Sulfur isotope homogeneity of oceanic DMSP and DMS

Significance

Oceanic emissions of volatile dimethyl sulfide (DMS) represent the largest natural source of biogenic sulfur to the global atmosphere, where it mediates aerosol dynamics and may affect climate. Sulfur isotope ratios (³⁴S/³²S) offer a way to estimate oceanic DMS contribution to aerosols. We used a unique method for the analysis of ³⁴S/³²S of DMS and its precursor, dimethylsulfoniopropionate (DMSP), in a range of marine environments. Surface water collected from six different ocean provinces revealed a remarkable consistency in ³⁴S/³²S ratios of DMS and DMSP ranging between +18.9 and +20.3‰. The ³⁴S/³²S of oceanic DMS flux to the atmosphere is thus relatively constant and distinct from anthropogenic sources of atmospheric sulfate, thereby enabling estimation of the DMS contribution to aerosols.

Abstract

Oceanic emissions of volatile dimethyl sulfide (DMS) represent the largest natural source of biogenic sulfur to the global atmosphere, where it mediates aerosol dynamics. To constrain the contribution of oceanic DMS to aerosols we established the sulfur isotope ratios (³⁴S/³²S ratio, δ³⁴S) of DMS and its precursor, dimethylsulfoniopropionate (DMSP), in a range of marine environments. In view of the low oceanic concentrations of DMS/P, we applied a unique method for the analysis of δ³⁴S at the picomole level in individual compounds. Surface water DMSP collected from six different ocean provinces revealed a remarkable consistency in δ³⁴S values ranging between +18.9 and +20.3‰. Sulfur isotope composition of DMS analyzed in freshly collected seawater was similar to δ³⁴S of DMSP, showing that the in situ fractionation between these species is small (<+1‰). Based on volatilization experiments, emission of DMS to the atmosphere results in a relatively small fractionation (−0.5 ± 0.2‰) compared with the seawater DMS pool. Because δ³⁴S values of oceanic DMS closely reflect that of DMSP, we conclude that the homogenous δ³⁴S of DMSP at the ocean surface represents the δ³⁴S of DMS emitted to the atmosphere, within +1‰. The δ³⁴S of oceanic DMS flux to the atmosphere is thus relatively constant and distinct from anthropogenic sources of atmospheric sulfate, thereby enabling estimation of the DMS contribution to aerosols.

Dimethylsulfoniopropionate (DMSP), a metabolite of marine phytoplankton, is one of the major organosulfur compounds produced in the oceans (1). One of its degradation products, dimethylsulfide (DMS), is volatile and supersaturated in all marine surface waters. Large amounts of DMS (0.55–1.1 Tmol S⋅y⁻¹) are released from the ocean to the atmosphere (2), where it contributes to the formation and growth of aerosol particles (3). Over remote oceans, distant from anthropogenic sulfur inputs, DMS is the major source of nonsea-salt sulfate aerosol (4, 5). The oxidation of DMS to submicron sulfate aerosols was suggested to increase cloud-condensation nuclei and the albedo of clouds, leading to a potential climate feedback loop operating through the DMS-producing biota in the surface ocean (6). Although evidence for the direct, local climate feedback via DMS is limited (7), DMS likely contributes, together with organic and sea-spray aerosol sources, to the complex dynamics of climate-active aerosols in the lower atmosphere (3, 7).

Sulfur isotope measurements (³⁴S/³²S ratio, δ³⁴S) may offer a way to constrain the contribution of ocean-derived DMS to global sulfur cycling and aerosol budgets. Recently, Oduro et al. (8) used a multistep extraction of large seawater volumes (e.g., 50 L) coupled with Raney nickel dehydrosulfurization and subsequent fluorination for the S isotope analysis of DMSP in intertidal macroalgae (+17.3 to +19.3‰) and estuarine phytoplankton blooms (+19 to +20‰). However, there are still no direct DMS and DMSP δ³⁴S data in oligotrophic oceans owing to their typically low concentrations of 0.5–5 and 5–80 nM, respectively (9). We recently developed a sensitive method for δ³⁴S analysis in individual compounds at the picomole level, which couples gas chromatography (GC) to multicollector inductively coupled plasma mass spectrometry (MC-ICPMS) (10). Based on this method we further developed a simple and robust technique requiring only a few milliliters of seawater for DMS and DMSP isotopic analysis, using purge-and-trap of seawater and subsequent injection of DMS into a GC coupled to MC-ICPMS (11).

