Coastal salt marshes as global methyl halide sources from determinations of intrinsic production by marsh plants

Emissions of CH3Cl, CH3Br and CH3I were measured biweekly for 12‐ to 24‐month periods between March 2002 and March 2005 from monospecific stands of four dominant southern California coastal salt marsh plants. These measurements revealed large inherent differences between species and more detailed patterns of seasonal production than previously reported. Marsh plants displayed intrinsic abilities to produce methyl halides. Salt marsh plants produced 92% of CH3Cl and 90% of CH3Br emitted and only 41% of the emitted CH3I. Unvegetated areas emitted 7.9% of CH3Cl, 9.9% CH3Br, and 59% of the emitted CH3I. The accuracy of the estimated methyl halide emissions from a coastal marsh and probably other ecosystems can be dramatically improved with increasing the number of species being measured and including emission from barren (mudflats and soil) areas. Estimates of global salt marsh emissions based on vegetated and barren area are 130, 21, 5.5 (mg m−2 yr−1) for CH3Cl, CH3Br, and CH3I, respectively, or 1.2, 3.9, and 0.8% of total global fluxes of these gases.


Introduction
[2] Methyl chloride (CH 3 Cl), methyl bromide (CH 3 Br) and methyl iodide (CH 3 I) are chemically important atmospheric constituents. They are also significant vectors of halogen transport (especially for bromine and iodine) from the terrestrial and marine environment to the atmosphere. Accurate knowledge of CH 3 Br and CH 3 Cl sources is particularly needed because their relatively long residence times allow them to reach the lower stratosphere where halogen radicals produced from their photodissociation and OH attack catalytically destroy ozone. Despite its short residence time, CH 3 I can contribute to stratospheric ozone destruction [Solomon et al., 1994], strongly affect tropospheric chemistry, especially of the marine boundary layer, and indirectly promote marine aerosol and cloud condensation nuclei formation [Davis et al., 1996;Carpenter et al., 1999;O'Dowd et al., 2002].
[3] Methyl halides (CH 3 X) are naturally produced by photoautotrophs and certain fungi. The major mechanism of biosynthesis is enzymatic [Wuosmaa and Hager, 1990], although an additional abiotic mechanism in leaf litter has been proposed [Hamilton et al., 2003]. The relative environmental importance of higher plant methyl halide emissions, especially from marshes [Varner et al., 1999;Rhew et al., 2000;Cox et al., 2004], agriculture [Redeker et al., 2000] and tropical plants [Yokouchi et al., 2002] has only recently been appreciated. Vegetation from temperate coastal salt marsh plants was estimated to produce 10% of the global source strength for the two gases CH 3 Cl and CH 3 Br [Rhew et al., 2000].

Incubation Conditions
[5] Methyl halide emissive flux was determined by enclosing a single plant species in an incubation chamber and periodically withdrawing a gas sample into a valved preevacuated canister or lockable syringe. The bottom was sealed by the surrounding mud and soil (>0.5 cm deep). Different sized chambers were used to accommodate the different sized plants. The smallest was a 4L modified glass beaker (14 cm ID) with sampling port, thermometer port and an internal battery powered fan. The sampling port contained a 20-cm-long stainless steel tube (1/16 inch OD, dead volume, 0.1 mL) which extended into the center of the chamber. The portion extending outside the chamber had a glass sleeve with septum for withdrawing a 30-mL gas sample into a syringe. The larger chambers were made of stackable sections of polycarbonate tubes (29.5 cm ID, 30 cm height) with polyethylene connectors, mounted on a permanent polyvinyl chloride chamber bases (15 cm height) to yield sealed chambers volumes of 24 -90 L [Redeker et al., 2000]. These large chambers had a sealed lid containing two ports, one for sampling and the other for a thermometer. The sampling line was a 30-cm-long stainless steel (1/4 inch OD, <5 mL dead volume) tube extending in to the center of the chamber with an Ultra-Torr 1 fitting for sampling with the attachment of a 500-mL stainless steel canister. These chambers did not inherently emit or absorb significant amounts of methyl halides [see Redeker et. al., 2000, note 8]. Control incubations were performed similarly by positioning the same type of chamber over unvegetated areas close to, and at the same elevation as, the other chamber avoiding all aerial plant material. The green algae Enteromorpha sp., at times present on marsh mudflats, was avoided as a Spartina control site.
[6] Permanent chamber bases were used for Spartina and Salicornia incubations because bimonthly removal of Spartina foliosa was destructive to the habitat and because the woody basal and migrating stems of Salicornia virginica made it difficult for the glass chamber to form a seal in the soil without damaging the plants. No biomass was harvested from within the permanent bases after incubations. Control incubations used chambers without permanent bases. We began our study of Spartina by placing the chamber base over new shoots on the immediate edge of a pure stand. After 2 years, at the end of our measurements, the chamber base was well inside the Spartina bed, >3 m from the growing edge. Independent of the Salicornia incubations using the larger chambers, nine additional incubations of Salicornia were performed using the small glass chamber followed by tissue harvesting to determine biomass specific production rates. Small glass chambers were used for Batis and Frankenia incubations. After the incubation period, all aerial Batis and Frankenia biomass, including woody biomass, was harvested for fresh weight determinations.
