interactions between specific phytoplankton and bacteria affect lake bacterial community succession

Interactions between specific phytoplankton andbacteria affect lake bacterial community succession

Sara F. Paver,1 Kevin R. Hayek,2 Kelsey A. Gano,4

Jennie R. Fagen,4 Christopher T. Brown,4

Austin G. Davis-Richardson,4 David B. Crabb,4

Richard Rosario-Passapera,4 Adriana Giongo,4

Eric W. Triplett4 and Angela D. Kent1,3*1Program in Ecology, Evolution, and ConservationBiology, 2School of Integrative Biology, and 3Departmentof Natural Resources and Environmental Sciences,University of Illinois, Urbana, IL, USA.4Department of Microbiology and Cell Science,University of Florida, Gainesville, FL, USA.

Summary

Time-series observations and a phytoplanktonmanipulation experiment were combined to test thehypothesis that phytoplankton succession effectschanges in bacterial community composition. Threehumic lakes were sampled weekly May–August andcorrelations between relative abundances of specificphytoplankton and bacterial operational taxonomicunits (OTUs) in each time series were determined. Toexperimentally characterize the influence of phyto-plankton, bacteria from each lake were incubatedwith phytoplankton from one of the three lakes orno phytoplankton. Following incubation, variation inbacterial community composition explained by phy-toplankton treatment increased 65%, while the varia-tion explained by bacterial source decreased 64%.Free-living bacteria explained, on average, over 60%of the difference between phytoplankton and corre-sponding no-phytoplankton control treatments. Four-teen out of the 101 bacterial OTUs that exhibitedpositively correlated patterns of abundance with spe-cific algal populations in time-series observationswere enriched in mesocosms following incubationwith phytoplankton, and one out of 59 negatively cor-related bacterial OTUs was depleted in phytoplanktontreatments. Bacterial genera enriched in mesocosmscontaining specific phytoplankton assemblagesincluded Limnohabitans (clade betI-A), Bdellovibrioand Mitsuaria. These results suggest that effects of

phytoplankton on certain bacterial populations,including bacteria tracking seasonal changes in algal-derived organic matter, result in correlations betweenalgal and bacterial community dynamics.

Introduction

Microbial contributions to energy flow and biogeochemicalcycling are fundamentally important to aquatic ecosystemfunction. In particular, individual bacterial taxa havedistinct combinations of metabolic capabilities (e.g.

decomposition of specific carbon compounds, nitrogentransformations), such that alterations to bacterial com-munity structure yield changes in rates of functions per-formed by bacteria (Bertilsson et al., 2007; Ishida et al.,2008; Strickland et al., 2009). Species interactions (e.g.competition, predation, facilitation) and environmentalconditions that influence bacterial community compositionand temporal dynamics therefore have potential im-plications for ecosystem function. In lakes, interactionsbetween phytoplankton and bacteria have been proposedto influence pelagic bacterial community dynamics (Kentet al., 2007; Šimek et al., 2008; Paver and Kent, 2010;Paver et al., 2013). Establishing the effect of phytoplank-ton assemblages on bacterial community temporal pat-terns and investigating potential mechanisms of algalinfluence will provide insight into the ecology of microor-ganisms, as well as enhance the ability to predict micro-bial response to environmental change.

Effects of phytoplankton on lake bacterial communitydynamics have been inferred from environmental obser-vations and bacterial response to phytoplankton mani-pulation. Lake phytoplankton communities undergo sea-sonal succession due to factors that include changes innutrient availability (e.g. silica, phosphorus, nitrogen) andgrazing pressure (Sommer et al., 1986; Graham et al.,2004). Changes in phytoplankton community compositionhave been observed to correlate with changes in bacterialcommunity composition (Kent et al., 2004; 2007). In astudy of six humic lakes, over half of the variation inbacterial community composition over a 3-month periodwas explained by temporal patterns in the phytoplanktoncommunity and covariation between changes in phyto-plankton and the environment (Kent et al., 2007). Whilecorrelated patterns suggest that phytoplankton influencebacterial community dynamics, they do not necessarily

Received 28 November, 2012; accepted 22 March, 2013. *Forcorrespondence. E-mail [email protected]; Tel. (+1) 217 333 4216;Fax (+1) 217 244 3219.

bs_bs_banner

Environmental Microbiology (2013) doi:10.1111/1462-2920.12131

© 2013 John Wiley & Sons Ltd and Society for Applied Microbiology

mailto:[email protected]

result from a causal relationship. Environmental time-series observations and experimental manipulations needto be combined to establish phytoplankton influence onbacterial community dynamics. For example, at the popu-lation level, bacteria within the genus Limnohabitansincreased in abundance in a reservoir during periodsdominated by cryptophytes (Šimek et al., 2008), and bac-terial isolates from this genus exhibited enhanced growthwhen co-cultured with certain algae or amended with algalexudates (Šimek et al., 2011). A combined approachinvestigating community-level responses to phytoplank-ton has yet to be reported.

The influence of phytoplankton on bacterial communitydynamics may be exerted through a number of potentialmechanisms. Bacteria rapidly utilize exudates releasedby phytoplankton (e.g. sugars, amino acids) (Maurinet al., 1997; Sadro et al., 2011). The composition of phy-toplankton exudates is species-specific (Fogg, 1983)and it has been shown that taxonomic composition ofseawater bacterial cells taking up algal exudates andbacterial community composition following incubationwith exudates depends on the phytoplankton species thatproduced the exudates (Sarmento and Gasol, 2012).Detritus following algal cell death is another source ofphytoplankton-derived organic matter that has a species-specific effect on bacteria (van Hannen et al., 1999). Suc-cession of the abundance and composition of genesassociated with the utilization of algal-derived organicsubstrates, changes in enzyme expression, and changesin dissolved organic matter uptake patterns in freshwaterand marine systems lend further support to organic-matter mediated influence of phytoplankton (Arrieta andHerndl, 2002; Lau and Armbrust, 2006; Lau et al., 2007;Paver and Kent, 2010; Eckert et al., 2012; Paver et al.,2013; Teeling et al., 2012). In addition to providing asource of organic matter, phytoplankton can provide ahabitat to endophytic bacteria living within algal cells andepiphytic bacteria that live in the phycosphere surround-ing algal cells (Cole, 1982). Marine bacteria in close asso-ciation with phytoplankton differ in composition amongalgal cells of different species and are distinct from thefree-living assemblage (Grossart et al., 2005; Jasti et al.,2005). Phytoplankton can also negatively affect popula-tions within the bacterial community through nutrient com-petition, antibiotic release and bactivory by mixotrophs(Cole, 1982; Nygaard and Tobiesen, 1993).

Our objective was to determine whether taxon-specificinteractions with phytoplankton contribute to temporalpatterns in lake bacterial community composition. Weselected three humic lakes where Kent and colleagues(2007) previously observed correlated phytoplankton andbacterial community dynamics as study sites. Local simi-larity analysis was used to detect bacterial operationaltaxonomic units (OTUs) whose relative abundances have

been correlated to the abundance of certain populationsof phytoplankton in weekly time-series observations fromeach lake. To experimentally assess the effect of phyto-plankton on bacterial community composition, bacterialcommunities from each lake were subsequently incu-bated with the phytoplankton assemblage from one ofthe three lakes or no phytoplankton (control) in a 7-daymesocosm experiment. If interactions between specificphytoplankton and bacterial populations drive bacterialcommunity dynamics, then: (i) correlated patterns ofabundance between certain phytoplankton taxa and bac-terial OTUs will be detected in environmental time-seriesobservations, (ii) bacterial communities will diverge incomposition when incubated with different phytoplanktonassemblages, and initially distinct bacterial communitieswill become more similar when incubated with the samephytoplankton assemblage, and (iii) bacterial OTUs thatrespond to particular phytoplankton species, e.g. Gym-nodinium fuscum, will increase (or decrease) in abun-dance when incubated with phytoplankton assemblagescontaining G. fuscum relative to no-phytoplankton controltreatments and will exhibit a seasonal pattern of relativeabundance that is positively (or negatively) correlated withG. fuscum abundance. After establishing the effect of phy-toplankton assemblages on bacterial community compo-sition, we used Illumina sequencing to determine whichbacterial genera were enriched in phytoplankton treat-ments relative to no-phytoplankton control treatments.

Results and discussion

Environmental correlations between algal andbacterial OTUs

Over 2 years of weekly environmental observationsbetween late May and mid-August in Crystal Bog (CB),South Sparkling Bog (SSB) and Trout Bog (TB), differ-ences in dominant phytoplankton taxa were observedacross lakes (Table 1). Some differences in phytoplanktontaxa were also observed between years within the samelake, though to a lesser extent than among-lake differ-ences (Table 1). Local similarity analysis identified 108significant positive correlations and 60 significant nega-tive correlations between the abundance of a specificalgal taxon (characterized morphologically) and the rela-tive abundance of a specific bacterial OTU (definedoperationally by unique ARISA fragment length) in the sixseasons of observation (three lakes, 2 years) (Fig. 1). Sixpairs of algal and bacterial taxa were positively correlatedand one algal–bacterial pair was negatively correlated inboth years within a given lake. One algal–bacterial pair,G. fuscum and bacterial ARISA fragment 684, was signifi-cantly positively correlated in SSB in 2003 and TB in2008, and exhibited a generally positive relationship infive of the six environmental time series (Fig. 2).