Here we describe the δ³⁴S values of DMSP obtained in five oligotrophic regions and one mesotrophic ocean region, encompassing a wide range of hydrological, meteorological, and biological conditions. Whereas DMSP can be reliably preserved for months (12, 13), DMS had to be analyzed within 24 h of sample collection to minimize storage artifacts. Taking advantage of our geographical proximity to the Red Sea and the Mediterranean, we analyzed the δ³⁴S of freshly collected DMS together with DMSP to evaluate the in situ isotopic fractionation resulting from their multiple transformations. We then examined the S isotopic fractionation during DMS volatilization in the laboratory. Taken together, our data reveal rather uniform δ³⁴S values of surface ocean DMSP and small S isotopic fractionation during DMSP degradation and DMS volatilization. These data provide a constraint on the marine contribution of the δ³⁴S value of DMS emitted to the atmosphere.

Results and Discussion

Homogeneity of DMSP Sulfur Isotopes in Seawater from Different Ocean Basins.

To evaluate the range of DMSP δ³⁴S values in seawater, we sampled six open ocean sites around the globe including Hawaii [St. HOT (Hawaii Ocean Time-series)], Bermuda (hydrostation S), East Mediterranean (Haifa), Eilat (northern Gulf of Aqaba, Red Sea), Gulf of Mexico (Mobile Bay–Alabama shelf), and the north Atlantic between Greenland and Iceland, hereafter called Greenland (Fig. 1A; see also SI Appendix, S1–S10 for site description and raw data). These sites were chosen to encompass variations in ocean basin (Atlantic, Pacific, and Indian), latitude (tropical, subtropical, and temperate), biogeochemical provinces (oligotrophic and mesotrophic), season, and phytoplankton composition. At most sites, we evaluated the sulfur isotope values of DMSP as a function of depth, and in Eilat we also examined seasonal variations in δ³⁴S values of DMSP (Fig. 1).

Fig. 1.

Sulfur isotope ratio (δ³⁴S) and concentration (nanomolar) of dimethylsulfoniopropionate (DMSP) in seawater from different ocean basins. (A) Location map and a legend of the sampling sites. (B) DMSP concentrations and δ³⁴S values of water collected throughout the photic zone in oligotrophic open ocean sites and during phytoplankton bloom events in Greenland and Eilat (C). The shaded area represents the range of δ³⁴S values obtained for surface water (≤5-m depth) from all locations.

Recent analytical developments facilitating a high throughput and enhanced sensitivity and precision (0.1–0.3‰ for >20 pmol S injected) (11) enabled us to successfully obtain an extensive dataset of oceanic DMSP δ³⁴S values (Fig. 1). Although the vertical distribution of S isotopes potentially holds many insights, we focus here on the surprising homogeneity in δ³⁴S values of surface ocean DMSP, best illustrated by plotting the isotopic composition versus DMSP concentrations (Fig. 1B). DMSP obtained from all sites, seasons, and depths had δ³⁴S values that fell within the narrow range of +17.8 to +20.5‰. Homogeneity was even more pronounced in near-surface seawater (0–5 m), with DMSP δ³⁴S ranging between +18.9 and +20.3‰ and averaging +19.7 ± 0.5‰ (shaded area, Fig. 1B). The observed consistency of δ³⁴S values is maintained at both low DMSP concentrations (<17 nM) measured in low-productivity waters and at high concentrations (up to 152 nM) measured during phytoplankton blooms (Fig. 1B).

