Radiocarbon isotopic evidence for assimilation of 1 atmospheric CO 2 by the seagrass Zostera marina 2 3

Submerged aquatic vegetation takes up water-column dissolved inorganic carbon (DIC) as a 11 carbon source across its thin cuticle layer. It is expected that marine macrophytes also use 12 atmospheric CO2 when exposed to air during low tide, although assimilation of atmospheric 13 CO2 has never been quantitatively evaluated. Using the radiocarbon isotopic signatures (ΔC) 14 of the seagrass Zostera marina, DIC and POC, we show quantitatively that Z. marina takes up 15 and assimilates atmospheric modern CO2 in a shallow coastal ecosystem. The ΔC values of 16 the seagrass (−40‰ to −10‰) were significantly higher than those of aquatic DIC (−46‰ to 17 −18‰), indicating that the seagrass uses a C-rich carbon source (atmospheric CO2, +17‰). 18 A carbon-source mixing model indicated that the seagrass assimilated 0–40% (mean, 17%) of 19 its inorganic carbon as atmospheric CO2. CO2 exchange between the air and the seagrass 20 might be enhanced by the presence of a very thin film of water over the air-exposed leaves 21 during low tide. Our radiocarbon isotope analysis, showing assimilation of atmospheric 22 modern CO2 as an inorganic carbon source, improves our understanding of the role of 23 seagrass meadows in coastal carbon dynamics. 24


Introduction
Submerged aquatic vegetation assimilates dissolved inorganic carbon (DIC) from the water column as a carbon source.Seagrasses take up DIC across their thin cuticle layer (Hemminga and Duarte, 2000), as their leaves lack stomata despite being angiosperms (Larkum and Den Hartog, 1989).An alternative carbon source, atmospheric CO 2 (C air ), can-not directly reach seagrasses when they are completely submerged; however, seagrasses can take up C air when their leaves are exposed to air during low tide (Clavier et al., 2011;Jiang et al., 2014;Leuschner and Rees, 1993).Seagrasses rely largely on aqueous CO 2 (CO 2 (aq)) as a carbon source for photosynthesis in nature (Beer and Koch, 1996).Some seagrass species, however, can use bicarbonate ions (HCO − 3 ) as a major carbon source (Beer and Rehnberg, 1997;Beer et al., 2002), although there is considerable interspecific variation in HCO − 3 utilization (Campbell and Fourqurean, 2013).As CO 2 (aq) is in limited supply under normal seawater conditions (pH ≈ 8), comprising only 1 % (roughly 10-15 µmol L −1 ) of the DIC pool, photosynthesis in seagrasses under high light conditions is frequently limited by carbon availability (Campbell and Fourqurean, 2013;Invers et al., 2001;Zimmerman et al., 1995).Under normal seawater pH conditions, the bicarbonate ion (HCO − 3 ) is the most abundant inorganic carbon species, accounting for nearly 90 % of the DIC pool (Plummer and Busenberg, 1982;Zeebe and Wolf-Gladrow, 2001).Some seagrass species indirectly use HCO − 3 under low-CO 2 (aq) conditions (Beer et al., 2002;Campbell and Fourqurean, 2013), using one or both of the following suggested mechanisms: (1) extracellular dehydration of HCO − 3 into CO 2 (aq) via membrane-bound enzymes (Beer and Rehnberg, 1997) and/or (2) electrogenic proton (H + ) extrusion into an boundary layer on the leaf surface, facilitating HCO − 3 /H + cotransport (Hellblom et al., 2001).Diffusion of CO 2 in water is much slower than that in air.During low tide, air-exposed aquatic macrophytes have a thin film of water between the air and their leaves, which promotes the uptake of C air , in contrast to high tide, when there is a thick water layer inhibiting the uptake of C air (Ji Published by Copernicus Publications on behalf of the European Geosciences Union.and Tanaka, 2002).Previous studies have shown the possibility of C air uptake by seagrasses by using evidence from stable carbon isotope ratios (δ 13 C) in seagrasses and the two carbon sources (DIC and C air ;Clavier et al., 2011;Cooper and McRoy, 1988;Raven et al., 2002).However, the 13 C method has considerable uncertainty because in addition to the source of carbon, the δ 13 C values of seagrasses are also determined by other factors such as the chemical species of DIC (CO 2 (aq) or HCO − 3 ) and photosynthetic carbon demand.The chemical species in the carbonate system (CO 2 (aq), HCO − 3 , and carbonate ion; CO 2− 3 ) have distinct δ 13 C values, and isotopic fractionations change depending on pH and temperature (Zeebe and Wolf-Gladrow, 2001;Zhang et al., 1995).Because the δ 13 C of HCO − 3 (0 ‰) is isotopically distinct from that of both CO 2 (aq) (−9 ‰) and C air (−8 ‰) under normal seawater conditions (pH ≈ 8), δ 13 C values in seagrasses become higher with increasing of HCO − 3 use (Campbell and Fourqurean, 2009;Hemminga and Mateo, 1996;Raven et al., 2002).However, quantification of the contribution of C air is impossible because of the δ 13 C value overlap between CO 2 (aq) and C air although low δ 13 C in seagrasses could be explained by the assimilation of either 13 C-depleted CO 2 (aq) or C air .Also, changes in the photosynthetic carbon demand driven by irradiance fluctuations affect the isotopic fractionation factor (Hemminga and Mateo, 1996;Raven et al., 2002).
