s, 92uC for 45 s; 35 (first PCR) or 20 (second PCR) (SAS version 9.2; SAS Institute, Cary, NC, USA). Non-metric
cyclesof65uCfor30s,72uCfor90s,92uCfor45s;followedby multidimensional scaling (NMDS) analyses were performed using
65uCfor30sand72uCfor10min.PCRproductswereseparated MetaMDS functions within the vegan package of R software (R
byelectrophoresisusing1.5%agarosegelsin0.5xTBE,andthe Core Development Team, 2005) [33] based on dissimilarities
PCR products were purified from the gel using the Qiagen calculatedusingtheBray–CurtisindexobtainedforT-RFLPand
QIAquick Gel Extraction kit (Qiagen Gmbh, Germany) and pyrosequencing results, using 1,000 runs with random starting
furtherpurifiedwithAgencourtHAMPureHXPPCRpurification configurations, and environmental variables (agronomic and
system (Agencourt Bioscience Co., Beverly, MA, USA). The 96 climatic characteristics and soil physicochemical properties) were
quantifiedsampleswerefinallymixedatequimolarconcentrations fitted using the envfit routine. For data derived from pyrose-
prior to sequencing. GATC Biotech (Constanz, Germany) quencing analysis only the Glomeromycota sequences were used.
performed sequencing procedures as custom service using a OrdinationsfortheBray–Curtisdissimilarityderivedfromrelative
Genome Sequencer FLX System and Titanium Series reagents frequency of OTUs in thepyrosequencing analysis did not reach
(Roche Applied Science, Indianapolis, IN, USA). Sequencing of acceptable [34] stress and stability levels and was not performed.
96samples wasperformed as apart of a biggerdataset. Instead, a Multivariate Regression Tree (MRT) was calculated.
MRT are a statistical technique that can be used to explore,
Processing of Pyrosequencing Data and phylogenetic describe,andpredictrelationshipsbetweenmultispeciesdataand
analysis environmental characteristics. MRT forms clusters of sites by
Pyrosequencing data were processed as described by Fierer et repeated splitting of the data, with each split defined by a simple
al. [25] using the Quantitative Insights Into Microbial Ecology rule based on environmental variables. The splits are chosen to
(QIIME) toolkit [26]. In brief, fungal sequences were quality minimizethedissimilarityofsiteswithinclusters[35].Thesumsof
trimmed,assignedtorhizospheresamplesbasedontheirbarcodes squares multivariate regression tree was calculated within the
and denoised using default parameters. Chimeras were identified mvpartpackagewiththeRsoftware,usingtheone-standarderror
with uclust_ref software [27] and removed, and the remaining ruleonthecross-validatedrelativeerrortodeterminethenumber
sequenceswerebinnedintoOTUsusinga97%identitythreshold of terminal nodes[35].
with uclust_ref software. Then, to take into account the different
number of sequences obtained for each orchard sample in the Results and Discussion
pyrosequencing analysis we estimated the relative frequency of
Diversity of olive AMF communities
each OTU in each orchard. Next, the most abundant sequence
fromeachOTUwasselectedasarepresentativesequenceforthat i) T-RFLPanalysis.Atotalof36uniqueTRFsprofileswere
OTU and deposited in the Genbank database under accessions consistently identified in the 384 rhizosphere samples
numbersKF831296-KF831328andtheentiredatasetofreadsin analyzed by T-RFLP analysis, with 30 TRFs (83.3%) found
the Sequence Read Archive of Genbank under BioProject ID in a reproducible manner in 93 of the 96 olive orchards
PRJNA237741. Alpha diversity statistics including Richness sampled. Mean Richness values ranged from 1 to 15
(numbersofOTUs)andtheShannonindexwerealsodetermined depending of the rhizosphere sample with an average of 5
for orchard sampleswith at leastfive sequences. TRFsperoliveorchard.Thistranslatesintoanestimatedtotal
PLOSONE | www.plosone.org 3 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
of30differentOTUspresentacrossallsampledorchardswith samples (with a mean of 11.4 sequences per orchard) (Figure 1;
13 and 3 OTUs being common in at least 25% or 50% of Table1;FigureS2;TableS2).Meanrichnessvaluesrangedfrom
olive orchards, respectively. Mean Shannon diversity index 1 to 9 with an average of 2.7 OTUs per olive orchard. Mean
values ranged from0.3 to 2.2 (Figure S1;Table S2). We did Shannon diversity index values ranged from 0 (1 single OTU) to
notfindsignificantdifferences(P.0.120)inAMFrichnessor 2.7 (Figures S1 and S2; Table S2). We did not find significant
Shannon diversity indexes derived from T-RFLP analysis differences (P.0.230) in AMF richness or Shannon diversity
according to the orchard management system, use of indexes derived from pyrosequencing analysis according to the
irrigation, presence of vegetative cover, olive tree variety or farm soil management system, use of irrigation, presence of
olive age (Figure S1). It has been shown that although vegetative cover, olivetree varietyor olive age(Figure S1).
