Shipboard measurements of atmospheric dimethylsulfide and hydrogen sulfide in the Caribbean and Gulf of Mexico

. Simultaneous shipboard measurements of atmospheric dimethylsulfide and hydrogen sulfide were made on three cruises in the Gulf of Mexico and the Caribbean. The cruise tracks include both oligotrophic and coastal waters and the air masses sampled include both remote marine air and air masses heavily influenced by terrestrial or coastal inputs. Using samples from two north- south Caribbean transects which are thought to represent remote subtropical Atlantic air, mean concentrations of DMS and Hz S were found to be 57 pptv (74 ng S rne3, CJ = 29 ppW, n = 48) and 8.5 pptv (11 ng S me3, c = 5.3 pptv, n = 36), respectively. The ranges of measured concentrations for all samples were O-800 pptv DMS and O-260 pptv H*S. Elevated concentrations were found in coastal regions and over some shallow waters. Statistical analysis reveals slight nighttime maxima in the concentrations of both DMS and HzS in the remote marine atmosphere. The diurnal nature of the Hz S data is only apparent after correcting the measurements for interference due to carbonyl sulfide. Calculations using the measured ratio of HIS to DMS in remote marine air suggest that the oxidation of HIS contributes only about 11% to the excess (non-seasalt) sulfate in the marine boundary layer.


Introduction
Oceanic emissions of reduced sulfur compounds constitute a major flux of sulfur to the atmosphere on a global basis. The atmospheric oxidation of these compounds may account for a significant portion of the sulfur dioxide (SOz) and sulfate observed in marine air, particularly in highly productive and remote regions of the oceans. Dimethylsulfide (DMS) is known to be the most abundant reduced sulfur gas in surface seawater and its distribution has been studied in many regions of the oceans (Andreae et al., 1986;Bates et aZ., 1987;Turner et al., 1987). Observations of DMS in the marine troposphere support the concept that DMS emission from the sea surface constitutes a major global flux of sulfur to the atmosphere (Andreae et al., 1985). Model calculations indicate that the magnitude of this flux may be sufficient to account for all of the sulfate in marine air; however, there are large errors associated with such mass balance calculations and other sulfur species may be important.
Hydrogen sulfide (HzS) emissions from shallow coastal waters, exposed coastal muds and salt marshes have been extensively documented, (Jorgensen and Okholm-Hansen, 1985;de Mello et al., 1987; and references therein), but have not generally been assumed to be a significant oceanic source for two reasons. First, H$ is thought to be produced solely as a result of anaerobic metabolism in anoxic sediments and, therefore, is unlikely to have a sign&ant source in the open ocean. Second, until recently, it was thought to be rapidly oxidized in oxygenated waters (e.g. Ostlund and Alexander, 1963). However, several factors argue for the possibility that H$ could be an important sulfur species in the oceans. These include a recent reexamination of the' oxidation rate of H$j supporting a half life of 50 * 16 hours at 25 " C (Miller0 et al., 1987), and calculations of the rate of hydrolysis of OCS which indicate a potential though quantitatively minor abiotic source (Elliot et af., 1987). The most direct evidence published to date is based on the observations of H# in the marine boundary layer, at levels which imply that HzS must be present in significant concentrations in the surface ocean, and constitute an important sea/air flux of sulfur relative to DMS. Several investigators have observed H# in oceanic air. Slatt et al. (1978) measured concentrations ranging from 5 to 50 pptv (7 to 65 ng mM3) at Barbados and Sal Island in the North Atlantic. Delmas and Servant (1982) found a mean concentration of 14 pptv (18 ng mm3) in the Gulf of Guinea over productive waters of the west African coastal upwelling area. Herrmann and Jaeschke (1984) reported concentrations ranging from 11 to 118 pptv (15 to 154 ng mm3) over the Atlantic Ocean.
Recently, Cutter and Oatts (1987) have measured nanomolar concentrations of sulfide in surface seawater. However, the speciation and, hence, volatility of the sulfide has not yet been established. The observations of HzS in marine air and waters raise interesting questions regarding its origin and relative importance compared to DMS as a precursor of atmospheric SO2 and non-seasalt sulfate over the open ocean.
