Dispersal limitation and the assembly of soil Actinobacteria communities in a longterm ...

12 Pages·2013·0.28 MB·English

Dispersal limitation and the assembly of soil Actinobacteria
communities in a long-term chronosequence
SarahD.Eisenlord1,DonaldR.Zak1,2 &RimaA.Upchurch1
1SchoolofNaturalResourcesandEnvironment,UniversityofMichigan,AnnArbor,MI48109
2EcologyandEvolutionaryBiology,UniversityofMichigan,AnnArbor,MI48109
Keywords Abstract
Actinobacteria,chronosequence,microbial
biogeography. It is uncertain whether the same ecological forces that structure plant and ani-
mal communities also shape microbial communities, especially those residing in
Correspondence soil. We sought to uncover the relative importance of present-day environmen-
SarahD.Eisenlord,SchoolofNatural tal characteristics,climatic variation, and historical contingencies in shaping soil
ResourcesandEnvironment,Universityof actinobacterialcommunitiesinalong-termchronosequence.Actinobacteriacom-
Michigan,2560DanaBuilding,440Church
munitieswerecharacterizedinsurfacesoilsamplesfromfourreplicateforeststands
Street,AnnArbor,MI48109.
withnearlyidenticaledaphicandecologicalproperties,whichrangefrom9500to
Tel:[phone];
14,000 years following glacial retreat in Michigan. Terminal restriction fragment
Fax:[phone];
E-mail:[email protected] length polymorphism (TRFLP) profiles and clone libraries of the actinobacterial
16SrRNAgenewereconstructedineachsiteforpheneticandphylogeneticanal-
FundedbygrantsfromtheNationalScience ysis to determine whether dispersal limitation occurred following glacial retreat,
FoundationandtheU.S.Departmentof orifcommunitycompositionwasdeterminedbyenvironmentalheterogeneity.At
EnergyOfficeofBiologicalandEnvironmental
everylevelofexamination,actinobacterialcommunitycompositionmostclosely
Research.
correlated with distance, a surrogate for time, than with biogeochemical, plant
community,orclimaticcharacteristics.DespitecorrelationwithleaflitterC:Nand
Received:13December2011;Accepted:28
December2011 annualtemperature,thesignificantandconsistentrelationshipofbiologicalcom-
munitieswithtimesinceglacialretreatprovidesevidencethatdispersallimitationis
EcologyandEvolution2012;2(3):538–549 anecologicalforcestructuringactinobacterialcommunitiesinsoiloverlongperiods
oftime.
doi:10.1002/ece3.210
ever, evidence is accumulating that some microorganisms
Introduction
do exhibit biogeographical patterns across time and space
Biogeographyisthestudyofgeographicaldistributionofor- (Fulthorpeetal.1998;ChoandTiedje2000;Whitakeretal.
ganisms over the Earth in both time and space. Ecologists 2003).Itiscurrentlyunderdebatewhethervariationinmi-
seektounderstandhowbiologicaldiversityisgeneratedand crobialcommunitiesoverspaceresultsfromenvironmental
maintained, especially in the light of a changing environ- filtering,orifgeographicbarriersandotherhistoricalcontin-
ment.Formicrobialbiogeography,thetraditionalviewhas gencies contribute to spatial structure in community com-
held that “Everything is everywhere, but the environment position through limiting dispersal (Horner–Devine et al.
selects”(BaasBecking1934).Thelargepopulationsizeand 2004;Martinyetal.2006;RametteandTiedje2007a;Geetal.
shortgenerationtimestypicalofmicrobialcommunitieslead 2008). If not all microbes are equally and evenly dispersed
torapidgeneticdivergence,potentiallyresultinginbiogeo- over time, it would suggest that forces structuring micro-
graphic patterns (Green and Bohannan 2006). However, it bialcommunitiesaremorecomplexthanadaptiveevolution
has been assumed that unlimited microbial dispersal leads throughnaturalselection.Historicalcontingenciescouldgive
toconstantinputofnewmembers,increasinggeneflowand risetocompositionalpatternsthroughisolationandgenetic
overwhelmingtheforcesofgeneticdrift(RobertsandCohan divergence.
1995;RametteandTiedje2007b).Globalstudiesofmicrobial We address this issue by examining the community pat-
diversityinaquaticandsoilcommunitiessupportthistheory terns of a deeply diverse and divergent phylum, the Acti-
(FiererandJackson2006;VanderGuchtetal.2007);how- nobacteria,inanorthernhardwoodforestchronosequence.
538 (cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd.
ThisisanopenaccessarticleunderthetermsoftheCreativeCommonsAttribution-NonCommercialLicense,whichpermits
use,distributionandreproductioninanymedium,providedtheoriginalworkisproperlycitedandisnotusedforcommercial
purposes.
S.D.Eisenlordetal. ActinobacterialPhylogeography
Actinobacteriaareimportantorganismsmediatingplantlit- beaphylogeneticsubsetofoldercommunitiesinthesouth.
ter decay and the subsequent formation of soil organic Moreover,ifdispersallimitationisnotafactorshapingthese
matter in terrestrial ecosystems (Paul and Clark 1996; communities as the Baas-Becking theory predicts, then we
DeAngelis et al. 2011). This phylum is phylogenetically di- would expect similar communities in all sites. This alter-
vergent and the closest prokaryotic relative has yet to be nativepredictsthatvariationinactinobacterialcommunity
identified (Embley and Stackebrandt 1994; Ventura et al. compositionshouldbestructuredbyenvironmentalfactors
2007).Actinobacteriaexpressavarietyofmorphologiesand suchasoverstoryplantcommunitycompositionandbiogeo-
life-historytraits,includingsporulation,whichcouldbead- chemicalcharacteristicsofthesoil.
