1

Isolation of Chemoautotrophic Sulfur-Oxidizing Denitrifying Bacteria 1

from a Stratified Marine Ecosystem: Salt Pond 2

Steven James Hallam1,2 3

4

5

1 Dept. of Microbiology & Immunology, University of British Columbia, Vancouver BC V6T1Z3, Canada 6

2 2008 Microbial Diversity Course, Marine Biological Laboratory, Woods Hole MA 02543 USA 7

8

9

10

Pages: 17 Figures: 8 11

Tables: 0 12

13

Abstract: 243 words, 1603 characters 14

Text: 3,900 words, 22,656 characters 15

16

Running Title: 17

18

Microbial community structure within a meromictic lake ecosystem 19

20

21

22

23

24

* To whom correspondence should be addressed. E-mail: shallam@interchange.ubc.ca25

2

Abstract 1

Meromictic lakes are stratified ecosystems characterized by clines of oxygen, salinity and 2

temperature with concomitant changes in redox chemistry spanning the vertical depth 3

continuum. In order to better understand the microbial community structure associated with low 4

oxygen metabolism within these ecosystems, a cultivation-independent survey targeting the 5

bacterial small subunit ribosomal RNA V6 region coupled to physical and chemical analyses was 6

conducted in Salt Pond on the coast of Falmouth Massachusetts. Analysis of V6-tag 7

pyrosequence reads derived from six depths spanning the oxycline identified Actinobacteria, 8

Bacteroidetes, Cyanobacteria, alpha and gamma Proteobacteria as abundant community 9

members within the oxic mixolimnion. Within the oxycline, cyanobacteria gave way to green 10

sulfur and purple sulfur bacteria. Within the monimolimnion green sulfur bacteria represented 11

over 60% of all V6-sequence tag data followed by firmicutes, alpha, delta, epsilon and gamma 12

Proteobacteria. Survey data was complimented by efforts to isolate sulfur-oxidizing microbes 13

with the capacity to use nitrate as the sole terminal electron acceptor. Sulfide concentrations 14

ranging between 0.4-1.2 mM were measured within the anoxic monimolimnion. Efforts to 15

cultivate sulfur-oxidizing organisims below the oxycline yielded three strains of Thiomicrospira 16

closely related to Thioalkalimicrobium aerophilum nested within the gamma Proteobacteria and 17

two divergent groups of epsilon Proteobacteria related to Sulfurimonas autotrophica and 18

Arcobacter nitrofigilis respectively. Overall, microbial community structure in Salt Pond was 19

dominated by phototrophic organisms spanning the vertical depth continuum with chemotrophic 20

energy metabolism based in part on the conversion of reduced sulfur compounds increasing with 21

depth below the oxycline. 22

23

3

Introduction 1

Meromictic lake ecosystems create stratified niche space for a wide-array of 2

photosynthetic and chemoautotrophic microorganisms. Broadly defined, they can be separated 3

into an upper oxic mixolimnion and a lower anoxic monimolimnion. Between these two regions 4

is a sharp and relatively stable redox transition zone of decreasing oxygen tension called the 5

oxycline or chemocline hosting an array of photosynthetic and chemosynthtic microorganisms 6

[1-8] Meromictic lakes are useful systems to investigate microbial community structure along 7

defined redox gradients based on the residual chemical energy of ammonia, methane, transition 8

metals including iron and manganese, hydrogen sulfide and molecular hydrogen [9-16], and 9

resemble the stratified chemical gradients typically associated with anoxic marine sediments 10

[17]. Therefore, biogeochemical and microbial ecological studies of meromictic lakes should be 11

extensible to stratified aquatic ecosystems including marine oxygen minimum zones (OMZ) 12

associated with enclosed basins and coastal fjords. 13

OMZs are a recurring oceanographic feature. They provide a physical and chemical 14

boundary layer for many organisms, expose large sections of the seafloor to long-term anoxic 15

conditions and support the biogeochemical transformation of reduced nitrogen compounds and 16

methane to nitrous oxide and carbon dioxide respectively. Recent cultivation-independent 17

molecular surveys of OMZ systems in Namibia [18], Peru [19] and the subarctic Pacific Ocean 18

