Marine Bacterioplankton Community Turnover Within Seasonally Hypoxic Waters of a Subtropical Sound: Devil’S Hole, Bermuda

bs_bs_banner

Environmental Microbiology (2014) doi:10.1111/1462-2920.12445

Marine bacterioplankton community turnover within seasonally hypoxic waters of a subtropical sound: Devil’s Hole, Bermuda

Rachel J. Parsons,1* Craig E. Nelson,2,3 community changes, including both direct counts and Craig A. Carlson,1,2 Carmen C. Denman,4,5 rRNA gene sequencing. During stratification, the Andreas J. Andersson,1,6 Andrew L. Kledzik,7 surface waters were dominated by the SAR11 clade of 4 1,8 Kevin L. Vergin, Sean P. McNally, Alphaproteobacteria and the cyanobacterium Syne- 4,9 4 Alexander H. Treusch and Stephen J. Giovannoni chococcus. In the suboxic bottom waters, cells 1Bermuda Institute for Ocean Science (BIOS), St. from the order Chlorobiales prevailed, with gene George’s, GE 01, Bermuda. sequences indicating members of the genera Chlo- 2Department of Ecology, Evolution and Marine Biology robium and Prosthecochloris – anoxygenic photo- and Marine Science Institute, University of California, autotrophs that utilize sulfide as a source of electrons Santa Barbara, CA, USA. for photosynthesis. Transitional zones of hypoxia also 3Center for Microbial Oceanography: Research and exhibited elevated levels of methane- and sulfur- Education, Department of Oceanography, University of oxidizing bacteria relative to the overlying waters. The Hawai‘i at Ma¯noa, Honolulu, HI, USA. abundance of both Thaumarcheota and Euryarcheota 4Department of Microbiology, Oregon State University, were elevated in the suboxic bottom waters (> 109 Corvallis, OR, USA. − cells l 1). Following convective mixing, the entire water 5London School of Hygiene and Tropical Medicine, column returned to a community typical of oxygen- London, UK. ated waters, with Euryarcheota only averaging 5% of 6Scripps Institution of Oceanography, University of cells, and Chlorobiales and Thaumarcheota absent. California San Diego, San Diego, CA, USA. 7Department of Marine and Environmental Systems, Introduction Florida Institute of Technology, Melbourne, FL, USA. 8College of the Environment and Life Sciences, The The importance of planktonic Bacteria and Archaea University of Rhode Island, Kingston, RI, USA. (referred to henceforth as bacterioplankton; Pace, 2006) 9Department of Biology, Nordic Centre for Earth and their role in ocean biogeochemical cycles are Evolution, University of Southern Denmark, Odense, well established; they are involved in virtually all the Denmark. biogeochemical reactions occurring in the oceans (Kirchman, 2008). High primary productivity in stratified Summary shallow coastal waters can lead to the seasonal development of suboxic and anoxic bottom waters. The develop- Understanding bacterioplankton community dyna- ment of these oxygen minimum zones is increasing in mics in coastal hypoxic environments is relevant strength, duration and extent because of the combined to global biogeochemistry because coastal hypoxia effects of eutrophication and climate change (Diaz and is increasing worldwide. The temporal dynamics Rosenberg, 2008; Stramma et al., 2010; Howarth et al., of bacterioplankton communities were analysed 2011; Wright et al., 2012). Physical mixing serves to ven- throughout the illuminated water column of Devil’s tilate portions of the water column and, if sufficiently deep, Hole, Bermuda during the 6-week annual transition can entrain nutrient rich deep water to the surface, eradi- from a strongly stratified water column with suboxic cating both oxygen minimum zones and redox gradients. and high-pCO2 bottom waters to a fully mixed and For example, the hurricanes in the Gulf of Mexico are ventilated state during 2008. A suite of culture- often strong enough to disrupt hypoxia on the Louisiana independent methods provided a quantitative spat- shelf (Rabalais et al., 2009). Alternating seasonal patterns iotemporal characterization of bacterioplankton of mixing followed by restratification can consequently regulate nutrient fluxes (Zehr and Ward, 2002) and create Received 6 November, 2013; accepted 23 February, 2014. *For cor- respondence. E-mail [email protected]; Tel. 1 441 297 1880 ecosystem shifts in bacterioplankton community structure x726; Fax 1 44 297 8143. (Giovannoni and Vergin, 2012).

Hypoxic environments are characterized by the deple- Despite the fact that microbial processes dominate in tion of oxygen with some demonstrating elevated con- anoxic and suboxic ecosystems, temporal patterns of centrations of sulfides (Stewart et al., 2007). Sulfide bacterial succession have been studied in only a handful production can result from sulfate reduction coupled of systems (Stal et al., 1985; Crump et al., 2007; Zaikova to the oxidation of reduced carbon compounds such as et al., 2010). Bacterial community succession has been methane (Stewart et al., 2007) or through the oxidation of associated with substrate concentrations in soils (Noll organic matter where sulfate is used as the electron et al., 2004), oxygen availability in estuaries (Crump et al., acceptor rather than oxygen (Fenchel et al., 2012). 2007), oxygen and phosphate gradients in fjords (Zaikova Oxygen is the most thermodynamically favoured terminal et al., 2010), and nitrogen fixation and light availability in electron acceptor followed by nitrate, nitrite then sulfate cyanobacteria mats within sandy sediments (Stal et al., and finally carbon dioxide (Wright et al., 2012). This 1985). sequence is used to define specific microbial niches that Devil’s Hole, Bermuda (32¡19.399′ N, 64¡43.08′ W) span oxygen ranges divided into oxic (> 90 μmol l−1); was used as a natural laboratory to examine the succes- dysoxic (20Ð90 μmol l−1); suboxic (1Ð20 μmol l−1) and sion of the bacterioplankton community structure as anoxic (< 1 μmol l−1) conditions (Wright et al., 2012). dysoxia and suboxia developed and subsequent turnover

The removal of oxygen, increase in pCO2 and subse- occurred. The study site was within the euphotic zone and quent decrease in pH all can play important roles in con- experiences seasonal thermal stratification resulting in trolling bacterioplankton community structure and function hypoxia below the base of the subthermocline layer. within this redoxcline (Brune et al., 2000; Taylor et al., Devil’s Hole is located in the shallow coastal embayment 2001). Adaptations to fluctuations in oxygen conditions, of Harrington Sound, Bermuda. Harrington Sound is specifically at the oxycline, in combination with elevated approximately 5 km2 with a mean bottom depth of 14.5 m

H2S and ammonium concentrations can enable spe- (Fig. 1). Devil’s Hole is the deepest part of the sound, a cialized bacterioplankton populations, such as H2S and former sinkhole with a bottom depth of ∼25 ± 1 m below ammonium-oxidizing bacterioplankton to colonize micro- the surface (depending on tidal height). Harrington Sound niches (Brune et al., 2000) and contribute to organic is linked to Bermuda’s North Shore reef platform via a matter production via chemoautotrophy (Taylor et al., narrow inlet (Flatts Inlet, < 20 m wide) providing the main 2001; Francis et al., 2007). In systems where the source of surface water tidal exchange (along with addi- redoxcline occurs in the euphotic zone (e.g. the Black tional subterranean water exchange). The nominal resi- Sea), organisms such as anaerobic photolithoautotrophic dence time of water in the sound is approximately 30 days green sulfur bacteria populate the oxycline (Repeta et al., with respect to the semidiurnal tidal ﬂow, but effective tidal 1989). For example, Chlorobium sp. are low light-adapted mixing can be as long as 4 months (Brown, 1978; 1980). photolithotrophic oxidizers of H2S that are capable of The physical and chemical properties of Harrington

ﬁxing CO2 to organic matter under suboxic or anoxic water Sound (and Devil’s Hole) have been studied extensively conditions (Kim and Chang, 1991; Overmann et al., 1992; for over 40 years (Thorstenson and Mackenzie, 1974; Taipale et al., 2009). Morris et al., 1977; Andersson et al., 2007).

Fig. 1. Inset Ð Map of Bermuda. Outset Ð Harrington Sound Map showing depth in contours (every 10 ft or 3 m) with Devil’s Hole located in the Southeast corner with the deepest depth (80 ft or 24.4 m). The only exchange with offshore water is through Flatts Inlet. The map is adopted from (Andersson et al., 2007).

This study consisted of a short time series in which 2004). TRFLP, FISH and ﬂow cytometry data were used water column biogeochemistry and bacterioplankton com- to select a subset of samples for deeper community in- munity structure were monitored at Devil’s Hole from terrogation using both Sanger sequencing and 454 August to November 2008, a period during which hypoxia pyrosequencing of 16S rRNA gene amplicons. By com- formed under stratiﬁed conditions (AugustÐSeptember) bining these approaches with a comprehensive suite of and was eroded under convective mixing conditions microbial, hydrographic and biogeochemical measure-

(OctoberÐNovember), a cycle that occurs annually. We ments (including temperature, salinity, pCO2,O2, viral employed multiple complementary culture-independent and bacterial densities, and inorganic nitrogen and phos- methods to characterize the succession- and mixing- phorous concentrations), we are able to present a cohe- induced change of the bacterioplankton communities sive picture of bacterioplankton community succession associated with the seasonally hypoxic bottom waters in as the deep waters transitioned from oxic to hypoxic Devil’s Hole. Community fingerprinting at high spatial and conditions. temporal resolution (terminal restriction fragment length polymorphism, TRFLP; Liu et al., 1997) was used to iden- Results tify zones and times of community assemblage shifts Physical and chemical properties throughout the time series. Selected bacterial and archaeal taxa were directly enumerated at the same high In early September 2008 (days 249Ð268), the water spatial and temporal resolutions using fluorescent in situ column was well stratified with temperatures ranging hybridization (FISH; Amann et al., 1990) and Catalyzed from 28 to 30¡C in the surface and a pronounced ther- Reporter Deposition FISH (CARD-FISH; Teira et al., mocline from 23 m to the bottom (Fig. 2A). Convective

Fig. 2. Contour plots showing day of year on the x axis and depth (m) on the y axis with colour contours for measurements of (A) −1 Temperature (¡C); (B) dissolved oxygen (μmol l ), (C) pCO2 (μatm) and (D) pH as well as biological parameters, (E) bacterioplankton abundance (×109 cells l−1) and (F) virioplankton abundance (×1010 VLP l−1) for all Devil’s Hole sampling dates from 5 September 2008 (day 249) to 8 October 2008 (day 282).