In this study we report simultaneous measurements of DMS and HZS in marine air from three cruises in the Gulf of Mexico and the Caribbean Sea. Cruise tracks are shown in Figure 1. The first of these cruises, to the Tongue of the Ocean, Bahamas, took place during June of 1985 aboard the R/V Cape FZorida. The second cruise was to the Gulf of Mexico, also aboard the Cape Fiorida, during October 1985. These two cruises were shakedown cruises which resulted in the acquisition of relatively limited data sets. The locations and meteorological conditions were such that they are not representative of open ocean conditions. However the results are of some interest with regard to coastal regions and will be briefly discussed in this paper. The third cruise was from Miami to the Cariaco Trench, off the coast of Venezuela, aboard the R/V Columbus Iselin during February and March of 1986. On this cruise a fairly extensive set of measurements was obtained along a north-south transect of the eastern Caribbean Sea, with prevailing winds such that oceanic air masses were sampled much of the time.  Mexico, 9-22 September 1985. (C)  All atmospheric samples were collected on the flying bridge of the ships, forward of potential contamination from stack and sewage emissions. Samples were collected only when the relative wind aboard ship was within 90 degrees of the bow. With only a few exceptions, the samples were collected while the ship was underway. In many cases the opportunity for sampling was limited, as these cruises were primarily run for oceanographic purposes which required long periods of sampling on station.

Dirnethylsu&de
Samples were collected for analysis of DMS by drawing air through cryotraps immersed in liquid oxygen (-183 C). The cryotraps were constructed from Teflon PFA tubing (l/4" OD, 5/32" ID, 16" long) and packed loosely with Teflon wool. The air flow through the trap was controlled and measured using a mass flow controller which was modified to operate in the integrating mode. The flow rate during sampling was initially 500 cm3 min-'; but decreased as the trap became blocked with ice. Typical air volumes used in this study ranged from 4 to 10 liters depending on the atmospheric DMS concentration. At the end of the sampling period, the level of the liquid oxygen was raised, and the cryotrap was capped with stainless steel end fittings for transfer to the shipboard lab for analysis. During early tests it was observed that severe losses of DMS and other sulfur compounds occurred in the cryotraps, presumably as a result of the cotrapping of atmospheric oxidants or oxidant precursors such as O3 or NO.r. This problem was overcome by using Na&03 prefilters as described by Ammons (1980). Both Ammons (1980) and Andreae et aI. (1985) have noted a similar difficulty with losses of sulfur compounds trapped on gold foil or gold coated sand. While effective, the prefilters have a relatively limited capacity for oxidant removal. On the Bahamas and Gulf of Mexico cruises we used to prehlters (47 mm diameter) in series in Teflon PFA filter housings, Each pair of prefilters was used for no more than 12 liters of air. On the Cariaco Trench cruise (Caribbean Transect) we used glass tubes packed with Na$Os coated Chromosorb W/AW (Supelco Inc., Bellefonte, PA) as oxidant scrubbers because of their high capacity compared to the glass fiber filters. Each tube was used for approximately 30 liters of air. These tubes are identical to those used by Ammons (1980). The scrubbers used in the present study were shown to prevent losses of preconcentrated DMS during sampling of relatively unpolluted marine air. Improved scrubbers have been used in subsequent studies (Cooper and Saltzman, 1988) and will be described in more detail in the future. DMS samples were analysed by gas chromatography, with flame photometric detection. The column used was 12' x l/8" Chromosil 330 packed in Teflon (Supelco, Inc. Bellefonte, PA). After sampling, the cryotraps were connected to the gas chromatographic system via a lo-port hastalloy valve (Valco Inc.), purged with dry nitrogen to remove residual oxygen, and the sample injected onto the column by immersing the loop in hot water. The carrier gas was 30 cm3 mine1 nitrogen, and the column was temperature programmed for an initial hold of two minutes at 25 *C followed by heating to 100 C at 25 *C min-' . The analytical system had a detection limit of about 5 x lo-'* moles of sulfur injected. This corresponds to a detection limit of 12 pptv DMS (16 ng S me3) for a 10 liter air sample.