(unpublished observations) have recovered an abundance of small subunit ribosomal DNA (SSU 19

rDNA) sequences closely affiliated with chemoautotrophic sulfur-oxidizing bacteria including 20

free-living relatives of hydrothermal vent clam and mussel endosymbionts. The widespread 21

occurrence of putative sulfur-oxidizing bacteria within marine OMZs raises fundamental 22

4

questions about the mechanism and biogeochemical role of sulfur cycling within the interior 1

ocean. 2

Salt Pond is a shallow seasonally stratified meromictic lake situated on the Massachusetts 3

coast near the town of Falmouth. Based on previous studies, the water column chemistry of Salt 4

Pond resembles that of other stratified marine ecosystems [20, 21]. Ease of access combined 5

with representative water column chemistry make Salt Pond an ideal location for exploring 6

microbial community structure along defined physical and chemical gradients. Previous studies 7

of the Salt Pond microbiota focused on enrichment of magnetotactic bacteria with special 8

emphasis on morphotypic variation and contribution to chemical cycling [20-22]. Although non-9

magnetotactic bacteria were identified as background in these studies, to date there have been no 10

efforts to explore the overall microbial community structure of the Salt Pond ecosystem. 11

Here we report the results of high resolution studies aimed at profiling the Salt Pond 12

microbiota along defined gradients of oxygen, temperature, light intensity, salinity, and sulfide 13

using a combination of V6-tag pyrosequencing and cultivation techniques using sulfide or 14

thiosulfate as electron donors and nitrate as sole electron acceptor to better understand 15

biogeochemical controls on microbial community structure and metabolic potential. The data 16

reflects summer stratification conditions during field collection in July 2008. 17

18

Methods 19

Sampling depths began at the base of the mixolimnion (1.82 m), and continued through 20

the oxycline (2.2 m, 2.9 m, 3.1 m), down to the lower monimolimnion (3.4 m, and 3.7 m). Water 21

samples were collected using a peristaltic pump (Cole-Parmer) system powered by a marine 22

battery pack and DC power converter with ¼” tubing directly tethered to a CTD array. The array 23

5

was manually deployed from a small rowboat in the northwestern quadrant of Salt Pond. 1

Temperature, Salinity, depth, light intensity and oxygen were recorded at approximately 0.25 m 2

intervals down to base of the monimolimnion (4.04 m). Water samples were pumped directly 3

into autoclaved 1 L bottles for environmental DNA (eDNA) extraction and cultivation work. 4

Aliquots from each bottle were transferred into 15 ml conical tubes amended with either 1% 5

formaldehyde (final concentration) for cell counts or zinc acetate for downstream sulfide 6

measurements, according to the methods of Cline [23]. 7

8

Sulfide Analysis 9

900µl from each zinc acetate treated sample was transferred into 2.0 ml disposable cuvettes and 10

amended with solutions 1 and 2 using a hydrogen sulfide detection kit (HS-WR) according to the 11

manufacturer’s instructions (Hach). Samples were subsequently analyzed on a Carey 50-scan 12

UV-visible spectrophotometer at a wavelength of 670 nanometers. 1:100 dilutions for samples 13

below the transition zone (3.4 m, and 3.7 m ) were also prepared and measured in order to 14

remain with the dynamic range of the spectrophotometer. A standard curve was generated based 15

on serial dilutions of a 100mM Na2S stock solution. 16

17

Environmental DNA Extraction 18

100 ml from each depth interval was filtered onto 47mm 0.2µM GTTP membranes (Millipore). 19

Membranes were subsequently sectioned and transferred into 2.0ml PowerBead tubes (MoBio) 20

and homogenized for 30 seconds using a bead beater device (Biospec Products). eDNA was 21

extracted according to manufacturer’s instructions, eluted in 100µl of 10mM Tris and stored at -22

20°C until needed for downstream PCR experiments. eDNA quality was determined by running 23

6

5µL samples and 100ng of 1KB+ Labber (New England Biolabs) on 1% agarose gels in 0.5x 1

TBE for 10 hours at 50 volts. 2

3

Environmental DNA Amplification and Colony PCR 4

For quality control purposes prior to V6-tag sequencing, small subunit (SSU) ribosomal DNA 5