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology 4 R. J. Parsons et al. mixing in late September (day 274) disrupted the fivefold. On day 268, phosphate levels were 40-fold stratification, and by early October (day 282), complete higher at 24 m than within the surface waters, increasing overturn had occurred with temperatures averaging from 0.04 μmol l−1 at the surface to 0.97 μmol l−1 at 24 m. 26Ð27¡C throughout the water column (Fig. 2A). Salinity After overturn on day 282, nitrite, nitrate and phosphate was constant throughout the water column (36.31 ± 0.13 concentrations were uniform throughout the water practical salinity units) for all sampling times. Dissolved column, while ammonia concentrations were still slightly oxygen (DO) was elevated and relatively uniform elevated (3.92 μmol l−1) at depth (Table 1). throughout the surface 20 m during stratification (180Ð Although photosynthetically available radiation (PAR) 215 μmol l−1) decreasing steadily from 20 m to below the measurements were not made during this study, a detection level at 24Ð25 m (Fig. 2B); after overturn, DO PAR profile conducted in August 2008 (32¡19.346′ N, concentration were > 180 μmol l−1 throughout the water 64¡43.36′ W) showed that the attenuation coefficient for column (Table 1; Fig. 2B). Harrington Sound was 0.24 (Manuel et al., 2013), indicat- Winkler titrations are the standard method but have a ing that at the bottom depth of 25 m, the light level was detection limit of μmol l−1 (Ulloa et al., 2012) and may 0.25% of surface irradiance. The 0.1% light level is nec- have interference from other chemical species such as essary for photosynthesis in many photoautotrophs. - NO2 and H2S (Revsbech et al., 2009). Thus, measurements below 1 μmol l−1 are referred to as below detection Bacterioplankton and virioplankton abundance and the (Tables 1 and 2), and we refer to these zones as suboxic virus-to-bacterium ratio following Wright and colleagues (2012), although qualita- Bacterioplankton cell densities during stratification were tive indications based on smell and precipitation of HgS(s) uniform from surface to 22 m (3.71 ± 0.82 × 109 cells l−1; (see later discussion) indicated presence of H S and 2 n = 20) and then increased by nearly an order of magni- anoxia. tude to 1.72 ± 0.40 × 1010 cells l−1; n = 8), within the During stratification, dissolved inorganic carbon (DIC) dysoxic/suboxic layer below 22 m (Table 2; Fig. 2E). In and total alkalinity (TA) averaged at 2016 ± 30 μmol l−1 Devil’s Hole during stratification, surface virioplankton and 2293 ± 8 μmol l−1, respectively, within the surface abundances ranged from 1 to 4 × 1010 virus-like particles layer but increased to > 2806 and > 2524 μmol l−1 within (VLPs) l−1 (Table 2; Fig. 2F). Virioplankton densities were the subthermocline layer by day 268 (Table 1). However, elevated in the surface 20 m and ranged from 2 to TA samples collected from depths where DO was meas- 3 × 109 VLP l−1 during stratified periods and decreased to ured as zero using Winkler titration are considered under- less than 0.5 × 109 VLP l−1 in the bottom waters and after estimates of the true alkalinity as these samples were also overturn in the autumn (Fig. 2F). As a result, the virus-to- associated with precipitation of HgS (s) upon addition of bacterium ratio (VBR) also decreased with depth during aqueous HgCl to the samples and reaction with H S. 2 2 stratification, with a ratio as low as 0.09 in the thermocline. Consequently, we are not able to accurately calculate The VBR ratio decreased temporally from 11.25 in the the CO parameters for these samples (referred to as 2 surface on day 249Ð281 in the surface on day 268. N/A in Table 1). Seawater pCO values increased from 2 After convective overturn, the bacterioplankton abun- ∼380 μatm in the surface 20 m to levels ranging from 811 dances returned to uniform concentrations (2.8 ± 1.1 × to > 8759 μatm within the thermocline (Table 1; Fig. 2C). 109 cells l−1; n = 10) (Fig. 2E), while virioplankton abun- During stratification, pH ranged from 7.9 to 8.0 in the top dances and VBR remained uniform and low throughout 20 m and dropped to 6.8Ð7.3 within the suboxic layer the water column (Fig. 2F). (Fig. 2D). After convective overturn, both DIC and TA values were distributed homogeneously throughout the Direct enumeration of bacterioplankton populations by water column and consistent with surface waters values FISH, CARD-FISH and flow cytometry during stratification (Table 1). After overturn, water column pCO2 levels were greater than equilibrium with the atmos- We used the probe Cren-537 to enumerate the phere (561 ± 75 μatm), and pH values were 7.92 ± 0.04 Thaumarchaeota in this study. The probe was designed to throughout the water column (Fig. 2C and D). detect sequences of environmental Crenarchaea, specifi- Inorganic nutrient chemistry was measured on days cally Marine Group I, and has been used for this purpose 255, 268, 274 and 282 (Table 1). During stratification, in multiple studies (DeLong et al., 1999; Karner et al., concentrations of nitrite, nitrate and ammonia re- 2001; Teira et al., 2004). Recently, the Thaumarchaeota mained low (< 1 μmol l−1) in the surface 20 m and (i.e. ammonia-oxidizing Archaea) have been recognized increased (> 1 μmol l−1) in the deeper suboxic bottom as their own phylum distinct from the Crenarchaea waters (Table 1). Peak ammonium concentrations were (Brochier-Armanet et al., 2008). 10.9 μmol l−1 in the suboxic layer (24 m) on day 255 and Cells belonging to the SAR11 clade of Alphapro- increased to 50.8 μmol l−1 on day 268, an increase of teobacteria dominated the known bacterioplankton in the

Table 1. Table of physical and chemical parameters measured for all Devil’s Hole sampling dates from 5 September 2008 (day 249) to 8 October 2008 (day 282).

Sample Depth Date DO Temperature Salinity pCO2 DIC Alkalinity Nitrite Nitrate Ammonium Phosphate ID m mm/dd/yyDay of year (μmol l−1) ¡C psu pH (μatm) (μmol l−1)(μmol l−1)(μmol l−1)(μmol l−1)(μmol l−1)(μmol l−1)

DH1 SFC 1.0 09/05/08 249 197.08 29.11 36.32 7.953 504 2006 2296 N/A N/A N/A N/A DH1 10 10.0 09/05/08 249 205.21 28.93 36.32 7.935 527 2006 2283 N/A N/A N/A N/A DH1 15 15.0 09/05/08 249 192.53 28.98 36.32 7.942 520 2015 2299 N/A N/A N/A N/A DH1 17 17.0 09/05/08 249 189.63 28.98 36.33 7.939 524 2017 2299 N/A N/A N/A N/A DH1 19 19.0 09/05/08 249 183.51 28.91 36.33 7.944 518 2014 2298 N/A N/A N/A N/A DH1 22 22.0 09/05/08 249 83.16 28.50 36.27 7.885 617 2077 2327 N/A N/A N/A N/A DH1 23 23.0 09/05/08 249 48.05 28.25 36.14 7.790 811 2158 2358 N/A N/A N/A N/A DH1 24 24.0 09/05/08 249 BD 25.98 36.32 7.146 4376 2627 2563 N/A N/A N/A N/A DH2 SFC 1.0 09/11/08 255 214.41 29.93 36.30 8.026 408 1952 2294 0.10 0.13 0.30 0.06 DH2 10 10.0 09/11/08 255 210.21 29.63 36.32 8.018 417 1957 2291 N/A N/A N/A N/A DH2 16 16.0 09/11/08 255 174.84 29.06 36.31 7.971 476 1987 2287 0.30 0.92 0.36 0.05 DH2 20 20.0 09/11/08 255 99.50 28.82 36.31 7.943 520 2020 2304 N/A N/A N/A N/A DH2 21 21.0 09/11/08 255 86.77 28.81 36.30 7.772 831 2118 2312 N/A N/A N/A N/A DH2 22 22.0 09/11/08 255 71.46 28.95 36.28 7.755 880 2148 2335 3.96 4.51 0.55 0.07 DH2 23 23.0 09/11/08 255 19.20 27.87 36.20 7.586 1376 2246 2355 3.35 5.06 0.88 0.07 DH2 24 24.0 09/11/08 255 BD 26.49 36.31 7.382 2379 2432 2460 1.18 0.67 10.91 0.23 DH3 SFC 1.0 09/18/08 262 197.00 28.64 36.02 7.941 517 1996 2271 N/A N/A N/A N/A DH3 10 10.0 09/18/08 262 193.42 28.74 36.46 7.914 562 2030 2297 N/A N/A N/A N/A

DH3 16 16.0 09/18/08 262 185.52 28.83 36.39 7.904 577 2035 2296 N/A N/A N/A N/A waters hypoxic seasonally within turnover Bacterioplankton DH3 22 22.0 09/18/08 262 182.88 28.66 36.39 7.887 605 2048 2298 N/A N/A N/A N/A DH3 23 23.0 09/18/08 262 174.00 28.55 36.39 7.877 624 2056 2301 N/A N/A N/A N/A niomna Microbiology Environmental DH3 24 24.0 09/18/08 262 177.61 27.19 36.33 7.653 1166 2242 2376 N/A N/A N/A N/A DH3 25 25.0 09/18/08 262 BD 25.46 36.22 N/A N/A 2896 N/A N/A N/A N/A N/A DH4 SFC 1.0 09/24/08 268 197.93 27.52 36.34 7.939 523 2017 2285 0.11 0.70 1.28 0.04 DH4 16 16.0 09/24/08 268 181.59 27.06 36.34 7.941 521 2024 2291 0.14 0.55 0.47 0.05 DH4 20 20.0 09/24/08 268 171.09 26.72 36.42 7.914 562 2043 2293 N/A N/A N/A N/A DH4 22 22.0 09/24/08 268 154.75 26.97 36.42 7.873 629 2066 2296 N/A N/A N/A N/A DH4 23 23.0 09/24/08 268 128.08 26.89 36.41 7.805 759 2111 2308 0.25 0.68 7.52 0.07 DH4 24 24.0 09/24/08 268 61.01 26.74 36.41 7.639 1192 2221 2345 0.23 0.23 14.42 0.10 DH4 24.5 24.5 09/24/08 268 BD 25.32 36.39 6.833 8759 2672 2454 0.21 0.39 50.80 0.97 DH4 25 25.0 09/24/08 268 BD N/A 36.34 N/A N/A 2807 N/A N/A N/A N/A N/A DH5 SFC 1.0 09/30/08 274 202.53 27.82 36.22 7.943 515 2004 2275 0.14 0.79 0.04 0.04 DH5 16 16.0 09/30/08 274 149.30 27.89 36.41 7.901 580 2038 2290 0.17 0.73 7.86 0.09 DH5 22 22.0 09/30/08 274 148.45 27.77 36.42 7.872 630 2058 2294 0.19 1.00 0.73 0.06 DH5 25 25.0 09/30/08 274 138.54 27.70 36.42 7.867 642 2073 2306 0.20 1.04 2.50 0.07 DH5 25.5 25.5 09/30/08 274 137.91 27.58 36.43 7.841 707 2143 2366 0.22 0.70 8.94 0.14 DH6 SFC 1.0 10/08/08 282 194.72 26.70 35.69 7.953 503 2014 2283 0.18 1.42 0.90 0.06 DH6 16 16.0 10/08/08 282 194.81 26.68 36.28 7.943 516 2017 2280 0.14 1.18 0.67 0.05 DH6 22 22.0 10/08/08 282 192.17 26.48 36.27 7.950 507 2017 2283 0.13 1.22 0.79 0.05 DH6 23 23.0 10/08/08 282 190.92 26.62 36.29 7.959 494 2011 2283 0.11 1.18 0.84 0.04 DH6 24 24.0 10/08/08 282 184.54 26.59 36.29 7.943 520 2030 2294 0.22 1.42 3.92 0.06

Sampling information includes sample ID, Depth (m), Date (mm/dd/yy) and Day of year. Salinity is measured in practical salinity units (psu). BD denotes below the detection level of the analysis. N/A denotes that the measurement is not available. Light grey denotes sampling dates during stratiﬁcation. Medium grey denotes partial overturn. Dark grey indicated complete overturn. On day 282, the salinity was at its lowest (35.69 practical salinity unit) at the surface as it was taken 2 days after a large rain event (2.6 inches; http://www.weather.bm/climate.asp). 5 6 .J Parsons J. R. Table 2. Bacterioplankton parameters measured for all Devil’s Hole sampling dates from 5 September 2008 (day 249) to 8 October 2008 (day 282).