[7] All incubations were performed between 0930 and 1400 local time, at a period of low tide. Weather conditions and air temperatures were recorded. Incubation sampling duration and frequency ranged from 5 to 15 min at 1-to 5min intervals depending on the plant species and chamber used.

Methyl Halide Analysis
[8] Analysis was performed as described by Redeker et al. [2000] using gas chromatography. A known amount (by pressure) of chamber air sample was preconcentrated in a glass-bead-packed stainless steel loop at 77 K. It was then directly injected onto a 50 m Â 0.53 mm Poraplot-Q column (Chrompack) and detected by electron-capture (Hewlett Packard 5890 Series II). Standards in N 2 (9.68 ppb CH 3 Br and 479 ppb CH 3 Cl; 25.4 ppb CH 3 I) from Scott Specialty Gases were injected at low pressures to create a linear calibration curve with a range of absolute amounts bracketing sample amounts.

Halide Extraction and Analysis
[9] Halide analysis was intermittently performed on tissue harvested throughout the study period. The chambered Batis and Frankenia tissue harvested for biomass determination was so analyzed. Spartina and Salicornia tissue immediately adjacent to the permanent chamber base was used for halide analysis. Tissue samples were weighed, rinsed in deionized water to remove any external salts and dried to constant weight at 55°C. The dried tissue was weighed, milled into a fine powder (40-mm mesh) and stored in a capped glass vial under desiccation. Solid content of tissue (% dwt) was calculated.
[10] Halides were extracted from the milled plant tissue by adding the powder (0.5 g) to 50 mL of boiling deionized water in a 150-mL flask for 10 min. Boiled tissue was then vacuum filtered and the filtrate collected and stored frozen until halide analysis. The extraction procedure was repeated 5 times on the same sample to ensure complete halide removal from the tissue. Greater than 98% Cl À , 97% Br À , and 99% I À was removed from the tissue after three sequential extractions and the three-step extraction procedure became the standard method. Values were not corrected for the 1 to 3% loss. The halide concentration in each extract was determined using a specific ion meter/autotitrator in conjunction with the appropriate halide electrode using Gran's known addition method [Redeker and Cicerone, 2004;Gran, 1952]. This method allowed for the measurement of halides near their limit of detection: 5 Â 10 À5 M Cl À ; 5 Â 10 À6 M Br À ; and 5 Â 10 À8 M I À (Thermo Orion specifications) and minimized the interfering effects of other ions and compounds. Following the autotitration the calculated initial concentration reported in mM was converted to tissue halide content as percent dry weight. The mean standard check for accuracy (with precision as %S.D.) of the analysis was 98% for Cl À (5.7%), 88% for Br À (11%) and 95% for I À (8.0% [Ralph and Manley, 2006]).

Results
[11] All emission rates reported are net rates (plant chamber minus control) unless otherwise indicated. The two-point averaged net emission rates normalized to the marsh area enclosed by the chamber, from sampling twice a month are shown in Figures 2a through 2d Figure 2a, the actual measured CH 3 Br and CH 3 I fluxes from Spartina have been multiplied by factors of 4 and 2, respectively, to give the plotted values.
[12] We analyzed the data by determining the linear correlation coefficients between monthly emission rates and monthly mean temperature, and monthly insolation. (Fitting the data to several nonlinear regression models such as exponential and rectangular hyperbola, yielded poorer correlations.) In no instances did the same effector correlate equally with emissions of all three methyl halides from a given plant. Only weak correlations (r < 0.7) were observed, except for the correlation of CH 3 I emission to temperature by Spartina (r = 0.83) and Batis (biomass-based; r = 0.75). Mean monthly insolation was most often more highly correlated to emissions than mean monthly temperature. Mean monthly temperature changes, however, rarely exceeded 10°C and were usually not highly correlated with methyl halide emissions. In most instances, either temperature or insolation, for the month in which the emissions were measured, displayed the highest levels of correlation. Occasionally, it was the previous month's temperature or insolation that more closely correlated with emissions. Only at shifts of À1 month was r sometimes greater than r when no shift was applied. A shift in the emission data to one month previous insolation resulted in an increased correlation for CH 3 Br produced by Batis and Spartina only (r = 0.7 for both). Salicornia emissions were poorly correlated with monthly mean temperature and insolation (r < 0.36).
[13] The dramatic increase in biomass normalized CH 3 Cl and CH 3 Br emission by Frankenia was correlated with flowering ( Figure 3a). Flowering in Batis corresponded to a striking increase in biomass normalized CH 3 I emission and a secondary spike in CH 3 Br emission (Figure 3b  Salicornia flowering corresponded with a peak in CH 3 I emissions ( Figure 2b).
