"RICHARDS, S. R., C. A. KELLY, AND JOHN W"
Limnol. Oceanogr., 36(3), 1991, 468-482 0 199 1, by the American Society of Limnology and Oceanography, Inc. Organic volatile sulfur in lakes of the Canadian Shield and its loss to the atmosphere S. R. Richards and C. A. Kelly Department of Microbiology, University of Manitoba, Winnipeg R3T 2N2 John I%‘.M. Rudd Freshwater Institute, Central and Arctic Region, Department of Fisheries and Oceans, 50 1 University Crescent, Winnipeg, Manitoba R3T 2N6 Abstract Identities, concentrations, and fluxes of volatile sulfur compounds (VSCs) were determined in 11 lakes in northwestern Ontario. Carbonyl sulfide (COS: up to 1.1 nM) and dimethyl sulfide (DMS: up to 11 nM) were present in surface waters during most of the ice-free season. Depth profiles showed accumulations below the mixed layer of methane thiol (MSH), DMS, and dimethyl disulfide (DMDS). There was no effect of low pH or increased [SOd2-] (from 2.4 to 12.3 mg liter-‘) on the concentrations and identities of surface-water VSCs. Accumulation of DMS below the mixed layer was 9 x higher in an acidified system, however, when compared to an unacidified reference lake. Estimates of flux from two stratified lakes indicated that volatilization was not an important sulfur loss mechanism compared to others such as sedimentary S0,2-. reduction. Concentrations of MSH, DMS, and DMDS in shallow Lakes 114 and 303 were often 5-30 X higher than in any other lake studied, and the flux from Lake 114 was -0.96-l .6 x the estimated oceanic DMS flux per unit of area. On a regional basis, in areas where SOd2- in precipitation is low, export of VSCs to the atmosphere from shallow lakes and bog pools may be significant compared to sulfur inputs through precipitation. The recognition that biogenic sulfur gases and coastal marine environments (e.g. An- are important in the cycling of sulfur and dreae and Barnard 1984). Lakes have re- in global climate regulation has prompted ceived almost no attention. In this study we considerable interest in their production and have identified and quantified VSCs in the fate. Volatile sulfur compounds (VSCs) typ- water column and sediments of lakes of the ically identified in environmental studies Canadian Shield. Rates of VSC flux to the include hydrogen sulfide (H2S), dimethyl atmosphere, the importance of this loss to sulfide (DMS), methane thiol (MSH), car- lake SO,*- budgets, and sites of VSC pro- bony1 sulfide (COS), carbon disulfide (CS,), duction in lakes were also studied. and dimethyl disulfide (DMDS). All these compounds can be produced biologically, Materials and methods although at least two (DMDS and COS) can Site description - Two sets of lakes were form from abiotic reactions (e.g. Bremner monitored from May to October 1988 for and Steele 1978; Kadota and Ishida 1972). the presence of VSCs. The first set consisted Much research on VSCs has focused on of several small (< 5 5 ha) lakes (226 South, their production and flux from the oceans 302 South, 114, 239, and 303) at the Ex- -- perimental Lakes Area (ELA) in north- Acknowledgments western Ontario (Brunskill and Schindler We thank S. Kasian for assistance with statistical 197 1). Two of these (302 South and 114) analyses, W. Buchannon for help with mass spectrom- have received experimental additions of eter analyses, C. Anema for obtaining samples from sulfuric acid (Table 1). The second set was the Red Lake Series lakes, B. Flett and A. Furutani for advice on analytical procedures, and P. Ramlal for located in a geologically homogeneous area assistance with analyzing anemometer data. We also of the Canadian Shield near Red Lake, On- thank M. 0. Andreae and two anonymous reviewers tario (50°N, 94”w). This set comprises larg- for their criticism of the manuscript. R. Hecky, D. W. er lakes ranging in surface area from 88 Schindler, and R. Hesslein provided comments on the (Green Lake) to 34,700 ha (Trout Lake) (Ta- first draft. Funding for this research was provided by NSERC ble 1). Samples were taken from surface wa- grant OGPGP 0 10. ter at the site of maximal depth in all lakes. 468 Volatile sulfur in lakes 469 Table 1. Morphometric parameters of the lakes from cells (4 ml each) O-20 cm below the studied and treatment information pertinent to 1988. sediment surface was pooled for analysis. A c: Treatment Extraction of VSCs- VSCs were extract- Depth (m) AO v,* (if appli- ed and trapped cryogenically. The extrac- Lake (IO4 mz) Mean Max (m-l) cable) tion apparatus consisted of a boiling flask, ELA lakes cold finger (water trap), and U-trap. To 2263 7.8 6.3 11.6 0.09 - monitor sample extraction efficiency, we 302s 10.9 5.1 10.6 0.10 H2S04t added an internal standard (DES in glycol, 114 12.1 1.7 5.0 0.59 H,SO,$ 239 56.1 10.5 30.4 0.07 - Holdway and Nriagu 1987). Glassware was 303 9.9 1.5 2.5 0,66 - silanized with 10% dimethyldichlorosilane Red Lake area (Sigma Chem. Co.) (Deprez et al. 1986). Green 88 7.7 18 0.08 - In order to minimize manipulation, we Orange 169 14.4 28 0.05 - did not filter water samples. There was little Linge 706 8.4 22 0.11 - Musclow 2,219 19.3 43 0.03 - alteration in DMS concentration when fil- Sydney 5,750 20.0 71 0.05 - tered samples were compared to unfiltered Trout, deep 34,700 13.7 47 0.07 - samples (data not shown), and MSH and Trout, shallow - 10 DMDS showed no consistent change in con- * ELA data based on Fee 1979. centration. This result suggests that the ex- t Rudd et al. 1990 (Lake 302s is currently acidified). $ Schindler and Turner 1982 (Lake 114 was acidified from 1979 to 1987). traction procedure itself does not produce DMS artificially by disrupting dimethyl sul- foniopropionate (DMSP)-containing algal In Trout Lake, a second site was in the cen- cells-a concern when analyzing for this gas ter of a large, unstratified bay (z,.,,,~= 10 m). in seawater. For Lakes 226 South and 302 South, we To extract sulfur gases, we heated a 250- obtained depth profiles. ml water sample (analyzed in duplicate) to Sample collection -Surface samples were -65OC and sparged it with UHP nitrogen collected at a depth of 0.2 m by hand in (60 ml min-l) for 50 min. VSCs were trapped silanized glass bottles. Samples from other in a U-shaped tube that was filled with silan- depths were taken with a peristaltic pump ized glass wool and immersed in liquid ni- and were overfilled three times before stop- trogen. After extraction, the U-trap was pering. After collection, all samples were evacuated, sealed, and heated to -5OOC. To immediately placed in the dark on ice. Anal- inject the contents onto the CC column, we yses were conducted within 24 h of sam- fitted the trap into a Carle GC sampling pling; tests showed little alteration in sam- valve with luer connections. ple integrity over this period (data not Detection and quant$cation of volatile shown). sulfur species-Sulfur gases were analyzed Because of the high solubility of DMS, with a Varian 3700 gas chromatograph (GC) dip sampling of water compares well with equipped with a dual-flame, photometric other sampling methods. COS is the least detector (FPD) at 2OO”C, with a Spectra- soluble, but solubility data for this com- Physics model 4290 integrator, Optimal flow pound also indicate that it should not be conditions were: H, (UHP) 140 ml min-l; affected by dip sampling (Wilhelm et al. air No. 1 (zero zero), 80 ml min-*; air No. 1977). 