SHORT COMMUNICATION

Heterogeneity in the nutrient limitation of differentbacterioplankton groups in the EasternMediterranean Sea

Marta Sebastian and Josep M GasolDepartament de Biologia Marina i Oceanografia, Institut de Ciencies del Mar, CSIC, Barcelona,Catalunya, Spain

The heterotrophic bacterial community of the Eastern Mediterranean Sea is believed to be limited byphosphorus (P) availability. This observation assumes that all bacterial groups are equally limited,something that has not been hitherto examined. To test this hypothesis, we performed nutrientaddition experiments and investigated the response of probe-identified groups using microautor-adiography combined with catalyzed reporter deposition fluorescence in situ hybridization. Ourresults show contrasting responses between the bacterial groups, with Gammaproteobacteria beingthe group more affected by P availability. The Roseobacter clade was likely colimited by P andnitrogen (N), whereas Bacteroidetes by P, N and organic carbon (C). In contrast, SAR11 cells wereactive regardless of the nutrient concentration. These results indicate that there is highheterogeneity in the nutrient limitation of the different components of the bacterioplanktoncommunity.The ISME Journal (2013) 7, 1665–1668; doi:10.1038/ismej.2013.42; published online 14 March 2013Subject Category: Microbial ecology and functional diversity of natural habitatsKeywords: nutrient limitation; heterotrophic bacteria; phosphorus; Mediterranean Sea; MARFISH

Although inorganic nutrient limitation of bacterio-plankton was considered rare about a decade ago(Caron et al., 2000), we now know it frequentlyoccurs in oligotrophic waters. Phosphorus (P), forexample, is often the primary limiting nutrient inthe Atlantic and Mediterranean (Cotner et al., 1997;Pinhassi et al., 2006). Nevertheless, most studieshave considered heterotrophic bacteria as a homo-geneous black box, while the bacterial community iscomposed by cells expressing high metabolic diver-sity (for example, Musat et al., 2008; Alonso-Saezet al., 2012), which likely experience differentdegrees of limitation and stress. The fact thatnutrient availability plays an important role inniche partitioning supports this hypothesis(Pinhassi et al., 2006), but the variability in thestress responses among different bacterial groupshas been hitherto ignored.

The Eastern Mediterranean is one of the mostoligotrophic and P-starved marine systems on Earth(for example, Tanaka et al., 2007), where bacterio-plankton is often P-limited (for example, Thingstadet al., 2005). Here we assessed whether nitrogen (N)

and/or organic carbon (C) could be colimiting thebacterial community, and whether different bacter-ial groups responded similarly to the variousnutrient additions (see Supplementary material fordetails).

Heterotrophic bacterial activity was significantlystimulated (Po0.05, Dunnett’s test) in all the treat-ments that contained P (Figure 1). Bacterial activitydoubled in the P and PC treatments, and additionalincreases occurred in the NP (5� higher than thecontrol) and NPC treatments (10� higher). Thesefindings indicate that the heterotrophic bacterialcommunity as a whole was primarily limited by P,but that these waters are a nearly balanced system,where addition of P leads to shifts from one type oflimitation to another. This hypothesis is supportedby the results obtained with the phosphate turnovertime (Supplementary Figure S1).

To investigate whether all the bacterial groupsresponded equally to the nutrient additions, weused microautoradiography combined with cata-lyzed reporter deposition fluorescence in situ hybri-dization (MARFISH). Community composition 2days after inoculation was not strongly affected bythe nutrient amendments, and was similar to that atthe beginning of the experiment, that is, SAR11dominated in all the treatments, followed byGammaproteobacteria (Supplementary Figure S2).Non-EUB cells decreased dramatically in the NP andNPC treatments (from 20% to 0%), suggesting that

Correspondence: M Sebastian, Departament de Biologia Marina iOceanografia, Institut de Ciencies del Mar, CSIC, Pg Marıtim de laBarceloneta 37-49, Barcelona, Catalunya E08003, Spain.E-mail: msebastian@icm.csic.esReceived 1 August 2012; revised 2 February 2013; accepted 11February 2013; published online 14 March 2013

The ISME Journal (2013) 7, 1665–1668& 2013 International Society for Microbial Ecology All rights reserved 1751-7362/13

www.nature.com/ismej

dormant cells may have become active uponnutrient additions, as observed in other studies(for example, Eilers et al., 2000).

