DMS and MSA measurements in the Antarctic boundary layer: impact of BrO on MSA production

In situ measurements of dimethyl sulphide (DMS) and methane sulphonic acid (MSA) were made at Halley Station, Antarctica (75 ◦ 35 ′ S, 26 ◦ 19W) during February 2004– February 2005 as part of the CHABLIS (Chemistry of the Antarctic boundary layer and the interface with snow) project. DMS was present in the atmosphere at Halley all year 5 (average 38.1 ± 43 pptV) with a maximum monthly average value of 113.6 ± 52 pptV in February 2004 coinciding temporally with a minimum in sea extent. Whilst seasonal variability and interannual variability can be attributed to a number of factors, short term variability appeared strongly dependent on air mass origin and trajectory pressure height. The MSA and derived non-sea salt sulphate (nss-SO 2 − 4 ) measurements 10 showed no correlation with those of DMS (regression R 2 = 0.039, and R 2 = 0.001, respectively) in-line with the complexity of DMS ﬂuxes, conﬂicting oxidation routes, transport of air masses and variable spatial coverage of both sea-ice and phytoplankton. MSA was generally low throughout the year, with an annual average of 42 ng m − 3 (9.8 ± 13.2 pptV), however MSA: nss-SO 2 − 4 ratios were high implying a dominance of 15 the addition oxidation route for DMS. Including BrO measurements into MSA production calculations demonstrated the signiﬁcance of BrO on DMS oxidation within this region of the atmosphere in austral summer. Assuming an 80% yield of DMSO from the reaction of DMS + BrO, an atmospheric concentration of BrO equal to 3 pptV increased the calculated MSA production from DMS by a factor of 9 above that obtained 20 when considering only reaction with the hydroxyl radical.

MSA in near-coastal ice cores has been proposed as an indicator of past sea ice conditions (Curran et al., 2003), although it has to be used with care (Abram et al., 2007). The interpretation relies on the ice core concentration (hence the airborne concentration) being determined mainly by production, and to a lesser extent by transport. If the chemical transformation of DMS to MSA was affected by changing conditions, this 15 would further modify the interpretation of this proxy, so this provides another motivation for understanding the processes in this oxidation.
The longest record of specifically DMS measurements in the Southern Hemisphere (S.H.) is from Cape Grim, Tasmania and dates back to 1976 (Ayers et al., 1995a(Ayers et al., , 1997. Although a great deal of information has been gained from these measurements the 20 complexity of the oxidation processes is such that the fate of DMS is some way from being fully understood. In the Antarctic there have been various measurement campaigns for DMS on ships in the Weddell sea (Davison et al., 1995;Staubes and Georgii, 1993) and comprehensive sulphur studies such as ISCAT 1998/2000 (Investigation of Sulfur chemistry in the Antarctic Troposphere) and SCATE (Sulphur Chemistry in the 25 Antarctic Troposphere Experiment) (Mauldin-III et al., 2001;Berresheim et al., 1998a) have taken place at Amundsen-Scott (South Pole, 90 • 0 ′ 0 ′′ S, 139 • 16 ′ 0 ′′ W) (Davis et al., 2004), Palmer Station (64 • 46 ′ S, 64 • 03 ′ W) (Berresheim et al., 1998b), and also at Dumont d'Urville (66 • 40 ′ S, 140 • 1 ′ E) ( Jourdain and Legrand, 2002;Legrand et al., 2001)  but have focussed mainly on austral summer. Main outcomes of these large campaigns show that those coastal stations such as Dumont d'Urville (DDU) and Palmer Station (Antarctic Peninsula) typically see highly variable DMS concentrations which are influenced strongly by meteorological conditions and specifically by intensive low pressure storms which provide a means for strong vertical exchange between ocean and atmo-5 sphere for this species (Berresheim et al., 1998b). At the South Pole station DMS was typically low or below the instrumental detection limit, with its oxidation products DMSO and MSA also low. Although Halley appears to be in a coastal position it is still 15 km away from the Weddell Sea which for the most part is not considered open ocean; due to the location of Halley DMS concentrations could be expected to display features of 10 both types of sites. Sulphur containing compounds in aerosol, such as sulphate and methane sulphonate (MSA) have been measured year-round since 1983 at Neumayer (Minikin et al., 1998), at Mawson since 1987 (Savoie et al., 1993), and began at DDU in 1991 as part of the CAASC (Coastal Antarctic Aerosol and Snowfall Chemistry) project (Wolff et 15 al., 1998). These studies find that although MSA concentrations are low during sea ice periods, relatively high values have been seen throughout the year indicating potential origins from either DMS oxidation or as by-products from long-range transport of DMS (Ayers et al., 1997;Minikin et al., 1998). There is still however very little literature on long-term measurements of atmospheric DMS for the Antarctic region. 