2 S. F. Paver et al.

© 2013 John Wiley & Sons Ltd and Society for Applied Microbiology, Environmental Microbiology

These observed taxon-level correlations support thehypothesis that interactions between specific phytoplank-ton and bacteria underlie observed correlations betweenphytoplankton and bacterial community dynamics (Kentet al., 2007). Correlations between the abundance of spe-cific phytoplankton taxa and bacterial relative abundance,including a positive correlation between G. fuscum andARISA fragment 684, have previously been detected inCB in 2002 (Newton et al., 2006). Two of the caveats to acorrelational approach are that not all algal–bacterialinteractions will be detected as correlations and thatcorrelations do not necessarily result from interactingpopulations. Correlations between the abundance ofG. fuscum and ARISA fragment 684 in time-series obser-vations (Fig. 2) illustrate how some algal–bacterial pairsmay be generally positively correlated, but not pass thesignificance threshold. Additionally, a bacterial taxa mayconsume exudates derived from multiple phytoplanktonspecies (Sarmento and Gasol, 2012), making it unlikely todetect correlation with specific phytoplankton species in amixed community. Population dynamics can also be con-trolled by other factors such as grazers, viruses and com-peting bacteria (Kent et al., 2006; Wei et al., 2011).Correlations do, however, provide an incomplete list ofpotentially interacting bacteria and phytoplankton that canthen be tested experimentally.

Bacterial response to phytoplankton manipulation

Bacteria from each of our study lakes were combined withphytoplankton from one of the three lakes or no phyto-plankton as a control. Initial bacterial community compo-sition in each mesocosm was primarily explained by thesource lake of the bacteria (CB, SSB or TB) and reflecteda minor influence of the phytoplankton treatment (CB,SSB, TB or control) (Table 2) (Fig. 3). Bacterial communi-ties in mesocosms with bacteria from CB were similar incomposition to those with bacteria from SSB, and com-paratively distinct from communities in mesocosmscontaining TB bacteria (average Bray-Curtis similarities:CB-SSB = 57%, CB-TB = 30%, SSB-TB = 32%). CB andSSB phytoplankton assemblages were dominated bySynura and contained similar taxa, while the phytoplank-ton assemblage from TB was distinct and contained taxathat were not detected or in low abundance in the otherlakes including Asterionella sp., G. fuscum and Mallo-monas sp. (Table 3).

Mesocosm microbial communities were contained in 10l LDPE cubitainers and incubated at the surface in thecentre of Crystal Bog where temperatures averaged21.8°C for 7 days (range: 20.0–23.6°C; hourly tempera-ture data from CB instrumented buoy: http://lter.limnology.wisc.edu/data/filter/10815). Following incuba-tion, the variation in mesocosm bacterial communityTa

ble

1.La

kech

arac

teris

tics

durin

gen

viro

nmen

talt

ime-

serie

ssa

mpl

ing

in20

03an

d20

08.

Lake

Latit

ude

Long

itude

Max

dept

hS

urfa

cear

ea(H

a)pH

Tota

lN(m

gl-1

)To

talP

(mg

l-1)

DO

C(m

gl-1

)P

hyto

plan

kton

taxa

with

high

est

biov

olum

e

Cry

stal

Bog

(CB

)46

°00′

26.8

″N89

°36′

22.5

″W2.

50.

565.

162

9(2

61)a

623

(221

)b22

(7)

28(2

3)9.

518

.2(1

4.6)

Syn

ura,

P.lim

batu

m,

P.um

bona

tum

,P.

cinc

tum

P.um

bona

tum

,G

.fus

cum

,S

ynur

a,P.

limba

tum

Sou

thS

park

ling

Bog

(SS

B)

46°0

0′13

.6″N

89°4

2′19

.9″W

8.0

0.44

5.1

642

(161

)a

585

(371

)b22

(20)

40(4

6)11

.215

.4(1

4.4)

P.lim

batu

m,

Syn

ura,

Mal

lom

onas

,G

.fus

cum

P.lim

batu

m,

G.f

uscu

m,

Syn

ura,

Trac

helo

mon

as

Trou

tB

og(T

B)

46°0

2′27

.5″N

89°4

1′09

.6″W

7.9

1.01

4.8

815

(237

)a

692

(314

)b32

(22)

50(1

15)

28.0

23.9

(16.

9)G

.fus

cum

,Ast

erio

nella

,M

allo

mon

as,

Cry

ptom

onas

G.f

uscu

m,

Cry

ptom

onas

,G

loeo

cyst

is,A

ster

ione

lla

Abb

revi

atio

ns:

tota

lN,

tota

lnitr

ogen

;to

talP

,to

talp

hosp

horu

s;D

OC

,di

ssol

ved

orga

nic

carb

on;

P.,

Per

idin

ium

;G

.,G

ymno

dini

um.

Num

bers

inpa

rent

hese

sre

pres

ent

the

rang

eof

obse

rved

valu

es(m

axim

umva

lue–

min

imum

valu

e).

a.F

irst

valu

ein

each

cell

is20

03da

tafr

omK

ent

and

colle

ague

s(2

007)

.b

.S

econ

dva

lue

inea

chce

llis

2008

data

.

Phytoplankton affect bacterial community dynamics 3


http://lter.limnology.wisc.edu/data/filter/10815

http://lter.limnology.wisc.edu/data/filter/10815

composition explained by phytoplankton treatmentincreased while the variation explained by bacterialsource lake decreased (Table 2). Bacteria incubated withphytoplankton from CB or SSB developed similarly,whereas incubation with the TB phytoplankton assem-blage had a unique effect on bacterial community compo-sition (Fig. 3).

Shifts in bacterial community composition in responseto incubation with various phytoplankton assemblagesprovide experimental evidence of phytoplankton assem-blages shaping bacterial community development. Thisevidence supports the hypothesis that phytoplankton suc-cession drives seasonal patterns in bacterial communitiespreviously observed in these lakes (Kent et al., 2007;

788 8 627 891547 603 790377389 369823917572

874

675 988749

Oocystis Euglena

528

485P. umb

484388

483

387 876

537 553540 684380 393638Mallomonas

925758 747 768 566 945 913 615

P. cinctum

600 650 840

744

320

Synura

Cryptomonas583763

596

780

598910

588

584

765

625494

545386

G.fuscum

383

821

UCCGloeocystisP. limbatum

881

Crucigenia

565

347

401 864

548

602

954

Scenedesmus792

794

558931

781

Dinobryon813

Cyclotella715

Asterionella

836 943

P. umb. umb.

385

888618

693606

Tabellaria

775805

Fig. 1. Network of all positive (solid lines) and negative (dotted lines) correlations between algal and bacterial taxa detected in CB, SSB andTB in 2003 and 2008. Thick lines indicate correlations observed in two seasons (e.g. CB 2003 and CB 2008, TB 2003 and SSB 2008).

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1 0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1 0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

Bac

teria

684

rela

tive

abun

danc

e(p

ropo

rtion

of m

axim

um)

Gymnodinium fuscum biovolume(proportion of maximum)

Bac

teria

684

rela

tive

abun

danc

e(p

ropo

rtion

of m

axim

um)

Bac

teria

684

rela

tive

abun

danc

e(p

ropo

rtion

of m

axim

um)

Gymnodinium fuscum biovolume(proportion of maximum)

Gymnodinium fuscum biovolume(proportion of maximum)A. B. C.

r=-0.26

r=0.51

r=0.88*

r=0.23 r=0.80

r=0.99*

Fig. 2. Relative abundance of bacterial ARISA fragment 684 as a function of the biovolume of phytoplankton Gymnodinium fuscum in CB (A),SSB (B) and TB (C) in 2003 (black squares) and 2008 (grey circles). Statistically significant r-values according to local similarity analysis areindicated by bold font and an asterisk.



Paver and Kent, 2010). The similar effects of CB and SSBphytoplankton and unique effect of TB phytoplankton islikely due to the similarity of the major phytoplankton taxain CB and SSB assemblages and the distinct assemblagecharacterized in TB (Table 3). Observed changes in bac-terial communities induced by the phytoplankton assem-blage corroborate prior observations of the effect ofphytoplankton on bacterial community development fol-lowing incubation with detritus from different algal speciesin lake water mesocosms (van Hannen et al., 1999) andincubation with phytoplankton assemblages dominated byeither diatoms or phytoflagellates in seawater meso-cosms (Pinhassi et al., 2004).

To better understand how phytoplankton assemblagesinfluenced bacterial communities, we classified bacterialOTUs according to whether they were abundant in the> 5 mm fraction of the community compared with the totalcommunity (> 0.22 mm) in each treatment as a proxy forphytoplankton attachment before and after incubation. Wethen determined what percentage of the post-incubationdifference between each treatment and the correspondingno-phytoplankton control could be explained by bacterialOTUs in each category (Table 4). Free-living bacteria,those with higher relative abundance in > 0.22 mm sub-samples than > 5 mm subsamples during the experiment,explained, on average, over 60% of the differencebetween treatments and the corresponding no-phytoplankton control. Phytoplankton-derived organicmatter including exudates likely facilitated the growth ofphytoplankton-enriched free-living taxa (Sarmento andGasol, 2012). Free-living bacteria that were depleted inrelative abundance in phytoplankton treatments com-pared with no-phytoplankton controls may have beendirectly negatively impacted by phytoplankton [e.g. preda-tion by mixotrophs (Stoecker, 1999), antibiotics (Cole,1982)] or indirectly negatively impacted by organisms withpositive growth responses to phytoplankton. Depletion inrelative abundance in phytoplankton treatments couldpotentially be an artefact of using relative abundancedata; bacteria with constant or even slightly increasingpopulation sizes may appear to decrease if other popula-

tions are increasing rapidly. Alternatively, in the absenceof labile phytoplankton-derived substrates in control treat-ments, phototrophic bacteria or bacteria that can utilizemore recalcitrant carbon pools may have had a competi-tive advantage, explaining an increase in relative abun-dance in control treatments.