2 (zero zero), 170 ml min-l; He (UHP) car- Pore water was obtained in two ways. In rier, 40 ml min-l. one method, sediment was collected with A 2-m X 0.64-cm-o.d., Teflon-coated an Eckman corer. The surface 6 cm were aluminum column packed with 20% SE-30 transferred to silanized glass reagent bottles on Chromsorb P AW/DMCS 60/80 mesh without headspace. Subsamples were cen- (Analabs) was used with temperature pro- trifuged and the supernatant decanted and graming: 35°C (l-min hold), 25°C min-l to analyzed. The second method used a pore- 105°C (4-min hold). These conditions pro- water equilibration sampler (Hesslein 19 76) vided adequate separation of H$, COS, with Nuclepore (2 pm) membrane, equili- MSH, DMS, DES, and DMDS except at brated in the sediment for 1 week. Water high H,S concentrations when the COS peak 470 Richards et al. Table 2. Levels of organic VSCs in pore waters of Lake 114. Concn (nM) Pore water : Method Species Surface Pore water Surface water --- 26 Jul Centrifuged cos 0.65 Not resolved* Sediment MSH 3.7 53 14 DMS 6.2 11 1.8 DMDS 2.8 Not detected 3 Aug In situ cos 0.49 Not resolved Pore water MSH 14 120 8.9 Equilibration DMS 11 38 3.6 DMDS 2.0 Not detected - ---- * The high methane and H,S levels found in the pore water masked detection of any COS present due to their similar retention times. was obscured (e.g. hypolimnetic samples in GC- mass spectrometry analysis -To late summer). confirm the identity of the sulfur peaks, we Sulfur compounds were identified by periodically conducted GC-mass spectrom- comparing retention times to those of known etry. The column was a 10-m x 0.32-mm sulfur standards. Confirmation was ob- fused silica PoraPLOT Q capillary column tained by GC-mass spectrometry (see be- (Chrompack). The temperature program low). DMS and CS2 had retention times of was: 40°C (l-min hold), 20°C min--l to 4.6 and 4.8 min, but no CS, was detected 22O”C, then 1°C min-l to 240°C (1 -min with the PPD or during mass spectrometry. hold). Water samples were extracted as de- Because the thiol moiety is oxidized scribed above. readily, it is possible that the presence of GC-MS analyses, performed on the ELA DMDS in environmental samples resulted lake samples at three different times, have from the oxidation of MSH during extrac- confirmed the presence of H2S, COS, MSH, tion. In some environmental samples, how- DMS, and the internal standard, IDES. No ever, e.g. Lake 114 pore water (Table 2) and volatile sulfur species were detected during hypolimnetic Lake 226 samples (Table 3), the GC-MS analyses that had not previously MSH was detected in the absence of DMDS. been identified by retention-time compar- This result suggests that MSH oxidation was isons with sulfur standards using the PPD. limited. Calibration -Standard curves (Tanger- H2S was not detected by this purge-and- man 1986) were determined each day of trap method in the epilimnetic waters of the analysis in order to quantify the sulfur peaks lakes studied. Accurate measurement at the extracted. Sulfur standards used routinely high levels found in the lake hypolimnion were COS (96+%), MSH (99.5+%), DMS was not possible at the GC conditions used (99+%), DMDS (99+%), and DES (98%) for the other VSCs. Therefore, H2S values (all from Aldrich Chem. Co., Inc.). are not reported. Determination of extraction emiencies - Extraction efficiency of each compound was Table 3. Accumulation of organic VSCs below the determined to estimate recovery from water mixed layer. samples. Liquid sulfur standards were pre- Accumu- pared by dissolution in anoxic ethylene gly- lation rate Hypolimnetic conditions* co1 (Andreae and Barnard 198 3) and dilu- (mmol w2-1 tion in distilled HZ0 to concentrations Lake Compound m-2 d- I) PH (mg liter-‘) similar to lake water. Recoveries were as 302s MSH 0.98 5.46-6.33 0.74-9.84 follows: COS, 95.0&18’.8%; MSH, DMS 0.57 66.6+7.4%; DMS, 96.7+5.1%; DMDS, DMDS 0.19 2268 MSH 0.81 6.17-6.35 0.44-l .8 98.7+6.9% (mean + 1 SD, ri > 5 for each DMS 0.066 determination). -- Preparation of the internal standard -The * The given range represents the lowest and highest values determined at a depth of 10 m from 10 May to 30 August 1988 for Lake 302 South, DES internal standard was prepared in an- and from 30 May to 19 September 1988 for Lake 226 South (ELA, chemical data). oxic ethylene glycol (Andreae and Barnard Volatile sulfur in lakes 471 1983; Holdway and Nriagu 1987) (99+%, diffusivity (at an annual mean temperature spectrophotometric grade, Aldrich Chem. of 15°C) of each compound was calculated Co., Inc.) with silanized, 8-ml, glass, screw- by the method of Wilke and Chang (195 5), capped vials fitted with Teflon-faced sili- updated by Hayduck and Laudie (1974). The cone septa. A primary standard of 300 ng estimated error of these diffusivities is ml-l was prepared gravimetrically. A work- - 10%. ing standard in a second vial was made by The mean mass transfer coefficient on further dilution to a final concentration of each day of analysis was used to estimate 4 ng S ml-l; 25 ml of the working standard the flux of volatile sulfur species found that was added to environmental samples to de- day in lake surface waters. On dates when termine extraction efficiency. Internal stan- the above lake windspeed was estimated dard recovery was 92.4 + 6.7% (mean + SD, from above land wind data, a mean daily n = 35). windspeed was used to calculate the mass Measurement of windspeed - Windspeed transfer coefficient for that date. The same was measured on Lake 302 South with an mass transfer coefficients were used to es- anemometer (Belfort) equipped with a mod- timate