(rDNA) bacterial sequences were amplified via PCR using B27F 6

(5’AGAGTTTGATCCTGGCTCAG) and the universal primer 1492R primer 7

(5’GGTTACCTTAGTTACGACTT) using the following PCR profile: a 2 min hot start at 95ºC 8

followed by 30 cycles of denaturation (20 sec at 95ºC), annealing (20 sec at 55ºC), and extension 9

(1 min at 72ºC) and a final extension at 72ºC for 3 minutes. Each 25µL reaction contained 1µL 10

of template DNA, in 1X master mix (Promega) supplemented with 1.25 µL of reverse and 11

forward primer (10 pmol each). The same primer pair and amplification procedure was used in 12

colony PCR experiments to identify Salt Pond isolates (see below) following 10 minute 13

denaturation in 0.05% NP40 solution at 95°C. 14

15

PCR Amplicon Library Construction for Pyrosequencing 16

Primers targeting the V6 hypervariable region of bacterial 16S rRNAs (Escherichia coli positions 17

967-1046) were used to generate amplicons for pyrosequencing [24]. The oligonucleotide design 18

included 454 Life Science's A or B sequencing adapter (shown in lowercase in the following) 19

fused to the 5′ end of primer A-967F, 5′-gcctccctcgcgccatcag-CAACGCGAAGAACCTTACC-3′20

, and B-1046R, 5 ′ -gccttgccagcccgctcag-CGACAGCCATGCANCACCT-3 ′ . PCR amplicon 21

libraries for individual depths included linker barcodes enabling multiplexed sequencing 22

7

reactions [25]. Amplification, quality control and data analysis procedures followed previously 1

described methods for V6-tag sequences [24]. 2

3

Enrichment, Isolation and Culture Conditions 4

To enrich for sulfur-oxidizing denitrifiers, a modular medium buffered with 10mM MOPS pH 5

7.2 containing 1X seawater base amended with salts (10mM NH4Cl, 1.5 mM KH2PO4), cofactors 6

(12X Vitamins, B12, and trace elements) was used. The carbon source supplied to the medium 7

was 2mM Na2HCO3. Electron donors included 8mM Na2S or 30 mM NaS2O3 with 20mM 8

NaNO3 as sole electron acceptor. 9

Primary enrichment cultures were generated at 26°C or 30°C using a modified version of 10

the agarose-plug overlay method developed by Holger Jannasch (1% GTG agarose in plug 11

containing sulfide and bicarbonate, and 0.16% in overlay containing modular medium mixed 12

with Salt Pond inocculum) to isolate individual colony forming units along oxygen-sulfide 13

gradient. Primary isolates obtained from enrichments were subsequently transferred to 500 µl of 14

1X seawater base and used to inoculate fresh overlays or 3% GTG agarose plates containing 15

modified overlay media (30 mM thiosulfate substituted for sulfide as electron donor with 40 mM 16

Na2HCO3 as carbon source and pH buffer under anaerobic growth conditions). 17

18

Phylogenetic Analysis 19

Individual colonies from selected plates were selected for colony PCR as described above. 20

Sequence traces were obtained using BigDye chemistry on an ABI Prism 3730 sequencer, 21

trimmed using a Unix implementation of Phred [26] (phred -id chromat_dir -pd phd_dir -sd 22

edit_dir -trim_alt "" -trim_phd -trim_fasta -trim_cutoff .01 -qd qual_dir) and analyzed using the 23

8

ARB package for phylogenetic sequence analysis [27]. Sequences were imported into the full-1

length SILVA 92 database (www.arb-silva.de) [28] and aligned to the closest relative. For 2

phylogenetic analysis, Salt Pond and selected reference sequences were exported from ARB and 3

realigned using the multi-sequence alignment tool MUSCLE [29]. The resulting alignment was 4

imported into MacClade [30] for manual refinement. Phylogenetic trees were inferred by 5

PHYML [31] using an HKY + 4Γ + I model of nucleotide evolution where the α parameter of 6

the Γ distribution, the proportion of invariable sites, and the transition/transversion ratio were 7

estimated for each dataset. The confidence of each node was determined by assembling a 8

consensus tree of 100 bootstrap replicates. 9

10

Results 11

12

Objectives and Workflow 13

14

The primary objectives of this study were to (i) profile the microbial community structure of Salt 15