The natural abundance of radiocarbon ( 14 C) has recently been used to assess food web structures (Ishikawa et al., 2014) and the origin and components of organic-matter pools (Goñi et al., 2013), as carbon sources have specific 14 C concentrations ( 14 C).The 14 C of inorganic carbon also has specific values depending on the source, such as DIC or C air .The 14 C of DIC generally differs from that of atmospheric CO 2 because of the longer residence time of carbon in aquatic ecosystems than in the atmosphere (Ishikawa et al., 2014;Stuiver and Braziunas, 1993).Moreover, the calculation of 14 C by internal correction using δ 13 C values eliminates any effects from isotopic fractionation (Stuiver and Polach, 1977), overcoming one of the major uncertainties in the conventional δ 13 C approach.This study is the first to show quantitative evidence that the seagrass Zostera marina assimilates modern C air , based on the 14 C values of the seagrass and two carbon sources.

Field surveys
Field surveys were conducted in 2014 during the growing season of Z. marina (May, July, September and November) in Furen Lagoon, Japan (Fig. 1; 43 Furen, Yausubetsu, and Pon-Yausubetsu rivers.The lagoon is covered by large seagrass meadows (67 % of the total area) dominated by Z. marina.The offshore of the lagoon (Sea of Okhotsk) is influenced by the dynamics of both the Oyashio and the Soya warm current.Surface water samples (depth, 0.1 m) for DIC (concentration and isotopic signatures) and total alkalinity (TA) in the water column were collected from a research vessel along the salinity gradient at seven stations in the lagoon (Fig. 1; stations F1-F7).At each station, one water sample was collected for measuring DIC and TA, and the salinity of the surface water was recorded with a conductivity-temperature sensor (COMPACT-CT; JFE Advantech, Nishinomiya, Japan).The samples for isotopic analysis of DIC were collected into 500 mL hermetically sealed glass bottles (Duran bottle; SCHOTT AG, Mainz, Germany), which were poisoned by adding saturated mercuric chloride solution (400 µL per bottle) to prevent changes in DIC due to biological activity.The samples for measuring DIC concentration and TA were collected into 250 mL Duran bottles (SCHOTT AG), which were poisoned with saturated mercuric chloride solution (200 µL per bottle).Seagrass (Z.marina) leaves were collected at four stations covered by Z. marina meadows (Fig. 1; stations F3, F4, F8 and F9) along the salinity gradient.The stations were located in subtidal zones (mean water depth, 0.83-1.12m).The above-ground wet-weight biomass of the seagrass, estimated from randomly thrown quadrats (0.0625 m 2 ), ranged from 500 to 6800 g m −2 .Three or four independent samples of seagrass leaves were collected at each station.Both the biofilm and epiphytes covering the leaves were gently removed by hands with powder-free groves and washed off using ultrapure water (Milli-Q water; Millipore, Billerica, MA, USA).To estimate the 14 C of C air , leaves of a terrestrial plant (giant reed, Phragmites australis) were collected near the lagoon.Plant samples were freeze-dried and subsamples were homogenized.To remove carbonate, the plant samples were acidified with 1 N HCl and dried again.