Calibration of the DMS chromatographic system was performed by injecting lo-200 ul volumes of a liquid standard into a 5 ml glass bubbler containing 1 ml of water. The standard was purged with nitrogen onto a cryotrap and analyzed in the same manner as atmospheric samples. A 60 ppm stock solution of DMS was prepared by diluting the pure liquid in methanol. A dilute working standard was prepared daily by dilution (1 : 1000) of the stock in distilled water. The calibration curves generated during this work were reproducible both within and between cruises. During the first few days of each cruise the detector underwent a gradual increase in sensitivity, then remained constant for the remainder of the cruise. The cause of this phenomenon is not known, but may be related to the loading of the column with high boiling sulfur compounds which slightly raise the background level of sulfur in the detector as they elute from the column.
During earlier laboratory and field studies we had obtained good agreement between low rate permeation tubes (GC Industries Inc., Chatsworth, CA), high permeation rate tubes (Kin-Tek Laboratories, Inc., Texas City, TX) and the liquid standards. Our field system was originally designed for use with the low rate, low temperature coefficient permeation tubes. However, during the course of the Bahamas cruise these tubes underwent significant changes in permeation rate and were abandoned in favor of the more reliable liquid standards.
The analytical procedure described above is also suitable for carbon disulfide and dimethyl disulfide. We have made numerous measurements of these compounds in terrestrial wetland environments using the same techniques (Cooper et ul., 1987;de Mello et ul., 1987). During the present study the system was calibrated for these gases as well as DMS, although they were rarely observed. Particular care was taken to assure that the system was free of CSz contamination, as one of our goals was to put at least an upper limit on its abundance in marine air. Standard additions of these compounds to air samples in the field demonstrated that the detection limit of the system was lower than 5 pptv in a 10 liter air sample. The lower detection limit compared to DMS results from the fact that CSl and DMDS have two atoms of sulfur per molecule and from the non-linear response of the FPD detector, which varies roughly as the square of the sulfur atom concentration.
Other potentially important sulfur compounds such as HxS, OCS, and mercaptans are not detectable at low levels with this system because of adsorptive losses, and because of quenching of the detector signal due to co-elution of the sulfur compounds with early eluting gases such as CO1 and/or hydrocarbons. Typical chromatograms of a mixed standard and an atmospheric sample are shown in Figure 2.

Hydrogen Suede
Hydrogen sulfide was collected and analysed using a method similar to that of Natusch et al. (1972). An air stream was drawn through a silver nitrate impregnated (Whatman 41) filter in a 47 mm Teflon PFA filter holder. Samples were collected at flow rates of 8 liter mine1 for times ranging from 30-100 min depending on the ambient H$ concentration.
Sulfide was recovered from the silver nitrate filters by washing in 0.1 M NaOH/NaCN solution to a total volume of 20 ml and analysed by the fluorescence quenching of dilute fluorescein mercuric acetate (20 nmol added per sample). A Turner Designs model lo-853 fluorometer was used with a blue lamp and narrow (10 nm) bandpass filters at 500 nm excitation and 520 nm emission wavelengths. The cuvettes were 30 ml capacity and the instrument was modified to allow rapid mixing of the solution in the cell using a magnetic stirrer.
Calibration was performed by standard addition of fresh sodium sulfide solution to blank filters. A 1 mM stock solution is made in distilled water containing 0.1 N NaOH and 0.1 mM EDTA. The stock solution is diluted (1 : 50) with 0.1 N NaOH/NaCN to make the working standard. The glassware used for preparation of the sulfide standards is conditioned and stored containing sulfide solution. In our experience, the sulfide stock remains intact for a period not exceeding one week. The routine calibration procedure serves primarily to verify that the fluorometer has retained its alignment and sensitivity. We have obtained virtually identical calibration curves over a period of several years. This is probably due to the fact that the sulfide concentration is a function of fluorescence decrease. Since the fluorescence decrease is measured as a percentage of fluorometer full scale, slight changes in the absolute response of the instrument do not affect the calibration. On the Cariaco Trench cruise, we were able to make fresh standards during the cruise using pre-weighed NaZS crystals, which were stored under nitrogen in sealed glass ampules. The results verified the calibration made at the start of the cruise using the initial sulfide stock solution.