diversity indexes are useful in describing community charac-
teristics, they do not provide information of important iii) Comparison of T-RFLP and pyrosequencing anal-
compositional features of biodiversity relating to the abun- ysis.Althoughinthepyrosequencinganalysiswegotsome
dances of shared taxa [36] and statistical analyses that orchard samples with no Glomeromycota sequences and this
incorporate taxon abundance and identity are more appro- technique is costly, labour-intensive and allows lower
priate to specifically assess changes in microbial community number of samples to be processed, it provides some
compositionandtoidentifytheexistence,ifany,ofagronomic advantages over the T-RFLP analysis. For example, the
or environmental gradients. latter technique does not provide any information on the
taxa identified, different taxa (species) may share similar
ii) Pyrosequencing analysis. The pyrosequencing approach
TRFsinelectropherograms,andmultipleTRFsprofilescan
of the 96 composite samples yielded a total of 13,772 high-
exist within a single species [41]. Furthermore, the lack of
quality reads after denoising, with a length .100 bp and ,
specififity of the primers used for T-RFLP have also been
550 bp, a mean of 147sequences per orchard field, and two
shown in other studies and may also be a source of errors
samples with no sequences (Table S2). Of these, most of
when PCR products serve as basis for those fingerprinting
sequences 47.2% could not be assigned to any Eukaryota
approaches [39,41]. Consequently, data dervived from T-
Phyla,29.6%wereassignedtothePhylumMetazoa,whereas
RFLP analysis should be complemented with other
10.2% (1,108 reads) could be assigned to OTUs from fungal
techniques that provide taxa identity information such as
families (Figure 1), which indicates that primers NS31 and
library cloning or pyrosequencing as was the case of this
AML2arenotenoughspecificforamplifyingAMFsequences.
study.
Otherstudiesusingpyrosequencinganalysishavealsoshown
that in spite of the supposed AMF primer specificity,
‘contaminant’ sequences belonging to taxa different from
Species identity of olive AMF communities
Glomeromycotaaredetected.Forexample,O¨piketal.[37]using
In our study the 33 OTUs identified represented most of the
sameNS31andAML2primersandBallestrinietal.[38]and
major AMF lineages, including Paraglomus spp. (family Paraglo-
Luminietal.[20]usingNS31andAMmixprimersalsofound
meraceae; two OTUs, comprising 4.20% of reads and 6.90% of
about .55% of amplified sequences belonging to taxa from
fields), Glomus group A (family Glomeraceae; 22 OTUs, compris-
non-Glomeromycota fungi. In a recent work Kohout et al. [39]
ing 59.73% of reads and 81% of fields), Glomus group B (family
alsodemonstratedthatacombinationofuptofiveprimersets
Claroideoglomaceae; four OTUs, 26.40% of reads and 38% of
specifically designed to amplify Glomeromycota, including
fields), Glomus group C (family Diversisporaceae; three OTUs,
primer NS31/AML2, co-amplifies to a high extend non-
6.03% of reads and 14% of fields), and Archaeospora (two OTUs,
targetAMFsequencesincludingplant,Asco-andBasidiomy-
3.62% and 12% of fields) (Figures 1 and 2; Table 1); with
cota(ca.20to50%,dependingoftheprimersetused).Inour
Archaeospora spp., Diversispora spp. and Paraglomus spp. being first
study, due to the fact that we did not retrieve Glomeromycota
recordsinolive.ItshouldbenotedthatOTUOAMF127clustered
sequences from some olive orchards our results could be
withthevirtualtaxonsequenceAF131054thathasbeenrecently
somehow biased. Consequently, to improve the number of
proposed as a potential new taxon (new family or even order)
sequences and molecular species characterization of AMF,
withinGlomeromycota [37].