The different bacterial groups showed very con-trasting responses. Gammaproteobacteria activity(in terms of % cells taking up leucine) wasstimulated in all treatments containing P (Dunnett’stest, Po0.05, Figure 2). This group was the oneresponding more clearly to P additions, suggestingthat Gammaproteobacteria experienced more severeP limitation than the other groups. Indeed, recentmetatranscriptomic data show that Gammaproteo-bacteria transcript pools are enriched in genes forP acquisition when compared with other phyloge-netic groups (Gifford et al., 2012). Nevertheless, thefurther increase observed in the total number ofactive cells in the PC, NP and NPC treatments(Supplementary Figure S3) indicated that

Gammaproteobacteria were also able to rapidly takeadvantage of additional inputs of N and C, andincrease their abundance and production. Thestrong relationship found between the bulkincorporation of leucine and the abundance ofleucine-incorporating gammaproteobacterial cells(Supplementary Figure S4) suggests that Gamma-proteobacteria likely accounted for the activitychanges of the bulk bacterial community. MARFISHmicrographs showed that gammaproteobacterialcells had larger per-cell silver grain clusters thanthe dominant SAR11 (Figure S5), which indicateshigher assimilation per cell (Nielsen et al., 2003),and were notably larger in the nutrient-enrichedtreatments. Gammaproteobacteria were dominatedby the NOR5/OM60 clade in the control, P and Ntreatments, whereas Alteromonadaceae dominatedin the NPC treatments (Figure S6).

The percentage of cells taking up leucine withinthe Bacteroidetes group was generally low, due totheir known preference for high-molecular-weightcompounds (Cottrell and Kirchman, 2000). Theactivity of this group was only significantlystimulated in the NPC treatment, although onaverage the activity increased in all the treatmentsthat contained P. Therefore, Bacteroidetes cellswere likely limited by P, but secondarily colimitedby N and C.

Roseobacter activity was stimulated ca. twofold inthe P and PC treatments and eightfold and fivefoldin the NP and NPC treatments, respectively(Figure 2). However, this stimulation was onlysignificant in the NP and NPC treatments, suggest-ing that this group was colimited by P and N.Nevertheless, the Roseobacter clade dominatesamong algal-associated bacteria (Buchan et al.,2005), and phytoplankton at the moment of the

0

1

2Bacteroidetes

% a

ctiv

e of

tota

l pro

kary

otes

*

*

**

0

10

20

30

40Gammaproteobacteria

Cont. PC NPN NPC0

10

20

30

40SAR11

Cont. PC NP NPC

0

2

4

6

8

10Roseobacter

** *

**

% a

ctiv

e of

tota

l pro

kary

otes

P NP

Figure 2 Percentage of active cells within each probe-identified group in relation to total prokaryotes (DAPI counts). Labels as definedin Figure 1. Asterisks denote significant differences in relation to the control (*Po0.05, **Po0.001). Note the different y axis scale forRoseobacter and Bacteroidetes.

Leuc

ine

inco

rpor

atio

n(n

mol

Leu

h-1

L-1)

0

0.2

0.4

0.6

0.8

1

1.2

806040200

Time (h)

Cont.PNPCNPNPC

Figure 1 Bacterial heterotrophic activity in the addition experi-ments. Samples were amended with phosphate (P), ammonia (N)and organic carbon (C), or with combinations of these nutrients(PC, NP and NPC). Cont.: control treatment (no amendments).Each data point represents the average of the two replicates. Errorbars represent the s.d. Arrow highlights the time-point whensamples were taken for the MARFISH analyses.

Heterogeneous limitation of bacterioplankton groupsM Sebastian and JM Gasol

1666

The ISME Journal

study was colimited by P and N (Pitta et al.,unpublished). Hence, it is possible that the strongpositive response of this group to N and P additionsis a consequence of dissolved organic carbon releaseby the stimulated phytoplankton.

