A preliminary study of carbon dioxide and methane emissions from patchy tropical seagrass meadows in Thailand

Study area

Sampling was conducted Koh Mook, an island off the south-west coast of Thailand. The island is part of the Haad Chao Mai National Park in the Andaman Sea (Fig. 1A). Seven seagrass species have been recorded at Koh Mook, including E. acoroides, T. hemprichii, C. rotundata, C. serrulata, H.ovalis, H. uninervis and S. isoetifolium (Nakaoka & Supanwanid, 2000). Three species are dominant (E. acoroides, T. hemprichii, and H. ovalis) (Nakaoka & Supanwanid, 2000). The intertidal flat at Koh Mook is approximately 400 m from the shoreline (Nakaoka & Supanwanid, 2000), where the majority of the above species can be found. Samples were collected from areas of patchy E. acoroides, patchy T. hemprichii, and bare sand in the same seagrass meadow. Three benthic chambers were deployed at each area (Figs. 1B and 1C), for a total of nine chambers (Fig. 1B). The data were collected in June 2023, which is the beginning of the rainy season in Thailand (Thai Meteorological Department, 2023). Water temperatures during data collection ranged from 24.2° to 33.6°C while the average light intensity during the day was 94.93 μmol m⁻² s⁻¹.

Design of benthic chamber, experimental setup and data collection

The transparent PVC benthic chambers were 120 cm in length for E. acoroides sampling, and 61.5 cm for T. hemprichii and bare sand sampling. They contained an air pocket and modified sampling ports for collecting gas and water samples (Fig. 2A).

Benthic chambers were installed approximately 15 cm into sediment (Fig. 2A). Incubation was conducted for 18 h, and samples were collected after 10 min of chamber set-up for the first sample collection at t0, 12 h after (t1) and 18 h after (t2). Approximately 5 mL of gas sample and 20–25 mL of water sample were collected from each chamber, using a syringe inserted into the sampling port (Fig. 2B). Samples were transferred to modified blood tube that were sealed with silicone to prevent sample leakage. The tubes were stored upside-down in a cold box/refrigerator (4 °C) until analysis in the lab (Asplund et al., 2022). Temperature and light intensity at each sampling location were recorded with a HOBO logger during the incubation time.

CO₂ and CH₄ emissions, carbon metabolism with measuring changes in dissolved oxygen concentration

CO₂ and CH₄ emissions from seagrass to the atmosphere were estimated following the method of Asplund et al. (2022). In the laboratory, samples were analyzed by using a method described in Kitpakornsanti, Pengthamkeerati & Limsakul (2022). A gas chromatograph (Agilent 6890N GC, Agilent, Santa Clara, CA, USA) equipped with an electron capture detector was used to determine CO₂ and CH₄ production from sediment. Quality control was checked using 1,000 rpm CO₂ and CH₄ on each sample, and standard deviation of replicates was less than 5%. The CO₂ and CH₄ emissions from seagrass sediment were calculated using the following equation (Eq. (1)) from Bulmer et al. (2017),

(1) $$F=\left({\displaystyle \frac{dG}{dt}}\right)\times V\times P/\left(R\times T\times A\right)$$where F is the CO₂ and CH₄ emissions (mmol m⁻² h⁻¹); dG/dt is the slope of the linear regression between gas concentrations and deployment time (ppmv h⁻¹); V is the headspace volume (L) of the chamber; P is the barometric pressure (1 atm); R is the ideal gas constant (8.205746 × 10⁻⁵ atm m² K⁻¹ mol⁻¹); T is the average temperature (^oK) inside the closed chamber at each measurement in the field; A is the sectional area (m²) of the bottom of the closed chamber.

Furthermore, community carbon metabolism was estimated from the dissolved oxygen (DO) concentration (Jiménez-Ramos et al., 2022) determined using the Winkler titration method. The collected water samples were kept in a cold box/refrigerator (4 °C) until they were analyzed in the laboratory (Egea et al., 2019b). Hourly community respiration rates (CR) were estimated as the difference in DO concentration between t1 and t0 samples divided by the time elapsed between sampling, using the following equation (Eq. (2)) (Jiménez-Ramos et al., 2022):

(2) $$CR\left(\frac{mmol{O}_2}{m^{2}d}\right)=\frac{D{O}_{t1}\left(\frac{mg{O}_2}{1}\right)-D{O}_{t0}\left(\frac{mg{O}_2}{1}\right)}{\Delta T_{t0}-T_{t1}\left(h\right)}\times \frac{volume\left(l\right)}{Area\left(m^2\right)}\times \frac{1mmol{O}_2}{32mg{O}_2}$$where DO_t1 and DO_t0 are the DO concentrations at t1 and t0, ΔT is the time elapsed between sampling events, and volume and area are for benthic chambers.

