the Fungal Symbionts of Bark and Ambrosia Beetles Within Asian Pines

12 Pages·2017·0.97 MB·English

by  James Skelton

MicrobialEcology
https://doi.org/10.1007/s00248-018-1154-8
HOSTMICROBEINTERACTIONS
Detecting Symbioses in Complex Communities: the Fungal Symbionts
of Bark and Ambrosia Beetles Within Asian Pines
JamesSkelton1&MichelleA.Jusino2&YouLi1&CraigBateman3&PhamHongThai4&ChengxuWu5&
DanielL.Lindner2&JiriHulcr1,3
Received:15August2017/Accepted:31January2018
#SpringerScience+BusinessMedia,LLC,partofSpringerNature2018
Abstract
Separatingsymbiosesfromincidentalassociationsisamajorobstacleinsymbiosisresearch.Inthissurveyoffungiassociated
with Asian bark and ambrosia beetles, we used quantitative culture and DNA barcode identification to characterize fungal
communitiesassociatedwithco-infestingbeetlespeciesinpines(Pinus)ofChinaandVietnam.Toquantitativelydiscernlikely
symbioses from coincidental associations, we used multivariate analysis and multilevel pattern analysis (a type of indicator
species analysis). Nearly half of the variation in fungal community composition in beetle galleries and on beetle bodies was
explained by beetle species. We inferred a spectrum of ecological strategies among beetle-associated fungi: from generalist
multispeciesassociatestohighlyspecializedsingle-hostsymbiontsthatwereconsistentlydominantwithinthemycangiaoftheir
hosts.Statisticallysignificantfungalassociatesofambrosiabeetlesweretypicallyonlyfoundwithonebeetlespecies.Incontrast,
bark beetle-associated fungi were often associated with multiple beetle species. Ambrosia beetles and their galleries were
frequently colonized by low-prevalence ambrosia fungi, suggesting that facultative ambrosial associations are commonplace,
and ecological mechanisms such as specialization and competition may be important in these dynamic associations. The
approachusedherecould effectively delimit symbiotic interactionsinany systemwhere symbioses are obscuredbyfrequent
incidental associations. It has multiple advantages including (1) powerful statistical tests for non-random associations among
potentialsymbionts,(2)simultaneousevaluationofmultiple co-occurringhostandsymbiontassociations,and(3)identifying
symbiontsthataresignificantlyassociatedwithmultiplehostspecies.
Keywords Barkbeetles .Microbiome .Mutualism .Ophiostomatales .Platypodinae .Scolytinae
Introduction
Electronicsupplementarymaterial Theonlineversionofthisarticle
(https://doi.org/10.1007/s00248-018-1154-8)containssupplementary
material,whichisavailabletoauthorizedusers. Identifyingsymbiosesunderconditionsofnaturalcomplexity
isdifficult.SimplydefiningthetermBsymbiosis^hasbeena
* JiriHulcr contentious topic since its inception, though the current
[email protected] trend—andourpreference—isforabroaddefinitionsuchas
De Bary’s original usage, which includes interactions with
1 SchoolofForestResourcesandConservation,UniversityofFlorida, positive as well negative outcomes [1–3]. Regardless of the
Gainesville,FL32603,USA
precisedefinitionused,thefirstconditionnecessaryforsym-
2 UnitedStatesForestService,NorthernResearchStation,Centerfor
biosis is that the organisms occur in proximity in nature.
ForestMycologyResearch,OneGiffordPinchotDrive,
Surveys of microbial symbionts often verify this first condi-
Madison,WI53726,USA
tionalone.Here,wegoonestepfurtherandemployempirical
3 DepartmentofEntomologyandNematology,InstituteofFoodand
tests for significant statistical relationships among hosts and
AgriculturalSciences,UniversityofFlorida,Gainesville,FL,USA
potentialsymbionts.Wepositthatspeciesthatareco-adapted
4 VietnamNationalMuseumofNature,VietnamAcademyofScience
and/orco-dependentupononeanotherwillbefoundtogether
andTechnology,CauGiay,Hanoi,Vietnam
more often than expected by chance, given their respective
5 KeyLaboratoryofForestProtectionofStateForestAdministration,
frequencies among samples. The null hypothesis of chance
ResearchInstituteofForestEcology,EnvironmentandProtection,
ChineseAcademyofForestry,Beijing,China associations for observed associations can be tested with
SkeltonJ.etal.
freely available permutation-based statistical software and stage, management regime, land cover, or experimental treat-
quantitative survey data. We demonstrate the value of this ments,butanygroupingschemecanbeusedtoidentifyspecies
approach in a survey of the fungi associated with co- withnon-randomassociationstofactorsofinterest[35,36].We
infesting bark and ambrosia beetles of recently dead pines used MPA to test for non-random associations among host
(Pinus)inAsianforests,thoughthesameapproachcouldbe beetle species and potential fungal symbionts, while simulta-
appliedtoanysystemwheremicrobialsymbiosesarediverse, neously evaluating multiple associations to characterize the
variableinthebreadthofspeciesinvolved,andhiddenamong
breadthofassociationsamongmultiplebeetleandfungaltaxa.
frequentincidentalassociations.