SAR11 activity, expressed as percentage of activecells, was relatively high in all the treatments(Figure 2). The fact that this group was activeregardless of the nutrient concentration contradictsthe common belief that SAR11 cells are strictoligotrophs. Instead, we believe that the conserva-tive metabolism of SAR11 cells allows them tothrive well in a wide range of trophic conditions.This is consistent with a recent metatranscriptomicstudy showing that SAR11 cells have limitedcapacity to sense and respond to environmentalchanges, which suggests that this phylogeneticgroup has evolved to maintain consistent growthindependent of environmental conditions (Giffordet al., 2012), something that had already been hintedin studies with a cultured representative of thisclade (for example, Rappe et al., 2002). Thecell-associated silver grain clusters were not largerin the nutrient amended treatments (Figure S5),suggesting that the per-cell activity within theSAR11 group did not notably increase, again agreeingwith the view that SAR11 cells would not takeadvantage of nutrient pulses (Gifford et al., 2012) andin general have low nutrient requirements (Sebastianet al., 2012). Yet, it should be noted that at leastcertain SAR11 phylotypes may display high growthrates (Malmstrom, et al., 2005, Campbell et al., 2011).

For many decades, the Liebig’s law of the mini-mum, which states that only a single resource islimiting, was the dominant theory shaping howoceanographers viewed phyto- and bacterioplanktonecology and their impact on biochemical cycles(Arrigo, 2004). This view has been recently chal-lenged by the concept of colimitation or multiplenutrient limitations (Saito et al., 2008). Here weshow that the reality is even more complex, withdifferent bacterial groups experiencing differenttypes of limitations under the same environmentalconditions.

Conflict of Interest

The authors declare no conflict of interest.

Acknowledgements

This experiment in the Cretacosmos facility was arrangedin co-operation by project Nutritunnel financed by theResearch Council of Norway and the MESOAQUAnetwork (FP7/2007-2013, grant agreement no. 228224).We thank Frede T Thingstad and Paraskevi Pitta forinviting MS to participate in this experiment and RunarThyrhaug for performing the heterotrophic bacterialactivity measurements. Processing of the samples wassupported by the grants FOSMICRO (CTM2009-07679-E)and STORM (CTM2009-09352/MAR), funded by the

former Spanish Ministry of Science and Innovation. MSwas supported by a ‘Juan de la Cierva’ award.This paper isdedicated to the memory of Runar Thyrhaug who passedaway shortly after the experiment.

References

Alonso-Saez L, Sanchez O, Gasol JM. (2012). Bacterialuptake of low molecular weight organics in thesubtropical Atlantic: are major phylogenetic groupsfunctionally different? Limnol Oceanogr 57: 798–808.

Arrigo KR. (2004). Marine microorganisms and globalnutrient cycles. Nature 437: 349–355.

Buchan A, Gonzalez JM, Moran MA. (2005). Overview ofthe marine Roseobacter lineage. Appl Environ Micro-biol 71: 5665–5677.

Campbell BJ, Yu L, Heidelberg JF, Kirchman DL. (2011).Activity of abundant and rare bacteria in a coastalocean. Proc Natl Acad Sci USA 108: 12776–12781.

Caron DA, Lim EL, Sanders RW, Dennett MR,Berninger UG. (2000). Responses of bacterioplanktonand phytoplankton to organic carbon and inorganicnutrient additions in contrasting oceanic ecosystems.Aquat Microb Ecol 22: 175–184.

Cotner JB, Ammerman JW, Peele ER, Bentzen E. (1997).Phosphorus-limited bacterioplankton growth in theSargasso Sea. Aquat Microb Ecol 13: 141–149.

Cottrell MT, Kirchman DL. (2000). Natural assemblages ofmarine proteobacteria and members of the Cytophaga-Flavobacter cluster consuming low-and high-molecu-lar-weight dissolved organic matter. Appl EnvironMicrobiol 66: 1692–1697.

Eilers H, Pernthaler J, Amann R. (2000). Succession ofpelagic marine bacteria during enrichment: a closelook at cultivation-induced shifts. Appl EnvironMicrobiol 66: 4634–4640.

Gifford SM, Sharma S, Booth M, Moran MA. (2012).Expression patterns reveal niche diversification in amarine microbial assemblage. ISME J 7: 281–298.

Malmstrom RR, Cottrell MT, Elifantz H, Kirchman DL.(2005). Biomass production and assimilation of dis-solved organic matter by SAR11 bacteria in theNorthwest Atlantic Ocean. Appl Environ Microbiol71: 2979.