Pond as a function of oxygen, salinity and sulfide gradients using V6-tag pyrosequecning 16

approaches, and (ii) to develop and ground truth methods for the isolation of chemoautotrophic 17

sulfur-oxidizing denitrifying bacteria extensible to stratified aquatic ecosystems in general. 18

Figure 1 provides a general workflow for the study. Following site selection and chemical 19

profiling, both cultivation-independent and cultivation-dependent methods were employed. Both 20

methods intersected at the use of SSU rDNA sequence information to infer phylogenetic 21

relationships among taxa at both the community and isolate levels. 22

23

9

Water Chemistry and Sample Selection 1

2

Figure 2 describes water column chemistry collected on the afternoon of July 5th 2008. Density 3

profiles through the water column reveal a dramatic shift in salinity between 1.0 and 1.5 m with 4

a corresponding thermocline ranging between 26.25 °C at the surface to 16.01 °C at 4.03 m 5

corresponding to the base of the monimolimnion (Figure 3). Salt Pond waters at the time of 6

sampling were brackish peaking at 25.38 ppt salinity at the base of the monimolimnion. The 7

oxycline covered a 1.6 m interval between 1.5 and 3.1 m. Sulfide concentrations increased 8

sharply below 3.1 m, reaching 1.26 mM at the base of the monimolimnion. Water samples for 9

eDNA extraction and cultivation work we selected on the basis of water column chemistry to 10

bracket the oxycline and monimolimnion (see methods). 11

12

Microbial Community Diversity and Structure 13

14

To evaluate overall microbial community diversity and structure high resolution V6-tag 15

pyrosequencing was conducted on eDNA derived from six depths spanning the oxycline and 16

monimolimnion (see methods). A total of 210,632 tag sequence reads were generated according 17

to the following distribution (1.8 m = 45,633, 2.2 m = 38,945, 2.9 m = 41,523, 3.1 m = 26717, 18

3.4 m 29840 and 3.7 m = 27974 reads). Community diversity was evaluated using rarefracton 19

analysis while community structure was evaluated at the phylum and order levels to provide both 20

course and fine scale resolution within the accuracy bounds of variable-tag identification 21

schemes (see methods). Overall, rarefraction curves for each of the six depth intervals exhibited 22

similar slopes at the 3% identity cut-off (Figure 4). At this level of genetic distance, the curves 23

10

predict that additional sampling will lead to only modest increases in estimates of total diversity 1

consistent with near saturation coverage of the V6 region for the six depth intervals. 2

Depth specific trends in the distribution of taxa spanning the vertical depth continuum 3

were evident (Figures 5). At the phylum level (Figure 5A), Actinobacteria comprised between 4

22-27% of tag sequences in the first two depth intervals before falling to 3% within the 5

monimolimnion. Cyanobacteria, Bacteroidetes, Verrucomicrobia, alpha and gamma 6

Proteobacteria accounted for the remainder of tags within both intervals. Within the oxycline at 7

2.9 m, cyanobacteria gave way to green sulfur (Chlorobi) and purple sulfur bacteria 8

(Chromatiales). Within the monimolimnion green sulfur bacteria represented over 60% of all 9

V6-sequence tag data followed by firmicutes, alpha, delta, epsilon and gamma Proteobacteria. A 10

closer look within the Proteobacteria revealed depth specific increases of V6-tags affiliated with 11

known sulfate-reducing bacteria within the order Desulfobacteriales reaching 5.47% of total tags 12

at the base of the monimolimnion (Figure 5B). V6-tags affiliated with members of the 13

Desulfuromanales were also identified below the 2.2 m depth interval. However, 14

Desulfuromanales comprised less than 1% of tag sequences to the base of the monimolimnion. In 15

addition to sulfate reducing bacteria, V6-tags affiliated with known sulfur-oxidizing bacteria 16

including Arcobacter (Camylobacterales) and Thiomicrospira (Thiotrichaes) were also detected 17

at low levels throughout the depth continuum. 18

19

Cultivation of sulfur-oxidizing bacteria 20

21

Figure 6 provides a general workflow for the cultivation approach developed in this study. 22

Primary enrichments in agar-plug overlay tubes containing a sulfide gradient (revelaed by the 23