Sample Depth Date Day of DO Bacterioplankton Virioplankton Chlorobium Thaumarchaea Euryarchaea SAR11 Synechococcus Alteromonas Bacteroidetes Rhodobacteraceae ID m mm/dd/yyyear (μmol l−1) (cells l−1) (VLP l−1) (cells l−1) (cells l−1) (cells l−1) (cells l−1) (cells l−1) (cells l−1) (cells l−1) (cells l−1)

DH1 SFC 1.0 09/05/08 249 197.08 3.4E + 09 3.8E + 10 5.1E + 07 3.8E + 07 2.0E + 08 9.2E + 08 2.4E + 08 1.9E + 08 3.2E + 08 1.3E + 08 al. et DH1 10 10.0 09/05/08 249 205.21 2.7E + 09 2.1E + 10 1.7E + 07 1.0E + 08 1.8E + 08 7.3E + 08 2.5E + 08 4.1E + 07 1.6E + 08 8.8E + 07 DH1 15 15.0 09/05/08 249 192.53 2.2E + 09 1.9E + 10 4.3E + 07 4.5E + 07 7.9E + 07 4.7E + 08 2.1E + 08 6.6E + 07 1.1E + 08 8.3E + 07 DH1 17 17.0 09/05/08 249 189.63 1.9E + 09 2.1E + 10 0.0E + 00 8.0E + 06 7.4E + 07 3.3E + 08 2.1E + 08 9.7E + 07 8.4E + 07 3.1E + 07 DH1 19 19.0 09/05/08 249 183.51 2.1E + 09 1.2E + 10 0.0E + 00 3.4E + 07 3.1E + 07 5.5E + 08 1.9E + 08 4.3E + 07 2.3E + 07 1.7E + 07 DH1 22 22.0 09/05/08 249 83.16 2.9E + 09 2.2E + 10 4.8E + 07 3.5E + 08 3.5E + 07 3.6E + 08 9.9E + 07 5.1E + 06 3.2E + 07 2.6E + 07 DH1 23 23.0 09/05/08 249 48.05 3.3E + 09 1.7E + 10 1.5E + 08 6.2E + 08 1.7E + 08 4.5E + 08 7.0E + 07 2.6E + 07 1.6E + 07 1.8E + 07

04SceyfrApidMcoilg n onWly&Sn Ltd, Sons & Wiley John and Microbiology Applied for Society 2014 © DH1 24 24.0 09/05/08 249 BD 2.1E + 10 1.4E + 10 1.3E + 10 8.7E + 08 1.7E + 09 1.4E + 09 7.3E + 07 7.6E + 08 1.9E + 08 4.3E + 07 DH2 SFC 1.0 09/11/08 255 214.41 2.6E + 09 2.0E + 10 2.4E + 07 2.1E + 08 1.8E + 08 5.4E + 08 5.3E + 08 2.3E + 08 2.5E + 08 1.7E + 08 DH2 10 10.0 09/11/08 255 210.21 2.3E + 09 1.8E + 10 0.0E + 00 3.4E + 08 1.4E + 08 4.5E + 08 5.0E + 08 2.4E + 07 1.8E + 08 1.6E + 08 DH2 16 16.0 09/11/08 255 174.84 2.6E + 09 1.6E + 10 2.8E + 07 3.3E + 07 1.9E + 08 5.0E + 08 3.1E + 08 1.2E + 08 2.4E + 08 8.9E + 07 DH2 20 20.0 09/11/08 255 99.50 2.7E + 09 1.4E + 10 0.0E + 00 5.9E + 07 9.6E + 07 3.7E + 08 5.3E + 07 5.3E + 06 2.6E + 08 2.0E + 08 DH2 21 21.0 09/11/08 255 86.77 3.6E + 09 1.2E + 10 0.0E + 00 3.8E + 07 1.2E + 08 4.6E + 08 5.9E + 07 1.5E + 06 2.4E + 08 2.2E + 08 DH2 22 22.0 09/11/08 255 71.46 3.8E + 09 9.0E + 09 0.0E + 00 1.6E + 08 2.8E + 08 3.4E + 08 5.4E + 07 2.7E + 08 3.3E + 08 2.7E + 08 DH2 23 23.0 09/11/08 255 19.20 1.5E + 10 6.2E + 09 1.7E + 09 8.7E + 08 1.5E + 09 1.5E + 09 3.8E + 07 2.7E + 08 8.3E + 08 9.7E + 08 DH2 24 24.0 09/11/08 255 BD 1.9E + 10 6.3E + 09 6.0E + 09 1.5E + 09 1.9E + 09 1.3E + 09 1.6E + 07 6.4E + 08 7.4E + 08 7.3E + 08 DH3 SFC 1.0 09/18/08 262 197.00 4.4E + 09 1.1E + 10 3.9E + 07 1.3E + 08 2.4E + 08 1.0E + 09 4.3E + 08 1.9E + 08 2.3E + 08 2.3E + 08 DH3 10 10.0 09/18/08 262 193.42 4.8E + 09 1.7E + 10 1.7E + 07 7.1E + 08 4.2E + 08 1.1E + 09 4.5E + 08 2.9E + 08 1.8E + 08 2.4E + 08 DH3 16 16.0 09/18/08 262 185.52 4.8E + 09 2.0E + 10 0.0E + 00 1.7E + 08 1.7E + 08 9.9E + 08 3.1E + 08 1.6E + 08 2.2E + 08 3.9E + 08 DH3 22 22.0 09/18/08 262 182.88 3.2E + 09 1.0E + 10 4.8E + 06 1.2E + 08 1.2E + 08 6.7E + 08 1.7E + 08 3.2E + 07 2.3E + 08 2.1E + 08 DH3 23 23.0 09/18/08 262 174.00 3.6E + 09 4.1E + 09 2.5E + 06 2.2E + 08 7.9E + 07 5.7E + 08 1.6E + 08 3.3E + 07 1.4E + 08 1.7E + 08 DH3 24 24.0 09/18/08 262 177.61 1.1E + 10 3.3E + 09 7.6E + 08 5.6E + 08 7.0E + 08 5.1E + 08 1.0E + 08 6.2E + 07 1.3E + 08 6.5E + 07 DH3 25 25.0 09/18/08 262 BD 2.1E + 10 2.3E + 09 6.5E + 09 2.5E + 09 5.2E + 08 5.7E + 08 6.6E + 07 1.4E + 08 1.9E + 08 4.7E + 08 DH4 SFC 1.0 09/24/08 268 197.93 3.5E + 09 9.9E + 09 7.3E + 07 8.7E + 07 2.3E + 08 8.6E + 08 2.8E + 08 1.4E + 08 9.2E + 07 3.4E + 08 DH4 16 16.0 09/24/08 268 181.59 3.4E + 09 1.3E + 10 8.8E + 07 6.7E + 07 1.7E + 08 6.8E + 08 2.9E + 08 1.6E + 08 3.9E + 08 2.4E + 08 DH4 20 20.0 09/24/08 268 171.09 2.7E + 09 1.5E + 10 3.9E + 07 N/A N/A 5.1E + 08 2.1E + 08 2.1E + 08 3.6E + 08 2.1E + 08 DH4 22 22.0 09/24/08 268 154.75 3.0E + 09 1.7E + 10 7.5E + 07 1.2E + 08 9.9E + 05 6.8E + 08 1.8E + 08 2.9E + 08 2.5E + 08 1.9E + 08 DH4 23 23.0 09/24/08 268 128.08 5.2E + 09 8.1E + 09 1.3E + 08 4.7E + 08 2.1E + 07 1.1E + 09 1.1E + 08 2.2E + 08 3.2E + 08 3.0E + 08 DH4 24 24.0 09/24/08 268 61.01 5.8E + 09 7.1E + 09 4.8E + 08 1.0E + 09 2.3E + 08 1.1E + 09 6.5E + 07 2.4E + 08 1.8E + 08 1.7E + 08 DH4 24.5 24.5 09/24/08 268 BD 1.7E + 10 4.6E + 09 3.1E + 09 4.2E + 08 2.3E + 08 9.4E + 08 7.9E + 07 3.0E + 08 9.7E + 08 3.6E + 08 DH4 25 25.0 09/24/08 268 BD 1.3E + 10 8.1E + 09 3.6E + 09 1.8E + 08 1.6E + 08 6.1E + 08 6.9E + 07 2.3E + 08 1.3E + 09 1.7E + 08 DH5 SFC 1.0 09/30/08 274 202.53 3.6E + 09 2.6E + 09 0.0E + 00 2.0E + 08 2.6E + 08 8.9E + 08 3.6E + 08 2.2E + 08 2.6E + 08 5.9E + 07 DH5 16 16.0 09/30/08 274 149.30 3.5E + 09 5.0E + 09 0.0E + 00 6.0E + 07 9.1E + 07 6.4E + 08 1.2E + 08 1.6E + 08 1.5E + 08 1.3E + 08 DH5 22 22.0 09/30/08 274 148.45 2.7E + 09 3.0E + 09 0.0E + 00 3.0E + 07 6.0E + 07 4.7E + 08 4.5E + 07 8.2E + 07 1.0E + 08 8.7E + 07 DH5 25 25.0 09/30/08 274 138.54 4.9E + 09 3.6E + 09 2.1E + 08 4.2E + 08 2.3E + 08 9.5E + 08 4.4E + 07 4.8E + 07 3.1 E + 08 3.2E + 08 niomna Microbiology Environmental DH5 25.5 25.5 09/30/08 274 137.91 3.2E + 09 3.3E + 09 3.7E + 08 3.4E + 08 2.4E + 07 5.8E + 08 4.3E + 07 1.5E + 08 2.8E + 08 2.1E + 08 DH6 SFC 1.0 10/08/08 282 194.72 2.6E + 09 2.3E + 09 2.7E + 06 4.5E + 07 3.8E + 08 6.6E + 08 2.3E + 08 2.1E + 08 2.4E + 08 1.6E + 08 DH6 16 16.0 10/08/08 282 194.81 1.2E + 09 2.8E + 09 1.0E + 06 1.6E + 07 1.7E + 07 2.7E + 08 2.2E + 08 1.3E + 08 3.6E + 07 7.1E + 07 DH6 22 22.0 10/08/08 282 192.17 2.0E + 09 2.1E + 09 5.7E + 06 1.9E + 07 7.9E + 07 5.3E + 08 2.0E + 08 1.3E + 08 7.9E + 07 1.4E + 08 DH6 23 23.0 10/08/08 282 190.92 1.4E + 09 2.2E + 09 3.0E + 06 1.4E + 07 2.2E + 07 3.7E + 08 2.1E + 08 5.5E + 07 1.4E + 08 8.9E + 07 DH6 24 24.0 10/08/08 282 184.54 3.1E + 09 2.1E + 09 1.4E + 07 5.8E + 07 2.7E + 08 6.9E + 08 2.0E + 08 2.2E + 08 3.7E + 08 2.1E + 08

BD denotes below the detection level of the analysis. N/A denotes that the measurement is not available. VLP denoted virus-like particles. Light grey denotes sampling dates during stratiﬁcation. Medium grey denotes partial overturn. Dark grey indicated complete overturn. Bacterioplankton turnover within seasonally hypoxic waters 7

Fig. 3. Contour plots of abundance of various members of the bacterioplankton community: (A) SAR11; (B) Chlorobiaceae; (C) Synechococcus; (D) Thaumarcheota; (E) Bacteroidetes; (F) Alteromonas; (G) Rhodobacteraceae; (H) Euryarcheota (×108 cells l−1) for all Devil’s Hole sampling dates from 5 September 2008 (day 249) to 8 October 2008 (day 282). Note: the bacterioplankton name reflects the specificity of the probe used for FISH analysis (Supporting Information Table S1). surface 20 m during stratification averaging 6.7 ± 2.3 × SAR11 and Synechococcus abundances decreased 108 cells l−1 (Table 2; Fig. 3A). This represented 22.2% ± with depth during stratification. Thaumarcheota increased 3.6% of the total bacterioplankton community (Supporting in abundance between 22 and 23 m to a maximum of Information Fig. S1). Synechococcus, determined using 2.97 × 109 cells l−1 at 25 m on day 262 (Table 2; Fig. 3D), flow cytometry, contributed ∼20% to the total bacterio- averaging 9.8% ± 5.6% of the total bacterioplankton com- plankton community, while members of Bacteroidetes, munity (Supporting Information Fig. S1). Between 16 and Alteromonas and Rhodobacteraceae comprised up to 22 m, Euryarcheota decreased by an order of magnitude 13%, 9% and 9% of the total bacterioplankton community, averaging only 2.3% ± 1.5% of the total bacterioplankton respectively, in the surface 20 m during stratification community and then increased within the suboxic layer (Supporting Information Fig. S1). During stratification, reaching a maximum of 1.91 × 109 cells l−1 at 24 m on day Thaumarcheota and Euryarcheota were less abundant 255 comprising 10.1% of the total bacterioplankton popu- averaging < 2 × 108 cells l−1 from surface to 20 m (Table 2; lation (Table 2; Fig. 3H, Supporting Information Fig. S1). Fig. 3D and H). Within the suboxic layer, SAR11 was initially abundant at

