Shifts among eukaryotes, bacteria, and archaea define the vertical organization of a lake sediment

16 Pages·2017·0.99 MB·English

by  Christian Wurzbacher

Wurzbacheretal.Microbiome (2017) 5:41
DOI10.1186/s40168-017-0255-9
RESEARCH OpenAccess
Shifts among Eukaryota, Bacteria, and
Archaea define the vertical organization
of a lake sediment
ChristianWurzbacher1,2,3*,AndreaFuchs4,5,KatrinAttermeyer5,6,KatharinaFrindte5,7,
Hans-PeterGrossart5,8,MichaelHupfer1,PeterCasper5andMichaelT.Monaghan1,2
Abstract
Background: Lake sedimentsharbordiversemicrobialcommunitiesthatcyclecarbonandnutrientswhilebeing
constantlycolonizedandpotentiallyburiedbyorganicmattersinkingfromthewatercolumn.Theinteractionof
activityandburialremainedlargelyunexploredinaquaticsediments.Weaimedtorelatetaxonomiccompositionto
sedimentbiogeochemicalparameters,testwhethercommunityturnoverwithdepthresultedfromtaxonomic
replacementorfromrichnesseffects,andtoprovideabasicmodelfortheverticalcommunitystructure insediments.
Methods: Weanalyzedfourreplicatesedimentcorestakenfrom30-mdepthinoligo-mesotrophicLakeStechlinin
northernGermany.Each30-cmcorespannedca.170yearsofsedimentaccumulationaccordingto137Csdatingand
wassectionedintolayers1–4cmthick.WeexaminedafullsuiteofbiogeochemicalparametersandusedDNA
metabarcodingtoexaminecommunitycompositionofmicrobialArchaea,Bacteria,andEukaryota.
Results: Communityβ-diversityindicatednearlycompleteturnoverwithintheuppermost30cm.Weobserveda
pronouncedshiftfromEukaryota-andBacteria-dominatedupperlayers(<5cm)toBacteria-dominatedintermediate
layers(5–14cm)andtodeeplayers(>14cm)dominatedbyenigmaticArchaeathattypicallyoccurindeep-sea
sediments.Taxonomicreplacementwastheprevalentmechanisminstructuringthecommunitycompositionandwas
linkedtoparametersindicativeofmicrobialactivity(e.g.,CO andCH concentration,bacterialproteinproduction).
2 4
Richnesslossplayedalesserrolebutwaslinkedtoconservativeparameters(e.g.,C,N,P)indicativeofpastconditions.
Conclusions: Byincludingallthreedomains,wewereabletodirectlylinktheexponentialdecayofeukaryoteswith
theactivesedimentmicrobialcommunity.ThedominanceofArchaeaindeeperlayersconfirmsearlierfindingsfrom
marinesystemsandestablishesfreshwatersedimentsasapotentiallow-energyenvironment,similartodeepsea
sediments.Weproposeageneralmodelofsedimentstructureandfunctionbasedonmicrobialcharacteristicsand
burialprocesses.Anupper“replacementhorizon”isdominatedbyrapidtaxonomicturnoverwithdepth,high
microbialactivity,andbioticinteractions.Alower“depauperatehorizon”ischaracterizedbylowtaxonomicrichness,
morestable“low-energy”conditions,andadominanceofenigmaticArchaea.
Keywords: Archaea,Eukaryota,Bacteria,Community,Freshwater,Lake,DNAmetabarcoding,Beta-diversity,
Sediment,Turnover
*Correspondence:[email protected]
1Leibniz-InstituteofFreshwaterEcologyandInlandFisheries,
Müggelseedamm301,12587Berlin,Germany
2BerlinCenterforGenomicsinBiodiversityResearch,Königin-Luise-Str.6-8,
14195Berlin,Germany
Fulllistofauthorinformationisavailableattheendofthearticle
©TheAuthor(s).2017OpenAccessThisarticleisdistributedunderthetermsoftheCreativeCommonsAttribution4.0
InternationalLicense(http://creativecommons.org/licenses/by/4.0/),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinktothe
CreativeCommonslicense,andindicateifchangesweremade.TheCreativeCommonsPublicDomainDedicationwaiver
(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated.
Wurzbacheretal.Microbiome (2017) 5:41 Page2of16
Background A basic mechanism thought to explain the vertical
Thecontinuousdepositionoforganicandinorganicpar- distribution of microbes is simply the one-way input
ticles to sediments is an important process in all aquatic of new organisms attached to OM that sinks from the
ecosystems. Approximately one third of the terrestrial water column. We hypothesize that this process would
organicmatter(OM)thatentersfreshwaterissequestered result in two simplified, competing structural models,
in sediments [1], although the total amount of OM that wherein the microbial community (1) consists exclu-
reaches the sediments is much greater than the amount sively of sinking colonizers, with the result being a
that is actually sequestered [2]. This is because micro- fully nested community structure in which the commu-
bial activity is responsible for the cycling of carbon, nity gradually changes from a complex and rich com-
including methane emission [3]. In lake sediments, a munity at the surface to an increasingly depauperate
proportion of newly settled OM is rapidly recycled and community with increasing sediment depth, dominated
subsequently transformed into secondary compounds, by progressive cell death, or (2) is structured by niche
resulting in a distinct uppermost sediment zone of high specialists at various layers that are well adapted to
heterotrophic activity [4, 5]. This is thought to lead to the specific environmental conditions including redox
the structuring of microbial communities along envi- gradients, OM, and nutrient (C, N, P) concentrations.
ronmental gradients that are much steeper than those These two models are suitably analogous to the recently
in marine sediments, with narrower vertical sequences developed β-partitioning of the total β-diversity of a
of electron acceptors [6]. The nature of this gradi- community, in which the taxonomic turnover is math-
ent influences the carbon, nitrogen, and sulfur cycles ematically separated into richness and nestedness com-
[7–9] and potentially affects the microbial community ponents (see [31] for a theoretical framework and [32]
structure[10]. forapplications).