Another 5.1% of the difference in bacterial communitycomposition between phytoplankton and control treat-ments was explained by bacterial OTUs that wereenriched in phytoplankton treatments and presumed tohave colonized algal cells during the experiment due totheir appearance in the > 5 mm fraction of phytoplanktontreatments following incubation. Phytoplankton-enrichedfree-living taxa consistently explained more of the differ-ences between phytoplankton and control treatmentsthan did phytoplankton-enriched colonizing bacteria,

Table 2. Results from PERMANOVA testing the effect of bacterial source lake (CB, SSB, TB) and phytoplankton treatment (CB, SSB, TB, control)on bacterial community composition before (day 0) and after (day 7) incubation.

dfSum ofsquares Pseudo-F P R2

Day 0 Bacteria 2 3.9 263.8 < 0.001 0.84Phytoplankton 3 0.8 36.7 < 0.001 0.17Bacteria*Phytoplankton 6 -0.2 -5.0 1.000 -0.05Residuals 24 0.2 0.04

Day 7 Bacteria 2 2.0 13.0 < 0.001 0.30Phytoplankton 3 1.8 8.0 < 0.001 0.28Bacteria*Phytoplankton 6 0.9 2.0 < 0.002 0.14Residuals 24 1.8 0.28

CBSSBTB

CBSSBTB

Bacteria

TimeDay 0Day 7

controlAxis 1

Axi

s 2

Phytoplankton

Stress=0.14

Fig. 3. Non-metric multidimensional scaling ordination of bacterialcommunities from Crystal (CB), South Sparkling (SSB) and Trout(TB) Bogs incubated with phytoplankton from CB, SSB, TB or ano-phytoplankton control before (day 0 – open symbols) and afterincubation (day 7 – filled symbols). Symbols represent the averagebacterial community composition in three replicate mesocosmscharacterized using ARISA community fingerprinting, and barsindicate standard error along each axis. Shape indicates bacterialsource, colour indicates source of the phytoplankton treatment.Before incubation, bacterial community similarity was determinedby bacterial source, as illustrated by solid ellipses. Followingincubation, the phytoplankton source had a pronounced effect onbacterial community composition as illustrated by dotted ellipses.



but the relative differentiating power of these groupsvaried depending on the specific treatment. Bacterialcommunities incubated with TB phytoplankton, whichexhibited the most distinct community composition post-incubation (Fig. 3), were particularly strongly differenti-ated from control treatments by phytoplankton-enrichedcolonizing bacteria. Replicate observations of putativecolonization of phytoplankton assemblages by specificbacterial OTUs can be explained by species-specificitybetween phytoplankton and their bacterial associates(Grossart et al., 2005; Jasti et al., 2005). Presumedphytoplankton-colonizing bacteria also potentially includebacteria that formed aggregates upon exposure toalgal extracellular products. Micro-aggregate formationoccurred when marine bacteria were incubated with dis-solved organic carbon from the diatom Chaetoceros sp.;however, aggregate formation was determined to beatypical since it did not occur when bacteria were incu-bated dissolved organic carbon from five other phyto-plankton species studied (Sarmento and Gasol, 2012).

Combining observational and experimental approachesat the OTU level

Fourteen of the bacterial OTUs that were positively cor-related with phytoplankton populations in environmentaltime-series observations were also enriched in meso-cosms following incubation with phytoplankton relative tono-phytoplankton controls (Table 5). One bacterial OTUwas negatively correlated with phytoplankton populationsin the environment and also decreased in relative abun-dance following incubation with phytoplankton relative tono-phytoplankton controls. Most of these bacterial OTUswere categorized as free-living, but four were identifiedas potential phytoplankton colonists and ARISA frag-ment 684 was phytoplankton-associated throughout theexperiment.

The observed correspondence between environmentaland experimental results at the OTU-level supports thehypothesis that phytoplankton affect the dynamics of spe-cific bacterial populations, thereby effecting change inbacterial community composition. Positive correlationsbetween bacterial OTUs and phytoplankton were morenumerous, repeatable between years, and overlapped toa greater degree with results from the mesocosm experi-ment than negative correlations, suggesting that facilita-tive effects of phytoplankton on bacteria are moreimportant that inhibitory effects in contributing to bacterialcommunity patterns in this system. While there were morefree-living bacterial taxa that exhibited consistentresponses to phytoplankton, phytoplankton-associatedbacteria also appear to contribute to bacterial communitydynamics. The relationship between phytoplankton-associated bacteria 684 and G. fuscum was the mostTa

ble

3.P

hyto

plan

kton

biov

olum

efr

omm

esoc

osm

sas

sem

bled

with

phyt

opla

nkto

nfr

omea

chla

keat

the

star

tof

the

incu

batio

n(m

m3

ml-1

�S

E).

Ast

erio

nella

Cry

ptom

onas

Din

obry

onG

.fus

cum

Glo

eocy

stis

Mal

lom

onas

P.ci

nctu

mP.

limba

tum

Syn

ura

CB

–2.

1�

1.3

¥10

45.

5�

2.4

¥10

2–

–4.

2�

4.2

¥10

2–

3.4

�1.

6¥

105

1.6

�0.

3¥

107

SS

B–

9.6

�2.

3¥

103

1.7

�0.

8¥

103

––

–8.

6�

1.8

¥10

46.

1�

1.4

¥10

67.

9�

0.8

¥10

6

TB

3.1

�0.

9¥

103

7.8

�0.

9¥

103

2.5

�0.

6¥

103

3.8

�0.

8¥

105

1.0

�0.

7¥

103

1.2

�0.

1¥

105

4.5

�4.

5¥

103

1.3

�1.

0¥

105

1.1

�1.

1¥

104

Abb

revi

atio

ns:

G.f

uscu

m,

Gym

nodi

nium

fusc

um;

P.ci

nctu

m,

Per

idin

ium

cinc

tum

;P.

limba

tum

,P

erid

iniu

mlim

batu

m.

Ooc

ystis

(2.1

�2.

1¥

102 )

was

dete

cted

inC

B,

Eug

lena

(4.7

�4.

7¥

102 )

was

dete

cted

inT

B.



consistent algal–bacterial relationship inferred fromobservations of these lakes (Fig. 2) (Newton et al., 2006).This consistency is likely due to bacteria 684 living in or onG. fuscum as suggested by the enrichment of bacteria684 in the phytoplankton-associated fraction. Šimek andcolleagues (2008; 2011) have also reported correspond-ence between observational and experimental results.Bacteria within the genus Limnohabitans increased inabundance during periods dominated by cryptophytes(Šimek et al., 2008) and isolates from two lineages withinthis genus exhibited enhanced growth when co-culturedwith certain phytoplankton including the cryptophyte Cryp-tomonas (Šimek et al., 2011). Taken together, the resultsfrom the current study and studies conducted by Šimekand colleagues demonstrate that phytoplankton affect therelative abundance of specific bacterial taxa and growthrate of bacterial strains and that these phytoplanktoninteractions scale up to effect seasonal patterns ofchange in bacterial communities.

Identification of bacterial taxa responding tophytoplankton manipulation

Illumina sequencing was used to identify bacterial taxaresponding to phytoplankton treatments in the mesocosmexperiment. Abundant bacterial clades in mesocosms,defined as those averaging greater than 1% of sequencesacross all treatments either before or after incubation,included representatives of the Alphaproteobacteria, Bet-aproteobacteria, Gammaproteobacteria, Actinobacteria,Bacteroidetes and Verrucomicrobia (Table 6). The twomost abundant clades were Pnec and acI-B. The betapro-teobacterial Pnec clade contains endosymbiotic as wellas free-living representatives of the Polynucleobacter

genus, which are ubiquitously detected in lakes and areecologically diverse (Vannini et al., 2007; Jezbera et al.,2011; Newton et al., 2011). The second most abundantgroup of sequences at the clade level was acI-B. Actino-bacteria within the acI lineage are also ubiquitous in fresh-water lakes (Newton et al., 2007).

A subset of identified taxa increased or decreased inrelative abundance in phytoplankton treatments whileremaining constant or exhibiting the opposite change inthe corresponding no-phytoplankton control (Fig. 4). Con-sistent enrichment of certain bacterial taxa was observedacross bacterial communities originating from differentlakes that were incubated with the phytoplankton assem-blage from a particular lake. For example, the proportionof Mitsuaria sequences increased in all mesocosms con-taining SSB phytoplankton, and Bdellovibrio increased inall mesocosms containing TB phytoplankton while notchanging or decreasing in no-phytoplankton controls.Notably, about half of the taxa that were enriched ordepleted in phytoplankton treatments could not be identi-fied to lineage or clade by the system described byNewton and colleagues (2011). Some of these identifiedgenera are members of minor freshwater lake phyla, whileothers belong to well-described freshwater phyla, butfamilies and genera that are not commonly detected infreshwater (Newton et al., 2011). Detection of taxa previ-ously not detected in freshwater is likely a result ofincreased sequencing depth facilitated by high-throughput sequencing approaches. Eiler and colleagues(2012) characterized lake environmental time-seriessamples using 454 pyrosequencing of 16S rRNA andfound that 38.3% of sequences could not be classified tothe described freshwater tribes and clades. Phytoplank-ton enrichment of atypical lake bacterial taxa suggests

Table 4. Per cent of the difference between each phytoplankton treatment and the corresponding no-phytoplankton control explained byphytoplankton-enriched or phytoplankton-depleted taxa categorized by their association with phytoplankton as determined by SIMPER analysis.

Bacterial source

CB phytoplankton SSB phytoplankton TB phytoplankton

CB SSB TB CB SSB TB CB SSB TB Avg.

