2.1 Study site and sample collection

Samples were collected at Al Kharar, a lagoon on the Saudi coast of the central Red Sea in February 2018. Two H. stipulacea meadows at a depth of 2–3 m, S1 (22^∘56^′46.5^′′ N, 38^∘52^′40.6^′′ E) and S2 (22^∘54^′44.5^′′ N, 38^∘53^′50.9^′′ E), were chosen to represent a range of organic matter content in the sediment and were selected to evaluate greenhouse gas fluxes. Moreover, the H. stipulacea meadow in the middle of the lagoon (S2) with higher biomass density (Table 1) was chosen as the study site to experimentally assess the role of temperature and darkness in greenhouse gas fluxes. The seagrass and sediment community was sampled using translucent cylindrical polyvinyl chloride (PVC) cores (26 cm length and 9.5 cm in diameter). The sharpened edge of the core was carefully pushed approximately 10 cm into the sediment with a rubber hammer so that the structure of leaves, roots and sediment stayed intact. A rubber stopper was then placed on top, before the core was carefully pulled out of the sediment without disturbing the structure and another rubber stopper was placed on the bottom of the core. The sediment cores were immediately transported to the laboratory.

Table 1Summary of vegetated sediment, bare sediment and seagrass leaves characteristics in the study sites (S1 and S2). NA: not available.

Download Print Version | Download XLSX

2.2 Sediment and plant characterization

Once the cores were opened, the first 10 cm of the sediment and the plant biomass from the same cores were collected and dried at 60 ^∘C to a constant dry weight (DW). To characterize the two different H. stipulacea meadows, sediment and plant biomass samples were then ground to conduct sediment composition and nutrient analyses. A 50 mL tube was filled with sediment from the first 10 cm, and the contents were dried at 60 ^∘C to a constant dry weight and weighed to determine the sediment bulk density (g sediment cm⁻³). Organic matter content was analyzed by loss on ignition (LOI; Dean, 1974). Approximately 5 g of dried sediment was placed in a muffle furnace and burned at 450 ^∘C for 5 h. The organic matter content was calculated as

The carbonate content was estimated using a Pressure Gauge Calcimeter. Approximately 1 g of sample was placed in the calcimeter, and the recipient was filled with 10 % hydrochloric acid (HCl). The mass of CaCO₃ in the sample (g) was then calculated as follows:

where p is the pressure recorded (ppm), b is the slope and a the intercept derived from the calibration curve, and w is the exact weight of each sediment sample (g). The percentage of CaCO₃ in the sample (% DW) was then calculated using Eq. (3):

Dried sediment and plant samples were digested using US Environmental Protection Agency (EPA) method 3052 and analyzed with nitric acid (HNO₃) and HCl using US EPA method 200.7 following manufacturer instructions. The phosphorus content (% DW) was analyzed using inductively coupled plasma optical emission spectroscopy (ICP-OES) on an Agilent 5110 ICP-OES (Agilent Technologies, Santa Clara, CA, USA). The C and nitrogen (N) concentration of both plants and sediments was analyzed after acidification with HCl (Hedges and Stern, 1984), using a FLASH 2000 Organic Elemental Analyzer (CHNS/O-2, Thermo Fisher Scientific, Waltham, MA, USA). The isotopic signature of ¹³C in sediment organic matter was analyzed, using cavity ring-down spectroscopy (CRDS G2201-I, Picarro Inc., Santa Clara, CA, USA), from the ¹³C of CO₂ released by a combustion module (Costech Analytical Technologies Inc., CA, USA) delivering the CO₂ resulting from combusting the sediment organic matter to the CRDS instrument.