Hourly rates of net community production (NCP) were estimated as the difference in DO concentration between t2 and t1 samples divided by the time elapsed between sampling using the following equation (Eq. (3)):

(3) $$NCP\left({\displaystyle \frac{mmol{O}_2}{m^{2}d}}\right)={\displaystyle \frac{D{O}_{t2}\left({\displaystyle \frac{mg{O}_2}{1}}\right)-D{O}_{t1}\left({\displaystyle \frac{mg{O}_2}{1}}\right)}{\mathrm{\Delta }{T}_{t1}-T_{t2}\left(h\right)}\times {\displaystyle \frac{volume\left(l\right)}{Area\left(m^2\right)}\times {\displaystyle \frac{1mmol{O}_2}{32mg{O}_2}}}}$$where DO_t2 and DO_t1 are the DO concentrations at t2 and t1, ΔT is the time elapsed between sampling events, and volume and area are for benthic chambers.

Hourly rates of community gross primary production (GPP) were calculated as the sum of hourly rates of CR and NCP:

(4) $$GPP=CR+NCP.$$

Finally, daily rates of community GPP, CR, and NCP were estimated following from:

(5) $$GPP=GPP\times photoperiod\left(h\right)$$

(6) $$CR=CR\times 24h$$

(7) $$NCP=GPP-CR$$where photoperiod corresponds to the number of sunlight hours measured on each sampling day. Metabolic rates in DO units were converted to carbon units assuming photosynthetic and respiratory quotients of 1 (values are used widely in seagrass studies) (Egea et al., 2019b).

Statistical analysis

Data were checked for normality and transformed as necessary before further analysis to meet the assumptions of the ANOVA. Data estimates throughout this work were presented as averages with the standard deviation (SD). ANOVA was used to test significant differences between sampling areas (bare sand, E. acoroides and T. hemprichii) and CO₂ emissions, and between sampling times (t0, t1 and t2) and CO₂ emissions. A post-hoc test was performed to compare differences in CO₂ emissions between the analyzed groups, including sampling times across all three sampling areas. Furthermore, MANOVA was used to test significant differences between carbon metabolism with both sampling area and sampling time. All statistical analyses were done in R statistical software version 4.3.1 (R Studio: Integrated Development for R, 2021), using the following packages: tidyverse 2.0.0 (Wickham et al., 2019), agricolae 1.3-6 (de Mendiburu, 2023), emmeans 1.8.8 (Lenth, 2023), multcomp 1.4-25 (Hothorn, Bretz & Westfall, 2008), car 3.1-2 (Fox & Weisberg, 2019), ggplot2 3.4.3 (Wickham, 2016), MASS 7.3-60 (Venables & Ripley, 2002), mvoutlier 2.1.1 (Filzmoser & Gschwandtner, 2021), pastecs 1.3.21 (Ibanez, 2018), mvnormtest 0.1-9 (Jarek, 2012), and reshape2 1.4.4 (Wickham, 2007).

Carbon dioxide emissions from Koh Mook seagrass meadows

ANOVA indicated a significant difference (p < 0.001) between CO₂ emissions and sampling area, where the average CO₂ emissions were significantly higher (p < 0.05) in the E. acoroides area (0.143 ± 0.000788 μmol CO₂ m⁻² h⁻¹) than in the T. hemprichii area (0.0489 ± 0.000403 μmol CO₂ m⁻² h⁻¹). Meanwhile, in the bare sand area, the average CO₂ emissions were significantly the lowest (p < 0.05) at 0.0082 ± 0.0000404 μmol CO₂ m⁻² h⁻¹ (Fig. 3).

Furthermore, the post-hoc test showed that CO₂ emissions were not significantly different (p > 0.05) between sampling times across all three sampling areas (Fig. 4). However, in the E. acoroides area, the average CO₂ emissions were slightly higher at t1 (0.1440 ± 0.00080 μmol CO₂ m⁻² h⁻¹) than at t0 and t2 (0.1424 ± 0.00080 and 0.1431 ± 0.00080 μmol CO₂ m⁻² h⁻¹, respectively). The same pattern was found in the bare sand area, where CO₂ emissions were also slightly higher at t1 (0.00825 ± 0.00042 μmol CO₂ m⁻² h⁻¹) than t0 (0.00816 ± 0.00042 μmol CO₂ m⁻² h⁻¹) and t2 (0.00819 ± 0.00042 μmol CO₂ m⁻² h⁻¹). In the T. hemprichii area, CO₂ emissions were only recorded at t0 (0.0485 ± 0.02817 μmol CO₂ m⁻² h⁻¹) and t1 (0.0491 ± 0.02817 CO₂ m⁻² h⁻¹) as the benthic chambers were slightly damaged at t2, and CO₂ emissions could not be measured.