Carbon isotope analysis
We determined the stable carbon isotope ratios (δ 13 C) and radiocarbon concentrations ( 14 C) of seagrass leaves, terrestrial plant leaves, DIC samples and POC samples.Prior to 14 C and δ 13 C measurements, samples were subjected to graphite purification as follows.DIC samples for 14 C and δ 13 C analysis were acidified (pH < 2) with H 3 PO 4 and sparged using ultra-high purity mixed N 2 /H 2 gas.The powdered plant leaves and POC samples for 14 C and δ 13 C analysis were combusted in an elemental analyzer (either a Euro EA3000, EuroVector, Milan, Italy; or a Flash 2000, Thermo Fisher Scientific, Inc., Waltham, Massachusetts, USA).For each process, the CO 2 evolved was collected cryogenically and purified in a vacuum line.The purified CO 2 was then reduced to graphite using hydrogen and an iron catalyst at 650 • C for 10 h.The 13 C and 14 C concentrations were measured using an accelerator mass spectrometer (AMS).The AMS results are reported as 14 C (‰) values (Stuiver and Polach, 1977) as follows: (1) The 14 C values have been corrected by the radioactive decay of an international standard (oxalic acid) since AD 1950 (Stuiver and Polach, 1977).The δ 13 C values are reported relative to Vienna Pee Dee Belemnite.δ 13 C data were corrected using an internal standard.The analytical precision of the AMS was within 0.7 ‰ for δ 13 C and 3 ‰ for 14 C.

Data analysis
Because DIC taken up by seagrasses is a mixture of DIC from two sources (terrestrial and oceanic) each having distinct 14 C values, it is reasonable to use salinity as a proxy for the extent of mixing of these two sources as well as for the salinity gradient-based comparison between 14 C of DIC and seagrass.This comparison was therefore possible even though DIC and Z. marina samples were not necessarily collected from the same stations (Fig. 1).Analyses of covariance (ANCOVA) were used to examine the difference in 14 C value between seagrass leaves and DIC.These differences provide evidence that the seagrasses assimilate C air .We selected salinity, categorical data (seagrass leaves or DIC) and the interaction term as the explanatory variables.
The relative contribution of C air to assimilated seagrass carbon was calculated by a two-carbon-source mixing model using the 14 C values of DIC ( 14 C DIC ), C air ( 14 C air ), and the seagrass ( 14 C seagrass ) at each of four stations as follows: C air (% contribution) = ( 14 C seagrass − 14 C DIC )/ ( 14 C air − 14 C DIC ) • 100. (6) 14 C air was estimated from the 14 C value of the sampled terrestrial plants ( 14 C = +17.2‰).The 14 C values of DIC as the carbon source for Z. marina in the mixing model were estimated from the linear model fitted with the ANCOVA.

Results and discussion
Our radiocarbon isotopic analysis shows quantitatively that the seagrass Z. marina uses C air in a shallow lagoon (Fig. 2a).In May and July 2014, 14 C seagrass was significantly higher  than 14 C DIC even if the effects of salinity were considered (ANCOVA, P < 0.001), and the interaction term was not significant (ANCOVA, P > 0.05).Our results indicate that the changes in 14 C DIC are regulated mostly by mixing between high-14 C river water and low-14 C seawater: the seagrass uses aquatic DIC as the main carbon source, as expected from previous studies (Campbell and Fourqurean, 2013;Hemminga and Duarte, 2000;Invers et al., 2001).The 14 C seagrass reflects 14 C DIC from May to July because Z. marina leaves start to grow in early May when sea ice is thawing at the study site, with the turnover time of leaves being 30-90 days (mean, 60 days; Hosokawa et al., 2009).Furthermore, the negative relationship between salinity and 14 C seagrass cannot be explained by any residual contamination from terrestrial organic carbon on the leaves because the terrestrial POC was 14 C-depleted (mean 14 C of terrestrial POC, −74.7 ± 23.4 ‰).
The significantly higher values in 14 C seagrass than 14 C DIC shows that the seagrass assimilates 14 C-rich C air ( 14 C around 17 ‰; Fig. 2a).The two-carbon-source mixing model indicated that the seagrass assimilated 0-40 % (mean ±SD, 17 ± 12 %) of its inorganic carbon as C air ; the contribution was 20 ± 12 % in the low-salinity zone (salinity, 12-15) and 13 ± 12 % in the high-salinity zone (salinity, 25-29; Fig. 2b).The contribution of C air as a carbon source varied greatly even between samples from the same station (Fig. 2b).Because we did not determine the exposure time of each shoot in this study, we are unable to quantify any relationship between the contribution of C air and air exposure time; however, the exposure time would mediate the assimilation of C air (Clavier et al., 2011).