vantageousforlong-distancedispersal.Thereisnoconsen- To test these alternatives, we initially characterized acti-
suswhetherActinobacteriaexhibitendemismorhaveacos- nobacterialcommunitiesusing16SrRNAgeneterminalre-
mopolitandistribution(Gløckneretal.2000;Wawriketal. striction fragment length polymorphism (TRFLP) finger-
2007). Here, we evaluate whether dispersal limitation is a prints. Using this information, we further refined the test
factorstructuringthecommunityofsoilActinobacteriafol- of our hypothesis via cloning and sequencing of the acti-
lowing glacial retreat in a present-day forest ecosystem in nobacterial16SrRNAgeneandsubsequenttaxonomicand
northeasternNorthAmerica. phylogeneticanalysis.Actinobacterialcommunitycomposi-
Previousworkprovidesevidencethatsoilactinobacterial tioninourfourstudysiteswasassessedbyexaminingcom-
communities exhibit regional biogeography, wherein com- munitysimilarity,identifyingadistance–decayrelationship,
munitymembershipchangesacrossthenorth–southdistri- andtestingtherelatednessofcommunitypatternstoenviron-
butionofanorthernhardwoodecosystemintheUpperGreat mentalvariation,climaticfactors,andgeographicdistance,
Lakes region of the U.S. (Eisenlord and Zak 2010). Across asaproxyforsiteage,throughmultivariatestatistics.Here,
thisgeographicregion,theperiodicretreatofglaciationca. weprovideevidencethatdispersallimitationisamechanism
14,000yearsagooccurredinasouthtonorthdirection.Over shapingActinobacteriacommunitiesinanorthernhardwood
aperiodof5000years,newlandscapeswererevealedforming forest ecosystem over a relatively long-time frame (i.e., ca.
achronosequence,inwhichsoilswereformedfromsimilar 5000years).
parent material, yet differ in time since deglaciation set in
motion the process of soil formation. According to pollen
records,forestsdominatedbyAcersaccharumMarsh.(sugar Methods
maple)establishedatthebeginningoftheHoloceneinthe
Studysitesandsampling
UpperLakeStatesregion.Thesepollenrecordsindicatesugar
maple became dominant ca. 4000 years following the re- ThebiogeographyofActinobacteriawasexaminedinthesur-
treatofglacialice(Davis1983),leavingbehindalong-term facesoiloffoursugarmapledominatedforestsontheLower
chronosequence. andUpperPeninsulaofMichigan(Fig.1).Thesesiteswere
Alongthischronosequence,wepreviouslylocatedecolog- selected from 31 candidate sites based on their ecological
ically and edaphically matched sugar maple stands, which andedaphicsimilarity,whichwereassessedbymultivariate
providesauniqueopportunitytostudythestructuringforce analyses of plant community composition, stand age, and
oftimeontheassemblyofsoilmicrobialcommunities.Repli- soil properties (Burton et al. 1991). Soils are well-drained
catesamplingofActinobacteriacommunitieswithinthesame sandy,typichaplorthodoftheKalkaskaseriesandoverstory
habitattypeinfourdifferentgeographiclocationsallowsus biomass is dominated by sugar maple (∼70–85%). These
to determine if there is a “distance effect” (Martiny et al. sites form a long-term chronosequence due to their simi-
2006). Because each geographic location corresponds with larity of environmental, ecological, and edaphic character-
timeelapsedfollowingglacialretreat,weconsidereddistance istics,yetthousandsofyearselapsedfollowingdeglaciation
tobeasurrogatefortime.Duetotheperiodicnatureofglacial andestablishmentofforestsateachsite.Thesouthernmost
retreat,distanceandtimedonotfollowalinearrelationship. siteDwasice-freeapproximately13,500yearsbeforepresent
Ifdispersallimitationwasaforcestructuringsoilmicrobial (BP)followedbymapleforestestablishment3500yearslater
communitiesoverlongtimeframes,dispersalofactinobac- (Evenson et al. 1976; Davis 1983; Drexler et al. 1983). Site
terial propagules would be limited in the more northern Cislocated83kmnorthofsiteDandwasdeglaciatedap-
sites because they are the youngest. Therefore, differences proximately13,000yearsBP,followedbymapleforestestab-
in community composition should correlate with distance, lishment4000years later(Evenson et al. 1976; Davis1983;
after controlling for present day environmental variability. Drexleretal.1983).SiteB,located150kmnorthofSiteC,
Furthermore, if the source of actinobacterial communities was uncovered approximately 11,000 years BP, and pollen
originatedfromtheoldersites,thenthedistributionofthese recordsindicatemapleforestestablishment4000yearslater.
communitiesshouldbeclusteredonaphylogenetictree.That Finally, the northernmost site A, 343 km northwest of site
is,youngeractinobacterialcommunitiesinthenorthshould B,wasice-free9500yearsBP,withmapleforestestablishing
(cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd. 539
ActinobacterialPhylogeography S.D.Eisenlordetal.
in each plot were composited and passed through a 2-mm
sieve in the field. From the sieved composite sample, a 5-
g subsample was removed for DNA extraction. By pooling
the 10 soil cores, our sampling scheme aggregated small-
scalespatialheterogeneityatthescaleofindividualplots.We
did so because our goal was to characterize the actinobac-
terialcommunityatthescaleofentireforeststands,andto
exploreregionaltrendsincommunitysimilaritythatmany
bestructuredbyhistoricalcontingencesandenvironmental
factors.
Samples were placed on ice in DNA extraction vials and
immediatelytransporttotheUniversityofMichigan,where
theywereheldat–80◦CpriortoDNAextraction.InMayof
2007,SiteBwasdefoliatedbythecanopyconsuminginsect,
Operophterabruceata,whichdepositedlargeamountsofin-
sectfrassandgreen-leaffragmentsontheforestfloor(D.R.
Zak, pers. obs.). Because these insects dramatically altered
thebiochemicalconstituents,theamount,andtimingofleaf
litterfall,weeliminatedtheMay2007SiteBsamplesfrom
ouranalyses.
Figure1. Forestsitescomposingalong-termchronosequencesfollow-
ingglacialretreat.SouthernmostSiteDistheoldest,whereasnorthern-
mostSiteAistheyoungest.Detailsregardingsiteageandenvironmental
characteristicscanbefoundinthetext,alongwithourmethodsforiden- DNAextractionandpolymerasechain
tifyingecologicallysimilarsitesacrossthisregion.
reaction(PCR)protocol
Soilssampledin2006wereusedtocharacterizetheactinobac-
terial community using TRFLP, whereas we used cloning
3500yearslater(Evensonetal.1976;Davis1983;Drexleretal. andsequencingtofurthercharacterizethecommunityfrom
1983). samples collected in the subsequent year. Microbial DNA
Thesesitesarewellcharacterizedintermsoftheirclimate, extraction and actinobacterial 16S rRNA gene amplifica-
plant community, and biogeochemical characteristics. tionfollowedsimilarprotocolstothosepreviouslydescribed
For example, daily air temperature, soil moisture, and inEisenlordandZak(2010).Briefly,microbialcommunity
soil temperature along with annual measurements of tree DNAwasextractedfromour2006samplesintriplicatefrom
species, diameter and height, leaf litter biomass, water 0.25to1gofsoilusingtheUltracleanSoilDNAextractionkit
balance, and leaf litter production have been recorded (MoBioLaboratories,Calsbad,CA,USA)forTRFLPanalysis.
since1994andareavailableattheMichiganGradientweb- MicrobialcommunityDNAwasextractedfromour2007soil
site (http://www.webpages.uidaho.edu/nitrogen-gradient/ samplesusing5-gsurfacesoilsubsampleswithMoBioPow-
Default.htm). Forests on our study sites were harvested ca. erMaxSoilDNAisolationkits(MoBioLaboratories)within
1900–1910 and have not experienced human disturbance one week of field collection for clone analysis. Actinobac-
sincethattime;tothebestofourknowledge,theyallhave terial 16S rRNA genes were amplified from total commu-
beenexposedtothesamedisturbanceregimeandsharethe nityDNAwithprimersEub338F-ACGGGCGGTGTGTACA
sameland-usehistory. andAct1159R-TCCGAGTTRACCCCGGC(Blackwoodetal.