Barkandambrosiabeetles,whichcompriseall~7400spe-
cies of the Curculionidae subfamilies Scolytinae and Methods
Platypodinae[4],haveassociationswithfungithatrangefrom
incidentalcommensalism,toparasitism,toco-evolvedobligate StudySitesandCollectionMethods
nutritional mutualisms [4–7]. Like many symbioses, interac-
tions between beetles and fungi may change with ecological Weconductedreplicatedquantitativeculture-basedsamplingof
context[8–13].ThetermBambrosiabeetle^referstoscolytine beetle bodies and galleries to test for significant associations
andplatypodinebeetlesthatfarmgardensofBambrosia^fungi betweenbeetlespeciesandfungiinAsianpineforestsinTam
within the galleries they excavate through the nutrient-poor DaoNationalPark,HồSơn,VinhPhucProvince,Vietnam,and
xylem or pith of their plant hosts [14]. Ambrosia beetles do Fuzhou, Fujian Province; Guiyang and Weining, Guizhou
noteatwood,butinsteadeatfungieitherexclusivelyornearly Province;andNanshan,ShaanxiProvince,China.Duringtwo
exclusively.Reciprocally,ambrosiafungidependonbeetlesfor collecting trips in March 2016 in Vietnam, we collected bark
dispersal and colonization of new plant tissues [15–17]. The andambrosiabeetlesandtheirgalleriesfromthetrunksofeight
fungalmutualistsaretransportedtonewgardensinsidepouch- dead or moribund pines within 1 km2 in Tam Dao National
or pit-like organs, termed mycangia, which support growing Park. Plantations of Chinese red pine (Pinus massoniana)
ambrosiafungi[17].Theambrosiahabithasevolvedindepen- formedmostoftheforestcanopyatthissite.Duringcollections
dentlyinatleast12lineagesofscolytineandplatypodinebee- fromAugust2015toDecember2015inCentralandSouthern
tles [18, 19] and in at least 9 lineages of fungi [20–24], and China, Xyleborus pinicola and Orthotomicus chaokhao were
thoseestimatesarelikelytoincreaseasmoretaxaarestudied collectedfromChineseredpineinGushanMt.,Fuzhou,Fujian
[4].Manyambrosiabeetlespeciesappeartobeobligatelyas- Province;X.pinicolaandEuwallaceafornicatuswerecollected
sociatedwithoneorafewfungaltaxa[21,24],thoughtheyare fromChineseredpineinHuaxi,Guiyang,GuizhouProvince;
oftenassociatedwithadditionalcommensals[25,26],parasites Polygraphus spp., Tomicus yunnanensis, Tomicus minor, and
[27],andfacultativemutualists[8,15,28,29]. HylurgopslongipilluswerecollectedfromYunnanpine(Pinus
Incontrasttoambrosiabeetleswhichboreintoxylem,bark yunnanensis) in Weining, Bijie, Guizhou Province; and
beetlesconsumerelativelynutrient-richphloemandtypically DendroctonusarmandiwascollectedfromChinesewhitepine
donotdependexclusivelyonfungifornutrition[5].However, (Pinusarmandii)inHuoditang,Nanshan,ShaanxiProvince.
somebarkbeetlespeciesbenefitfrommutualisticnutritional Bark-dwellingbeetleswerelocatedbyremovalofsections
fungi as a key component of their diet [30–33], while other ofbark.Xylem-inhabitingambrosiabeetlesandtheirgalleries
associatedfungimaybeparasitesorcompetitorsofbarkbee- were excised from the trees using chisels. Individual beetles
tlesandtheirmutualists[34].Thus,therelationshipsbetween and sections of their galleries were placed aseptically into
barkbeetlesandfungiareoftenmorevariableanddependent sterile 1.5-ml centrifuge tubes modified with small holes
on context. Fungal mutualisms as well as parasitic associa- throughthecaptopermitgaseousexchange.Tomaintainsta-
tions may be more prevalent among bark beetles than we tisticalindependenceamongreplicatedobservations,onlyone
currently know, but they are difficult to identify because of representativebeetleandgallerysampleweretakenfromex-
inherentvariabilityandtheubiquitouspresenceofincidental tensivegallerieswithmultiplebeetles,andgalleriesweresam-
environmental taxa. Consequently, improved observational pled from multiple trees for each beetle species. Beetles
methods,replicatedsampling,andrigoroustestsofappropri- remainedintheirgalleriesandwerestoredalive,inthedark,
atenullhypothesesareneededtoteaseapartthesecomplexes atambienttemperatureuntilprocessing,whichwasnolonger
andoftenobfuscatedrelationships. than6daysaftercollection.Wealsosampleddispersingbee-
Inthisstudy,weapplymultivariateanalysesandmultilevel tles using bottle traps (design from [37, 38]). Traps were
patternanalysis(MPA)toreplicatedquantitativeculturedatato baited with 95% ethanol and a racemic mixture of alpha-
test for statistically significant associations between co- and beta-pinene (Synergy Semiochemicals Corporation,
occurring beetle and fungal taxa found in Asian pine forests. Burnaby, BC,Canada; itemno. 3076).Ourobjectivewasto
In typical applications of MPA to community data, sampling broadly sample the diversity of fungi and beetles present in
unitsaregroupedbyhabitatcharacteristicssuchassuccessional these pine forests. Therefore, our sampling effort was
DetectingSymbiosesinComplexCommunities:theFungalSymbiontsofBarkandAmbrosiaBeetlesWithinAsian...
designedtocaptureasmuchdiversityaspossiblefrommany sampleforfungalmorphotypeassignmentandcounting.The
pinetreesandtrapsbaitedwithpine-emulatinglures.Sample number ofcolony-forming units was used as a surrogate for
sizes variedasa natural consequence ofvarying abundance/ theoriginalfungalabundance,whichwasestimatedviamul-
detectabilityamongbeetlespeciesatthetimeofsampling.To tiplication of the number of colonies per morphotype by the
avoidconflatingsimilarityinhosttreepreferenceswithsym- inverse of the dilution. Isolates recovered from embedded
biotic affinities between bark beetle and fungal taxa, only woodchipswerecombinedwithmorphotypecountsfromcor-
pine-specific beetle species were included in the presented respondingdilutionplates.Fromeachrepresentativeplate,we
analysis. Consequently, fungi that are prevalent in pine, but selectedanisolateforeachfungalmorphotype.Isolateswere
not biologically dependent on any beetle species, will be then subcultured on a fresh plate and photographed after
foundassociatedwitheachbeetlespeciesataratethatisnot 5–7 days,andanothersubcultureofeachwastransferredtoa
significantlydifferentfromrandom,giventhefrequenciesof sterile 2-mlPDAmini-slantfor transportationtoa Biosafety
eachfungusandbeetleinthedata.Conversely,fungithatare Level2quarantinefacilityattheDivisionofPlantIndustries,
dependentoncertainbeetlespecieswillbefoundinassocia- Florida Department of Agriculture, Gainesville, FL, USA.
tion with those beetles at a rate greater than expected by Isolateswerethenre-platedonPDA,incubatedfor4–7days
chance. Traps were checked at 2–6-day intervals. Beetles to at 25 °C, and compared to photographic and descriptive ac-
besampledforfungiwerestoredliveinsterilemicrocentrifuge countsoftheoriginalisolates.Isolateswerediscardedincases
tubeswithasmallpieceoflightlymoistenedsteriletissuepaper ofmismatchesorcontamination.
priortoprocessing.