1.4 × 109 cells l−1 on day 249, decreasing to barely de- used to symbol/colour code samples (Fig. 4C): ‘surface tectable levels prior to overturn on day 268, while stratified’, deep ‘transitional’ (DO 90Ð160 umol l−1), Synechococcus did not contribute significantly to the total ‘dysoxic’ (DO 20Ð90 umol l −1) and ‘suboxic’ bacterioplankton community (< 1%) within suboxic deep (DO < 20 umol l −1), and post-overturn ‘mixed’. Commu- waters. In the suboxic depth horizon, Bacteriodetes nity differentiation by habitat was tested using analysis of ranged from < 1% to as much as 9.8% of the total similarity (ANOSIM); all five habitats differed strongly and bacterioplankton community (Supporting Information significantly (pairwise R > 0.5, P < 0.001) with suboxic Fig. S1). and dysoxic habitat classifications markedly distinct from Cells belonging to the family Chlorobiaceae increased the remaining habitats and from each other (pairwise with depth and dominated suboxic layer bacterioplankton. R > 0.8). These differences are further clarified through Chlorobiaceae cell densities were negatively correlated visualizations in the cluster dendrogram (Fig. 4A) and with oxygen concentration (r =−0.7063, P < 0.0001, NMDS ordination (Fig. 4B). Samples associated with n = 32) and greatest under completely suboxic con- the surface 20 m during the stratified period (surface ditions. However, the contribution of Chlorobiaceae stratified; AugustÐSeptember) were relatively similar to changed temporally within the suboxic layer from an initial samples observed throughout the water column following contribution of 58.9% of the bacterioplankton community convective mixing (mixed; OctoberÐNovember). In con- (Fig. 3B, Supporting Information Fig. S1), declining tem- trast, deep water samples associated with dysoxic and porally to 27.8% of the bacterioplankton community prior suboxic zones as well as the adjacent- and post-dysoxic to overturn (Fig. 3B, Supporting Information Fig. S1). ‘transitional’ zones were clearly distinct from surface Early in the study, Alteromonas increased with depth, water samples and from each other (Fig. 4B). Multiplex reaching a maximum of 7.6 × 108 cells l−1 at 24 m on day 454 pyrosequencing of the rRNA genes of 10 representa- 249 (Table 2; Fig. 3F), while members of the Rhodobacte- tive samples (one to three samples from each of the five raceae family reached their highest abundance of TRFLP community types) confirmed differential commu- 9.7 ± 0.5 × 108 cells l−1 at 23 m on day 255 (Fig. 3G). As nity composition, with significant Unifrac complete linkage the suboxic layer deepened, Alteromonas and Rhodo- clustering (SIMPROF P < 0.05) separating suboxic, bacteraceae showed a temporal decline (Fig. 3). dysoxic, transitional and surface/mixed communities at After convective overturn, SAR11 again dominated, the 95% operational taxonomic unit (OTU) sequence averaging 6.1 ± 2.1 × 108 cells l−1 throughout the water identity level (Fig. 5). Consensus phylotypes defined from column (Table 2; Fig. 3A) and comprising 22.1% ± 3.6% alignment of pyrosequencing reads to the SILVA 16S of the total bacterioplankton community (Supporting rRNA gene database and naïve Bayesian classification Information Fig. S1). After overturn on day 282, Synecho- (Wang et al., 2007) generally sorted into two categories: coccus averaged 11.7% ± 4.7%, Bacteroidetes com- those that were most enriched in the surface communities prised 7.5% ± 3.9%, Alteromonas averaged 7.2% ± (with oxygen > 150 μmol l−1) and those that became domi- 2.4%, Rhodobacteraceae contributed to 6.4% ± 0.7%, nant mainly in dysoxic and suboxic bottom waters (Fig. 5). Thaumarcheota comprised only 1.1% ± 0.7%, and Eury- The former taxa included many oligotrophic ocean arcahea averaged 5.9% ± 5.5% of the total bacterio- bacterioplankton clades known from the surrounding Sar- plankton community throughout the water column, while gasso Sea, including SAR11, SAR86, SAR116, SAR324, Chlorobiaceae levels were barely detected (Supporting SAR406, Synechococcus, Roseobacteraceae, OM190 Information Fig. S1). Cells belonging to the Vibrio spp. and the oligotrophic NS4, and NS5 clades of the were not detected in Devil’s Hole with the FISH probes Flavobacteria (Fig. 5). used in this study (Supporting Information Table S1). 454 pyrosequences of 16S rRNA gene amplicons from suboxic/dysoxic water samples were enriched in several taxa relative to samples collected from the Bacterial community structure using TRFLP analysis oxygenated mixed water column. The deep suboxic and pyrosequencing of the 16S rRNA gene layer included sequences matching the Chlorobium Community structure variation among samples measured and Prosthecochloris genera of the Chlorobiales, the by TRFLP was analysed using multivariate distance matri- Desulfobacteraceae and Desulfobulbaceae clades of ces (BrayÐCurtis dissimilarity among 51 TRFLP samples the Deltaproteobacteria, the SVA0071 clade of the in restriction fragment relative abundance space; Nelson, Gammaproteobacteria, the Thermosinus clade of the 2009) and subsequently visualized using both hierarchical Firmicutes, the candidate divisions OP3 and OD1, clustering (Fig. 4A) and non-metric multidimensional and the VC2.1 clade of the Bacteriodetes. Dysoxic and scaling (NMDS; stress = 0.09; Fig 4B) algorithms. A priori transitional samples were somewhat less enriched in designations of five distinct habitat classifications based Chlorobiales than the suboxic samples and additionally on thermoclines (Fig. 2A) and oxyclines (Fig. 2B) were exhibited elevated levels of the SUP05 subclade of

Fig. 4. Bacterial community types through time and space defined by TRFLP. Samples are complete-linkage hierarchically clustered (A) according to TRFLP fragment relative abundance and coded according to a priori community types defined by thermocyclines and oxyclines: suboxic (black cross); dysoxic (grey triangle); surface stratified (green square); mixed (blue circle); and transitional (orange open square). Dendrogram branches highlighted in red are not significantly different (SIMPROF P > 0.05). Samples starting with ‘W’ are those in JulyÐAugust and NovemberÐDecember outside of the high-resolution sampling period (C). NMDS Ordination (B) shows the relative similarity of community types, with samples selected for 16S amplicon pyrosequencing circled and labelled (two-dimensional solution stress = 0.09). Community types (C) shown in spatiotemporal context throughout the time series according to sampling date (x-axis) and sampling depth (y-axis) in Devil’s Hole. Cruises are labelled across the top of the time series for reference and a legend matching community type clusters from (A) is included. DH1 corresponds to day 249; DH2 to day 255; DH3 to day 262; DH4 to day 268; DH5 to day 274 and DH6 to day 282.

Oceanospirillales [widespread pelagic suboxic sulfur- of Chlorobiaceae within the suboxic waters and a oxidizing bacterioplankton; sensu (Walsh et al., 2009)] as dominance of SAR11 within the other water types (Fig. 6). well as two poorly classified genera of the Gamma- Synechococcus was present in the surface stratified and proteobacteria (Arenicella and Thiobos), Planctomyces mixed water types for both methods (Fig 6). Vibrio species and two clades of Sphingobacteria (Fig. 5). were not measured in either analysis (Fig. 6). Clone library frequency distributions reflected those of the FISH and pyrosequencing analyses. In the surface waters, the Comparison of community structure from majority of clones (38.3%) belonged to the SAR11 clade pyrosequencing, clone libraries and FISH of Alphaproteobacteria, with the next most abundant The community structure was tested by aggregating 16S clone belonging to the Rhodobacteraceae (17.3%). The rRNA gene libraries according to the level of taxonomic highest percentage abundance (48%) on day 268 (DH4) resolution offered by each FISH probe (Supporting Infor- within the suboxic layer at depth 24.5 m matched mation Table S1) in order to determine the consistency Chlorobium and was used to develop the FISH probe as and quantitative replicability of the pyrosequencing and noted earlier. The next most abundant clone (29.4%) in FISH analysis. Both methods showed a dominance the DH4 suboxic waters was related to Desulfocapsa.

Fig. 5. Heat map of relative abundances of dominant phylotypes of bacteria among Devil’s Hole samples analysed by 16S amplicon pyrosequencing. Samples cluster together (bottom) according to a priori community types via complete linkage hierarchical clustering using either weighted Unifrac distances or BrayÐCurtis distances calculated from raw phylotype relative abundance. Taxa are organized according to relative abundance and relative enrichment in oxic surface waters or sub-bottom waters, with the most abundant taxa in either habitat at either end. The 43 phylotypes shown are those comprising 1% of total sequences in at least one of the 10 samples sequenced. Each phylotype is annotated according to the SILVA v115 (2013) taxonomy and condensed for simplicity to one broad grouping (phylum or class), one clade assignment (approximating order/family level where appropriate) and a putative genus assignment where at least 70% of sequences met the consensus bootstrapping conﬁdence criteria. Community types (symbol/colour annotated at bottom) are as in Fig. 4: suboxic (black cross); dysoxic (grey open triangle); surface stratiﬁed (green square); mixed (blue circle) and transitional (orange open square). DH1 corresponds to day 249, DH2 to day 255, DH3 to day 262, DH4 to day 268, DH5 to day 274, and DH6 to Day 282.

Fig. 6. Bacterial community structure as determined by 454 pyrosequencing data analysis and FISH analysis for a subset of the Devil’s Hole samples taken in 2008. Ten samples were chosen that represented the following types: suboxic (black cross), dysoxic (grey triangle), surface stratiﬁed (green square), mixed (blue circle) and transitional (orange open square). The top two legend rows represent the six taxa targeted with both FISH (Table 2 and Fig. 3) and pyrosequencing (Fig. 5). For comparability, pyrosequencing data were taxonomically aggregated to correspond to the phylogenetic resolution of the FISH/Flow Cytometry direct counts (e.g. all sequences classiﬁed in the phylum Bacteroidetes were summed together for comparison with the phylum-level Bacteroidetes FISH probe). A subset of order-level clades comprising > 5% of the community of one or more samples (Fig. 5) are included in the pyrosequencing data with the remaining sequences labelled ‘other’. For FISH, probed abundances are relativized to total bacterial counts (bacterioplankton minus archaeal) with the remainder labelled ‘not probed’. DH1 corresponds to day 249, DH2 to day 255, DH3 to day 262, DH4 to day 268, DH5 to day 274, and DH6 to day 282.

Discussion another eutrophic sound, the Chesapeake Bay, surface water populations were dominated by Synechococcus Bacterioplankton community changes when oxygen sp., the SAR86 clade and SAR11 clade (Crump et al., becomes limiting 2007). In the Chesapeake Bay, oxygen limitation resulted Many bacterioplankton processes are affected either in a bacterioplankton community initially dominated by directly or indirectly via changes in seawater chemistry denitriﬁers and successively, after the depletion of nitrate, and vice versa (Liu et al., 2010). Bacterioplankton pro- by sulfate reducers (Crump et al., 2007). Populations cesses help to drive the oxidative state of the carbon cycle that developed in suboxic water were most similar (Giovannoni and Vergin, 2012) and are major players in (< 92% similarity) to uncultivated Firmicutes, uncultivated the global biogeochemical cycles including the nitrogen Bacteroidetes, Gammaproteobacteria in the genus cycle (Francis et al., 2007). Thioalcalovibrio and the uncultivated SAR406 cluster SAR11, Synechococcus, Rhodobacteraceae and (Crump et al., 2007). These results were similar to those Bacteroidetes abundances were present within the strati- from Devil’s Hole (Fig. 5); however, the deep suboxic layer ﬁed layer of Devil’s Hole (Table 2; Figs 3 and 5). Within also included Chlorobiaceae, the Desulfobacteraceae