We collected surface soil horizons (Oe, Oa, and A hori- 2005). The PCR protocol followed 95◦C for 5 min for ini-
zons) on three separate dates (June 2006, October 2006, tial denaturing, then 25 rounds of amplification (94◦C for
andMay2007)tocharacterizeactinobacterialcommunities. 30 sec, 57◦C for 30 sec, 72◦C for 90 sec) followed by 10
In each of the four sites, there are three randomly located min at 72◦C for elongation, and finally held at 6◦C before
30-m×30-mreplicateplotsranging15–150mapart.Wehave removal (adapted from Blackwood et al. 2005). All PCRs
previouslyandcontinuouslyquantifiedecological,edaphic, wereconductedinduplicateandproductswerepooledbe-
andbiogeochemicalcharacteristicsforeachplotinallfour forepurificationwithMoBioUltraCleanPCRCleanupKit
studysites(Burtonetal.1993;Reedetal.1994;Burtonetal. (MoBioLaboratories,Calsbad,CA,USA)accordingtoman-
2004;Pregitzeretal.2004).Ineach30-m×30-mplot,we ufacturer’sinstruction.ForTRFLPthePCRreactiondiffered
collected10soilsamplesusinga2.5-cmdiametersoilcore, from above by having a 6-Carboxyflurescein (6-FAM) at-
whichextendedtoadepthof5cm.The10surfacesoilsamples tachedtheEub338Fprimer.
540 (cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd.
S.D.Eisenlordetal. ActinobacterialPhylogeography
Communitycharacterizationusingterminal sion3.69(Felsenstein2005),usingtheJukesCantoralgorithm
restrictionfragmentlengthpolymorphism of substitution. Mothur (Schloss et al. 2009) was then em-
(TRFLP) ployed to assign OTUs at 90%, 93%, 95%, 97%, and 99%
similarity using the average neighbor algorithm. The rela-
Following PCR clean up of the actinobacterial 16S rRNA
tive abundance of OTUs at each similarity level was exam-
geneamplicon,approximately200–500ngofpurifiedPCR
product,asdeterminedbyPicogreen(cid:2)R analysis(Invitrogen; ined to address the argument that the resolution at which
microbial communities are analyzed influences results and
asinstructedbythemanufacturer),wasdigestedwith5Uof
subsequentlytheirinterpretation(ChoandTiedje2000).At
TaqI(Promega)at65◦Cfor1h.PassingdigeststhroughaMi-
97% similarity, Mothur was used for taxa-based alpha and
croconYM-30filter(Millipore)desaltedthemandremoved
b(cid:2)eta diversity estimates within and across sites and to run
enzymesfromthereaction.Eachsamplewassubmittedindu-
-LIBSHUFF (Schloss et al. 2004), a program which uses
plicateforgenotypingconductedattheUniversityofMichi-
coveragecurvestostatisticallydetectiftwoormoremicro-
gan’sCoreSequencingFacilityusinganABI3730XLDNASe-
bial communities are similar using the Cramer-von Mises
quencerwitha96capillaryarray.Rox1000(Bioventures)was
test statistic (Schloss et al. 2009). OTU sequences at 97%
usedasastandardtodeterminerestrictionfragmentlengths.
similarityweregeneratedbyconsensusofclonesequencesin
Electropherograms were inspected using Genemarker 1.60
Geneious.
(SoftGenetics). We required a peak height 50 fluorescence
Referencesequencesand56actinobacterialOTUsdefined
units and the appearance of each restriction fragment in
at 97% similarity were then realigned in Geneious with
bothduplicatesforoursubsequentanalyses.Eachterminal
ClustalW for phylogenetic analysis. Because phylogenetic
restrictionTRFwithapeakheightthatof1%orgreaterofthe
analysesaresensitivetotreetopology,RaxMLwasusedtose-
totalintensitywerescoredintothepresence–absencematrix
lectthebest-fittreewiththeMaximumLikelihoodalgorithm;
(Hassettetal.2009).
Staphylococcusaureus wasusedtorootthetree.Differences
inthephylogeneticpatternsineachstudysitewerequanti-
16S rRNAgenecloningandphylogenetic
fiedwiththeonlinestatisticaltoolUniFrac(Lozuponeetal.
analysis
2006).PhylogeneticdistancesmatricesreportedbyUniFrac,
Actinobacterial16SrRNAgeneswereclonedwiththeInvit- alongwiththerelativeabundancesofOTUsatallfivesim-
rogenTOPOTAcloningkitusingTOP10chemicallycompe- ilarity levels, were used for multivariate statistics described
tentcells(Invitrogen).InsertsweresequencedattheGeorgia below.
