Short-term variability in biogenic sulfur emissions from a Florida Spartina Alterniflora marsh

Emissions of biogenic sulphur gases from a Florida Sparrina ulfernifloru zone were measured over several tidal and diel cycles using a dynamic flow chamber technique, corroborating recently published i~or~tion in the literature. The flux of hydrogen sulfide from individuaf measurements is shown to vary by over four orders of magnitude, and correhtes primarily with the stage of the tidal cycle. In contrast, the fluxes ofdimethyt sulphide, carbon disulphideand dimethyl disulphide vary by less than an order of magnitudeand correlate primarily with !he diurnal temperature changes in the sediment surface. These diflerences are discussed in terms of the various biological and physical parameters which may regulale the release of reduced sulphur compounds to the atmosphere. Key word index: Bioaenic sulphur emissions, salt marshes, hydrogm sulphide, dimcthyl sulphide, carbon


INTRODUCTION
Rexxnt studies on the problem of acid precipitation have focused largely on the biogeochemical cycling of SuIphur~on~ining ~om~unds. The cont~bution of biogenic reduocd S gases to the atmospheric S burden has been and remains an area of major concern (Marouiis and Bandy, 1977;Adams et al., 1981;Aneja er al., 1981;Cline and Bates, 1983;Andreae and Raemdonck, 1983;Steudler and Peterson, 1984). Oxi~tion of these com~unds leads to the formation of sulphate, which is precipitated as sulphuric acid in rainwater. Tidally influenced marine sediments and salt marshes are a major source of atmospheric H2S (Georgii. 1977;Hansen er al., 1978;Jaeschke et al., 1980;lngvorsen and J&gensen, 1982;Aneja. 1984;Will er a/., 1978) as a result of anaerobic bacterial sulphate reduction and decomposition of organic material. Salt marshes have also been shown to be a major source of organosulphur compounds (Adams et al., 1981;Aneja et 01.. 1981;Steudler and Peterson, 1984). The range of reported emission rates is large, and extreme variability at a given sampling site is often found. Consequently, estimates of the annual emissions of H2S and organosulphur compounds are still very uncertain, and are, in fact, the major holdback in attempts to accurately quantify the atmospheric S cycle.
The spatial and temporal emission patterns of biogenic S compounds reflect the combined effect of numerous biological and physical parameters. Different metabolic pathways produce di#Terent S compounds (Aneja, 1984). and the ultimate release to the atmosphere may depend on both the emciency of remineralization processes (Hansen et al., 1978) and the effect of physical parameters such as tidal inundation (Aneja, 1984) and temperature (Hansen et al., 1978). In this paper, we present flux measurements of four reduced S gases which were measured at three different en~ronments within a Spurt~nu u~~er~i~~ra zone on the Gulf coast of Florida. The fluxes of hydrogen sulphide (H2S), dimethyl sulphide (DMS), carbon disulphide (CS,) and dimethyl disulphide (DMDS) are assessed in terms of the relative importance of tidal and die1 cycles.

Study site
Emission measurements were made on M-17 May and 7-8 October 1985at St. Mark's National Wildlife RcJuge,Florida, U.S.A. (Fig. I). A stand of short Sparrina u~rer~~or~ (30-50 cm tali) extended about IS m below the high tide mark. The substrate was fine-grained quartz sand with an organic content of =z 1%. Measurements were made at three different locations; over the S. ahem$ora. on bare sand adjacent to S. alremiflora, and on exposed mudflats below the S. crlrernilyora stand. Sediment temperatures ranged from 23"Cat night to 36°C in the early afternoon in May.and from I9 to 29-C in October. The tidal range was 80-100 cm, exposing about 100 m of mud Rats. The anaerobic zone in the sand, evidenced by black colouration, with the characteristic odour of HIS, WBS typically less than 2 cm below the surface, and occasionally broke the surJace. The depth of this anaerobic region was observed 10 be much deeper within the S. altern#foru stand @@tally 5-10 cm). detail elsewhere (Cooper, 1986). The chamber was placed on !he surface of interest, !akingcare no! to damage any foliage, and depressed slightly in the sediment to ensure a good seal. Ambient air was used as a sweep gas a! a flow rate of 2.2-3.2 / min' '. This air was passed through gas scrubbers comaining silica gel, molecular sieve, activated charcoal and palladiumcaaled molecular sieve in order to remove ambient reduced S compounds, SO2 and atmospheric oxidanls before entering the chamber. An equilibration time of 20 min was required before rnatsurements were begun. Samples were drawn from the top of the chamber at a flow rate of