Phytoplankton-enrichedPhyto-associated 18.7 8.0 8.6 2.4 6.2 3.8 4.0 8.3 0.2 6.7Phyto-colonizer 8.4 2.6 3.2 0.6 0.1 1.0 6.9 13.9 9.0 5.1Particle-associated 0.0 1.5 9.1 0.0 0.0 1.9 3.4 5.6 18.5 4.4Free-living 17.0 31.4 34.6 50.5 36.9 48.0 29.1 24.6 23.8 32.9Other 0.0 0.0 0.0 0.0 1.9 0.0 0.1 0.0 0.1 0.2

Phytoplankton-depletedPhyto-associated 3.7 5.5 0.0 5.8 7.1 0.2 5.7 4.3 1.2 3.7Phyto-colonizer 0.3 1.7 0.0 8.3 1.7 5.1 1.1 0.6 15.8 3.8Particle-associated 19.2 4.4 9.6 5.0 2.1 6.9 7.6 2.6 4.1 6.8Free-living 25.0 35.4 30.7 20.2 32.9 25.8 37.2 32.0 21.6 29.0Other 7.8 9.6 4.2 7.2 11.2 7.3 5.1 8.2 5.7 7.4

Definitions: Phyto-associated, higher in relative abundance in the > 5 mm subsample compared with the > 0.22 mm subsample on day 0;Phyto-colonizer, no difference in relative abundance on day 0 and higher in relative abundance in the > 5 mm subsample on day 7; Particle-associated, higher in relative abundance in the > 5 mm subsample in both phytoplankton treatment and no-phytoplankton control on day 7;free-living, higher in abundance in the > 0.22 mm subsample relative to the > 5 mm subsample; other, no difference in abundance detected between> 5 mm and > 0.22 mm subsamples of phytoplankton treatments.



Tab

le5.

Bac

teria

lOT

Us,

defin

edby

AR

ISA

frag

men

tle

ngth

(AF

L),

that

sign

ifica

ntly

incr

ease

d(+

)or

decr

ease

d(-

)ov

erth

em

esoc

osm

incu

batio

npe

riod

inph

ytop

lank

ton

trea

tmen

tsre

lativ

eto

no-p

hyto

plan

kton

cont

rols

and

who

seab

unda

nce

was

posi

tivel

y(+

)or

nega

tivel

y(-

)co

rrel

ated

with

abun

danc

eof

spec

ific

phyt

opla

nkto

nin

one

orm

ore

wee

kly

envi

ronm

enta

ltim

e-se

ries

obse

rvat

ions

ofC

B,

SS

Ban

dT

Bfr

omM

ay–A

ugus

t20

03an

d20

08.

AF

L

Mes

ocos

mex

perim

ent:

Phy

topl

ankt

onan

dba

cter

ialt

reat

men

tsE

nviro

nmen

talt

ime

serie

s:P

hyto

plan

kton

spec

ies

CB

phyt

opla

nkto

n

SS

B

phyt

opla

nkto

n

TB

phyt

opla

nkto

n

Cru

cige

nia

Cyc

lote

llaC

rypt

omon

asC

ST

Din

obry

onC

ST

Glo

ecys

tisT

G.

fusc

umT

Mal

lom

onas

(C)T

Ooc

ystis

(C)

P. cinc

tum

S(T

)

P. limat

umC

ST

P. umbo

natu

m

P. umb.

umb.

Syn

uraC

S(T

)Ta

bulla

riaC

BS

SB

TB

CB

SS

BT

BC

BS

SB

TB

389

+b+b

+49

4+

+-

--d

548

++

+-

+55

8+

+-

+56

5+

+-

+d

598

-+

-+

++

+-

627

-b+

684

+a+c

+d

744

++

+74

9+

+76

8+

-80

5+

+b+

821

+c-

+-

-88

1+

++

++

++

--

-89

1+

--

910

++

+-

+-

-91

7-

++

954

++

+-

-98

8+b

+c-

a.P

hyto

plan

kton

-ass

ocia

ted.

b.

Phy

topl

ankt

on-c

olon

izer

.c.

Par

ticle

-ass

ocia

ted.

d.

Cor

rela

tion

dete

cted

mul

tiple

times

.C

,S,T

Spe

cies

was

dete

cted

inph

ytop

lank

ton

asse

mbl

age

from

CB

,S

SB

and

TB

resp

ectiv

ely.

()D

etec

tion

inm

esoc

osm

phyt

opla

nkto

nas

sem

blag

eat

low

abun

danc

e.A

bbre

viat

ions

:C

B,

Cry

stal

Bog

;S

SB

,S

outh

Spa

rklin

gB

og;

TB

,Tr

out

Bog

;G

.fus

cum

,G

ymno

dini

um;

P.ci

nctu

m,

P.lim

batu

man

dP.

umbo

natu

m,

Per

idin

ium

.



that previously overlooked taxa may play significant eco-logical roles and contribute to temporal patterns of bacte-rial communities observed in lakes (Kent et al., 2007;Shade et al., 2007; Nelson, 2009).

Many of the phytoplankton-enriched bacterial taxa inthis study include host-associated organisms, with arange of hosts that include phytoplankton. Phytoplankton-enhanced growth has been observed for organismswithin the Limnohabitans (clade betI-A) (Šimek et al.,2011) and Campylobacter (Axelsson-Olsson et al.,2010) genera. Additionally, Acidovorax facilis has beendetected on diatom microaggregates (Knoll et al., 2001).Phytoplankton-enriched taxa also included a prevalenceof genera associated with oral and intestinal floraincluding Bacteroides, Capnocytophaga, Eikenella, Eu-bacterium, Faecalibacterium, Fusobacterium, Gemella,Kingella, Leptotrichia, Neisseria, Porphyromonas,Prevotella, Selenomonas, Tannerella and Veillonella(Christersson et al., 1991; Kamiya et al., 1993; Tyrrellet al., 2003; Rudney et al., 2005; Liu et al., 2008; Jiaet al., 2010). Other genera enriched in these mesocosmsinclude organisms previously associated with cock-roaches [Candidatus Rhabdochlamydia (Corsaro et al.,2007)], amoeba [Campylobacter (Axelsson-Olsson et al.,2010)], lichen [Methylobacterium (Hodkinson and Lutzoni,2009)] and plant roots [Azohydromonas and Ideonella(Coelho et al., 2008)]. Growth of enteric bacteria has pre-viously been observed to coincide with the bloom of analgal mat in a pristine stream (McFeters et al., 1978).These coliform bacterial isolates were then demonstratedto grow on filter-sterilized extracellular products ofChlorella. While a bacterial genus may include a variety oforganisms with distinct metabolic capabilities and eco-logical strategies, the prevalence of host-associatedorganisms with close relation to phytoplankton-enrichedbacteria suggests that organisms within certain lineageshave evolved to interact with a range of hosts.

While this study focused on demonstrating the effect ofphytoplankton on bacterial community dynamics, bacte-rial communities also have the potential to affect phyto-plankton community dynamics. Some genera enriched inphytoplankton treatments contain organisms with thatnegatively affect specific phytoplankton. Acidovorax con-tains a species capable algacidal activity (Kang et al.,2008), and both Bdellovibrio (Burnham et al., 1976) andAchromobacter (Wang et al., 2010) contain organismsthat have an algalytic effect on specific cyanobacteria.Enrichment of certain bacterial taxa may have positiveor negative feedback on the growth of particularphytoplankton similar to the framework developed forunderstanding interactions between plants and soil micro-organisms (Bever et al., 2010).

Limitations

There are potential limitations in our study design thatmerit discussion. ARISA community fingerprinting wasselected to characterize changes in OTU relative abun-dance across time-series observations and to link envi-ronmental and experimental observations. One drawbackto ARISA fingerprinting is that some fragment lengthsrepresent multiple, diverse taxa (Newton et al., 2006).Based on 16S rRNA + ITS clone libraries constructedfrom selected CB samples collected over a 3-year span,about 30% of ARISA fragment lengths were assigned tomultiple taxa (Newton et al., 2006). For the majority oftaxa in these lakes, using ARISA data to characterizepopulation dynamics is a useful approach. ARISA hasbeen shown to be comparable to a quantitative PCRapproach in its ability to represent relative abundance ofdominant Candidatus Accumulibacter clades (He et al.,2010). Strong correlations have also been observedbetween Prochlorococcus cell counts and abundanceprofiles from ARISA (Brown et al., 2005).

Table 6. Average relative abundance before (one replicate per treatment) and after (three replicates per treatment) incubation of the mostabundant bacterial clades detected in mesocosms.

Phylum Clade Linnaean classification

Average relative abundance

Day 0 Day 7

Alphaproteobacteria alfI-A Rhizobiales 0.005 0.021alfIV-A Novosphingobium 0.001 0.024

Betaproteobacteria betIII-A Alcaligenaceae 0.040 0.062Pnec Polynucleobacter 0.269 0.126

Gammaproteobacteria gamIII-A Acinetobacter 0.002 0.010Verrucomicrobia verI-A Xiphinematobacter 0.027 0.007Actinobacteria acI-A Actinomycetales 0.011 0.001

acI-B Actinomycetales 0.207 0.039acV-A Acidomicrobiales 0.033 0.001

Bacteroidetes bacI-A Sediminibacterium 0.004 0.021bacVI-A Mucilaginibacter 0.004 0.044

Clade classification and corresponding Linnaean classification to the most resolved level available is based on Newton and colleagues (2011).