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology 12 R. J. Parsons et al. and Desulfobulbaceae families of the Deltaproteo- sulfur compounds, is occurring at Devil’s Hole (Table 1; bacteria, while Firmicutes played a minor role. Fig. 5). Previous work in Harrington Sound, Bermuda suggested that anoxygenic photoautotrophic populations Viral dynamics within the suboxic layer were responsible for elevated chlorophyll A concentrations at the suboxic interface (Smetacek et al., 1981). Initially, in Devil’s Hole, the VBR was as high as 11.42 Using clone libraries, 454 pyrosequencing and quantifi- within the stratified layer, and by day 262 as low as 0.09 cation via FISH, we confirmed that the anoxygenic within the suboxic layer. This low VBR in the suboxic photolithotroph Chlorobiaceae were present within Devil’s layer is not typical of similar systems. In suboxic environ- Hole (Figs 3 and 6). This observation was consistent with ments like the Baltic Sea, viral decay rates are slow, and studies from anoxic zones of the Black Sea that show that the VBR and viral production are high (Pradeep Ram Chlorobium sp, a photolithoautotrophic green sulfur bac- et al., 2009). In the Baltic Sea, the VBR was as high as terium containing bacteriochlorophyll e, is prevalent 50 in the suboxic zone with virus-induced mortality at the suboxic/dysoxic transition zone and well adapted of bacterioplankton reaching 51% within this layer to extraordinarily low light levels (Overmann et al., (Weinbauer et al., 2003). However, in deep Mediterra- 1992; Taipale et al., 2009). In the Cariaco Basin, nean sediments, viruses have had little effect on bacterial bacterioplankton chemoautotrophy is driven by reduced abundance (Danovaro and Serresi, 2000). These sedi- sulfur species within the redoxcline (Taylor et al., ments also had a low VBR (< 1), and the bacterial and 2001) and members related to the sulfate-reducing Del- viral abundances did not significantly correlate suggest- taproteobacteria, Desulfobulbus mediterraneus were ing that the number of viruses may be dependent upon present within this zone (Rodriguez-Mora et al., 2013). other factors such as redox conditions (Danovaro et al., Sulfate-reducing taxa such as Desulfatibacillum, 2002). Low VBRs have also been seen in anoxic Desulfobacterium, Desulfococcus, Syntrophobacter and freshwaters and sediments where viruses removed only Desulfovibrio species have also been observed in the 9% of bacterial production in deep anoxic waters and 3% oxygen minimum zone off the Chilean Coast (Canfield in the sediments (Pradeep Ram et al., 2009). Thus, the et al., 2010). Sulfate reducers belonging to Deltapro- role played by viruses in the suboxic layer at Devil’s Hole teobacteria were present in both Devil’s Hole appeared to have a smaller impact on bacterial removal (Desulfocapsa and Desulfobacula in Fig. 5) and the than in the stratified layer. Any nutrient enrichment by redoxcline of the Cariaco Basin (Lin et al., 2006). viral lysis within the suboxic layer would be minimal and An uncultivated microbe referred to as SUP05 has viral abundance may well be impacted by redox condi- been found in the seasonally anoxic Saanich Inlet tions within this suboxic environment. However, it must (British Columbia, Canada) (Zaikova et al., 2010). be noted that total VLP and total prokaryotic abundance The metagenome of SUP05 has been analysed and often obscure the variability associated among specific found to include a diverse range of genes involved viral and prokaryotic taxa (Parsons et al., 2011), and in chemolithotrophic oxidation of reduced sulfur com- further work is required to better understand the controls pounds (Walsh et al., 2009). In suboxic and anoxic of viral dynamics in hypoxic systems. waters, SUP05 was dominant with Desulphobacte- raceae, Arcobacteraceae, Bacteroidetes, Thaumar- Bacterioplankton community structure and nutrient cheota and even SAR11 thriving in oxygen deficient cycling within the suboxic layer waters (Wright et al., 2012). Because SAR11 has a role in mediating the demethylation of dimethylsul- Devil’s Hole is fairly unusual when compared with other foniopropionate to methylmercaptopropionate; this may seasonal oxygen minimum zones because the suboxic indicate a role for this group in sulfur cycling when layer is within the photic zone. As a result, Chlorobiaceae, oxygen becomes limiting (Wright et al., 2012). SUP05 a family of green sulfur bacteria that uses hydrogen is important in sulfur cycling within oxygen minimum sulfide as an electron source and carbon dioxide as its zones (Wright et al., 2012). The SUP05 taxa were also carbon source, dominated within the suboxic layer found to be dominant and active within the oxygen (Table 2; Figs 3, 5 and 6). Thaumarchaeota, chemo- minimum zones community off the Chilean Coast lithoautotrophic ammonia-oxidizers, were abundant at the (Canfield et al., 2010; Stewart et al., 2012). Members of oxic-anoxic interface or oxycline (Table 2; Fig. 3). Oxygen the SUP05 subclade of Oceanospirillales were abundant stress tolerance may have enabled microbes to colonize within the transitional and dysoxic zones of Devil’s Hole microniches, especially at the oxycline (Brune et al., (Fig. 5). The presence of both SUP05 and DO within the 2000). Other adaptations to this zone include sulfide transitional zone suggests that the sulfur cycle and, in oxidation, sulfate reduction, methane oxidation and particular, the chemolithotrophic oxidation of reduced methanogenesis, ammonia oxidation and anammox path-

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology Bacterioplankton turnover within seasonally hypoxic waters 13 ways (Brune et al., 2000). As an example, sulfide oxida- bacterial sequences within the oxygen minimum zone tion occurred at the oxycline within a specific niche (Stewart et al., 2012). Anammox transcripts dominated dependent on oxygen and sulfide availability (Nelson the transcriptome in this study (Stewart et al., 2012). et al., 1986). After their initial discovery, Thaumarchaeota were Previous studies in Harrington Sound, Bermuda have thought to thrive in oxygenated oligotrophic conditions measured elevated hydrogen sulfide and methane con- (Martens-Habbena et al., 2009). However, Thaumar- centrations in sediments from Devil’s Hole (Thorstenson chaeota instead are often prominent within the oxycline and Mackenzie, 1974), and these compounds could well (Wright et al., 2012) including those of the South Pacific be present within the suboxic layer of the water column. (Belmar et al., 2011), the Black Sea (Lin et al., 2006) Besides hyperthermophilic Euryarcheota and Crenar- and the Saaniach Inlet (Zaikova et al., 2010). Thaumar- chaeota, members of the Desulfobacterales are known to chaeota were present within the oxycline of Devil’s Hole reduce sulfate (Muyzer and Stams, 2008), and members where ammonia levels were elevated and oxygen levels of these taxa were present within the suboxic zone were reduced (Tables 1 and 2; Figs 3 and 6; Supporting (Figs 3, 5 and 6; Supporting Information Fig. S1). Anaero- Information Fig. S1 and Table S2), demonstrating that bic methane oxidation takes place in marine sediments some lineages appear to be adapted to ammonia rich where a high rate of sulfate reduction also takes place and low oxygen conditions as well. Similar observations (Iversen and Jorgensen, 1985). Many anaerobic sulfate- have been made in oxygen minimum zones, where reducing microbes form aggregates, resulting in higher Thaumarchaeota densities were greatest below the tolerance to oxygen exposure (Krekeler et al., 1998; euphotic zone and when oxygen levels are below Wieringa et al., 2000). Anaerobic oxidation of methane 1 μmol l−1 (Sinninghe Damste et al., 2002; Francis et al., can be coupled to sulfate reduction (Muyzer and Stams, 2005; Lin et al., 2006; Labrenz et al., 2007; Lam et al., 2008) by the formation of cell aggregates with sulfate- 2007; Mincer et al., 2007; Beman et al., 2008; Martens- reducing bacteria (Levipan et al., 2007). Species from the Habbena et al., 2009; Belmar et al., 2011). Previous work phylum Euryarcheota (Methanosarcinaceae) are known has shown that Thaumarchaeota were remarkably suc- to be methylotrophic methane-producing Archaea (van cessful under low oxygen conditions in the Black der Maarel et al., 1999; Levipan et al., 2007). Archaeal Sea (Lin et al., 2006; Francis et al., 2007; Stoica and methanotrophs were identified in the Black Sea (Vetriani Herndl, 2007). With Thaumarchaeota preferring low light et al., 2003) where they may form aggregates with and low oxygen conditions as suggested earlier and with sulfate-reducing bacteria (Durisch-Kaiser et al., 2005). the water column at Devil’s Hole falling within the photic With a Euryarcheota-Desulfobacterales aggregate reduc- zone, the abundance of Thaumarchaeota at this interface ing sulfate to hydrogen sulfide and Chlorobiaceae could be an indication that oxygen and ammonia concen- anaerobically oxidizing this sulfide to elemental sulfur and trations are more important in shaping the ecological back to sulfate (Muyzer and Stams, 2008), all three organ- niche for Thaumarchaeota than the presence of light in isms would thrive and, in fact, do thrive within the suboxic Devil’s Hole. During stratification in late September (Day layer at Devil’s Hole (Table 2; Figs 3, 5 and 6; Supporting 268), cells detected by the Cren-537 probe reached a Information Fig. S1). The relative abundance of Chloro- maximum of 17% of the total prokaryoplankton population biaceae and Desulfobacterales were both at their highest at the oxycline at 24 m (Fig. 3 and Supporting Information within the suboxic layer on day 249 (DH1) and at their Fig. S1). The waters were still oxygenated (61.01 μmol l−1) lowest in the surface on day 268 (DH4; Figs 5 and 6). at this depth and had high levels of ammonia The oxidation of ammonia to nitrite and nitrate plays a (14.4 μmol l−1). Thus, the precursors required by critical role in the nitrogen cycle (Francis et al., 2005). ammonia-oxidizing bacterioplankton would have been Anaerobic ammonium oxidation (anammox) where present. After overturn, ammonia levels declined at depth, ammonia and nitrite are converted anaerobically to N2 gas oxygen levels returned to surface concentrations (Fig. 2), by bacteria (Strous et al., 1999) and ammonia oxidation and Thaumarchaeota were barely detected. within the domain Archaea have been recognized as two new links in the global nitrogen cycle (Zehr and Ward, Comparison of cultivation independent techniques 2002; Francis et al., 2007). Members of Thaumarchaeota (Brochier-Armanet et al., 2008), previously included in the Here, we showed good agreement between the relative phylum Crenarchaea, are capable of ammonia oxidation contribution of several major bacterioplankton clades (i.e. Nitrosopumilus maritimus) (Könneke et al., 2005; enumerated by FISH and that determined from 454 Beman et al., 2008). While both bacteria and archaea can pyrosequencing (Fig. 6). Culture-independent methods oxidize ammonia, studies off northern Chile have indi- can provide vast information about bacterioplankton cated that crenarchaeal ammonia monooxygenase diversity (Rappe and Giovannoni, 2003), biogeography sequences were expressed significantly more than the (Dolan, 2005; Kirchman et al., 2005), temporal succession