In contrast to the wealth of studies on marine sed- The decomposition rate of settled or buried pelagic
iments (cf. the 65 studies of [11]), few studies have dead and living organisms is thought to depend primar-
examined the vertical microbial community structure ilyontheactivityoftheindigenousmicrobialcommunity
of freshwater sediments (e.g., [12–16]). The community rather than on chemical processes (e.g., depurination;
of sediment microbes was thought to be dominated by [33]). The decrease in DNA with depth that has been
Bacteria,togetherwithasmallerfractionofmethanogenic reported for freshwater sediments (e.g., [34]) is likely to
Archaea(reviewsin[6,17,18]).Thisviewhasbeenchal- resultofnucleicaciddegradationofdeadorganisms,par-
lenged by the recent discovery of an abundance of non- ticularly eukaryotes, whose biomass also decreases with
methanogenicArchaeainmarinesediments[19,20].They depth. As a result, the decomposition of buried organ-
are assumed to be adapted to low-energy environments, ismsshouldbeafunctionoftheactivecommunitywhich
and at least one lineage seems to be specialized in inter is itself buried over time. Vice versa, temporal patterns
alia amino acid turnover [20]. This discovery has led to of sedimentation will also influence the active micro-
arevisedperceptionofmicrobialcommunitiesinmarine bial community, for example by shifting the redox gra-
sediments, where Archaea appear to be as abundant as dient.Historicallychanginglakeconditionsarerecorded
Bacteriaandincreaseinrelativeabundancewithsediment in lake sediments as DNA and as chemical parameters
depth [11]. Data on sediment Archaea in freshwater are (e.g., [34]). An important question that remains is how
scarce,andthecausesofthesignificantvariationobserved decompositionprocesseswithinthesedimentredoxgra-
amongstudiesremainlargelyunknown(e.g.,[14,21–23]). dient are related to the burial of OM, eukaryotes, and
Prokaryoticactivity,biomass,andcellnumbersdecrease prokaryotes[35,36].
withdepthinmanyfreshwaterandmarinesediments(e.g., Weexaminedthebiogeochemicalpropertiesandmicro-
[24,25]),althoughotherstudiesreportrelativelyconstant bial community composition (Eukaryota, Bacteria, and
proportionsofactivecellswithdepthandfindnoaccumu- Archaea[37])ofsedimentsintheoligo-mesotrophichard-
lationofdeadcellsindeepersediments[5,26].Despitethe water Lake Stechlin in northeast Germany. Our aims
continuouspresenceofvegetativecellsandrestingstages, weretoevaluate(1)whethermicrobialcommunitieswere
recentstudiesofmarinesystemsindicatethatthemajority nested or structured (to test the competing models,
of microbial cells in energy-deprived horizons consist of above),(2)howsedimentparametersreflecting“present”
microbialnecromass[27,28]andtheproportionofliving and “past” conditions influence the overall community
organisms decreases with the increasing age of the sedi- structure(Table1),and(3)whetherrecentlyreportedver-
ment[29].Thevertical,progressivetransformationofOM tical patterns of marine Archaea [11] can predict those
anddepletionofelectronacceptorsmayeventuallyleadto observedinfreshwatersediments.Wetookfourreplicate
anextremelylow-energyenvironmentindeepersediment 30-cm sediment cores from ca. 30-m water depth. 137Cs
layers with very low growth rates similar to sub-seafloor dating indicated the cores include sediments deposited
sediments[30]. overthepastca.170years.
Wurzbacheretal.Microbiome (2017) 5:41 Page3of16
Table1Definitionof“present”and“past”sedimentparameters and archaeal OTUs recovered in each sample layer is
Wedefinepresentparametersastheprincipalcomponentsofallcontext providedinAdditionalfile1.
dataderivedfrom(a)porewateranalysis,whichindicatesthatchemical When pooling all replicates for analysis of the com-
gradientsarecausedbytheconsumptionandproductionofongoing
munity matrix with a cluster analysis, the microbial
biologicalprocesses(e.g.,sulfateandmethane),andfrom(b)directly
measuredparametersofmicrobialactivities(e.g.,bacterialprotein communitiesweregroupedintothreemajorclusterscor-
production).Thepresentparametersarethereforeanexpressionof responding to depths of 0–5, 5–14, and 14–30 cm, with
recentmicrobialprocesses.
apronouncedseparationat14cm(Fig.1).Thesediment
Pastparametersaretheprincipalcomponentsofconservative communities in layers between 14–30-cm depth were
parameters,whichonceintroducedintothesedimentswillnotchange moresimilartooneanother(> 65%ofcommunitystruc-
significantlyandarethereforeanexpressionofthelake’shistory(e.g.,
heavymetals).Here,wealsocategorizethetotalamountofelemental ture) than layers in the upper two clusters (< 50%). All
carbon,nitrogen,hydrogen,andsulfurasmainlyconservativeparameters. α-diversity indices (inverse Simpson, evenness, and esti-
Thepastparametersarethereforeanexpressionofhistoricalchanges.
mated Chao index) decreased with depth (Fig. 1). This
patternalsooccurredintherarefactionanalysisoftheHill
indices[39]andconfirmedasignificantseparationofthe
Results threedepthclustersbymajortaxonomicchanges(Fig.2).