new designed specific AMF-primers should be tested in
The fact that most sequences from our study belonged to the
complex matrixes such as soil or rhizosphere [39,40], or
Glomeraceae family (Glomus group A) that contains several cryptic
deeper sequencing effort should be done in future studies to
taxa with differences in ecological properties agrees with other
face this problem and to capture the total AMF diversity in
studies that have found many isolates of this group in different
olive.
locations through the world, including Mediterranean-type
ThefactthatweextractedDNAfromtheoliverhizosphere(i.e., environments,onbothnaturalwoodlandstohighinputmanaged
soil tightly adhered to roots) might have accounted for the low agro-ecosystems [20,42,43,44] suggesting that these taxa have a
presence of AMF sequences; probably extracting DNA from generalist ruderal style with tolerance to disturbance such as in
washed or entire roots may enhance the specificity of AMF agricultural ecosystems.
amplification.Inourstudy59.84%ofsequencesassignedtofungi Interestingly,wefoundthatthoseAMFspeciescommonlyused
belongedtoGlomeromycota(Figure1),with38oliveorchardsamples as olive inoculants or previously isolated from olive roots (i.e.,
retrievingnoGlomeromycotasequences(TableS2).Wedidnotfind Claroideglomus claroideum, Funneliformis mosseae, Rhizophagus clarus, R.
anypatternforthelackofGlomeromycotasequencesinthosesamples intraradices, and Septoglomus viscosum; see [7,8,9,12,13]) showed low
with any of the agronomic, climatic or soil physicochemical abundancesincetheywerepresentin1.7to18.97%offields.This
propertiesofthesampledorchards(datanotshown).Whendefining isinagreementwiththefactthatAMFbelongingtoGlomeraceae
anOTUasbelongingtoAMFonthebasisofhavingatleast97% family colonize preferentially the roots and might be present in
similaritytosequencesclassifiedasAMFintheSilvaandGenBank lower densities intherhizospheresoil [45].On thecontrary,two
databases we identified 33 OTUs that could be unequivocally AMF sequences including OAMF216 belonging to Claroideoglo-
assigned to the Glomeromycota in 58 out of 94 olive rhizosphere meraceae (18.6% of sequences), and OAMF91246, a new
PLOSONE | www.plosone.org 4 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
Figure 1. Proportion of overall phyla and disectioning of the fungal and Glomeromycota phyla detected by pyrosequencing
analysiswithprimersNS31/AML2fromrhizospheresamplesobtainedfrom96oliveorchardsinAndalusia,southernSpain.