A second limitation with ARISA, which is also applicableto the 16S rRNA sequencing approach used here, is theuse of relative abundance data. Since these relativeabundance approaches are not quantitative, it is possiblefor an OTU to decrease in relative abundance while main-taining a consistent population size or even increasing inabundance if other populations are increasing morerapidly. Despite the potential to misclassify the responsesof certain weakly responding OTUs, a relative abundance

approach is useful for identification of the bacterial OTUsthat are the most responsive to phytoplankton, and nec-essary for interpretation of observed patterns withinthe context of previous studies that have used relativeabundance approaches (e.g. Kent et al., 2007; Shadeet al., 2007; Nelson, 2009; Eiler et al., 2012). Additionally,based on previous experimental work conducted by Kentand colleagues (2006) in Crystal Bog using the samemesocosm containers and incubation time as the current

-0.5 0change in relative abundance

0.5

firm Eubacteriumbac Tannerella

bet Bergeriella

chl C. Rhabdochlamydia

ver MC18

bac PorphyromonasbacII Capnocytophagafuso LeptotrichiabetI Neisseria

epsilon Campylobacter

bet Eikenella

alfI Rhodoplanesfuso Fusobacteriumfirm Gemellabet Kingellafirm Veillonella

ac Actinomycesfirm Selenomonas

alfV-A Rhizobiales

alfI-B RhizobialesbetII RubrivivaxbetII Mitsuaria

firm Streptococcus

delta Bdellovibrio

bacI-A Sediminibacteriumacido C. Solibacterbet Ideonella

alfIII-A SphingomonasalfI MethylobacteriumbetIII Achromobacter

bac Prevotella

betI-A LimnohabitansbetI Acidovorax

betVII-B Herbaspirillum

gamIII-A Acinetobacter

CB SSB TB

Bacteria

Phytoplankton

CB

SS

BTB C

BS

SB

TB CB

SS

BTB C

BS

SB

TBClade GenusbetIII Azohydromonas

betI Aquitalea

gamII-A Serratiafirm ClostridiumbetI Paucibacter

betVII-A Oxalobacteraceae

bacI-B Chitinophagaceae

firm Faecalibacterium

firm RuminococcusbetI-B Rhodoferax

alf Caulobacter

betIII PigmentiphagabetIII TetrathiobacterbetIII-A Alcaligenaceaegam Aquamonas

thaum C. Nitrososphaerapnec Polynucleobacter

ac Corynebacterium

alf Telmatospirillumgam Actinobacillus

ControlFig. 4. Change in relative abundance overthe 7-day incubation (final relative abundance–initial relative abundance) of bacterial taxasignificantly enriched or depleted in at leastone phytoplankton treatment (CB, SSB, TB)relative to the corresponding no-phytoplanktoncontrol (control) based on EDGE analysis ofIllumina sequencing data. Circles indicatewhether each taxon was enriched byphytoplankton (green), depleted byphytoplankton (white), or both (half and half).Taxa were identified to clade using thenaming scheme described by Newton andcolleagues (2011) and genus using Linnaeanclassification. Where clade or genus nameswere not available, taxa were identified to themost resolved level available. Magnitude ofchange has been relativized acrosstreatments for each taxon. Order of taxa isbased on complete-linkage clustering.Abbreviations: C., Candidatus; alf,Alphaproteobacteria; bet, Betaproteobacteria;gam, Gammaproteobacteria; delta,Deltaproteobacteria; epsilon,Epsilonproteobacteria; ac, Actinobacteria;acido, Acidobacteria; bac, Bacteroidetes; chl,Chlamydiae; firm, Firmicutes; fuso,Fusobacteria; thaum, Thaumarchaeota; ver,Verrucomicrobia.



study, bacterial abundance is expected to remain rela-tively constant over time in the absence of added nutrients(Fig. S1), lessening the potential for OTU response mis-classification.

Phytoplankton collection for the mesocosm experimentusing filtration made it possible to use native phytoplank-ton assemblages, but it also introduced potential con-founding effects of including additional organisms. A smallnumber of rotifers identified as Keratella sp. (0.1 � 0.1cells ml-1) were retained in the 20–100 mm size fraction.Rotifers are filter feeders known to ingest bacteria andsmaller phytoplankton and may selectively graze bacteria> 0.5 mm (Arndt, 1993), which could decrease certainbacterial and algal populations. In addition to rotifers,phytoplankton treatments also contained bacteriaattached to phytoplankton. Phytoplankton-associatedbacterial OTUs that were higher in abundance in phyto-plankton treatments than no-phytoplankton controlsexplained 6.7% of the difference between phytoplanktonand no-phytoplankton treatments. While 6.7% is a not anegligible percentage, it does not invalidate the observa-tion of phytoplankton effects on bacterial community com-position and dynamics.

Effect of species interactions on planktoniccommunity dynamics

Experimental and observational results from this studysupport the hypothesis that taxon-specific interactionswith phytoplankton are an important ecological driverof bacterial community dynamics in lakes. Bacterialresponses to phytoplankton appear to be primarily posi-tive and to involve both loosely associated free-livingbacteria and, to a lesser degree, bacteria closely asso-ciated with phytoplankton. Our sequencing results cor-roborate the observations of Šimek and colleagues(2008; 2011) that the betI-A clade, which encompassesthe Limnohabitans genus, responds positively to incuba-tion with phytoplankton and contributes to correlatedpatterns of algal and bacterial community change.Sequencing results also identified additional taxa thatwarrant further investigation into their relationship withphytoplankton. From a community ecology perspective,this work provides an example of the importance ofbiotic drivers and potential for facilitation to structurecommunities. Future work is needed to determine howthe magnitude of the effect of phytoplankton assem-blages compares to other factors that influence temporalpatterns of bacterial communities [e.g. temperature(Shade et al., 2007; Adams et al., 2010)], how the effectof phytoplankton on bacterial communities may bealtered by interactions with the environment, and tocharacterize functional consequences of phytoplankton-induced shifts in bacterial communities.

Experimental procedures

Environmental observations

Integrated epilimnion samples were collected from CrystalBog (CB), South Sparkling Bog (SSB) and Trout Bog (TB),three hydrologically isolated humic lakes in the NorthernHighland Lake District of Wisconsin (Table 1). Sample col-lection occurred weekly from the end of May through Augustin 2003 and 2008 using PVC pipe integrated samplers asdescribed by Kent and colleagues (2004). Microorgani-sms were collected onto 0.22 mm filters (Supor-200; PallGelman, East Hills, NY, USA) and frozen at -20°C. Sub-samples were preserved in 2% glutaraldehyde for phyto-plankton identification and enumeration. Total N, total P andDOC were determined biweekly according to standard pro-tocols of the North Temperate Lakes Long Term EcologicalResearch project (http://lter.limnology.wisc.edu/research/protocols/) (Kent et al., 2007).

Algal exchange mesocosm experiment

Bacterial communities from CB, SSB and TB were incubatedwith native algal assemblages from CB, SSB, TB,or filter-sterilized and autoclaved CB water (as a no-phytoplankton control) in a full factorial design with threereplicates of each treatment. Microorganisms were collectedfrom CB, SSB and TB on 22 July 2009 using an integratedepilimnion sampler (Kent et al., 2004). Bacterial communi-ties from each lake were separated from larger aquaticorganisms by filtration through a 1 mm Polycap AS cartridgefilter (Whatman, Piscataway, NJ, USA). Native algal assem-blages were collected by filtering through a 100 mm nylonmesh (Spectrum Laboratories, Rancho Dominguez, CA,USA) to remove zooplankton and collecting, then rinsingalgal cells captured on a 20 mm nylon mesh with filter-sterilized and autoclaved CB water (Spectrum Laboratories),which allowed smaller organisms such as heterotrophicnanoflagellates and bacteria to pass through. A smallnumber of rotifers (0.1 � 0.1 cells ml-1) were also included in20–100 mm size fractions. Phytoplankton collected on the20 mm mesh were resuspended in 0.2 mm filter-sterilized,autoclaved CB water, concentrating phytoplankton from 50 lof lake water to 3 l of sterilized water. All combinations ofbacteria (5 l of 1 mm filtered water) from each lake and 0.3l of concentrated phytoplankton from each lake or ano-phytoplankton control (0.3 l of 0.2 mm filter-sterilized andautoclaved CB water) were combined in 10 l LDPE cubitain-ers (I-Chem, Rockwood, TN, USA) and incubated at thesurface in the centre of Crystal Bog for 7 days. A 500 mlsubsample was collected from each mesocosm on days 0, 3and 7. Microorganisms from 100 ml of subsamples of eachmesocosm were collected onto 0.22 mm filters (Supor-200;Pall Gelman, East Hills, NY, USA) for characterization of thewhole community, and 5 mm Isopore filters (Millipore, Bill-erica, MA, USA) for characterization of particle-associatedorganisms and frozen at -20°C. A 50 ml subsample waspreserved in 2% glutaraldehyde for phytoplankton enumera-tion. A 2 ml subsample was observed under a Leica MZ 12.5stereomicroscope at 50 ¥ magnification to confirm the pres-ence of living phytoplankton.