20 In recent years the reaction of DMS + BrO has been explored and the rate of this reaction found to be a factor of 10 3 faster than that of DMS+OH, having significant impact on the rate of production of DMSO and hence also of MSA (Toumi, 1994;Ingham et al., 2007). Measurements of BrO at ppt levels alongside measurements of OH and DMS made during NAMBLEX on the West Coast of Ireland, led Saiz-Lopez et al. 25 to calculate the relative oxidation rates of DMS using these laboratory study data and found that, at the levels of BrO and OH present there, the oxidation of DMS by BrO was an order of magnitude faster than that of DMS by OH (Saiz-Lopez et al., 2004).
Due to the seasonal inaccessibility of Antarctica long-term trends of BrO in this region have mainly been studied using integrated measurements obtained from satellites in space (Hegels et al., 1998). In situ measurements of BrO in the Antarctic are limited and have been generally of a short-term nature although significant in their findings. Kreher et al. (1997) used a zenith sky DOAS instrument to identify sudden 5 large BrO column enhancement during spring 1995 and during four successive austral springs (1999)(2000)(2001)(2002) at Niemeyer, Friess et al. (2004Friess et al. ( , 2005 used a similar instrument to find that BrO was present in almost all air masses which had previously been in contact with sea-ice surfaces. Recent improvements in measurement techniques such as atomic resonance fluorescence and differential absorption spectroscopy (DOAS) have since led to the possibility of more intense long-term studies in this region and during CHABLIS in situ measurements of tropospheric BrO were made for the first time in the Antarctic boundary layer using a LP-DOAS instrument (Saiz-Lopez et al., 2007). Most previous work on DMS and its oxidation products in Antarctica has focussed on the reactions with the OH radical (Berresheim et al., 1998b;Jourdain and Legrand, Introduction

Experimental
DMS was measured every hour using a gas chromatograph with flame ionisation detection (see Read et al., 2007). The detection limit for this species was 5 pptV and the 5 uncertainty on the measurement <5 pptV. DMS was calibrated on the basis determined in the laboratory that its response is 0.65 times that of propane. Propane was calibrated using a commercial 28 hydrocarbon component in nitrogen calibration cylinder (NPL, UK) at the ppbv level, in addition to a diluted pptV NMHC standard. The relationship between DMS and propane was verified in laboratory experiments both before and after the CHABLIS campaign. Main precision errors in the measurement of DMS and propane arise from peak integration (Read et al., 2007). Size-segregated aerosol samples were collected on a 7-stage Sierra Instruments high-volume cascade impactor mounted on the roof of the CASlab. Filters were usually changed at fortnightly intervals, and were stored frozen until they could be transported 15 to the UK for analysis. Major anions (chloride, fluoride, MSA, sulphate and nitrate) and cations (sodium, calcium, magnesium and potassium) were analysed using Dionex ion chromatographs in a clean room at the British Antarctic Survey laboratories in Cambridge. Rankin and Wolff (2003) describe the cascade impactor set-up and the analysis techniques in more detail. 20 The Differential Optical Absorption Spectroscopy (DOAS) instrument was located in the Clean Air Sector Laboratory (CASLAB). An effective light path of 8 km at a height of 5 m above the snowpack was set up between the CASLAB and a retro-reflector array positioned 4 km to the east. Further information on the instrumental design and spectral de-convolution procedures can be found in Plane and Saiz-Lopez (2006).