2.3 Assessment of carbon dioxide and methane air–seawater fluxes

In February 2018, triplicate cores from vegetated and adjacent bare (about 5 m from the edge of the seagrass patch) sediments were collected from sites S1 and S2 and transferred to incubation chambers (Percival Scientific Inc., Perry, IA, USA) set at 25 ^∘C with a 12 h light (up to 70 µmol photons m⁻² s⁻¹) : 12 h dark (12 h L : 12 h D) cycle to measure the greenhouse gas (CO₂ and CH₄) fluxes supported by these communities. Before measuring fluxes, the water overlying the sediment inside the cores was carefully siphoned until only 5 mm of water remained over the sediment surface; fresh seawater was carefully siphoned in the core to avoid disturbing the redoxcline, leaving a headspace of approximately 5–6 cm; and the cores were closed again with stoppers containing gas-tight valves. The cores were left for 1 h to allow for equilibration between the seawater and the headspace air. We then sampled 10 mL of air from each core using a syringe and injected the air sample in a cavity ring-down spectrometer (Picarro Inc., Santa Clara, CA, USA) through a Small Sample Isotopic Module extension (SSIM A0314, Picarro Inc., Santa Clara, CA, USA), which provided both the partial pressure and the isotopic carbon composition of the CO₂ and CH₄ in the air sample. One sample from each core was taken at the start (T₀), after 12 h of light (T₁) and after 12 h of dark (T₂). The daily CO₂ and CH₄ fluxes were calculated from the difference between T₀ and T₂ taking into account the core surface area (µmol m⁻² d⁻¹). Before and after each sampling, two standards were measured (A: 750 ppm CO₂, 9.7 ppm CH₄; B: 250.5 ppm CO₂, 3.25 ppm CH₄).

2.4 Effect of warming on carbon dioxide and methane air–seawater fluxes

In March 2018, we collected 18 vegetated and 18 bare sediment cores from site S2 to evaluate the response of greenhouse gas fluxes to warming. The sampling was performed as described above. The cores were transferred to the Coastal and Marine Resources Core Lab (CMOR, KAUST, Saudi Arabia). Nine vegetated and nine bare sediment cores each were placed in two aquaria with flow-through seawater set at in situ temperature (25 ^∘C) and with a 12 h L (up to 200 µmol photons m⁻² s⁻¹) : 12 h D cycle. One aquarium was maintained at 25 ^∘C over the entire duration of the experiment to serve as a control for temperature-independent variability in fluxes. The temperature in the second aquarium was increased at a rate of 1 ^∘C d⁻¹ to allow for acclimatization of the vegetated and bare cores. CO₂ and CH₄ fluxes were measured at every 2 ^∘C from 25 to 37 ^∘C, with parallel measurements conducted on the cores maintained at 25 ^∘C. After a 1 d acclimation period at each new temperature, the cores were closed with the stoppers and transferred to incubation chambers (Percival Scientific Inc., Perry, IA, USA) set at the target temperature for CO₂ and CH₄ flux measurements as described above. After the 24 h measurements, the cores were returned to the aquaria. An additional core kept at each of the constant temperature and warming sets was sampled every 4 d (i.e., at 4 ^∘C temperature intervals in the warming treatment) to analyze sediment composition. The cores used for flux estimates were opened after the final measurement (20 d since collection) to estimate the plant biomass and analyze the sediment composition at the end of the experiment.

2.5 Effect of darkness on carbon dioxide and methane air–seawater fluxes

In May 2018, six vegetated and six bare sediment cores were collected from site S2 and kept at a constant 25 ^∘C with a 24 h dark cycle. Only during the measurements in the incubators were the cores exposed to a 12 h L : 12 h D cycle, allowing fluxes to be compared with those measured in cores permanently maintained with the 12 h L : 12 h D photoperiod. CO₂ and CH₄ fluxes were measured after the first day of acclimation and then kept in the aquaria until signs of seagrass mortality started to become apparent, which occurred after 1 week in the dark. CO₂ and CH₄ fluxes were measured at alternate days as detailed above. At the end of the experiment (21 d since collection), the cores were opened and sampled to assess plant biomass and sediment composition.

2.6 Measurements of carbon dioxide and methane air–seawater fluxes

The concentration of CO₂ in the seawater after equilibrium was calculated based on the concentration of CO₂ in the headspace (ppm) measured by CRDS according to Sea et al. (2018) and Wilson et al. (2012):

where β is the Bunsen solubility coefficient of CO₂ (mol mL⁻¹ atm⁻¹), m_a is the CO₂ concentration measured in the headspace (ppm) and p_dry is the atmospheric pressure of dry air (atm). The Bunsen solubility coefficient of CO₂ was calculated using Eq. (5):

where H^cp is the Henry constant (mol mL⁻¹ atm⁻¹) calculated using the marelac R package (Soetaert et al., 2016). R is the ideal gas constant (0.082057459 atm L mol mL⁻¹ K⁻¹), and T is the standard temperature (273.15 K).