Sediments in the cores were black in color, with no Communityturnover(distance)increasedwithdepth,fol-
visible lamination. Water content was 93–97%. No lowingadistancedecaycurveandapproachingadistance
macrozoobenthic organisms were visible, although DNA of 1 (i.e., no shared taxa) for the comparison of the low-
metabarcoding (see below) detected the presence of est layer (30-cm depth) with the surface (Table 2). Upon
nematodes in addition to microbes (Additional file 1). partitioningtheβ-diversityamongsamplelayersintotax-
There was an exponential increase of dissolved refrac- onomicrichnessandreplacementeffects[32],taxonomic
tory carbon with sediment depth (fluorescence index replacementwasconsistentlyhighandwassignificantfor
(FI) = 1.69–2.01; Fig. 1) across all four cores, indicat- multiplesamplelayersabove12cm(Table2).Incontrast,
ing the enrichment of fulvic acids [38]. Prokaryotic cell theeffectofrichnessincreasedwithdepth(R2 =0.96,F =
numbers were on average 1.8 ± 0.5 × 109 ml−1 wet 230.8,dF = 11, see Additional file 3) and it was signif-
sediment and were highest in the upper sediment lay- icantly elevated in the deepest layer (26–30-cm depth)
ers. Bacterial biomass production as carbon (BPP-C) (Table2).
(range 0–282 μgCml−1d−1) decreased rapidly with The OTUs that were the most influential in struc-
depth, approaching zero below 10 cm. Total DNA turing the microbial community across our 15 sedi-
concentration (range <0.3–17.6 μgml−1 sediment) was mentlayerswereidentifiedbycalculatingspecies(OTU)
negatively correlated with FI (r = −0.886) and fol- contribution to β-diversity [32]. This analysis identi-
lowed an exponential decay function. DNA half-life was fied 96 “structuring” OTUs (see Additional file 4) whose
inferred to be t1/2 = 22 a (corresponding to 5.4 cm; identity reflected the interrelationship of domains. The
f(DNA) = 13.9 × e−0.128x,r2 = 0.81). RNA content was mostinfluentialphylawereEuryarchaeotaandThaumar-
lower than DNA content in all layers, with DNA:RNA chaeota (Archaea) as well as Chloroflexi, Proteobacteria,
ratios ranging from 2.3 at the surface to 20.8 at 20-cm andPhycisphaerae(Bacteria)(Additionalfile4).Structur-
depth (Fig. 1). The sediment exhibited a typical electron ing OTUs included both redox-dependent groups (13%
acceptor sequence (Fig. 1) with a mean oxygen pene- of structuring OTUs could be clearly assigned to redox
tration depth of 4.6 mm (SD 1.4). Nitrate and nitrite processes by their classification, e.g., Nitrospiraceae,
were immediately depleted at the sediment surface, sul- Desulfobacteraceae, and Methylococcales) and redox
fateapproachedaconstantminimumconcentrationafter irrelevant groups (e.g., Eukaryota, Bacteriovoraceae). A
5 cm, soluble reactive phosphorous (SRP) and ammo- numberofthe96structuringOTUsweresignificantlyele-
+
nium (NH ) increased with sediment depth, N O gas vatedinoneormorezones(Fig.3).Eukaryoticandbacte-
4 2
wasnotdetected,CH increasedlinearlywithdepth,and riallineageswerecharacteristicfortheuppermostcluster
4
CO exhibited minima at the surface and at a depth of (cluster a in Fig. 1), whereas Archaea and Bacteria were
2
10 cm (Fig. 1). More detailed profiles of all measured elevatedinthelowestzone(clusterc)(Fig.3).Forclusterb,
parameters can be found in the supplemental material onlytwostructuringarchaealOTUswereidentified.The
(Additionalfile2). residual OTUs from cluster b were significantly elevated
Total taxon richness across the 60 samples was esti- eitherintheuppertwoclusters(mainlyBacteria)orinthe
mated (Chao) to be 8545 (SE = 173) operational taxo- lower two clusters (mainly Archaea). Only one structur-
nomic units (OTUs). The proportion of sequences with ingOTU(Methylococcales)wassignificantlydifferentin
nocloserelativesintheSILVAreferencedatabase(<93% itsrelativeabundanceinallthreeclusters(Fig.3).
sequencesimilarityusingBLAST)washighestatadepth Sequence proportions of Archaea, Bacteria, and
of9-10cm(40±4%).Anoverviewofeukaryotic,bacterial, Eukaryota(A:B:E)shiftedfrom10:70:20at0cmto50:50:0
Wurzbacheretal.Microbiome (2017) 5:41 Page4of16
Fig.1Depthprofilesofmicrobialcommunityclusteringandkeybiologicalandchemicalcharacteristicsofthesedimentcorestakenfrom30-m
depthinLakeStechlin.Microbialcommunitieswereclusteredbysimilarity(averageclustering)intothreegroups(a,b,c)correspondingtodifferent
depthhorizons(upper-leftpanel).Dates(y-axis)werecalculatedusing137Csmeasurements.Valuesaremeans(±1SE)fromfourreplicatecores.
