Introduction
Formation of oxygen-depleted water masses in bottom environments is a widespread phenomenon in coastal areas around the world (Diaz and Rosenberg 2008). Oxygen-depleted water mass, in other word, bottom-water hypoxia develops where the consumption of oxygen by organisms and chemical processes in water and sediment exceeds the supply of oxygen. Although the contribution of oxygen consumption in water column to total oxygen consumption is often higher than sediment oxygen consumption (SOC) (Murrell and Lehrter 2011), SOC accounted for 20-81% of total O2 consumption below pycnocline (Dortch et al. 1994; Rivera et al. 2010; Murrell and Lehrter 2011).
Hence SOC can significantly contribute to the formation of bottom-water hypoxia in coastal area.
However, it has been noticed that SOC apparently decline at the hypoxic conditions (Lichtschlag et al. 2015). On the other hand, sulfide derived from activity of sulfate reducing bacteria would built up in the pore water of sediment, diffuse out into the overlying water under oxygen depleted condition and reach the upper-oxygenated water, resulting in the maintenance and expansion of bottom-water hypoxia (Roden and Tuttle
40 1992).
Murrel and Lehrter (2010) estimated SOC with collected sediment cores from the hypoxic Louisiana Continental Shelf, and found SOC increased when bottom-waters were re-oxygenated. This indicates that SOC rate potentially increases under bottom-water hypoxia due to accumulation of reduced compounds such as sulfide. The increased potential SOC may contribute to re-formation of bottom-water hypoxia. Indeed, prompt reformation of bottom-water hypoxia are often observed (e.g. Rabalais et al. 2007) after temporal mixing of water column and replenishment of oxygen due to strong winds.
However, little is known about the relationship between potential SOC and bottom-water hypoxia. In addition, SOC should be composed of biological O2 consumption (BOC), which is mainly driven by aerobic microbial community respiration, and chemical O2
consumption (COC), which is mainly caused by oxidation of sulfide ions derived from sulfate reducing bacterial activity. However, as far as I know, potential BOC and COC have not so far been evaluated separately in the previous studies on coastal hypoxia. In order to clarify the relationship between SOC and bottom-water hypoxia, it is instrumental to clarify dynamics of potential BOC and COC under bottom-water hypoxia.
Here I examined that temporal change in the potential Whole SOC (WSOC), COC and BOC separately under bottom-water hypoxia in an enclosed bay, Omura Bay, Japan.
41
I used INT reduction method that was described in chapter II.
42 Material and methods
Study site and sampling
Surface sediment samples and environmental parameters in the water column were collected in a center region of Omura Bay (St. 21, 32°55.390ʹN, 129°51.350ʹE, Fig. 1).
During summer months from 2011–2013 (Table 3), ten sediment cores were collected at each sampling from the center location at 20 ± 1 m depth with an acrylic pipe (31 cm long with 26 mm inner diameter) by scuba diving or TFO gravity corer. Reference cores were also collected from 11 m depth at a fringe site (32°51’.520” N, 129°52’.210” E, in Fig. 1) 8.3 km away from the center (July, November of 2011 and August of 2013). The vertical profiles of DO, temperature, salinity and chlorophyll a in the sampling sites were obtained using a Conductivity Temperature Depth (CTD) profiler (AAQ, JFE-Advantec Co, Kobe, Japan).
All core samples were kept at in situ temperature and carefully brought to the laboratory within 3 h after sampling, during which time samples were handled carefully to avoid direct exposure to sunlight and other physical disturbances. Upon return to the laboratory, three replicate sediment cores were extruded from cores down to either 5 or 7 mm depth from the top. The top 0–7 mm (in 2011) or 0–5 mm (in 2012 and 2013) layer of sediment was pooled and regarded as the uppermost sediment layer. The pooled
43
sediment layers from three cores were stored -20°C until DNA extraction and total organic carbon (TOC) measurement. A portion of the pooled sediment was fixed with glutaraldehyde for bacterial counting (See “Bacteria counting” for more detail). It was unfortunately not possible to keep the depth range of the uppermost surface and subsurface layers constant throughout the study, mainly because different devices were used to dissect the cores (Wada et al. 2012). Nevertheless, the uppermost sediment layers (0–7 mm) in 2011 mostly overlapped with those in the other 2 years (0–5 mm), and the number of cores in 2011 contributed only 30% of the total number (6 out of 20) of cores examined. Therefore, it does not seem unrealistic to assume that combining the 2011 data with those of 2012 and 2013 would not be comparable to combining data among sediment layers of identical depth (0–5 mm).