The lower limit of detection of the method is limited by the variability of filter blanks. Individual batches of filters had blanks ranging from O-200 pmol S'-per filter (Cooper, 1986), but variability within each batch was never greater than about 40 pmol. This gave a practical detection limit of about 10-r' moles HIS per filter which is equivalent to 3 pptv for a 100 min sample.
Contamination of the filters and the analytical glassware was a major problem on the Bahamas cruise, primarily because of a leak in the ships plumbing which vented some H$ into the ships lab late in the cruise. For the subsequent cruises the sulfur analyses were relocated into a van which was bolted onto the deck of the ship, and was thus removed from the plumbing system. We also modified the sample preparation procedures. To minimize sample handling, the filters were extracted in seaIed scintillation vials, rather than a Buchner funnel that had been used previously.
The NaOH/NaCN washing solution was stored in and delivered from a glass repipet. No further contamination problems were encountered during the laboratory procedures as demonstrated by regular analysis of blank lilters.
Laboratory and field tests have demonstrated that a single silver nitrate filter is >98Oh efficient for HIS removal under the sampling conditions used for terrestrial environments (Cooper, 1984). However, aboard ship, where longer sampling times were necessary to detect background levels of HZS, we consistantly observed a signal on back-up filters. This artifact sulfide has been attributed to the retention and subsequent hydrolysis of carbonyl sulfide on the basis of subsequent laboratory experiments. A temperature dependance of the artifact generation is evidenced by the diurnal nature of the signal found on the back-up filters . The total concentration of sulfide measured on the front filter is assumed to include an artifact equivalent to that on the back-up filter. The measured sulfide is corrected by subtracting the sulfide content of the back-up filter, as discussed in our earlier paper.
Back-up filters were only used during the Cariaco Trench cruise, where the mean sulfide concentration of these back-ups was found to be 4.9 pptv (6.4 ng S me3) and to be independent of that on the front filters. Hydrogen sulfide data obtained on the Bahamas and Gulf of Mexico cruises, where only single stage filters were used, are probably overestimated by a similar amount.
The question of contamination of the air by H$ released by the ship is a critical one which is difhcult to dismiss out of hand. We have attempted to minimize this problem by sampling only when the ship was underway, and by monitoring the relative wind speed and direction to assure headwinds during sampling. Perhaps the most convincing evidence that the measurements represent the composition of marine air is the data itself With the exception of very few samples, large atmospheric H$S concentrations were normally only observed downwind of shallow waters or exposed mudflats. When sampling oceanic trajectories the HzS concentrations invariably decreased markedly. No differences in H$ levels were observed between samples collected on the bow of the ship and those collected on the flying bridge. There is no evidence to suggest that the measurements do not reflect the composition of the ambient marine air sampled.

Bahamas Cruise
The Bahamas cruise track was from Miami to the Tongue of the Ocean, an elongate, deep basin which intrudes into the Bahamas platform (Figure 1, inset). Wind directions during the cruise ranged from southeasterly to southwesterly (120 to 250 degrees) during the first four days, after which southeasterly winds predominated. The predominant trajectory appeared to be down the axis of the Tongue of the Ocean, however, several shifts in wind direction brought air which had passed over Andros lsland or portions of the Bahama Banks.
The concentrations of DMS and HzS ranged from 25-300 pptv (33-392 ng S me3) and lo-260 pptv (13-340 ng S mw3) respectively on this cruise (Figure 3). The two large peaks in H$S on days 2 and 4 coincided with slight changes in the wind direction which shifted from along the length of the Tongue of the Ocean to trajectories skirting Andros (day 3) and the northeastern Great Bahama Bank (day 5). A similar effect appears in the DMS data, with the peak concentration on day 6 accompanying a brief change from southeasterly winds to almost northerly winds, from the northeastern Bahama Bank. Interestingly, the changes in the DMS and H$ concentrations on days 5 and 6 did not occur simultan- eously, suggesting that their sources are geographically or temporally different. This is in agreement with results from numerous emissions studies of various coastal environments (Jorgensen and Okholm-Hansen, 1985;de Mello et al., 1987; and references therein).