[14] Initial chamber temperature and average chamber temperature during the incubation period were weakly or not correlated to emissions for all three methyl halides for all plants. Spartina CH 3 Br and CH 3 I emissions during non-flowering periods were poorly correlated to average chamber temperature (r = 0.5 and 0.3, respectively), however, CH 3 Cl emissions were weakly correlated (n = 30; r = 0.70). Because flowering appeared to be an important determinant for certain methyl halide emissions, the emission data for each plant during the flowering period were correlated to average chamber temperature. In most instances the correlation was extremely weak (r < 0.4). Average chamber temperature was moderately correlated with emissions during flowering for (1) CH 3 I emissions by Spartina during 2003 (n = 8; r = À0.71), (2) CH 3 Br emissions by Salicornia (n = 8; r = À0.71) and (3) CH 3 Cl emissions by Frankenia during 2003 (n = 8, r = +0.69). However, CH 3 I emissions by Spartina during the flowering period of 2004 and CH 3 Cl emissions by Frankenia during the flowering period of 2002 were not correlated with average chamber temperature. Methyl halide emissions by Spartina (located low in the marsh) and Frankenia (located in mid and high marsh) were not correlated with height of last high tide, time since last high tide or tidal strength (height of last high tide Ä time since last high tide; r < 0.05).
[15] Annual emissions of CH 3 Cl and CH 3 Br from the soil and mud (control incubations) were smaller than their emissions from the plant by approximately an order of magnitude (Table 1). Methyl iodide emissions from Spartina and Salicornia and their controls were similarly related. The area surrounding Spartina (mud) and Salicornia (soil/mud) produced low levels of CH 3 I throughout the year with peak emissions in the late summer. The emission of CH 3 I from the soil where Batis and Frankenia were located was greater than or equal to plant emissions and peaked in late summer. Net consumption of CH 3 I by soil or mud rarely occurred, whereas consumption of the other two gases occurred more frequently, primarily in the winter.
[16] Annual plant methyl halide emission rates normalized to grams fresh weight and dry weight (gfwt and gdwt, respectively) are shown in Table 2 for these four species. The succulent plants (i.e., Batis and Salicornia) have very high water content compared to nonsucculent plants (Spartina and Frankenia), therefore, a more accurate comparison of their intrinsic methyl halide emission rates is revealed when expressed per gram dry wt. Annual biomass normalized methyl halide emission rates for Batis and Frankenia are the mean of all direct measurements. The annual biomass-normalized methyl halide emission rates for Salicornia and Spartina were indirectly determined because the plants were not harvested. The yearly mean Salicornia biomass density of 750 gdwt m À2 , determined at Mugu Lagoon (Ventura County, California [Boyer et al., 2001]), and the yearly mean Spartina biomass density of 171 gdwt m À2 , determined from the Tijuana Estuary (San Diego County, California [Zedler and Nordby, 1986]), were therefore used. Biomass normalized emissions (n = 9) from Salicornia were also directly measured using the small glass chamber followed by harvesting. The rates were similar to those in Table 2 ranging from 0.5-18 mg CH 3 Cl gdwt À1 yr À1 , 0-1.3 mg CH 3 Br gdwt À1 yr À1 , and 0.11-3.5 mg CH 3 I gdwt À1 yr 1 , even though it was difficult to ensure a tight seal (the reason we changed to a permanent base). Spartina biomass normalized rates were determined 3 times (November, June and August) using biomass density determinations near the site where emission fluxes were measured. The rates were similar to those in Table 2 ranging from 0 to 38 mg CH 3 Cl gdwt À1 yr À1 , 1.4-12.8 mg CH 3 Br gdwt À1 yr À1 , and 4-13 mg CH 3 I gdwt À1 yr 1 .
[17] Tissue halide content for these plants is also shown in Table 2. Although the infrequent sampling probably masked any significant seasonal changes, values are shown to confirm that halides are present in high concentrations. The succulent species Batis and Salicornia contain more halides than the nonsucculent plants.
[18] Estimates of total annual emissions from two coastal salt marshes, one in the Upper Newport Bay (UNB) and the other in the Tijuana Estuary (TE, San Diego County, California), are based on descriptions of marsh area and percent plant cover of each marsh region (UNB [Vogl, 1966; U.S. Army Corp of Engineers et al. 2000]; TE [Zedler, 1977]). The description of each marsh varies dramatically and therefore different techniques were used to estimate total plant cover. Annual emissions (mg m À2 yr À1 ) were calculated for the four dominant marsh plant species studied from integration of the seasonal emission profile. The limited data set of Rhew et al. [2000] for another marsh plant present, Monathochloe littoralis was also incorporated. It was assumed that emissions from these plants overwhelm emissions from any other species present. Furthermore, it is assumed that emissions are constant throughout a 24-hour day. Rhew et al. [2002] have shown putative diurnal CH 3 Cl and CH 3 Br emissions from Batis. The calculated yearly mean emission from the Spartina controls was used for an estimate of emission from unvegetated marsh mudflat of UNB.