The atmospheric pressure of dry air (p_dry) was calculated as follows:

where p_wet is the atmospheric pressure of wet air. Boyle's law was applied as gas was collected several times from the same core.

The concentration of dissolved CO₂ in seawater before equilibrium was then calculated using Eq. (7):

where V_w is the volume of seawater (mL) and V_a is the volume of the headspace (mL). The units were then converted to nanomolar (nM):

The daily CO₂ fluxes were calculated from the difference between T₀ and T₂ taking into account the core surface area (µmol m⁻² d⁻¹).

Daily CH₄ fluxes were estimated using the same calculations as for the CO₂ fluxes with the exception of the Bunsen solubility coefficient. The Bunsen solubility coefficient was calculated as a function of the seawater temperature and salinity following Wiesenburg and Guinasso (1979). The total CO₂ greenhouse-equivalent fluxes were calculated assuming CH₄ to have a greenhouse potential 28-fold greater than that of CO₂ per mol of C (Myhre et al., 2013).

2.7 Isotopic composition of carbon dioxide (δ¹³C-CO₂) and methane (δ¹³C-CH₄)

The isotopic signature of CO₂ and CH₄ produced during the incubations was estimated using Keeling plots following Garcias-Bonet and Duarte (2017). δ¹³C of CO₂ and CH₄ produced in our incubations was extracted from the intercept of the linear regression between the inverse of the gas concentration (ppm⁻¹) and the isotopic signature measured from the discrete samples in the CDRS instrument.

2.8 Data analysis

The data were analyzed for normality using the Shapiro–Wilk test. The Mann–Whitney test and Student's t test were used to test for differences in seagrass and sediment composition between sites and between vegetated and bare sediments, and analysis of variation (ANOVA) and the Kruskal–Wallis test were used to test for differences between vegetated and bare sediments and both sites. To assess differences in greenhouse gas fluxes between different H. stipulacea communities, differences in CO₂ and CH₄ fluxes were analyzed between sites and between vegetated and bare sediment by using the Kruskal–Wallis test. Trends in the flux between the communities experiencing warming and the ones maintained at 25 ^∘C, as well as in the isotopic signature of δ¹³C-CO₂ and δ¹³C-CH₄ over time, were analyzed by linear regression. When assessing the effect of darkness on greenhouse gas fluxes, the trend of CO₂ and CH₄ fluxes and their isotopic signatures were analyzed by linear regression. The statistical analyses were conducted in PRISM 5 (GraphPad Software, La Jolla, CA, USA) and JMP Pro 13.1.0 (SAS Institute Inc., Cary, NC, USA). The data are openly available from Burkholz et al. (2019b).

3.1 Seagrass and sediment composition

Carbon, nitrogen and phosphorus (P) concentrations in seagrass leaves were low, but they were 4- to 40-fold higher than vegetated and bare sediment concentrations (Table 1). Seagrass sampled in site S1 had the highest C, N and P concentrations in the leaves, while sediment C and P concentrations differed significantly between sites (ANOVA, p<0.05, and Kruskal–Wallis, p<0.001, respectively), with the highest C and the lowest P concentrations found in the sediment of S2 (Table 1). There were no consistent differences in C, N and P concentration in bare and vegetated sediments (Table 1).

The sediments had high, but variable, carbonate concentrations, which differed between sites (Kruskal–Wallis, p<0.0001; Table 1). The organic matter content was slightly increased in S2 compared to S1, in both vegetated (t test, p<0.0001) and bare (t test, p<0.001) sediments (Table 1). Sediment bulk density was similar in both S1 and S2 sites, but vegetated sediments in S1 showed significantly lower bulk density compared to bare sediments (t test, p<0.05; Table 1). Seagrass biomass density was significantly higher in S2 compared to S1 (t test, p<0.05). The isotopic signature of sediment organic carbon ranged across sites from $-\mathrm{15.77}\pm \mathrm{0.07}$ ‰, in vegetated sediments, to $-\mathrm{16.36}\pm \mathrm{0.28}$ ‰, in bare sediments (Table 1). The carbon isotopic signature of seagrass leaves from the same location has been recently reported as $-\mathrm{7.96}\pm \mathrm{0.27}$ ‰ by Duarte et al. (2018).

