Spatial and diel variability in the emissions of some biogenic sulfur compounds from a Florida Spartinaalterniflora coastal zone

Aaorpkrr LPirommr Vol. 21. No. 1. pp. 9al-990. ooo4-6981/%7 Printed in Gre.al Bnuin. Q 1987 Pergamon Journals Ltd. SPATIAL AND DIEL VARIABILITY IN THE EMISSIONS OF SOME BIOGENIC SULFUR COMPOUNDS FROM A FLORIDA SPARTINA ALTERNIFLORA COASTAL ZONE WILLIAM Z. DE MELLO, DAVID J. CCK~PER t, WILLIAM J. COOPER *, ERIC S. SALRMAN, ROD G. ZIKA, DENNIS L. SAVOIE and JOSEPH M. PROSPERO Rosenstiel School of Marine and Atmospheric Science, Univcnity of Miami, Miami, FL 33149, U.S.A. and *Drinking Water Research Center, Florida International University, Miami, FL 33199. U.S.A. (Firsr receiued 27 May 1986 and injinolform I I August 1986) Abstract-Emission rates of the biogcnic reduced sulfur gases dimethyl sulfide, dimethyl disulfide, carbon disulfide and hydrogen sulfide were measured from several environments within a Florida Sparrino olfernijoro coastal zone. Spatial and diel variability was observed in the emission rates of all the sulfur gases. The speciation and magnitude ofsulfur emissions can be related to site elevation and the spatial variability of vegetation coverage. Dimethyl sulfide appears to be a metabolic byproduct of S. olrerniflora. Key word index: Dimethyl sulfide, hydrogen sulfide, dimethyl disulfide. carbon disulfide. biogenic sulfur emissions, Sparlina olferniflora. INTRODUCTION In recent yean, interest in the natural input of volatile sulfur gases to the atmosphere has increased because of their importance in balancing the global S cycle and their contri- bution to the acidity of rainfall. Emissions of biogcnic S compounds have been measured directly over continental and coastal areas (Steudler and Peterson. 1985; Jdrgensen and Okholm-Hansen. 1985; Cooper er cl., 1987; and references therein). Most of these measurements were conducted over salt marshes. in particular Sportino alternifora marshes, and areas of mudfiats. Dimethyl sulfide was usually found to be the predominant compound emitted from vegetated areas; from mudflats the major compound emitted was H,S. Other than this broad generalization. however, there is a lack of data concerning dicl or spatial variability in the reported emission rates. Seasonal changes in the emission rates of reduced S compounds from a New England S. olferniflora marsh were reported by Steudler and Peterson (1985). and extremely short term, tidally induced, changes in HxS emission rates from a Florida S. olrernifloro zone were reported by Cooper er al. (1987). This report extends our previous data set of the emission fluxes of dimethyl sulfide (DMS), HIS, carbon disulfide (CS,)and dimethyl disulfide (DMDS) to include the elfects of spatial variability and of temporal variability on a diel scale. STUDY SITE The present study wasconducted during May and October 1985 and January 1986 at the St. Marks National Wildlife Refuge, Florida (Fig. 1). Stands of the saltmarsh cordgrass S. olrernifloro ranged from 30 to SO cm tall and occupied a zone about IS m wide seaward of the extreme high tide mark. At this locale the substrate is a fine-grained quartz sand. Unvegetated areas within the cordgrass had a thin oxic top layer, about 0.5 cm thick. and an underlying dark layer t To whom correspondence should be addressed. Fig. 1. Sampling site location. St. Marks National Wildlife Refuge, Florida, U.S.A. charactcrimd by a strong odor of HIS. The thickness of the oxic layer within the cordgrass was variable but generally ranged between 5 and 1Ocm. Samples were taken at two different locations in the S. alrernfloro: a wet site flushed by at least one high tide each day; and a drier, more oxic, site close to the extreme high tide mark. Samples were also taken at an adjacent bare sand site within the wet S. ulrernifloro stand. SAMPLING AND ANALYTICAL METHODS The sampling and analytical techniques have been de- scribed in detail previously (Cooper cr cl/., 1987). Briefly, samples of S compounds emitted from the soil surface were taken using Teflon lined dynamic flow chambers. DMS. CS2 and DMDS were trapped on Teflon loops at - 183°C. and analyxcd gas chromatographically using a Chromosil 330 column (Supelco, Inc.. Rellefonte, PA) and a S specific Ilame