DNAExtraction,Amplification,andSequencing
FungalIsolation
Genomic DNAwas extracted from approximately 5 μl of
We used standard sampling techniques for the isolation and myceliaof1–2-week-oldpureculturesin20μlofextraction
quantification of fungal propagules from subcortical beetles solution (Extract-N-Amp; Sigma-Aldrich, St. Louis, MO,
(e.g.,[26,39–41]).Weusednon-selectivemediabecauseour USA)andincubatedat96°Cfor10min.Then,20μlof3%
goalwastocaptureasmuchofthenaturalfungaldiversityas BSAwasaddedandthemixturewasvortexedat~1500rpm
possible.Asterilescalpelwasusedtoasepticallyexcisethree for5sandthencentrifugedat2000rpmfor30s.Thesuper-
small (~1–3 mm) woodchips from the interior surfaces of natant(20μl)wasusedasagenomicDNAtemplateforPCR.
each gallery. Excised chips from each gallery were placed Partialsequences for the 28S largesubunit (LSU)ribosomal
togetherin1.5-mltubeswith500μlofsterile1Xphosphate- DNA(rDNA)wereobtainedfromallisolatesforfungaliden-
buffered saline (PBS) and vortexed at ~1500 rpm for 60 s. tification using the primers LR3 (CCGTGTTTCAAGAC
Gallery samples were then serially diluted to 1:100 and GGG) [42] in the Vietnam study, LR5 (ACCCGCTG
1:1000 concentrations and plated on sterile potato dextrose AACTTAAGC; http://sites.biology.duke.edu/fungi/mycolab/
agar(PDA)media,usingoneplateperdilutionfactor.Chips primers.htm) in the China study, and LROR (ACCCGCTG
werealsoembeddedinPDA.Forbeetlespeciesthatwerenot AACTTAAGC) [42] in both studies. PCR reactions
known to have a mycangium or with a mycangium that is contained the following: 1 μl template DNA, 1 μl forward
widelyopenedtothesurface(e.g.,Dinoplatypusflectus),the and reverse primers (10 mM), 0.125 μl of Taq polymerase
fungi from the beetle body were isolated as a Bbody wash^ (Takara Ex Taq), 2.5 μl PCR buffer (15 mM MgCl ), 2 μl
2
sampleinwhichentirebeetleswerevortexedindividuallyin dNTPs(2.5mMeach),and20μlmoleculargradeDNA-free
500μlofsterile1XPBSandseriallydilutedto1:10and1:100 water. Thermocyclerprofilewasasfollows:3 min at94°C,
concentrations prior to plating on sterile PDA. For beetles followedby34cyclesof94°Cfor45s,55°Cfor45s,and
withoralmycangia(i.e.,XyleborusandEuwallacea),thehead 72 °C for 90 s. LROR was used in sequencing reactions.
wasasepticallyremoved,placedin500μlofsterile1XPBS, Sanger sequencing was performed by the Interdisciplinary
crushed within the 1.5-ml tube using a sterile plastic pestle, CenterforBiotechnologyResearch(ICBR)attheUniversity
vortexed,andseriallydilutedto1:100and1:1000concentra- of Florida. We chose LSU over the more commonly used
tions prior to plating on PDA. Negative controls were pre- internaltranscribedspacer(ITS)barcoderegionbecauseam-
pared simultaneously using the same methods and material, plificationofITSisdifficultandinconsistentformanyofthe
without adding beetle or gallery material. The 1:10 dilution ophiostomatoidfungi known toassociate with barkand am-
platesweremadebyplating50μlofinitial500μlPBSwash. brosiabeetles.
The1:100and1:1000dilutionplatesweremadebytransfer-
ring 50 μl of original wash to 450 μl of fresh sterile PBS, TaxonomicAssignments
vortexing,andplating50and5μl,respectively.
Plates were incubated for 4–7 days in the dark at 25 °C. Sequencechromatogramswereinspectedforqualityandman-
One representative dilution plate was chosen from each uallybinnedintooperationaltaxonomicunits(OTUs)at99%
SkeltonJ.etal.
similarityusingSequencher5.4.5.Ophiostomatoidfungiwere Heatmapswereusedtovisualizethefrequencyandabun-
binnedat100%similaritybecause these taxaoftenhadhigh danceoffungiassociatedwitheachbeetlespecies.Asample-
sequencesimilaritybutdisplayed variation ata few locithat by-fungus matrix was aggregated by mean CFUs for each
was consistent among replicates within collection sources. fungaltaxonperbeetlespeciesandsampletype(i.e.,gallery
Coarse identification of consensus sequences was made to versus beetle). We performed hierarchical clustering of all
the lowest possible taxonomic rank via GenBank National combinations of beetle species and sample types according
Center for Biotechnology Information (NCBI) Basic Local toRaup-Crickdistancesusingthehclust()functioninthestats
Alignment Search Tool (BLAST). We further classified packageofR[47].Thishierarchicalclusteringwasthenused
ophiostomatoid fungi using phylogenetic analysis because to sort the columns of the fungal taxon frequency matrix.
these fungi are known to be widely important to bark and Rows were sorted by their weighted average scores using
ambrosia beetle biology and because this group is not well thefunctionwascores()intheveganpackageofR[49].The
represented or curated in GenBank. Sequences that were sortedmatriceswerethenvisualizedasaheatmapusingcus-
identified within the Ophiostomatales using BLAST were tomRscripts.
aligned and inspected in Geneious 7.1.9. Additional LSU We used MPA to identify associations between beetle and
sequences wereincludedfromrepresentativetaxawithinthe fungaltaxathatoccurredsignificantlymorefrequentlythanex-
Ophiostomatales from a recent phylogenetic study [43]. pectedbychance.Multivariateanalyses(e.g.,PERMANOVA,
Bayesianinference analyseswereperformed using MrBayes ANOSIM,BETADISPER)canbeusedtotestfordifferencesin
3.2.6 [44] on the University of Florida supercomputer symbiontcommunitycomposition,butthesetestsalonedonot
(HiPerGator2.0).Thenucleotidesubstitutionmodelwascho- identify specific significant associations among taxa. For our
sen using the Akaike information criterion from jModelTest purposes,MPAofferedadvantagesoverothercommonlyused
2.1.4[45,46].Tenrunswereconductedwith5Mgenerations analyticalframeworkssuchasANOVAandG-tests(e.g.,[9])or
andsubsamplingevery1000thgeneration.Thetreewasedited multiplegeneralizedlinearmodels(e.g.,Bmvabund^)[52]be-
in FigTree 1.4.3 (http://tree.bio.ed.ac.uk/software/figtree/). causeoflessstringentassumptionsaboutunderlyingdatatype
Isolate names were assigned based on monophyly and (i.e.,occurrenceversusabundance)anddistributions,andMPA
closestrelatives.Forisolatesthatwerelosttocontamination, allowsforasimultaneoustestofassociationsbetweenallcom-
identification was inferred by sequences of representatives binations of taxa with appropriate type I error rates [35].
from the same morphotype (< 10% of isolates). Previous work has also attempted to decompose animal-
Representative sequences for each OTU was uploaded to associated microbial assemblages into Bcore^ and Bsatellite^
GenBank (accession numbers MF543573–MF543598 and communitymembersbasedonspeciesabundancedistributions
LC314669–LC314681;TableS1). [9, 53]. However, this approach requires precise estimates of
species abundances and requires difficult-to-test assumptions
DataAnalysis aboutunderlyingspeciesabundancedistributions[54].