OH and HO 2 were measured from a 5 m high inlet on the roof of a containerised laboratory adjacent north of the CASLab. The technique of on-resonance laser-induced fluorescence was used to detect OH, with HO 2 measured following chemical conversion through addition of NO. Calibrations were carried out using photolysis of water 5 vapour at 185 nm and yielded detection limits of 4.8×10 −3 and 0.068 pptv for HO and HO 2 , respectively. Overall uncertainty for the measurements was calculated to be ±27%. Bloss et al. (2007) describe the experimental set-up in more detail. Higher concentrations of DMS were expected and observed in the summer associated with both a decrease in sea ice cover and with highest biogenic emissions (Ayers et al., 1995b;Read et al., 2007). The summer (December-February) average seen 15 in this study of 46±50 pptV (<5-286 pptV) was lower than previously quoted summer values of 94 pptV (6-595 pptV) from measurements made by Berresheim et al. (1998b) at Palmer Station during 1994 and of 290±305 pptV (34-2923 pptV) from Legrand et al. (2001) at DDU in 1999. At Halley values for DMS stayed high well into March with the maximum in measurements (418.1 pptV) occurring early in this month. Halley is sit-20 uated further "inland" than most other Antarctic sites; for example, 4 • (500 km) further south than the site at Neumayer (Minikin et al., 1998); and so the effect of the ice retreat occurs later and the later peak levels in DMS at Halley appear to reflect this. A temporally comparable study involving a cruise in the Drake passage and around Antarctica in March-April 1986 reported an average value of 106 pptV (4.4 nmol m −3 ) for DMS Introduction  Berresheim et al. (1987), also showed high scatter similar to this study with data varying between 16.8 pptV (0.7 nmol m −3 ) and 235.2 pptV (9.8 nmol m −3 ), which they attribute to changes in wind speed and wind direction.
Although sea ice cover is vast around the continent for most of the year, even in

Ions
The seasonal variations of the ions are shown in Fig. 2 and mean concentrations for the ions from this experiment are detailed within Table 1 which also shows averages for the same ions in 2001 (Rankin and Wolff, 2003). Sodium (Na + , a measure of sea salt) concentrations measured during CHABLIS 20 were much more in-line with values reported for Halley between the years 1991-1993 by Wagenbach et al. (1998) (62 ng m −3 Na + , 200 ng m −3 total salt) rather than with the 2001 study which saw a mean value of over a factor of two higher. As stated within the 2001 study, aerosol concentrations are highly variable which leads to large interannual variability in the measurements and it was thought that the 2001 study was 25 characterised by an isolated air mass depositing an unusually high loading of these species onto only one set of filters (Rankin and Wolff, 2003 (9.8±13.2 pptV), significantly lower than the 84 ng m −3 measured by Rankin and Wolff in 2001 but similar to a study by Minikin in 1998 in which three Antarctic sites (Neumayer, Dumont d'Urville, Halley) were compared and Halley (data from 1991-1992) was shown to display the lowest values for this species of 3.5 pptV (15 ng m −3 ) (Minikin et al., 1998). . Along with Na + , the Cl − ion was most dominant on the larger size fractions whereas particles such as MSA and SO 2− 4 , formed from gas phase species such as DMS were observed within the submicron fraction (Fig. 3). This was typical behaviour for these latter species although in contrast to previous studies the majority of the MSA and SO 2− 4 was found in the <0.25 µm mode rather than in the 0.3 µm 15 (Jourdain and Legrand, 2002;Rankin and Wolff, 2003).
Sulphate (SO 2− 4 ) was also present from sea salt. Equation (1) below can be used to provide a derived quantity non-sea-salt sulphate as an indicator of biogenic activity.