sulfur fluxes from Lakes 302 South, el DPlOl one channel time of event re- 226 South, 239, and 114. The diffusive corder (Omnidata). The anemometer was boundary-layer thicknesses calculated in this placed near the center of the lake; distance study agreed well with those determined by from the anemometer cups to the lake sur- Emerson et al. (1973) on another lake at the face was 1 m. Windspeed was recorded from ELA. July to October. Estimation of sulfur flux to the atmo- Before July, above lake windspeed was sphere-Estimates were made of sulfur loss estimated from above land windspeed at a to the atmosphere with the stagnant-film nearby meteorological site (K. Beaty pers. model of gas exchange (Lewis and Whitman comm.), and the relationship between the 1924). Flux was calculated according to two windspeeds developed from measure- ments made when both sites were operating. F = k(C - Co) (2) Estimation of mass transfer coeficient - The relationship between diffusive bound- where F is the gas exchange per unit area, ary-layer thickness and windspeed deter- k the mass transfer coefficient, c the gas con- mined by Broecker et al. (1980) uses wind centration in the fluid phase, and c, the at- velocity at a height of 10 m. The power-law mospheric concentration of the gas. model given in Eq. 1 (Panofsky and Dutton The mass transfer coefficient was calcu- 1984) was used to convert the l-m wind- lated as described above, and C, was as- speed measurements to the estimated wind- sumed to be negligible, based on estimates speed at a height of 10 m: of atmospheric DMS concentrations above the ocean (Andreae et al. 1985; Barnard et VJV1 = (z&)p (1) al. 1982). COS has an atmospheric concen- tration of - 5 12 parts per trillion by volume where V, is mean windspeed at height zr, (pptv; Torres et al. 1980), but this consti- V, is mean windspeed at height z2, z, is tutes only 3-8% of the mean aqueous COS height above the lake surface, z2 is height concentrations in the ELA lakes. This above the lake surface, and p is the power- amount is within the error of the calcula- law exponent. The value of p (0.24) was tions and was therefore not included in the estimated from the relationship between equation. Fluxes on each day were averaged power-law exponents and lake surface area to generate the mean flux (see Table 6). Each (A. Solinske unpubl.). Diffusive boundary- lake in the ELA area experiences essentially layer thickness was estimated for each 4-h the same above the lake wind. Therefore, period, and mass transfer coefficients per- differences in sulfur fluxes among lakes were taining to these 4-h periods were then de- dependent on differences in concentrations termined with the calculated molecular dif- of sulfur species. For this reason concentra- fusivity of each species. The molecular tion data were analyzed in a Tukey multi- 472 Richards et al. Table 4. Time-weighted mean surface water WC levels. X-Mean value from May to October given in nM (SD in parentheses); R-range; ND-not detectable (detection limit - 1 ng S); NR-not resolved due to high H,S and CH, concn. Lake. cos DMS DMDS MSH -- ELA lakes 302s x 0.36(0.11) 1.3(0.60) 0.09 l(O.090) ND R 0.23-0.94 0.32-2.7 ND-O.3 1 226s x 0.31(0.16) 1.3(0.5) 0.065(0.083) o.orJs;.o60) R ND-O.91 0.7 l-2.7 ND-O.28 ND-O.45 239 x 0.26(0.12) l.l(O.4) 0.020(0.050) ND R ND-O.53 0.36-2.6 ND-O.27 ND 114 x 0.61(0.13) 6.6(2.9) 1.4(0.7) 8.6t5.8) R 0.37-l. I 0.75-l 1 ND-3.0 ! ND-32 303* x 0.38(0.15) 1.2(0.4) 0.043(0.043) l.l(O.4) R 0.23-0.9 1 0.33-2.0 ND-O. 170 ND-2.8 239, NE bog x NR 3.9(2.5) 0.097(0.059) 13(6) R NR 0.65-l 1 ND-O.25 ND-38 Red Lake area Green x 0.26(0.09) 0.95(0.37) ND ND R ND-O.51 0.24-2.2 ND ND Orange 55 0.12(0.16) 0.84(0.37) ND ND R ND-O.69 0.26-2.3 ND ND Linge x 0.18(0.13) 0.65(0.20) ND ND R ND-O.55 0.36-1.8 ND ND Musclow x 0.13(0.11) 0.48(0.19) ND ND R ND0.31 ND-O.94 ND ND Sydney R 0.16(0.08) 0.74(0.19) ND ND R ND-O.26 0.42-l .3 ND ND Trout, deep 37 0.19(0.10) 1.3(0.59) ND ND R ND-O.5 1 0.28-3.6 ND ND Trout, shallow x 0.091(0.10) 1.6(1.1) ND ND R ND-O.26 0.28-5.1 ND ND * Lake 303 values based on four measurements from August to October. comparison test (P < 0.05) to determine 4). The epilimnia of most lakes contained differences in lake sulfur fluxes. A test of detectable concentrations of only COS (lake homogeneity of coefficients of polynomial means: 0.09-0.61 nM, Table 4) and DMS equations (Milliken 1989) was used to test (lake means: 0.48-1.6 nM, Table 4). I-%$, for differences in DMS temporal patterns (P MSH, and DMDS were found generally only < 0.05). below the mixed layer (Figs. 1, 2), although Calculation of accumulation rates below low levels ( < 0.3 1 nM) of DMDS were oc- the mixed layer -Accumulation rates were casionally detectable in the surface waters calculated by determining the total mass of of Lakes 239, 226 South, and 302 South organic volatile sulfur below the mixed lay- (Fig. 3A,B). An exception to this general er at two different times and dividing the pattern was found in a shallow, holomictic difference by the sediment surface area and lake (Lake 114), where MSF and DMDS the elapsed time. The mass of each species were easily detectable (up to 32 and 3.0 nM, was determined by multiplying the concen- respectively) at the lake surface, with COS tration in each depth interval by the volume and DMS also present (up to 1.1 and 11 of water in that interval and summing the nM, Fig. 3C, Table 4). Another shallow lake, values. Lake 303, also had elevated levels of MSH (Table 4). Results With a few exceptions, in particular Lake Sutfur species and concentrations in sur- 114 (Fig. 3C), DMS was the tjrost abundant face waters -Five VSCs were identified: species (Figs. 3A,B and 4, Table 4). Con- H$, COS, MSH, DMS, and DMDS (Table centrations ranged from -0.:3 1 to 2.8 nM Volatile sulfur in lakes 473 D DMDS I 0 10 20 30 40 Concentration (nM) Fig. 1. Depth profiles of Lake 302s determined on two dates in 1988. Solid lines- 15 June; dotted lines- 24 August. The bottom of the mixed layer for each date indicated by horizontal lines: solid- 15 June; dashed- 24 August. in most lakes, but Lake 114 commonly ex- in the Red Lake area (Table 4) and the for- hibited higher levels (up to 11 nM, Fig. 3C). mer was detectable only occasionally in Both the deep and shallow stations on Trout Lakes 226 South, 302 South, and 239, and Lake (the largest lake) also had high DMS at levels not exceeding 0.3 1 nM (Fig. 3, Ta- concentrations on at least one sampling date ble 