Atmospheric Exchange of Reduced Sulfur Compounds in Natural Ecosystems By

Terrestrial-Atmospheric Exchange of Reduced Sulfur Compounds in Natural Ecosystems

Mary Elizabeth Whelan

A dissertation submitted in partial satisfaction

of the requirements for the degree of

Doctor of Philosophy

Geography

in the

Graduate Division

of the

University of California, Berkeley

Committee in charge:

Professor Robert C. Rhew, Chair Professor Kurt Cuffey Professor Allen Goldstein Professor Ron Amundsen

Fall 2013

Abstract

Terrestrial-Atmospheric Exchange of Reduced Sulfur Compounds in Natural Ecosystems

Mary Elizabeth Whelan

Doctor of Philosophy in Geography

University of California, Berkeley

Professor Robert C. Rhew, Chair

The sulfur biogeochemical cycle includes biotic and abiotic processes important to global climate, atmospheric chemistry, food security, and the study of related cycles. The largest flux of sulfur on Earth is weathering from the continents into the sulfate-rich oceans; one way in which sulfur can be returned to land is through transport of reduced sulfur gases via the atmosphere. Here I developed a method for quantifying low-level environmental fluxes of several sulfur-containing gases, H2S, COS, CH3SCH3 (DMS), and HSCH3, between terrestrial ecosystems and the atmosphere.

COS is the most prevalent reduced sulfur gas in the atmosphere, considered to be inert in the troposphere except for its uptake in plant leaves and to a smaller extent aerobic soils. This dissertation reports two surprising cases that go against conventional thinking about the sulfur cycle. We found that the common salt marsh plant Batis maritima can mediate net COS production to the atmosphere. We also found that an aerobic wheat field soil produces COS abiotically when incubated in the dark at > 25 °C and at lower temperatures under light conditions.

We then sought to separately quantify plant and soil sulfur gas fluxes by undertaking a year-long field campaign in a grassland with a Mediterranean climate, where green plants were present only half of the year. We measured in situ soil fluxes of COS and DMS during the non- growing dry season, using water additions to simulate soil fluxes of the growing, wet season. COS and CO2 are consumed in a predictable ratio by enzymes involved in photosynthetic pathways; however, while CO2 is released by back diffusion and autorespiration, COS is usually not generated by plants. Using measurements during the growing season, we were then able to calculate gross primary production by using the special relationship between CO2 and COS.

This dissertation has developed a greater understanding of the vagaries of the atmospheric-terrestrial sulfur cycle and explored using that cycle as a tool for studying the carbon cycle.

This dissertation is dedicated to Mr. James Chalfant.

Table of Contents

1. Introduction 1

2. Materials and Methods 8

3. Salt marsh vegetation: a carbonyl sulfide (COS) source to the atmosphere 20

4. Carbonyl sulfide produced by abiotic thermal and photo-degradation of 38 soil organic matter from wheat field substrate

5. Exchange of carbonyl sulfide (COS) between a grassland and the 53 atmosphere

6. Conclusion 72

1. Introduction

Sulfur is an element essential to all life: proteins usually contain cysteine and methionine, sulfur-containing amino acids with specific structural and biochemical roles (Levine et al., 1996). Sulfur is constantly weathered from the continents and transported to the ocean via river runoff. Many terrestrial primary producers must therefore rely on airborne sulfur sources (Hawkesford and de Kok, 2007). This dissertation contributes to a better understanding of the movement of sulfur gases between terrestrial ecosystems and the atmosphere.