3.2 Carbon dioxide and methane air–seawater fluxes

The daily CO₂ flux was up to 6-fold higher in vegetated compared to bare sediments and tended to be generally higher in S2 compared to S1, where bare sediments showed net CO₂ uptake, although differences were not significant (Kruskal–Wallis, p>0.05; Fig. 1a). At both sites, S1 and S2, the daily net CH₄ flux was 10- to 100-fold higher in vegetated compared to adjacent bare sediments with generally higher fluxes at S2 (Kruskal–Wallis, p<0.01; Fig. 1b). The total CO₂ equivalent (CO₂-eq) fluxes varied between sites and were higher in the vegetated compared to the bare sediments (Kruskal–Wallis, p<0.01; Fig. 1c).

Figure 1Mean + SE (a) CO₂, (b) CH₄ and (c) CO₂-eq fluxes in vegetated (green) and adjacent bare (yellow) sediments at two sites (S1 and S2) in the central Red Sea.

Download

3.3 Effect of warming on carbon dioxide and methane air–seawater fluxes

The CO₂ fluxes in vegetated sediments increased greatly with warming (R²=0.38, p<0.001) but decreased over time when the community was maintained at 25 ^∘C (R²=0.30, p<0.01; Fig. 2a, Table S1 in the Supplement), shifting from sediments showing net CO₂ emission to net CO₂ uptake. Similar responses were observed in the bare sediments, where CO₂ fluxes increased with warming (R²=0.54, p<0.0001), while the community tended to shift over time from supporting net CO₂ emission to net CO₂ uptake when maintained at 25 ^∘C (Fig. 2b). The net CO₂ flux – i.e., the difference between the CO₂ fluxes in warming sediments and those at sediments maintained at 25 ^∘C – increased significantly with warming in both the vegetated and the bare sediment (R²=0.74 and p<0.05, and R²=0.91 and p<0.001, respectively; Fig. 2c).

Figure 2Mean ± SE CO₂ (left) and CH₄ (right) fluxes in (a, d) vegetated and (b, e) bare sediments. Symbols indicate each replicate of the community experiencing warming from 25 to 37 ^∘C (red) and the community maintained at 25 ^∘C (blue) over the experimental period (number of days since the onset of the experiment). An outlier at 33 ^∘C in vegetated sediments in shown in green. (c, f) Mean ± SE (c) CO₂ and (f) CH₄ net fluxes in vegetated (green) and bare (yellow) sediments over the experimental period (number of days since the onset of the experiment). The second x axis indicates the experimental temperature for the community exposed to warming from 25 to 37 ^∘C. Lines represent a fitted linear equation.

Download

CH₄ fluxes declined over time when the sediments were maintained at 25 ^∘C, both in vegetated (R²=0.43, p<0.001; Fig. 2d) and, less strongly, bare sediments (R²=0.24, p<0.01; Fig. 2e, Table S2). In contrast, CH₄ fluxes tended to increase with temperature in vegetated (Fig. 2d) and bare (Fig. 2e) sediments gradually warmed from 25 to 37 ^∘C, although it was not significant (p>0.05 and p>0.05, respectively; Table S2). The net CH₄ fluxes – i.e., the difference between the CH₄ fluxes in sediments exposed to warming and sediments maintained at 25 ^∘C – increased significantly over time (i.e., with warming) in vegetated (R²=0.69, p<0.05) but not in bare sediments (p>0.05; Fig. 2f). An outlier in the vegetated sediment at 33 ^∘C supporting extreme emissions (CO₂ flux of 55 170 µmol CO₂ m⁻² d⁻¹ and CH₄ flux of 699.8 CH₄ µmol m⁻² d⁻¹) was observed on day 14 in one of the replicates of the warming treatment where seagrass had died (Fig. 2a and d); it was excluded from the regression analyses reported above.