Musat N, Halm H, Winterholler B, Hoppe P, Peduzzi S,Hillion F et al. (2008). A single-cell view on theecophysiology of anaerobic phototrophic bacteria.Proc Natl Acad Sci USA 105: 17861.

Nielsen JL, Christensen D, Kloppenborg M, Nielsen PH.(2003). Quantification of cell-specific substrateuptake by probe-defined bacteria under in situconditions by microautoradiography and fluore-scence in situ hybridization. Environ Microbiol 5:202–211.

Pinhassi J, Gomez-Consarnau L, Alonso-Saez L, Sala MM,Vidal M, Pedros-Alio C et al. (2006). Seasonalchanges in bacterioplankton nutrient limitation andtheir effects on bacterial community composition inthe NW Mediterranean Sea. Aquat Microb Ecol 44:241–252.

Rappe MS, Connon SA, Vergin KL, Giovannoni SJ. (2002).Cultivation of the ubiquitous SAR 11 marine bacter-ioplankton clade. Nature 418: 630–633.

Heterogeneous limitation of bacterioplankton groupsM Sebastian and JM Gasol

1667

The ISME Journal

Saito MA, Goepfert TJ, Ritt JT. (2008). Some thoughts onthe concept of colimitation: three definitions and theimportance of bioavailability. Limnol Oceanogr 53:276–290.

Sebastian M, Pitta P, Gonzalez JM, Thingstad TF, Gasol JM.(2012). Bacterioplankton groups involved in the uptakeof phosphate and dissolved organic phosphorus in amesocosm experiment with P-starved Mediterraneanwaters. Environ Microbiol 14: 2334–2347.

Tanaka T, Zohary T, Krom MD, Law CS, Pitta P,Psarra S et al. (2007). Microbial communitystructure and function in the Levantine Basin ofthe eastern Mediterranean. Deep Sea Res I 54:1721–1743.

Thingstad TF, Krom MD, Mantoura RFC, Flaten GAF,Groom S, Herut B et al. (2005). Nature of phosphoruslimitation in the ultraoligotrophic eastern Mediterra-nean. Science 309: 1068–1071.

Supplementary Information accompanies this paper on The ISME Journal website (http://www.nature.com/ismej)

Heterogeneous limitation of bacterioplankton groupsM Sebastian and JM Gasol

1668

The ISME Journal

SUPPLEMENTARY  MATERIAL    

Methods  Incubations  were  performed  in  2L  duplicate  polycarbonate  bottles  (Nalgene)  

that   were   suspended   inside   a   large   concrete   tank   with   running   water   that  controlled   temperature.   The   experiment   lasted   3   days.   Nutrients  were   added   to  final  concentrations  of  10  µM  C  (glucose),  1.6  µM  N  (NH4Cl)  and  0.1  µM  P  (KH2PO4)  to   yield   the   following   treatments:   control   (with   no   additions),   +P,   +N,   +PC,   +NP  +NPC.    

Bacterial  heterotrophic  activity  was  determined  in  all  mesocosms  every  day  by   incorporation   of   tritium-­‐labeled   leucine   (Kirchman   et   al.,   1985)   using   the  centrifugation  procedure.  Triplicate  samples  and  one  prefixed  control  sample  were  incubated  with  3H-­‐Leucine  (4.27  TBq  mmol-­‐1,  Perkin  Elmer,  Boston,  USA)  at  a  final  concentration   of   60   nM.   Incubation   was   performed   in   the   dark   at   in   situ  temperature   for  1  h  and  stopped  with  5%  TCA,   final   concentration.  The  samples  were  then  centrifuged  at  16000  x  g  for  10  min  before  removal  of  the  supernatant.,  were  washed  twice  by  adding  5%  TCA,  vortexed,  centrifuged  and  the  supernatant  removed.   Counting   cocktail   (Ecoscint   A,   National   Diagnostics,   Atlanta,   USA)  was  added  and  the  incorporation  of  radioactive  leucine  measured  by  liquid  scintillation  counting.    