GenomicsFacilityattheUniversityofGeorgia(Athens,Geor-
gia).Thisstudyexpandedourprevioussequencingeffortsof
Environmentalvariables
33 clones in each plot (Eisenlord and Zak 2010; Genbank
accession FJ661107-FJ662388) to include an additional 63 Environmentalcharacteristicswereassembledintofourdata
clonesfromeachofthe12samples(i.e.,threeplotsineachof sets: (1) a biogeochemical data set composed of factors
fourstudysites),totaling1152sequences(Genbankaccession which we selected a priori that are relevant to soil mi-
HQ845548-HQ845603). crobial communities, (2) plant community composition,
Sequences were manually edited in Geneious v.5.0.2 (3) climatic characteristics, and (4) distance which repre-
(BiomattersLtd.)and727highqualitycontiguoussequences sented time since glacial retreat. The biogeochemical data
were generated from forward and reverse sequences. The matrix included soil pH and moisture content (measured
top-typespeciesmatcheswereretrievedfromtheRibosomal from our 2007 samples), and previously collected values
Database Project (RDP; Cole et al. 2009) for all sequences, for leaf litter C content, leaf litter C:N ratio, total leaf lit-
and50representativesequencesfromeverymajorgroupof ter mass, C:N ratio of soil organic matter, and extractable
−
theActinobacteriaphylawereretrievedfromtheNCBITax- soil NO (Table 1; Burton et al. 1993; Reed et al. 1994;
3
onomyBrowserforuseasreferences.Cloneandreferencese- Burton et al. 2004; Pregitzer et al. 2004). All metadata
quenceswerealignedusingClustalW(Thompsonetal.1994) are available at http://www.webpages.uidaho.edu/nitrogen-
intheprogramGeneious.Referencesequenceswereincluded gradient/Default.htm. All environmental data used in this
inthealignmenttobuildphylogeneticbackbonesupportby studywereaveragesovergrowingseasonfromtheyears2005
preservingspatialheterogeneityinthe16Ssequences.Align- to 2008. The second matrix represented the plant commu-
ments were manually edited to remove gaps and ambigu- nity based on the relative importance (i.e., basal area of a
ouslyalignedsequences.Referencesequenceswereremoved species/basal area of all species) of overstory and under-
fromtheclonealignmentbeforeoperationaltaxonomicunits storyspecies(TableS1).Ourthirddatamatrixcharacterized
(OTUs)weredetermined. climatic variation by including temperature, precipitation,
Theclonesequencealignmentwasusedtogenerateadis- andambientNdepositiontoidentifytheroleofclimatein
tancematrixinPhylogenyInferencePackage(PHYLIP)ver- shaping these actinobacterial communities (Table 1). The
(cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd. 541
ActinobacterialPhylogeography S.D.Eisenlordetal.
Table 1. Site averages for age, climate, and environmental the distance–time matrix was generated with the great cir-
characteristics. cledistance(Vincety1975).Theplantcommunitysimilarity
matrixwasgeneratedwiththeBray–Curtismetric(Brayand
A B C D
Curtis 1957). Biogeochemical, plant community, climatic,
Glacialretreat(yearsBP) 9500 11,000 13,000 13,500 anddistance–timedatawerevisualizedwithnonmetricmul-
Climate tidimensionalscaling(nMDS;Fig.2).
Meantemperature1(◦C) 4.82 6.06 6.49 7.65 With site as the main factor, an analysis of similarity
MeanPrecipitation(cm) 91.87 93.28 92.81 86.63
(ANOSIM) test was used to compare communities across
AmbientNdeposition1(kg/ha) 5.89 6.07 7.37 7.37
sites(n=4),withindividualplotsasreplicateswithineach
Environment site(n=3).TheMantel-typetest,RELATE,wasusedincon-
Leaflitter[C]1(g/kg) 458 456 453 455
junction with the Spearman rank correlation coefficient to
LeaflitterC:N1 63.68 57.06 52.91 43.41
Leaflittermass1(g) 412.7 396.3 591 550.2 determineifthereweresignificantcorrelationsbetweenthe
ExtractableDOC(mg/L) 5.5 2.79 5.85 9.95 biological data (TRFLP, OTU, and Phylogenetic distances)
ExtractableNO−(mg/L) 0.08 0.55 0.86 0.92 and the biogeochemical, plant community, climatic, and
3
SOMC:N 13.12 22.55 15.9 11.42 distance–time data sets. The RELATE test is similar to the
SOM[N](mg/g) 1.84 1.36 1.83 1.73 Manteltest,inthatituseselement-by-elementcorrelations
pH1 4.55 4.7 4.41 4.61
of similarity matrices. Though instead of Pearson correla-
Moisturecontent1(%) 23 24 18 14
tionsusedbytheManteltest,RELATEusesSpearman’srank
1ParamatersincludedindatasetsforRDAanalysis. correlationcoefficients,asismoreappropriatefortheinter-
pretationofourdata(ClarkandGorley2006).Ranksimilar-
itiesbetweensiteaverageswereusedinthisanalysistocorrect
primary historical event taken into consideration for this forthedifferentscalingofeachcorrelationcoefficient.The
studyistheperiodicretreatoftheWisconsinicesheetacross distance–decayrelationshipwasexploredwiththe2006TR-
lower and upper Michigan. Because distance between sites FLPdatabyplottingthelogtransformedaverageSørensen
overlaystimesincedeglaciationinourchronosequence,we communitysimilaritymetric(gainedfromPRIMER),against
useddistance–timeasourfourthdataset;itwascomposed thelogtransformedgeographicdistancebetweentheplots.
ofglobalpositioningsystem(GPS)coordinatestakenatthe Additional statistics were conducted in the R (R Devel-
centerofeachsampleplot(TableS2).Thechosenvariables opment Core Team 2008) package vegan (Oksanen et al.
foreachsetofdatawereassignedtobiogeochemical,plant 2011). Environmental vectors, of biogeochemical, climatic,
community,climatic,anddistance–timedatasetsformulti- and distance–time data sets, were fit to nMDS ordinations
variatestatisticalanalysis. of biologicaldata,whichidentifiedthe individualvariables
correlated with community patterns. Redundancy analysis
Multivariate statisticalanalysis
(RDA)wasusedtoexaminethecorrelationsbetweenspecies
ItisplausiblethatsoilActinobacteriabiogeographyisshaped patternsandenvironmentalvariablestoevaluatewhichvari-
by local environmental conditions, historical factors, or by ablesexplainedsignificantproportionsofvariationinActi-
both. Following the framework of Martiny et al. (2006), nobacteriacommunitycomposition.Distance-basedredun-
weusedmultivariateanalyses(PRIMERv6;Plymouth,UK) dancyanalysis(db-RDA;LegandreandAnderson,1999)was
inorderto identifysignificantcorrelationsbetweenfactors appliedusingtheBray–Curtisdistancemetrictodetermine
composingbiogeochemical,plantcomposition,climatic,and ifdistance–timesignificantlyaccountedadditionalbiological
distance–timedatamatrices. variation,afterthevariationduetoenvironmentalvariables
TRFLP fingerprint data matrices, OTU relative abun- washeldconstant.
dance matrices, and phylogenetic distances were treated
similarly as “biological” data. Similarity, matrices for TR- Results
FLPfingerprintsweregeneratedusingtheBray–Curtissim-
At all levels of examination, actinobacterial communities
ilarity metric (Bray and Curtis 1957) on non-transformed
werecompositionallydifferentineachofourfourforestsites,
presence–absence data. Relative abundances of OTUs were
andvariationinthecommunitieswassignificantlycorrelated
squareroottransformedtolessentheemphasisofthemost
withtimesinceglacialretreat(i.e.,distance;Table2).
abundantspeciespriortothegenerationofsimilaritymatri-
ceswiththeBray–Curtiscoefficient.Phylogeneticdistances
TRFLPcommunitycomparison
weregeneratedwiththeonlinepackageUniFrac(Lozupone
etal.2006). Basedongreaterthan1%contributiontototalTRFs,there
Biochemicalandclimaticsimilaritymatricesweregener- were27uniqueTRFsinJuly,and26inOctober.Rarefaction
atedwithEuclidiandistancesofstandardizeddata,whereas curvesgeneratedinEstimateS(Colwell2009)approachedan
542 (cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd.