Despite CO₂ and CH₄ fluxes showing the same response to warming in both types of sediment, vegetated sediments held higher fluxes than bare sediments. The relationship between net CO₂ and CH₄ fluxes in bare vs. vegetated sediments showed that both bare and vegetated communities tended to act as net CO₂ sinks at 25 ^∘C, but tended to act as CO₂ sources at warmer temperatures (Fig. 3a), whereas net CH₄ fluxes were 3- to 8-fold higher in vegetated compared to bare sediments (Fig. 3b). The CH₄ / CO₂ ratio declined in the vegetated sediments exposed to warming from 7 to 0.8 %. For CO₂ and CH₄ fluxes in vegetated sediments, the Q₁₀ value for the temperature range 25–37 ^∘C was 9 and 1.5, respectively, while the Q₁₀ value for bare sediments was 13.8 and 4.2, respectively.

Figure 3Relationship between vegetated and bare sediments for (a) CO₂ and (b) CH₄ fluxes. Symbols indicate different temperatures ranging from 25 to 37 ^∘C; the dashed line indicates 1 : 1; and dotted lines show 2 : 1, 4 : 1 and 8 : 1.

Download

3.4 Effect of darkness on carbon dioxide and methane air–seawater fluxes

The vegetated sediment shifted over time from showing net CO₂ uptake to net CO₂ emission when maintained in the dark (R²=0.70, p<0.05), while the increase in the bare sediment was not significant (p>0.05; Fig. 4a, Table S3). In contrast, when vegetated sediment was maintained at 25 ^∘C with a 12 h L : 12 h D photoperiod, the community shifted from net CO₂ emission to net CO₂ uptake (Mann–Whitney, p<0.05; Fig. 5a). In bare sediments, CO₂ fluxes showed the same trend in cores maintained at 25 ^∘C with a 12 h L : 12 h D photoperiod as under dark conditions (Fig. 5b).

Figure 4Mean ± SE (a) CO₂ and (b) CH₄ fluxes in vegetated (green) and bare (yellow) sediments of communities exposed to prolonged darkness over the experimental period (number of days since the onset of the experiment).

Download

Figure 5Comparison of mean ± SE CO₂ (left) and CH₄ (right) fluxes in (a, c) vegetated and (b, d) bare sediments maintained at 25 ^∘C with a 12 h L : 12 h D photoperiod (white) and communities kept at 25 ^∘C with a 24 h D period (black) over the experimental period (number of days since the onset of the experiment). Dots indicate mean values, and error bars indicate standard error of the mean.

Download

When vegetated sediments were kept in the dark, net CH₄ fluxes decreased 5-fold over time (R²=0.99, p<0.0001; Fig. 4b, Table S4). However, the CH₄ fluxes did not differ significantly between vegetated cores maintained at 25 ^∘C in the 12 h L : 12 h D photoperiod or in the dark (Mann–Whitney, p>0.05), showing the same trend of decreasing CH₄ fluxes (Fig. 5c). In the bare sediment, CH₄ fluxes in sediments kept in the dark were higher than those at 25 ^∘C with a 12 h L : 12 h D photoperiod, with significant differences only observed on days 14 and 20 (Mann–Whitney, p<0.05 and p<0.05, respectively; Fig. 5d).

3.5 Isotopic composition of carbon dioxide (δ¹³C-CO₂) and methane (δ¹³C-CH₄)

The isotopic signature of the δ¹³C-CO₂ became heavier with warming in the bare sediment, increasing from $-\mathrm{22.36}\pm -\mathrm{4.97}$ ‰ δ¹³C at 25 ^∘C to $-\mathrm{9.01}\pm \mathrm{0.98}$ ‰ δ¹³C at 37 ^∘C (R²=0.91, p<0.001), while the other treatments showed similar values over time, ranging from a minimum average of $-\mathrm{17.89}\pm \mathrm{1.81}$ ‰ to a maximum average of $-\mathrm{11.55}\pm \mathrm{5.32}$ ‰ δ¹³C (Fig. 6a–d).

Figure 6Mean ± SE isotopic signature of CO₂ (δ¹³C-CO₂) and CH₄ (δ¹³C-CH₄) in the communities experiencing warming from 25 to 37 ^∘C (left) and the communities maintained at 25 ^∘C (right). (a–d) δ¹³C-CO₂ is shown for the (a, b) vegetated and (c, d) bare sediments over the experimental period (number of days since the onset of the experiment). (e–h) δ¹³C-CH₄ is shown for the (e, f) vegetated and (g, h) bare sediment over the experimental period. The second x axis indicates the temperature increase for the community experiencing warming.