[19] Vogl [1966] used a customary zonal classification scheme: littoral zone or salt marsh proper which is subject to tidal submersion, splash and spray, and the maritime zone comprising the bluff and dune vegetation. He also represented plant cover in the UNB salt marsh as a percent cover in three littoral zones, lower, upper and high, but he did not attempt to quantify the area of each zone. For this we relied upon the description in a report by U.S. Army Corp of Engineers (Upper Newport Bay Ecosystem Feasibility Study Final Report, 2003, available at http:// www.ocwatersheds.com/watersheds/pdfs/Ecosystem_ Restoration_Feasibility_Study.pdf) (Tables 3a and 3b) which states that there is 9.7 Â 10 5 m 2 (240 acres) of intertidal mudflats. On the basis of Vogl's description we calculate 3.2 Â 10 5 m 2 barren area or mudflat in the low salt marsh (Table 3b). This discrepancy is based on Vogl's description ''the littoral zone was delimited by the plant species present, since plants were considered better indicators of zonation than physical factors. . ..'' The lower limit  did not therefore include the mudflat exposed during low tides below the range of Spartina; the outer edge of Spartina delineated the lower marsh. The Corps of Engineers defined intertidal mudflats as bare areas between À1.3 m and +0.46 m mean sea level (MSL) and the low salt marsh between +0.46 m and +0.91 m above MSL. (J. B. Zedler empirically defined the lower marsh edge at +0.3 m MSL). Therefore, Vogl's barren area is most likely near or above +0.3 m MSL and excludes much of the estuarine mud that the Corps of Engineers included. Using the plant area coverage for UNB salt marsh and the calculated annual methyl halide emissions for each species, we estimated the total methyl halide flux coming from these plants ( Table 4). The total contribution of each plant species is dependent on the emission rate from its tissue and the total area that it covers.
[20] An estimate of the total methyl halide emission from the entire Upper Newport Bay littoral zone, which, therefore, excludes salt pan (2.8 Â10 4 m 2 ), uplands (in the sense of Vogl's maritime zone, 2.3 Â 10 5 m 2 ) and fresh water marsh (7.1 Â 10 5 m 2 ), and which must exclude open water (8.5 Â 10 5 m 2 ) production for which we have no estimate, is shown in Table 5. To account for emissions from plants we did not measure, we assumed an emission rate equal to the total area weighted emission from the known plants (S plant emission, g yr À1 , normalized to total area covered by these plants; e.g., 410 mg CH 3 Cl m À2 yr À1 ). Estimated annual contribution by plants was about 91% for CH 3 Cl and CH 3 Br. Plants accounted for only 41% of CH 3 I emissions. The emission rate from mid and an upper marsh soil was determined from the mean emission from the controls from Salicornia, Batis and Frankenia. Soil in the mid/upper marsh accounted for 50% of CH 3 I emissions, much greater than that emitted from estuarine mud. The emission rate for estuarine mud was determined from the emissions from the Spartina control. Total mudflat area was equal to that of the ''intertidal mudflat'' plus the ''barren low'' littoral zone (Tables 3a and 3b). Mudflats accounted for less than 10% of the total CH 3 I released and less than 1% for the other gases.
[21] In estimating the total coverage of each plant species in TE salt marsh we relied on the detailed measurements of Zedler [1977] (see the auxiliary material 1 ).

Discussion
[22] This work extends and refines the research of Rhew et al. [2000] in the estimation of methyl halide emission from coastal salt marshes. This present study differs signif-icantly from the former in that only single plant species were chambered (no species mixtures), transparent chambers were used allowing for plant photosynthesis, and the sampling frequency was much greater and covered an entire year or more. Measuring emissions from individual species permitted direct interspecies comparisons, which, in conjunction with plant coverage data, allowed for a more accurate assessment of salt marsh methyl halide emissions. We also included production of methyl halides from barren areas in our estimations.
[23] The use of transparent chambers ensured that the plants were photosynthetically active. If methyl halide production is coupled to C fixation, the use of transparent chambers would maintain this relationship. Methyl halides are formed enzymatically (methyl transferases) using Sadenosyl-L-methionine (SAM) as the methyl donor to methylate halides [Wuosama and Hager, 1990;Manley, 2002]. Such methylations, however, are generally not directly coupled to primary metabolism (e.g., photosynthesis). The lack of coupling between methyl halide production and photosynthesis is supported by the observations that algal and rice methyl halide production is not enhanced in the light [Manley and Dastoor, 1987;Redeker and Cicerone, 2004]. Diurnal patterns in methyl halide emissions have been reported for various ecosystems including rice paddies [Redeker and Cicerone, 2004] and salt marshes [Rhew et al., 2000[Rhew et al., , 2002. However, some of these patterns may have been in response to changing ambient temperature rather than irradiance [Redeker and Cicerone, 2004].  Open water 8.5 Â 10 5 (209 acres) Intertidal mudflat 9.7 Â 10 5 (240 acres) Low salt marsh 5.9 Â 10 5 (146 acres) Middle salt marsh 6.2 Â 10 5 (154 acres) High salt marsh 4.0 Â 10 4 (10 acres) Total salt marsh 1.3 Â 10 6 [24] Higher sampling frequency and determination of emissions on the basis of chambered area and plant biomass allowed for a more accurate description of the seasonal patterns of methyl halide emissions. Emissions showed a pronounced seasonality caused not only by seasonal changes in biomass density, but also due to a physiological activation of the emission process. Such an approach identified flowering as an important event (see below).