Turnover   time   of   phosphate   (Pi)  was   estimated   using  33PO43-­‐.  H333PO4   (40-­‐158Ci  mg-­‐1;  Perkin  Elmer)  was  diluted  in  distilled  water  and  15-­‐µL  aliquots  were  added  to  duplicate  9-­‐mL  subsamples  to  give  final  concentrations  ranging  between  25  and  108  pM.  Duplicate  killed  controls  were  included  with  each  set  of  samples.  Killed  controls  were  amended  with  paraformaldehyde  (2%  final  concentration)  30  min   before   the   addition   of   the   isotopic   tracer.   Incubations  were   done   in   15-­‐mL  Falcon   tubes   at   room   temperature   and   subdued   light.   The   duration   of   each  incubation  varied  depending  on  the  expected  turnover  time,  ranging  from  20  min  to  2  h.  Incubations  were  terminated  by  the  addition  of  paraformaldehyde  (2%  final  concentration)   and   filtered,   within   30   min,   onto   a   0.2   µm   polycarbonate   filter,  which  was  placed  on  top  of  a  Whatman  (GF/C)  glass  fiber  filter  saturated  with  100  mmol   L-­‐1   KH2PO4.   To   stop   incubations,   fixation   was   chosen   over   cold-­‐chase  addition  of  cold  PO43-­‐  to  be  consistent  with  the  methodology  employed  during  the  MARFISH  analyses  (see  below).  Fixation  has  an  experimental   limitation,  which   is  that   some   of   the   accumulated   33P   could   leak   out   of   fixed   cells   after   the   cell  membranes   become   compromised.   Talarmin   et   al.   (2011)   recently   reported   that  up   to   40%   of   the   33Pi   label   is   lost   from   fixed   heterotrophic   bacterial   cells  immediately   after   fixation   and   Casey   et   al.   (2009)   estimated   that   ∼25%   of   the  isotope  in  Pi  incubations  could  leak  out  from  the  cells  within  24h.  To  minimize  this  leakage,   samples   were   filtered   within   1h   after   stopping   the   incubation.   After  filtration,  the  filters  were  rinsed  twice  with  sterile  Milli-­‐Q  water  and  transferred  to  scintillation   vials   with   1   mL   Ultima   Gold   scintillation   cocktail.   Aliquots   (50   µL)  from   the   subsamples   incubated   with   33PO43-­‐   were   transferred   directly   to  scintillation  vials  and  mixed  with  1  mL  scintillation  cocktail   to  measure   the   total  added   radioactivity.   Samples   were   radioassayed   in   a   Packard   Tri-­‐Carb   4000  scintillation  counter.  Turnover  times  were  calculated  using  the  equation  T  =  t  /  [-­‐ln  (1-­‐R)],   where   t   =   incubation   time   and   R=   consumed   fraction   of   added   tracer  (Thingstad  et  al.,  1993).  Killed  controls  were  subtracted  prior  to  the  calculations.    

 

 For   MARFISH   analyses   30-­‐mL   subsamples   were   spiked   with   3H-­‐leucine  