INTRODUCTION
In recent yean, interest in the natural input of volatile sulfur gases to the atmosphere has increased because of their importance in balancing the global S cycle and their contribution to the acidity of rainfall. Emissions of biogcnic S compounds have been measured directly over continental and coastal areas (Steudler and Peterson. 1985; Jdrgensen and Okholm-Hansen. 1985; Cooper er cl., 1987; and references therein). Most of these measurements were conducted over salt marshes. in particular Sportino alternifora marshes, and areas of mudfiats. Dimethyl sulfide was usually found to be the predominant compound emitted from vegetated areas; from mudflats the major compound emitted was H,S. Other than this broad generalization. however, there is a lack of data concerning dicl or spatial variability in the reported emission rates.
Seasonal changes in the emission rates of reduced S compounds from a New England S. olferniflora marsh were reported by Steudler and Peterson (1985). and extremely short term, tidally induced, changes in HxS emission rates from a Florida S. olrernifloro zone were reported by Cooper er al. (1987). This report extends our previous data set of the emission fluxes of dimethyl sulfide (DMS), HIS, carbon disulfide (CS,)and dimethyl disulfide (DMDS) to include the elfects of spatial variability and of temporal variability on a diel scale.

STUDY SITE
The present study wasconducted during May and October 1985 and January 1986 at the St. Marks National Wildlife Refuge, Florida (Fig. 1). Stands of the saltmarsh cordgrass S. olrernifloro ranged from 30 to SO cm tall and occupied a zone about IS m wide seaward of the extreme high tide mark. At this locale the substrate is a fine-grained quartz sand. Unvegetated areas within the cordgrass had a thin oxic top layer, about 0.5 cm thick. and an underlying dark layer t To whom correspondence should be addressed. charactcrimd by a strong odor of HIS. The thickness of the oxic layer within the cordgrass was variable but generally ranged between 5 and 1Ocm. Samples were taken at two different locations in the S. alrernfloro: a wet site flushed by at least one high tide each day; and a drier, more oxic, site close to the extreme high tide mark. Samples were also taken at an adjacent bare sand site within the wet S. ulrernifloro stand.

SAMPLING AND ANALYTICAL METHODS
The sampling and analytical techniques have been described in detail previously (Cooper cr cl/., 1987). Briefly, samples of S compounds emitted from the soil surface were taken using Teflon lined dynamic flow chambers. DMS. CS2 and DMDS were trapped on Teflon loops at -183°C. and analyxcd gas chromatographically using a Chromosil 330 column (Supelco, Inc.. Rellefonte, PA) and a S specific Ilame photometric detector. H,S was trapped on a silver nitrate impregnated filter and anaiyxed using the fluorescence quenching of dilute fluorescein mercuric acetate. Samples for HIS were only taken at times of low tide in order to avoid the tidal etfects reported by Cooper er al. (1987). These methods have detection limits for DMS, CSz. DMDS and H,S of 0.01. 0.03,0.06 and 0.03 pg m -x h-', except in the case of the bare sand site. Here, the sample volumes were limited to 500 cm'. and the detection limits of DMDS and CS2 increased to 0.1 and 0.2 pg S m -z h -', respectively.
Soil surface temperature was recorded using a mercury thermometer. The difference between the soil temperature inside and outside the chamber was less than 1°C at night, but was as high as 6'C at the unvcgctated site in the afternoon. Solar flux was monitored using a radiometer (The Epplcy Laboratory, Inc., RI) equipped with a wide band solar spectrum filter (300-800 nm). Tide height was measured using a staff gauge. Table I. Most of the data sets for the individual sites represent diurnal studies; sampling was only ceased at times of tidal inundation. Several general observations can be made concerning the range of emission measurements presented here. The most obvious are that DMS is the predominant species emitted from both the wet and dry Sporrincl sites and that the emissions of all compounds from the dry site arc significantly higher than those from the wet site.