doi:10.1371/journal.pone.0096397.g001
unidentified Glomeraceae (18.9% of sequences) that formed a variables on the estimated alpha-diversity indexes. Consequently,
separate cluster from other well-known AMF Glomeraceae taxa, in a second approach to specifically assess changes in AMF
were identified in 27.6% and 36.21% of olive orchards, community composition (incorporating taxon abundance (fre-
respectively (Figure 2; Table 1). This might indicate that AMF quency)andidentity),weusedNMDSordinationtorepresent,in
can exert higher host specificity in olive than previously thought two dimensions, the pairwise Bray-Curtis dissimilarities between
whichmayhaveimplications fortheolive nurseryprocess.Thus, AMF communities derived from T-RFLP analyses. Then, we
someauthorshavereportedadifferentialgrowthresponseofolive projected each of the environmental and agronomic variables
cultivars to AMF inoculation where this responsiveness to independentlyontotheNMDSordinationtoidentifyhypothetical
mycorrhization has been found to depend on both the AMF gradientslikelyrelatedtothedifferentiationinAMFcomposition
speciesandtheplantgenotype[7,46].Inourstudy,wedidnotfind (Figure 3; Table 2). In relation to agronomic variables AMF
anycleardifferencesbetweentheAMFsequencesdetectedinthe communities were differentiated according to the cultivar geno-
rhizosphere of wild olives and those found in the cultivated ones typeandageofplantationandtheirrigationregimenoftheolive
(Figure S2). This could be due to the small number of sequences orchard, in that order, whereas the grouping according to the
that we sampled from wild olives which deserves further studies orchardmanagementsystemorpresenceofavegetativecoverwas
since wild olives have been shown to be a potential reservoir for not significant (Table 2). Thus, there was a tendency to locate
discovering microbial species of diverse biotechnological and rhizospheresamplesintheNMDSordinationfromoliveorchards
commercial interest [14,47]. ,15 year old at the bottom quandrant of Y=0 (with only two
exceptions), whereas olive orchards of 15 to 30 year old were all
Factors shaping the structure of AMF communities in locatedontheleftcuadrantofX=0(Figure3).Theeffectofolive
gentoype in affecting soil biota has also been shown in a recent
olive rhizosphere
study [48] which demostrated that olive genotypes significantly
It has been shown that although diversity indexes (such as
influencedthenematodeassemblagespresentintheirrhizospheric
RichnessandShannonusedinourstudy)areusefulindescribing
soil.
community characteristics they do not provide information of
WealsoidentifiedC:Nratio,soilCandorganicmattercontent
important compositional features of microbial diversity related to
and pH as the environmental variables better explaining (P,
the abundances of shared taxa [36] which migth explain that we 0.001; 0.2836.r2.0.1301) the AMF community composition
did not find an effect of the environmental and agronomic
amongtheoliveorchards,inthatorder(Figure3,Table2).Other
PLOSONE | www.plosone.org 5 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
inthesampledolive AMMaarj Archaeosporasp.VT5 ArchaeosporatrappeiVT245 ParaglomuslaccatumVT281 ParaglomusbrasilianumVTX239 Glomussp.groupBVT193 GlomusgeosporumVT65 Glomussp.groupBVT56 Nomatch Glomussp.MO-GC1VT60 Glomussp.WirselOTU20VT62 Diversisporasp.VT62 Glomussp.Dictamnus2VT163 Glomussp.Douhan6VT143 Glomussp.VT145 Glomussp.VT153 Glomussp.VT118 GlomusWirselVT137 GlomusJP3VT128 Glomussp.Alguacil09bVT109 Glomussp.Glo45VT109 Glomussp.Glo24VT105 GlomusintraradicesVT113 VT113Glomusintraradices
e
c
n
equencyofoccurre bClosesttaxa Silva108 Glomeromycotasp. Archaeosporatrappei Fungi Fungi Glomusetunicatum Glomusetunicatum Glomusetunicatum Glomusetunicatum Glomuseburneum Glomuseburneum Glomuseburneum Glomeromycota UnculturedGlomussp. UnculturedGlomussp. Glomeromycota UnculturedGlomussp. UnculturedGlomussp. UnculturedGlomussp. Glomeromycota UnculturedGlomussp. UnculturedGlomussp. Rhizophagusintraradices Rhizophagusintraradices
fr
a,
ax of%)