Our conception of the sulfur cycle has changed considerably, often informed by technical advances in measuring specific compounds. In the early 20th century, researchers associated with the United States Geological Survey suggested that volcanoes dominated the atmospheric sulfur burden; sulfate was reduced in the ocean to pyrite, returning sulfur to ocean sediments and the rock cycle. In the 1970s, the role of volcanoes was often neglected and a greater importance wrongly assigned to biogenic hydrogen sulfide (H2S) emissions (Brimblecombe, 2003). One of the most prevalent procedures for measuring H2S, the Natusch method (Natusch et al., 1972), inadvertently included interference from carbonyl sulfide (COS), a more abundant and pervasive reduced sulfur gas. This called into question much of the H2S gas phase data collected over marine ecosystems (Cooper and Saltzman, 1987). Similarly, many early chamber measurements of COS fluxes used sulfur-free sweep air which was later shown to lead to COS flux values of the opposite sign than what was actually occurring in many ecosystems (Castro and Galloway, 1991; de Mello and Hines, 1994). There is still a scarcity of methane thiol (CH3SH) related data, perhaps because of difficulties in measuring this reactive compound.

In our current understanding, atmospheric sulfur emissions arise from human industry, volcanic activity, formation of sea salt aerosols, aeolian processes, and gas production in natural ecosystems (Watts, 2000). Examining the exchange of different sulfur species over ecosystems gives us information about their origins and fate. There are four sulfur compounds considered in this dissertation: hydrogen sulfide (H2S), carbonyl sulfide (COS), dimethyl sulfide (DMS or CH3SCH3), and carbon disulfide (CS2) (see Table 1). The production and consumption of H2S is related to oxidation-reduction potential, but its atmospheric budget is not well constrained on the global scale (Watts, 2000). DMS and CS2 tend to be from marine and coastal sources and their oxidation yields about half of the COS in the atmosphere (Barnes et al., 1994; L. Wang et al., 2001). COS is the longest lived and most abundant sulfur compound, inert in the troposphere but for a large vegetative sink (Montzka et al., 2007). Investigating what controls these fluxes will help clarify how atmospheric sulfur gases interact with terrestrial ecosystems; in particular how plants and soils could act as biogenic sources of sulfate aerosol precursors.

1.1 Human perturbation of the sulfur cycle Natural emissions of sulfur gases will become more important over time as anthropogenic inputs to the sulfur cycle decrease. Humans have perturbed the biogeochemical sulfur cycle, arguably more than any other major element cycle. Some estimates suggest that human industrial emissions of sulfur exceed natural emissions globally by a factor of 2 or 3 (Rodhe, 1999).

Overall, the human contribution to the atmospheric sulfur burden are declining because of SO2 regulation. Acid rain formation was first related to human-made SO2 emissions in the 19th century (Smith, 1872), though it was not until 1980 that the United States began to investigate the need for a policy response. SO2 is the most highly soluble of the reduced sulfur compounds (Table 1), reacting with atmospheric water droplets to form sulfuric acid. Water droplets with a pH below 5.6 constituted “acid rain” or snow found downwind of fossil fuel-burning smoke stacks (Likens and Bormann, 1974). Multiple sulfur emissions controls were put in place in the late 1980s and has ameliorated this problem (Lackey and Blair, 1997).

Anthropogenic SO2 emissions abatement in Europe and the United States have reduced total atmospheric sulfate by approximately 27% since 1980 (Forster et al., 2007). New developments in emerging economies have caused an overall increase of anthropogenic SO2 in the Southern hemisphere, though the increase is small (2%) for the global budget (Stern, 2005).

Diminishing anthropogenic atmospheric sulfur inputs increases the influence of natural emissions on the remaining budget and reduces the total input of sulfur on terrestrial ecosystems. Deposition of sulfate aerosols and SO2 to some vegetated areas is an important source of sulfur to soils and plants (Hawkesford and de Kok, 2007), indicating a potential food security problem as the anthropogenic sulfur source wanes. compound molecular boiling point vapor Henry’s law Redox Lifetime in the weight (°C, pressure constant state atmosphere (g/mol) approximate) (bar at (mol/L at of S 20°C) 25°C) H2S 34.1 -60 18.2 0.086 -II ~3 days (CH3)2S 62.1 37 0.6 0.474 -II ~2 days (or DMS) CS2 76.1 46 0.4 0.054 -II ~7 days SO2 64.1 -10 3.4 1.23 IV 15 min to 4 days COS 60.1 -50 12.5 0.022 -II ~2-7 years Table 1. Most abundant sulfur-containing gases in Earth’s atmosphere. Lifetimes are taken from Brimblecombe 2003, except for COS, taken from Xu et al. 2002. The rest of the table is adapted from Hawkesford and de Kok et al. 2007.