Parametersandunits:FI=fluorescenceindex;cells[106ml−1];BPP-C=bacterialproteinproductionincarbon[μgCml−1d−1];DNAextract[ng
μl−1];thesharedchaoindex(Rveganpackage,[106]);lowsim.=proportionofsequenceswithnocloserelative[%ofsequences];SRP[mgl−1];
NH4[mgl−1];SO24−[mgl−1];DOC[mgl−1];CH4[μmoll−1];CO2[mmoll−1];P[mgg−1dryweight];Fe[mgg−1dryweight];Pb[mgg−1dry
weight].AdditionalinformationforallmeasuredvariablesperindividualcoreareprovidedintheAdditionalfiles2and12
at10cmand60:40:0at30-cmdepth(Fig.4a).Theeukary- as a function of the occurrence of Eukaryota (75.6%
oticproportionswerecorrelatedwithDNAconcentration of the variation) and Bacteria (10.0% of the variation;
(r = 0.869) and decayed exponentially with depth. Mul- model:R2 = 0.856,p < 0.001,Additionalfile5).Amul-
tiple linear regression could predict DNA concentration tivariate ordination of all samples confirmed the strong
Wurzbacheretal.Microbiome (2017) 5:41 Page5of16
Table2Microbialcommunityturnoverandrichnessforeachsamplinglayer
Depth[cm] Distance Repl. Richness LCrepl LCrich OTUs Coverage
0–1 NA NA NA 0.083*** 0.039 1787 0.825
1–2 0.683 0.600 0.083 0.086*** 0.001 1566 0.867
2–3 0.668 0.662 0.006 0.068 0.050 1724 0.828
3–4 0.688 0.659 0.029 0.057 0.079 1870 0.818
4–5 0.735 0.733 0.003 0.064 0.052 1759 0.833
5–6 0.739 0.709 0.030 0.070* 0.024 1625 0.859
6–7 0.761 0.722 0.040 0.069 0.019 1611 0.860
7–8 0.780 0.751 0.029 0.066 0.030 1709 0.841
8–9 0.822 0.749 0.072 0.071** 0.010 1567 0.861
9–10 0.824 0.753 0.071 0.071** 0.010 1557 0.863
10–14 0.860 0.782 0.078 0.074*** 0.011 1518 0.853
14–18 0.902 0.720 0.183 0.063 0.075 1268 0.880
18–22 0.920 0.715 0.206 0.061 0.113 1242 0.891
22–26 0.930 0.689 0.241 0.056 0.183 1087 0.905
26–30 0.934 0.649 0.285 0.041 0.304* 1019 0.903
DistancetotalJaccard-baseddistancemeasuresbetweensurface(0–1cm)andeachlowerlayer,Repl.replacementcomponentoftheJaccarddistance,Richnessrichness
componentoftheJaccarddistance,LCRepllocalcontributionofthereplacementcomponent,LCRichlocalcontributionoftherichnesscomponent,OTUsnumberof
observedOTUs,Cov.samplecoverage.AsterisksindicatesignificantlyincreasedLCvalueswithp<0.05*,0.01**,and0.001***,respectively
vertical gradient in microbial community structure, F = 12.3,p = 0.0005). The variance in cluster c was
reflected in the distance between the surface and deep reduced compared to the other clusters (betadispersal
sediments on axis 1 (Fig. 4b, Mantel correlation: r = analysis: Tukey’s honest significant differences between
0.735,p < 0.001).The three distinct clusters at different groups,p<0.01),confirmingthegreatersimilaritiesseen
depths (above) were recovered using adonis (Fig. 4b, in the previous cluster analysis. Microbial community
Fig.2Hilldiversitiesofthethreedepthhorizonclusters.Absolute(upperpanel)andaverage(lowerpanel)HilldiversityforthethreeHillnumbersq
resemblingrichness(0),exponentialShannonIndex(1),andInverseSimpsonindex(2)[39].Theabsolutedataisbasedonallsequencesobtainedfor
eachhorizon;theaveragediversityisbasedonallsequencesfromeachsamplingdepth,groupedashorizons,normalizedtoasamplingcoverage
of0.9.TheaveragediversitywassignificantdifferentbetweendepthhorizonsforallHillnumbers,ANOVA:F(2,12)=27.9,59.8,67.5forq=0,1,2;
respectively;p<0.001
Wurzbacheretal.Microbiome (2017) 5:41 Page6of16
and “past” principal components with comparable effect
sizes;seeAdditionalfile6).
In order to examine links between the two parame-
ter types and microbial community structure, we used
fuzzy set analysis to test whether past parameters were
correlated with the richness component and present
parameters with the replacement component of the
microbial community. We partitioned the whole dataset
into replacement and richness matrices and correlated
these with the present and past parameters. The rich-
ness community submatrix was strongly correlated with
the past parameters (two-dimensional fuzzy set ordina-
tionwiththefirsttwoaxesofthePCA,r =0.99),andthe
replacement community submatrix was correlated with
the present parameters (one-dimensional fuzzy set ordi-
nationwiththefirstaxisofthePCA,r =0.65,p<0.001).
Discussion
Comprehensivestudiesofthemicrobialcommunitiesand
physical-chemicalcharacteristicsoffreshwatersediments
are scarce, and general concepts are often transferred
Fig.3OverviewofthesedimentstructureinLakeStechlin.The
from marine systems without validation. This is despite
clusteranalysisseparatesthreedepthhorizons:theredox-stratified
the fact that salinity and sulfate concentrations are very
zone(0–5cm),whichincludesathinlayerofoxygen.Afewfauna
different in the two ecosystem types [40, 41] and can
speciesexistinthiszone,i.e.,Nematoda,Gastrotricha,and
microeukaryotes(e.g.,Ciliophora),inadditiontolargenumbersof haveaprofoundinfluenceonbiologicalcommunitiesand
highlyactiveBacteria.Below5cm,where 50%oftheDNAisalready biogeochemical processes. We addressed fundamental
decomposed,thesystementersthetransitionzone.Thiszoneis questions regarding the vertical structure, organization,
situatedbelowthesulfate-methanetransition.Below14cm,wefind
and inter-relationships among microbial communities
thedepauperatehorizon,whichextentsinthedeepersediment,in
and biogeochemical parameters in freshwater sediments
whichArchaeadominatethecommunity.Inanextrapolationofthe
richnesscomponentofthecommunitystructure,thelossofrichness and establish a structural model that can potentially be
wouldcompletelydominate(100%)themicrobialcommunityat1-m appliedtootheraquaticsediments.