4). MSH was found in the epilimnion (3.6 and 5.1 nM for the deep and shallow of Lake 226 South only in late October (Fig. stations; Fig. 4C, Table 4). COS, the other 3A). species detected routinely in surface waters, Surface waters of acidified Lake 302 South was usually at lower concentrations than contained concentrations of DMS, COS, and DMS. ELA lakes tended to have higher COS DMDS very similar to the unacidified ref- and DMS concentrations than the Red Lake erence lake, 226 South (Fig. 3A, Tables 4 area lakes, which are larger (Figs. 3 and 4, and 5). Therefore, neither lower pH nor el- Table 4). evated SOd2- concentrations (from 2.4 to DMDS and MSH, which were consis- 12.3 mg liter-‘) had an obvious effect on tently found in Lake 114, were usually not VSC levels. Accumulation rates of DMS and detectable in the surface waters of the other DMDS below the mixed layer were greater, lakes, except for Lake 303, which is also however, in the acidified lake (see below). shallow and unstratified (Figs. 3 and 4, Ta- The very high mean concentrations of ble 4). In Lake 114, MSH concentrations DMS and DMDS found in the surface wa- were often higher than DMS, reaching up ters of shallow Lake 114 (Table 4, Fig. 3C) to 32 nM. DMDS and MSH were not de- were 5-70 x greater than the deeper lakes. tected at any time in the epilimnia of lakes Mean DMS and DMDS concentrations were 474 Richards et al. 9, ---------- F ‘II& .. .... ..*..._.......................~ ,.,..... +......-...... -4 0 I I 4 I, 6 1 I 12 I , 16 ( , 20 , , 24 I 0 20 40 60 I--- 60 1 Concentration (nM) Fig. 2. As Fig. 1, but for Lake 2268. Solid lines-7 July; dotted lines-7 September. Horizontal lines: solid- 7 July; dashed-7 September. also 2--l 4 x higher than found in the bog For example, some lakes showed a peak in pool sampled (Table 4). MSH concentra- DMS concentration about midway through tions were similar in Lake 114 and the bog summer (e.g. Lakes 239, 226 South, 302 pool (Table 4); this species was usually un- South, 114, Green Lake, Orange Lake, and detectable in the surface water of the deeper Trout Lake, shallow station; Figs. 3 and lakes. In order to determine if the concen- 4A,C). In other lakes, however, DMS fluc- trations in Lake 1 14 were representative of tuated or remained relatively constant (e.g. unstratified, shallow lakes, we performed Sydney Lake, Linge Lake, Musclow Lake, analyses (August-October 198 8) on another and Trout Lake, deep station; Figs. 3C and shallow ELA lake, Lake 303 (Table 1). Dur- 4B,C,D). COS had a midsummer minimum ing this period, DMS and COS levels in in most lakes (8 ofthe 11 studied; Figs. 3A,B Lake 303 were not significantly different in and 4). the deeper, stratified lakes (Table 4), but A test of homogeneity of coefficients of MSH, which probably has a significant sed- polynomial equations (Milliken 1989) was iment origin, was very much higher in Lake applied to the data to test for statistically 303 than in the deeper stratified lakes. The significant differences (P < 0.05) in seasonal following year (1989), both DMS and MSH patterns of the ELA lakes. Quadratic equa- concentrations in Lakes 114 and 303 were tions of the form high (Lake 114: DMS-up to 10.9 nM; log y = BO + B,(log x) + B,(log x)~ (3) MSH-4.2 nM; DMDS-6.3 nM; Lake 303: DMS,-up to 9.1 nM; MSH-5.5 nM; were fitted to DMS concentration (as y) and DMDS-4.2 nM). day number (as X) for Lakes 302 South, 226 Some seasonal patterns for VSC concen- South, 114, and 239 (data shown in Fig. 3). trations in surface water were discernible. Significant fits were found fqr Lakes 114, Volatile sulfur in lakes 475 Lakes 302s * 1 (bold) AA & 2268 Legend cos -.-.-.-.m-.-.-.-.-.-.e.e.w.e. MSH ........................................... .............*.. DMS DMDS --------------- Sampling Date (1988) Fig. 3. Surface VSC concentrations in four ELA lakes. Species differentiation for all graphs given in legend. Sydney and Linge Lakes B (bold) 6 C D Trout Lake Musclow Lake cos -.-.-.-.- DMS - Sampling Date (1988) Fig. 4. Surface VSC concentrations in six lakes in the Red Lake area. Trout Lake panel shows shallow (bold) and deep stations. Species differentiation for all graphs given in legend to panel D. L 476 Richards et al. Table 5I Comparison of epilimnetic SOd2-concentrations and pH in Lakes 302 South and 226 South (I 988). PH [S042-] (mg liter ‘) Lake Range Mean Range Mean 2268 30 May-24 Ott 6.68-7.10 6.9 2.30-2.53 2.43 302s 10 May-26 Ott 4.59-5.25 4.7 10.20-14.20 12.29 302 South, and 226 South. The ANCOVA pore water by factors of 1.8 and 3.6 (Table suggested by Milliken (1989) was used to 2). No DMDS was detectable in these anal- test whether the coefficients BO, B1, and B, yses. Any COS that might have been present were significantly different for these three was obscured by high concentrations of H,S lakes. This analysis of temporal patterns and CH,. suggested that the DMS concentrations in Estimates of VSCjlux to the atmosphere Lakes 302 South, 226 South, and 114 fol- and its sign$cance in the sulfur budgets of lowed the same pattern, with Lakes 302 fresh water lakes -Estimates were made of South and 226 South described by a com- sulfur losses to the atmosphere in four lakes mon equation. A significantly larger B0 term (Table 6). The stratified lakes were not sig- was determined for Lake 114. The COS and nificantly different from each other (P < DMDS data could not be fitted to smooth 0.05) in mean fluxes of COS, DMS, DMDS, curves, which prevented a similar set of or MSH. The total fluxes from these lakes analyses on these compounds. ranged from 804 to 1,170 nmol S m-‘* d-l. VSC accumulation below the mixed layer The sulfur flux from unstratified Lake 114 in stratiJied lakes-In the two lakes where was significantly greater (P < 0.05) by an measurements below the epilimnia were order of magnitude (12,500 nmol m-* d-l, made (Lakes 302 South, acidified, and 226 Table 6). This disparity was due principally South, reference), the deeper strata con- to consistently higher concentrations of tained significantly higher concentrations of DMS, MSH, and DMDS. MSH, DMS, and DMDS during summer Relationship to sulfate budgets -The sig- stratification (Figs. 1,2). MSH accumulated nificance of sulfur volatilization in the SOa2- at the highest rate in both lakes (0.98 mmol budgets of lakes was estimated (Table 6). MSH .m-* d-l in Lake 302 South and 0.81 The total flux could account for only 0.3 in Lake 226 South, Table 3). These rates and 1.O% of Sod*- lost via SOa*- reduction were similar despite the lower pH of Lake in Lakes 302 South and 239. Further, the 302 South and elevated S04*- concentra- loss of MSH, DMS, and DMDS from the tions resulting from experimental acidift- deeper strata to the atmosphere at fall over- cation (Rudd et al. 1990; Table 3). Accu- turn represented only 0.07% of the Sod*- mulation rates of DMS and DMDS, lost in Lake 302 South (Table 6). In Lake however, were very different in the two lakes. 