As 14 C DIC was significantly lower than 14 C air , the contribution of C air can be determined for May and July 2014 (Fig. 2a).This radiocarbon isotopic approach would be useful in the high latitudes of the Pacific Ocean where surface seawater is 14 C-depleted ( 14 C DIC < 0 ‰; Talley, 2007).In contrast, the 14 C DIC in surface seawater is generally higher than 14 C air in other regions of the Pacific Ocean because of bomb-derived 14 C (Talley, 2007).
In any case, the 14 C approach is potentially applicable to other regions by using the 14 C gradient.However, the seasonal dynamics of 14 C DIC would affect the application of this approach because it is only applicable when the 14 C values for end-members (seawater DIC, freshwater DIC, and C air ) are distinct (not overlapping) as they were in May and July 2014 during this study.We could not use the 14 C approach to quantify the C air contribution in September or November 2014 in Furen Lagoon because the 14 C DIC of seawater increased to near 14 C air and there was overlap between the two (Fig. 3).The overlapping in the range of values, induced by variations in the 14 C DIC of seawater, was likely caused by the dynamics of the Oyashio (mean 14 C DIC , −41 ‰; Aramaki et al., 2001Aramaki et al., , 2007) ) and the Soya warm current (mean 14 C DIC , 52 ‰; Aramaki et al., 2007;Kumamoto et al., 1998;Figs. 1 and 4).According to the dis- tribution of sea surface temperature derived from the Moderate Resolution Imaging Spectroradiometers (MODIS) images in 2014 (Fig. 4; http://oceancolor.gsfc.nasa.gov/cms/), the oceanic boundary of Furen Lagoon was the Oyashio throughout the year except from late summer to autumn when the Soya warm current intrudes into the boundary (Oguma et al., 2008;Takizawa, 1982).The oceanic endmember 14 C DIC would reflect the value of the Oyashio from January to August 2014, when the Soya warm current did not reach the oceanic boundary of Furen Lagoon (Fig. 4).The oceanic end-member 14 C DIC would, therefore, not overlap with 14 C air during January to August 2014, which includes the whole period of a 1-3-month seagrass leaf growing prior to sampling, i.e., February to July, indicating that the uptake of C air by the seagrass is a robust estimate during the period.Even if the sporadic upwelling had occurred during the study period, our determination of the C air contribution here would be underestimated because the 14 C DIC of the upwelling deep-sea water is lower than that of surface water (Aramaki et al., 2001(Aramaki et al., , 2007)).Nevertheless, the applicability of the 14 C technique is dependent on the 14 C dynamics of end-members.
Our 14 C analysis considerably reduces the limitations and uncertainties of conventional methods such as that using only δ 13 C (Clavier et al., 2011;Cooper and McRoy, 1988;Raven et al., 2002).In particular, the use of 14 C has the advantage of avoiding effects of isotopic fractionation (Stuiver and Polach, 1977); the use of δ 13 C does not and therefore generates large uncertainties.The δ 13 C of the seagrass was low (−14.0 ± 2.4 ‰) in the low-salinity zone (salinity, 12-15) and high (−8.8 ± 1.9 ‰) in the high-salinity zone (salinity, 25-29; Fig. 2c).There were significant correlations between salinity and δ 13 C of DIC, HCO − 3 , CO 2 (aq) and the seagrass (Pearson's correlation coefficient: P < 0.001; Fig. 2c,  d).As the δ 13 C of HCO − 3 was isotopically distinct from δ 13 C of both CO 2 (aq) and C air (Fig. 2d) and as Z. marina also uses HCO − 3 as a carbon source under low-CO 2 (aq) conditions (Beer and Rehnberg, 1997), the δ 13 C of the seagrass should change depending on the contribution of HCO − 3 as a carbon source (Campbell and Fourqurean, 2009;Hemminga and Mateo, 1996;Raven et al., 2002).However, it is not possible to distinguish the contribution of C air from that of other carbon sources because the δ 13 C of C air overlapped those of both HCO − 3 and CO 2 (aq) (Fig. 2d).Furthermore, δ 13 C of both HCO − 3 and CO 2 (aq) changes through mixing between low-δ 13 C river water and high-δ 13 C seawater in brackish areas (Fig. 2d; Hemminga and Mateo, 1996;Simenstad and Wissmar, 1985).