(Perkin  Elmer)  to  yield  0.5  nM.  The  samples  were  incubated  for  2.5  h.  One  simple  was   killed   with   paraformaldehyde   before   the   addition   of   the   radiolabeled  substrate  and  was  used  as  a   control.  At   the  end  of   the   incubation,   samples  were  fixed  with  paraformaldehyde,  allowed  to  sit  in  the  dark  for  at  least  one  hour,  and  then  portions  of  5-­‐10-­‐mL  were  filtered  onto  three  different  0.2  µm  polycarbonate  filters.   Filters  were  washed   twice  with   sterile  Milli-­‐Q  water   and   frozen   at   -­‐80°C  until  processing  in  the  lab.  Filters  were  then  hybridized  following  the  CARD-­‐FISH  protocol   (Pernthaler   et   al.,   2002)   to   identify   the  different   bacterial   groups.  After  thawing,   the   filters   were   dipped   in   0.1%   agarose,   dried   at   37°C,   and   then  dehydrated  with  95%  ethanol.  This  allowed  attachment  of   the  cells   to   the   filters.  Then,   cell  walls  were   permeabilized  with   lysozyme   (1   h)   and   achromopeptidase  (30  min)  at  37°C.  Filters  were  cut  into  multiple  pieces  and  hybridized  with  one  of  the   following  horseradish  peroxidase   (HRP)-­‐labeled  probes:  EUB338   I-­‐II  and  –III  (targets  most  Eubacteria,  Daims  et  al.,  1999),  GAM42a  together  with  its  unlabeled  competitor  probe  (targets  most  Gammaproteobacteria,  Manz  et  al.,  1992),  CF319a  (targets   many  members   of   the   Bacteroidetes   group,   Manz   et   al.,   1996),   ROS537  (targets  members  of  the  Alphaproteobacteria  Roseobacter-­‐Sulfitobacter-­‐Silicibacter  group,   Eilers   et   al.,   2000),   SAR11-­‐441R   (targets   the  Alphaproteobacteria   SAR11,  Morris   et   al.,   2002),   ALT1413   (targets  Alteromonadaceae,   Eilers   et   al.,   2000),   or  NOR5-­‐730   (targets   the   NOR5/OM60   clade,   Eilers   et   al.,   2000).   Specific  hybridization   conditions   were   established   by   addition   of   formamide   to   the  hybridization  buffers   (45%   formamide   for   the   SAR11  probe,   50%   for   the  NOR5-­‐730   probe,   60%   for   the   ALT1413   probe,   and   55%   for   the   other   probes).  Hybridization   was   performed   overnight   at   35°C.   For   amplification,   we   used  tyramide  labeled  with  Alexa  488.  After  processing,  a  small  portion  of  the  filter  was  cut  and  stained  with  4’,6-­‐diamidino-­‐2-­‐phenylindole  (DAPI,  final  concentration  1  µg  mL-­‐1)  to  quantify  the  abundance  of  the  different  phylogenetic  groups  in  relation  to  total   prokaryotic   counts.   The   rest   of   the   filter   was   glued   onto   a   glass   slide   and  subsequently   processed   for   microautoradiography   as   described   in   detail   in  Alonso-­‐Sáez  and  Gasol  (2007),  which  is  a  modification  of  the  protocol  described  by  Alonso   and   Pernthaler   (2005).   Exposure   times   were   determined   empirically   by  following   changes   in   number   of   cells   taking   up   the   substrate   over   time.  Optimal  exposure   times   were   selected   once   the   number   of   cells   taking   up   the   substrate  reached   a   plateau   but   accumulation   of   silver   grains   still   allowed   visualization   of  the   cells   associated   to   them.   Cells   were   counted   in   an   Olympus   BX61  epifluorescence   microscope.   Cells   touching   or   overlapping   silver   grains   after  developing   of   the   emulsion   were   considered   as   active   cells   or   MAR+   cells.   For  abundance  of  probe-­‐positive  cells,  between  500  and  1000  DAPI-­‐positive  cells  were  counted  manually   in  a  minimum  of  10  fields.  Killed  controls  were  evaluated  with  the  probe  EUB338  I-­‐II  and  –III.  The  proportion  of  labeled  cells  in  the  killed  controls  was  2%.  This  proportion  was  not  subtracted  from  the  percent  of  cells  taking  up  3H-­‐leucine  in  the  live  incubations.    

References  Alonso-­‐Saez,  L.,  and  Gasol,   J.M.   (2007)  Seasonal  variation   in   the  contribution  of  different  bacterial  groups  to  the  uptake  of  low  molecular  weight-­‐compounds  in  NW  Mediterranean  coastal  waters.  Appl  Environ  Microbiol  73:  3528-­‐3535.  