Download

The isotopic signature of δ¹³C-CH₄ decreased over time in both vegetated and bare sediments, whether they were maintained at constant temperature or experienced warming (Fig. 6e–h). The isotopic signature in the vegetated sediment exposed to warming decreased significantly from −50.8 ‰ to −54.06 ‰ (R²=0.67, p<0.001).

The δ¹³C isotopic composition of both CO₂ and CH₄ became heavier over time when the community was kept in the dark (Fig. 7), with a significant increase of δ¹³C-CH₄ in bare sediments (R²=0.94, p<0.01; Fig. 7d).

Figure 7Mean ± SE isotopic signature of CO₂ (δ¹³C-CO₂, left) and CH₄ (δ¹³C-CH₄, right) in (a, c) vegetated and (b, d) bare sediments exposed to prolonged darkness over the experimental period (number of days since the onset of the experiment).

Download

4.1 Carbon dioxide and methane air–seawater fluxes

The values reported for CO₂ and CH₄ fluxes varied greatly between the two sites studied here, with higher fluxes in the more organic sediments with higher biomass density (S2). CO₂ and CH₄ fluxes were also highly variable over time at the studied site, as the first evaluation of fluxes at the same location delivered rates up to 100-fold above the rates of the second measurement 1 week later. Hence, organic matter availability along with temperature may account for the large variation in CO₂ and CH₄ fluxes. Additionally, the variability of CO₂ and CH₄ fluxes could also be supported by infaunal species present in the cores that were not recorded in this study. These trends were similar to results reported in previous studies, as a high variability between species and locations was found (cf. Table 1 in Garcias-Bonet and Duarte, 2017).

Even though there were some differences, carbon, nitrogen and phosphorus concentrations were generally similar, and they did not seem to have an effect on CO₂ and CH₄ fluxes. Carbon, nitrogen and phosphorus concentrations were low compared to mean values (carbon: 33.6±0.31 % DW; nitrogen: 1.92±0.05 % DW; phosphorus: 0.23±0.011 % DW) reported for seagrass leaves by Duarte (1990). Serrano et al. (2018) explained the discrepancy between Red Sea data and global data with the extreme conditions in the Red Sea, such as low nutrient input and high temperatures, as well as a limited data set favoring high carbon stocks in the Mediterranean.

The results presented here add to the findings of Garcias-Bonet and Duarte (2017) identifying Red Sea seagrass communities as a significant source of CH₄. The presence of seagrass resulted in a larger organic matter supply to the sediments, favoring the presence of methanogens, which led to higher CH₄ fluxes compared to those fluxes supported in bare sediments (Barber and Carlson, 1993; Bahlmann et al., 2015), consistent with the up-to-100-fold-higher CH₄ fluxes supported by vegetated compared to bare sediments in this study. Additionally, higher fluxes in vegetated cores could be an indicator of direct effects resulting from the presence of seagrass, as vascular plants on land have been shown to have varying effects on methane emissions caused by differences in biomass and gross photosynthesis (Öquist and Svensson, 2002).

Similar trends were also seen by Garcias-Bonet and Duarte (2017), who reported an increase in CH₄ fluxes with increasing organic matter content in Red Sea seagrass sediments. They reported organic matter contents in Red Sea seagrass sediments ranging from 2.33±0.07 % (Halodule uninervis) to 12.42±0.23 % (Enhalus acoroides), including a mixed meadow with H. stipulacea and H. uninervis showing slightly increased organic matter content of 3.51±0.17 % compared to vegetated sediments at S2. Moreover, they found the highest CH₄ fluxes in meadows with the highest biomass, confirming our findings of higher fluxes at study site S2.

In terms of CO₂ equivalents, only the bare sediment maintained at 25 ^∘C seemed to act as a C sink over the experimental period, while the vegetated sediments, both maintained at 25 ^∘C and exposed to warming, acted as sources of greenhouse gases. A sublethal disturbance, such as warming below the lethal threshold, can therefore lead to a shift of seagrass ecosystems from acting as net sinks to net sources of greenhouse gases, as demonstrated experimentally here.