© 2014 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology 14 R. J. Parsons et al. in the context of biogeochemistry (Morris et al., 2005; cycling have been hypothesized at global scales to Carlson et al., 2009; Treusch et al., 2009; Giovannoni and explain the dramatic transition in atmospheric oxygen Vergin, 2012; Vergin et al., 2013) and bacterioplankton levels at the boundary between the Proterozoic and networks (Fuhrman and Steele, 2008) that provide a Phanerozoic eras (Johnston et al., 2009). Thus, as a framework for inferring metabolic system functions physically circumscribed system that undergoes reliable (Zaikova et al., 2010). Previous studies that examined transitions, Devil’s Hole holds promise as a natural labo- bacterial community succession in a paddy soil oxygen ratory for studying microbial processes that are postulated gradient (Noll et al., 2004), and in a seasonally anoxic fjord to increase in importance as ocean surface warming pro- (Zaikova et al., 2010), used culture-independent molecu- motes the spread of oxygen minimum zones. lar methods. However, few studies have used both microscopy and molecular techniques to fully characterize the communities. FISH yields quantitative data that can Experimental procedures be used to verify conclusions based on TRFLP and Sample collection pyrosequencing data. Previous studies have raised the issue of polymerase chain reaction (PCR) bias when Samples profiles were collected from five to eight depths at ′ ′ assessing relative contribution based on fragment length Devil’s Hole (32¡19.399 N, 64¡43.08 W) on a weekly basis during September and October 2008, with two additional contribution, clone libraries or sequences (Suzuki and sampling events in August (stratified presuboxic) and Novem- Giovannoni, 1996; Lee et al., 2012). Bias implicit in PCR ber (mixing post-suboxic) only at the surface and bottom reactions required for pyrosequencing can distort the (24 m) depths. Discrete water samples were collected by observed community structure (Lee et al., 2012). Here, hydrocast with a Niskin bottle. Samples were collected and we found good agreement between relative contributions analysed for concentrations of DO, DIC, TA and salinity as of several major bacterioplankton clades enumerated described previously (Andersson et al., 2007). Continuous in by FISH and those determined from 454 pyro- situ profiling of DO, salinity, pH and temperature were conducted with a tethered YSI 556 multiprobe (Yellow Springs, sequencing (Fig. 6). This study combined a suite of OH, USA). Additional water samples were collected via culture-independent methods to provide a robust and Niskin bottle from various depths for bacterial and viral abun- quantitative spatiotemporal characterization of bacterio- dance (preserved at 10% and 1% formaldehyde, respectively, plankton community succession, including FISH, Flow and frozen at −80¡C until final slide preparation), nitrate, Cytometry, amplicon pyrosequencing, clone libraries and nitrite, ammonia and phosphate concentrations, and filtration TRFLP fingerprinting. We compared the broad clade-level for DNA extraction. compositions with those derived from FISH for the 10 DO samples were immediately drawn into 115 ml ground- glass stoppered BOD bottles and fixed with Winkler reagents samples analysed by pyrosequencing (Fig. 6). There were (Andersson et al., 2007). Samples for DIC and TA were similarities between the relative contribution of those drawn after the DO samples into 200 ml glass sample bottles groups enumerated directly with FISH or flow cytometry (Kimax, Kimble Chase, Vineland, NJ, USA) and fixed with a and those evaluated with pyrosequencing data includ- saturated solution of mercuric chloride (HgCl2; 100 μl). Each ing Rhodobacteraceae, SAR11, Chlorobiaceae, Syne- bottleneck was taped with Teflon tape prior to sampling to chococcus, Alteromonas, Vibrio and Bacteroidetes insure a tight seal (Andersson et al., 2007). Samples for (r = 0.9553, P = 0.007, n = 56). salinity were drawn into 250 ml clear borosilicate bottles with plastic insert and screw caps. Nutrient samples were drawn from the Niskin through a polycarbonate 0.8 μm membrane Summary filter to remove particulates and collected in 60 ml high- density polyethylene bottles kept on ice until shore side and Devil’s Hole forms a natural laboratory that can demon- frozen at −20¡C until time of analysis. strate how the bacterioplankton community changes in response to low pH, and O2 and increased levels of pCO2. The bacterioplankton community changed as oxygen Chemical analyses levels were depleted within the thermocline, pH levels Salinity was analysed using an autosalinometer according decreased and pCO2 levels built up levels above to the Bermuda Atlantic Time Series Study (BATS) method- 8000 μatm. These observations of bacterioplankton com- ology (Knap et al., 1997). Seawater DO was determined by munity turnover provide a clearer view of the complex automatic Winkler titrations based on a UV end-point detec- changes in community structure that occur in illuminated tor system according to the BATS protocol (Knap et al., water columns when lower depths transition from oxic to 1997). DIC and TA were analysed according to Andersson and colleagues (2007). DIC parameters pCO2 and pHTOT suboxic conditions. This in turn leads to the development + (pH defined on a total H scale) were calculated at in situ of hypotheses regarding the biogeochemical relationships temperature and salinity conditions based on TA and DIC that emerge within seasonally suboxic waters. Similar data (Andersson et al., 2007). Nutrient chemistry was per- transitions in microbial communities and geochemical formed via flow-injection analysis on a Lachat Quickchem

8000 by the University of California, Santa Barbara Marine wavelengths as previously described. The image capturing Science Institute Analytical Laboratory (http://analab.msi was performed using a Toshiba (Irvine, CA, USA) CCD .ucsb.edu/). video camera with a Pro-series capture kit version 4.5 (Media Cybernetics, Bethesda, MD, USA) and processed with Image Pro software (version 4.5; Media Cybernetics) Bacterioplankton abundance as previously described (Carlson et al., 2009; Parsons Bacterioplankton samples were thawed, and 1Ð5 ml was fil- et al., 2011). tered onto 0.2 μm filters prestained with Irgalan Black (0.2 g in 2% acetic acid) under gentle vacuum (∼100 mmHg) and TRFLP fingerprinting of the 16S rRNA gene post-stained with 0,6-diamidino-2-phenyl dihydrochloride −1 (5 μgml , DAPI, SIGMA-Aldrich, St. Louis, MO, USA) DNA was extracted using the phenol chloroform method (Porter and Feig, 1980). Slides were then enumerated using (Giovannoni et al., 1990). Templates from the mixed commu- an AX70 epifluorescent microscope (Olympus, Tokyo, nities were amplified using PCR with primers 27F-FAM Japan) under UV excitation at 100× magnification. At least (5’FAM-AGRGTTYGATYMTGGCTCAG) and 519R (GWAT 400 cells per slide (12 fields) were counted. Virioplankton TACCGCGGCKGCTG) (SIGMA Biosynthesis, St. Louis, MO, abundance was enumerated according to the methods of USA) (Morris et al., 2005). Amplicons were gel purified (Noble and Fuhrman, 1998). Briefly, water samples were with Qiagen QiaQuick PCR product kit (Venlo, Limburg, The filtered on to 0.02 μm Anodisc aluminum oxide filters Netherlands) following the manufacturer’s instructions to (Whatman, Kent, UK), stained with 1X SYBR Green I remove both impurities and primers and digested with restric- (Molecular Probes, Inc., Eugene, OR, USA) and enumer- tion enzyme Hae III (NEB, Ipswich, MA, USA). Fragment ated via epifluorescence microscopy (Parsons et al., 2011) analysis of denatured products in formamide with a custom At least 200 VLP per slide (12 fields) were counted. 30 −600 bp size standard (Bioventures, Murfreesboro, TN, Synechococcus abundances were fixed with 1% 0.02 μm USA) was conducted at the UC Berkeley DNA Sequencing formalin and analysed on a Becton Dickinson, Franklin Facility on an Applied Biosystems 3730XL capillary Lakes, NJ, USA (formerly Cytopeia, Inc., Seattle, WA, USA) sequencer. Data analysis proceeded according to previously Influx™ flow cell sorter using a 488 nm blue excitation laser published methods (Nelson, 2009). (Casey et al., 2009); the coefficient of variation of triplicate samples was < 5% for cell concentrations > 200 cells ml−1. 16S Amplicon clone library development

Enumeration by FISH and CARD-FISH Clone libraries were constructed from samples collected at the surface and at the oxycline (24 m) on day 268 (DH4; The FISH probes used for this study included AC137, during anoxia) and at the same depths on day 282 (DH6; CF319a, CF319b, Cren537, Eury806, Negative 338, after mixing by convective overturn). Clone libraries were NON338, Roseo 536, SAR11_152, SAR11 441, SAR11 542, constructed from PCR amplicons (using the above primers) SAR11 732 and Vib Spl 127 (Supporting Information using pGEM-T-Easy (Promega Corporation, Madison, WI, Table S1). Chlorob 441 was designed for this study using USA) vector following the manufacturer’s instructions. Trans- ARB software (Ludwig et al., 2004) and the full-length SILVA formed colonies (100) selected from each cruise were cul- database (Pruesse et al., 2007) to which clone library tured and processed as previously described (Vergin et al., sequences were aligned (described later). Probes were vali- 2001). QIAprep Spin Miniprep Kit (Qiagen) was used to purify dated in silico for specificity using Probe Match on the Ribo- plasmid DNA to prepare for sequencing. Once restricted, somal Database Project (Cole et al., 2009), and TestProbe products were cleaned and sequenced at the Center for and Probebase on the SILVA website (Loy et al., 2007) and Genome Research and Biocomputing at Oregon State presented in Supporting Information Table S1. Bacterial University. abundance water samples (1Ð5 ml) were filtered onto 0.2 μm polycarbonate filters under gentle vacuum (∼100 mmHg) and − stored at 20¡C with desiccant. Quarter filters were washed Pyrosequencing of 16S amplicons in 95% ethanol and then probed according to (Morris et al., 2002). Archaeal enumeration was performed using CARD- PCR amplicons of the V1-V2 hypervariable regions of the FISH (Teira et al., 2004; Herndl et al., 2005). Permea- 16S small ribosomal RNA subunit gene were pyrosequenced bilization of the cell membrane was conducted using 0.1N from a set of 10 representative DNA samples. PCR reactions HCl with no prior embedding in agarose. The hybridization were carried out as described earlier using reaction condi- and wash conditions with all probe sequences are described tions and primers as previously described (Hamady et al., in Supporting Information Table S1. The resulting filters 2008; Nelson and Carlson, 2012; Nelson et al., 2013). from FISH and CARD-FISH were mounted with 20 μlof Sequences were dereplicated, aligned, clustered into OTUs, 1.67 μgml−1 0, 6-diamidino-2-phenylindole dihydrochloride and analysed for diversity and phylogenetic similarity among (DAPI, SIGMA-Aldrich) in citiflour solution (Ted Pella, Inc., samples within the software environment MOTHUR v32 Reading, CA, USA) and sealed with nail polish and stored (Schloss et al., 2009) according to Nelson and colleagues frozen in the dark (Parsons et al., 2011). (2013). Median read lengths were ∼350 bp, and most reads Image analysis coupled with epifluorescence microscopy spanned the full amplicon. A total of 16,564 full-length high- (Olympus AX70 microscope) was used to process FISH and quality reads were acquired from the 10 samples after CARD-FISH slides excited with Cy3 (550 nm) and UV denoising and removing chimeras via MOTHUR (detailed in