114, however, total sulfur flux accounted for The DMS accumulation in Lake 302 South almost 15% of the “lost” SOd2-. Therefore, (0.57 mmol DMS m-* d-l, Table 3) was sulfur volatilization was an important term about 9 x higher than the DMS rate in Lake in the SO,*- budget of Lake 114. 226 South (0.066 mmol DMS m-* d-l, Ta- ble 3). No DMDS could be detected above Discussion the sediments in Lake 226 South, whereas Concentrations and sources -In this an accumulation rate of 0.19 nmol DMDS study, DMS concentrations (lake means: m-2 d-l was calculated for Lake 302 South. 0.3 l-6.6 nM, Tables 4 and ,7) were com- After fall overturn in Lake 302 South, the parable to levels found in Hamilton Harbor, only species detectable throughout the en- Lake Ontario (Table 7; Holdway and Nri- tire water column were COS and DMS, agu 1987), but about lo-fold greater than which were uniform in concentration. concentrations in Lake Mendota surface Pore water --In Lake 114, pore water had waters (Table 7; Zinder and Brock 1978a). MSH concentrations 8.9-l 4 x higher than Two Antarctic lakes, Burton and Organic the surface waters. DMS was also higher in (Deprez et al. 1986), contain DMS at con- Volatile sulfur in lakes 477 Table 6. Estimates of sulfur flux to the atmosphere and its significance in lake sulfate budgets (assuming a negligible atmospheric concentration and an ice-free period of 8 months). (ND-not detectable.) s flux (nmol S mm2 d-l) S lost via volatili7ation Lake Species Range Mean (mol yr-‘) %S* 302s cos 108-540 250 6.5 0.066 DMS 49-1,230 730 19 0.19 DMDS 155-1,190 170 4.3 0.043 MSH - - - - Total 1,150 29 0.30 Loss at fall overtumt (species below mixed layer) 7.5 0.07 239 cos ND-330 170 23 0.22 DMS 54-l ,200 600 82 0.79 DMDS ND-410 34 4.7 0.04 MSH - - - - Total 804 110 1.0 2268 cos 7 l-930 230 4.3 - DMS 106-2,370 820 16 - DMDS ND-430 100 1.9 - MSH ND-220 17 0.32 - Total 1,170 23 114 cos 85-l ,400 440 13 0.52 DMS 320-l 6,000 4,000 120 4.7 DMDS ND-7,500 1,900 54 2.2 MSH ND-25,000 6,200 190 7.4 Total 12,500 380 15 * SOAX- lost within the lake as dcfincd by I - 0 - Am = S where I is input, 0 is outflow, and Am is delta mass. Lake 302s: S = 9,810 mol yr-’ (Rudd et al. 1990); Lake 239: S = 20.9 kcq (annual mean for 1981-1983; Schindler et al. 1986); Lake 114: S = 2,450 mol (preacidihcation value from 1978; Schindler and Turner 1982). t Loss at overturn assumes that the volatile species below the mixed layer were lost to the atmosphere and chemical oxidations and conversions during mixing were negligible. Thus, it reprcscnts a maximal loss. Table 7. Concentrations of VSCs determined in various environments. (ND-not detectable.) Species Concn (nM) Environment Reference DMS 2.63 Ocean surface waters Andreae and Barnard 1984 5.77 Wetlands (Canada) Nriagu et al. 1987 1,130 Eutrophic freshwater pond Bechard and Raybum 1979 5-10 Salt pond, epilimnion Wakeham et al. 1987 4.8-22.5 Burton L. (Antarctica) (under-ice) Deprez et al. 1986 3.2 Organic L. (Antarctica) (surface) Deprez et al. 1986 11.3 Schlcinsee-hypolimnion (max value) Henatsch and Juttner 1988 0.63-1.15 Hamilton Harbor, L. Ontario (surface) Holdway and Nriagu 1987 0.10 Desjardins Canal (single date) Caron and Kramer 1989 0.016-0.161 L. Mendota (surface) Zinder and Brock 1978a 0.32-l 1 ELA lakes- surface This study ND-5.1 Red Lake area- surface This study cos 0.23-0.32 Burton L.-hypolimnion Deprez et al. 1986 “trace” Schleinsee - hypolimnion Henatsch and Juttner 1988 0.79 Desjardins Canal (single date) Caron and Kramer 1989 0.10-0.37 Delaware Bay (surface) Ferek and Andreae 1984 0.02-0.06 Pacific Ocean off Peru (surface) Ferek and Andreae 1983 ND-l.13 ELA lakes- surface This study ND-O.69 Red Lake area- surface This study MSH 42-63 Schleinsee - hypolimnion Henatsch and Juttner 1988 ND Stratified lakes (surface) This study ND-32 Unstratified lakes (surface) This study 0.76 Desjardins Canal (single date) Caron and Kramer 1989 478 Richards et al. centrations higher than most lakes included Henatsch and Juttner 1988), COS has been here (Table 7). DMS concentrations deter- detected only in the hypolimnion. The re- mined1 for other environments, such as oce- ported hypolimnetic levels of COS in Lake anic surface waters, salt marshes, and wet- Burton (Deprez et al. 1986; Table 7), how- lands, are in some cases higher and in some ever, are comparable to our surface-water cases lower than in the lakes of our study concentrations; we could not quantify COS (Table 7). in the deeper strata of our lakes due to in- The main precursor of DMS in marine terference from H2S and CH,. In eutrophic systems is DMSP, a compound considered Desjardins Canal, COS concentrations of to play an important role in algal osmoreg- -0.79 nM have been found (Caron and ulation (e.g. Vairavamurthy et al. 1985). At- Kramer 1989), which agrees well with the tempts have therefore been made to cor- range we found (Table 4). In addition, oce- relate :DMS concentrations with salinity, but anic studies have found COS at concentra- salinity alone is not a good predictor (Iver- tions comparable to the epilimnia in our son et al. 1989). This lack of fit is under- study (Table 7). scored by the freshwater lakes at the ELA, MSH was not usually found in the epilim- which have DMS concentrations similar to nia of stratified lakes but was generally con- the mean open-ocean value (2.6 nM, Table fined to the thermocline and hypolimnion 7), but which differ in salinity from the ocean where levels > 300 nM were found (Fig. 1B). by four orders of magnitude. The ratio of Henatsch and Juttner (1988) also found DMS : Chl a in the lakes [0.2-l mmol DMS MSH restricted to the hypolimnion in (mg Chl a)-‘] was much smaller, however, Schleinsee, with maximal concentrations of than in the ocean (from -4 up to 220, Iver- -41.6-62 nM (-5 x lower than maximal son et al. 1989) and consistent with ratios levels found in Lake 302 South, Fig. 1B). observed along estuarine salinity gradients CS, has been found in lake systerns previ- up to 25-307~ (Iverson et al. 1989). Thus, ously (Deprez