S.D.Eisenlordetal. ActinobacterialPhylogeography
Figure2. nMDSaveragedbysite(stress<0.05),errorbarsrepresentstandarderrorofthreebiologicalreplicateswithineachsite.(a,b)Biogeochemical
andclimaticdatasetsseparatesitesA,B,C,andDbasedonEuclidiandistances.(c)distancedata,asaproxyforsiteage,arerepresentedbythegreat
circledistancebetweensites.(d,e)biologicaldatamatrices,JulyTRFLPandOctoberTRFLP,weregeneratedusingtheBray–Curtissimilaritymetric.(f)
PhylogeneticdistancesbetweensitesA,C,andDbasedonUniFracgeneticdistances.AllimagesweregeneratedinPRIMERv.6andeditedinExcel
2007.
Table2. ANOSIM(withsiteasamainfactor)andMantel-typetestRELATEresultsforTRFLP,phylogeneticdistance,andOTUrelativeabundancesof
fivelevelsofsimilarity.TRFLPdataarepresence–absencefrom2006.Phylogeneticdistancesarebasedon97%similarity.AllfivelevelsofOTUrelative
abundancesweresquareroottransformedbeforeanalysis.SpearmanmetricstatisticisrelatedasRho.
RELATE
ANOSIM Biogeochemical Plantcommunity Climate Distance-time
Data RStatistic P-value Rho P-value Rho P-value Rho P-value Rho P-value
TRFLPJune 0.45 0.002 0.18 0.115 0.04 0.54 0.02 0.436 0.35 0.022
TRFLPOctober 0.29 0.015 0.31 0.015 0.03 0.52 0.14 0.144 0.32 0.027
Phylogeneticdistance 0.58 0.004 0.17 0.223 0.15 0.29 0.17 0.159 0.24 0.064
OTU90% 0.06 0.320 −0.10 0.680 0.5 0.51 0.08 0.672 0.00 0.479
OTU93% 0.33 0.020 0.19 0.155 0.5 0.52 0.15 0.165 0.15 0.186
OTU95% 0.35 0.050 0.02 0.474 0.5 0.48 0.16 0.133 0.36 0.043
OTU97% 0.69 0.004 0.21 0.138 1.0 0.18 0.33 0.032 0.35 0.036
OTU99% 0.56 0.001 0.12 0.222 0.5 0.51 0.38 0.041 0.32 0.031
BolddesignatessignificantPvaluesaslessthan0.050.ItalicizedPvaluesareconsideredsuggestiveaslessthan0.075.
asymptoteandprovidedinSupportingInformation(Fig.S1). incompositiontheclosertheyaregeographicallyandinage
TRFLP actinobacterial community similarity, based on the (Table2).TherewasnoevidenceintheJulysamplesthatcom-
Sørensenmetric,hadasignificantnegativerelationshipwith munitieswithsimilarbiogeochemicalorclimaticcharacter-
timesinceglacialretreat(z-score=–0.094,P=0.02,Fig.3). isticshavesimilaractinobacterialTRFLPprofiles(P=0.115,
Thisdistance–decay,ortime–decay,relationshipheldtruefor P=0.436).ThefittingofallenvironmentalvectorstotheJuly
bothJulyandOctobersamplingdates;asdistanceincreased, nMDS,displayedinFigure4,revealeddistance–timetobesig-
communitysimilaritydecreased(Julyslope=–0.11,Octo- nificantlycorrelatedwithcommunitypatterns(P =0.014),
berslope=–0.03).Therewasnosuchrelationshippresent noneoftheothervariablesweresignificant(P valuesrange
whenbiogeochemicalcharacteristicswereregressedagainst 0.174–0.671). The db-RDA analysis revealed distance–time
geographicdistance(P=0.29).RELATEresults,basedonthe accounted for an additional 17% of community variation,
Bray–CurtissimilaritymatricesofJulyandOctoberTRFLP after variation correlated with environmental and climatic
data,indicateactinobacterialcommunitiesaremoresimilar factorswasheldconstant(P=0.048).
(cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd. 543
ActinobacterialPhylogeography S.D.Eisenlordetal.
(P=0.015).However,db-RDArevealed,whenvariationdue
tolitterC:Nwasheldconstant,distance–timedidnotsignif-
icantlyexplainanymoreofthevariationinthecommunities
(P=0.665),andthesamewastrueforlitterC:Naftervari-
ationduetodistancewasheldconstant(P =0.263).There
wasnorelationofactinobacterialcommunitiestotheplant
communityineitherJuly(Spearman=0.04,P =0.544)or
October(Spearman=0.03,P=0.524).
Taxonomicalphaandbetadiversity
Analysisof727clonedactinobacterialsequencesfromMay
2007,samplesitesA,C,andDresultedin56OTUsgrouped
at97%similarity.WeidentifiedOTUsin16of39actinobac-
terial families, classified with the RDP (Table 3). For the
most abundant OTUs, the closest similarity to known or-
Figure 3. Distance–decay relationship displaying the log transformed
community similarity based on the Sørensen metric of averaged July ganisms(cid:2)was90%tomembersoftheThermomonosporaceae
andOctoberTRFLPprofilesplottedagainstthelogtransformeddistance family. -LIBSHUFFresultsrevealedsignificantdifferences
betweensites.Slopeis0.0946,Yinterceptis2.07,andP=0.02. in community membership between sites A and D (P <
0.001),sitesCandD(P=0.007),andbetweensitesAandC
(P =0.015).Diversityestimates,AceandChao1,indicated
Incontrast,basedonRELATEanalysis,Octobersamples thattheoldestsiteDwasmorediversethanthetwonorth-
with similar biogeochemical characteristics did have simi- ern and younger sites, but this difference was not resolved
laractinobacterialprofiles(Spearman=0.309,P =0.015). when the 95% confidence intervals were considered. Rar-
Environmental vector fitting revealed both distance–time efactioncurves(seeFig.S1)alsoindicatetheoldestsite(Site
(P = 0.030) and litter C:N (P = 0.043) were correlated D) contained a greater richness than the younger sites. Al-
withcommunitypatterns;noothervariablesweresignificant thoughtherarefactioncurvesapproachedanasymptote,we
(P = 0.174–0.797). The RDA found distance–time signif- didnotcapturethefulldiversityoftheactinobacterialcom-
icantly accounted for 18% of community variation (P = munity. When examining the families found at each site,
0.026) and litter C:N accounts for 18% of the variation the oldest site (D) contained 13 unique OTUs from four
Figure4. Environmentalvectorfittingonan
nMDSplotofJuly2006TRFLPActinobacteria
communitiescalculatedwiththeBray–Curtis
dissimilaritymetric.Environmentalvariables
includedwereaprioridecidedbiogeochemical
factors,climaticvariabletemperature,and
distance–time.Distance–timewastheonly
variablesignificantlycorrelatedwith
communitycomposition,designatedby*(P=
0.014).