The INT reduction method with sediment core
Sediment INT reduction method was performed to reveal dynamics of SOC in Omura Bay. I followed the protocol described by Wada et al. (2012) with some modifications. Triplicate cores were used for measurement of whole INT reduction rate (WIR). After the overlying water had been replaced with 40 mL of aerated and filter-sterilized (0.22 μm) artificial seawater (TetraMarine Salt Pro, Tetra), 5 mL of 0.1% INT
44
solution (w/v) and 5 mL of MQ water were added into overlying water and then gently mixed by pipetting up and down (for WSOC). In contrast, triplicate or replicate cores were used for measurement of chemical INT reduction rate (CIR). After the overlying water had been removed, 40 mL of aerated and filter-sterilized (0.22 μm) artificial seawater (TetraMarine Salt Pro, Tetra) and 5 mL of formalin were added into core and waited for 10 min. Subsequently, 5 mL of 0.1% INT solution (w/v) was added into overlying water and then gently mixed by pipetting up and down. Upon replacing the water, special care was taken to avoid disturbance at the sediment surface. Core samples in acrylic tubes for each sampling were incubated in the laboratory at 26°C in dark conditions for 24 h. After the incubation, the overlying water was siphoned out to a sterile plastic tube and 20 mL was filtered through a cellulose acetate membrane filter (25 mm in diameter, pore size 0.22 μm, Advantec, A020A025A) and kept below -20°C until analysis. The sediment cores were then vertically extruded and sliced into horizontal sections with 0-5 mm depth. The sediment slices were kept below -20°C. INT-F extraction and following calculation were done as described above (See “Whole INT reduction versus oxygen consumption experiments” in chapter II).
45
Biological INT reduction rate (BIR) was calculated by subtracting CIR from WIR.
In order to avoid possible complication, calculated WIR, CIR and BIR as a proxy for SOC were represented as “WOCINT”, “COCINT” and “BOCINT”, respectively.
Sediment organic carbon and acid-volatile sulfides analysis
TOC of the sediment samples was treated according to the method of Wada et al.
(2016). Briefly, TOC of the sediment samples was determined on a Perkin Elmer 2400 CHNS/O Series II Analyzer (Perkin Elmer, Shelton, CT, USA) with acetanilide (C = 71.09%, N = 10.36%) as a standard. Prior to analysis, sediment samples were freeze-dried in a vacuum low-temperature oven (DRV320DA, Advantech, Tokyo, Japan) then passed a sieve with 500-μm-mesh and pulverize with a porcelain mortar and pestle. Samples were then put into silver capsules (Santis, Teufen, Switzerland) and treated with 1 M HCl for 24 h to remove carbonates.
Acid-volatile sulfides (AVS) in sediment samples were fixed with zinc acetate, and were measured spectrophotometrically by using the methylene blue method (Kondo et al.
1990).
46 Bacteria counting
Each sediment slice was mixed with filter-sterilized sea water containing glutaraldehyde (final conc. 2 %) and kept at 4°C until use. The fixed sediments were treated according to the method of Wada et al. (2016) with slight modifications. Briefly, to liberate the bacterial cells from sediment particles, Tween 20 (Bacto Tween 20, Difco) was added to the sample at a final concentration of 1 mg L−1. After mixing for 1 min, the sample was subjected to ultra-sonication (Q125, QSONICA) with five cycles of 5 s at 40% amplitude followed by centrifugation at 1600×g for 30 s. The supernatant was diluted 50-fold with filtered sterilized seawater. A 200 μL sample of the supernatant was then mixed with 4’,6-diamidino-2-phenylindole (DAPI), with a final concentration of 5 μg mL−1, and kept for 30 min at room temperature (ca. 25°C). The filter was then placed on a glass slide and embedded in non-fluorescent immersion oil. Bacterial cells were examined with an epifluorescent microscope (BX51, Olympus, Tokyo, Japan) with UV excitation. At least 400 single cells were counted on each slide at a magnification of 1000×.
Statistical analysis
To assess effect of environmental parameters on potential SOC, I used a backward stepwise regression to select the most predictive variables for SOC (WOCINT, COCINT
47
and BOCINT), which were performed using the statistical program R (R Core Team 2015).
Environmental data were log10-transformed before performing the regression analysis.