Here k is equal to 0.252, the weight ratio between sulphate and sodium in seawater 20 (Wagenbach et al., 1998). There has been discussion as to whether a lower ratio should be used to account for the finding that sea ice surfaces, with a lower sulfate/Na ratio, act as a source of sea salt (Wagenbach et al., 1998); however, the difference does not affect the main results of this paper. The calculated nss-SO 2− 4 is also plotted on Fig. 2 1 and 2). This rapid decrease in concentrations at the end of austral summer was also observed in a study by 5 Prospero et al. (1991) and implies perhaps that the process responsible for converting DMS into MSA is less efficient at this time. Regression analysis highlights no correlation between DMS and the two particulates (R 2 =0.039, and R 2 =0.001) throughout the year. The background concentrations in DMS throughout winter were also not mirrored by the MSA concentrations, a finding similar to that of Jourdain et al. (2001), who attributed the winter DMS to either small emissions from open water present in sea ice located offshore and/or advection from further north in conjunction with a longer lifetime of DMS (Jourdain and Legrand, 2001). Davison et al. (1996) on a cruise between the UK and the Antarctic observed poor regression (R 2 =0.3) between DMS and MSA measurements even though in that study concentrations appeared to coincide. In this study 15 the poor correlation is thought to be a consequence either that (a) the DMS observed at Halley is emitted from the Weddell Sea region but the MSA is a result of processing of DMS emitted from areas substantially further north or (b) that emissions vary little throughout the summer months (December-March). Levels of oxidants such as OH and BrO measured at the site show little variation between January, 0.018±0.013 pptV

Factors affecting DMS and MSA concentrations and their interannual variability
Although Halley is ideally situated on the ice shelf in close proximity to the Weddell 15 Sea, a number of conditions exist which will affect the DMS concentrations measured at the CASLab. These will include physical, meteorological, and chemical factors. A consequence of the varying conditions is that it is difficult to compare measurements from different sites and/or from ships; and interannual variability also arises as a result of these factors. 20 Extensive ice cover during most of the year minimises gas exchange between seawater and the overlying atmosphere. Variability in the extent of cover spatially, affects both local concentrations for these species and accounts for a large majority of the year to year interannual variability of DMS observed in this region of the atmosphere (Cosme et al., 2005 If the sea-ice conditions are favourable for DMS emissions, there still has to be a sufficient distribution and quantity of phytoplankton present in the sea water within the exposed areas. Studies have shown that phytoplankton blooms are often associated 5 with sea ice retreat, shallow waters, with areas of strong upwelling and regions of high eddy kinetic energy; conditions prevalent in the Weddell Sea region (Comisco et al., 1993). Figure 4a shows chlorophyll data, (an indicator of phytoplankton blooms) from seawifs satellite data (http://seawifs.gsfc.nasa.gov/) for austral winter 2004 and 2005 which highlights variability in both the chlorophyll and the sea ice cover from year to 10 year. It is impossible to identify whether these variations are the reason for the interannual variability in the DMS concentrations without additionally considering the air mass origins. Additional complications arise from the fact that different species of phytoplankton have different abilities to produce DMS. Following from this Fig. 4b shows that there were far fewer directly advected air masses from ocean sectors which had passed over 15 highly-concentrated-chlorophyll regions during the second year. Southwest of the site is Precious Bay, an area which often has open water even in winter, and during winter peaks in DMS corresponded to these air masses or occurred during small low pressure storms in which there were higher temperatures experienced at the site. During the SCATE campaign low pressure storm systems were also shown to be a transport 20 mechanism for DMS within the free troposphere influencing the concentrations of its oxidation products by downward mixing (Berresheim and Eisele, 1998). Unfortunately due to technical difficulties DMS data was not obtained during the major low pressure storm which occurred in May 2004 during this campaign .
Finally en route to the site there must also be the right chemical conditions to (a) 25 allow DMS to be transported without oxidation in order to measure DMS and/or to (b) allow oxidation to DMSO and then to MSA for MSA to be measured on the filters. These conditions include the various oxidant concentrations and distribution and will impact on the lifetime and concentration of both DMS and therefore its oxidation products.

Printer-friendly Version
Interactive Discussion

Oxidation of DMS by OH
Much focus has in the past been on the reaction of DMS + OH as the primary oxidation route for this species (Berresheim et al., 1998b;Jourdain and Legrand, 2001;Legrand et al., 2001), particularly with respect to the production of DMSO. The DMSO can then be oxidised further to produce either MSA or nss-SO 2− 4 , and it is these oxidised 5 products which can be found in ice cores and used as indicators of air origin. The reaction of DMS+OH proceeds via two independent channels (Hynes et al., 1986).