1.2 Atmospheric sulfate in the sulfur cycle Much of the sulfur entering the atmosphere starts in the form of SO2 or sulfate -2 (SO4 ). Many reduced sulfur gases are oxidized in the atmosphere. These airborne sulfate particles are the principal component of sulfate aerosols, which are then returned to the ocean or land reservoirs through wet or dry deposition.

Dimethyl sulfide (DMS) is thought to be responsible for non-sea-salt sulfate near marine ecosystems. Non-sea salt aerosols can act as cloud condensation nuclei (CCN), changing the lifetime and precipitation patterns of clouds. (Andreae and Crutzen 1997). The theoretical increase in DMS production by phytoplankton in warmer oceans was hypothesized to indirectly counteract global warming. In 1987, Robert Charlson, James Lovelock, Meinrat Andreae, and Stephen Warren wrote a paper suggesting that the climate may be “biologically regulated” by DMS production, popularly known as the CLAW hypothesis. More atmospheric DMS leads to more CCN, more clouds, higher albedo, and a cooler climate (Charlson et al., 1987). The CLAW hypothesis spurred a great deal of scientific research which confirmed many of the links in the proposed feedback mechanism. However, after decades of research, cloud dynamics still possess the largest uncertainty in our understanding of climate change and the CLAW hypothesis has not been proven nor disproven (Ayers and Cainey, 2007).

COS is the primary precursor of sulfate aerosols in the stratosphere, notwithstanding direct injections from major volcanic eruptions (Brühl et al., 2012; Crutzen, 1976). In the mid-latitude stratosphere these aerosols are implicated in ozone destruction by interacting with NOy reservoirs (Andreae and Crutzen, 1997; Chin and Davis, 1995). Sulfate aerosols can also backscatter incoming solar radiation (Charlson et al., 1992), contributing to global dimming. However, the global cooling effect of COS- sourced stratospheric sulfate aerosols is approximately offset by the COS global warming potential (Brühl et al., 2012).

The primary means of atmospheric sulfur uptake by ecosystems is wet and dry deposition of sulfate. However, COS is not readily oxidized in the troposphere and is therefore the most abundant sulfur gas, with ambient concentrations around 500 ppt and the longest lifetime of 2-7 years (Xu et al., 2002) (See Table 1). Once in the atmosphere, the largest sinks for COS in the troposphere is uptake by plants.

1.3 Linking sulfur and carbon cycles Another motivation for studying the terrestrial-atmospheric COS exchange is its relationship to the carbon biogeochemical cycle. Carbonyl sulfide has been found to react irreversibly with the enzymes involved in photosynthesis (Protoschill-Krebs and Kesselmeier, 1992; Protoschill-Krebs et al., 1996). While CO2 is both assimilated and respired by the plant, COS is only absorbed (Seibt et al., 2010; Stimler et al., 2010). The ratio of COS uptake to CO2 is a sensitive indicator of the ratio between plant carbon uptake and respiration (Berry et al., 2013). Simultaneously observing terrestrial CO2 and COS fluxes could act as a measure of net and gross CO2 exchange, barring other sources or sinks of COS, e.g. soil interactions.

Considering the available data at the time, Watts (2000) suggested that COS should be produced by anoxic soils and consumed by oxic soils. Field measurements of oxic soils do exhibit small net COS consumption (Castro and Galloway, 1991; Steinbacher et al., 2004; White et al., 2010; Xu et al., 2002; Yi et al., 2007), with the exception of a wheat field (Billesbach et al., 2014). If we develop a greater understanding of soil COS exchange for different ecosystems, COS could prove a powerful tool for independently estimating gross primary production.