The Akaike information criterion (AIC), which balances the fit of a model against the number of parameters, was used to select the best fit model. In addition, a generalized linear model was compared to examine the relationship between environmental parameters, potential SOC. Outliers in the data set were detected using Grubbs’ outlier test at the 5% significance level before regression analysis.
48 Results
Environmental variables in overlying water and sediment
Figure 6 shows changes in dissolved oxygen (DO), temperature and salinity in the bottom water overlying the sediment (average value from 0 to 50 cm above sediment) at the center of Omura Bay during the field surveys in the years 2011–2013. In this study, the DO concentration range was divided into three categories according to the criteria used by Wright et al. (2012) with slight modifications: oxic (>90 μM O2), dysoxic (20–
90 μM O2) and suboxic or anoxic (<20 μM O2). Water temperature at the bottom of Omura Bay center ranged from 14 to 27.3°C and peaked in September each year. Salinity at the bottom ranged from 28.5 to 33, with a decreasing trend during summer months (June–
September). The TOC content ranged between 22.0-39.5 mg g−1 (dry weight sediment) in the center region and 21.3-33.1 mg g−1 in the south fringe. The TOC content in the sediment was relatively high from June to late August, started to decline from late August to September, and reached a minimum by the end of September each year (Table 3).
Bacterial cell abundance in the center region ranged from 4.3 × 109 to 1.7 × 1011 cells g−1 and that in the south fringe from 4.4 × 108 to 1.7 × 1010 cells g−1, respectively. Bacterial abundance also covaried between the two layers (r = 0.98, P < 0.01). AVS in the center region of 2013 ranged from 0.02 to 0.52 mg S g-1. AVS increased from July to August,
49 and abruptly decline in October.
SOC measured by INT reduction assay
Figure 7 and Table 3 show changes in WOCINT, COCINT and BOCINT at the center of Omura Bay during the field surveys in the years 2011–2013. WOCINT ranged between 0.34-3.82 mmol INTF m-2 day-1, while COCINT ranged between 0.32-2.37 mmol INTF m
-2 day-1. BOCINT ranged from -0.62 to1.67 mmol INTF m-2 day-1. A total of 20% of BOCINT
represent negative value (4 out of 20) due to higher COCINT, which were not used for further statistical test. All of SOC abruptly increased between June and August, and declined after September. WOCINT and BOCINT peaked at August in 2012 when complete anoxia was found.
Multiple linear regression analysis between SOC (WOCINT, COCINT and BOCINT) and environmental variables revealed that the combination of temperature and DO would explain more than 30% of variation of WOCINT and COCINT (Table 4). In contrast, DO alone would explain 19% of variation BOCINT. It was further demonstrated that temperature alone can explain partly, but significantly the variation of WOCINT and COCINT, whereas temperature poorly explained the variation of BOCINT (Fig. 8).
Although DO alone can also explain a part of variation of WOCINT, COCINT and BOCINT,
50
the relationship between BOCINT and DO was week. In contrast, bacterial abundance significantly explained 20–50% of variation of each SOC (Fig. 8). Sulfide contents measured in only 2013 were strongly correlated with COCINT (r = 0.82, P < 0.01).
51 Discussion
It was clearly demonstrated that both WSOCINT and COCINT increased and peaked during hypoxic period. Even though incubation conditions for INT reduction assays were not necessary representing in situ environment, WSOCINT and COCINT were clearly correlated with in situ ambient water temperatures and DO at the time of sampling. These results are consistent with previous reports on the environmental factors including water temperature, bottom-water oxygen concentration and availability of organic matter govern the rate of SOC that (Kristensen 2000; Glud 2008; Middelburg and Levin 2009).
Because COCINT depends on the amount of reduced compounds such as sulfide that have accumulated in sediment under low DO period before the assay incubation, COCINT is likely to reflect anaerobic bacterial metabolism, in particular, sulfate reduction in the sediment. As sulfate reduction rate can vary depending on the temperature under anaerobic condition (Kristensen 2000; Robador et al. 2009), it is reasonable to consider that increase in bottom water temperature during hypoxia enhanced COCINT in Omura Bay.
While correlation between BOCINT and DO was neither strong nor significant, BOCINT tended to increase towards low DO. In addition, BOCINT was significantly correlated with bacterial abundance (Fig. 8). From these observations, changes in BOCINT
52
can be explained as follows. Firstly, preservation of organic matter under low DO condition may have contributed to enhance catabolic activity of facultative anaerobic bacteria during incubation of sediment for INT reduction method. As a number of studies have reported, degradation rates of labile organic matter are comparable between oxic and anoxic conditions and those of refractory organic matter can be enhanced when anaerobic sediment is exposed to oxygen (Lee 1992; Hulthe et al. 1998; Arndt et al. 2013).