OH + CH 3 SCH 3 → CH 3 SCH 2 + H 2 O (abstraction) k 1 (265 K) = 4.50 × 10 −12 cm 3 molecule −1 s −1 (R1) OH + CH 3 SCH 3 → CH 3 S(OH)CH 3 (addition) k 2 (265 K) = 2.21 × 10 −11 cm 3 molecule −1 s −1 (R2) The addition channel leads to CH 3 S(OH)CH 3 or DMSO and subsequently to a greater relative production of MSA rather than nss-SO 2− 4 , as compared to the abstrac-10 tion channel (Fig. 5). There do exist alternative routes for the production of MSA, either through the abstraction channel or via the reaction of DMSO+OH to give MSIA and then MSA, however for the purposes of this analysis we have assumed those routes to be a negligible source of MSA.
The MSA:nss-SO 2− 4 ratio, R (Legrand and Pasteur, 1998) can provide much informa-15 tion as to the oxidation pathways of DMS, although care must be taken to interpret the various contributions to the nss-SO 2− 4 concentrations from non-DMS sources such as sulphate from continental areas and downward transport from the stratosphere. Using 210 Pb levels Minikin et al. (1998) Fig. 6 along with R b . During January-February 2004 and 2005 the average R b of 0.65±0.13 5 was higher than the peak value for Halley of 0.49±0.05 reported for January 1992 and also substantially higher than the annual peak values of 0.46±0.17 (Neumeyer, January 1983-1995 and of 0.28±0.05 (Dumont d'Urville, March 1983-1995 (Legrand and Pasteur, 1998), implying a dominance in the addition pathway for these measurements of DMS at Halley. In addition it is commonly thought that the addition channel 10 is faster at low temperatures (<285 K) (Arsene et al., 1999) which could explain the higher ratios observed in this study.
If we want to investigate the MSA production from DMS then we must consider the addition route to DMSO. If we first assume that the concentration of DMSO in the atmosphere is controlled by (a) production via the DMS/OH pathway, (b) reaction with OH 15 in the gas phase (k 3 =1×10 −10 s −1 Hynes and Wine, 1996), and (c) loss onto aerosols, we can calculate [DMSO] by knowing the yield (β 1 ) of the DMS+OH reaction, together with k 4 , the heterogeneous loss rate of DMSO.
For this analysis we are focussing on the summer (December-March) concentrations of DMS and MSA and the apparent interannual variability between 2004 and 2005. A 20 study at Palmer Station quotes a yield (β 1 ) of DMSO from the DMS+OH reaction of 0.8 and uses a k 4 =2.4×10 −4 s −1 (Legrand et al., 2001) and we have also used this data for our analysis. DMSO is not thought to be directly emitted from the oceans and no consideration has so far been given to the impact of other oxidants such as Cl, and NO 3 on the concentration of DMSO, although the reaction with these species is at least 25 2 orders of magnitude slower than with OH (Falbe- Hansen, 2000).
The following equation Eq.
(2) can be used to obtain a time dependent level of DMSO using the measured DMS concentrations in this study. Although OH was measured for CHABLIS during December 2004-February 2005, for consistency between the two summers we instead used a 24 h average value of 3×10 5 molecules cm −3 (0.012 pptV) for the entire analysis (Bloss et al., 2007). Calculations were initiated applying an initial level of DMSO of 3 pptv (average measured concentration Legrand et al., 2001) and actual DMS concentrations were inputted when they were available, usually every hour.
Roughly one third of the DMSO (DMSOp) produced is thought to be rapidly oxidised into MSA aerosol via MSIA (Legrand et al., 2001), as in Eq. (3).
[MSA] = 0.33k 4 [DMSO p ]∆t (3) ∆t had to be assumed and so 4 h was used based on calculations from Jourdain and 10 Legrand (2001). This assumption obviously leads to potential errors in the calculation of absolute MSA however the importance of this work is to demonstrate the impact of the additional oxidation routes not solely to reproduce the absolute concentrations of MSA.