544 (cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd.
S.D.Eisenlordetal. ActinobacterialPhylogeography
Table3. Atotalof727clonesfromthreesites(A,C,andD)grouped The UniFrac distance matrices were analyzed with
into56OTUsat97%similarity.WeidentifiedActinobacteriain16of39 PRIMERtodetermineifsampleswithsimilarbiogeochem-
actinobacterialfamiliesbasedonRDPvaluesandphylogeneticanalysis. ical, plant community, climatic, or distance–time charac-
AllgroupingsexcepttheAcidimicrobialesarewithintheorderActino-
teristics also contained closely related communities. The
mycetales.Wereporttotalnumberofclonesandtheirabundanceat
ANOSIM, with site as a main factor, indicated that plots
eachsiteaswellasthetotalnumberofOTUsandtheirabundanceat
withinaparticularsiteweremoresimilarthanplotsbetween
eachsite.
different sites (P = 0.004); the nMDS ordination of acti-
Clones OTUs
nobacterial phylogenetic distances is displayed in Figure 2.
Family Total A C D Total A C D As with the TRFLP fingerprint data, the 97% similarity
phylogenetic distances were significantly correlated with
Micromonospora 42 20 6 16 5 4 2 3 distance–time(Spearman=0.24,P =0.064),butnotwith
Actinospicaceae 8 5 1 2 4 3 1 0
biogeochemical(Spearman=0.17,P =0.223),plantcom-
Catenulisporaceae 5 0 0 5 1 0 0 1
munity(Spearman=0.15,P=0.293),orclimatic(Spearman
Pseudonocardiaceae 56 26 16 14 4 4 3 3
Corynibacterineae* 87 29 27 31 4 2 2 4 =0.17,P=0.159)variables.Fittingofallenvironmentalvari-
Thermomonosporaceae 334 109 119 106 6 4 4 5 ablestothenMDSdemonstratedleaflitterC:N(P=0.028),
Streptosprangeaceae 1 0 0 1 1 0 0 1 temperature(P=0.062),anddistance–time(P=0.081)all
Nocardioidaceae 5 1 1 3 3 1 1 3 correlatedwithpatternsofphylogeneticdistancewhencon-
Microbacteriaceae 17 8 6 3 3 1 2 3
sideredindependently.Whenvariationduetoenvironmen-
Micrococaceae 3 0 0 3 1 0 0 1
talvariableswasheldconstantinthedb-RDA,distance–time
Streptomycetaceae 11 4 2 5 3 2 1 3
didnotexplainadditionalvariationinthecommunities(P=
Nakamureliaceae 8 0 3 5 1 0 1 1
Frankia1 1 0 0 1 1 0 0 1 0.33).Interestingly,whenvariationduetodistance–timewas
Kinosproaceae 1 0 1 0 1 0 1 0 heldconstant,noneoftheenvironmentalvariablescouldac-
Geodermaceae 5 1 2 2 2 1 1 1 countforvariationinthecommunitieseither(Pvaluesrange
Acidimicrobium2 143 42 59 42 17 9 11 11 0.16–0.25).
Becausethereismuchdebateaboutthescaleatwhichto
1Designatesgroupingtosuborder.
examinerelationshipsbetweenmicrobialcommunitiesand
2DesignatesorderAcidimicrobiales.
environmentalcharacteristics(ChoandTiedje2000;Bissett
et al. 2010), we examined the relative abundance of OTUs
familiesnotfoundinanyoftheyoungersites.Thenextoldest at 90%, 95%, 97%, and 99% similarity and their relation-
site(C)containedoneuniquefamily,whereastheyoungest ship to biogeochemical, plant community, climactic, and
site (A) contained no unique families (Table 3). We indi- distance–timedatasets(Table2).Relativeabundanceswere
viduallyregressedthemostabundantanddiversegroupsof square root transformed to minimize the impact of abun-
Actinobacteria, the Micromonospora, Pseudonocardia, Ther- dantspeciesandallowforhighercontributionofthemore
momonopsora,andAcidimicrobiumagainstpH,DOC,SOM rarespecies.Whentheactinobacterialsequenceswereexam-
Ncontent,leaflittermass,andC:Nratioandfoundnosig- inedat90%similarity,therewerenosignificantdifferences
nificantrelationshipinanycase(datanotshown). intherelativeabundancesofthesephylotypesbetweensites,
asdetectedbyANOSIM;however,ateachofthehighersimi-
Phylogeneticcommunityanalysis
larities,plotsgroupedmorecloselywithinsitesthanbetween
The UniFrac metric was used to identify unique phyloge- sites. Furthermore, there were no instances in which bio-
neticbranchlengthbelongingtoactinobacterialcommuni- geochemicalorplantcommunitycharacteristicssignificantly
ties within each site when compared with each other site, correlatedwithOTUrelativeabundances(P =0.068–0.13;
as well as when compared to the entire community. The Table2).Climatecharacteristics,specificallyannualtempera-
youngest site (A) and the oldest site (D) each had signif- ture,werecorrelatedwithactinobacterialrelativeabundance
icantly unique lineages when compared against the entire at 97% and 99% similarity and distance–time significantly
phylogenetictree(P=0.03,P=0.02,respectively);however, correlatedwithrelativeabundancesofActinobacteriaat95%,
thesiteofintermediateage(C)didnot(P=0.42).UniFrac 97%,and99%similarity(Table2).