11

color phase of the redox indicator resazurin ) were used as seed stock for secondary enrichment 1

and isolation on thiuosulfate containing plates (see methods). Although many isolates were 2

obtained from enrichments spanning the depth continuum, only a five were selected from 3

downstream characterization based on their capacity to form colonies on agarose plates. In 4

general, colonies formed in regions of the overlay that seemed to recapitulate aspects of their in 5

situ environment. For instance, tubes inoculated with water from the 3.7 m depth interval formed 6

colonies closer to the agar plug, whereas tubes inoculated with water from the 1.8 m depth 7

interval formed colonies closer further away from the agar plug. In general, colonies formed 8

within the agar-plug overlay medium over the course 3-5 days. On plates incubated in anaerobic 9

jars colony forming units grew more slowly over the course of 7-10 days. In all cases but one, 10

colonies appeared white when viewed under direct lighting presumably due to the precipitation 11

of elemental sulfur granules. Photomicrographs appear in the inverse due to bottom illumination 12

through the dissecting microscope (Figure 7). Although nitrate was used as sole electron acceptor 13

in both agar-plug overlay and plate growth media, it remains formally possible that strains 14

isolated in this study are capable of utilizing molecular oxygen as a terminal electron acceptor. 15

Physiological characterization of the five isolates remains the object of ongoing investigation. 16

However, phylogenetic classification based on SSU rDNA sequence (see methods) placed all 17

five within known sulfur-oxidizing lineages (Figure 8). Three of the five isolates were affiliated 18

with Thiomicrospira within the gamma proteobacteria. The remaining two were most closely 19

related to environmental clones affiliated with Sulfurimonas and Arcobacter respectively within 20

the epsilon Proteobacteria. These results are consistent with V6-tag sequencing data indicating 21

the presence of all three groups within the Salt Pond depth continuum. Based on this observation, 22

12

it appears that the methods developed in this study for the isolation of sulfur-oxidizing bacteria 1

are robust. 2

3

Conclusions 4

5

The current study provides a high-resolution look at microbial community diversity spanning 6

defined gradients of salinity, oxygen and sulfide in the meromictic Salt Pond. Overall, microbial 7

community structure is dominated by the phototrophic bacteria, with oxygenic photosynthesis 8

giving way to low-light adapted anaoxygenic photosynthesis within the oxycline and 9

monimolimnion. It will be of interest to determine the taxonomic identity of green sulfur and 10

purple sulfur bacteria resident with the depth continuum and to relate their V6-tag abundance 11

with lipid biomarker or pigment analysis. Likewise a more thorough evaluation of V6-tag 12

sequences in relation to published studies of magnetotactic bacterial enrichments from Salt Pond 13

is warranted. 14

Based on V6-tag sequence analysis the Salt Pond site harbors a low level representation 15

of indigenous chemoautotrophic sulfur oxidizing bacteria. Through the combined use of agar-16

plug and plate isolation techniques 5 isolates were obtained from within these lineages (three 17

Thiomicrospira all close relatives of a described lineage, Thioalkalimicrobium aerophilum [32], 18

one Sulfurimonas, and one Arcobacter sp). The later two isolates are novel with ~8% sequence 19

divergence from known representatives. Taken together, the techniques employed in this study 20

appear robust, capable of selecting specific physiological groups from within a diverse 21

community background and should be tunable to other stratified aquatic ecosystems. Future 22

studies focused on determining the precise metabolic capabilities of sulfur oxidizing isolates 23

13

obtained from Salt Pond including oxygen tolerance, growth rates and range of electron 1

acceptors is necessary. Given their capacity to grow as colony forming units on agarose plates 2

the potential for genetic analysis exists should they prove metabolically interesting. Future 3

studies of Salt Pond focusing on the biogeochemical roles of phototrophic, sulfur-oxidizing and 4

sulfate reducing community members have the potential to shed light on sulfur cycling within 5

other stratified aquatic ecosystems including coastal waters and enclosed basins. 6

7

Acknowledgements 8

9

Special thanks to the MBL Microbial Diversity 2008 course directors Tom Schmidt and William 10