omeacht Frequencyorchards( 10.34 1.72 1.72 6.90 18.97 27.59 1.72 1.72 27.59 3.45 12.07 1.72 13.79 1.72 5.17 5.17 1.72 36.21 36.21 15.52 1.72 3.45 8.62 6.90 3.45 8.62 18.97
fr
nces of(%)
ue ncyces
q en
e ue
ofs Freqsequ 3.47 0.15 4.07 0.15 7.84 18.25 0.15 0.15 18.55 2.87 3.02 0.15 3.31 2.41 1.66 4.07 3.47 0.15 18.70 18.85 8.90 1.36 1.81 1.66 2.11 0.30 4.37 6.79
er
b
m
nu ofes
n, ernc
affiliatio Numbseque 23 1 1 27 52 121 1 1 123 19 20 1 21 16 11 27 23 1 124 125 59 9 12 11 14 2 29 45
c
mi er
b
o m
on Nu 299 323 300 322 310 306 324 315 327 311 302 303 318 317 316 328 325 320 305 319 307 301 308
x 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
s,ta Acc. KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83 KF83
e
pl
m
a
s
e
er
h 9 9 8 7 4 6 6 9
p 3 0 5 5 3 4 9 2
os 127 646 131 600 262 216 71 333 798 264 156 190 443 43 359 912 73 499 213 452 226 15 227
hiz de MF MF MF MF MF MF MF MF al MF MF MF al MF MF al MF MF MF al MF MF MF MF MF MF MF al
ere. Co OA OA OA OA OA OA OA OA Tot OA OA OA Tot OA OA Tot OA OA OA Tot OA OA OA OA OA OA OA Tot
inolivequenc aation group
taxadetectedrelatedAMFs OTUidentific Phylogenetic Ia Ib IIa IIb IIIa IIIb IVa IVb V VI VIIa VIIb VIIc VIIIa VIIIb IXa
otaest
cs
yo
mcl
erohe ae
GlomTable1.orchards,andt Family Archaeosporaceae Paraglomeraceae Claroideoglomerace (GlomusgroupB) Diversisporaceae (GlomusgroupC) Glomeraceae (GlomusgroupA)
PLOSONE | www.plosone.org 6 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
e
AMMaarj Glomussp.VT94 Nomatch Glomussp.Glo7VT214 FunneliformiscaledoniumVT65 FunneliformismosseaeVT67 SeptoglomusviscosumVT63 Glomussp.VT301 Glomussp.VT301 GlomusAlguacil09cGlo4VT166 GlomusAlguacil09cGlo4VT166 whereOAMFrefersto‘oliv
#)
S
#
p. p. p. p. p. p. F
s s s s s s M
mus mus mus mus mus mus OA
bClosesttaxa Silva108 UnculturedGlo UnculturedGlo UnculturedGlo Glomussp. Glomussp. UnculturedGlo UnculturedGlo UnculturedGlo Glomeromycota Glomeromycota assignedaCode( nasinFigure2.
of%) ywas show
uencyofFrequencyences(%)orchards( 6.90 3.45 8.62 1.72 1.72 10.34 1.72 3.45 3.45 18.97 1.72 18.97 sequencefoundinthestudtheQIIMEsoftware.theMaarjAMdatabaseare
NumberofFreqsequencessequ Acc.Number KF83131281.21 KF83129720.30 KF831321172.56 KF83129640.60 KF83132650.75 KF831314131.96 KF83131310.15 KF83130440.60 50.75 KF831309304.52 KF83129810.15 314.68 yesiananalysisshowninFigure2.EachAMFOTUAMFOTUsderivedfromuclust_refanalysiswithNumericalcodesof‘virtualtaxa’VTasappearin
Cont.Table1. aOTUidentificationFamily PhylogeneticgroupCode IXbOAMF293 IXcOAMF123 IXdOAMF521 XaOAMF102182 XbOAMF77556 XcOAMF311 XIOAMF3 OAMF20 Total XIIOAMF25566 OAMF125 Total aThephylogeneticgroupswerearbitrarilynamedaccordingtotheirpositionintheBa#arbuscularmycorrhizalfungi’andtothenumberassignedtoeachrepresentativebClosesttaxaassignedbyBLASTanalysisusingtheSilva108orMaarjAMdatabase.doi:10.1371/journal.pone.0096397.t001
PLOSONE | www.plosone.org 7 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
PLOSONE | www.plosone.org 8 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
Figure 2. Phylogenetic relationships of nuclear small subunit ribosomal RNA (SSU rRNA) gene sequences of Glomeromycota
referencesequencesderivedfromuclust_refsearchwiththosethatmatchedthesilva_108andMaarjAMdatabases,thereference
AMFdatabasefromRedeckerandRaab[28]andthosereportedinolivefromCalventeetal[7]andpresentintheGenBank.Bayesian
50%majorityruleconsensustreeasinferredfromnSSUrRNAsequencesalignmentsunderthegeneraltimereversible+G+Imodel.Numbersonthe
nodesindicateBayesianposteriorprobabilities(.50%).ThephylogramwasrootedwithParaglomeraceaesequences.Numericalcodesinboldname
eachrepresentativeAMFOTUsfromoliverhizospherederivedfromuclust_refanalysiswiththeQIIMEsoftwareandarelabelled(OAMF#S#)where
OAMFrefersto‘olivearbuscularmycorrhizalfungi’and#tothenumberassignedandgroupsidentifiedandtheremainingcodereferstothesoil
sample.Phylogeneticgroups(ItoXd)werearbitrarilydescribedandareshowninTable1.(*)Althoughthissequencewasoriginallyidentifiedas
belongingtoR.clarusbyCalventeetal.[7]itsclosesttaxonomicaffiliationistoFunneliformissp.andclearlydiffersfromsequencesAJ276084and
AJ852597ofR.clarus.