1.4 Terrestrial-atmospheric sulfur gas exchange Here we investigate exceptions to the current understanding of terrestrial- atmospheric sulfur gas exchange. To this end, we have developed a method to observe fluxes of H2S, CH3SH, DMS, CS2, and COS simultaneously in situ and in the laboratory (Chapter 2). Using static flux chambers, we explored a surprising instance where a plant species acted as a large net source of COS to the atmosphere instead of a sink (Chapter 3). We then performed lab incubations with agricultural soils that acted as a COS source (Billesbach et al., 2014), contrary to previous studies of oxic soils. Our lab measurements agree with field observations and our evidence suggests an abiotic COS production mechanism related to thermal and photo- degradation (Chapter 4). To separate the influence of soils and plants in the field, we measured annual production and consumption of COS in a grassland ecosystem with green plants present only part of the year (Chapter 5). A summary of the results and how they contribute to our conception of terrestrial-atmospheric sulfur exchange is outlined in Chapter 6.

Works Cited Andreae, M.O., Crutzen, P.J, 1997. Atmospheric aerosols: biogeochemical sources and role in atmospheric chemistry. Science 276, 1052–1058. Ayers, G.P., Cainey, J.M., 2007. The CLAW hypothesis: a review of the major developments. Environmental Chemistry 4, 366–374. Barnes, I., Becker, K.H., Patroescu, I., 1994. The tropospheric oxidation of dimethyl sulfide: A new source of carbonyl sulfide. Geophysical Research Letters 21, 2389–2392. Berry, J., Wolf, A., Campbell, J.E., Baker, I., Blake, N., Blake, D., Denning, A.S., Kawa, S.R., Montzka, S.A., Seibt, U., Stimler, K., Yakir, D., Zhu, Z., 2013. A coupled model of the global cycles of carbonyl sulfide and CO2: A possible new window on the carbon cycle. Journal of Geophysical Research: Biogeosciences 118, 842–852. Billesbach, D.P., Berry, J.A., Seibt, U., Maseyk, K., Torn, M.S., Fischer, M.L., Abu-Naser, M., Campbell, J.E., 2014. Growing season eddy covariance measurements of carbonyl sulfide and CO2 fluxes: COS and CO2 relationships in Southern Great Plains winter wheat. Agricultural and Forest Meteorology 184, 48–55. Brimblecombe, P., 2003. The Global Sulfur Cycle, in: Holland, H.D., Turekian (Eds.), Treatise in Geochemistry. pp. 645–682. Brühl, C., Lelieveld, J., Crutzen, P.J., Tost, H., 2012. The role of carbonyl sulphide as a source of stratospheric sulphate aerosol and its impact on climate. Atmospheric Chemistry & Physics 12, 1239–1253. Castro, M.S., Galloway, J.N., 1991. A comparison of sulfur-free and ambient air enclosure techniques for measuring the exchange of reduced sulfur gases between soils and the atmosphere. Journal of Geophysical Research 96, 15427–15. Charlson, R. J., Schwartz, S.E., Hales, J.M., Cess, R.D., Coakley, J.A., Hansen, J.E., Hofmann, D.J., 1992. Climate Forcing by Anthropogenic Aerosols. Science 255, 423–430. Charlson, R. J., Warren, S.G., Lovelock, J.E., Andreae, M.O., 1987. Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate. Nature 326, 655–661. Chin, M., Davis, D.D., 1995. A reanalysis of carbonyl sulfide as a source of stratospheric background sulfur aerosol. Journal of Geophysical Research 100, 8993–9005. Cooper, D.J., Saltzman, E.S., 1987. Uptake of carbonyl sulfide by silver nitrate impregnated filters: Implications for the measurement of low level atmospheric H2S. Geophysical Research Letters 14, 206–209. Crutzen, P.J., 1976. The possible importance of CSO for the sulfate layer of the stratosphere. Geophysical Research Letters 3, 73–76. De Mello, W.Z., Hines, M.E., 1994. Application of static and dynamic enclosures for determining dimethyl sulfide and carbonyl sulfide exchange in Sphagnum peatlands: Implications for the magnitude and direction of flux. Journal of Geophysical Research 99, 14–601.