Besides, alternating oxic-anoxic conditions will greatly enhance organic matter mineralization in compared with permanently anoxic condition (Aller 1994). The oxygenated condition under which the INT reduction method was conducted may have favored organic matter degradation and hence led to increase in BOCINT under hypoxic period. Secondly, increased bacterial abundance during hypoxia may also have contributed to the enhancement of BOCINT during hypoxia. It has been known that bacterial biomass and growth rates in anoxic environments can be as great as, or even greater than, those under oxic conditions (Cole and Pace 1995; Bastviken and Tranvik 2001; Bastviken et al. 2001). Consistent with this notion, peak of bacterial abundance in surface sediment of Omura Bay was found in anoxic condition (Table 3).
In 4 out of 20 cases, BOCINT showed negative values due to unexpectedly higher COCINT than WSOCINT (Table 3). Other studies that employed formalin to determine
53
COC and BOC in sediment samples also reported negative BOC values (Sweerts et al.
1991; Zimmerman and Benner 1994). Addition of formalin could have enhanced penetration of oxygen further into the sediment (Sweerts et al. 1991), thereby increased COCINT and hence resulted in negative BOCINT. However, the ratios of COCINT to WSOCINT were on average 83%, which was comparable to those of other coastal areas such as the southwestern part of Suo-Nada, Japan (78%, Kamizono et al. 1996) and Hiroshima Bay (30-50% from summer to autumn, 70-100% in winter, Seiki et al., 1994).
In addition, if the R/ INT-F ratio for WSOC (29.1) estimated in chapter II was used to convert the amount of INT-F to that of O2, WSOC in the center of Omura Bay ranged from 9.8 to 111.1 mmol O2 m-2 day-1 with a mean of 35.4 mmol O2 m-2 day-1. This is comparable to WSOC estimates of re-oxygenated sediment in other coastal area (26-48 mmol O2 m-2 day-1; Murrell and Lehrter, 2010, 31.7 mmol O2 m-2 day-1; Miller-Way et al.
1994). Additionally, the present WSOCINT was also comparable to WSOC obtained with direct O2 measurement in other coastal areas such as the East China Sea (3.6-17.6 mmol O2 m-2 day-1, Song et al., 2016) , Northern Gulf of Mexico ( 0.82-56.4 mmol O2 m-2 day
-1, Rowe et al. 2002) and Seto Inland Sea (3.6-17.6 mmol O2 m-2 day-1, Nakamura 2003).
Therefore, apart from the unexpected negative BOCINT, the overall estimates of SOC in
54
the present study are reasonably acceptable for the assessment of microbial respiratory metabolism in Omura bay sediment.
55
Fig. 6 Dissolved oxygen (DO) concentration (μM) (filled circle), Temperature (°C) (open triangle) and salinity (inverted filled triangle) in overlying water at the center of Omura bay in (a) 2011, (b) 2012 and (c) 2013. Oxygen conditions were divided into oxic (>90 μM O2), dysoxic (20–90 μM O2) and suboxic (<20 μM O2) conditions.
56
Fig. 7 SOC estimated by INT reduction assay. WOCINT = whole oxygen consumption, COCINT = chemical oxygen consumption, BOCINT = biological oxygen consumption.
Error bars for WOCINT represent standard errors from 3 replicate cores. Error bars for COCINT from July in 2011 through July in 2012 represent ranges of two replicate cores while those for COCINT after August in 2012 represent standard errors from 3 replicate cores. Asterisks indicate sampling days when BOCINT have negative value.
57
Fig. 8 Scatter plots depicting the relationship between SOC and environmental parameters. Scatter plots of SOC versus DO (a-c). Scatter plots of SOC versus
temperature (d-f). Scatter plots of SOC versus salinity (g-i). Scatter plots of SOC versus TOC (j-l). Scatter plots of SOC versus bacterial abundance (m-o). Filled circle and open circles represent the value of the center and south fringe, respectively. Solid lines
represent the best fit of a linear model to the data of the central Omura Bay.
58
Table 3 Environmental parameters and sediment oxygen consumption rate in Omura Bay.