Seasonal Emission Pattern
[25] Our degree of sampling allowed for the detection of a strong seasonal signal and monthly differences. Monthly emissions of CH 3 Cl, CH 3 Br from Spartina and Frankenia were more highly correlated with monthly insolation than monthly temperature because of its dramatic effect on seasonal biomass via growth. These two species show a pronounced biomass change with the seasons, with presumably day length having the greatest effect on growth. During the winter, aboveground biomass of Spartina senesced and disappeared, with new shoots emerging. During the winter, green tissue of Frankenia also disappeared leaving only woody twigs. Whereas CH 3 I emission was strongly correlated to monthly temperature for Spartina, that was not the case for Frankenia. Clearly CH 3 I emissions from these two species are controlled by different physiological processes than are the emissions of the other methyl halides. Batis showed little change in biomass with seasons, and the areabased emissions were less highly correlated with either temperature or insolation as compared to the other plants. Biomass-based CH 3 I emissions by Batis do show a strong temperature influence suggesting greater metabolic activity in the production of this gas during the warmer months. Salicornia is a plant with a woody base supporting succulent tissue and the ratio of woody stem to succulent tissue increases dramatically in the winter [Boyer et al., 2001]. In the winter there was a decrease in CH 3 I emissions; CH 3 Br  Total area equal to that of intertidal mudflat plus barren lower littoral (Table 3); total emission rate calculated as integrated yearly emission from Spartina controls (Table 1). c Total area equal to mid and high littoral; total emission calculated as mean of integrated yearly emission from Salicornia, Batis and Frankenia controls (Table 1). emissions showed a slight decline and CH 3 Cl emission were relatively unchanged. There was much more biomass and succulent tissue present in the spring of 2005 than present in the spring and summer of 2004 which corresponded to overall greater methyl halide emissions in the spring of 2005. Although there were poor correlations between methyl halide emissions from Salicornia and monthly temperature, CH 3 Cl and CH 3 Br emissions did correlate slightly with insolation showing succulent tissue dependency. A stronger correlation may have been masked by the large increase in emissions and biomass that occurred in 2005 after heavy summer rains.
[26] We could not correlate emissions with tidal or rainfall patterns on a daily, weekly or monthly scale. Presumably tides and rainfall did not influence emissions directly, apart from their effect on overall plant health and growth. In most cases, mean chamber incubation temperatures were weakly correlated to daily emissions, even when those emissions were separated into flowering and nonflowering categories. Positive correlations between emissions (e.g., CH 3 Cl emissions from Spartina, nonflowering) and temperature may have been due to the effect of temperature on diffusion and the rate of enzyme catalyzed methyl halide synthesis. Negative correlation between emissions (e.g., CH 3 I emissions by Spartina during 2003 flowering and CH 3 Br emissions by flowering Salicornia) and temperature may have been a result of inhibition of methyl halide biosynthesis at high temperature. The highest chamber temperatures were not generally associated with average monthly temperatures (e.g., chamber temperature for Spartina of 34.6°C occurred 12/03) and thus did not amplify the seasonal signal. The negative correlations between temperatures and emissions during flowering period may have resulted in an underestimation of emission peaks associated with flowering. The use of temperature controlled chambers would have been beneficial.

Intrinsic Plant Variation in Methyl Halide Emissions
[27] Calculating the yearly biomass specific methyl halide emission rates, especially those based on dry weights (Table 2) demonstrates an inherent difference in emissions amongst species. Frankenia and Batis biomass specific emission rates for CH 3 Cl and CH 3 Br were at least an order of magnitude greater than the rates for the other two species. The biomass specific emission rates for CH 3 I by Frankenia were also greater than the rates from Spartina and Salicornia, but the CH 3 I emissions from Batis exceeded only those of Salicornia. The values were empirically determined for Batis and Frankenia; they were calculated based on seasonal determinations of biomass for Salicornia and Spartina from other studies. The biomass of above ground tissue of S. virginica in a southern California salt marsh reaches a maximum in the summer ($1000 gdwt m À2 , 20 -40% succulent tissue) compared to the winter ($500 gdwt m À2 , 3% succulent tissue), with slightly more succulent tissue peaking in June and woody tissue peaking in August [Boyer et al., 2001]. Biomass normalized emissions calculated for Salicornia (Table 2) were very similar to those few direct measurements. The mean biomass density from these limited determinations was 649 gdwt m À2 , which is similar to the yearly mean of 750 gdwt m À2 [Boyer et al., 2001], validating our approach. S. foliosa aerial biomass (using height and numbers of stems as a proxy) in a southern California salt marsh peaks in July and again in September when fruiting, and is at a minimum throughout the winter [Zedler and Nordby, 1986;Covin and Zedler, 1988]. The limited measurements of Spartina biomass densities (570 gdwt m À2 ) was 3 times the yearly mean determined from the work of Zedler and Nordby [1986] but well within reported growing season maximums of 211 -898 g dwt m À2 [Zedler and Nordby, 1986;Covin and Zedler, 1988]. The mean annual biomass density calculated for Batis and Frankenia were 520 and 470 gdwt m À2 , respectively.