doi:10.1371/journal.pone.0096397.g002
environmental factors showing a significant (P,0.034) but lower OTUsandtwogroupsofsoilscharacterizedbytwospecificOTUs
effectincludedclay,sandandNcontent,extractablePandannual each (Figure4).
evapotranspitation and minimum temperature of the sampled In our study we were able to identify some agronomic and
locatios (Table 2). environmental gradients driving the AMF community differenti-
Amultivariateregressiontreewasalsocalculatedtosummarize ation; however, we found some differences when using data
the relationships between AMF community composition derived derived from T-RFLP or pyrosequencing analysis. This could be
from pyrosequencing analysis and environmental and agronomic duetotheunequalsamplesizeincludedineachdataset(93vs.58
variableswiththemostinformativevariablesineachsplitshownin olive orchards, respectively). Another factor that is likely to have
Figure 4. The tree explained .30% of the variability in AMF contributedtothosedifferencesistheusageoftwodifferentrRNA
profiles,muchofwhichwereaccountedbythefirstsplitbasedon regions and methods, SSU for pyrosequencing and LSU for T-
exchangeable K and in a lower extend by altitude and sand and RFLP. These regions differ in their phylogenetic resolutions, and
clay content (Figure 4). Then, climatic variables from sampled themethodsofT-RFLPversusfullampliconsequencingdifferas
locations including total rainfall and evapotranspiration (ETP) wellinthisrespectasfoundinpreviousstudies[49].Consequently,
followed by soil pH were the next best predictors forthe second- in our study results from the pyrosequencing analysis should be
order split. Two climatic variables (altitude and average rainfall), interpretedwithcautionduetothesmallerdatasetanalysedand
and nutrient contents of soil samples (including OM, C and the possibility of introducing some bias due to the fact that from
extractablePandtheC:Nratio)alloweddifferentiatingfivegroups some olive orchards we did not amplified any Glomeromycota
ofsoilsthatincludedthreegroupsofsoilsshowinghighrichnessin sequences. Nevertheless, we retrieved consistent results with both
techniques. Thus, soil pH, textural characteristics, nutrient
Figure3.NMDSbiplotofaBray-CurtisdissimilaritymatrixofT-RFLPanalysis.Thefittedvectorsofenvironmentalandphysicochemicalsoil
variablesandtheagronomicvariableageofplantation(indicatedwithdifferentsymbols)mostsignificantlyandstronglyassociated(P,0.05)withthe
ordinationandshowninTable2arealsorepresented.SizeofsymbolsisproportionaltoAMFrichnessinthoseoliveorchards.
doi:10.1371/journal.pone.0096397.g003
PLOSONE | www.plosone.org 9 May2014 | Volume 9 | Issue 5 | e96397
ArbuscularMycorrhizalCommunitiesinOliveOrchards
Table2.Summaryofrelationshipsabetweenagronomic,soilandenvironmentalfactorsandAMFcommunitiesassessedbyT-RFLP
analysis.