The calculated MSA concentrations are plotted on Fig. 7  likely to have originated from fresh emissions rather than having been transported, whilst the MSA may have been transported from further afield. This is in line with the air mass trajectories for this time which show much of the air has travelled over the Weddell Sea region (Fig. 4b) During January-February 2005 DMS concentrations were much lower and there is substantial underestimation by the calculated [MSA] when compared to the measured. The air mass trajectories show more air has travelled over the continent in 2005 rather than from the northern sea regions (Fig. 4b) implying that the site received more aged air and so the DMS concentrations measured at the site may not be representative of 5 those at the source. During January-February 2005 the sea ice coverage en route may also have been more substantial leading to smaller fresh emissions of DMS. Underestimation of the calculated MSA implies either that the OH and/or DMS concentrations we have used in our calculations are too low, that additional oxidation routes other than that of OH+DMS are responsible for the MSA production, or that the ∆t we have used 10 is too low.

Impact of BrO concentrations on the oxidation of DMS
Reaction rates for DMS with species other than OH are detailed in Table 2. The obvious competitor to OH is NO 3 but during the whole of CHABLIS it was not seen to be above the instrumental detection limit of the LP-DOAS instrument (2 pptv). BrO however 15 was measurable in concentrations up to 8 pptV in summer which, when considered in context with a rapid rate of DMS+BrO, represents a significant sink for DMS. IO measurements were also made and were present in similar concentrations as BrO however they undergo a slower rate of reaction with DMS. Cl atoms were not measured directly but inferred from non-methane hydrocarbons and found to be at a maximum of 20 3.4×10 4 atom cm −3 during spring (Read et al., 2007). Cl is likely to be at least a factor of 5 lower during summer, and therefore the loss of DMS through reaction with Cl is likely to be less than 20% of that through reaction with BrO during summer.
Von Glasow (2002b) showed that including halogen chemistry and specifically BrO into atmospheric chemistry transport models, led to an increase of 63% DMSO glob- 25 ally. Barnes et al. (1991) found that BrO also adds to DMS which has a major impact on our OH dominated calculations for MSA and could account for the high MSA:nss-SO 2− 4 ratios (Fig. 6). It could also explain the underestimation in the calculated [MSA] 2672 versus measured [MSA] and, assuming variable levels, the lack of correlation between our atmospheric DMS and particulate MSA. A study by Gondwe et al. (2004) which compares modelled MSA:nss-SO 2− 4 ratios with measured, underestimates the ratio for measurements made in the Southern Ocean and Drake Passage. Higher measured ratios compared to model predictions such as these which neglect to include BrO, may 5 also imply an influence from this source.
We can include the reaction of BrO+DMS into our time dependent DMSO equation to give the following: 10 We have used measured BrO concentrations for the two summer periods (Saiz-Lopez et al., 2007) whilst the OH concentration is as we treated before, set at 3×10 5 molecules cm −3 (Bloss et al., 2007) as this was the average of measurements throughout the 2005 summer months. Using an estimated yield (β 2 ) of 0.8, the exact yield not yet having been determined (Toumi, 1994), the DMSO and resultant MSA 15 concentration increase on average by a factor of 9 (Table 3).
Although measurements of BrO were used for this study, they are in situ levels and so may not be representative of the concentrations of this species at the time of DMS oxidation. Due to the nature of its sources and sinks the atmospheric BrO concentration is likely to be more variable than that of the OH and so the sensitivity of the analysis to 20 changing BrO has also been calculated and is also detailed in Table 3.
With the inclusion of the DMS+BrO reaction, the calculated MSA during most of February-March 2004 are now in good agreement with the measured (Fig. 8) Interactive Discussion measured at the site are not representative of the emissions at source or that the time allowed for the build-up of the DMSO on the aerosol is too short. The measured MSA concentrations are relatively high (∼40 pptv) however which is consistent with assumptions that the DMS has had time to oxidise en route and it is clear from this analysis that the DMS + BrO reaction is important in its oxidation.
For the Summer 2005 period the relative rate of oxidation of DMS by OH and BrO is calculated using the measurements of DMS, BrO and OH when all three were measured simultaneously and this is shown in Fig. 10. The rate of oxidation of DMS by BrO dominates the DMS oxidation by on average a factor of 7 throughout the summer.
A change in the yield of either the DMS+BrO reaction or the DMS + OH reaction 10 changes the factor increase in the DMSO and therefore in the MSA concentrations and so knowledge of the yields of these reactions is crucial to this analysis. Table 4 details the factor increase in MSA assuming BrO and OH at constant levels of 3 pptV and 0.012 pptV respectively with changing yields. Assuming that the MSA from the abstraction route is negligible then this analysis 15 shows that the MSA concentration is far more sensitive to the yield of BrO+DMS rather than to the yield of OH+DMS.