Nelson et al., 2013); samples were rarefied to 850 sequences Brochier-Armanet, C., Boussau, B., Gribaldo, S., and per sample for subsequent analyses with a total of 3357 Forterre, P. (2008) Mesophilic Crenarchaeota: proposal for unique sequences classified into 1369 OTUs at the 95% a third archaeal phylum, the Thaumarchaeota. Nat Rev sequence identity level among the 10 samples. Complete Microbiol 6: 245Ð252. linkage clustering on OTU-weighted Unifrac distances Brown, F. (1978) Mixing processes. The Bermuda marine (Lozupone and Knight 2005) within the software package environment. Barnes, J.A., and Bodungen, B.V. (eds). 2: PRIMER (Version 6, Primer-E, Plymouth, Lutton, Ivybridge, 10Ð30. UK, 2006) was used to analyse multivariate community struc- Brown, F. (1980) The nitrogen cycle and heat budget of ture differentiation among all pyrosequencing community a subtropical lagoon, Devil’s Hole, Harrington Sound, samples. The SIMPROF bootstrapping routine in PRIMER Bermuda: implications for nitrous oxide production and was used within hierarchical clustering algorithms to test consumption in marine environments. PhD Thesis. whether individual samples differed significantly in commu- Evanston, IL, USA: Northwestern University. nity composition (P < 0.05). All statistics were done in the Brune, A., Frenzel, P., and Cypionka, H. (2000) Life at the SAS programming language via the software package JMP oxic-anoxic interface: microbial activities and adaptations. (Version 9. SAS Institute, Inc., Cary, NC, USA; 1989Ð2010). FEMS Microbiol Rev 24: 691Ð710. Pyrosequencing runs have been accessioned in the National Canfield, D.E., Stewart, F.J., Thamdrup, B., De Brabandere, Center for Biotechnology Information Sequence Read L., Dalsgaard, T., Delong, E.F., et al. (2010) A cryptic sulfur Archive (http://trace.ncbi.nlm.nih.gov/Traces/sra) as run cycle in oxygen-minimum-zone waters off the Chilean accession SRR828415, with multiplexing barcodes for the coast. Science 330: 1375Ð1378. samples analysed here listed in Supporting Information Carlson, C.A., Morris, R., Parsons, R., Treusch, A.H., Table S2. Giovannoni, S.J., and Vergin, K. (2009) Seasonal dynamics of SAR11 populations in the euphotic and mesopelagic zones of the northwestern Sargasso Sea. ISME J 3: 283Ð Acknowledgements 295. Casey, J.R., Lomas, M.W., Michelou, V.K., Dyhrman, S.T., We thank John Casey and Rob Condon for flow cytometry Orchard, E.D., Ammerman, J.W., and Sylvan, J.B. (2009) analysis, James Biggs for assistance with microscopy, Elisa Phytoplankton taxon-specific orthophosphate (Pi) and ATP Wallner at the University of California, Santa Barbara for utilization in the western subtropical North Atlantic. Aquat assistance with nutrient analysis, and Marie Johnson at Microb Ecol 58: 31Ð44. Oregon State University for clone library technical assis- Cole, J.R., Wang, Q., Cardenas, E., Fish, J., Chai, B., Farris, tance. We would like to thank the Department of Conserva- R.J., et al. (2009) The Ribosomal Database Project: tion Services, Government of Bermuda for their assistance improved alignments and new tools for rRNA analysis. with researching Devil’s Hole studies and with supplying Nucleic Acids Res 37: D141ÐD145. background light data. This research was supported by the Crump, B.C., Peranteau, C., Beckingham, B., and Cornwell, NSF-REU grant (OCE-0552453) to the Bermuda Institute of J.C. (2007) Respiratory succession and community suc- Ocean Sciences, The Gordon and Betty Moore Foundation cession of bacterioplankton in seasonally anoxic estuarine (GBMF607.01), and by NSF Oceanic Microbial Observatory waters. Appl Environ Microbiol 73: 6802Ð6810. OCE-0801991 to CAC and SJG. This is contribution number Danovaro, R., and Serresi, M. (2000) Viral density and 9082 from the School of Ocean and Earth Science and Tech- virus-to-bacterium ratio in deep-sea sediments of the nology (SOEST) of the University of Hawai‘i at Manoa and Eastern Mediterranean. Appl Environ Microbiol 66: 1857Ð contribution number 2037 from the Bermuda Institute of 1861. Ocean Sciences. Danovaro, R., Manini, E., and Dell’Anno, A. (2002) Higher abundance of bacteria than viruses in deep Mediterranean sediments. Appl Environ Microbiol 66: 1857Ð1861. References DeLong, E.F., Taylor, L.T., Marsh, T.L., and Preston, C.M. Amann, R.I., Krumholz, L., and Stahl, D.A. (1990) (1999) Visualization and enumeration of marine planktonic Flourescent-oligonucleotide probing of whole cells for archaea and bacteria by using polyribonucleotide probes determinative, phylogenetic, and environmental studies in and fluorescent in situ hybridization. Appl Environ Microbiol microbiology. J Bacteriol 172: 762Ð770. 65: 5554Ð5563. Andersson, A.J., Bates, N.R., and Mackenzie, F.T. (2007) Diaz, R.J., and Rosenberg, R. (2008) Spreading dead zones Dissolution of carbonate sediments under rising pCO2 and and consequences for marine ecosystems. Science 321: ocean acidification: observations from Devil’s Hole, 926Ð929. Bermuda. Aquat Geochem 13: 237Ð264. Dolan, J.R. (2005) An introduction to the biogeography of Belmar, L., Molina, V., and Ulloa, O. (2011) Abundance and aquatic microbes. Aquat Microb Ecol 41: 39Ð48. phylogenetic identity of archaeoplankton in the permanent Durisch-Kaiser, E., Klauser, L., Wehrli, B., and Schubert, C. oxygen minimum zone of the eastern tropical South (2005) Evidence of intense archaeal and bacterial Pacific. FEMS Microbiol Ecol 78: 314Ð326. methanotrophic activity in the Black Sea water column. Beman, J.M., Popp, B.N., and Francis, C.A. (2008) Molecular Appl Environ Microbiol 71: 8099Ð8106. and biogeochemical evidence for ammonia oxidation by Fenchel, T., Blackburn, H., and King, G.M. (2012) Bacterial marine Crenarchaeota in the Gulf of California. ISME J 2: Biogeochemistry: The Ecophysiology of Mineral Cycling, 429Ð441. 3rd edn. New York, NY, USA: Academic Press, pp. 1Ð312.

Francis, C.A., Roberts, K.J., Beman, J.M., Santoro, A.E., and Labrenz, M., Jost, G., and Jürgens, K. (2007) Distribution of Oakley, B.B. (2005) Ubiquity and diversity of ammonia- abundant prokaryotic organisms in the water column of the oxidizing archaea in water columns and sediments of the central Baltic Sea with an oxic-anoxic interface. Aquat ocean. Proc Natl Acad Sci USA 102: 14683Ð14688. Microb Ecol 46: 177Ð190. Francis, C.A., Beman, J.M., and Kuypers, M.M.M. (2007) Lam, P., Jensen, M.M., Lavik, G., McGinnis, D.F., Muller, B., New processes and players in the nitrogen cycle: the Schubert, C.J., et al. (2007) Linking crenarchaeal and bac- microbial ecology of anaerobic and archaeal ammonia oxi- terial nitrification to anammox in the Black Sea. Proc Natl dation. ISME J 1: 19Ð27. Acad Sci USA 104: 7104Ð7109. Fuhrman, J.A., and Steele, J.A. (2008) Community struc- Lee, C.K., Herbold, C.W., Polson, S.W., Wommack, K.E., ture of marine bacterioplankton: patterns, networks, and Williamson, S.J., McDonald, I.R., and Cary, S.C. relationships to function. Aquat Microb Ecol 53: 69Ð81. (2012) Groundtruthing next-gen sequencing for microbial Giovannoni, S.J., and Vergin, K.L. (2012) Seasonality in ecology-biases and errors in community structure esti- ocean microbial communities. Science 335: 671Ð676. mates from PCR amplicon pyrosequencing. PLoS ONE 7: Giovannoni, S.J., DeLong, E.F., Schmidt, T.M., and Pace, e44224. N.R. (1990) Tangential flow filtration and preliminary Levipan, H.A., Quinones, R.A., Johansson, H.E., and Urrutia, phylogenetic analysis of marine picoplankton. Appl Environ H. (2007) Methylotrophic methanogens in the water Microbiol 56: 2572Ð2575. column of an upwelling zone with a strong oxygen gradient Hamady, M., Walker, J.J., Harris, J.K., Gold, N.J., and off Central Chile. Microbes Environ 22: 268Ð278. Knight, R. (2008) Error-correcting barcoded primers for Lin, X., Wakeham, S.G., Putnam, I.F., Astor, Y.M., Scranton, pyrosequencing hundreds of samples in multiplex. Nat M.I., Chistoserdov, A.Y., and Taylor, G.T. (2006) Compari- Methods 5: 235Ð237. son of vertical distributions of prokaryotic assemblages in Herndl, G.J., Reinthaler, T., Teira, E., van Aken, H., Veth, C., the anoxic Cariaco Basin and Black Sea by use of fluores- Pernthaler, A., and Pernthaler, J. (2005) Contribution of cence in situ hybridization. Appl Environ Microbiol 72: archaea to total prokaryotic production in the deep Atlantic 2679Ð2690. Ocean. Appl Environ Microbiol 71: 2303Ð2309. Liu, J., Weinbauer, M.G., Maier, C., Dai, M., and Gattuso, J.P. Howarth, R., Chan, F., Conley, D.J., Garnier, J., Doney, S.C., (2010) Effect of ocean acidification on microbial diversity Marino, R., and Billen, G. (2011) Coupled biogeochemical and on microbe-driven biogeochemistry and ecosystem cycles: eutrophication and hypoxia in temperate estuaries functioning. Aquat Microb Ecol 61: 291Ð305. and coastal marine ecosystems. Front Ecol Environ 9: Liu, W.T., Marsh, T.L., Cheng, H., and Forney, L.J. (1997) 18Ð26. Characterization of microbial diversity by determining ter- Iversen, N., and Jorgensen, B.B. (1985) Anaerobic methane minal restriction fragment length polymorphism of genes oxidation rates at the sulfate-methane transition in marine encoding 16s rRNA. Appl Environ Microbiol 63: 4516Ð sediments from Kattegat and Skagerrak (Denmark). Limnol 4522. Oceanogr 30: 944Ð955. Loy, A., Maixner, F., Wagner, M., and Horn, M. (2007) Johnston, D.T., Wolfe-Simon, F., Pearson, A., and Knoll, A.H. probeBase Ð an online resource for rRNA-targeted oligo- (2009) Anoxygenic photosynthesis modulated Proterozoic nucleotide probes: new features 2007. Nucleic Acids Res oxygen and sustained Earth’s middle age. PNAS 106: 35: 800Ð804. 16925Ð16929. Ludwig, W., Strunk, O., Westram, R., Richter, L., Meier, H., Karner, M., DeLong, E.F., and Karl, D. (2001) Archaeal domi- Yadhukumar, et al. (2004) ARB: a software environment for nance in mesopelagic zone of the Pacific Ocean. Nature sequence data. Nucleic Acids Res 32: 1363Ð1371. 409: 507Ð510. Lozupone, C., and Knight, R. (2005) Unifrac: a new Kim, B., and CHang, H. (1991) Removal of hydrogen sulfide phylogenetic method for comparing microbial communities. by Chlorobium thiosulfatophilum in immobilized-cell and Appl Environ Microbiol 71: 8228Ð8235. sulfur-settling free-cell recycle reactors. Biotechnol Prog 7: van der Maarel, M.J.E.C., Sprenger, W., Haanstra, R., and 495Ð500. Forney, L.J. (1999) Detection of methanogenic archaea in Kirchman, D.L. (2008) Microbial Ecology of the Oceans, 2nd seawater particles and the digestive tract of a marine fish edn. Hoboken, NJ, USA: John Wiley & Sons. species. FEMS Microbiol Lett 173: 189Ð194. Kirchman, D.L., Malmstrom, R.R., and Cottrell, M.T. (2005) Manuel, S., Coates, K., Kenworthy, W., and Fourqurean, J. Control of bacterial growth by temperature and organic (2013) Tropical species at the northern limit of their range: matter in the Western Arctic. Deep Sea Res Part 2 Top composition and distribution in Bermuda’s benthic habitats Stud Oceanogr 52: 3386Ð3395. in relation to depth and light availability. Mar Environ Res Knap, A.H., Michaels, A.F., Steinberg, D., Bahr, F., Bates, N., 89: 63Ð75. Bell, S., et al. (1997) BATS Methods Manual. Woods Hole, Martens-Habbena, W., Berube, P.M., Urakawa, H., de la AM, USA: U.S. JGOFS Planning Office. Torre, J.R., and Stahl, D.A. (2009) Ammonia oxidation Könneke, M., Bernhard, A.E., de la Torre, J.R., Walker, C.B., kinetics determine niche separation of nitrifying archaea Waterbury, J.B., and Stahl, D.A. (2005) Isolation of an and bacteria. Nature 461: 976Ð979. autotrophic ammonia-oxidizing marine archaeon. Nature Mincer, T.J., Church, M.J., Taylor, L.T., Preston, C., Karl, 437: 543Ð546. D.M., and DeLong, E.F. (2007) Quantitative distribution of Krekeler, D., Teske, A., and Cypionka, H. (1998) Strategies of presumptive archaeal and bacterial nitrifiers in Monterey sulphate-reducing bacteria to escape oxygen stress in a Bay and the North Pacific Subtropical Gyre. Environ cyanobacterial mat. FEMS Microbiol Ecol 25: 89Ð96. Microbiol 9: 1162Ð1175.