also revealed that the oldest site (D) had unique lineages
whencomparedwiththetwoyoungersites(D–CP =0.06;
Discussion
D–AP=0.03).ThePtestfurtherrevealedthatphylogenetic
clusteringofsiteswithinthephylogenetictreedidnotoccur Microorganisms are believed to be globally distributed by
(P=0.15),anduponvisualinspectionofourtree,therewas prevailing winds (Griffin et al. 2002) and community pat-
noevidencethatactinobacterialcommunitiesintheyounger ternsinspaceandtimearethoughtto resultfrombarriers
siteswereasubsetofthoseintheoldestsite. todispersal,physiologicalrequirements,resourceavailability,
(cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd. 545
ActinobacterialPhylogeography S.D.Eisenlordetal.
competition,orsomecombinationthereof(Whitakeretal. sideredtogetherorindependently,despitethefactthatthese
2003;PapkeandWard2004).Severalfactorsleadustorea- edaphiccharacteristicscanshapesoilmicrobialcommunities
sonthattheregionalspeciespoolofActinobacterialiestothe (Ba˚a˚thandAnderson2003;VanderGuchetal.2007;Lauber
westofourstudysitesandprovidedpropagulesinaconsis- et al. 2008). When variation related to these variables was
tentmanneraseachsitewasfreedfromglacialiceoverthe heldconstant,distance–timestillaccountedforcloseto20%
pastca.14,000years.First,theprevailingwindsateachstudy oftotalvariationinourcommunities.
sitecomefromthewest,acrosslargebodiesofwater(i.e.,Lake Wealsocandispelthenotionthatsubtlevariationinplant
MichiganandLakeSuperior).Windcanbeanagentoflong- communitycompositioninfluencessoilactinobacterialcom-
distancedispersalforActinobacteria,aswellasotherbacteria munities,becausewefoundnorelationshipbetweentheplant
(Pearceetal.2009),andeachstudysiteshouldhavereceived communityandactinobacterialcommunitiesateverylevelof
wind-blownpropagulesfromthesameregionalspeciespool investigation.Whenweexaminedtherelationofactinobac-
duetotheirperpendicularorientationtoprevailingwinds.If terialcommunitiesat90%,93%,95%,and99%DNAsimi-
indeed“everythingiseverywhere”andtherearenodisper- larity,therewerenodetectabledifferencesbetweenthesitesat
sallimitationswithintheActinobacteria,theoryfollowsthat thecoarsestlevel(90%)ofsimilarity.Patternsemergedonly
eachecologicallyequivalentstudysitewillhavesimilaracti- whenthecommunitieswereconsideredatfinerphylogenetic
nobacterialcommunitiesduetonearidenticalenvironmental resolutions.Thisimpliesallofourstandshavecommunity
variables,whicheliminateenvironmentalfilteringaswellas members from the same pool of actinobacterial suborders,
constantadditionsbytheregionalspeciespool.Conversely, andthedifferencesinthecommunitiesoccuratthefamily
Bissettetal.(2010)describedahypothesis“whereveryougo, andgenuslevels,representedbyouranalysisofhigherDNA
that’swhereyouare”implyingthatbeyondstrongenviron- similaritypercentages.Thevariationincommunitiesatthese
mentalselection,otherfactors(i.e.,dispersalorcolonization finer levels of genetic resolution was consistently related to
limitationandevolutionaryevents)playasignificantrolein distance–time,litterC:N,andtemperature.Despiteourbest
shaping microbial communities. If this is true, and not all effortstoholdedaphicpropertiesconstanttominimizethe
sitesreceivedconstantadditionsofActinobacteriaasglaciers effectofenvironmentalfilteringandallowforthedetection
receded from the region due to dispersal limitation, then ofpossibledispersallimitation,changesinthecommunities
distance–time would be detectable as a significant force in over the growing season lead to correlations with the mi-
structuring the assembly of these communities. Consistent norchangesleaflitterC:NratioandtemperatureinourMay
withthisexpectation,ouranalysesrevealedthatdistance,a communities,andlitterC:NinOctoberbiologicaldataset.In
surrogatefortime,wasasignificantfactorshapingactinobac- bothcases,whenvariationduetodistance–time(17%–18%)
terial communities in soil, thereby providing evidence that washeldconstant,litterC:Nandtemperaturenolongerex-
dispersallimitationwasanecologicalforcestructuringthese plainedasignificantproportionofthevariation.Regardless,
communities. by examining these communities at many levels of resolu-
Tobetterunderstandtheimportanceofdispersallimita- tion,wehaverevealeddistance(i.e.,time)wasconsistently
tionasanecologicalforce,wesoughttoidentifythedegreeto correlatedwithvariationinactinobacterialcommunitycom-
whichenvironmentalheterogeneity,climaticvariation,and position, indicating both environmental heterogeneity and
distanceinfluencedactinobacterialcommunitiesinsoil.We historicalcontingenciesplayaroleinshapingthesemicrobial
purposely held ecological and edaphic factors as constant communities.
aspossibleacrossourstudysitestominimizedifferencesin AlthoughtheActinobacteriacommunitiesintheyounger
habitat characteristics. Distance was used as a proxy of the sites did not phylogenetically cluster as a subset within the
time since glacial retreat exposed new landscapes for colo- oldersites,theoldestsiteDhadthehighestspeciesrichness
nization.Asdistanceincreased,thecommunitysimilarityof estimatesandrarefactioncurvesbasedonTRFLPandtaxo-
Actinobacteriasignificantlydecreasedwithaz-scoresimilar nomicdatasets.Furthermore,phylogeneticanalysisrevealed
tothosefoundinastudybyMartinyetal.(2011)ofsaltmarsh thisoldestsitecontainedmembersoffourfamiliesandone
microbialcommunities.Ifenvironmentalconditionsbecame suborder,whichdidnotoccurintheothersites.Ouroldest
increasinglydifferentoverdistanceaswell,themostlogical site also has the highest proportion of unique community
explanation for this distance–decay relationship would be members,andthistrendwasfurthersupportedbyasignifi-
that species are adapted to, and structured by, their niche cantamountofuniquephylogeneticlineagewhencompared
requirements.However,ourforestsiteswerechosentocon- to the younger sites. It is plausible that the higher diver-
strain differences in edaphic and ecological characteristics, sityinouroldestsitecouldresultfromlongertimeelapsing
andasdistanceincreasedbetweensites,thesepropertiesdid since deglaciation, allowing more time to accumulate ad-
not become increasingly different. Furthermore, July acti- ditional species from the regional species pool, as well as
nobacterialcommunitycompositionwasnotcorrelatedwith moretimeforlocaladaptationordrifttooccur.Althoughwe
anyofourmeasuredenvironmentalcharacteristicswhencon- concedeamorein-depthandthoroughevaluationofthese
546 (cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd.