High concentrations of DMSP and DMS are often observed in seawater during phytoplankton blooms, leading to high DMS fluxes to the atmosphere (14). There are no quantitative data about the contribution of phytoplankton blooms to the global DMS flux, but it is commonly believed that blooms are a major source (1, 2). Evaluating the sulfur isotope composition of DMSP/DMS during blooms is therefore of specific interest.

Two blooms were analyzed in this study, one from the North Atlantic near Greenland (July 2012) and another from the northern Red Sea near Eilat (May 2012). The prolific DMS(P) producer Emiliania huxleyi dominated the Greenland bloom and resulted in high DMSP concentrations (79–152 nM) and δ³⁴S between +19.5 and +22.1‰ (Fig. 1C). The Eilat bloom consisted of small eukaryotes (<8 µm) and Synechococcus spp. with maximum DMSP concentrations of 110 nM and δ³⁴S between +18.8 and +19.2‰. Despite some variations in δ³⁴S values of the Greenland bloom, δ³⁴S in surface samples (0–5 m) were similar to those obtained in other oligotrophic oceans (Fig. 1B, shaded area). It thus seems that phytoplankton blooms do not substantially alter the δ³⁴S of DMSP compared with nonbloom conditions.

To extend our observations to coastal environments, we conducted a surface transect in the northern Gulf of Mexico from the mouth of Mobile Bay to the edge of the continental shelf (December 2012; SI Appendix, S8). All offshore samples (water depth of 20–30 m) with salinity >34 g/kg had very similar δ³⁴S values, averaging +19.8‰ ± 0.4‰, in accord with the values obtained at the other oceanic stations (Fig. 1B). At sites with lower salinity inside the mouth of Mobile Bay, DMSP δ³⁴S was lower at +17.8‰ (Fig. 1B). Even lower δ³⁴S of DMSP (+11.3‰) was recorded in a salt marsh on Dauphin Island (SI Appendix, S8). These results demonstrate that there can be a larger variability of δ³⁴S in DMSP in estuarine and coastal waters, depending on location and local conditions, as has been shown previously for bulk S isotope analysis of different algae species (15, 16) and DMSP of estuarine environments (8). Nevertheless, from a global perspective, the δ³⁴S values of DMSP and its volatile breakdown product DMS are likely overwhelmingly dominated by the open ocean DMSP/DMS signature. Crude estimates suggest 20- to 30-folds higher emissions of reduced S gases (mainly DMS) in open oceans relative to coastal and wetland water (17).

Summarizing the δ³⁴S DMSP data, we have provided an extensive dataset of δ³⁴S values collected in different seasons from various ocean ecozones typified by varying phytoplankton composition and abundance, all yielding remarkably consistent δ³⁴S values of DMSP in surface seawater. These values of +19.7 ± 0.5‰ are only slightly depleted relative to δ³⁴S of seawater sulfate (+21.1‰).We next turn to examine possible S isotope fractionations in the conversion of DMSP to DMS, which is of interest for understanding the oceanic DMS(P) cycling and its emission to the atmosphere.

Sulfur Isotope Fractionation Between DMSP and DMS.

Sulfur isotope analysis of freshly collected (<24 h) surface water taken from the eastern Mediterranean (Haifa, April 2013) showed identical δ³⁴S values of DMS (+19.4 ± 0.1‰, n = 3) and DMSP (+19.3 ± 0.3‰, n = 10). Similarly, we observed small isotopic fractionation (+0.5‰) between DMSP and DMS in surface water from Eilat (11).

Depth profiles offer an additional perspective on this issue because the relative importance of the processes shaping δ³⁴S of DMS and DMSP such as photochemical oxidation, bacterial consumption, and volatilization change with depth (18–20). Thus, samples from different depths may reveal the overall in situ δ³⁴S fractionation between DMSP and DMS. Two depth profiles were conducted in Eilat during September 2012 and April 2013 (Fig. 2). The phytoplankton-sourced DMS and DMSP were indeed found mostly above the thermocline and their concentrations covaried (Fig. 2). The δ³⁴S values of DMSP and DMS ranged between +19 and +21‰, and DMS was consistently enriched in ³⁴S relative to its DMSP precursor by <+1‰ at specific depths, averaging at +0.6‰ throughout the water column (Fig. 2). Further examination of these profiles revealed interesting DMS/P dynamics and links to other measured parameters (SI Appendix, S3–S8) that are outside the scope of this paper.