Morris, B., Barnes, J., Brown, F., and Markham, J. (1977) The Rabalais, N.N., Turner, R.E., Díaz, R.J., and Justic«, D. (2009) Bermuda Marine Environment. A report of the Bermuda Global change and eutrophication of coastal waters. ICES inshore waters investigations 1976Ð1977. Bermuda Bio- J Mar Sci 66: 1528Ð1537. logical Station Special Publication 15: 1Ð120. Rappe, M., and Giovannoni, S. (2003) The uncultured micro- Morris, R., Rappé, M., Connon, S., Vergin, K., Siebold, W., bial majority. Annu Rev Microbiol 57: 369Ð394. Carlson, C., and Giovannoni, S. (2002) SAR11 clade domi- Repeta, D.L., Simpson, D.L., Jorgensen, B.B., and Jannasch, nates ocean surface bacterioplankton communities. Nature H.W. (1989) Evidence for anoxygenic photosynthesis from 420: 806Ð810. the distribution of bacteriochlorophylls in the Black Sea. Morris, R.M., Vergin, K.L., Cho, J.C., Rappe, M.S., Carlson, Nature 342: 69Ð72. C.A., and Giovannoni, S.J. (2005) Temporal and spatial Revsbech, N.P., Larsen, L.H., Gundersen, J., Dalsgaard, T., response of bacterioplankton lineages to annual convec- Ulloa, O., and Thamdrup, B. (2009) Determination of ultra- tive overturn at the Bermuda Atlantic time-series study site. low oxygen concentrations in oxygen minimum zones by Limnol Oceanogr 50: 1687Ð1696. the STOX sensor. Limnol Oceanogr Methods 7: 371Ð381. Muyzer, G., and Stams, A.J.M. (2008) The ecology and Rodriguez-Mora, M.J., Scranton, M.I., Taylor, G.T., and biotechnology of sulphate-reducing bacteria. Nat Rev Chistoserdov, A.Y. (2013) Bacterial community composi- Microbiol 6: 441Ð454. tion in a large marine anoxic basin: a Cariaco Basin time- Nelson, C. (2009) Phenology of high-elevation pelagic bac- series survey. FEMS Microbiol Ecol 84: 625Ð639. teria: the roles of meteorologic variability, catchment inputs Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., and thermal stratification in structuring communities. ISME Hartmann, M., Hollister, E.B., et al. (2009) Introducing J 3: 13Ð30. mothur: open-source, platform-independent, community- Nelson, C.E., and Carlson, C.A. (2012) Tracking differential supported software for describing and comparing microbial incorporation of dissolved organic carbon types among communities. Appl Environ Microbiol 75: 7537Ð7541. diverse lineages of Sargasso Sea bacterioplankton. Sinninghe Damste, J.S., Rijpstra, W.I., Hopmans, E.C., Prahl, Environ Microbiol 14: 1500Ð1516. F.G., Wakeham, S.G., and Schouten, S. (2002) Distribution Nelson, C.E., Carlson, C.A., Ewart, C.S., and Halewood, E.R. of membrane lipids of planktonic Crenarchaeota in the (2013) Community differentiation and population enrich- Arabian Sea. Appl Environ Microbiol 68: 2997Ð3002. ment of Sargasso Sea bacterioplankton in the euphotic Smetacek, V., von Bodugen, B., Bolter, M., Knopers, B., zone of a mesoscale mode-water eddy. Environ Microbiol Pollehne, F., and Odebrecht, C. (1981) The pelagic system 16: 871Ð887. in Harrington Sound. BBSR Special Publication 19: 77Ð81. Nelson, D.C., Revsbech, N.P., and Jorhensen, B.B. (1986) Stal, L.J., van Gemerden, H., and Krumbein, W.E. (1985) Microoxic-anoxic niche of Begiatoa spp.: microelectrode Structure and development of a benthic marine microbial survey of marine and freshwater strains. Appl Environ mat. FEMS Microbiol Lett 31: 111Ð125. Microbiol 52: 161Ð168. Stewart, F.J., Ulloa, O., and DeLong, E.F. (2012) Microbial Noble, R.T., and Fuhrman, J.A. (1998) Use of SYBR Green I metatranscriptomics in a permanent marine oxygen for rapid epifluorescence counts of marine viruses and minimum zone. Environ Microbiol 14: 23Ð40. bacteria. Aquat Microb Ecol 14: 113Ð118. Stewart, K., Kassakian, S., Krynytzky, M., DiJulio, D., and Noll, M., Matthies, D., Frenzel, P., Derakshani, M., and Murray, J. (2007) Oxic, suboxic, and anoxic conditions in Liesack, W. (2004) Succession of bacterial community the Black Sea. In The Black Sea Flood Question: Changes structure and diversity in a paddy soil oxygen gradient. in Coastline, Climate, and Human Settlement. Yanko- Environ Microbiol 7: 382Ð395. Hombach, V., Gilbert, A., Panin, N., and Dolukhanov, P. Overmann, J., Cypionka, H., and Pfennig, N. (1992) An (eds). Dordrecht, The Netherlands: Springer, pp. 1Ð21. extremely low-light adapted phototrophic sulfur bacterium Stoica, E., and Herndl, G.J. (2007) Contribution of from the Black Sea. Limnol Oceanogr 37: 150Ð155. Crenarchaeota and Euryarchaeota to the prokaryotic Pace, N.R. (2006) Time for a change. Nature 441: 289Ð plankton in the coastal northwestern Black Sea. J Plankton 289. Res 29: 699Ð706. Parsons, R.J., Breitbart, M., Lomas, M.W., and Carlson, C.A. Stramma, L., Schmidtko, S., Levin, L.A., and Johnson, G.C. (2011) Ocean time-series reveals recurring seasonal pat- (2010) Ocean oxygen minima expansions and their biologi- terns of virioplankton dynamics in the northwestern Sar- cal impacts. Deep Sea Res Part 1 Oceanogr Res Pap 57: gasso Sea. ISME J 6: 273Ð284. 587Ð595. Porter, K.G., and Feig, Y.S. (1980) The use of DAPI for Strous, M., Kuenen, J.G., and Jetten, M.S.M. (1999) Key identifying and counting aquatic microflora. Limnol physiology of anaerobic ammonium oxidation. Appl Oceanogr 25: 943Ð948. Environ Microbiol 65: 3248Ð3250. Pradeep Ram, A.S., Sabart, M., Latour, D.S., and ime- Suzuki, M., and Giovannoni, S.J. (1996) Bias caused by Ngando, T. (2009) Low effect of viruses on bacteria in deep template annealing in the amplification of 16S rRNA genes anoxic water and sediment of a productive freshwater res- by the PCR. Appl Environ Microbiol 62: 625Ð630. ervoir. Aquat Microb Ecol 55: 255Ð265. Taipale, S., Jones, R.I., and Tiirola, M. (2009) Vertical diver- Pruesse, E., Quast, C., Knittel, K., Fuchs, B.M., Ludwig, sity of bacteria in an oxygen-stratified humic lake, evalu- W.G., Peplies, J., and Glockner, F.O. (2007) SILVA: a com- ated using DNA and phospholipid analyses. Aquat Microb prehensive online resource for quality checked and aligned Ecol 55: 1Ð16. ribosomal RNA sequence data compatible with ARB. Taylor, G.T., Iabichella, M., Ho, T.-Y., Scranton, M.I., Nucleic Acids Res 35: 7188Ð7196. Thunell, R.C., Muller-Karger, F., and Varela, R. (2001)

Chemoautotrophy in the redox transition zone of the Zaikova, E., Walsh, D.A., Stilwell, C.P., Mohn, W.W., Tortell, Cariaco Basin: a significant midwater source of organic P.D., and Hallam, S.J. (2010) Microbial community dynam- carbon production. Limnol Oceanogr 46: 148Ð163. ics in a seasonally anoxic fjord: Saanich Inlet, British Teira, E., Reinthaler, T., Pernthaler, A., Pernthaler, J., Columbia. Environ Microbiol 12: 172Ð191. and Herndl, G.J. (2004) Combining catalyzed reporter Zehr, J.P., and Ward, B.B. (2002) Nitrogen cycling in the deposition-fluorescence in situ hybridization and ocean: new perspectives on processes and paradigms. microautoradiography to detect substrate utilization by Appl Environ Microbiol 68: 1015Ð1024. bacteria and archaea in the deep ocean. Appl Environ Microbiol 70: 4411Ð4414. Thorstenson, D.C., and Mackenzie, F.T. (1974) TIme variability of pore water chemistry in recent corbonate sediments, Supporting information Devil’s Hole, Harrington Sound, Bermuda. Geochim Additional Supporting Information may be found in the online Cosmochim Acta 38: 1Ð19. version of this article at the publisher’s web-site: Treusch, A.H., Vergin, K.L., Finlay, L.A., Donatz, M.G., Burton, R.M., Carlson, C.A., and Giovannoni, S.J. (2009) Fig. S1. Bacterioplankton community structure as deter- Seasonality and vertical structure of an ocean gyre micro- mined by FISH and CARD-FISH analyses taken during bial community. ISME J 3: 1148Ð1163. this study in 2008. Presented as % of the total Ulloa, O., Canfield, D.E., DeLong, E.F., Letelier, R.M., and bacterioplankton community as determined by direct counts Stewart, F.J. (2012) Microbial oceanography of anoxic (x-axis) versus depth in metres (y-axis). Top panel represents oxygen minimum zones. PNAS 109: 15996Ð16003. profiles from DH1 (day 249) to DH4 (day 268) during stratifi- Vergin, K.L., Rappé, M.S., and Giovannoni, S.J. (2001) cation and bottom panel represents profiles from DH5 (day Streamlined method to analyze 16S rRNA gene clone 274) and DH6 (day 282) after overturn. % contribution of libraries. Biotechniques 30: 938Ð944. Synechococcus (green), SAR11 (blue), Rhodobacteraceae Vergin, K.L., Beszteri, B., Monier, A., Thrash, J.C., (pink), Bacteroidetes (yellow), Alteromonas (orange), Chloro- Temperton, B., Treusch, A.H., et al. (2013) High-resolution biacea (black), Thaumarchaeota (aqua), Euryarchaeota SAR11 ecotype dynamics at the Bermuda Atlantic (purple) and not probed (grey) are shown. time-series study site by phylogenetic placement of Table S1. Table showing probes used in this project includ- pyrosequences. ISME J 7: 1322Ð1332. ing their sequences, Target species including class level and Vetriani, C., Tran, H.V., and Kerkhof, L.J. (2003) Finger- what % of hits are the target species according to the SILVA printing microbial assemblages from the oxic/anoxic database. Additional information on the % formamide used in chemocline of the Black Sea. Appl Environ Microbiol 69: the hybridization solution, the hybridization temperature (¡C), 6481Ð6488. the NaCl concentration (M) in the wash solution and the wash Walsh, D.A., Zaikova, E., Howes, C.G., Song, Y.C., Wright, temperature (¡C) are also included. J.J., Tringe, S.G., et al. (2009) Metagenome of a versatile Table S2. Table showing the subset of samples analysed chemolithoautotroph from expanding oceanic dead zones. using pyrosequencing and the multiplexing barcode associ- Science 326: 578Ð582. ated with this analysis. The table includes sample ID, actual Wang, Q., Garrity, G.M., Tiedje, J.M., and Cole, J.R. (2007) sampling date, the day of year when the samples were taken Naive bayesian classifier for rapid assignment of rRNA and the multiplexing barcode. Pyrosequencing runs have sequences into the new bacterial taxonomy. Appl Environ been accessioned in the National Center for Biotechnology Microbiol 73: 5261Ð5267. Information Sequence Read Archive (http://trace.ncbi.nlm Weinbauer, M.G., Brettar, I., and Hofle, M.G. (2003) .nih.gov/Traces/sra) as run accession SRR828415. Lysogeny and virus-induced mortality of bacterioplankton Table S3. Clone library identifications as determined by in surface, deep, and anoxic marine waters. Limnol Sanger sequencing and phylogenetic reconstruction using Oceanogr 48: 1457Ð1465. ARB and the SILVA 100 database. The NCB taxonomy was Wieringa, E.B., Overmann, J., and Cypionka, H. (2000) used to name the clones. Samples were taken during this Detection of abundant sulphate-reducing bacteria in study in 2008. Presented are all clones sequenced and in amrine oxic sediment layers by a combined cultivation bold are the clones chosen for development of novel FISH and molecular approach. Environ Microbiol 2: 417Ð probes. Top half represents clones from the library con- 427. structed from samples from cruise DH4 (day 268) during Wright, J.J., Konwar, K.M., and Hallam, S.J. (2012) Microbial stratification and bottom half of table represents clones from ecology of expanding oxygen minimum zones. Nat Rev library DH6 (day 282) after overturn. Also listed are the Microbiol 10: 381Ð394. GenBank accession numbers.