MPA is an expanded form of indicator species analysis
Principal coordinate analysis (PCoA) was used to visualize (ISA). Although earlier versions of ISA had several limita-
differencesinthecompositionofthefungalcommunitiesas- tions such as strict hierarchical groupings of sampling loca-
sociated with each beetle species. PCoAwas implemented tions, issues caused by variation in niche breadth among
usingthecmdscale()functioninthebasepackageofR[47]. species, and significance tests with inflated type I error
TheRaup-Crickmetricwasusedforalldistance-basedanal- [35, 55], MPA offers improvements over traditional ISA
yses.Thismetricismorerobusttovariationinalphadiversity methods bytestingfor significantassociations between each
amongsamplesthanothercommonlyusedincidencemetrics species and all possible combinations (or a user-defined
suchasJaccardandBray-Curtis[48].Distancematriceswere subset)ofsamplinggroupsandthusaccommodatingvariation
generatedfromfungalOTUestimatesforeachuniquecombi- in niche breadth among species and using improved
nationofbeetlespeciesandsampletypeusingtheraupcrick() permutation-based significance tests with appropriate type I
function in the vegan package of R [49]. Permutation-based errorrates[35].Inshort,dataforfungalcommunitiescultured
multivariate analysis of variance (PERMANOVA) [50] was from beetle bodies and galleries were grouped according to
usedtotestformainandinteractiveeffectsofbeetlespecies beetle species. Then, matrix permutations were used to
andsampletype(galleryversusbeetlebody).PERMANOVA identifyfungalspecies thatwereassociatedwitheachbeetle
was implemented using the adonis() function in the vegan species,aswellaswithallcombinationsofbeetlespecies,ata
package of R [49]. Because PERMANOVA is sensitive to frequency greater than predicted by chance, given the
variation in multivariate dispersion among groups, we also frequencyofeachspeciesinthedata[36,56].
conducted a test of homogeneity of multivariate disper- TheVietnamandChinadatasetswereanalyzedseparately
sion [51] using the betadisper() function in the vegan toavoidtheconfoundinginfluenceofsamplingregiononthe
package of R [49]. degreeofassociationbetweentherespectivebeetlesandfungi
DetectingSymbiosesinComplexCommunities:theFungalSymbiontsofBarkandAmbrosiaBeetlesWithinAsian...
sampled. For instance, if the data were combined prior to significant differences in fungal community composition
analysis, a ubiquitous environmental species in one region among beetle taxa (Fig. 1, Table 1). Although there was a
might be found to have a statistically significant association significanteffectofsampletype(galleryversusbeetlebody)
withallbeetlespeciessampledinthatregion.However,such and a significant interactionbetween sample typeand beetle
an association would not reflect any degree of ecological taxa,thoseeffectsweresmallincomparisontothemaineffect
relevanceforbeetlesandfungiotherthangeographiccoinci- of beetle species, which accounted for 41% of the variation
dence.WeconductedMPAusingthemultilevelpatternfunc- among samples. A test for homogeneity of multivariate dis-
tion in the indicspecies package of R (multipatt() function) persion did not detect significant differences in multivariate
[57].Becauseoursamplingeffortvariedacrossbeetlespecies, dispersionamongbeetlespecies(F =0.89,p=0.487),in-
5,65
we used group-equalized indicator value indices (IndVal.g) dicatingthatthe significant effects found byPERMANOVA
[35, 58] and blocked samples by type (gallery versus beetle reflecteddifferencesinfungalcommunitycompositionamong
body)duringpermutationtestsforsignificancetoavoidcon- beetle species and not differences in variability among sam-
flating the effects of sample type (i.e., beetle bodies versus ples(i.e.,multivariatedispersion)withinbeetlespecies.
galleries)withtheeffectsofbeetlespecies[59]. Of the 39 observed unique associations between beetles
andfungalOTUs,weidentified9(23%)associationsassig-
nificantly non-random (Table 2). Each of the beetle species
Results
sampledinVietnamwassignificantlyassociatedwithatleast
onefungalOTU.Themajorityoffungiisolatedfromgalleries
In both the Vietnamese and Chinese datasets, beetle species werealsoisolatedfrombeetlebodysampleswhenbothwere
wasahighlysignificantpredictoroffungalspeciescomposition sampled(seethedendrograminFig.2).Anotableexception
onbeetlebodiesandwithinbeetlegalleries.Despiteconsider- was R. c.f. arxii sp. 1. This OTU was abundant in 15 of 17
able overlap in fungal taxa recovered among beetle species, mycangium samples from X. pinicola (prevalence=0.88,
beetle species explained more than 40% of the variation in meanabundance=2559CFUs),butnotdetectedintheirgal-
fungalcommunitycompositionamongsamplesinbothstudies leries(0of8samples).
(Table 1). Importantly, quantitative sampling methods and
MPAallowedustotestforstatisticallysignificantassociations China
amongmultiplebeetleandfungaltaxaandidentifiedsignificant
associations even when few beetles could be examined We recovered 22 fungal OTUs from 37 beetle and gallery
(Table 2). Bark beetles tended to harbor communities com- samples taken from pines in China. Similar to observations
posed of several fungi in the genera Ophiostoma, Sporothrix, fromVietnam,beetlespeciesexplainedthelargestfractionof
andLeptographium,whereasambrosiabeetleswerefrequently variation in community composition among samples (49%),
associatedwithspeciesinthegenusRaffaeleasensulato.The in addition to weaker, but significant, main and interactive
ambrosia beetle Xyleborus pinicola and the bark beetle effects of sample type (Table 1). There was no significant
Orthotomicus chaokhao were sampled in both countries. difference in multivariate dispersion among beetle species
Xyleborus pinicola was significantly associated only with (F =1.85, p=0.11), indicating that significant effects
7,29
Raffaeleac.f.arxiisp.1inbothcountriesandwithsimilarindi- found by PERMANOVA reflect differences in species com-
catorvalues(IndVal.g=0.78inVietnamandIndVal.g=0.75in positionamongbeetlespecies.AsinVietnam,R.c.f.arxiisp.