http://lter.limnology.wisc.edu/research/protocols/

http://lter.limnology.wisc.edu/research/protocols/

Microbial community analysis

Dominant phytoplankton species were identified microscopi-cally from glutaraldehyde-preserved pre-incubation samplesand samples collected before and after a 7-day pilot experi-ment (Table S1), and counts were transformed using biovol-ume estimates generated for species common to these lakes(Graham et al., 2004). DNA was extracted from filters usingFastDNA Spin Kits (MP Biomedicals, Solon, OH, USA). 2003time-series phytoplankton counts and DNA samples were thesame as those in Kent and colleagues (2007). The bacterialcommunity from environmental and experimental sampleswas characterized using Automated Ribosomal IntergeneicSpacer Analysis (ARISA) (Fisher and Triplett, 1999) usingconditions described by Kent and colleagues (2007). The16S–23S rRNA intergenic spacer region was PCR-amplifiedfrom 20 ng of each DNA sample with 6-FAM labelled, univer-sal 16S rRNA primer 1406F (5′-TGYACACACCGCCCGT-3′)and 23SR (5′-GGGTTBCCCCATTCRG-3′) targeting thebacterial 23S rRNA gene. The polymerase chain reactioncontained 1 mM deoxynucleoside triphosphates (Promega,Madison, WI, USA), 0.4 mM of each primer, PCR buffer with2.5 mM MgCl2 (Idaho Technology, Salt Lake City, Utah, USA)and 0.05 U ml-1 GoTaq (Promega). PCR cycles had an initialdenaturation at 94°C for 2 min, followed by 30 cycles of 94°Cfor 35 s, 55°C for 45 s and 72°C for 2 min, with a final exten-sion carried out at 72°C for 2 min carried out in an EppendorfMasterCycler Gradient (Eppendorf AG, Hamburg, Germany).Fluorescently labelled ARISA PCR amplicons were combinedwith a custom 100–1250 bp Rhodamine X-labelled internalsize standard (Bioventures, Murfreesboro, TN) and analysedby the Keck Center for Functional Genomics at the Universityof Illinois via denaturing capillary electrophoresis usingan ABI 3730XL Genetic Analyser (Applied Biosystems,Carlsbad, California, USA).

Size calling, profile alignment and grouping peaks into binsof operational taxonomic units were carried out usingGeneMarker version 1.75 (SoftGenetics, State College, PA,USA). To include the maximum number of peaks whileexcluding background fluorescence, a threshold of 500 fluo-rescence units was used. ARISA fragments known to corre-spond to chloroplasts were removed from the analysis. Thesignal strength of each peak was normalized to account forrun-to-run variations in signal detection by dividing the area ofindividual peaks by the total fluorescence (area) detected ineach profile.

Illumina sequencing

High-throughput Illumina sequencing of 16S rRNA geneswas conducted on one representative replicate from eachtreatment sampled on day 0 and all three replicates sampledon day 7, except in two treatments where one mesocosmcontainer was determined to leak following the experimentand the affected replicate was identified as an outlier. Illu-mina sequencing was conducted as described by Fagenand colleagues (2012). Briefly, the V4 region of the 16SrRNA gene was amplified from DNA samples using universalprimers 515F and 860R (Caporaso et al., 2011) with anadded barcode sequence and Illumina adapters. PCRcycles consisted of an initial denaturation at 94°C for 3 min,followed by 20 cycles of 94°C for 45 s, 50°C for 30 s and

65°C for 90 s, and a final elongation step of 65°C for 10 min.PCR products were purified using the Qiagen (Valencia, CA,USA) PCR purification kit according to the manufacturer’sprotocol. Sequencing was conducted on an Illumina IIx (Illu-mina, CA, USA) with two 101 base pair, paired read cycles.Sequences were first quality trimmed using a modifiedversion of Trim2 (Huang et al., 2003), and the first 11 basesof the primer region of each paired read were removed toprevent biases based on the degenerate bases in the primersequences. A minimum read length of 70 was required aftertrimming, and only complete pairs were used for analysis.Sequences from the multiplexed Illumina run were sortedinto their respective samples of origin based on theirbarcode sequences. The average number of reads persample was 35 323 and ranged from 1203 to 87 194 reads.The database used for 16S rRNA analysis was compiled bythe McMahon and Bertilsson research groups and containedfreshwater 16S rRNA sequences based on Ribosomal Data-base Project release 8.0 (Maidak et al., 2001), additionalfreshwater-specific sequences (Newton et al., 2011), and the97% operational taxonomic unit Greengenes databasedated 4 February 2011 (Werner et al., 2012). Freshwater16S rRNA sequences were classified according to thelineage/clade/tribe naming convention proposed by Newtonand colleagues (2011) while non-freshwater sequencesretained the conventional family/genus/species namingscheme. The database was formatted using TaxCollector(github.com/audy/taxcollector) (Giongo et al., 2010) andsequences were compared with the database using CLCreference assembly version 3.1 using paired reads param-eters and a global alignment. To classify the sequences,98% of each sequence had to align to a reference in thedatabase. An 80% similarity cut-off was applied at thephylum level and a 95% similarity cut-off was appliedat the clade/genus level. Sequences were uploadedto MG-RAST (Meyer et al., 2008) (IDs: 4508634.3-4508679.3).

Data analysis

Environmental time-series observations were analysed usinglocal similarity analysis (LSA) and Pearson correlations todetect correlated phytoplankton and bacterial taxa from aspecific lake in a given year (e.g. CB 2003) (Ruan et al.,2006). Significance of local similarity scores was based on1000 permutations. Correlations with local similarity scoresgreater than 0.3 and P-values < 0.001 were considered sig-nificant (Shade et al., 2010). LSA, Pearson correlations andcalculation of corresponding P- and q-values (calculatedto adjust for multiple comparisons) were performed usingthe LSA code developed by Ruan and colleagues (2006) inthe R statistical environment (R Development Core Team,2010). LSA networks were visualized using Cytoscape v.2.6.1 (Shannon et al., 2003). Correlated populations werethen compared among the six time series (three lakes, 2years) and to taxa enriched in algal addition mesocosmexperiments.

Pairwise Bray-Curtis similarities were calculated for everycombination of samples using ARISA relative fluorescence.Non-metric multidimensional scaling (MDS) was used todisplay Bray-Curtis similarities in multidimensional space



http://github.com/audy/taxcollector

using PRIMER version 6 (PRIMER-E, Plymouth MarineLaboratory, UK) (Clarke and Warwick, 2001). A two-waycrossed (orthogonal) permutational analysis of variance(PERMANOVA) was used to test the effect of bacterial andphytoplankton treatments on mesocosm bacterial communitycomposition before and after incubation. PERMANOVA is anon-parametric multivariate analysis of variance that gener-ates P-values using permutations (Anderson, 2001; McArdleand Anderson, 2001). PERMANOVA tests were run using theadonis function from the vegan package (Oksanen et al.,2011) in the R statistical environment (R Development CoreTeam, 2010). The SIMPER analysis in PRIMER version 6was used to determine taxa differentiating the > 5 mm fractionfrom the whole community (> 0.22 mm) for each treatment ondays 0, 3 and 7 and each phytoplankton assemblage treat-ment from the corresponding negative control on days 0 and7 (PRIMER-E, Plymouth Marine Laboratory, UK) (Clarke andWarwick, 2001).

Extraction of Differential Gene Expression (EDGE) soft-ware (version 1.1.291) was used to identify bacterial taxadisplaying differential responses in phytoplankton treatmentscompared with the corresponding no-phytoplankton control(Storey et al., 2005; Leek et al., 2006). Analyses in theEDGE package were designed for identification of differen-tially expressed genes across treatments, and have previ-ously been used to detect bacterial taxa with differentialresponses to incubation in epilimnion and hypolimnion lakelayers (Shade et al., 2010). Time-course differential analysiswas used to compare each treatment with the correspondingno-phytoplankton control over time (Leek et al., 2006). Anabundance table of Illumina sequence data classified atthe genus/clade level was subsampled using daisychop-per (http://www.genomics.ceh.ac.uk/GeneSwytch/) so thatsequencing depth was equal for each mesocosm samplewith the same bacterial source (Eiler et al., 2012). Unclas-sified sequences, determined as the number of sequencesclassified at the phylum level but not at the genus/cladelevel, were included in the subsampled table. Three repli-cates of each treatment on day 7 were compared with theinitial community (day 0). For analysis of ARISA communityfingerprints, relative fluorescence data from all replicates ondays 0, 3 and 7 were used and replicate was specified as acovariate. EDGE analyses were run using default param-eters of 1000 permutations and a significance threshold ofq < 0.10. The q-value is the false discovery rate analogueof the P-value, which is calculated to account for multiplecomparisons (Storey, 2002). Change in the abundance ofsubsampled Illumina sequences of bacterial taxa with sig-nificantly different response to phytoplankton and controltreatments was visualized by creating a heatmap using thegplots package in R (R Development Core Team, 2010). Tobe included in the heatmap, each clade or genus had to bedetected in at least two treatments on day 7, exhibit signifi-cant differences from the corresponding control in at leasttwo treatments as determined by EDGE, and be classifiedas enriched or depleted in at least one phytoplankton treat-ment. Enriched and depleted taxa were defined as thosethat increased or decreased (|change| > 0.005), respectively,in relative abundance in at least one phytoplankton treat-ment while exhibiting the opposite change in the corre-sponding negative control.

Acknowledgements

We thank N. Hibbard, E. Wendorf, J. Read, S. Reuter and S.Yeo for field assistance, T. Meinke and P. Montz for technicalassistance, and the UW Madison Trout Lake ResearchStation for logistical support. We thank E. Wendorf, J.Graham and L. Graham for phytoplankton count data; A.Shade and K. McMahon for 2008 DNA samples; C. Cáceres,D. Keymer, M. Lemke, W. Metcalf, R. Whitaker and A. Yan-narell for helpful discussion and B. Crary, L. Hu, J. Koval, D.Li, H. Lin, K. McMahon, A. Peralta, N. Youngblut and anony-mous reviewers for thoughtful comments on earlier versionsof this manuscript. Funding was provided by NSF Grant MCB0702653 and an O’Dell Fellowship from the Department ofNatural Resources and Environmental Sciences at the Uni-versity of Illinois to S.F.P.