depth(approx.500a).FollowingthedecaycurveoftheDNA,
99.99999%oftheDNAwouldbetransformedatthatdepth.Onthe
VerticalorganizationofLakeStechlinsediments
rightside,thetenmoststructuringOTUs(fromAdditionalfile4)are
The sediment habitat is thought to be autonomous
listed,whichweresignificantlyelevatedinthecorrespondinghorizon
(onlyresultswithp<0.01intheTukeyHSDposthoctestwere in terms of species richness and community structure
included).ThebracketsabandbcmarkthoseOTUsthatwere [17, 18], despite the constant colonization by microbes
elevatedintheuppertwoorlowertwozones,respectively.Onlytwo thatdescendfromthewatercolumnwithsinkingorganic
OTUswereelevatedinthetransitionzone.Thegrayboxmarksthe
particles(Additionalfile4).InLakeStechlin,weobserved
singletaxonthatwassignificantlydifferentinallthreehorizons.
ahighspecies(OTU)β-diversitywithdepthinlakesed-
Taxonnamesarecolorcodedaccordingtotheirclassificationor
phototrophyifapplicable:phototrophicorganism(green),Eukaryota iment,leadingtonearlycompletetaxonomicturnoverof
(black),Bacteria(red),andArchaea(blue) themicrobialcommunitywithin30-cmdepth.Suchhigh
turnover may be a common feature of vertical sediment
profilesandhasbeenreportedforbacterialtaxaincoastal
marine sediments [42], for marine Archaea and Bacteria
structure was correlated with sediment parameters [43], and for freshwater Archaea [21]. Previous studies
representative of both “present” (Mantel correlation: of freshwater sediments that reported moderate species
r = 0.527,p < 0.001)and“past”(r = 0.459,p < 0.001) turnover were restricted to low-resolution methods
conditions (see Table 1). These two parameter sets were [15, 21, 44]. Our study differed from previous efforts
nearly orthogonal in ordination (Fig. 4b). Apart from in that most studies have focused on either Bacteria or
the betadispersal analysis, we came to the same conclu- Archaea and not on all three domains simultaneously,
sion when we applied weighted phylogenetically based and none of the previous marine or freshwater studies
UniFrac distances instead (significant structuring along have partitioned β-diversity into richness and replace-
the depth gradient, significant separation of the three ment components. The former allowed us to examine
depth clusters, significant correlation with the “present” whole-community patterns and potential interactions,
Wurzbacheretal.Microbiome (2017) 5:41 Page7of16
Fig.4Microbialproportionsandcommunitystructure.aDepthprofilesofthemicrobialcommunity(Eukaryota,Bacteria,andArchaea)presentedas
relativeproportionstoeachother,whichwasdeterminedbyrelativepyrosequencingreadspermicrobialfraction.bNMDSordinationofthevertical
sedimentmicrobialcommunitystructure.TheclustersfromFig.1arepresentedasstandarddeviationaroundthegroupcentroid.Thecolorscaleof
thedotsrepresentssedimentdepth.Communitycompositionwashighlycorrelatedwithsedimentdepths(onaxis1;Mantelcorrelation:
r=0.735,p<0.001).Thethreedistinctclustersatdifferentdepthsweresignificantlydifferent(adonis,p=0.0005)withareducedvariancein
clusterc(Tukey’shonestsignificantdifferencesbetweengroups,p<0.01).Theprincipalcomponentsof“past”and“present”environmental
parameterswerecorrelatedwiththecommunitymatrix(Mantelcorrelation:“present”r=0.527,p<0.001and“past”r=0.459,p<0.001)andwith
theordination(envfit:“present”R2=0.658,p<0.001and“past”R2=0.547,p<0.001)
and the latter allowed to distinguish taxonomic changes in the redox-stratified zone, where most of the settled
that result from microbial activity from those related OMisreadilyavailable.Sulfatewasdepletedbelow5-cm
to sediment burial. We found the microbial commu- depth,andtherefore,mostredoxprocesseswilltakeplace
nity to be clearly delineated into three distinct clus- above.Themajorityoffreshwatersedimentstudiesexam-
ters, each spanning multiple sampling layers (details in inethiszoneingreatdetail(e.g.,[4,5,10]),includingthe
Additional file 7). Based on the significant contribu- identification of redox processes at the millimeter scale
tionoftaxonomicreplacement(Table2)andmeasurable [45, 46]. Active decomposition leads to high prokaryotic
microbial production in the upper two clusters (clus- cellnumbersclosetothesedimentsurface[5,25,47]and
ters a, b; Fig. 1), we term this part of the sediment the decreasingabundancewithdepth[15].Wealsoobserved
“replacement horizon” (Fig. 3). We term the lower lay- aninitiallossofmanytaxainthehighlyactiveoxycline—
ers, comprising the deepest cluster (cluster c; Fig. 1) the indicatedinourdatabyanoutliertotherichnesscommu-
“depauperate horizon,” based on the importance of tax- nitycomponent(Additionalfile3)—whichmayhavebeen
onomic richness (as opposed to turnover) in structuring intensified by active grazing by ciliates and copepods.
thecommunityandbasedontheconstancyofmostsed- Importantmethaneoxidationprocessesoccurinthetran-
imentbiogeochemicalparametersatthesedepths(Fig.3, sitionzone(belowthesulfate-methanetransition)(cluster
Additional file 7). Both horizons are discussed in more b, Additional file 7) [48]. At 10 cm, the CO minimum
2
detailbelow. could indicate the start of hydrogenotrophic methano-
genesis (Fig. 1) as previously described for profundal
The replacement horizon We define the replacement Lake Stechlin sediments [49]. The archaeal contribution
horizon (Fig. 3) as comprising the two microbial com- roseconstantlyinthetransitionzone,whileBacteriaand
munity clusters in which taxonomic turnover was most Eukaryota decreased and the general activity measures
pronounced (a, b in Fig. 1). This horizon is further sub- declinedrapidly.
divided by the sulfate-methane transition into what we According to DNA concentrations, more than 85% of
term “redox-stratified” and “transition” zones (Informa- settled organisms were decomposed within this replace-
tionBox,Additionalfile7).Bacterialactivitywashighest ment horizon, which spanned approximately 60 years.