et al. 1986; Henatsch and it is important to note that low salinity does Juttner 1988) but was not detected here. not mean that DMS will be low. Also, fresh- The two potential sites of volatile sulfur water algae do release DMS with the ad- production in freshwater systems are the dition of hydroxide (Challenger et al. 1957; sediments and the water column. We have S. Richards unpubl.), although it is not clear evidence for important contributions by if this release indicates the presence of both sources. A significant sediment source DMSP or of another DMS compound. was suggested by the following: late summer Despite the similarity in surface-water concentrations of MSH, DMS, and DMDS DMS concentrations in stratified lakes and below the mixed layer significantly greater the open ocean, the thinner diffusive bound- than concentrations in the epilimnia (Figs. ary-layer thickness in the ocean results in 1, 2); mean surface water concentrations of higher fluxes of DMS to the atmosphere than MSH, DMS, and DMDS in unstratified, estimated for our stratified lakes (Table 6). shallow Lake 114 higher than in any strat- Total flux of sulfur from unstratified Lake ified lake (Table 4); and pore-water analyses 114 was -0.96-1.6 x the estimated oceanic in Lake 114 indicating concentrations of DMS flux rate, however, per unit of area MSH and DMS -2-14 x higher than in (7,900-l 3,000 nmol me2 d-l; Barnard et al. overlying water (Table 2). MSH was de- 1982) and is comparable to the mean sulfur tected only in water with close or recent flux from a freshwater pond on Cedar Is- sediment contact such as the surface waters land, North Carolina (7,100 nmol rnh2 d- ’ ; of Lakes 114 and 303 (Fig. 3C, Table 4), Lamb et al. 1987). water below the mixed layer (Figs. 1,2), and Previously, volatile species other than Lake 226 South surface waters after fall DMS were virtually unstudied in lakes. We overturn (Fig. 3A). The production of vol- also routinely found COS in surface waters, atile sulfur species by microbial decompo- at roughly constant levels and with mean sition in the sediments, then, had an im- epilimnetic values of -0.09-0.61 nM (Ta- portant effect on both the conc,entration and ble 4). In other studies (Deprez et al. 1986; speciation of the VSCs detected. Pore water Volatile sulfur in lakes 479 of the organic-rich, flocculent sediment from column are the breakdown of sulfur-con- Lakes 114 and 303 contained concentra- taining organic matter due to zooplankton tions of MSH and DMS up to 7 X greater grazing (Dacey and Wakeham 1986), mi- than the pore water of epilimnetic, sandy crobial decomposition (Kadota and Ishida sediments in Lakes 239,240, and 302 North 1972; Bremner and Steele 1978), or abiotic (data not shown). chemical reactions. COS can be produced Other studies have also found significant microbially or may originate from the volatile sulfur concentrations in anoxic ma- chemical oxidation of sulfur-containing or- rine and lake waters and sediments (e.g. ganic matter (Ferek and Andreae 1984; Tur- Henatsch and Juttner 1988). Our data do co et al. 1980). In seawater, COS production contrast, however, with the findings of Zin- is unrelated to the activity of living organ- der and Brock (1978a,b) and Wakeham et isms (Andreae 1990). al. (1987). In the study of Wakeham et al., Seasonal patterns - As described, there sediments in a coastal salt pond were not a was no correlation between the mean epi- net source of DMS. The difference may be limnetic COS and DMS concentrations and due to disparity in microbial populations, lake size (Tables 1 and 4). Time series dem- since they found anaerobic phototrophic onstrated a tendency for the smaller lakes bacteria capable of consuming DMS in the (< 200 ha) each to have a midsummer max- hypolimnion of the salt pond. Zinder and imum in DMS concentration (Figs. 3 and Brock (1978a, b) did not find any VSCs oth- 4A) that corresponded to a midsummer peak er than H2S associated with the sediments in microbial respiration in both the water of Lake Mendota and suggested that sedi- column and in the sediments (unpubl.). ments acted as a sink for DMS. Our results Where no seasonal peak was observed (Figs. demonstrate this function is not universal 4B,D), perhaps microbial utilization or gas for lake sediments. exchange had increased in addition to pro- Although sediment production apparent- duction or no peak in microbial production ly contributed to surface-water VSC con- occurred. COS levels remained relatively centrations in our shallow lakes, there was constant in each lake throughout the study also good evidence for a water-column but had midsummer minima in 8 of the 11 source. In all the stratified lakes, regardless lakes. This pattern was unexpected because of size (7.8-34,700 ha; Table 4), there was biological activity is highest at that time and close similarity in mean surface-water DMS COS is most likely produced by photooxi- and COS concentrations. This result would dation of dissolved organic matter (Ferek not be expected unless the dominant site of and Andreae 1984) or the oxidation of re- production was the water column, with sim- duced sulfur molecules such as CS, and ilar production rates per unit of volume. mercaptans (Turco et al. 1980). Except for Lakes 114 and 303, all of these Efect of pH and SOd2- concentration-- lakes had A, : V, ratios of 0.03-O. 11 m-l An experimentally acidified lake (302 South) (Table 1). In addition, DMS concentrations was compared to an unacidified reference in Lake Ontario were similar (Table 7), and lake (226 South) to determine if there was its A, : V’ ratio is even smaller (6.63 x 1O-5 an effect on surface VSC concentrations by m-l, based on data from M. Nielson; U.S. either acidification or an increase in SOd2- EPA/Environ. Can. 1987). Lakes 114 and to levels found in acidified lakes (Table 5). 