The data from this cruise is, therefore, not representative of 'background' marine conditions. It does, however, clearly demonstrate the significance of island effects on the ambient reduced sulfur levels in this region. On the last night of the cruise, measurements were made in the Gulf Stream with southerly winds. These winds should have brought air from south of the Bahamas platform, and, hence, provided an environment more representative of true marine conditions. Both DMS and H$ dropped to values similar to those measured in more remote locations.

Gulp-of Mexico Cruise
During the first portion of this cruise, data was obtained only for DMS. Concentrations ranging from 25-70 pptv (33-91 ng S mm3, Figure 3b) were measured as the ship steamed from Miami southwest offshore of the Florida keys and then northwest along the western shelf of Florida in the Gulf of Mexico. On the shelf, DMS concentrations were low, with more than half the samples below the detection limit, and a range from 5-25 pptv (7-33 ng S mm3) in the remainder. The low concentrations probably reflect the influence of continental air. Aerosol data collected simultaneously on this cruise show a threefold increase in nitrate and a tenfold increase in non-seasalt sulfate over remote oceanic values during this time. Weather patterns during the period indicate that the pollutant sources of oxidized sulfur are probably the industrialized areas on the west coast of Florida. The influence of continental air has been previously shown to decrease the measured DMS concentrations. This may be a result of increased oxidant generation in the polluted air (Andreae el uZ., 1985), and/or failure of the oxidant scrubbers under these conditions (Cooper and Saltzman, 1988).
Atmospheric sampling was interrupted during the middle portion of this cruise while the ship was on station for water column studies and because of following winds and severe weather. Towards the end of the cruise several days were spent at anchor off the Dry Tortugas. Atmospheric sampling was possible because of steady southeasterly winds. DMS concentrations were very low, often below the limit of detection (indicated by the zero concentrations in Figure 3b). In order to verify that the analytical technique was functioning properly, some measurements were made by difference, using standard additions of DMS to air samples. The results indicate that the analysis was functioning properly, and that DMS concentrations were as low as 3 pptv. The cause of these low values is not known, but slightly elevated aerosol nitrate and non-seasalt sulfate values indicate that a continental component to the sampled air may still have been present. H$S concentrations were consistently greater then those of DMS at this station, ranging from 30-90 pptv (39-li8 ng S mm3). The source of this H$S was presumably the shallow muds and tidal flats of nearby Florida Bay and the Florida Keys. H$ concentrations of up to 24 ppb (31 pg me3) have been measured in a mangrove habitat in this region (Cooper, 1986).

Cariaco Trench Cruise (Caribbean Transect)
The Cariaco Trench cruise consisted of two transects obtained while the ship was underway from Miami to the coast of Venezuela and back to Miami (Figure 1, track C). No atmospheric measurements were made during the two week period in which the ship was on station at the Trench. East to northeasterly winds predominated for the duration of the cruise. Data from this cruise (Figure 3c) are therefore representative of true marine background conditions, except for three notable periods. The concentration of DMS was markedly higher during passage over shallow waters at the still of the Cariaco Basin (days 5-6 and 23) and also during transit near the Turks and Caicos Islands (day 26). Hydrogen sulfide concentrations were not noticeably affected at these times, but several small 'spikes' in the H$S data may be a result of ship contamination at times of low relative wind speed.
DMS concentrations were found to be consistently higher than H$j during this cruise, with ranges of 16-104 pptv (2 l-136 ng S ms3) and O-20 pptv (O-26 ng S mm3), respectively, excluding the anomalously high values mentioned above. CS? was above the detection limit (approximately 1.5 pptv) in only a few samples, taken on day 28 over shallow waters in the vicinity of the Bahamas. The concentration range of 1.5-2.6 pptv (mean 2.2 pptv, 5.7 ng S mm3) is in good agreement with the results of Kim and Andreae (1987) and our own subsequent data (Cooper and Saltzman, 1988).

I. Processes Regulating Atmospheric Dh4.S' and H2S Concentrations
Marine air representing three distinct sets of boundary layer conditions has been sampled in this study. These are (1) coastal regions and shallow waters, (2) continentally influenced marine air, and (3) background marine air. Each of these environments may have different factors controlling the ambient reduced sulfur gas concentrations.