Implications of these findings
Although low temperatures such as those in experienced in the Antarctic can lead to a favoured addition route for the oxidation of DMS, it is more likely that the increased 20 MSA production observed at Halley is as a result of the presence of BrO. BrO strongly increases the importance of the addition branch and indeed the oxidation rate even at low atmospheric concentrations. There is therefore a likely underestimation of the total oxidation rates of DMS in studies which do not incorporate halogen chemistry. This finding also impacts on flux calculations as DMS from the ocean is often calculated based on atmospheric DMS mixing ratios and the assumption that reaction with OH is the only chemical loss. If BrO is a significant sink for DMS, the resulting DMS 2674 fluxes must be higher to yield the same atmospheric DMS mixing ratios.

Ice core implications
Both MSA and nss-SO 2− 4 , as well as their ratio, have been used in ice core studies to try to diagnose environmental conditions in the past. In particular, MSA has been proposed as an indicator of sea ice extent (Curran et al., 2003;Abram et al., 2007), while 5 there have been hopes that the ratio, if largely controlled by temperature, might yield information about the latitude and temperature history of sulphur emissions recorded in ice cores. If the presence of BrO can also play a substantial role in the relative and absolute production of MSA, then this raises a potential new complication in interpreting the ice core records of these species. In particular, we know that BrO is strongly associated with the presence of sea ice, although possibly strongly concentrated in the very lowest layers of the atmosphere. If DMS remained in the volume of atmosphere strongly influenced by BrO for a substantial fraction of its lifetime, then this could provide a further mechanism for an influence of sea ice on MSA concentrations recorded in ice cores. Modelling studies (Cosme et al., 2005) that include a realistic assess-15 ment of the spatial and vertical distribution of BrO would however be needed to assess whether this was really a major influence.

Conclusions
A year-long (February 2004-February 2005) time-series of DMS measurements in the Antarctic troposphere has been presented along with filter measurements of its oxi-20 dation product MSA and also with derived nss-SO 2− 4 . DMS shows a strong seasonal variation with an austral winter minimum and summer maximum which follows that of the sea-ice cover; but DMS was also observed to be present throughout the year due to its transport from biogenically active regions further north in association with an extended lifetime of 10 winter concentration of 1×10 5 molecules cm −3 ). The shorter-term variability in the DMS was found to be strongly due to air mass origin and trajectory pressure height. The biweekly filter data appears to correlate a little with the DMS data but in reality the data show very low regression values due to the infrequency and differences in sampling rates of the filter data, in addition to the various complexities in the rate of oxidation of 5 DMS to form these products.
Using the DMS concentrations measured at the site in conjunction with recent kinetic data MSA concentrations were calculated and found at times to differ from the filter measurements. This was postulated to be due to a number of factors including the strong possibility of different DMS values at source (which may be 3.4 days away 10 (OH=3×10 5 molecules cm −3 Bloss et al., 2007)), and the influence of locally produced DMS on the site measurements. Chemical influences were also discussed including the additional effect of BrO on the production of DMSO and hence of MSA. Although the yield of the BrO+DMS reaction is still undetermined the impact of up to 7 pptV of BrO on the calculated MSA atmospheric concentration was found to be highly signif- 15 icant, and will have a significant impact on model calculations for this region of the atmosphere. Gondwe, M., Krol, M., Klasses, W., Gieskes, W., and De Baar, H. Legrand, M., Sciare, J., Jourdain, B., and Genthon, C.: Subdaily variations of atmospheric dimethylsulfide, dimethylsulfoxide, methanesulfonate, and non-sea-salt sulfate aerosols in the atmospheric boundary layer at Dumont d'Urville (coastal Antarctica) during summer, J. Geophys. Res., 106, 14 409-14 422, 2001. Liss, P. S., Hatton, A. D., Malin, G., Nightingale, P. D., and Turner, S. M.: Marine sulphur  Fig. 9. Rate of production of DMSO from the reaction of DMS with OH and BrO.