References

Adams, H.E., Crump, B.C., and Kling, G.W. (2010) Tempera-ture controls on aquatic bacterial production and commu-nity dynamics in arctic lakes and streams. EnvironMicrobiol 12: 1319–1333.

Anderson, M.J. (2001) A new method for non-parametricmultivariate analysis of variance. Austral Ecol 26: 32–46.

Arndt, H. (1993) Rotifers as predators on components of themicrobial web (bacteria, heterotrophic flagellates, cilliates)– a review. Hydrobiologia 255: 231–246.

Arrieta, J.M., and Herndl, G.I. (2002) Changes in bacterialbeta-glucosidase diversity during a coastal phytoplanktonbloom. Limnol Oceanogr 47: 594–599.

Axelsson-Olsson, D., Olofsson, J., Svensson, L., Griekspoor,P., Waldenstrom, J., Ellstrom, P., and Olsen, B. (2010)Amoebae and algae can prolong the survival of Campylo-bacter species in co-culture. Exp Parasitol 126: 59–64.

Bertilsson, S., Eiler, A., Nordqvist, A., and Jorgensen, N.O.G.(2007) Links between bacterial production, amino-acid uti-lization and community composition in productive lakes.ISME J 1: 532–544.

Bever, J.D., Dickie, I.A., Facelli, E., Facelli, J.M., Klironomos,J., Moora, M., et al. (2010) Rooting theories of plant com-munity ecology in microbial interactions. Trends Ecol Evol25: 468–478.

Brown, M.V., Schwalbach, M.S., Hewson, I., and Fuhrman,J.A. (2005) Coupling 16S-ITS rDNA clone libraries andautomated ribosomal intergenic spacer analysis to showmarine microbial diversity: development and application toa time series. Environ Microbiol 7: 1466–1479.

Burnham, J.C., Stetak, T., and Locher, G. (1976) Extracellularlysis of blue-green alga phormidium-luridum by Bdellovi-brio bacteriovorus. J Phycol 12: 306–313.

Caporaso, J.G., Lauber, C.L., Walters, W.A., Berg-Lyons, D.,Lozupone, C.A., Turnbaugh, P.J., et al. (2011) Global pat-terns of 16S rRNA diversity at a depth of millions ofsequences per sample. Proc Natl Acad Sci USA 108:4516–4522.

Christersson, L.A., Zambon, J.J., and Genco, R.J. (1991)Dental bacterial plaques. J Clin Periodontol 18: 441–446.

Clarke, K.R., and Warwick, R.M. (2001) Change in MarineCommunities: An Approach to Statistical Analysis andInterpretation, 2nd edn. Plymouth, UK: PRIMER-E.



http://www.genomics.ceh.ac.uk/GeneSwytch/

Coelho, M.R.R., de Vos, M., Carneiro, N.P., Marriel, I.E.,Paiva, E., and Seldin, L. (2008) Diversity of nifH gene poolsin the rhizosphere of two cultivars of sorghum (Sorghumbicolor) treated with contrasting levels of nitrogen fertilizer.FEMS Microbiol Lett 279: 15–22.

Cole, J.J. (1982) Interactions between bacteria and algae inaquatic ecosystems. Annu Rev Ecol Syst 13: 291–314.

Corsaro, D., Thomas, V., Goy, G., Venditti, D., Radek, R., andGreub, G. (2007) ‘Candidatus Rhabdochlamydia crassifi-cans’, an intracellular bacterial pathogen of the cockroachBlatta orientalis (Insecta: Blattodea). Syst Appl Microbiol30: 221–228.

Eckert, E.M., Salcher, M.M., Posch, T., Eugster, B., and Pern-thaler, J. (2012) Rapid successions affect microbialN-acetyl-glucosamine uptake patterns during a lacustrinespring phytoplankton bloom. Environ Microbiol 14: 794–806.

Eiler, A., Heinrich, F., and Bertilsson, S. (2012) Coherentdynamics and association networks among lake bacterio-plankton taxa. ISME J 6: 330–342.

Fagen, J.R., Giongo, A., Brown, C.T., Davis-Richardson,A.G., Gano, K.A., and Triplett, E.W. (2012) Characteriza-tion of the relative abundance of the citrus pathogen Ca.Liberibacter asiaticus in the microbiome of its insect vector,Diaphorina citri, using high throughput 16S rRNA sequenc-ing. Open Microbiol J 6: 29–33.

Fisher, M.M., and Triplett, E.W. (1999) Automated approachfor ribosomal intergenic spacer analysis of microbial diver-sity and its application to freshwater bacterial communities.Appl Environ Microbiol 65: 4630–4636.

Fogg, G.E. (1983) The ecological significance of extracellularproducts of phytoplankton photosynthesis. Bot Mar 26:3–14.

Giongo, A., Davis-Richardson, A.G., Crabb, D.B., and Tri-plett, E.W. (2010) TaxCollector: modifying current 16SrRNA databases for the rapid classification at six taxo-nomic levels. Diversity 2: 1015–1025.

Graham, J.M., Kent, A.D., Lauster, G.H., Yannarell, A.C.,Graham, L.E., and Triplett, E.W. (2004) Seasonal dynamicsof phytoplankton and planktonic protozoan communities in anorthern temperate humic lake: diversity in a dinoflagellatedominated system. Microb Ecol 48: 528–540.

Grossart, H.P., Levold, F., Allgaier, M., Simon, M., andBrinkhoff, T. (2005) Marine diatom species harbour distinctbacterial communities. Environ Microbiol 7: 860–873.

van Hannen, E.J., Mooij, W.M., van Agterveld, M.P., Gons,H.J., and Laanbroek, H.J. (1999) Detritus-dependentdevelopment of the microbial community in an experimen-tal system: qualitative analysis by denaturing gradient gelelectrophoresis. Appl Environ Microbiol 65: 2478–2484.

He, S.M., Bishop, F.I., and McMahon, K.D. (2010) Bacterialcommunity and ‘Candidatus Accumulibacter’ populationdynamics in laboratory-scale enhanced biological phos-phorus removal reactors. Appl Environ Microbiol 76: 5479–5487.

Hodkinson, B.P., and Lutzoni, F. (2009) A microbiotic surveyof lichen-associated bacteria reveals a new lineage fromthe Rhizobiales. Symbiosis 49: 163–180.

Huang, X.Q., Wang, J.M., Aluru, S., Yang, S.P., and Hillier, L.(2003) PCAP: a whole-genome assembly program.Genome Res 13: 2164–2170.

Ishida, C.K., Arnon, S., Peterson, C.G., Kelly, J.J., and Gray,K.A. (2008) Influence of algal community structure on deni-trification rates in periphyton cultivated on artificial sub-strata. Microb Ecol 56: 140–152.

Jasti, S., Sieracki, M.E., Poulton, N.J., Giewat, M.W., andRooney-Varga, J.N. (2005) Phylogenetic diversity and spe-cificity of bacteria closely associated with Alexandrium spp.and other phytoplankton. Appl Environ Microbiol 71: 3483–3494.

Jezbera, J.J.J., Jezberova, J., Brandt, U., and Hahn, M.W.(2011) Ubiquity of Polynucleobacter necessarius subspe-cies asymbioticus results from ecological diversification.Environ Microbiol 13: 922–931.

Jia, W.J., Whitehead, R.N., Griffiths, L., Dawson, C., Waring,R.H., Ramsden, D.B., et al. (2010) Is the abundance ofFaecalibacterium prausnitzii relevant to Crohn’s disease?FEMS Microbiol Lett 310: 138–144.

Kamiya, S., Taniguchi, I., Yamamoto, T., Sawamura, S., Kai,M., Ohnishi, N., et al. (1993) Analysis of intestinal flora of apatient with congenital absence of the portal vein. FEMSImmunol Med Microbiol 7: 73–80.

Kang, Y.K., Cho, S.Y., Kang, Y.H., Katano, T., Jin, E.S., Kong,D.S., and Han, M.S. (2008) Isolation, identification andcharacterization of algicidal bacteria against Stephanodis-cus hantzschii and Peridinium bipes for the control offreshwater winter algal blooms. J Appl Phycol 20: 375–386.

Kent, A.D., Jones, S.E., Yannarell, A.C., Graham, J.M.,Lauster, G.H., Kratz, T.K., and Triplett, E.W. (2004) Annualpatterns in bacterioplankton community variability in ahumic lake. Microb Ecol 48: 550–560.

Kent, A.D., Jones, S.E., Lauster, G.H., Graham, J.M.,Newton, R.J., and McMahon, K.D. (2006) Experimentalmanipulations of microbial food web interactions in a humiclake: shifting biological drivers of bacterial communitystructure. Environ Microbiol 8: 1448–1459.

Kent, A.D., Yannarell, A.C., Rusak, J.A., Triplett, E.W., andMcMahon, K.D. (2007) Synchrony in aquatic microbialcommunity dynamics. ISME J 1: 38–47.

Knoll, S., Zwisler, W., and Simon, M. (2001) Bacterial colo-nization of early stages of limnetic diatom microaggre-gates. Aquat Microb Ecol 25: 141–150.

Lau, W.W.Y., and Armbrust, E.V. (2006) Detection of glyco-late oxidase gene glcD diversity among cultured and envi-ronmental marine bacteria. Environ Microbiol 8: 1688–1702.

Lau, W.W.Y., Keil, R.G., and Armbrust, E.V. (2007) Succes-sion and Diel transcriptional response of the glycolate-utilizing component of the bacterial community during aspring phytoplankton bloom. Appl Environ Microbiol 73:2440–2450.

Leek, J.T., Monsen, E., Dabney, A.R., and Storey, J.D. (2006)EDGE: extraction and analysis of differential gene expres-sion (vol 22, pg 507, 2006). Bioinformatics 22: 1412.