The highest levels of DMS and H$S were measured over shallow and coastal waters, but did not occur simultaneously. Concentrations of up to 300 pptv (392 ng S mm3) DMS and 260 pptv (340 ng S mm3) H$ measured on the Bahamas cruise can be attributed to the proximity of productive shelf waters. Extremely high DMS concentrations over shallow waters on the Cariaco Trench Cruise (up to 800 pptv, 1045 ng S me3) were measured simultaneously with background H$ concentrations (4-l 1 pptv, 6-14 ng S me3). The difference in spatial and temporal emissions of DMS and H$S probably reflects their different biological production processes. DMS appears to be a product of healthy aerobic algal metabolism and zooplankton grazing (Dacey and Wakeham, 1986). In contrast, HZS is formed primarily in anoxic environments, via either bacterial sulfate reduction or degradation of organic matter.
The lowest concentrations of DMS were measured in continentally influenced air. This trend is most evident during the Gulf of Mexico cruise when 19 samples were found to be below the limit of detection, Similarly, Andreae et ul. (1985) in the Bahamas and North Atlantic, found their lowest DMS concentrations during the passage of continental air associated with cold fronts. They attributed the low DMS levels to its rapid removal because of the enhanced concentration of oxidants in the continental air mass. It should be noted that sampling losses may also contribute to an apparent decrease in the DMS concentrations .
It is not clear from our data whether the destruction rate of atmospheric H$S is enhanced because of the increased oxidant levels. The only H$j measurements made where the influence of continental air was evident showed the reverse effect, with a much higher concentrations of H$S (28-77 pptv, 37-101 ng S mm3). However, these high levels probably reflect the presence of H$ from the coastal areas discussed in the previous section.
Remote marine conditions were only encountered during the Cariaco Trench cruise (Caribbean Transect) and are most apparent from the consistently low H$ concentrations. Although sharp peaks in DMS concentration were found during passage over shallow waters, no DMS samples were below the limit of detection.

Background Levels of DMS and H?S in Marine Air
Results from the statistical analysis of all the shipboard data are presented in Table I. The concentrations shown as representing 'background' conditions are calculated from the DMS and HIS data sets of the Caribbean Transect, after excluding all data points which fell outside the range of the mean &one standard deviation. These modified means are in excellent agreement with the median values of the unmodified DMS and H$S data sets. Both statistical treatments eliminate bias in averaging due to (1) the extremely high DMS concentrations measured over shallow waters, and (2) spikes in the H$S data, which may be the result of ship contamination. Average background concentrations of 57 pptv (74 ng S mm3) DMS and 8.5 pptv (11 ng S me3) H$S are calculated using both of these statistical procedures.
Our DMS data are in reasonable agreement with previous reports. In extensive studies of the marine atmosphere, Andreae el uZ. (1985) reported a mean DMS concentration of 107 pptv (140 ng S mm3), slightly higher than our mean for all cruises of 86 pptv (112 ng S me3). However, our median estimate of 57 pptv (74 ng S mm3) for the 'background' DMS concentration is significantly lower than their mean value. The difference may be due, at least in part, to the fact that many of their measurements were made over biologically productive waters and during the season of maximum phytolankton productivity.
Our estimate of 8.5 pptv (11 ng S mm3) 'background' HzS concentration falls within the range reported by Slatt et ul. (1978) for remote environments (medians 2-10 pptv). Delmas and Servant (1982) measured a mean concentration of 15 pptv (20 ng S mW3) over the Gulf of Guinea; this value is identical to our unadjusted mean. The most extensive data set previously reported for H$S in the marine atmosphere is that of Herrmann and Jaeschke (1984). They reported a mean concentration of 3 1 pptv (40 ng H$ mm3) from a data set of 39 samples judged to be representative of background conditions over the Atlantic Ocean. However, the detection limit of their method was 1 I pptv (15 ng S mm3), and they applied this value to all of the samples which had undetectable levels. The greater sensitivity of our H$ method allowed the measurement of H$S concentrations as low as 3 pptv. The majority of the samples considered representative of background conditions in our study had concentrations lower than the detection limit quoted by Herrmann and Jaeschke in their study.