Forster, P., Ramaswamy, V., Artaxo, P., Berntsen, T., Betts, R., Fahey, D., Van Dorland, R., 2007. Changes in atmospheric constituents and in radiative forcing, in: IPCC, 2007: Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Hawkesford, M.J., and de Kok, L.J. 2007. Sulfur in Plants: An Ecological Perspective. Springer. Lackey, R., Blair, R., 1997. Science, policy, and acid rain: lessons learned. Renewable Resources Journal 15, 9–13. Levine, R.L., Mosoni, L., Berlett, B.S., Stadtman, E.R., 1996. Methionine residues as endogenous antioxidants in proteins. PNAS 93, 15036–15040. Likens, G.E., Bormann, F.H., 1974. Acid rain: A serious regional environmental problem. Science 184, 1176–1179. Montzka, S. A., Calvert, P., Hall, B.D., Elkins, J.W., Conway, T.J., Tans, P.P., Sweeney, C., 2007. On the global distribution, seasonality, and budget of atmospheric carbonyl sulfide (COS) and some similarities to CO2. Journal of Geophysical Research 112. Natusch, D.F., Klonis, H.B., Axelrod, H.D., Teck, R.J., Lodge, J.P., Jr, 1972. Sensitive method for measurement of atmospheric hydrogen sulfide. Analytical Chemistry 44, 2067–2070. Protoschill-Krebs, G., Kesselmeier, J., 1992. Enzymatic pathways for the consumption of carbonyl sulphide (COS) by higher plants. Botanica Acta 105, 206–212. Protoschill-Krebs, G., Wilhelm, C., Kesselmeier, J., 1996. Consumption of carbonyl sulphide (COS) by higher plant carbonic anhydrase (CA). Atmospheric Environment 30, 3151–3156. Rodhe, H., 1999. Human impact on the atmospheric sulfur balance. Tellus A 51, 110– 122. Seibt, U., Kesselmeier, J., Sandoval-Soto, L., Kuhn, U., Berry, J. A., 2010. A kinetic analysis of leaf uptake of COS and its relation to transpiration, photosynthesis and carbon isotope fractionation. Biogeosciences 7, 333–341. Smith, R.A., 1872. Air and rain: the beginnings of a chemical climatology. Longmans, Green, and Company. Steinbacher, M., Bingemer, H.G., Schmidt, U., 2004. Measurements of the exchange of carbonyl sulfide (OCS) and carbon disulfide (CS2) between soil and atmosphere in a spruce forest in central Germany. Atmospheric Environment 38, 6043–6052. Stern, D.I., 2005. Global sulfur emissions from 1850 to 2000. Chemosphere 58, 163– 175. Stimler, K., Montzka, S.A., Berry, Joseph A., Rudich, Y., Yakir, D., 2010. Relationships between carbonyl sulfide (COS) and CO2 during leaf gas exchange. New Phytologist 186, 869–878. Wang, L., Zhang, F., Chen, J., 2001. Carbonyl sulfide derived from catalytic oxidation of carbon disulfide over atmospheric particles. Environmental Science & Technology 35, 2543–2547. 6