Metcalf for expert guidance and support, Mitch Sogin and the VAMPS team for access to the 11

pyrosequecning platform and informatics pipeline, James Saenz, Lydia Zeglin and David Walsh 12

for help in field collection and the rest of the course staff and students for helpful discussions and 13

camaraderie. This work was funded by grants from the Gordon and Betty Moore Foundation, 14

The Sloan Foundation, the Bernard Davis Endowed Scholarship Fund, the Frank R. Lillie 15

Fellowship and Scholarship Fund and the Natural Sciences and Engineering Research Council 16

(NSERC). 17

18

19

14

Figure Legends 1

2

Figure 1. A workflow summary of Salth Pond sampling and analysis based on a preliminary 3

chemical and physical profile followed by a combination of cultivation-independent molecular 4

surveys and targeted isolation methods associated with specific metabolic subsystems, elemental 5

cycles, or redox chemistries. 6

7

Figure 2. Opposing gradients of Salinity, Oxygen and Sulfide. Sampling depths for eDNA 8

extraction and cultivation work are indicated with arrows. 9

10

Figure 3. Changes in light intensity and temperature as a function of depth. 11

12

Figure 4. Rarefaction analysis for V6-tag sequences derived from six sampling intervals 13

spanning the Salt Pond depth continuum based on pairwise distance. Rarefaction is shown for 14

OTUs with differences that do not exceed 3%. 15

16

Figure 5. (A) Phylum level classification of V6-tag sequences based on best blast hit analysis 17

using the V6RefDB database (see methods) (B) Order level classification of V6-tags within the 18

Proteobacteria. Values on the X-axis are shown as a percentage of total tags for the 19

corresponding depth interval. NA represents unassigned tags within a given classification level. 20

21

Figure 6. A workflow summary of Salth Pond methods used to enrich and isolate sulfur-22

oxididizing bacteria from the Salt Pond depth continuum. P0 represents primary enrichment 23

15

followed by successive selection and re-isolation using the agar-plug overlay method. Individual 1

colonies derived from the primary enrichment were diluted in 1X seawater base and used to 2

simultaneously inoculate a second sulfide gradient tube and GTG agarose plate containing 3

thiosulfate as sole electron acceptor (see methods). Colony PCR was performed on selected 4

colonies to ascertain the taxonomic identity of isolates (see methods). 5

6

Figure 7. Colony morphology of selected Salt Pond isolates grown on thiosulfate. SP_A4 7

corresponds to Salt Pond cast A4. Isolate designations use a standard nomenclature where the 8

first number indicates depth interval: 1 = 1.82 m, 2 = 2.2 m 3= 2.9 m, 4 = 3.1 m, 5 = 3.4 m, and 9

6 = 3.7 m. The second letter indicates growth temperature (A = 26°C and B = 30°C). The third 10

number indicates isolate id followed by the number of plate passages. 11

12

Figure 8. Distance tree of Proteobacterial SSU rDNA sequences isolated from the Salt Pond 13

depth continuum. Bootstrap values (%) are based on 100 replicates using the maximum 14

likelihood method and are shown for branches with greater than 50% support (see methods). The 15

scale bar represents 0.05 substitutions per site. Isolate depths are color coded in the tree. 16

17

18

19

16

References 1

1. Bosshard, P.P., et al., Bacterial diversity and community composition in the chemocline 2 of the meromictic alpine Lake Cadagno as revealed by 16S rDNA analysis. FEMS 3 Microbiol Ecol, 2000. 31(2): p. 173-182. 4

2. Bosshard, P.P., R. Stettler, and R. Bachofen, Seasonal and spatial community dynamics 5 in the meromictic Lake Cadagno. Arch Microbiol, 2000. 174(3): p. 168-74. 6

3. Bowman, J.P., et al., The microbial composition of three limnologically disparate 7 hypersaline Antarctic lakes. FEMS Microbiol Lett, 2000. 183(1): p. 81-8. 8

4. Koizumi, Y., H. Kojima, and M. Fukui, Potential sulfur metabolisms and associated 9 bacteria within anoxic surface sediment from saline meromictic Lake Kaiike (Japan). 10 FEMS Microbiol Ecol, 2005. 52(3): p. 297-305. 11