Fig. 2.

Depth distribution of DMSP, DMS, light, and chlorophyll in the open water of Eilat, Red Sea. Depth profiles were collected in two seasons: late summer (September 2012) and spring (April 2013). (A) DMSP and DMS concentrations. (B) DMSP and DMS δ³⁴S values. (C) Chlorophyll a (green line) concentration and photosynthetically-active solar radiation (PAR, yellow line) between 400 and 700 nm.

Larger fractionation of opposite sign (−1 to −1.5‰) was reported by Oduro et al. (8) for laboratory incubations conducted in the dark where DMSP-containing macroalgae (+18.2 ± 0.6‰) released DMS (+16.6‰) (8). Our results from Eilat and the eastern Mediterranean suggest (i) a small (<+1‰) or no fractionation in the process of DMSP conversion to DMS in oceanic water or (ii) DMS removal processes in the mixed layer (photolysis, biological consumption, and ventilation), which turn over the mixed-layer DMS pool every 1–3 d (19) may re-enrich the DMS pool to values close to those of DMSP pool.

Whatever the reason is, the fact that DMSP and DMS δ³⁴S values are so similar suggests that DMSP can be used as a proxy for dissolved DMS in seawater. DMSP analysis is easier and it can be preserved for at least several months (12) provided colonial Phaeocystis spp. are not present (13), thereby enabling reliable seawater analysis even from remote locations where δ³⁴S analysis within 24 h is impossible.

Laboratory Experiments of S Isotopic Fractionation During Water–Air Exchange of DMS.

Examination of δ³⁴S of DMSP and DMS in depth profiles from Eilat (Fig. 2) reveal no distinct fractionation at the surface where DMS volatilization and emission to the atmosphere takes place. To strengthen our field observations we performed a laboratory study of the change in δ³⁴S during the volatilization of synthetic DMS with δ³⁴S of −3.0 ± 0.3‰ (Fig. 3, shaded area). Two experiments were performed with different initial DMS concentrations (26.6 nM and 100 nM) and different evaporation rates. Both experiments resulted in similar small fractionation factors following the Rayleigh distillation equation (Eq. 3, Methods). The combined results of both experiments yield an observable but small fractionation factor of −0.49‰ and R² of 0.71 (Fig. 3). Taking into consideration the overall experimental error (resulting mostly from 0.1 to 0.4‰ error in δ³⁴S values, SI Appendix, S11), the observed fractionation factor ranges between −0.28 to −0.69‰ or ∼−0.5 ± 0.2‰ based on the 95% confidence level. In the ocean, where only a fraction of the DMS is emitted to the atmosphere (1–30%), we expect this small δ³⁴S fractionation, in accord with previous estimations (21). Our results thus suggest that the δ³⁴S of DMS emitted from the sea surface is similar to the δ³⁴S of the seawater DMS pool, which in turn is isotopically similar to that of the DMSP pool (Fig. 4).

Fig. 3.

The effect of volatilization on the δ³⁴S of remaining aqueous DMS. The extent of volatilization was expressed as the natural log of the fraction of remaining DMS in solution [Ln(f)]. Initial concentrations of reagent DMS in artificial seawater were 100 nM (experiment A) and 26.6 nM (experiment B) and the initial δ³⁴S was −3.0 ± 0.3 ‰ in both experiments (shaded area). Error bars for experiment B represent the SD (1σ) between duplicates. The slope of the linear regression line is the fractionation factor following the Rayleigh distillation equation (Eq. 3).

Fig. 4.

Schematic representation of the predicted isotope fractionation associated with the conversion of DMSP_(aq) to DMS_(aq) in the surface ocean and the volatilization of DMS_(aq) to the atmosphere as DMS_(g).