[28] Normalizing emissions to the amount of chambered aerial tissue biomass revealed seasonal emissions inherent to each plant species independent of biomass density (Figure 3). This increase in methyl halide emission demonstrates physiological control of the process. Such control could involve the increase in the synthesis of the methyl transferase(s) involved in methyl halide production and/or increased activity of the enzyme. The seasonal emission peak of certain methyl halides corresponded with flowering in Batis (CH 3 I in August) and Frankenia (CH 3 Cl and CH 3 Br in late spring early summer). The CH 3 I emission peak for Spartina in August 2004 and for Salicornia in October 2004 also corresponded with flowering, but because data are not normalized to biomass but to chambered area, the peaks may be influenced by biomass increases. Clearly flowering is related in the production of certain methyl halides, depending on the plant species. Flowering in rice corresponded to a large increase in CH 3 Br emissions and a minor peak in CH 3 I emissions [Redeker et al., 2000].
[29] Plants contain many different types of SAM utilizing methyltransferases defined by the different organic substrates that they methylate [Manley, 2002]. A variety of purified methyltransferases (caffeic acid OMT, caffeoyl-CoA OMT, flavanol OMT, salicylic acid carboxyl MT) have been shown to catalyze the methylation of bromide in the presence of SAM (S. L. Manley, unpublished data, 2002). The latter enzyme had the highest specific activity in producing CH 3 Br. Salicylic acid carboxyl MT isolated from petals of the plant Clarkia breweri, functions to create the volatile compounds associated with this flower to attract insects [Ross et al., 1999]. Such a methyl transferase may be present during flowering of these salt marsh species and secondarily methylate the respective halides. Although Spartina is a wind pollinator and may not produce volatile insect attractants, increased CH 3 I emissions was correlated with flowering for both years studied.
[30] Intrinsic differences in methyl halide production by plants are also revealed by examination of the molar ratios of emissions (Table 6). Each plant has its own distinct ratio pattern. However, Spartina and Salicornia have similar CH 3 Cl: CH 3 Br ratios as do Frankenia and Batis. The latter two species also share similar CH 3 Cl:CH 3 I and CH 3 Br:CH 3 I ratios reflecting similar emission fluxes for all compounds.
[31] The linear correlation between CH 3 Cl and CH 3 Br monthly emissions was moderate to high (r > 0.7) for all plant species except Salicornia, suggesting synchronized production (Table 6). Frankenia showed the highest correlation between CH 3 Cl and CH 3 Br (r = 0.98) because the emission of these gases were enhanced during flowering, whereas with the other species only CH 3 I emissions were primarily enhanced during flowering. The correlation between CH 3 Cl and CH 3 Br for a Tasmanian coastal wetland site (Cape Grim, 40°S, 145°E) dominated by the succulent halophyte Pachycornia arbuscula (basionym Salicornia arbuscula) was also high (r = 0.86 [Cox et al., 2004]), however, we found the correlation based on our integrated mean with related species S. virginica low (r = 0.35).
[32] Except for the species Salicornia [Ralph and Manley, 2006], this is the first time all three halides have been measured in these halophytes (Table 3a). Batis and Salicornia contain more halides than Frankenia or Spartina, because of their strategy to survive in saline soils by maintaining a large amount of tissue water (succulence) with which to sequester salts in cell vacuoles [Jennings, 1968]. It is anticipated that other halophytes would also be prolific methyl halide producers because of their high tissue halide levels.
[33] The frequency of analysis for tissue halides may have missed any significant seasonal differences in tissue halide content for the various halophytes. Salicornia chloride and bromide levels were slightly lower in the late winter of 2003 as compared to summer of 2002, presumably as a response to increased succulence [Ralph and Manley, 2006]. This conceivably could affect CH 3 Cl and CH 3 Br emissions. However, in the winter, much of the green tissue of Salicornia disappeared, leaving mostly woody stems above ground, and the decline is most likely a response to overall lack of metabolic activity. In marked contrast, during the flowering period of September and October of 2002, iodide tissue levels increased by several orders of magnitude [Ralph and Manley, 2006]. The peak emission of CH 3 I also corresponded with flowering during a different year (Figure 2b). This excess iodide possibly in concert with flower methyl transferase activity may be responsible for the October peak in CH 3 I production in this species. The peak in CH 3 I in March of 2005 corresponded with the development of new green succulent tissue associated with renewed growth.