Watts, S.F., 2000. The mass budgets of carbonyl sulfide, dimethyl sulfide, carbon disulfide and hydrogen sulfide. Atmospheric Environment 34, 761–779. White, M.L., Zhou, Y., Russo, R.S., Mao, H., Talbot, R., Varner, R.K., Sive, B.C., 2010. Carbonyl sulfide exchange in a temperate loblolly pine forest grown under ambient and elevated CO2. Atmospheric Chemistry & Physics 10, 547–561. Xu, X., Bingemer, H.G., Schmidt, U., 2002. The flux of carbonyl sulfide and carbon disulfide between the atmosphere and a spruce forest. Atmospheric Chemistry and Physics Discussions 2, 181–212. Yi, Z., Wang, X., Sheng, G., Zhang, D., Zhou, G., Fu, J., 2007. Soil uptake of carbonyl sulfide in subtropical forests with different successional stages in south China. Journal of Geophysical Research 112, D08302.

2. Materials and Methods

This chapter describes field and lab materials and methods that have been used in measuring fluxes of carbonyl sulfide (COS), dimethyl sulfide (DMS), carbon disulfide (CS2), hydrogen sulfide (H2S), and methane thiol (MeSH). COS has a lifetime of 2-7 years (Xu et al., 2002) and a global atmospheric concentration of ≈500 parts-per- trillion (ppt or pmol mol-1) (Montzka et al., 2007). COS fluxes in natural systems range from uptake of -120 pmol COS m-2 sec-1 (a rainforest, see Chapter 6) to production of nearly +300 pmol COS m-2 sec-1 (a salt marsh, in DeLaune et al., (2002)). The lifetimes of the other reduced sulfur gases in the atmosphere are 2 to 7 days (see Table 1.1) and the extent of their ambient concentrations range from 1 to 500 ppt (Prinn, 2003), depending on the proximity of sources and sinks. Quantifying natural fluxes of reduced sulfur compounds requires a low method detection limit.

For field studies in this dissertation, flask air samples were collected from static flux chambers and sulfur gases were quantified by lab-based GC methods. For laboratory incubations, dynamic and static chambers were used and air samples were injected directly into a GC. The precision and method detection limits are listed in Table 2.1.

Compound Ambient GC method GC method Minimum flux Minimum flux concentration (ppt) detection precision from detectable, detectable, from Prinn, (2003) limit (Khan et al., 2012) lab-based field based COS 500 2% 0.05 DMS 10-100 120 ppt in pmol/min -2 -1 CS2 1-300 a 100 mL 0.6 pmol m sec

H2S, 5-500 air sample 7% 10 pmol/min CH3SH ? Table 2.1 Ambient concentrations and method detection limits for measuring sulfur gas fluxes from natural systems. No ambient global atmospheric concentration for CH3SH was reported in the literature.

COS can be quantified with stainless steel equipment because of its relatively high ambient concentration and low reactivity with steel. Of the suite of sulfur gases described here, CS2 is perhaps the second most stable in an air sampling flask, but is less abundant and environmental fluxes of the ecosystems studied were much lower compared to the other gases. Air samples of DMS may adsorb on steel parts or chemically combine into dimethyl disulfide (DMDS). H2S and MeSH are comparatively more reactive compounds with decreasing concentrations in air sampling flasks after only a few days. Sample collection and handling concerns are summarized in Table 2.1.

Compound Recommended Field or lab sampling materials flasks for whole air samples carbonyl sulfide electro-polished -stainless steel, silica-lined stainless steel, (COS) stainless steel or PTFE, FEP, PFA amorphous silica- lined flasks; -no rubber gaskets or ascarite traps (used concentrations in air for scrubbing CO2) sampling flasks are stable for at least 4 -when measuring high fluxes e.g., Whelan, months (Montzka et et al., (2013), the uncertainty introduced by al., 2004) using aluminum chamber materials may be small compared to the magnitude of fluxes.

carbon disulfide -stainless steel, silica-lined stainless steel, (CS2) PTFE, FEP, and PFA

-very low ambient concentrations that can cause instrument detection limit issues

dimethyl sulfide, amorphous silica- -PTFE, PFA, FEP and amorphous silica-lined CH3SCH3 (DMS) lined flasks; air tubing material samples are stable for at least 1 week (M A H -avoid prolonged exposure to stainless steel Khan et al., 2012)

hydrogen sulfide electro-polished -PTFE, PFA, and amorphous silica-lined (H2S) stainless steel flasks tubing material with 100 nm methane thiol, amorphous silica -Sealing materials for glass containers CH3SH (MeSH) coating; must be caused analyte loss analyzed within 3 days of collection (M -Lab-based incubations require above A H Khan et al., 2012). ambient concentration (≈10 ppb) to avoid sample loss despite using all PFA chambers.