OS-I.3 I min-'for H~Sanal~jsor~cm3mjn-'
for DMS, CSI and DMDS analysis. H,S emission measurements were made during a complete tidal cycle on 16-17 May 1985, while DMS, CS2 and DMDS were measured over IWO tidal cycles on 7-8 October 1985. Measurements were made during the rising tide until water surrounded the base of the chamber, and continued during the ebb tide until the water had receded below the sampling toca!ion. Tide height was recorded periodically during the study periods.
The analytical equipment was housed in a self-contained mobile laboratory located close 10 the sampling site. This allowed samples to be analyzed immediately after colleftion. photometric detector. This method has a detection limi! of less than 5 x IO-'* moles of.5 injected, which corresponds to a lower emission rate of approiimately 1-2 mg S r;l_ 2 a I. Calibration was performed by purging liquid standards onto the sample loop and, jnde~ndently, by using ~rrn~!jon tubes (GC Industries, Chatsworth, CA). However, the latter method could only be used over short time scales due to longterm changes in the measured permeation rates of the tubes.

H,S was analRed by a method similar to that of
Negative quenching of the FPD signal by CO2 and HCs in the early part of the chromatograms prevented use of the gas chromatographic method for HIS or COS analysis.

Emissions of hydrogen sulphide
The variations of H2S emission rates from the wet site over Spurtina alferntflora, the adjacent bare sand site within the S. ~~!erni~oru,and the interti~~ mudflat site on 16-17 May 198.5 are shown in Fig. 2. Tidal height data are plotted as the vertical distance from the minimum tide height recorded over the 2-day sampling period. Sampling was conducted during the times indicated by the broken lines on the tide height plot. Intersection of the solid and broken lines therefore represents the times of tidal inundation at the three sites.
It is evident in Fig. 2 that the H2S emission at all three sampling sites increases dramatically as the water approaches the chamber, and is a maximum as the tide reaches the base of the chamber. The e&c! is most pronoun~d at the wet sand site, where the flux increases by > 4 orders of magnitude from about 0.01 toover 100gS(H2S)m-2a-'.Thedatabetween 12:30 and 16:OO represent time that the water covered the sampling site. The chamber was floated during this time,andemissionof76-272 mgS(H2S)m-2a-' were measured from the sea surface. An enhancement in H,S emission is also evident at the time that the water leaves the sampling site (after 16:OO). In fact, because the chamber was re-equilibrated for 20 min after exposure of sediment prior to sampling, the enhancemen! on the falling tide was probably significantly greater than that shown in Fig. 2.
Complete tidal cycles were not studied in the case of the wet S. abernifloru site or the intertidat mudflat site. The plot of H2S emissions in Fig. 2 from the S. alrerniforu site is a composite of measurements made over the entire 2-day period, 16-17 May, with the 17 May data plotted to show the correct tidal height at the time of sampiing. Though less dramatic than at the wet sand site, the emission measurements at these sites show a similar tidally induced enhancement in the HLS flux as the water approaches the chamber. An emission range of 0.005-I .04 g S( H2S) m " * a -' was measured over the S. alternifloru, and 0.029-0.73 g S(H$) m -2 a-' on the mudflat.
A similar enhancement of H2S emissions from a N Carolina intertidal mudflat was reported by Aneja (19&Q, who suggested that hydrostatic pressure forced the rekase of gases. The above observations are in marked contrast to those made by Steudler and Peterson (1985) during a study of the diurnai cycle of reduced S emissions. Their data from a site over S. alrerniflora not only showed no tidal enhancement in the HIS flux, but, on the contrary, actually indicated the reverse effect at the incidence of high tides. The importance of the short-term enhan~ment in H2S emission for the calculation of an atmospheric HIS flux can be clearly demonstrated by integrating the data presented in Fig. 2. The net emission during the brief 12min period from 00:20 to OO:32 (942 pgrn -2, was 30 times greater than during the entire 7 h interval from 16:40 to 23:40 (31 pgrn-').