Alonso,  C.,  and  Pernthaler,   J.   (2005)  Incorporation  of  glucose  under  anoxic  conditions  by  bacterioplankton  from  coastal  North  Sea  surface  waters.  Appl  Environ  Microbiol  71:  1709-­‐1716.  Daims,  H.,  Brühl,  A.,  Amann,  R.,  Schleifer,  K.H.,  and  Wagner,  M.  (1999)  The  Domain-­‐specific  Probe  EUB338  is  Insufficient  for  the  Detection  of  all  Bacteria:  Development  and  Evaluation  of  a  more  Comprehensive  Probe  Set.  Syst  Appl  Microbiol  22:  434-­‐444.  Eilers,   H.,   Pernthaler,   J.,   Glockner,   F.O.,   and   Amann,   R.   (2000)   Culturability   and   in   situ  abundance  of  pelagic  bacteria  from  the  North  Sea.  Appl  Environ  Microbiol  66:  3044  -­‐3051.    Kirchman,  D.,  K'nees,  E.,  and  Hodson,  R.  (1985)  Leucine  incorporation  and  its  potential  as  a   measure   of   protein   synthesis   by   bacteria   in   natural   aquatic   systems.   Appl   Environ  Microbiol  49:  599-­‐607.  Manz,  W.,  Amann,  R.,  Ludwig,  W.,  Vancanneyt,  M.,  and  Schleifer,  K.H.  (1996)  Application  of  a  suite  of  16S  rRNA-­‐specific  oligonucleotide  probes  designed  to  investigate  bacteria  of  the  phylum   Cytophaga-­‐Flavobacter-­‐Bacteroides   in   the   natural   environment.   Microbiology  142:  1097-­‐1106.  Manz,   W.,   Amann,   R.,   Ludwig,   W.,   Wagner,   M.,   and   Schleifer,   K.H.   (1992)   Phylogenetic  oligodeoxynucleotide   probes   for   the   major   subclasses   of   proteobacteria:   problems   and  solutions.  Syst  Appl  Microbiol  15:  593-­‐600.  Morris,   R.M.,   Rappé,   M.S.,   Connon,   S.A.,   Vergin,   K.L.,   Siebold,   W.A.,   Carlson,   C.A.,   and  Giovannoni,   S.J.   (2002)   SAR11   clade   dominates   ocean   surface   bacterioplankton  communities.  Nature  420:  806-­‐810.  Pernthaler,  A.,  Pernthaler,  J.,  and  Amann,  R.  (2002)  Fluorescence  in  situ  hybridization  and  catalyzed   reporter   deposition   for   the   identification   of   marine   bacteria.   Appl   Environ  Microbiol  68:  3094-­‐3101.  Talarmin,  A.,  Van  Wambeke,  F.,  Duhamel,  S.,  Catala,  P.,  Moutin,  T.,  and  Lebaron,  P.  (2011)  Improved  methodology  to  measure  taxon-­‐specific  phosphate  uptake  in  live  and  unfiltered  samples.  Limnol.  Oceanogr.:  Methods  9:  443-­‐453  Thingstad,   T.F.,   Skjoldal,   E.F.,   and   Bohne,   R.A.   (1993)   Phosphorus   cycling   and   algal-­‐bacterial  competition  in  Sandsfjord,  western  Norway.  Mar  Ecol  Prog  Ser  99:  239-­‐259.  

     

                                   

 SUPPLEMENTARY  FIGURES    Figure  S1.-­‐  RESPONSE  OF  THE  TURNOVER  TIME  OF  PHOSPHATE  TO  NUTRIENT  

ADDITIONS  IN  P-­‐STARVED  MEDITERRANEAN  WATERS    To   assess   the   bulk   community   response   to   nutrient   additions  we   first   estimated  the   turnover   time  of  phosphate   (Pi)   to   see  how   fast   this  nutrient  was  utilized   in  these  P-­‐starved  waters,  and  how  the  different  treatments  affected  its  utilization.      

   

 Figure  S1  legend.  Phosphate  turnover  time  in  the  different  treatments  on  day  1  and  day  2  of  the  experiment.  Samples  were  amended  with  phosphate  (P),  ammonia  (N),  and  organic  carbon  (C),  or  with  combinations  of  these  nutrients  (PC,  NP,  NPC).  Cont.:  control   treatment  (no  amendments).  Each  data  point  represents   the  average  of   the  two  replicates.  Error  bars  represent  the  standard  deviation.      The   turnover   time   in   the   control   treatment  was  ~  1  h,  which  means   that   all   the  bioavailable   pool   of   Pi  would   be   used   in   the   timespan   of   1   h   if   the   supply   of   Pi  stopped.  All   the   treatments  where  P  was  added  resulted   in  a  notable   increase   in  the   turnover   time   of   Pi  within   the   first   24   h   of   the   experiment.  However,   the   Pi  turnover  time  in  the  NP  and  NPC  treatments  after  48  h  returned  to  values  close  to  those  observed  in  the  control.  These  results  imply  that  the  Pi  added  could  not  be  entirely   used   in   the   P   treatment   due   to   the   lack   of   enough   N   and   C,   which   is  supported   by   the   observation   that   heterotrophic   bacteria   in   these   waters  accumulated  polyphosphates  when  Pi  was  added  alone  (Sebastián  et  al.,  2012).      Sebastián,   M.,   Pitta,   P.,   González,   J.M.,   Thingstad,   T.F.,   and   Gasol,   J.M.   (2012)   Bacterioplankton  groups   involved   in   the   uptake   of   phosphate   and   dissolved   organic   phosphorus   in   a   mesocosm  experiment  with  P-­‐starved  Mediterranean  waters.  Environmental  Microbiology  14:  2334-­‐2347      