Seagrass community carbon metabolism

Overall, the MANOVA showed that there was no significant difference (p > 0.05) in daily carbon metabolism across the different sampling areas (Fig. 5). However, some slight trends were observed in GPP, CR, and NCP, such that the values were always higher in the vegetated areas than in the bare sand area (Fig. 5).

The daily average GPP was highest in the E. acoroides area, followed by the T. hemprichii and bare sand areas, with average values of 1.61 ± 3.17, 1.53 ± 0.62, and 0.71 ± 1.78 mmol C m⁻² d⁻¹, respectively (Fig. 5A). Daily CR was in the T. hemprichii area at –3.50 ± 0.75, followed by the E. acoroides area at −3.33 ± 2.62 and bare sand area at −2.30 ± 1.38 mmol C m⁻² d⁻¹ (Fig. 5B). Daily NCP was highest in the T. hemprichii area (5.04 ± 0.35 mmol C m⁻² d⁻¹), slightly lower in the E. acoroides area (4.95 ± 1.01 mmol C m⁻² d⁻¹) and lowest in the bare sand area (3.01 ± 0.41 mmol C m⁻² d⁻¹) (Fig. 5C). As they are >1, these values of NCP suggest that the seagrass meadows at Koh Mook are autotrophic.

MANOVA showed no significant difference (p > 0.05) between seagrass community carbon metabolism, including GPP, CR, and NCP, and sampling time (t0, t1 and t2) (Fig. 6). However, some patterns were observed across sampling times and areas. We found that the GPP and CR in the T. hemprichii and E. acoroides areas followed the same pattern (Figs. 6 A and 6B), with higher values at t1 (GPP: 1.575 and 2.007 mmol C m⁻² d⁻¹, CR: −2.836 and −1.556 mmol C m⁻² d⁻¹, respectively) than t0 (GPP: 1.499 and 1.223 mmol C m⁻² d⁻¹, CR: −4.171 and −5.116 mmol C m⁻² d⁻¹, respectively). However, there were some differences in GPP and CR between the two seagrass species. In the E. acoroides area, GPP was 0.78 mmol C higher at t1 than t0, and CR was 3.56 mmol C higher at t1 than t0. Meanwhile, in the T. hemprichii area, GPP was only 0.07 mmol C higher at t1 than t0, and CR 0.91 mmol C higher (Figs. 6A and 6B). On the other hand, in the bare sand area, GPP was slightly higher at t0 (0.754 mmol C m⁻² d⁻¹) than t1 (0.668 mmol C m⁻² d⁻¹), while CR was slightly higher at t1 (−1.919 mmol C m⁻² d⁻¹) than t0 (−2.681 mmol C m⁻² d⁻¹) (Figs. 6A and 6B). Meanwhile, the NCP values at all sampling areas were higher at t0 (3.435, 5.671, and 6.340 mmol C m⁻² d⁻¹, respectively) than t1 (2.587, 4.411, and 3.564 mmol C m⁻² d⁻¹, respectively) (Fig. 6C).

Methane emissions

CH₄ emissions to the atmosphere from the Koh Mook seagrass meadows were at low levels, only detected in samples from vegetated areas, where the values were higher in the E. acoroides area than the T. hemprichii area (Fig. 7A). Furthermore, CH₄ emissions were only detected at t0 (2.09 μg CH₄ m⁻² h⁻¹ in the E. acoroides area and 1.35 μg CH₄ m⁻² h⁻¹ in the T. hemprichii area) (Fig. 7B). Additionally, CH₄ emissions were undetected in the bare sand area (Figs. 7A and 7B).

Carbon dioxide emissions

The results showed that CO₂ emissions from seagrass meadows to the atmosphere in Koh Mook Island, Thailand, were relatively low (less than 10-fold) compared to emissions reported in several previous studies, but different methodologies were employed across these studies (Burkholz, Garcias-bonet & Duarte, 2020; Roca et al., 2022).