China). Conversely, the bark beetle O. chaokhao was signifi- 1 was consistently prevalent in mycangium samples from
cantly associated with two fungi in Vietnam (Ophiostoma ips Chinese X. pinicola (prevalence=0.833, mean abundance=
andSporothrixsp.1)butshowednosignificantassociationsin 7133CFUs),butnotdetectedintheirgalleries(zeroofseven
China. Significant associations between ambrosia beetles and samples).Of28uniqueassociationsobservedbetweenbeetle
ambrosia fungi tended to be exclusive. However, one species and fungaltaxainChina,we identified six (21%) assignifi-
ofambrosiafungi,Raffaeleac.f.fuscasp.1,wasfoundassoci- cant(Fig.3).OftheeightbeetletaxasampledinChina,four
atedwithbothX.pinicolaandD.flectus.Thistaxonwasfound werefound tohavesignificant associationswithatleast one
in5of17(29%)X.pinicolamycangiainVietnam,thoughits fungaltaxon(Table2).
association with X. pinicola was not statistically significant.
Raffaeleac.f.arxiisp.1wasnumericallydominantinthreeof
Discussion
thefourcaseswherebothwerepresent.
Vietnam Ourunderstandingofmutualismshasbeenevolvingfroman
earlierassumptionofrigidpairwiseinteractionstoaparadigm
We recovered 116 fungal isolates representing 26 fungal offlexibleandmorecomplexinteractionnetworksoftencom-
OTUs from 67 beetle and gallery samples. There were posed of generalists and specialist taxa. For instance, Sirex
SkeltonJ.etal.
Table1 Results of
permutation-based multivariate Study Modelterms df Meansquares PseudoF R2 pvalue
analysisofvariancetests
(PERMANOVA)for Vietnam Beetlespecies 5 2.151 9.815 0.410 0.001**
beetle-associatedfungal Sampletype 1 1.240 5.660 0.047 0.001**
communitiesobservedin
Species×type 3 0.580 2.647 0.066 0.002**
VietnamandChina,eachbased
Residuals 57 0.219
on10,000permutationsofthe
Raup-Crickdistancematrix Total 66
China Beetlespecies 7 1.101 4.180 0.491 0.001**
Sampletype 1 0.688 2.799 0.044 0.014*
Species×type 2 0.398 1.634 0.052 0.034*
Residuals 26 0.269
Total 36
Wetestedformainandinteractiveeffectsofbeetlespeciesandsampletype(galleryversusbeetlebody)onfungal
communitycomposition.R2 valuesinitalicshighlightthelargestfractionofexplainedvariationamongsamples
inbothstudies
*α=0.05,significant;**α=0.01,significant
wood wasps were thought to have strictly species-specific pinesinChinaandVietnammirrorthisemergingperspective
nutritional symbioses with Amylosterium fungi, until recent ofambrosiasymbiosesasoftenvariableatthelevelofspecies
invasions of European Sirex led to rampant symbiont associations, while generally conserved at broader phyloge-
swapping with North American wasps [60, 61]. Likewise, neticlevels.Thesefindingsdepictbeetle-associatedfunginot
experimental studies of the camouflage symbiosis between assimplecropsthatarestrictlymanagedbythebeetlefarmers
sepiolid squid and strains of bioluminescent Vibrio bacteria butasdynamiccommunitymemberswithcontext-dependent
have revealed hierarchies of competitive interactions among competitiveabilitiesandavarietyoflifehistorystrategiesfor
symbiotic strains within the light organ [62–64]. A similar garneringbeetle-facilitateddispersal.
trend isunfolding inbarkand ambrosiabeetlesymbiosis re-
search.Improvedmethods,taxonomy,andmountingobserva- AmbrosiaBeetlesandFungi
tionsandexperimentshaverevealedthatthesesymbiosesare
oftendiffuseinteractionsinvolvingmultiplefungiandbeetle Aninferenceofco-evolvedambrosialmutualismsistypically
species [4, 9, 15, 24]. Beetles can thrive while consuming consideredjustifiediftwoconditionsaremet:highprevalence
alternativefungalfoodsources[15,29],andcomparativephy- andabundanceofviablesporesfromafungalspecieswithin
logenetic analysessuggestfrequentsymbiontswappingover the mycangium of a single beetle species, and consistency
evolutionary timescales [65]. The results of our surveys in acrossthatbeetle’s geographicrange[9,23, 24, 26, 41].We
Table2 Resultsofmultilevelpatternanalysis(MPA),atypeofindicatorspeciesanalyses,offungalassociationsobservedinpineforestsofVietnam
andChina
Study Beetletaxa Fungaltaxa Freq.ingalleries(%) Freq.onbeetles(%) IndVal.g pvalue
Vietnam Cyrtogeniusluteus Ophiostomatalessp. 12.5 30.8 0.58 0.007
Crypturgussp. Nectriaceaesp.1 n.a. 40.0 0.63 0.014
Dinoplatypusflectus Raffaeleac.f.fuscasp.2 11.1 28.6 0.50 0.024
Euwallaceac.f.andamanensis Sporothrixsp.2 80 50.0 0.92 0.001
Xyleboruspinicola Raffaeleac.f.arxiisp.1 0 87.5 0.78 0.014
Cyrtogeniusluteus+Orthotomicussp. Ophiostomaips 25.0 50.0 0.72 0.004
Cyrtogeniusluteus+Orthotomicussp. Sporothrixsp.1 37.5 25.0 0.62 0.021
China Dendroctonusarmandi Ophiostomasp.1 33.3 60.0 1.0 0.001
Tomicusyunnanensis Sporothrixsp.3 n.a. 80.0 0.89 0.005
Xyleboruspinicola Raffaeleac.f.arxiisp.1 0 83.3 0.75 0.015
Hylurgopslongipillus+Tomicusminor Ophiostomasp.3 100 75 0.82 0.003
Only statistically significant associations are shown. Degree of association was determined by the group-equalized indicator statistic (IndVal.g).
Significanceofeachassociationwasassessedbythepermutationtestwith10,000matrixpermutationsforeachputativeassociation(α=0.05)
n.a.notavailable
DetectingSymbiosesinComplexCommunities:theFungalSymbiontsofBarkandAmbrosiaBeetlesWithinAsian...