Preliminary analysis of our HzS data revealed that OCS interferes in the determination of low concentrations . The background data presented here have been corrected for this effect. It is impossible to assess the magnitude of this artifact in the earlier studies because of its temperature dependence, but it is probable that the previously reported concentrations of atmospheric H*S have included a similar artifact component.

Diurnal Cycles of Atmospheric DA4S and H2S
To discern any diurnal cycles in the ambient concentrations of DMS and HIS, the scatter in the data necessitates the use of an averaging technique. Figure 4 shows the data from the Caribbean Transect after grouping the points into fourhour intervals and calculating the mean concentration in each grouping. Anomalous data points, as discussed above, have been discarded for the purpose of this calculation. This treatment reveals a slightly higher concentration of DMS during the night, similar to that observed by Andreae et al. (1985) in remote marine air. Our hydrogen sulfide data also exhibit a slight diurnal cycle, with the highest concentration at night. Previously, this effect had only been found in coastal or terrestrial environments (Herrmann and Jaeschke, 1984;Cooper, 1986). The diurnal cycle in H$ in the remote marine atmosphere only becomes apparent after the data are corrected for the analytical artifact from OCS hydrolysis. The magnitude of this artifact is at a maximum during the daytime ( Figure 5). The sulfide concentrations measured on both front and back-up filters are grouped into four-hour mean values, shown along with the resultant corrected means (of Figure 4) obtained by simple subtraction of the back-up filter sulfide from the front filter sulfide. It is clear in Figure 5 that the uncorrected data show an anomalous daytime maximum in H$, and that the artifact component is greater than the actual concentration at this time. The uncorrected data are similar to those of Delmas and Servant (1982) suggesting that their results may include a similar artifact component.
The magnitude of the diurnal cycles in both HxS and DMS are smaller than predicted in photochemical models of their atmospheric oxidation (e.g. Graedel, 1979). The models assume removal to be due solely to photochemically generated OH and, hence, predict a large nocturnal accumulation. Andreae et ul. (1985) showed in a model that the nighttime reaction of DMS with NOs, as suggested by Winer et ul. (1984), could explain the small diurnal changes in ambient DMS. In contrast, NO3 is not thought to be of importance in the oxidation of H$ since the rate constant is less than 5 x lo-l6 cm3 molecule-i se& (Dlugokencky and Howard, 1988). Hence, an alternative explanation for the

Origin of H$ in the Marine Boundary Layer
The data presented here sheds little light on the source of hydrogen sulfide over the surface ocean. In nearshore environments, as in the Bahamas, H# emissions are dominated by coastal emissions presumably from anoxic intertidal and/or subtidals muds. Over oligotrophic waters, our data suggest that there is a marine source; however, the magnitude of this source is smaller than previousIy thought. The possibility certainly exists that unknown positive artifacts continue to contribute to our 'corrected' signal. The development of methods with higher selectivity is desirable, and we are currently working in this direction. During the Bahamas cruise, we made several attempts to directly measure the sea/air flux of HzS using a Teflon lined floating chamber (Cooper, 1986). No emissions were ever detected, yielding an upper limit of 0.3 mg S mm2 yr-' for the H$S flux. This result, although the experiment has not been repeated, adds to our conviction that the oceanic HzS flux is not large.

Implications for the Origin of Non-seasalt S&ate
The relative contributions of DMS and H2S to SO2 and nonseasalt sulfate in the remote marine atmosphere can be assessed by comparing calculated oxidation rates for the two species using our estimated background concentrations of DMS (57 pptv) and H# (8.5 pptv) from the Caribbean Transects. The major oxidation pathway of DMS in unpolluted air is via reaction with OH radical. This reaction yields primarily SO 2, which is then further oxidized by both homogeneous and heterogeneous pathways to sulfate. DMS also has an oxidation pathway to methanesulfonic acid; however, the branching ratio of MSA to SO* formed from DMS has been the subject of some controversy. It now appears that the high yields of MSA (>50%) observed by some workers (Hatakeyama and Akimoto, 1983;Hatakeyama et ul., 1985) resulted from the presence of nitrogen oxides in some laboratory experiments, and that the value at marine NOX levels is quite low (<lOOh; Grosjean, 1984b;Barnes et al., 1987). This conclusion is in general agreement with both kinetic studies (Hynes et ul., 1986) and held measurements of the ambient ratio of MSA to non-seasalt sulfate (Saltzman et ul., 1983). For the purpose of this calculation, we use a rate constant of 6 x lo-l2 cm3 molecule-' s-' (Hynes et ul., 1986) and assume an SO2 yield of 95Oh from the oxidation of DMS by OH. A rate constant of 5 x lo-l2 cm3 molecule-' s-' (Barnes et uZ., 1986, and references therein) is used for the oxidation of H$S by OH and assume that the reaction product is entirely SO2. Using these rate constants and the background concentrations above, the relative contributions of DMS and H2S to the production of SO* over the oceans are 87 and 1 lob, respectively.