In any case, there are large uncertainties when using δ 13 C to quantitatively estimate the contribution of C air as a carbon source because the isotopic fractionation that occurs in the steps between the carbon source and organic plant compounds changes depending on the photosynthetic carbon demand (Hemminga and Mateo, 1996;Raven et al., 2002).The radiocarbon isotopic approach can avoid the uncertainties derived from both the chemical species of DIC and the isotopic fractionation factor in carbon assimilation.
The seagrass leaves assimilated C air when exposed to air during low tide (Fig. 5).CO 2 exchange between the air and water would occur at the very thin film of water on the airexposed seagrass leaves (Fig. 5c), likely enhancing the passive uptake of C air by diffusion.Our high estimate of the C air contribution (mean, 17 %) was unexpected because prior works suggest that photosynthetic rates of seagrasses in in-tertidal zones decrease during air exposure (Clavier et al., 2011), particularly in cases of desiccation (Leuschner et al., 1998).However, the leaves of subtidal seagrass are never desiccated because of the presence of the thin film of water, which reduces the negative effects of air exposure (i.e., desiccation).
The net ecosystem production of seagrass meadows is a key factor determining whether they are sinks or sources of C air (Maher and Eyre, 2012;Tokoro et al., 2014;Watanabe and Kuwae, 2015).Previously, however, such an exchange of CO 2 has been thought to occur only via the air-water interface with subsequent exchange with seagrasses as DIC.This study using radiocarbon isotope analysis demonstrates the assimilation of modern C air by seagrass.Moreover, our radiocarbon isotopic approach has potential for application to other photoautotrophs living near the air-water interface, such as intertidal macroalgae and amphibious macrophytes.Other applications may include determining the origin of the DIC source (e.g., terrestrial or oceanic) in deeper seagrass systems.However, adequate determinations will require separation and stability in the end-member values (e.g., in oceanographic contexts and in the dynamics of 14 C in coastal waters).The relative contribution of gas exchange via the air-seagrass water film to the total exchange is still unknown.To understand the role of seagrass meadows in the global carbon cycle, it will be necessary in future studies to precisely measure CO 2 exchanges at both the air-water and air-seagrass water-film interfaces.
Figure 2. (a) Spatial distribution of the 14 C values of dissolved inorganic carbon (DIC) (blue open circles) and seagrass (red open circles) along the salinity gradient in May and July 2014 in Furen Lagoon, Japan.Blue and red solid lines represent the linear models fitted with analyses of covariance (ANCOVA) examined for DIC and seagrass, respectively.(b) Spatial distribution of the relative contribution of C air to total inorganic carbon assimilated by seagrass along the salinity gradient, as calculated by the two-carbon-source mixing model.(c) Spatial distribution of the δ 13 C values of DIC (blue open circles) and seagrass (red open circles) along the salinity gradient.(d) Spatial distribution of the δ 13 C values of bicarbonate ion (HCO − 3 ; green open circles) and aqueous CO 2 [CO 2 (aq)] (purple open circles) along the salinity gradient.The dashed line indicates the isotopic signature of atmospheric CO 2 (C air ).

Figure 3 .
Figure 3. Spatial distribution of the 14 C values of dissolved inorganic carbon (DIC; blue) and seagrass (red) along the salinity gradient in September (open circles) and November (open triangles) 2014 in Furen Lagoon, Japan.The dashed line indicates the 14 C of atmospheric CO 2 ( 14 C air ).

Figure 4 .
Figure 4. Seasonal dynamics of ocean currents affecting the oceanic boundary of Furen Lagoon.Red circle shows the location of Furen Lagoon.(a) From mid-November to August, the oceanic boundary is the Oyashio ( 14 C DIC , −60 to −10 ‰; Aramaki et al., 2001).(b) From September to early November, the Soya warm current ( 14 C DIC , 50 to 70 ‰; Aramaki et al., 2007) intruded into the boundary.
Figure 5. (a) Distant and (b) close-up views of the seagrass leaves exposed to the air during low tide in Furen Lagoon, Japan.(c) Conceptual diagram of the uptake of atmospheric CO 2 (C air ) across the surface-water film on the seagrass leaves during low tide (left), and the uptake of DIC during high tide (right).