S.D.Eisenlordetal. ActinobacterialPhylogeography
communitiesisneededbeforewecandrawfirmconclusions bacterialcommunitiesovermultiplespatialscales.Mol.Ecol.
fromthesetaxonomicandphylogeneticcommunitypatterns, 69:134–157.
theconsistentnatureofourresultsindicatehistoricalcontin- Blackwood,C.B.,A.Oaks,andJ.S.Buyer.2005.Phylum-and
genciesdoinfluenceActinobacteriacommunitycomposition class-specificPCRprimersforgeneralmicrobialcommunity
overlongperiodsoftimeregardlessofthehighamountof analysis.Appl.Environ.Microbiol.71:6193–6198.
unexplainedvariation. Bray,J.R.,andJ.T.Curtis.1957.Anordinationoftheupland
This study highlights the importance of examining the forestcommunitiesofsouthernWisconsin.Ecol.Monogr.
identity of organisms as well as how related they are to 27:325–349.
one another when studying microbial biogeography. Based Burton,A.J.,C.W.Ramm,K.S.Pregitzer,andD.D.Reed.1991.
on presence–absence and relative abundance of 16S rRNA Useofmultivariatemethodsinforestresearchsiteselection.
actinobacterialgenes,multivariatestatisticsindicatetheob- Can.J.ForestRes.21:1573–1580.
served distance–decay relationship was best explained by Burton,A.J.,K.S.Pregitzer,andN.W.MacDonald.1993.Foliar
distance–time, providing evidence that dispersal limita- nutrientsinsugarmapleforestsalongaregionalpollution-
tionstructuresactinobacterialcommunities(Whitakeretal. climategradient.SoilSci.Soc.Am.57:1619–1628.
Burton,A.J.,K.S.Pregitzer,J.N.Crawford,G.P.Zogg,andD.R.
2003).Furthermore,aftervariationduetodistance–timewas
Zak.2004.SimulatedchronicNO −additionreducessoil
heldconstant,biogeochemicalandclimaticvariationcould 3
respirationinnorthernhardwoodforests.GlobalChangeBiol.
notaccountforfurthervariationinthesecommunities.How-
10:1080–1091.
ever,furthermultivariateandphylogeneticanalysesrevealed
Ba˚a˚th,E.,andT.H.Anderson.2003.Comparisonofsoil
asignificantamountofuniquelineageattheyoungestsite,
fungal/bacterialratiosinapHgradientusingphysiologicaland
lackofclusteringalongthephylogenetictree,andthecorre-
PLFA-basedtechniques.SoilBiol.Biochem.35:955–
lationofgeneticdistancetoleaflitterC:Nandtemperature
963.
as well as distance–time, all indicate that a simple mecha-
Cho,J.C.,andJ.M.Tiedje.2000.Biogeographyanddegreeof
nism of time and dispersal limitation may not be the only
endemicityoffluorescentPseudomonasstrainsinsoil.Appl.
ecologicalfactorshapingthesecommunities.Therefore,we
Environ.Microbiol.66:5448–56.
suspectothermechanismscontributetothespatialpatterns
Clark,K.R.,andR.N.Gorley.2006.PRIMERv6:user
of soil Actinobacteria in our study: such as the lasting im-
manual/tutorial.PRIMER-ELtd.,Plymouth,U.K.
printof“priorityeffects”onmicrobialcommunityassembly
Cole,J.R.,Q.Wang,E.Cardenas,J.Fish,B.Chai,R.J.Farris,A.
(Fukamietal.2010),thepossibilitythatsubtledifferencein
S.Kulam-Syed-Hohideen,D.M.McGarrell,T.Marsh,andG.
leaflitterbiochemistrycouldaltercommunitycomposition,
M.Garrity.2009.Theribosomaldatabaseproject:improved
or changes in unmeasured variables such as the commu-
alignmentsandnewtoolsforrRNAanalysis.NucleicAcids
nity composition of other bacteria and fungi which share
Res.37(Databaseissue):D141–D145;doi:10.1093/nar/gkn879.
asimilarniche,allcouldimpactactinobacterialcommuni-
Colwell,R.K.2009.EstimateS:statisticalestimationofspecies
ties.Regardless,wehavestrongevidenceonmanylevelsof
richnessandsharedspeciesfromsamples.Version8.2.User’s
resolution establishing that time since glacial retreat leads
Guideandapplication.Availableathttp://purl.oclc.org/
todecreasedcommunitysimilaritywithoutdecreasingenvi-
estimate.
ronmentalhomogeneityacrossthischronosequenceofsugar
Davis,M.B.1983.Quaternaryhistoryofdeciduousforestsof
maple forest ecosystems, and that distance, a surrogate for
easternNorthAmericaandEurope.GardenJ.70:550–563.
time, has consistent and significant impacts on these Acti- DeAngelis,K.M.,M.Allgaier,Y.Charvarria,J.L.Fortney,P.
nobacteriacommunities. Hugenholtz,B.Simmons,etal.2011.Characterizationof
trappedlignin-degradingmicrobesintropicalforestsoil.PLoS
One.6(4):e19306,doi:10.1371/journal.pone.0019306.
Acknowledgments
Drexler,C.W.,W.R.Farrand,andJ.D.Hughes.1983.
OurresearchwassupportedbygrantsfromtheNationalSci- CorrelationofglaciallakesintheSuperiorBasinwitheastward
enceFoundationandtheU.S.DepartmentofEnergyOffice dischargeeventsfromLakeAgassiz.GeolAssocofCan
ofBiologicalandEnvironmentalResearch. 26:309–329.
Eisenlord,S.D.,andD.R.Zak.2010.Simulatedatmospheric
nitrogendepositionaltersActinobacterialcommunity
References
compositioninforestsoils.SoilSci.Soc.Am.74:1157–1166.
BaasBecking,L.G.M.1934.Geobiologieofinleidingtotde Embley,T.M.,andE.Stackebrandt.1994.Themolecular
milieukunde.W.P.VanStockum&Zoon,TheHague,the phylogenyandsystematicoftheactinomycetes.Annu.Rev.
Netherlands(inDutch). Microbiol.48:257–289.
Bissett,A.,E.Richardson,G.Baker,S.Wakelin,andP.H.Thrall. Evenson,E.,W.Farrand,D.Eschman,D.Mickelson,andL.
2010.Lifehistorydeterminesbiogeographicalpatternsofsoil Maher.1976.Greatlakeansubstage:areplacementforvalderan
(cid:2)c 2012TheAuthors.EcologyandEvolutionpublishedbyBlackwellPublishingLtd. 547