Emissbns ($ dimethyl sulphidc, curhnn disufphidp und dimethyl disulphide
The variations of the emission rates of DMS, CS2 and DMDS from the wet site over Spurrittu crlrcv-njflorrr and the adjacent bare sand site on 7-8 October 1985 are shown in Fig 3. It is clear that there is no significant increase in the emission as the tide reaches the chamber, indicated by the broken line in Fig. 3. Instead, the predominant feature of the emission pattern from both sampling sites is the steady decrease on both days through the afternoon and evening. This follows the decrease in sediment temperature shown in Fig. 3, measured both inside (IN) and outside (OUT) the chambers.
The variation of emission rate with temperature is similar to that reported by Aneja et al. (197QA who demonstrated a logarithmic dependance of flux on temperature for DMS from a S. alterniffora zone and for H2S + COS from an intertidal mudflat at low tide. This relationship may be the result of several different factors. The metabolic activity of soil bacteria is a function of temperature, which is conveniently quantified by a Qto (tem~rature coefhcient) factor. This factor, the increase in activity for a 10°C rise in temperature, is normally 2-3 for enzyme mediated processes. From the data presented here, thecalculated Qlo is significantly higher, greater than 10 in all cases. This suggests that other processes may also be contributing to the elevated emission of the S compounds. Two possible effects that may be important are the solubility of the S gases and the stability of their complexes. Both decrease with increasing temperature, but insufhcient measurements were made to assess the magnitude of the effect.
It is evident from the emission plots in Fig. 3 that the emissions of DMS and CS2 are greater from the S. olternifiora site than the bare sand site, while the flux of DMDS is similar from the two locations. DMS is the predominant species emitted from both sites, the flux being significantly higher than both CS2 and DMDS, in agreement with all the previous studies listed in Table I.
The only emission pattern that does not adhere closely to the temperature data in Fig. 3 is that of DMS from the S. alrernijloro site. This suggests that the release of this compound may not be entirely related to bacterial processes in the soil, but may be related to the metabolism of the S. akernifforu.
The same emissions pattern has been noted previously at the same study site and on a drier, infrequently flooded, site where the emission of DMS was found to be related to biomass of S, nlternifiora inside the chamber (de MelIo et al., 1987). It is suggested that the release of DMS may be related to the osmotic regulation of the S. u~rer~~oru in response to freshwater run-off from the interior of the marsh. Osmotic regulation with dimethyl propiothetin, the most probable biological precursor of DMS, has been found in certain Spartina species (Larher er al., 1977).
The range of measured fluxes is compared to that of hydrogen sulphide in Table 1 measured in this study fall toward the lower extreme ot earlier studies. N Florida is the southern extreme of Spartina species habitation, and the sampling site at St. Marks Refuge is well flushed tidally and very low in organic matter. Steudler and Peterson (1985) reported emissions of approximately an order of magnitude higher from a peaty New England Spartina marsh. This difference in substrate could explain both their higher emission rates of organosulphur compounds and lack of observed tidal effects. The peaty substrate would hold water more efficiently than a sandy substrate, minimizing the effect of the incoming tide flushing out pore waters or accumulated gases, while at the same time acting as a rich source of biodegradable organic matter.

Calculation of an atmospheric sulphur jiux
Each of the plots in Figs 2 and 3 can be integrated to obtain an atmospheric flux of reduced S compounds over the sampling period. Because different processes are found to explain the emission patterns, it is necessary to use different methods of calculation to arrive at a mean flux. The H2S data need to be integrated over a tidal cycle, while the DMS, C.!$ and DMDS must be integrated over a diurnal cycle. Mean flux estimates calculated in this manner are presented in Table 2. These calculations assume two equal tidal cycles per day; a diurnal flux is thus obtained by simply doubling the H2S emissions seen in Fig. 2 for the single tidal cycle. Table 2 shows that the predominant emission from the S. alterniflora site is DMS (448mgSm-*a-') being almost an order of magnitude greater than the combined flux of the other gases. The situation is markedly different at the unvegetated site in the S.

alterniflora,
where the HIS emission (756 mg S rn-' a -') is more than an order of magnitude greater than the combined organosulphur emission. This is in agreement with the results of Aneja et al. (1981) (Table  I) and Steudler and Peterson (1984). The higher llux of H,S from the sand relative to the other gases calculated here is probably a consequence of our complete study of the tidal cycle giving a greater integrated flux than the steady low tide measurements.
A possible source of error in this study is the enhancement of emissions by the use of S-free sweep gas in the chamber. In practical terms, however, the short-term changes in emission rates found at this site would make the use of ambient air as sweep gas impossibly complicated, since the residence time in the   chamber is long (7 min)compared to the sampling time at peak emission rates (30 s).

CONCLUSIONS
The data presented in this paper indicate that the major emission of H2S from the intertidal regions studied is not a steady release from the sediment surfaces exposed at low tides, but is concentrated in a narrow region at the water's edge as the tide rises and falls. The higher fluxes measured at a given sampling site occur for very brief periods of time. Consequently, comprehensive flux data on a short time scale are required in order to calculate average emissions on a longer time scale. By conducting emission measurements on expanses of exposed mudflats or marshlands at times of low tide, previous studies may have significantly underestimated HIS fluxes to the atmosphere. A mean flux of 0.756 gS(H,S)m-* a _ ' from bare sand in a S. alterntjlora zone is obtained by intcgwtion of the emissions data meusurcd over a complete tidal cycle. While this may be an underestimate due to the lack of data at the beginning of the cycle, it is still more than an order of magnitude higher than the directly determined flux for the steady emissions over 7 h of tidal exposure, which represents 92';/, of the sampling time.