5. Overmann, J., M.J. Coolen, and C. Tuschak, Specific detection of different phylogenetic 12 groups of chemocline bacteria based on PCR and denaturing gradient gel 13 electrophoresis of 16S rRNA gene fragments. Arch Microbiol, 1999. 172(2): p. 83-94. 14

6. Sabet, S., W. Chu, and S.C. Jiang, Isolation and genetic analysis of haloalkaliphilic 15 bacteriophages in a North American Soda Lake. Microb Ecol, 2006. 51(4): p. 543-54. 16

7. Sorokin, D.Y., et al., Sulfur-oxidizing bacteria in Soap Lake (Washington State), a 17 meromictic, haloalkaline lake with an unprecedented high sulfide content. Appl Environ 18 Microbiol, 2007. 73(2): p. 451-5. 19

8. Tuomi, P., et al., Bacterial population dynamics in a meromictic lake. Appl Environ 20 Microbiol, 1997. 63(6): p. 2181-8. 21

9. Winfrey, M.R. and J.G. Zeikus, Microbial methanogenesis and acetate metabolism in a 22 meromictic lake. Appl Environ Microbiol, 1979. 37(2): p. 213-21. 23

10. Deevey, E.S., Jr., N. Nakai, and M. Stuiver, Fractionation of Sulfur and Carbon Isotopes 24 in a Meromictic Lake. Science, 1963. 139(3553): p. 407. 25

11. Gorlenko, V.M. and E.N. Chebotarev, [Microbiologic processes in meromictic Lake 26 Sakovo]. Mikrobiologiia, 1981. 50(1): p. 134-9. 27

12. Gorlenko, V.M., M.B. Vainshtein, and E.N. Chebotarev, [Sulfur and iron cycling 28 bacteria in low-sulfate meromictic Lake Kuznechikha]. Mikrobiologiia, 1980. 49(5): p. 29 804-12. 30

13. Haberyan, K.A., S.P. Horn, and G. Umana, Basic limnology of fifty-one lakes in Costa 31 Rica. Rev Biol Trop, 2003. 51(1): p. 107-22. 32

14. Ivanov, M.V., et al., [Microbial processes of carbon and sulfur cycles in lake Mogil'noe]. 33 Mikrobiologiia, 2001. 70(5): p. 675-86. 34

15. Kallistova, A., et al., [Sulfate reduction and methanogenesis in the Shira and Shunet 35 meromictic lakes (Khakass Republic, Russia)]. Mikrobiologiia, 2006. 75(6): p. 828-35. 36

16. Likens, G.E. and P.L. Johnson, A Chemically Stratified Lake in Alaska. Science, 1966. 37 153(3738): p. 875-877. 38

17. Yen, T.F., Chemical aspects of marine sediments, ed. T.F. Yen. 1977, Ann Arbor, MI: 39 Ann Arbor Science Publishers. 1-38. 40

18. Woebken, D., et al., From the Namibian and Oregon coast upwelling systems and their 41 cross-comparison with planctomycete genomes. Isme Journal, 2007. 1(5): p. 419-435. 42

19. Stevens, H. and O. Ulloa, Bacterial diversity in the oxygen minimum zone of the eastern 43 tropical South Pacific. Environmental Microbiology, 2008. 10(5): p. 1244-1259. 44

17

20. Simmons, S.L., D.A. Bazylinski, and K.J. Edwards, Population dynamics of marine 1 magnetotactic bacteria in a meromictic salt pond described with qPCR. Environ 2 Microbiol, 2007. 9(9): p. 2162-74. 3

21. Simmons, S.L., et al., Spatiotemporal distribution of marine magnetotactic bacteria in a 4 seasonally stratified coastal salt pond. Appl Environ Microbiol, 2004. 70(10): p. 6230-9. 5

22. Simmons, S.L. and K.J. Edwards, Unexpected diversity in populations of the many-celled 6 magnetotactic prokaryote. Environ Microbiol, 2007. 9(1): p. 206-15. 7

23. Cline, J.D., Spectrophotometric determination of hydrogen sulfide in natural waters. 8 Limnology & Oceanography 1969. 14: p. 454-458. 9

24. Sogin, M.L., et al., Microbial diversity in the deep sea and the underexplored "rare 10 biosphere". Proc Natl Acad Sci U S A, 2006. 103(32): p. 12115-20. 11