Factorsb r2 P
Soilphysicochemicalproperties
Clay(%) 0.1087 0.003996 **
Sand(%) 0.0987 0.007992 **
OrganicC(%) 0.1915 0.000999 ***
OrganicN(%) 0.1006 0.010989 *
ExtractableP(ppm) 0.0735 0.02997 *
ExchangeableK(ppm) 0.0315 0.228771
CEC 0.0005 0.976024
C:Nratio 0.2623 0.000999 ***
pH(KCl) 0.1301 0.000999 ***
SOM(%) 0.1914 0.000999 ***
Climaticcharacteristics
TotalRainfall 0.0183 0.438561
AverageRainfall 0.0045 0.811189
ETP 0.0849 0.026973 *
Tmax 0.0486 0.117882
Tmin 0.0815 0.016983 *
Tmean 0.0412 0.154845
Altitude 0.0274 0.306693
Agronomiccharacteristics
Olivevariety 0.2836 0.000999 ***
Presenceofvegetativecover 0.0032 0.738262
Ageofplantation 0.1233 0.001998 **
Irrigationregimen 0.0401 0.033966 *
Orchardmanagementsystem 0.002 0.986014
aCorrelationswithallenvironmentalvariables(r2)wereobtainedbyfittinglineartrendstotheNMDSordinationobtainedwitheachrestrictionenzymeandsignificance
(P)wasdeterminedbypermutation(nperm =1000).
‘***’=P,0.001;
‘**’=P,0.01;
‘*’=P,0.05.Variableswithhighestsignificantweightareshowninbold.
bOrchardagronomicandclimaticcharacteristics,andsoilphysicochemicalpropertiesareshowninTableS1andsomeofthemwerereportedbefore[5,14].Climatic
variableswereobtainedfromSigMapa,GeographicInformationSystemfromtheSpanishMinistryof‘‘MedioAmbienteyMedioRuralyMarino’’(http://sig.mapa.es/
geoportal/)usingArcGIS10(ESRI,Redlands,California,EE.UU.).
doi:10.1371/journal.pone.0096397.t002
contents, and some climatic variables appear as the most AMFassemblagesamongdifferentagronomicorsoilusepractises
important vectors driving the AMF community differentiation in [20].Allthedataobtainedinthiswork,reinforcetheconceptthat
olive.Soiltexturehastypicallynotbeenidentifiedasbeingofgreat the general AMF assemblage structure and composition in olive
importance on AMF community composition until recently mightbeinfluencedprimarilybysoiltypeandclimateandatless
[43,50]. Landis et al. [51] identified a texture effect in an oak extentbyhostplantfeatures(age,vegetativestages,hostgenotype)
savannah ecosystem, but the effects of texture could not be oragriculturalpracticesasithasbeenshowninotherwoodycrop
separated from plant community composition and soil N. A such as vineyards [20,53]. To our best knowledge this study
separate study [52] did find closely influences of clay, moisture, provides the first evidence of a specific effect of such factors on
and pH in AMF composition in a maize agricultural system in AMF community composition in olive. Further research using a
Zimbabwe,alongwithstrongeffectsofsoilorganicCandtotalN. deeper pyrosequencing effort or more specific primers should be
On the other hand, Moebius-Clune et al. [51] studied AMF conducted to determine how this specific selection of AMF
communities in an assemblage of maize fields across an eastern communitiesbythedifferentolivevarietiesmayberelatedtoolive
NewYorkStatelandscapeandfoundsoiltexturalcomponentsas resiliencetomycorrhizationduringtheolivenurseryprocessorto
themoststronglyrelatedtoAMFcommunitydifferencesfollowed the successful establishment of those mycorrhized planting stocks
bynutrientconcentrations,particularlyMg,whereassoilPorpH, when transplanted to soils in the different biogeographical areas
were lessimportant. (as identified by climatic and soil physicochemical properties)
Theseresultsdemonstratethatwhentherearesmalldifferences present insouthern Spain.
in soil physicochemical characteristics, the composition of the
AMF communities might be similar with an overlapping in the
PLOSONE | www.plosone.org 10 May2014 | Volume 9 | Issue 5 | e96397