China Vietnam
1.0 DEuewndarlloaccteoan usps .armandi 1.0 CCyryrptotugregnuius ss plu.teus
Hylastini sp Dinoplatypus flectus
Orthotomicus chaokhao Euwallacea sp.
Polygraphus sp. Orthotomicus chaokhao
Tomicus minor Xyleborus pinicola
Tomicus yunnanensis
Xyleborus pinicola
5 5
0. 0.
%) %)
Co2 (10 0.0 Co2 (7.7 0.0
P P
5 5
0. 0.
− −
0 0
1. 1.
− −
−0.6 −0.4 −0.2 0.0 0.2 0.4 0.6 −0.6 −0.4 −0.2 0.0 0.2 0.4 0.6
PCo1 (19%) PCo1 (8.3%)
Fig.1 Principalcoordinateanalyses(PCoA)showingtheeffectofbeetle community composition, sample clouds for each species overlapped
speciesonfungalcommunitycompositionobservedfrombeetlebodies considerably, emphasizing the need for replicated sampling to
(blacksymbols)andbeetlegalleries(graysymbols)inapineforestin accurately characterize fungal taxa specifically associated with each
ChinaandVietnam.Overlappingpointsareslightlyjitteredforclarity. beetle taxon. Ordination was performed on a Raup-Crick distance
Although beetle species had a highly significant effect on fungal matrixofCFUestimates
Fig.2 Heatmapillustrating
associationsbetweenfungiand
beetletaxaobservedinVietnam. Fungi
Columnsrepresentall
Ceratocys(cid:2)opsissp.
combinationsofbeetlespeciesand
Hypocrealessp.
sampletypesthatwerecollected. Raffaeleac.f. amasae
Numbersinparenthesesindicate Raffaeleac.f. fuscasp. 2
Raffaeleac.f. arxiisp. 2
samplesizes.Dendrogramatthe
Exophialasp.
topshowsresultsofhierarchical Raffaeleac.f. arxiisp. 3
clusteringoffungalcommunities Ophiostoma ips
associatedwitheachbeetlespecies/ Ophiostomatalessp.
Sporothrixsp. 1
sampletypecombination.Rows
Leptographiumsp.
representfungaltaxaandare Nectriaceaesp. 2
arrangedbyweightedaverage Raffaeleac.f. fuscasp. 1
Fusarium solani
scoresobtainedfromhierarchical
Geosmithiasp.
clustering.Cellcolorrepresents Orbiliaceasp.
themeanlog-transformed Raffaeleac.f. arxiisp. 1
colony-formingunits;darker Veronaeacompacta
Xeonoacremoniumsp.
colorsaremoreabundant.Black
Fusariumsp.
outlinesindicatestatistically Graphiumsp.
significantassociationsidentified Schizopylumsp.
byindicatorspeciesanalysis Geotrichumsp.
Nectriaceaesp. 1
Pestalo(cid:2)opsissp.
Sporothrixsp. 2
SkeltonJ.etal.
Fungi
Sporothrixsp. 3
Ophiostomasp. 3
Ophiostomatalessp.
Ophiostomasp. 1
Fusarium sp.
Raffaelea c.f. subfusca
Fungi sp
Sarocladiumstrictum
Leptographiumsp
Ophiostomaips
Hypocrealessp
Raffaelea c.f.arxiisp1
Ulocladiumsp
Ophiostomasp. 2
Fig.3 Heatmapillustratingassociationsbetweenfungiandbeetletaxa averagescoresobtainedfromhierarchicalclustering.Cellcolorrepresents
observedinChina.Columnsrepresentallcombinationsofbeetlespecies thefrequencyofeachassociation;darkercolorsaremorefrequent.Black
and sample types. Numbers in parentheses indicate sample sizes. outlines indicate statistically significant associations according to
Dendrogramatthetopshowsresultsofhierarchicalclusteringoffungal indicator species analysis. BFungi sp.^ isa distinct, albeit unidentified
communities associated with each beetle species/sample type strain,notanaggregategroup
combination.Rowsrepresentfungaltaxaandarearrangedbyweighted
recovered high CFU estimates ofR.c.f. arxii sp. 1 from the show high prevalence throughout the ranges of their beetle
mycangiaofX.pinicolainbothChinaandVietnam,andthis associates, many more occur at lower frequencies that vary
funguswasnotobservedinassociationwithanyotherbeetle across the geographic range of beetle species. For instance,
species. Consequently, R. c.f. arxii sp. 1 is inferred to be a many species of Raffaelea and related fungi are associated
specialized, prevalent, and dominant mycangial symbiont of with the redbay ambrosia beetle (Xyleborus glabratus) in
X.pinicola.Surprisingly,R.c.f.arxiisp.1wasnotrecovered North America, often co-occurring within the mycangium
fromthe galleries ofX. pinicolaineitherChinaorVietnam. [39, 40] and showing variable prevalence across broad geo-
FourrelatedRaffaeleaspecieswererecoveredfromgalleries graphicspatialscales[8].WefoundR.c.f.fuscasp.1in29%
of X. pinicola and D. flectus; thus, it is not likely that the of X. pinicola mycangia in Vietnam, where it was the only
absence of this fungus in gallery samples is due to our ambrosia fungus recovered from two individuals and co-
methods. Instead, it is possible that R. c. f. arxii was absent occurredwithR.arxiisp.1inthemycangiaoffourindivid-
inaparticularperiodofthenaturalcommunitysuccessionthat uals.R.c.f.fuscasp.1wasnotrecoveredfromX.pinicolain
takes place in ambrosial gardens, during which mycangial China, suggesting that this fungus is a geographically local-
symbionts are overrun by secondary colonists [29, 66]. It is ized associate of X. pinicola. This species may represent a
alsopossiblethatthemostabundantandprevalentmycangial facultativelocalmutualist,aninnocuouscommensal,orafun-
symbiontisonlyonememberofacommunityoffungiintro- gal antagonist. Similarly, Dinoplatypus flectus was found
ducedandconsumedbythebeetle,andwhileitisacompet- associated with four Raffaelea species (three Raffaelea spp.
itive dominant in the mycangium, it is less competitive in s. s. and Raffaelea amasae from the BEstaya/Dryadomyces
wood.Ifthisscenarioistrue,suchfungithatarecompetitive group^)[22],thoughnoneofthesefungiwerehighlypreva-
in the mycangium but unimportant to beetle nutrition in the lentacrossindividuals.Theonlystatisticallysignificantasso-
gallerycouldrepresentparasiticcheaters. ciate of this beetle was R. c.f. fusca sp. 2, which was only
Maintainingadiversityoffungalassociationscouldbuffer recoveredfrom33%ofD.flectusbodysamplesand 11%of
ambrosiabeetlepopulationsfromchangesinlocalconditions gallerysamples,andnotfoundwithanyotherbeetlespecies.
or available host plants. While some ambrosial symbionts These observations suggest that D. flectus and, possibly,
DetectingSymbiosesinComplexCommunities:theFungalSymbiontsofBarkandAmbrosiaBeetlesWithinAsian...