Methyl transferase isolated from Batis shows a higher affinity with I > Br > Cl [Wuosmaa and Hager, 1990]. Our molar ratio of CH 3 Cl: CH 3 Br plant emission of 12 ± 3 (sd) compared to 17 ± 14 reported by Rhew et al. [2000] for all plants.

Methyl Halide Emission From Mud and Soil
[35] The soil/mud mixed control for Salicornia was more similar in methyl halide flux to the estuarine mud control for Spartina, except the Spartina control showed greater CH 3 I production. The Batis and Frankenia soil controls were also more similar in emissions suggesting similar production mechanisms. For our estimations of annual emissions from UNB (Table 5), however, we chose to categorize emissions from estuarine mud as equivalent to the Spartina control and those from marsh soil as equivalent to mean of the Salicornia, Frankenia and Batis controls. We did this because their CH 3 Cl: CH 3 Br: CH 3 I ratios were more similar, and the Salicornia control was closer to the other sites. The CH 3 Cl: CH 3 Br: CH 3 I molar ratio from mudflats (5.2:0.53:1) and mid marsh soil (10:1:1), respectively is much different from the plant ratio. These ratios are different than those of plants and may represent differences in production mechanisms or the effects of differential consumption processes. Biological degradation, however, has been shown to favor CH 3 Br [Schaefer et al., 2002], indicating the greater influence of emission differences.
[36] Emissions of CH 3 Cl and CH 3 Br from mudflats and mid marsh soil were smaller than that from plants (Table 5). A weak seasonal signal for CH 3 Cl emissions was detectable only for mid marsh soil and for CH 3 Br emissions from mud. The uptake of CH 3 Cl and CH 3 Br was more pronounced in the mid marsh soil occurring primarily, but not exclusively, in the winter. Rarely were these gases taken up by mud. Microbial production in mud throughout the year may have accounted for the very low level emission of these gases. Their emission from mid marsh soil during the summer may have also been microbially produced, with production ceasing in the winter. Emissions from mud in a tidal channel with the green seaweed Enteromorpha present [Rhew et al., 2000] were approximately 2.6 times greater than our average emissions of CH 3 Cl and CH 3 Br from mud. Green seaweeds are known producers of methyl halides [Manley et al., 1992;Nightingale et al., 1995].
[37] Mudflats contributed a significant amount of CH 3 I to the total salt marsh emissions (8.8%, Table 5). Its production occurred throughout the years studied (April 2004 was the only instance of no production) and consistently peaked in the summer. Benthic diatoms are a common component of estuarine mud [Admiraal, 1984] and micro algae, including diatoms, are known to produce CH 3 I [Tait and Moore, 1995;Moore, 1996, 1998;Manley and de la Cuesta, 1997]. Methyl iodide is the major alkyliodide found in estuarine sediments, with a concentration range of 1 -3.5 pg gdwt À1 sediment [Tessier et al., 2002]. It may also be formed abiotically during oxidation of organic matter in sediment [Keppler et al., 2000].
[38] Highly significant CH 3 I emissions also occurred from barren soil. The magnitude of emission (g yr À1 ) was  (Table 2). Also, a plant species that covers a relatively small area (e.g., Batis, Frankenia) can make a significant contribution to the total methyl halide emitted from the marsh (Table 4; also see the auxiliary material). Conversely, a plant species that covers a large area, such as Spartina, can be an overall minor contributor of methyl halides. The accuracy of the estimated methyl halide emissions can be dramatically improved by increasing the number of species being measured. Our estimation of methyl halide emissions from coastal salt marshes could still be low because we did not measure eight species present in UNB and TE, including Scirpus californicus (Bulrush) with 14% cover in the UNB upper marsh and Tiglochin maritime (Sea lavender) with 11% cover in UNB middle marsh.
[40] We did not account for possible diurnal patterns in methyl halide emission from plants. It is unclear if any or all halophytes exhibit this response, and if so, the amplitude of the response wave. Batis and Salicornia have shown increased CH 3 Cl and CH 3 Br emission with a peak corresponding with maximum daylight [Rhew et al., 2000[Rhew et al., , 2002. The magnitude of the response varied depending on the methyl halide emitted and the plant species examined. It is possible, however, that these diurnal emissions were artifactual owing to changes in plant metabolism. The chambers used by Rhew et al. [2000Rhew et al. [ , 2002 blocked the plants from sunlight during the incubation period of 15 to 40 min. During that period the plant would undergo a dramatic metabolic shift, from photosynthesis to respiration as indicated by an increase in chamber CO 2 . The effect of these metabolic changes on methyl transferase activity is unknown. Ambient air temperature also increased and may have had a direct effect on respiration and methyl halide production. Similar incubations in transparent chambers with regulated temperature would confirm the existence of a diurnal response. If the diurnal response is real, and occurs with other plant species, then our estimation of methyl halide emissions is high because our measurements were made during daylight hours.