303 have A, : V, ratios of 0.59 and 0.66 (Ta- The lakes were paired because of their sim- ble 1). Therefore, in lakes with A, : V, ratios ilar bathymetries (Table 1). The surface wa- below a value between 0.11 and 0.59, water- ters of both the acidified and unacidified column production may be most important lakes had remarkably similar concentra- in determining surface DMS concentration. tions of epilimnetic COS and DMS over Formation of DMS in the water column time (Fig. 3A), suggesting that acidification may be due to the breakdown of DMSP (e.g. had not affected epilimnetic concentrations Iverson et al. 1989), but the presence of this of these two VSCs. Vairavamurthy et al. osmolyte in freshwater algae has not been (198 5) similarly found that for Hymeno- studied. Other possible sources in the water monas carterae, DMS production was in- 480 Richards et al. dependent of environmental SOd2- concen- thermocline and hypolimnion (Table 6). tration, except at growth-limiting levels. Because volatilization of COS, MSH, DMS, The rates of DMS and DMDS accumu- and DMDS in these stratified lakes was not lation below the mixed layer were markedly an important loss route of reduced S, it higher in acidified Lake 302 South (Table would not be an important source of alka- 3), although the contrast between lake types linity in acidified lakes. In Lake 114, how- must be drawn cautiously since it is based ever, the entire mixed layer is in contact on just two systems. The higher accumu- with sediment and all four organic species lation rates in Lake 302 South were unex- were elevated. Sulfur volatilization was pected, however, because DMS production therefore an important term in the SOd2- in sediments was expected to originate from budge of Lake 114 (Table 6). Sediment con- decaying organic matter such as sediment- tact thus may be an important factor in de- ing algal cells, and there has been no in- termining the percentage SOd2- lost by vol- crease in primary production or sedimen- atilization. tation (Rudd et al. 1990) that would One route of volatile sulfur loss not stud- stimulate decomposition. ied was ebullition from sediments. There- Furthermore, since surface-water DMS fore diffusive fluxes (Table 6) may under- concentrations were similar in Lakes 226 estimate total fluxes. South and 302 South (Table 4, Fig. 3A), it Contribution to the atmosphere-Volatile is unlikely the higher accumulation rate re- sulfur release from both water and land sur- sulted from the decomposition of algae with faces can be oxidized in the atmosphere and increased DMSP concentrations. There has, redeposited in forms such as SOa2- (Charl- however, been an increase in dissimilatory son et al. 1987). The total annual wet de- SOd2- reduction rates (Rudd et al. 1990) position of SOa2- in the ELA area is 5-l 0 leading to increased hypolimnetic H2S pro- mmol mm2 yr-l (Barrie and Hales 1984). duction in Lake 302 South. This enhance- The estimated flux of sulfur from deep ELA ment could lead to a greater production of lakes (Table 6) therefore represented 2-6% DMS biologically from methylation of H2S of this direct SO, 2- deposition on the lake to MSH and then to DMS (Drotar et al. surface, and shallow Lake 114 returned as 1987) or abiotically via an interaction be- much as 30-60% per unit of area. These tween CH, and H&S (Sorensen 1988). There values agree well with estimates of the per- are not enough data, however, to delineate centage contribution of DMS from wetlands a relationship between H,S and the organic to SOa2- in precipitation (3-l 8%, Barrie et sulfur species. Others (Wakeham et al. 1987) al. in prep.; up to 30%, Nriagu et al. 1987). have found that H2S is not an important Andreae and Andreae (1988) found a com- precursor in the formation of DMS. parable return to the atmosphere of precip- In acidified systems, one might expect a itated SOd2- in the Amazon basin (35OYo). greater flux of COS to the atmosphere due Interestingly, the range in lacustrine sulfur to reduced hydrolysis of this compound. emissions (804-l 2,500 nmol me2 d-l, Table Lakes 302 South and 226 South had com- 6) is similar to the range of total volatile parable concentrations of this species sulfur emissions from midcontinent soils in throughout the study (Fig. 3A, Table 4), the United States (200-9,800 nmol me2 d-l; however, and therefore the estimated fluxes Lamb et al. 1987). Sulfur volatilization from were not significantly different (Table 6). soils and lakes in remote areas may there- Relationship to SOq2- budgets - SOa2- is fore return 2-60% of the sulfur in wet SOd2- removed in these lakes by both algal uptake deposition, depending on the proportion of and SOd2- reduction (Rudd et al. 1990). A soil and lake types. possible fate of this sulfur is volatilization Volatile species such as COS and DMS from decomposition of algal sulfur or from play an important role in cloud formation reaction of H,S with organic matter. This and climate (Bates et al. 1987; Charlson et loss route accounted for only a very small al. 1987; Turco et al. 1980). To construct portion of the SO, 2--budget, however, even models accurately predicting cloud forma- including losses at fall overturn from the tion and climate change, it is necessary to Volatile sulj “ur in lakes 481 know natural fluxes of volatile sulfur species ography and bathymetry of selected lake basins, from various aquatic and terrestrial ecosys- Experimental Lakes Area, northwestern Ontario. J. Fish. Res. Bd. Can. 28: 139-155. tems. Based on our data, we estimate the CARON, F., AND J. R. KRAMER. 1989. Gas chro- total organic volatile sulfur flux from Ca- matographic determination of volatile sulfides at nadian freshwater lakes to be 5-76 x lo6 trace levels in natural freshwaters. Anal. Chem. kg S yr-l. In particular, if the data from 61: 114-l 18. F., CHALLENGER, R. BYWOOD,P. THOMAS,AND B. J. Lakes 114 and 303 (Tables 4 and 6) char- HAYWARD. 1957. Studies on biological methyl- acterize other shallow lakes, there could be ation. 17. The natural occurrence and chemical a regionally important flux of VSCs in areas reactions of some thetins. Arch. Biochem. Bio- where such lakes are abundant, as in the phys. 69: 514-523. Hudson Bay lowlands, the largest wetland CHARLSON,R. J., J. E. LOVELOCK,M. 0. ANDREAE, AND S. G. WARREN. 1987. Oceanic phytoplank- in North America. Furthermore, the con- ton, atmospheric sulphur, cloud albedo and cli- struction of large, shallow reservoirs may mate. Nature 326: 655-66 1. increase inputs of VSCs to the atmosphere. DACEY, J. W. H., AND STUARTG. WAKEHAM. 1986. Already the surface area of artificial reser- Oceanic dimethylsulfide: Production during zoo- voirs built in Canada during the 1970s is plankton grazing on phytoplankton. Science 233: 1314-1316. equivalent to the surface area of Lake On- DEPREZ,P.P., P.D. FRANZMANN,ANDH.R. BURTON. tario (R. He&y pers. comm.), and this source 1986. Determination of reduced sulfur gases in is likely to increase in the future as hydro- Antarctic lakes and seawater by gas chromatog- electric development progresses. raphy after solid adsorbent Preconcentration. J. Chromatogr. 