4.2 Effect of warming

Both CO₂ and CH₄ fluxes were higher in vegetated compared to adjacent bare sediments, indicating elevated remineralization rates in vegetated sediments as well as a higher susceptibility of seagrass sediment to increasing temperatures. Vegetated sediments exposed to warming shifted from acting as a CO₂ sink to an increasingly intense source, while the CO₂ fluxes in vegetated sediments maintained at 25 ^∘C decreased over time. Warming leads to an increase in both community photosynthesis and respiration, with respiration's faster rate of increase (Harris et al., 2006) explaining the shift to a CO₂ source in sediments exposed to a thermal stressor. However, the fluxes maintained at 25 ^∘C showed a net CO₂ uptake with a mean of 464.78±156.6 µmol CO₂ m⁻² d⁻¹ (Table S1), while those reported in a mixed Halodule sp. and Halophila sp. meadow in India showed a net CO₂ release (dry season: 1190±1600 µmol CO₂ m⁻² d⁻¹; wet season: 18 400 ± 8800 µmol CO₂ m⁻² d⁻¹; Banerjee et al., 2018). Both values reported were measured at higher temperatures (dry season: 30±0.68 ^∘C; wet season: 27.94±0.72 ^∘C; Banerjee et al., 2018) compared to our fluxes measured at 25 ^∘C, also indicating that temperature might lead to higher fluxes. An initial high CO₂ flux measured on day two after sampling could be an indicator for the experienced disturbance due to sample collection and transportation even though we allowed the cores some time to adapt.

Mean CH₄ fluxes at in situ temperature (25 ^∘C) in vegetated sediments were lower than the mean value of 85.1±27.8 µmol CH₄ m⁻² d⁻¹ reported for other seagrass meadows in the Red Sea (Garcias-Bonet and Duarte, 2017). In contrast, the community exposed to warming reached a maximum average CH₄ flux almost 4-fold higher than the community held at 25 ^∘C and showed a clear increase with warming relative to sediments held at 25 ^∘C. The increase in CH₄ fluxes with warming was consistent with reports from Barber and Carlson (1993) for a Thalassia testudinum community in Florida Bay and from Garcias-Bonet and Duarte (2017) for Red Sea seagrass communities, who reported higher CH₄ fluxes at higher temperatures. Additionally, previous research has shown that methanogenesis has a higher thermal dependence than respiration and photosynthesis (Yvon-Durocher et al., 2014), confirming the trends seen here with increasing fluxes at higher temperatures. We also reported a 10-fold decline in CH₄ fluxes over time for sediment communities maintained at 25 ^∘C, which could be attributable to increased sulfate reduction, reduced CH₄ production or a combination of both. Methane is produced under anoxic conditions in marine sediments, yet only a small portion is released, as CH₄ production by methanogens is compensated for by CH₄ oxidation by sulfate-reducing bacteria (Barnes and Goldberg, 1976). Similar to the trends seen in CO₂ fluxes, the decrease in CH₄ fluxes could be attributable to an initial stress response to the disturbance caused by sample collection and transportation. While reduced photosynthetic activity and a degradation in biomass could result in higher CH₄ fluxes (Lyimo et al., 2018), the cores maintained at 25 ^∘C might show the effect of healthy conditions.

Increasing water temperature led to a decrease in the CH₄ / CO₂ ratio. While there was ∼7 % of sequestered carbon released as CH₄ to the atmosphere in vegetated sediments at 25 ^∘C (on day two ), it decreased to ∼0.8 % in vegetated sediments at 37 ^∘C. In contrast, Banerjee et al. (2018) reported ∼1 % of carbon being released as CH₄.

The isotopic composition of CO₂ in all treatments showed generally heavier isotopic signatures compared to previous reports of seagrass carbon (average δ¹³C value of $-\mathrm{7.73}\pm \mathrm{0.11}$ ‰ for Red Sea seagrass and $-\mathrm{7.57}\pm \mathrm{0.15}$ ‰ for H. stipulacea in the Red Sea; Duarte et al., 2018), indicating various organic matter sources such as macroalgae blades (13.38±0.3 ‰), mangrove leaves (26.58±0.13 ‰) and seston (25.43±0.42 ‰; Duarte et al., 2018). However, the mean δ¹³C value of Red Sea seagrass sediments was reported to be $-\mathrm{13.36}\pm \mathrm{0.4}$ ‰ (Garcias-Bonet et al., 2019a), similar to the results found in this study. Our chosen study sites were located in an enclosed lagoon with a high abundance of mangrove forests, leading to the conclusion that mangroves might be a major source of organic matter for our study sites. However, a recent study applying stable isotope mixing models found the major contributors to the organic matter in seagrass sediments in the Red Sea to be seagrass leaves and macroalgae blades, with contributions of 43 % and 37 %, respectively (Garcias-Bonet et al., 2019a).