24h48hday2

Cont. P N PC NP NPC

1000

100

10

1

0.1

Pi t

urn

ove

r tim

e (

h)

Treatment

Figure  S2.-­‐  RESPONSE  OF  THE  COMMUNITY  COMPOSITION  TO  THE  NUTRIENT  ADDITIONS    

   

 Figure  S2  legend.  Community  composition  before  the  start  of  the  experiment  and  in  the  different  treatments  at  the  moment  the  MARFISH  analyses  were  performed.  Data  are  presented  as  percent  contribution  of  the  probe-­‐identified  groups  to  total  number  of  prokaryotes  (DAPI  counts).  Treatments  correspond  to  the  following  amendments:  phosphate  (P),  ammonia  (N),  and  organic  carbon  (C),  or  combinations  of  them  (PC,  NP,   NPC).   Initial:   before   the   start   of   the   experiment.   Cont.:   control   treatment   (no  amendments).  

     

0

20

40

60

80

100

Initial Cont. P N PC NP NPC

GammaproteobacteriaBacteroidetesSynechococcusSAR11Roseobacterother EUBnon EUB

% D

AP

I

treatment

Figure  S3.-­‐  EFFECT  OF  NUTRIENT  ADDITIONS  IN  THE  TOTAL  ABUNDANCE  OF  ACTIVE  CELLS  WITHIN  EACH  PROBE-­‐IDENTIFIED  GROUP

   Figure  S3  legend.  Abundance  of  cells  active  in  3H-­‐leucine  incorporation  belonging  to  each   probe-­‐identified   group.   Labels   as   defined   in   Figure   S1.   Asterisks   denote  significant  differences  in  relation  to  the  control  (*:  p<  0.05,  **:  p<  0.001).  Insert  in  the  left   lower  panel   is  an  expanded  view  of   the  abundance  of  cells   in   the  Bacteroidetes  lineage.  Note  the  different  scale  in  the  Y-­‐Axis  for  Eubacteria  (upper  panel).  

     

0

Bacteroidetes

Gammaproteobacteria Roseobacter

0

Cont. P N PC NP NPC

SAR11

0

1.5

Cont. P N PC NP NPC

0.5

1

1.5

2

2.5

0.5

1

1.5

2

2.5

Cont. P N PC NP NPC

Abundance

of act

ive c

ells

(M

AR

+)

(x10

5 c

ells

)

Treatment

***

*

*

**

*

p = 0.05

p = 0.05

*

Eubacteria

0

1

2

4

3

**

Figure  S4.-­‐  RELATIONSHIP  BETWEEN  BACTERIAL  ACTIVITY  AND  NUMBER  OF  ACTIVE  CELLS  

 Figure   S4   legend.   Relationships   between   the   total   numbers   of   active   cells   within  each  probe-­‐identified  group  and  the  bulk  leucine  incorporation  measured  for  each  of  the   samples.   EUB:   Eubacteria,   Gamma:   Gammaproteobacteria,   Ros:   Roseobacter,  Bact:  Bacteroidetes.  Numbers  in  italics  represent  the  slopes  of  the  linear  fit.  Note  the  similarity  in  the  slopes  of  the  Gammaproteobacteria  and  total  bacteria  (Eubacteria)  relationships.        

R=0.87

R=0.96

R=0.85

R=0.92

R=0.85

EUB

Gamma

SAR11

Ros

Bact.

34x

31x

8x

1x

14x

0

10

20

30

40

0 0.2 0.4 0.6 0.8Leucine incorporation

(nmol Leu h -1 L-1)

Ab

un

da

nce

of

act

ive

ce

lls (

MA

R+

)

(x10

5 cells

)

       Figure  S6.-­‐  GAMMAPROTEOBACTERIA  COMPOSITION            

 Figure   S6   legend.   Composition   of   the   gammaproteobacterial   population   in   the  control  and  nutrient  amended  treatments.  

0

20

40

60

80

100

Cont. P N PC NP NPC

NOR5/OM60AlteromonadaceaeOther Gammap.

% o

f pro

be p

ositiv

e c

ells