Although the meadows at Koh Mook release CO₂, the levels are low and do not outweigh their role in carbon sequestration. The low CO₂ emissions at Koh Mook are suppressed by the condition of the seagrass meadows, which are not under localized threats significant enough to change the sediment condition from anoxic to oxic (Lovelock, Fourqurean & Morris, 2017). Potential disturbances in this area include local community activities, such as monthly gleaning and small boats, which do not threaten the seagrass carbon stock. Thus, they will continue to contribute to carbon sequestration as long as local conditions remain the same (Dahl et al., 2023).

There were significant differences in CO₂ emissions between sampling areas. Higher values were found in vegetated areas compared to unvegetated areas, which was the pattern reported by Latifah et al. (2023) in mixed seagrass meadows with five species (E. acoroides, T. hemprichii, C. rotundata, O. serrulata, and H. ovalis) at Karimun Jawa, Indonesia and by Burkholz, Garcias-bonet & Duarte (2020) in H. stipulacea meadows in the Red Sea. We suspect that this pattern is due to the presence of seagrass, which store organic carbon and increase oxygen input into the sediment, thereby driving remineralization, which causes an increase in the production of CO₂ (Dahl et al., 2023). Banerjee et al. (2019) proposed that CO₂ emissions were relatively higher in vegetated areas because the composition and action of the more active and dominant microbes drove GHG release. Furthermore, vegetated areas contribute to the production of particulate organic matter and trap terrigenous dissolved organic carbon (tDOC), facilitating the remineralization of particulate matter, which leads to CO₂ emissions from coastal waters (Wahyudi et al., 2021; Zhou et al., 2021). This process decreases the pH in shallow waters (seagrass meadows) and encourages the release of CO₂ into the atmosphere (Capelle et al., 2020; Zhou et al., 2021). Moreover, the photosynthesis that occurs only in vegetated areas consumes CO₂, driving the deposition of calcium carbonate (CaCO₃), which increases the carbonate (CO₃²⁻) saturation of the water and increases water pH (Buapet, Gullström & Björk, 2013; Latifah et al., 2023; Leclercq, Gattuso & Jaubert, 2000; Semesi, Beer & Björk, 2009). This process increases aquatic pCO₂ until the pressure is high enough to release CO₂ into the atmosphere (Kalokora et al., 2020; Latifah et al., 2023; Leclercq, Gattuso & Jaubert, 2000).

We identified a slight difference in CO₂ emission values between sampling times with higher values at t1 (around 6-am on the second day). However, the statistical results did not show significant differences between CO₂ emissions and sampling time. The same CO₂ emission pattern was reported by Ismail et al. (2023) in mesocosm experiments on different plant community compositions, including seagrass (T. hemprichii) and calcifying algae (Hydrolithon sp.). The factors that cause small fluctuations in CO₂ emissions are photosynthesis and respiration. The slightly increased CO₂ emission at t1 was probably caused by nocturnal respiration from plant communities and sediments. The lower CO₂ emission values at t2 (around 12 noon on the second day) were the result of diurnal photosynthesis which absorbs CO₂ (Ismail et al., 2023). The lowest CO₂ emissions were recorded at t0 (around 6 pm on the first day) because the net CO₂ release decreased in the afternoon, as was found in Ismail et al. (2023). Possibly due to DO production in the water by the photosynthesis process throughout the day encourages small CO₂ production in the late afternoon (Ismail et al., 2023).

Seagrass community carbon metabolism

Carbon metabolism in seagrass meadows is typically intensive, supporting high GPP (Hemminga & Duarte, 2000). The present results indicate that the average daily GPP ranged from 0.711 to 1.33 mmol C m⁻² d⁻¹, with daily CR between −2.30 and −3.70 mmol C m⁻² d⁻¹. Consequently, we observed that daily NCP ranged from 3.01 to 5.04 mmol C m⁻² d⁻¹, which is within the range (0.6 to 6.8) reported for Thallasia testudenum meadows in Texas (Ziegler & Benner, 1999) but lower than reported for Cymodocea nodosa meadows (18 ± 4 mmol C m⁻² d⁻¹) in winter at Cadiz Bay, Spain (Egea et al., 2020). The negative CR value indicates that Koh Mook seagrass meadows absorb more CO₂ than they release, promoting balance in the seagrass community carbon metabolism and increasing NCP. These results indicate that the seagrass meadows at Koh Mook generally exhibit autotrophic conditions (NCP > 1) or contribute to a carbon sink. This preliminary research suggests that these meadows have the potential to accumulate organic carbon.