In another experiment we examined the effect of biological processes on δ³⁴S of DMS, by incubating two fresh seawater samples from Eilat for 72 h in dark vials (preventing photochemistry) and without headspace (preventing volatilization). Throughout the incubation we observed minor isotopic changes of <+0.6 ‰, despite significant increases in DMS concentration (SI Appendix, S12). The increase in DMS with time likely reflects the breakdown of DMSP (22), and the lack of isotopic fractionation in this step further supports the homogeneity in S isotopes of DMS and DMSP in seawater.

Implications for the Isotope Mass-Balance Calculations of Sulfate Aerosols.

The GC-MC-ICPMS system has proved to be an efficient, sensitive, and precise approach for the analysis of trace amounts of DMS and DMSP from oligotrophic waters. In all locations (Eilat, Hawaii, Bermuda, Mediterranean, Greenland, and Gulf of Mexico) and all depths in the upper 120 m, δ³⁴S values of DMSP and DMS ranged between +17.8 to +20.5‰. A narrower range of +18.9 to +20.3‰ was found for surface (≤5m) waters. Only small differences in δ³⁴S (<+1‰) between DMSP and DMS, were observed in situ and during volatilization experiments of DMS. This suggests that the δ³⁴S of the DMS flux to the atmosphere is relatively constant from the open ocean. The uniform δ³⁴S of seawater DMS and DMSP should facilitate isotopic mass balance calculations to estimate their contribution to sulfate aerosols in the atmosphere, which show highly varied δ³⁴S values with relation to their sulfur sources (21, 23–30).

This isotopic approach assumes that the sources of sulfate aerosols (termed end members) have distinct δ³⁴S values from one another. Three main end members for sulfate aerosols are present in remote marine locations, assuming that volcanic gases—excluding those created in big eruptions—are localized events. They are anthropogenic sources, sea-salt sulfate, and sulfate originating from marine biogenic activity (mostly DMS). Sea-salt sulfate (+21.1‰) has a similar δ³⁴S value to DMS/DMSP (+18.9 to +20.3‰); therefore, nonisotope approaches such as size separation and sea-salt correction by means of Na⁺ measurements are needed to distinguish between these end members. Sea-salt sulfate usually forms relatively large-diameter aerosols (>3μm) and is less abundant in smaller particles. In contrast, nonsea-salt sulfate (i.e., DMS and anthropogenic) concentrates in smaller particles, mostly in the submicron diameters, and can thus be separated based on size (31). The fractional contribution of the remaining sea-salt sulfate to the total sulfate aerosol can be calculated based on Na⁺ measurements in the aerosol and applying the mass ratio of SO₄^2– to Na⁺ of 0.252 in bulk seawater (25). After accounting for the sea-salt sulfate in the mass balance calculations, two end members are left: oceanic DMS and anthropogenic sources. The δ³⁴S of anthropogenic sources is site-dependent but it usually converges to the range of 0 to +10‰ (25). This range is distinct from that of oceanic DMS; therefore, isotope mass balance calculations should allow estimation of the biogenic S contribution to sulfate aerosols. One caveat to the above statement concerns the possible contribution of marine cell components (e.g., protein) blown into sea spray to sulfate aerosols. Protein δ³⁴S values are likely similar to that of DMS/DMSP (32). The contribution of protein S to sulfate aerosols is currently unknown but is assumed to be small given current estimates of total particulate organic carbon emission from the ocean (33).

In summary the remarkably uniform δ³⁴S values of surface seawater DMS and its precursor DMSP found in this study is an improvement over previous estimations of DMS that ranged between +12.5 and +22‰ (21, 29). This in turn will decrease uncertainties in calculations of the sources of sulfate to the atmosphere when using the S isotope approach.

Methods

Seawater Sampling.