25. Huber, J.A., et al., Microbial population structures in the deep marine biosphere. 12 Science, 2007. 318(5847): p. 97-100. 13

26. Ewing, B. and P. Green, Base-calling of automated sequencer traces using phred. II. 14 Error probabilities. Genome Res, 1998. 8(3): p. 186-94. 15

27. Ludwig, W., et al., ARB: a software environment for sequence data. Nucleic Acids 16 Research, 2004. 32: p. 1363-1371. 17

28. Pruesse, E.C., et al., SILVA: a comprehensive online resource for quality checked and 18 aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res, 2007. 19 In press. 20

29. Edgar, R.C., MUSCLE: multiple sequence alignment with high accuracy and high 21 throughput. Nucleic Acids Res, 2004. 32(5): p. 1792-7. 22

30. Maddison, D.R. and W.P. Maddison, MacClade 4. 2003, Sunderland: Sinauer Associates, 23 Inc. 24

31. Guindon, S., et al., PHYML Online--a web server for fast maximum likelihood-based 25 phylogenetic inference. Nucleic Acids Res, 2005. 33(Web Server issue): p. W557-9. 26

32. Sorokin, D.Y., et al., Thioalkalimicrobium aerophilum gen. nov., sp. nov. and 27 Thioalkalimicrobium sibericum sp. nov., and Thioalkalivibrio versutus gen. nov., sp. 28 nov., Thioalkalivibrio nitratis sp.nov., novel and Thioalkalivibrio denitrificancs sp. nov., 29 novel obligately alkaliphilic and obligately chemolithoautotrophic sulfur-oxidizing 30 bacteria from soda lakes. Int J Syst Evol Microbiol, 2001. 51(Pt 2): p. 565-80. 31

32 33

Figure 1

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

0 2 4 6 8 10

Oxygen (mg/L)

16 18 20 22 24 26

Salinity (ppt)

Sulfide [mM]

Salinity (ppt)

Dep

th (m

)

Oxygen [mg/L]

0 0.2 0.4 0.6 0.8 1 1.2 1.4

Sulfide (mM)

1L Sample

Salt Pond SP_A4 Opposing Gradients of Salinity, Oxygen and Sulfide

Figure 2

Figure 3

2.5

3.0

3.5

4.0

0 2 4 6 8 10

PAR

Salt Pond SP_A4 Light Intensity vs Temperature

0 5 10 15 20 25

°C

0.0

0.5

1.0

1.5

2.0

Temperature (°C)

Dep

th (m

)

Light Intensity (PAR)

Figure 4

No. of V6-tags Sampled

Op

erat

ion

al T

axo

no

mic

Un

its

A

B

Figure 5

Figure 6

Figure 7

Thiothrix flexilis

Thiothrix nivea

Thiomicrospira kuenenii

Thiomicrospira crunogena XCL−2

Thiomicrospira psychrophila

Thiomicrospira sp. JB−A1F

SP_A4 5B_3.1

SP_A4 5B_4.1

SP_A4 5B_1

Thioalkalimicrobium aerophilum

Cycloclasticus oligotrophus

Methylophaga sulfidovorans

Burkholderia cepacia

Thiomonas arsenivorans

Nitrosomonas sp. K1

Nitrosospira multiformis ATCC 25196

Agrobacterium tumefaciens

Rhodospirillum rubrum ATCC 11170

Sulfurimonas autotrophica

uncultured epsilon proteobacterium 131631

SP_A4 6B_8.1

Sulfurimonas paralvinella

Thiomicrospira denitrificans ATCC 33889

Sulfurospirillum barnesii

Sulfurospirillum carboxydovorans

Sulfurospirillum arcachonense

uncultured eubacterium CHA3−124

SP_A4 4A_3

Arcobacter sp. BSs20195

Arcobacter nitrofigilis

Candidatus Arcobacter sulfidicus

Arcobacter cryaerophilum

Arcobacter butzler

Desulfobacterium indolicum

Desulfonema limicol

Desulfovibrio desulfuricans G20

50 changes

ε

α

β

δ

γ

100

100

100

100

100

100

100

100

100

64

81

99

97

99

93

99

93

100

80

100

1.8

2.2

2.9

3.1

3.3

3.7

Depth (m)

Proteobacteria