Wurzbacheretal.Microbiome (2017) 5:41 Page8of16
The decay of DNA, in combination with an enrichment lineagesidentifiedinAdditionalfile4.Anotherpotential
offulvicacidswithincreasingdepth(FIvalues),confirms mechanism—one thatmay bethe mostimportantin the
the assumption that DNA can serve as a proxy for the depauperate horizon—is differential cell replication. The
buried OM. We conclude that there is a gradient of OM resourcesforcellmaintenanceandgrowthshoulddepend
quality in addition to a gradient of electron acceptors in on cell size and complexity. This means that small cells,
thereplacementhorizon.Thismayfacilitatethestratifica- such as nano-Archaea (e.g., Candidatus Parvarchaeum),
tionofactivemicrobialtaxawithdepthbecauseOMcan should have a selective advantage because they can con-
simultaneously serve as an electron donor and acceptor tinuetogrowunderconditionsinwhichlargercellsmust
[50, 51]. OM quality can also modulate microbial redox switch to cell maintenance. This could be one explana-
processes [52], with apparent consequences for carbon tionfortheobserveddropinevennessinthedepauperate
turnoverrates[36,51]. horizon.
Inthemechanismofrandomappearances,theappear-
The depauperate horizon The depauperate horizon ance of taxa may be due to the disappearance of others.
(Fig. 3) was characterized by high concentrations of Because high-throughput sequencing methods produce
methane (CH ) and CO , a dominance of Archaea, relative (rather than absolute) data, it may superimpose
4 2
and low diversity compared to the replacement horizon. proportionsoverquantities.Forexample,theinitialdecay
Microbialcommunitycompositionwasmorenested,with of eukaryotes may have opened a niche for previously
a 9% relative nestedness here compared to only 2% in hidden rare taxa. Further, if there was no growth in the
the replacement horizon. By entering this horizon, the sediment, lineages that are potentially better suited for
DNA:RNA ratio doubled and Archaea replaced Bacteria long-term survival than others would appear, such as
asthedominantmicroorganisms.Webelievethisreflects spore-forming Bacteria (Firmicutes). However, we (and
an increase in the number of microbes entering a sta- others: [27, 34]) did not observe an enrichment of this
tionary state below this depth, where cell maintenance lineagewithdepth.Inaddition,theuseofreplicatecores
predominatesovercellsynthesisduetothelowavailability in combination with our conservative stripping (see the
ofterminalelectronacceptors.Thisisanalogoustowhat “Methods”section)shouldhaveremovedmostoftheran-
has been suggested for cells in low-energy marine envi- domeffects.Themostlyconstantcellnumberswithsmall
ronmentsinthedeepsub-seafloorsediment[27,30,53]. local maxima and the observed shifts in the evenness
The variability in community composition was very low support a non-random stratification of sediment com-
across replicates in the depauperate horizon. The nest- munities including cell replication. While the cellular
edness suggests the gradual disappearance of taxa with reproduction probably approaches stagnation for most
buryingageandarichnesscomponentofturnoversteadily microbes in the depauperate horizon, the slowly shift-
increasing to more than 20% (Table 2). If the richness ing redox conditions across seasons and years may be
componentwastofurtherincreaseinalinearmanner,it conducivetocolonizationofthereplacementhorizonby
wouldbethesolefactorstructuringthecommunitycom- different niche specialists. The low sedimentation rate
position deeper than 1 m (the total sediment depth of of Lake Stechlin (ca. 2 mm per year, as determined by
LakeStechlinis6m).Itisintuitivethattherichnesscom- 137Cs dating at the Federal Office for Radiation Protec-
ponent may be a function of the burying time and that tion,Berlin[courtesyofU.-K.Schkade],or0.4–2.1gm−2
it represents the fading signal of preserved organisms. It d−1,asdeterminedbysedimenttraps[59])maymitigate
remains unclear why it does not follow an exponential thisstratification,andthescaleofthehorizonsmaydiffer
decayfunctionanalogoustothatforDNA. insystemswithhigherorlowersedimentationrates.