Detailed descriptions of the sampling sites are provided in SI Appendix. Seawater samples were collected from the open waters of the Northern Gulf of Aqaba, Eilat, Israel (station A, September 2012 to May 2013; location map in SI Appendix, S1); Mediterranean, Haifa, northern Israel (April 2013); from station A off Hawaii as part of the Hawaii Ocean Time-series (HOT) cruise (September 2012); from hydrostation S off Bermuda (July 2012); Alabama Shelf off Dauphin Island (December 2012); and off Greenland (July 2012). For the Eilat samples, seawater was collected in Niskin bottles and transferred immediately to 40-mL amber glass vials with polytetrafluoroethylene septa in triplicate and filled to the top (no head space). Two of these vials were prefilled with 0.5 mL of concentrated (37%) HCl for long-term (months) preservation of DMSP_t (13). The other vial remained untreated and was used for dissolved DMS analysis. The vials were kept at the temperature of surface seawater (∼25 °C) until processing within 24 h of their collection. DMSP_t samples from other locations were collected similarly by filling polypropylene centrifuge tubes with unfiltered seawater directly from Niskin sample bottles and acidifying to preserve the DMSP. Samples from Hawaii, Alabama, and Bermuda were shipped by air express to Hebrew University for analysis.

Quantification of DMS and DMSP.

DMS was analyzed using a Teledyne Tekmar (Atomx) purge-and-trap sample concentrator with autosampler coupled to a Perkin-Elmer 580 GC equipped with flame photometric detector [described in detail in SI Appendix and in Said-Ahmad and Amrani (11)].

δ³⁴S Analysis by GC-ICPMS.

The S isotope results are expressed in conventional δ³⁴S notation as a per mil (‰) deviation from the international standard Vienna Canyon Diablo Troilite according to Eq. 1:

where ³⁴R is the integrated ³⁴S/³²S ion-current ratio of the sample and standard peaks. When the ratio between the heavy isotope (³⁴S) and the light isotope (³²S) differed between the initial (A) and the final (B) stages of reaction or process this is termed isotopic fractionation and is expressed as

The S isotope fractionation factor (ε) of DMS evaporation was calculated from the residual DMS in solution according to the Rayleigh distillation equation:

where δ_initial is the δ³⁴S value of the intial DMS and δ_residual is the δ³⁴S of the residual DMS in solution and f is the residual fraction of DMS in solution.

Details of the S isotope analysis of DMS and DMSP are presented elsewhere (11). Briefly, samples were purged and trapped and subsequently injected into the GC for separation to individual compounds. Sulfur isotopic composition of DMS was measured by a coupled multicollector ICPMS (Neptune Plus; Thermo). Seawater samples were removed from the original amber vial using a syringe (usually 5 mL) with minimal disturbance and injected gently into a new 40-mL sparging vial equipped with a Teflon septum. The vial was then sparged with He (99.995%) for 12 min at 45 mL/min. Water vapor was removed by a Nafion-membrane dryer (Perma Pure LLC) using dry N₂ or He as the counter flow. A Teflon sample loop in the flow path after the dryer was inserted in a Dewar of liquid N₂ to trap DMS. After sparging, the six-port valve (Valco Instrument Co.) was turned to the inject position and the sample loop transferred quickly from the liquid N₂ to hot water to introduce trapped gases into a Perkin–Elmer Velocity-1 capillary column (30 m × 0.32 mm i.d. × 1.0 μm), connected directly to the six-port valve). At the same time the GC (Clarus 580, Perkin-Elmer) and the MC-ICPMS were started (operating conditions for the GC and MC-ICPMS are shown in SI Appendix, S2). The DMS peak eluted at 2.4 min at 40 °C isothermal with 1.5 mL/min He carrier flow.

After separation by the GC, analytes were transferred to the MC-ICPMS via a heated (200 °C) transfer line. The sulfur-isotopic composition of individual GC peaks were measured by multicollector ICPMS (Neptune Plus; Thermo). A standard DMS sample with known δ³⁴S value was purged and trapped for calibration every four or five seawater samples and a bracketing technique was used to correct for instrumental mass bias (11).