X.pinicoladonotdependonasinglenutritionalsymbiontbut of beetles, each of which solely relies on this one fungus
insteadhavemultiplesymbionts.Someevidencesuggeststhat [69]. Resolving the ecological causes and evolutionary
interactionswithmultiplefunctionallydiversefungallineages consequencesofthesedifferingasymmetriesamongambrosia
minimizetheriskofDendroctonusbeetleslosingthebenefits symbioses remains an unexplored and likely fruitful avenue
ofsymbioticfungalpartnersoverarangeofvariableenviron- forfutureresearch.
mental conditions [67]. However, the ecological and evolu-
tionary consequences of having multiple fungal symbionts BarkBeetlesandFungi
remaincriticallyunderstudiedandpoorlyunderstood.
One unexpected association was the significant associ- Fidelity among phloem-feeding bark beetles and associated
ation between Euwallacea c.f. andamanensis and the fungi is even less settled [70]. Our results suggest that bark
ophiostomatoid fungus Sporothrix sp. 2, recovered from beetle-associatedfungiarelessrestrictedintheirassociations
multiple beetles and galleries, from multiple trees in than ambrosia beetle associates occurring in the same trees.
Vietnam. Euwallacea farm many Fusarium lineages that Whileallsignificantassociatesofambrosiafungiwererecov-
together form the monophyletic Bambrosial Fusarium eredonlyfromonebeetlespecies,severalfungalassociatesof
clade^ (AFC) within the Fusarium solani species com- bark beetles were typically found with multiple beetles, and
plex, sometimes co-occurring with Raffaelea species threeofsevensignificantbarkbeetleassociatesweresignifi-
[21]. Fungus-feeding bark and ambrosia beetle species cantly associated with multiple beetle species. None of the
that specialize on coniferous hosts are frequently associ- bark beetles sampled in this study are known to possess a
ated with ophiostomatoid fungi [5, 6], suggesting that mycangium, suggesting that transmission via external body
these fungi are particularly strong competitors within co- surfaces or within the gut is less selective. Another possible
nifers [5]. Perhaps Euwallacea form facultative fungal explanation for greater promiscuity among bark beetle-
associations that are contingent on the host tree, favoring associated fungi is simply the physical qualities of phloem,
ophiostomatoid fungi when colonizing coniferous hosts. which,unlikexylem,is(1)restrictedtoathinplane,(2)more
Recent comparative phylogenetic work suggests that accessible to diverse taxa, and (3) relatively high in nutrient
Euwallacea species have often switched nutritional sym- content. These qualities encourage higher beetle densities
bionts within the AFC [65], and other studies have found within the phloem and intermingling of galleries belonging
the ophiostomatoid genus Raffaelea in association with todifferentspecieswhichcouldleadtocommonplacefungal
Euwallacea. Thus, a switch to symbionts outside of the exchangesamongbarkbeetles.
AFC seems at least plausible.
Partnerfidelity,orone-for-onespeciesmatchingofmutu- MethodologicalConsiderations
alists(oftenreferredtoasBspecificity^),isacommonthemein
studiesofambrosiabeetlesandtheirfungi(e.g.,[4,24,68]). Objectively weighing the evidence for putative symbiotic
Studies including this one show that fidelity between associations is increasingly essential in the inference of
ambrosiabeetlesandfungiisasymmetrical,butinterestingly, symbiotic relationships, especially as powerful and cost-
differentambrosialsystemsappeartodisplaydifferentdirec- effective new techniques such as high-throughput meta-
tionsinasymmetry.Inthisstudy,ambrosiabeetleswereoften barcoded amplicon sequencing, metagenomics, and
foundwithmultiplespeciesofRaffaelea,whileeachambrosia metatranscriptomics, provide researchers with richer but
fungus was typically only found associated with one beetle potentially noisier datasets [9, 12, 13, 71]. Several factors
species. Four of the five Raffaelea OTUs recovered in the arelikelytointroducenoiseintomicrobialsymbiontcommu-
VietnamstudyandthesingleRaffaeleaOTUrecoveredfrom nity data including (1) symbiotic taxa that are not always
the China study were found associated with only one beetle present or detected, (2) ubiquitous environmental taxa that
species, despite the mere centimeters that separated the are introduced by stochastic mechanisms (e.g., wind-blown
galleriesofdifferentco-infestingambrosiabeetles.Similarly, spores,incidentalphoresy),and(3)non-indigenoussymbionts
the onlysignificantassociateofEuwallacea inVietnamwas thatmayinvadefromgalleriesofco-infestingbeetlespecies.
notrecoveredfromanyotherbeetle.Whileitispossiblethat Our results show that replicated sampling and permutation-
more extensive sampling could uncover shared ambrosia basedstatisticalanalysescanrapidlyandobjectivelyconfirm
fungi among beetle species, our results indicate that shared expected associations and identify unexpected associations
ambrosia fungi are not prevalent in this system. In contrast between hosts and microbial symbionts amidst a cacophony
to this study, fungi in other ambrosia systems are known to ofincidentalinterlopers.
utilizemanybeetlespecieswhilethebeetlesaretypicallyre- Identification of biologically significant associations
strained to one or a few fungal species. Examples include amongmanyincidentalassociationscanbeobscuredbynat-
members of the AFC [21] and the extreme Flavodon ural variation among fungal taxa in propagule production,
ambrosius (Basidiomycota) which colonizes many species amenabilitytocultureonmedia,andcompetitiveabilityunder
SkeltonJ.etal.
cultureconditions[5,39].Thisvariabilitymakesassessments Acknowledgements We thank Tran Thi Men for the assistance in the
field.WethankChingShanLinforthehelpwithbeetleidentification.