Liu, C., Finegold, S.M., Song, Y., and Lawson, P.A. (2008)Reclassification of Clostridium coccoides, Ruminococcushansenii, Ruminococcus hydrogenotrophicus, Ruminococ-cus luti, Ruminococcus productus and Ruminococcusschinkii as Blautia coccoides gen. nov., comb. nov., Blautiahansenii comb. nov., Blautia hydrogenotrophica comb.nov., Blautia luti comb. nov., Blautia producta comb. nov.,



Blautia schinkii comb. nov and description of Blautia wex-lerae sp nov., isolated from human faeces. Int J Syst EvolMicrobiol 58: 1896–1902.

McArdle, B.H., and Anderson, M.J. (2001) Fitting multivariatemodels to community data: A comment on distance-basedredundancy analysis. Ecology 82: 290–297.

McFeters, G.A., Stuart, S.A., and Olson, S.B. (1978)Growth of heterotrophic bacteria and algal extracellularproducts in oligotrophic waters. Appl Environ Microbiol35: 383–391.

Maidak, B.L., Cole, J.R., Lilburn, T.G., Parker, C.T., Saxman,P.R., Farris, R.J., et al. (2001) The RDP-II (RibosomalDatabase Project). Nucleic Acids Res 29: 173–174.

Maurin, N., Amblard, C., and Bourdier, G. (1997) Phytoplank-tonic excretion and bacterial reassimilation in an oligomes-otrophic lake: molecular weight fractionation. J PlanktonRes 19: 1045–1068.

Meyer, F., Paarmann, D., D’Souza, M., Olson, R., Glass,E.M., Kubal, M., et al. (2008) The metagenomics RASTserver – a public resource for the automatic phylogeneticand functional analysis of metagenomes. BMC Bioinfor-matics 9: 386.

Nelson, C.E. (2009) Phenology of high-elevation pelagic bac-teria: the roles of meteorologic variability, catchment inputsand thermal stratification in structuring communities. ISMEJ 3: 13–30.

Newton, R.J., Kent, A.D., Triplett, E.W., and McMahon, K.D.(2006) Microbial community dynamics in a humic lake:differential persistence of common freshwater phylotypes.Environ Microbiol 8: 956–970.

Newton, R.J., Jones, S.E., Helmus, M.R., and McMahon,K.D. (2007) Phylogenetic ecology of the freshwater Actino-bacteria acI lineage. Appl Environ Microbiol 73: 7169–7176.

Newton, R.J., Jones, S.E., Eiler, A., McMahon, K.D., andBertilsson, S. (2011) A guide to the natural history offreshwater lake bacteria. Microbiol Mol Biol Rev 75:14–49.

Nygaard, K., and Tobiesen, A. (1993) Bactivory in algae – asurvival strategy during nutrient limitation. Limnol Ocea-nogr 38: 273–279.

Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., O’Hara,R.B., Simpson, G.L., et al. (2011) vegan: CommunityEcology Package [WWW document]. URL http://CRAN.R-project.org/package=vegan.

Paver, S.F., and Kent, A.D. (2010) Temporal patterns inglycolate-utilizing bacterial community composition corre-late with phytoplankton population dynamics in humiclakes. Microb Ecol 60: 406–418.

Paver, S.F., Nelson, C.E., and Kent, A.D. (2013) Temporalsuccession of putative glycolate-utilizing bacterioplanktontracks changes in dissolved organic matter in a high-elevation lake. FEMS Microbiol Ecol 83: 541–551.

Pinhassi, J., Sala, M.M., Havskum, H., Peters, F., Guadayol,O., Malits, A., and Marrase, C.L. (2004) Changes in bac-terioplankton composition under different phytoplanktonregimens. Appl Environ Microbiol 70: 6753–6766.

R Development Core Team (2010) R: A Language and Envi-ronment for Statistical Computing. Vienna, Austria: R Foun-dation for Statistical Computing.

Ruan, Q.S., Dutta, D., Schwalbach, M.S., Steele, J.A.,

Fuhrman, J.A., and Sun, F.Z. (2006) Local similarity analy-sis reveals unique associations among marine bacterio-plankton species and environmental factors. Bioinformatics22: 2532–2538.

Rudney, J.D., Chen, R., and Zhang, G. (2005) Streptococcidominate the diverse flora within buccal cells. J Dent Res84: 1165–1171.

Sadro, S., Nelson, C.E., and Melack, J.M. (2011) Linking dielpatterns in community respiration to bacterioplankton in anoligotrophic high-elevation lake. Limnol Oceanogr 56: 540–550.

Sarmento, H., and Gasol, J.M. (2012) Use of phytoplankton-derived dissolved organic carbon by different types of bac-terioplankton. Environ Microbiol 14: 2348–2360.

Shade, A., Kent, A.D., Jones, S.E., Newton, R.J., Triplett,E.W., and McMahon, K.D. (2007) Interannual dynamicsand phenology of bacterial communities in a eutrophiclake. Limnol Oceanogr 52: 487–494.

Shade, A., Chiu, C.Y., and McMahon, K.D. (2010) Differentialbacterial dynamics promote emergent community robust-ness to lake mixing: an epilimnion to hypolimnion trans-plant experiment. Environ Microbiol 12: 455–466.

Shannon, P., Markiel, A., Ozier, O., Baliga, N.S., Wang, J.T.,Ramage, D., et al. (2003) Cytoscape: a software environ-ment for integrated models of biomolecular interaction net-works. Genome Res 13: 2498–2504.

Šimek, K., Hornak, K., Jezbera, J., Nedoma, J., Znachor, P.,Hejzlar, J., and Sed’a, J. (2008) Spatio-temporal patternsof bacterioplankton production and community compositionrelated to phytoplankton composition and protistan bacte-rivory in a dam reservoir. Aquat Microb Ecol 51: 249–262.

Šimek, K., Kasalicky, V., Zapomelova, E., and Hornak, K.(2011) Alga-derived substrates select for distinct betapro-teobacterial lineages and contribute to niche separation inLimnohabitans strains. Appl Environ Microbiol 77: 7307–7315.

Sommer, U., Gliwicz, Z.M., Lampert, W., and Duncan, A.(1986) The PEG-model of seasonal succession of plank-tonic events in fresh waters. Arch Hydrobiol 106: 433–471.

Stoecker, D.K. (1999) Mixotrophy among dinoflagellates. JEukaryot Microbiol 46: 397–401.

Storey, J.D. (2002) A direct approach to false discovery rates.J R Stat Soc Ser B Stat Methodol 64: 479–498.

Storey, J.D., Xiao, W.Z., Leek, J.T., Tompkins, R.G., andDavis, R.W. (2005) Significance analysis of time coursemicroarray experiments. Proc Natl Acad Sci USA 102:12837–12842.

Strickland, M.S., Osburn, E., Lauber, C., Fierer, N., and Brad-ford, M.A. (2009) Litter quality is in the eye of the beholder:initial decomposition rates as a function of inoculum char-acteristics. Funct Ecol 23: 627–636.

Teeling, H., Fuchs, B.M., Becher, D., Klockow, C., Gardebre-cht, A., Bennke, C.M., et al. (2012) Substrate-controlledsuccession of marine bacterioplankton populationsinduced by a phytoplankton bloom. Science 336: 608–611.

Tyrrell, K.L., Citron, D.M., Warren, Y.A., Nachnani, S., andGoldstein, E.J.C. (2003) Anaerobic bacteria cultured fromthe tongue dorsum of subjects with oral malodor. Anaerobe9: 243–246.

Vannini, C., Pockl, M., Petroni, G., Wu, Q.L.L., Lang, E.,Stackebrandt, E., et al. (2007) Endosymbiosis in statu nas-



http://CRAN.R-project.org/package=vegan

http://CRAN.R-project.org/package=vegan

cendi: close phylogenetic relationship between obligatelyendosymbiotic and obligately free-living Polynucleobacterstrains (Betaproteobacteria). Environ Microbiol 9: 347–359.

Wang, H., Liu, L., Liu, Z.P.P., and Qin, S. (2010) Investiga-tions of the characteristics and mode of action of an alga-lytic bacterium isolated from Tai Lake. J Appl Phycol 22:473–478.

Wei, Y., Kirby, A., and Levin, B.R. (2011) The population andevolutionary dynamics of Vibrio cholerae and its bacteri-ophage: conditions for maintaining phage-limited commu-nities. Am Nat 178: 715–728.

Werner, J.J., Koren, O., Hugenholtz, P., DeSantis, T.Z.,Walters, W.A., Caporaso, J.G., et al. (2012) Impact of train-ing sets on classification of high-throughput bacterial 16srRNA gene surveys. ISME J 6: 94–103.

Supporting information

Additional Supporting Information may be found in the onlineversion of this article at the publisher’s web-site:

Fig. S1. Total number of bacteria in mesocosms(average � standard error) with (+ NP) and without addednitrogen and phosphorus on days 1, 3 and 7 from experi-ments described by Kent and colleagues (2006) (unpublisheddata). This experiment was conducted in Crystal Bog usingthe same mesocosm containers and incubation length usedin the current study. The < 1 mm size fraction is analogous tothe no-phytoplankton control treatment in this study and< 243 mm size fraction is similar to phytoplankton assem-blage treatments. Without added nutrients, bacterial abun-dance remains relatively stable over the incubation period.Table S1. Initial and final phytoplankton cell counts (cellsml-1 � standard error) over a 7-day pilot experiment. No phy-toplankton cells were detected in initial samples collectedfrom no-phytoplankton control treatments.



interactions between specific phytoplankton and bacteria affect lake bacterial community succession

Documents