Seagrass carbon metabolism rates showed variations between the vegetated areas (patchy E. acoroides and T. hemprichii) and unvegetated bare sand. Although GPP, CR and NCP were not significantly different between sampling areas and sampling times, the presence of patchy E. acoroides and T. hemprichii influenced carbon metabolism at the study site. Higher carbon metabolism can be driven by seagrass biomass, as it provides a complex habitat that directly impacts carbon metabolism (Barrón et al., 2004; Egea et al., 2020; Johnson et al., 2020). Furthermore, the above-ground biomass of E. acoroides and T. hemprichii at Koh Mook contributed to carbon metabolism by providing photosynthetic tissue (Jiménez-Ramos et al., 2022).

Seagrass carbon metabolism, especially NCP, was higher at t0 (around 6 pm on the first day) than t1 (around 6 am on the second day). This pattern was possibly influenced by the short incubation time, as NCP tends to decrease with prolonged incubation (Olivé et al., 2016), which reduces oxygen levels in the benthic chamber, thereby reducing the NCP budget. Moreover, the period of darkness leading up to t1 increased CR, which utilized O₂ and released CO₂ (Olivé et al., 2016).

Methane emissions

Seagrass meadows are also considered a natural source of small amounts of CH₄ (Asplund et al., 2022). At the study site, sediment–water methane was only detected at t0 at the E. acoroides and T. hemprichii areas. CH₄ was not detected in the unvegetated area or at t1 and t2, most likely because the sampling method used was not sensitive enough to capture the low methane levels released at those sampling times, mainly when sampling is conducted at a specific time when methane production is low. The CH₄ values recorded in this study are within the range reported for Zostera marina meadows in Nordic waters (0.3–30 μg CH₄ m⁻² h⁻¹) (Asplund et al., 2022). Our results suggest that the seagrass meadows at Koh Mook contribute only slightly to CH₄ emissions. The low CH₄ production at the Koh Mook meadow might be related to the salinity of these waters. Higher salinity allows higher sulfate levels to develop in seawater, thus microbes that utilize sulfate become more dominant, and methanogenesis is inhibited (Schorn et al., 2022).

Even though CH₄ values were low, this gas is nonetheless a GHG with a Sustained-Flux Global Warming Potential (SGWP) 45–96 times greater than CO₂ (Neubauer & Megonigal, 2015). Therefore, small emissions of CH₄ can still offset the carbon sink in seagrass ecosystems and must be taken into consideration. For instance, CH₄ emissions from P. oceanica meadows could offset less than 1% of the carbon buried in sediment (Yau et al., 2022), while in mangrove ecosystems, CH₄ emissions could offset up to 18% of carbon burial (Rosentreter et al., 2018). Although CH₄ emissions were only detected in the vegetated areas at Koh Mook, the contribution of seagrass to CH₄ emissions should not be ignored since seagrass meadows can facilitate CH₄ production. Furthermore, under certain conditions, such as increasing temperature (Begum et al., 2020), low salinity (Schorn et al., 2022) and high organic carbon content (Asplund et al., 2022), CH₄ production can disrupt the role of seagrass as a GHG sink. According to Schorn et al. (2022), the release of CH₄ by seagrass will offset carbon burial by 4 to 5%. Finally, CH₄ emissions were only found at t0 after 10 min of chamber setup, and it could be that sediment disturbance during chamber setup released CH₄ into the water column.

Research limitations and improvements

This preliminary study has several limitations that need to be considered, including the duration of CO₂ and CH₄ emission measurements, which was only 18 h, thus not fully representing the daily emission values for 24 h.

We emphasize the importance of improving this study in the future by assessing long-term trends and exploring different environmental conditions, such as temperature and salinity, to ensure a more comprehensive understanding of the contribution of seagrass meadows in Thailand to GHG emissions. The benthic chamber used in our research proved to be a more practical, effective, and reliable method of measuring GHG emissions in remote areas, such as Koh Mook, than methods such as eddy-covariance (Tokoro et al., 2007). However, this method also has weaknesses because it can increase surface gas flux due to the pressure difference between water and air (Ismail et al., 2023).

In future research, several improvements and considerations must be made to ensure the validity of capturing GHG emissions from seagrass meadows with a benthic chamber. Local tides, waves, substrates, and coverage are important factors, and gas leakage from the chamber must be addressed. Additionally, this preliminary study had limited spatial sampling coverage and the number of benthic chambers. In the future, the number of chambers will increase to ensure that emissions measurements fully represent the scale of the large study area. Moreover, for a comprehensive understanding of the carbon dynamics in seagrass, it is also recommended that water- and air-pCO₂ be measured. Furthermore, for a comprehensive understanding of the carbon dynamics in seagrass, it is also recommended that water- and air-pCO₂ be measured.