Potential causes of the high taxonomic replacement Burialprocessesandmicrobialactivities
Many “present” parameters changed rapidly with depth, Themicrobialsedimentcommunityappearedtobehighly
particularly in the replacement horizon (e.g., DNA, FI, indigenous, and yet the constant arrival of sinking OM
BPP, electron acceptors), and this was a likely driver of couldburythemicrobialcommunity.Indeed,buriedDNA
the high degree of taxonomic turnover. Several mech- andorganismspreservehistoricalplanktoncommunities
anisms could be responsible for these patterns, namely thatcanbeindicativeofpastconditionsofthelakeecosys-
cellular turnover and random appearances. In cellular tem[34,60,61].Thesepastenvironmentalconditionsare
turnover,taxonomicreplacementispotentiallycausedby alsopartlypreservedasparticulatematter,whichisrela-
cellsynthesis,lysis,andrecyclingofdormantcells,which tivelyconservative.Wefoundseveraloftheseconservative
are assumed to be high in sediments [54, 55], particu- “past” parameters to correlate well with the present-day
larlyvirallysis[56,57].Wefoundindicationsforcellular communities.Althoughpreviousstudieshavefoundsed-
recycling caused by the predatory Bacteriovoracaceae iment parameters to influence community patterns in
(cf. [58]), which was one of the structuring bacterial marine systems [43], the parameters were not separated
Wurzbacheretal.Microbiome (2017) 5:41 Page9of16
into a present-past context, and microbial community theMCG(potentiallymethanogenic,[68])andtheMarine
turnover was not partitioned into richness and replace- Benthic Group D (MBG-D). Both groups are among the
mentcomponents.Ourresultsrevealedthattherichness most numerous Archaea in the marine sub-seafloor, and
component could be largely explained by the first two they are thought to metabolize detrital proteins ([20],
principalcomponentsofthepastparameters.Incontrast, discussed in more detail below). Interestingly, we also
thereplacementcomponentwasnotfullyexplainedbythe identified a MBG-B as structuring OTU for the transi-
present parameters, indicating that sources of variation tion zone (Fig. 3), a group which was recently described
otherthanenvironmentalparametersareimportant,such as eukaryotic progenitor from a hydrothermal vent field
asbioticinteractions.Strongbioticinteractionshavebeen (Lokiarchaeota, [75]). Several MCG OTUs belonged to
identifiedinaverticalprofileofameromicticlakewitha the top structuring taxa. MCG was recently named as
comparablechainofredoxprocessesastheyoccurinsed- Bathyarchaeotaby[76]foritsdeep-branchingphylogeny
iments [62]. Deep sediment layers may offer low-energy and its occurrence in deep subsurface environments—
niches that favor a large variety of syntrophic microor- environmental conditions that our cores (30-m water
ganisms [63, 64]. The Dehalococcoidales (Chloroflexi depthand30-cmlength)didnotmeet.
[65, 66]) and the Miscellaneous Crenarchaeotic Group Our results suggest that the specific niche adaptation
(MCG,[67,68])arepromisingcandidatesforsuchhybrid of these microbes is not necessarily related or restricted
forms of energy harvesting and were among the most to the deep biosphere but rather to a cellular state of
influential lineages in our data set (Additional file 4). “low activity” [77]. In this context, it is interesting that
The MCG co-occurred with Dehalococcoidales, similar single MCG OTU sometimes dominated the commu-
towhatwasfoundinmethanehydrate-bearingsediment nity in the deep horizons (up to 34% in core D at 26–
in Lake Baikal at >1500-m water depth [69]. Another 30 cm), resulting in a reduced overall evenness and a
indication that biotic interactions are important in Lake shift of the residual taxa to the rare biosphere, con-
Stechlin sediments is the appearance of the Candidatus trasting the potential random effects as discussedabove.
Parvarchaeum as a structuring lineage (Fig. 3). This lin- Another intriguing observation is the considerable over-
eage can exhibit cell-to-cell coupling that allows for lapofarchaealandpartiallybacteriallineagesbetweenour
thermodynamic processes that would otherwise not be study and deep-sea environments. Consequently, typical
possible[70]. marine lineages (e.g., Archaea in Additional file 4: MGI,
MCG [Bathyarchaeota], MHVG, DHVEG-1, DHVEG-6
Archaeainfreshwatersediments [Woesearchaeota], DSEG, MBG-A, MBG-B [Lokiar-
ThefactthatArchaeacanbeverynumerousinfreshwa- chaeota], MBG-D, MBG-E) are not as “marine” or
tersedimentsandevendominatemicrobialcommunities as “deep-sea” as previously thought. Given the high
is a rather new discovery, and data from comparative cost of deep-sea research [30], freshwater sediments
studies are lacking. Their recovery rate in relation to might literally pose a row-boat alternative for research
bacterial sequences or cell numbers varies between 3– questions targeting these “remote” and “extremophile”
12% ([71], cells), 5–18% ([21], qPCR), and 14–96% ([14], microorganisms.
qPCR), depending on the lake, sampled sediment hori-
zons,andmethodsemployed.Inmostcases,onlysurface Studylimitationsandperspectives
sedimentsampleshavebeenconsidered(e.g.,[72],1%of When we look at systems that sequester carbon, it is
cells), and the few studies involving vertical profiling to important to keep relic DNA in mind. Relic DNA is
date are ambiguous in finding an archaeal depth gradi- defined as DNA residuals that remain in the system
ent. Our results and those from cell counts from Lake after cell death. Its presence can inflate richness and
Biwa (Japan) of [71] suggest an increase in the propor- misrepresent relative abundances in some types of soils
tionofArchaeawithsedimentdepth.However,theresults when analyzed with DNA metabarcoding [78]. The very
obtainedbyquantitativePCRforLakeTaihu(China;[14]) few aquatic studies that investigated relic DNA reported
andLakePavin(France; [21])didnot reportsucha rela- largeamountsofextracellularDNAinmarineandfresh-
tionship.Archaeahave,onaverage,compactgenomes[73] water sediments, with fragment sizesof up to 10 kb, but
and a lower ribosomal copy number than Bacteria [74], with low amplification success [79, 80]. The degradation
whichmayleadtounderestimatesofarchaealabundance. of extracellular DNA varies widely in different environ-
Similar to our results, [21] found three sequential depth mentsandisdependentontheamountofOMandorganic
clusters in the archaeal community structure within the clay fractions present [81]. The low RNA content in the
first40cm,defininganintermediatelayerbetween4and deeperlayersinourstudymightindicatethepresenceof
12cm.Nexttowell-describedmethanogenicArchaea,we relic DNA; however, cell numbers in these deeper layers
mainly recovered archaeal lineages with no clear func- (>1.5 × 109 cells ml−1) were similar to those in upper
tionalassignmentthusfar(similarto[21]),i.e.,primarily layers (1.9–2.4 × 109 cells ml−1). The high DNA:RNA
Wurzbacheretal.Microbiome (2017) 5:41 Page10of16
ratiomaythereforeresultfromanabundanceofdormant been the subject of more than 55 years of research
and potentially dying cells [54], rather than extracellular [83]. Sediment cores were extracted from the southern
relicDNA.Inourlake,theextractableDNAseemstobe bay of the lake. Four adjacent sites were sampled to
rapidly decaying (similar to the sequences from eukary- account for spatial heterogeneity in the sediment (sites
otesaspotentialprogenitorsofrelicDNA),whichpoints A–D, Additional file 9). Pore water was collected by
to a short-lived fate of relic DNA in Lake Stechlin. We four in situ dialysis samplers, so-called peepers [84],
also found no evidence for fragmented DNA in our sed- which were deployed for 14 days using a frame (1 m2).