Two other oxidants have been proposed as potentially significant in DMS oxidation: NO3 (Winer et aI., 1984) and IO (Barnes et al., 1987). If operative, either mechanism for DMS oxidation would further decrease the relative contribution of H$ to the atmospheric sulfate. It is unlikely that either of these oxidants would react sufficiently rapidly to compete with OH in H$S removal. An upper estimate of 5 x 1 O-l6 cm3 molecule -' s-l for the rate constant of the reaction between NO3 and H2S has been reported by Dlugokencky and Howard (1988). The reaction of IO with DMS probably proceeds via a stabilized (conjugated) intermediate which breaks down to yield DMSO + I (Barnes et d., 1987). A similar stabilized intermediate does not exist for the IO + H$S reaction. It is not likely that either mechanism is particularly important in the removal of DMS from the remote marine atmosphere. NO3 radicals are associated with polluted air masses and are probably not present at sufficient concentration to significantly affect DMS levels in clean marine air (Winer et ul., 1984). IO radicals are generated from the photolysis of methyl iodide. This may be released from the sea surface in highly productive coastal regions with abundant planktonic and macro-algae, but is not likely to be significant in oligotrophic oceanic regimes.
A summary of the relative contributions of the various sulfur species present in background marine air to the background sulfate over the remote oceans is given in Table II. Oxidation of carbonyl sultide, the most abundant reduced sulfur gas in the atmosphere, accounts for less than 1% of the non-seasalt sulfate, based on a concentration of 500 pptv (Khalil and Rasmussen, 1984) and a rate constant of 2 x lo-l5 cm3 molecule-' se1 (Wahner and Ravishankara, 1987). The oxidation pathway of CSz is not clear at this time, and OCS may be produced in addition to SO*. A rate constant of 1.5 x lo-'* cm3 molecule-' s-l has been reported for the reaction of CS2 with OH (Hynes and Wine, 1987). Using this value, and assuming complete oxidation to SO2 of our measured concentration of 2.2 pptv CS2 (a probable upper limit) gives a contribution of less than 2% to the total non-seasalt sulfate.

Summary and Conclusions
Measurements of atmospheric DMS and H2S were made on three cruises in the Gulf of Mexico and the Caribbean. The ranges of measured concentrations were O-800 pptv (o-1045 ng S me3) DMS and O-260 pptv (O-340 ng S me3) H2S, with the highest concentrations found in coastal regions and over shallow waters. The lowest DMS concentrations were found in a continentally influenced air mass, whereas the lowest H2S concentrations were found in remote marine air. Concentrations of 57 pptv DMS (74 ng S me3) and 8.5 pptv H$ (11 ng S mm3) are calculated for conditions representing the remote marine boundary layer. These concentrations suggest that oxidation of DMS and H2S contribute approximately 87 and 1 l%, respectively, of the non-seasalt sulfate in the marine boundary layer.
Diurnal cycles were found in the atmospheric concentrations of DMS and H# in background marine conditions, exhibiting slight nighttime elevations of both compounds. The small magnitude of these variations cannot be fully explained.
and Tom Snowdon. Thanks are also due to the crew of the R/V Cape Florida and the R/V Columbus Iselin, and to Paul Llundgren and Ron Hutchinson of the RSMAS Marine Department. We are also grateful to Rod Zika for the use of his fluorometer, to Ken Mopper and Ollie Zafiriou for their efforts as chief scientists of the cruises, and to John Plane for helpful discussions. The work was supported by grants ATM 84-05921 and ATM 87-09802 from the National Science Foundation.