The isotopic signature of CO₂ released from bare sediments shifted with warming, suggesting a shift from seston, mangroves and macroalgae to seagrass carbon as the source of CO₂. In the vegetated cores, the isotopic composition of CO₂ stayed rather constant, indicating several sources of organic carbon with no clear shift, regardless of warming.

The isotopic signature of CH₄ in vegetated sediments confirmed its biogenic source as previous reports have shown that the isotopic signature of CH₄ from biogenic sources can range from −40 ‰ to −80 ‰, while the isotopic signature of CH₄ from geological and thermogenic sources ranges from −30 ‰ to −50 ‰ (Reeburgh, 2014). The isotopic composition of CH₄ in bare sediments was generally at the lower end of this range, with no clear shift with increasing temperature.

The isotopic composition of CH₄ can be determined by the production of CH₄ (methanogenesis) leading to lower δ¹³C values and the oxidation of CH₄ (methanotrophy) leading to higher δ¹³C values (Whiticar, 1990). Garcias-Bonet and Duarte (2017) reported fluctuations in the isotopic signature of CH₄ in Red Sea seagrass meadows, suggesting an indication of both processes. With exposure to increasing temperatures, we observed a shift to a lighter isotopic signature of CH₄ in vegetated sediments, thereby indicating an increasing CH₄ production by methanogens with warming.

4.3 Effect of prolonged darkness

Communities maintained at 25 ^∘C with a 12 h L : 12 h D photoperiod showed continuous net CO₂ uptake, while the communities kept in the dark shifted, as expected, to a heterotrophic state, acting as a CO₂ source. The net CO₂ production corresponded to community respiration rates, while that with a 12 h L : 12 h D photoperiod corresponded to the net community production.

We found, however, no effect of prolonged darkness on CH₄ fluxes, suggesting that the elevated CH₄ fluxes in vegetated sediments were not directly supported by fresh photosynthetic products, but rather by the elevated organic matter content in vegetated sediments compared to bare ones. These findings were in contrast to those reported by Lyimo et al. (2018), who found increased CH₄ fluxes under shading, indicating that degradation of belowground biomass might have been the key factor related to increased CH₄ fluxes. However, they also reported varying results for different shading intensities, with low intensity having similar fluxes compared to their control group (Lyimo et al., 2018). In contrast, Öquist and Svensson (2002) found that photosynthesis might be regulating methane fluxes in a subarctic peatland ecosystem, with lower photosynthesis resulting in lower methane fluxes.

4.4 Implications

Reports on greenhouse gas fluxes by seagrass ecosystems are limited (Oremland, 1975; Barber and Carlson, 1993; Alongi et al., 2008; Deborde et al., 2010; Bahlmann et al., 2015; Garcias-Bonet and Duarte, 2017; Banerjee et al., 2018; Lyimo et al., 2018), and no reports had been previously published on how increasing seawater temperatures might affect greenhouse gas fluxes by seagrass ecosystems. Blue carbon ecosystems have been shown to turn into C sources when disturbances lead to mortality (Macreadie et al., 2015; Lovelock et al., 2017; Arias-Ortiz et al., 2018), consistent with the very large CO₂ and CH₄ fluxes observed in one vegetated sediment where the seagrass died when warmed to 33 ^∘C. However, even where seagrass remained alive, warming led to elevated greenhouse fluxes. Additionally, the elevated nutrient and high organic matter stock in seagrass meadows, which supports 1.7 times more organic matter content than surrounding bare sediments, can promote an increase in CO₂and CH₄ fluxes following disturbance (Gonsalves et al., 2011; Sotomayor et al., 1994). Our results suggest that this stock in seagrass sediments may be remineralized to support net greenhouse gas fluxes at the warmer temperatures reached and with further warming of the Red Sea. Hence, warming may, as other disturbances (Lovelock et al., 2017), shift seagrass ecosystems from net sinks to net sources of greenhouse gases, thereby providing a feedback mechanism that may contribute to further enhancing global warming.