ofsignificancebasedsolelyonnumericaldominanceamong
WethankZhenZhang,ZaihuaYang,andZaihuaYangfromtheGuizhou
taxa within mixed samples tenuous. However, carefully
AcademyofForestry,Guiyang,fortheaidinChinabeetlecollections.
selectedanalyticalframeworkscanmediatebiasesassociated
with variation among taxa. Rather than the relative abun- FundingInformation ThisworkwasfinanciallysupportedbytheUSDA
Forest Service (USDA Farm Bill section 100007), the National
dances of multiple taxa within samples, our analysis using
Science Foundation, and the Florida Department of Agriculture,
MPAemphasizedtheabundanceandfrequencyofeachtaxon
Division of Plant Industry.
across replicates within groups of samples, independent of
otherfungaltaxa.Becausethestatisticalsignificanceofeach
associationisbeingassessedindependentofotherfungaltaxa,
References
theeffectsofrelativelylowpropaguleabundanceand/orpoor
cultural characteristics on the determination of an
association’sstatisticalsignificancearemitigated. 1. SaffoMB(1992)Comingtotermswithafield:wordsandconcepts
insymbiosis.Symbiosis14:17–31
ApplyingMPAtoquantitativeculture-baseddataprovided
2. MartinBD,SchwabE(2013)Currentusageofsymbiosisandas-
reasonablestatisticalpowerevenwithmoderatesamplesizes. sociatedterminology.InternationalJournalofBiology5:32–45
We were able to identify statistically significant associations 3. DeBaryA(1879)Thephenomenonofsymbiosis.KarlJTrubner,
withasfewasthreereplicatesinsomeinstances,thoughmore Strasbourg
4. HulcrJ,StelinskiL(2017)Theambrosiasymbiosis:fromevolu-
replicates would certainly strengthen these inferences.
tionary ecology to practical management. Annual Reviews of
Second,mostofthesignificantassociationsidentifiedinthis Entomology62:285–303
studyarecorroboratedbypreviousrecordsforrelatedbeetle 5. HarringtonTC(2005)Ecologyandevolutionofmycophagousbark
species. For instance, we found significant associations of beetlesandtheirfungalpartners.In:VegaFE,BlackwellM(eds)
X. pinicola and the platypodine D. flectus with two Ecologicalandevolutionaryadvancesininsect-fungalassociations.
OxfordUniversityPress,Oxford,pp257–291
Raffaelea species. Raffaelea species have previously been
6. SixDL(2012)Ecologicalandevolutionarydeterminantsofbark
foundassociatedwithotherXyleborusspecies[8,22,72],as beetle—fungussymbioses.Insects3:339–366
well as platypodines [73, 74]. Lastly, we consistently found 7. BeaverR(1989)Insect-fungusrelationshipsinthebarkandambro-
the same species, Raffaelea c.f. arxii sp. 1, as significantly siabeetles.In:WildingN,CollinsN,HammondP,WebberJ(eds)
associatedwithX.pinicolainVietnamandChina,recovering Insect-fungusinteractions.Academic,SanDiego,pp121–143
8. CampbellAS,PloetzRC,DreadenT,KendraP,MontgomeryW
nearly identical indicator values in both studies. Thus, even
(2016) Geographic variation in mycangial communities of
modestly replicated quantitative sampling, when combined Xyleborus glabratus. Mycologia 108:657–667. https://doi.org/
with appropriate statistical tests, can consistently reveal 10.3852/15-133
multiple associations of likely biological significance within 9. KostovcikM,BatemanCC,KolarikM,StelinskiLL,JordalBH,
HulcrJ(2015)Theambrosiasymbiosisisspecificinsomespecies
complexcommunitiesofmicrobesandinsects.
and promiscuous in others: evidence from community pyrose-
Thereareimportantlimitationstosurvey-basedsymbiosis quencing.TheISMEJournal9:126–138
studies such as the one reported here. Surveys alone cannot 10. Evans L,HofstetterR, Ayres M,Klepzig K (2011) Temperature
demonstratebiologicaloutcomessuchastheeffects(positive alterstherelativeabundanceandpopulationgrowthratesofspecies
or negative) of taxa on one another or if co-evolutionary within the Dendroctonus frontalis (Coleoptera: Curculionidae)
community.Environ.Entomol.40:824–834
histories exist between hosts and symbionts. Unknown
11. KlepzigKD,HofstetterRW(2011)Fromattacktoemergence:in-
exogenousfactorssuchassharedmicrohabitat,hostplantpref-
teractionsbetweensouthernpinebeetle,mites,microbes,andtrees.
erences, or synchronous successional changes in beetle and In:Coulson,RN,Klepzig,KD(eds.)SouthernPineBeetleIIGen
fungalcommunitieswithindeadtreescouldleadtostatistically TechRepU.S.DepartmentofAgricultureForestService,Southern
ResearchStation,pp.141–152
significant associations in the absence of truly symbiotic
12. Miller KE, Hopkins K, Inward DJ, Vogler AP (2016)
interactions. Thus, experimentation and phylogenetic recon-
Metabarcodingoffungalcommunitiesassociatedwithbarkbee-
structionsarerequiredtodemonstratewhattypeofecological tles.EcologyandEvolution6:1590–1600.https://doi.org/10.1002/
interactionsoccursandwhathasinfluencedtheirevolutionary ece3.1925
trajectories. Furthermore, relationships between animals and 13. DohetL,GregoireJ-C,BerasateguiA,KaltenpothM,Biedermann
PH (2016) Bacterial and fungal symbionts of parasitic
microbialsymbiontsmaychangerapidlywithcontextsuchas
Dendroctonus bark beetles. FEMS Microbiol. Ecol. 92.
season,treespecies,populationphase,lifehistorystage,pres-
https://doi.org/10.1093/femsec/fiw129
ence/absence of other fungi or animal associates [10, 29, 14. Schmidberger J (1836)Naturgeschichte desapfelborkenkä Apate
75–77], and even the sampling method [26]. Consequently, dispar.BeiträgezurObstbaumzuchtundzurNaturgeschictederden
ObstbäumenschädlichenInsekten4:213–230
once a putative association is discovered, extensive targeted
15. BatraLR(1966)Ambrosiafungi:extentofspecificitytoambrosia
sampling of specific host and symbiont species, followed by
beetles.Science153:193–195
careful analyses, environmental factors may reveal or explain
16. SixDL(2003)Barkbeetle-fungussymbioses. Insect Symbiosis
additionalandimportantlayersofcomplexity. 1:97–114