362: 9-2 1. DROTAR,A. M., G. A. BURTON,JR., J. E. TAVERNIER, References AND R. FALL. 1987. Widespread occurrence of ANDREAE,M. 0. 1990. Ocean-atmosphere interac- bacterial thiol methyltransferases and the biogenic tions in the global biogeochemical sulfur cycle. emission of methylated sulfur gases. Appl. Envi- Mar. Chem. 30: l-29. ron. Microbial. 53: 1626-l 63 1. - AND T. W. ANDREAE. 1988. Thecycleofbio- S., EMERSON, W. BROECKER, AND D. W. SCHINDLER. gehic sulfur compounds over the Amazon basin 1973. Gas-exchange rates in a small lake as de- 1. Dry season. J. Geophys. Res. 93: 1487-1497. termined by the radon method. J. Fish. Res. Bd. -, AND W. R. BARNARD. 1983. Determination Can. 30: 1475-1484. of trace quantities of dimethylsulfide in aqueous FEE, E. J. 1979. A relation between lake morphom- solutions. Anal. Chem. 55: 608-6 12. etry and primary productivity and its use in in- -,AND- 1984. The marine chemistry of terpreting whole-lake eutrophication experiments. DMS. Mar. Chkm. 14: 267-279. Limnol. Oceanogr. 24: 401-4 16. - AND OTHERS. 1985. Dimethyl sulfide in the FEREK,R. J., AND M. 0. ANDREAE. 1983. The su- marine atmosphere. J. Geophys. Res. 90: 12,89 l- persaturation of carbonyl sulfide in surface waters 12,900. of the Pacific Ocean off Peru. Geophys. Res. Lett. BARNARD,W.R.,M.O. ANDREAE,W.E. WATKINS,H. 10:393-396. BINGEMER,ANDH.-W. GEORGII. 1982. The flux -,AND- . 1984. Photochemical produc- of dimethylsulfide from the oceans to the atmo- tion of carbonyl sulphide in marine surface waters. sphere. J. Geophys. Res. 87: 8787-8793. Nature 307: 148-150. BARRIE, L. A., AND J. M. HALES. 1984. The spatial HAYDUCK, W., AND H. LAUDIE. 1974. Prediction of distributions of precipitation acidity and major diffusion coefficients for nonelectrolytes in dilute ion wet deposition in North America during 1980. aqueous solutions. AIChE J. 20: 611-615. Tellus 36: 333-355. HENATSCH,J. J., AND F. JUTI~\IER. 1988. Capillary BATES,T. S., R. J. CHARLSON, AND R. H. GAMMON. gas Chromatographic analysis of low-boiling or- 1987. Evidence for the climatic role of marine ganic sulphur compounds in anoxic lake-water by biogenic sulphur. Nature 329: 3 19-32 1. cryoadsorption. J. Chromatogr. 445: 97-105. M. BECHARD, J., AND W. R. RAYBURN. 1979. Volatile R. HESSLEIN, H. 1976. An in situ sampler for close organic sulfides from freshwater algae. J. Phycol. interval pore water studies. Limnol. Oceanogr. 21: 15: 379-383. 912-914. BREMNER,J. M., AND C. G. STEELE. 1978. Role of HOLDWAY,D.A., AND J.O. NRIAGU. 1987. A purge microorganisms in the atmospheric sulfur cycle, and trap gas Chromatographic method for di- p. 155-201. In M. Alexander [ed.], Advances in methyl sulfide in freshwater. Int. J. Environ. Anal. microbial ecology. V. 2. Plenum. Chem. 32: 177-186. BROECKER,W. S., T.-H. PENG, G, MATHIEU, R. IVERSON, R.L.,F.L. NEARHOOF,ANDM.O.ANDREAE. HESSLEIN,AND T. TORGERSEN. 1980. Gas ex- 1989. Production of dimethylsulfonium propio- change rate measurements in natural systems. Ra- nate and dimethylsulfide by phytoplankton in es- diocarbon 22: 676-683. tuarine and coastal waters. Limnol. Oceanogr. 34: BRUNSKILL,G. J., AND D. W. SCHINDLER. 1971. Ge- 53-67. 482 Richards et al. KAD~TA, H., AND Y. ISHIDA. 1972. Production of level by tenax trapping and gas chromatography. volatile sulfur compounds 159 1 by micro-organ- J. Chromatogr. 366: 205-216. isms. Annu. Rev. Microbial. 26: 127-138. A. TORRES, L., P. J. MAROULIS,A. B. GOLDBERG, AND LAMB, B., H. WESTBERG, ALLWINE, L. BAMESBER- G. A. R. BANDY. 1980. Atmospheric OCS mea- GER,AND A. GUENTHER. 1987. Measurement of surements on project Gametag. J. Geophys. Res. biogenic sulfur emissions from soils and vegeta- 85:7357-7360. tion: Application of dynamic enclosure methods TURCO, R. P., R. C. WHITTEN, 0. B. TOON, J. B. with Natusch filter and GC/FPD analysis. J. At- POLLACK,AND P. HAMILL. 1980. OCS, strato- mos. Chem. 5: 469-491. spheric aerosols and climate. Nature 283: 283- LEWIS,W. K., AND W. C. WHITMAN. 1924. Principles 286. ofgas adsorption. Ind. Eng. Chem. 17: 1215-l 220. U.S. ENVIRONMENTAL PROTECTION AGENCYAND ENVI- MILLIKEN, G. A. 1989. Analysis of covariance: Mul- RONMENT CANADA. 1987. The Great Lakes: An environmental atlas and resource book. tiple covariates, p, 51-60. In Proc. 14th Annu. VAIRAVAMURTHY, A., M. 0. ANDREAE, AND R. L. SAS Users Group Int. IVERSON. 1985. Biosynthesis of dimethylsulfide NRIAGU, J. O., D. A. HOLDWAY, AND R. D. COKER. and dimethylpropiothetin by Hymenomonns cur- 1987. Biogenic sulfur and the acidity of rainfall terse in relation to sulfur source and salinity vari- in remote areas of Canada. Science 237: 1189- ations. Limnol. Oceanogr. 30: 59-70. 1192. J. WAKEHAM,S. G., B. L. HOWES, W. H. DACEY, R. P. H. PANOFSIKY, A., AND J. A. DUTTON. 1984. The AND SCHWARZENBACH, J. ZEYER. 1987. Biogeo- power law for the wind profile, p. 130-131. In chemistry of dimethylsulfide in a seasonally strat- Atmospheric turbulence. Models and methods for ified coastal salt pond. Geochim. Cosmochim. Acta engineering applications. Wiley. 51: 1675-1684. RUDD, 3. W. M., C. A. KELLY, D. M. SCHINDLER, AND WILHELM,E., R. BATTINO,AND R. J. WILCOCK. 1977. M. TURNER. 1990. A comparison of the acidi- Low-pressure solubility of gases in liquid water. fication efficiencies of nitric and sulfuric acids by Chem. Rev. 77: 219-262. two whole-lake addition experiments. Limnol. WILKE, C. R., AND P. CHANG. 1955. Correlation of Oceanogr. 35: 663-679. diffusion coefficients in dilute solutions. AIChE J. SCHINDLER, W., AND M. A. TURNER. 1982. Bio- D. 1:264-270. logical, chemical and physical responses of lakes ZINDER,S. H., AND T. D. BROC:K. 1978~. Production to experimental acidification. Water Air Soil Pol- of methane and carbon dioxide from methane thi- 01 and dimethyl sulphide by anaerobic lake sedi- lut. 18: 259-27 1. ments. Nature 273: 226-228. - - M. P. STAINTON,AND G. A. LINSEY. -, AND -. 1978b. Methane, carbon diox- 1986. Natural sources of acid neutralizing capac- ide, and hydrogen sulfide production from the ter- ity in low alkalinity lakes of the Precambrian Shield. minal methiol group of methionine by anaerobic Science 232: 844-847. lake sediments. Appl. Environ. Microbial. 35: 344- J. SORENSEN, 1988. Dimethylsulfide and methane thi- 352. 01an sediment porewater of a Danish estuary. Bio- geochemistry 6: 20 l-2 10. Submitted: 26 December 1989 TANGERMAN, 1986. Determination of volatile sul- A. Accepted: 5 June 1990 phur compounds in air at the parts per trillion Revised: 21 February 1991