[41] The two southern California salt marshes that were modeled for plant-emitted methyl halides, UNB (Table 4) and TE (see the auxiliary material) have nearly identical species composition but different plant cover percentages. Spartina percent cover at both locations was similar but it was Salicornia that dominated the TE salt marsh. Monanthochloe was more prevalent at TE salt marsh. However, the total fluxes of methyl halides (mg m À2 yr À1 ; mean weighted flux, Table 4; see the auxiliary material) from the vegetated area in each marsh are similar because the most prolific producers, Batis and Frankenia, have similar emission rates (Table 1) and combined percent coverage (21% UNB and 19% TE).
[42] A more detailed methyl halide inventory was constructed for UNB salt marsh ( Table 5). The major contributor of CH 3 Cl and CH 3 Br were plants, producing over 90% of both gases. Mid and upper marsh soil contributed nearly all of the remaining 10%. In contrast, the mid and upper marsh soil accounted for 50% of the CH 3 I produced with 41% coming from the plants and 8.8% from mudflats. As previously discussed, CH 3 I production from soil may have come from the roots and the organisms associated in the rhizosphere (e.g., mycorrhizae). If this is the case, we have not accounted for the large CH 3 I emission (22 mg m À2 yr À1 , Table 5) by the soil under the plants in the mid and high salt marsh because control emissions (bare soil) were subtracted from chamber emissions (plant + soil). Therefore an additional 8.4 Â 10 3 g CH 3 I yr À1 , equal to the flux (22 mg m À2 y À1 ) Â the total area soil area covered by plants (3.8 Â 10 5 m 2 ), would need to be added to the total UNB CH 3 I emissions for a total of 2 Â 10 4 g CH 3 I yr À1 .
[43] On the basis of our analysis of methyl halide emissions in Table 5, the annual global emission of methyl halides from salt marshes was estimated (Table 7). When taking into account the total barren area of the salt marsh, including the large area of mudflats in the lower salt marsh, exposed at low tides, our estimate of 49 Gg CH 3 Cl yr À1 and 8.0 Gg CH 3 Br yr À1 is smaller than the 170 Gg CH 3 Cl and 14 CH 3 Br Gg yr À1 reported by Rhew et al. [2000]. If we extrapolate using only vegetated area as representative of the entire marsh, as did Rhew et al. [2000], then our estimate of 160 Gg CH 3 Cl yr À1 is very close to theirs while our estimate of 24 Gg CH 3 Br yr À1 is 1.7 times larger.
[44] Methyl halide fluxes have been measured from the coastal salt marsh in Cape Grim which is dominated by Pachycornia arbuscula (means fluxes were 2.6 mg CH 3 Cl m À2 yr À1 , 1.7 mg CH 3 Br m À2 yr À1 , 2.2 mg CH 3 I m À2 yr À1 ) and the emissions showed pronounced seasonality [Cox et al., 2004]. The ranges of values fit well with our findings for the closely related species Salicornia (Table 1) but were much lower than our estimated annual emission for UNB (Table 7), because of our inclusion of other more prolific methyl halide emitting species. Calculated from total production, Table 5. b Assuming significant rhizosphere production, see text.
[45] Estimates of global salt marsh methyl halide production are highly uncertain because of the error associated with estimates of global salt marsh area and methyl halide source strengths. The estimate of global marsh area (salt marsh + mangrove), 3.8 Â 10 11 m 2 [Woodwell et al., 1973] was stated to be highly unreliable, ±50%. Recent work by Duarte et al. [2005] estimates that salt marshes cover 4 Â 10 11 m 2 and global mangroves cover 2 Â 10 11 m 2 , an increase of >50%. Our estimation of global emissions does not consider the variability in species composition that exists between temperate salt marshes and mangroves, and even among temperate salt marshes worldwide. Indeed, cover estimates of a northern California salt marsh showed Salicornia comprising 80% of the marsh and Spartina only 14% [Zhang et al., 1997]. Certain genera are well distributed globally (e.g., Spartina), but many are not. The work reported here reveals the importance of determining emissions from all species present, not just those that dominate in coverage. Differences in growing season lengths were also not considered but may also dramatically impact estimates of methyl halide emissions from global salt marshes.
[46] The estimate of total global wetlands (freshwater, saline, inland, and coastal) is 1.3 Â 10 13 m 2 , and therefore they are ecologically highly significant [Finlayson et al., 1999]. Assessments of these wetland areas as methyl halide sources has begun, including fens [Varner et al., 1999], rice paddies [Redeker et al., 2000] and coastal wetlands [Rhew et al., 2000;Cox et al., 2004] (also this study). On the basis of size, yearly solar irradiance and coastal location, mangroves, coral reefs and seagrass-based systems need to be investigated for methyl halide emissions. Mangroves replace salt marshes at approximately the 20°C winter ocean isotherm and as such cover approximately 60 -70% of the coastline between 25°N and 25°S latitudes [Hogarth, 1999]. Their community structure is also much different than temperate salt marshes, with much greater plant biomass density and productivity. To date, no mangrove plant species has been studied for methyl halide emissions, although a strong CH 3 Cl signal has been detected from tropical coastal land [Yokouchi et al., 2000]. They could be very significant sites of methyl halide emissions.