Sampling site
Year Month/
Day
Overlying water Sediment
DO conditions
DO (μM)
Temp (°C)
Salinity (PSU)
WOCINT
(mmol INTFm-2
day-1)
COCINTl (mmol INTF
m-2 day-1)
BOCINT
(mmol INTF m-2
day-1)
TOC (mg g-1)
Sulfide (mg S g-1)
Cells abundance (Cells g-1)
Center 2011 7/5 Dysoxic 70.4 20.1 32.6 0.34 0.41 -0.07 24.1 NA 6.3 x 109
7/15 Dysoxic 65.2 21.0 31.2 0.51 0.37 0.14 NA NA 6.0 x 109
8/12 Suboxic 4.8 24.5 30.3 1.22 0.68 0.54 NA NA 7.2 x 109
9/14 Dysoxic 44.3 25.6 28.9 0.53 0.51 0.02 25.7 NA 7.2 x 109
11/9 Oxic 128.7 22.0 31.3 0.64 0.40 0.24 24.8 NA 7.6 x 109
12/21 Oxic 243.3 14.0 31.1 0.54 0.40 0.14 26.7 NA 4.3 x 109
2012 5/19 Oxic 157.6 15.7 32.9 0.69 0.53 0.16 31.3 NA 6.1 x 1010
6/20 Oxic 124.0 19.0 33.1 0.58 0.70 -0.12 32.2 NA 3.4 x 1010
7/18 Dysoxic 82.7 22.1 31.9 0.77 0.76 0.01 33.1 NA 3.0 x 1010
8/24 Suboxic* 0.0 24.1 31.3 3.82 2.15 1.67 30.6 NA 1.7 x 1011
9/20 Oxic 195.9 27.3 31.2 2.78 2.37 0.41 22.0 NA 1.7 x 1011
2013 6/6 Oxic 138.2 17.6 32.9 0.97 0.32 0.65 37.8 0.12 2.5 x 1010
6/28 Dysoxic 70.7 20.8 31.3 0.49 0.24 0.25 36.5 0.06 3.0 x 1010
7/12 Dysoxic 53.2 22.7 32.8 1.57 1.25 0.32 39.5 0.02 3.2 x 1010
7/26 Suboxic 1.6 22.3 30.4 1.56 1.41 0.15 37.4 0.32 3.1 x 1010
8/2 Suboxic 1.8 22.8 32.7 1.13 1.75 -0.62 38.3 0.37 2.5 x 1010
8/12 Suboxic 8.0 24.6 32.6 1.71 1.50 0.21 35.3 0.52 2.8 x 1010
8/28 Suboxic 3.1 26.4 32.7 1.68 1.44 0.24 31.8 0.44 3.2 x 1010
9/12 Suboxic 2.9 26.9 28.5 2.02 2.15 -0.14 36.4 0.43 2.7 x 1010
10/31 Oxic 166.7 22.1 30.6 0.77 0.45 0.32 37.0 0.03 3.2 x 1010
Average 78.1 22.1 31.5 1.22 0.99 0.23 31.9 0.3 4.0 x 1010
South fringe 2011 7/5 Oxic 159.4 22.0 31.6 0.27 0.16 0.11 33.1 NA 4.4 x 108
2011 11/9 Oxic 193.8 21.9 31.9 0.20 0.37 -0.17 21.3 NA 7.0 x 109
2013 8/28 Oxic 140.3 28.3 32.6 1.22 1.12 0.10 29.6 NA 1.7 x 1010
Average 164.5 24.1 32.0 0.6 0.5 0.0 28.0 NA 8.2 x 109
59
Table 4 Multiple linear regression using abiotic parameters to explain SOC.
WOCINT (mmol INTF m-2 day-1) COCINT (mmol INTF m-2 day-1) BOCINT (mmol INTF m-2 day-1) Explanatory
variable
Regression coefficient
P-value
Model statistics
Explanatory variable
Regression coefficient
P-value
Model statistics
Explanatory variable
Regression coefficient
P-value
Model statistics
DO -0.48 0.08
R2 = 0.41
DO -0.37 0.06
R2 = 0.52
DO -0.25 0.06
R2 = 0.24
Temp 3.54 0.19
R2adj. = 0.34
Temp 3.7 0.07
R2adj. = 0.44
Temp - -
R2adj.
= 0.19
Salinity - -
P = 0.011
Salinity - -
P = 0.003
Salinity - -
P = 0.055
TOC - - TOC - - TOC - -
Intercept -2.80 0.46 Intercept -3.38 0.23 Intercept 0.74 0.003
60
IV. Effects of bottom-water hypoxia on sediment bacterial