imentsamplesthatwouldindicatethepresenceoflarger Shortlybeforeretrievingthepeepers,foursedimentcores
quantities of extracellular DNA (Additional file 8). Relic were taken from each site with Perspex tubes (inner
DNA would have caused an overestimate of the rich- diameter 6 or 9 cm; length 60 cm) using gravity cor-
ness, in particular, the deeper layers would appear more ers (UWITEC™, Mondsee, Austria) at 30-m water depth
speciesrichthantheyare.TheuseofasingleSSUprimer (aphotic depth) on two subsequent days (March 26 and
pair in our study was also a compromise and underes- 28, 2012; peepers were retrieved on April 1, 2012).
timates the richness of metazoan groups [37, 82], which Two sediment cores (6-cm diameter) were stored in the
affectsthezooplanktonOTUsthatwefoundintheupper dark at 4 °C until oxygen penetration depth was mea-
layers. This means that the richness component of β- sured within the next 4 h (see below). The sediments
diversitymaybemoreimportantasaresult,leadingtoa of the 9-cm cores were sliced directly into 1-cm lay-
narrowertransitionzoneandgreaterdifferencesbetween ers for the uppermost 10 cm and then in 4-cm layers
the horizons. On the other hand, our study was also for sediment depths of 10–30 cm. One core was used
limited by resolution, since our pyrosequencing efforts for the analysis of the total sediment, and the other was
couldnotadequatelyanalyzetherarebiosphere,inwhich used for pore water, gas, and microbial analyses (see
we would suspect most signals from relic DNA. Future below). In May 2014, 24 additional cores were taken to
studieswithhighersequencingdepthsandseveralgroup- determine the age-depth correlation using the cesium
specificprimerpairswillbeabletofollowthefateofrelic 137technique.
DNAinmoredetailsas,e.g.,ofEukaryota.
Maximumoxygenpenetrationdepth
Conclusions Two initial cores were carefully transferred into 20-cm
Our results indicate that the sediments of Lake Stechlin shortcoreswithoutdisturbingthesedimentsurface.The
are a steady-state and chemostat-like environment with short cores were kept cool (4 °C) until measurements
ahighlystratifiedindigenousmicrobialcommunity.Sed- were taken. Oxygen microprofiles were performed using
iments were not a one-way system for the burial of twoClark-typemicroelectrodes(OX50oxygenmicrosen-
organic matter and were not composed purely of redox- sors, Unisense, Aarhus, Denmark) with a 50-μm glass
active taxa. We conclude that both processes take place tip.SensorTracePro2.0software(Unisense)wasusedfor
alongside a vertical gradient of electron acceptors and datastorage.Theelectrodeswerecalibratedbytwo-point
decomposing OM of decreasing quality with depth. The calibration.Foreachcore,wemeasuredatleastfourpro-
microbialcommunitywasstructuredintodistinctgroups, files. The sediment-water interface was defined as the
and both the microbial community and the sediment point where the oxygen depletion shifted from linear to
parameterscouldbedividedintocomponentsrelevantfor non-linear[85].
burial and past conditions as well as for recent carbon
turnoverprocessesandtheircontextdata.Bioticinterac- Porewateranalysis
tionsarelikelytoplayanimportantrole,andwewereable The sampled sediment horizons were centrifuged
toidentifyimportantsedimenttaxaforeachhorizon.We (13,250g for 10 min) to retrieve pore water (filtered
putaspotlightonthelargelyunexploredfreshwatersedi- through rinsed 0.45-μm cellulose acetate membranes,
mentsandconfirmedearlierfindingsthatwerepreviously Roth, Germany) for immediate analysis of the dissolved
describedonlyformarinesediments,suchastheimpor- organic carbon (DOC) and FI. DOC was measured
tanceofmarinearchaeallineagesandtheintroductionof as non-purgeable organic carbon with an organic car-
adepauperationzoneinwhichtheburialprocessbecomes bon analyzer (multi N/C 3100, Analytic Jena AG, Jena,
increasinglyimportant. Germany). FI was measured following the protocol of
[38]. Peeper samples were analyzed for concentrations
Methods of SRP and ammonium (NH+), dissolved iron (Fe2+/3+),
Samplingsiteandsamplingprocedures manganese (Mn2+), chloride4 (Cl−), nitrate (NO−), and
3
Lake Stechlin (latitude 53° 10 N, longitude 13° 02 E) sulfate(SO2−),followingDINENISO10304-1.SRPand
+ 4
is a dimictic oligo-mesotrophic lake (maximum depth NH werephotometricallydeterminedusingsegmented
4
69.5 m; area 4.23 km2) in northern Germany that has flow analysis (SFA, Skalar Sanplus, Skalar Analytical