Results of the emission flux measurements are summarixed in
The wet Sparha and the wet sand sites in May. October and January were sampled with two identical chambers less than I m apart. Despite the proximity of the sites, the emissions of DMS were more than an order of magnitude higher from the vegetated than from the bare site. The emission rates of the other gases showed no systematic difference between the two sites. In contrast to the emission of H2S, data from the October sampling trip (Cooper er al., 1987) indicate that tidal changes do not significantly atTect the emissions of the other S compounds on a short time scale.
In Fig. 2, WC present data obtained in October from two dry sites, less than I m apart, in what appeared to be a homogeneous stand of S. alterniflora. Simultaneous sampling showed a two-fold difference (site 2 > site I) in the emission rate of DMS between the twosites.Thcemission rateofCS,.a factor of IO lower. showed the reverse effect with Site I having twice the emission rate of Site 2. DMDS was comparable in magnitude at the sites. Careful inspection of the study sites later showed that although there was approximately equal S. olrernijbra biomass above the soil, the root biomass was substantially greater at site 2.
It is evident in Fig. 2 that the variation in emission rates of DMS, CS, and DMDS follow the trend of both soil surface temperature and solar flux. There is. however, an interesting observation that was made when sampling was conducted at low tide during the night. Figure 3 shows data obtained at the wet Spartino site in May. A daytime maximum similar to that of Fig. 2 can be seen, but a second maximum also occurred in the emission flux of DMS at night, which does not follow the temperature data. At the same time that this effect was occurring, the salinity of the sediment porcwater was decreasing as the tide receded, probably due to seepage of water from impounded marshland landward of the sampling site.
While temperatures on the May and October trips were comparable, it was significantly colder in January. Emissions of HIS, CSI and DMDS werecorrespondingly lower, and the lowest emission rates of DMS from all three sites were found at this time. However, the upper emission rate of DMS measured from the two vegetated sites was not noticeably different than on previous occasions, despite a difference of IO-19°C in the upper temperatures of the individual sampling sites.

Spatial variability in the emission rates of the biogcnic sulfurcompounds DMS,CSx and DMDS within this sandy S. o/terniJlora zoneappears to be related to two principle factors: (I) the amount of biomass enclosed by the chamber; and (2) the extent to which a site is influenced by daily tides. Diel variability, on the other hand, may be related either to temperature related changes in microbially mediated soil processes or to the transpiration/respiration cycle of the higher plants, i.e. photosynthesis in S. olrerniflora.
The processes regulating the formation and release of the various S compounds probably differ substantially. HIS is formed mainly by the anaerobic reduction of sulfate (Jdrgensen and Okholm-Hansen. 1985, and references therein), and its major release appears to be a consequena of tidal pumping (Cooper et al., 1987. and references therein). In contrast, previous studies (Bremner and Steele, 1978. and references therein) have shown that DMS, DMDS and CSI arc formed by the microbial degradation of amino acids in waterlogged soils. The increase in emissions of both CS* and  debris on the beaches and in the intertidal zone. The data presented here, however, are not consistent with the production of DMS solely by these microbial processes in the soil. If this was the only proaxss producing DMS. the drier site would be expected IO emit lexs DMS than the wet site, and the seasonal variability should be greater, i.e. more similar to that of the other compounds. The relationship between S. olremijfora biomass and DMS emission and the fact that acaxonal variability was evident at the unvcgctatcd site suggest that emissions of this gas are related to the physiology of the macrophytc. S. olrerni~oro.
The ionic sulfur containing compound, dimethylpropiothetin (DMPT), whose baxc catalyzed or enzymatic cleavage leads to DMS (Cantoni and Anderson, 1956)  To put the effect of tidal variation on H,S emissions in context here, our study of spatial and die1 variability precluded the use of H,S emissions data measured close to tidal inundation of the wet sites. For this reason the data of Table 1 do not clearly indicate the magnitude of the total flux of this gas. Cooper cr al. (1987) showed variation over four orders of magnitude at a bare sand site, with over 90% of emissions occurring in less than 10% of the tidal cycle. Similarly in January, at times of approaching inundation, the emission flux at the wet sand site increased from 0.6 to 8151 pgm-'h-l in less than I5 min. However, this elTect is only significant when calculating an integrated emission flux from unvegetated areas in the S. olrernifloru.
While the data presented here do not change the overall picture of biogcnic emissions from S. ahrrniflora salt marshes, it can be used to explain the large range of emission rates which have been reported in previous studies. The observed variability may be related mainly to the commonly used method of estimating emissions. The use of an encapsulating chamber is necessary to isolate the area of interest from other influences, but experimental and logistical constraints arc such that relatively small areas are enclosed at any one time (less than 0.1 m' in this study). These small areas are not necessarily representative of the larger ecosystem.