Table 2.1 The most abundant ambient reduced sulfur compounds and associated materials concerns in measuring fluxes at near ambient concentrations.

2.1 GC methods for quantification of reduced sulfur compounds

The work in this dissertation employed gas chromatograph (GC) methods in quantifying fluxes of reduced sulfur compounds between terrestrial ecosystems and the atmosphere. This required low detection limits (≈ 100 ppt), the ability to transport air samples, and use of gas handling equipment inert to the compounds of interest.

Two detectors were used in quantifying COS fluxes: a mass spectrometer (MS) and a sulfur chemiluminescence detector (SCD). The custom-built inlet system for the GC/MS has been described elsewhere (Rhew et al., 2007) and is similar to the system used to measure COS in the Siple Dome ice core (Aydin et al., 2002). The wetted parts of the GC/MS inlet system are mostly stainless steel, and reactions on the surface reduced the recovery rate of other sulfur compounds aside from COS.

To measure concentrations of the broader suite of sulfur gases, a GC/SCD was used. An SCD has the advantage of design simplicity and specificity for sulfur compounds. Gas samples are burned at 800 °C in a plasma of oxygen and hydrogen, converting sulfur compounds into SO2. The sample stream is then bombarded with ozone, causing the SO2 to reach an excited state, chemiluminescing in front of a photo- multiplier tube. The result is a linear, equimolar response to many sulfur compounds in a low concentration range (Benner and Stedman, 1989; Shearer et al., 1990; Yan, 2002). Coupling an SCD with a GC, all of our compounds of interest could be quantified from a 40 mL to 120 mL air sample.

An Agilent 355 SCD and an Agilent 7890A GC oven (Agilent Technologies, Santa Clara, CA, USA) were used with a DB-1 capillary column (30 m × 0.32 mm ID, 5 μm, J&W Scientiﬁc Inc, Folsom, CA, USA). The detector coupling temperature (between the column and the detector) was set to 200 °C; higher temperatures resulted in sample degradation, while lower temperatures resulted in ghost peaks. Helium flow on the column was set to 5 psi with an electronic pressure controller. A Restek silica-coated 6 port valve (Restek, Bellefonte, PA, USA) and a trap packed with silane-treated glass wool was then installed so that helium constantly flowed through the column (see Figure 2.2). The oven was programmed to hold at 30 °C for 5 minutes, then ramp up at 15 °C per minute until reaching 150 °C, where it was held for 3 additional minutes. The 16 minute run time is necessary to elute dimethyl disulfide at the end of the run, so it does not interfere with subsequent analyses. A chromatogram of the 5 component sulfur gas standard (Matheson Tri_gas Inc, Newark, CA, USA) and an ambient air sample from a marshland are shown in Figure 2.1.

COS

H2S DMS detector response (uV) response detector CS2

CS2

COS H2S DMS

CH3SH detector response (uV) response detector

retention time (min)

Figure 2.1 GC/SCD chromatograph of (top panel) 150 mL ambient air sample from a salt marsh near Fremont, CA and (bottom panel) 3 mL of a 5 component sulfur gas standard with 180 ppb of each compound (Matheson Tri-Gas Inc., Newark, CA, USA). Figure adapted from Khan et al., (2012).

To analyze an aliquot of air, the cryotrap and silica-coated tubing between the sample flask or lab incubation chamber were evacuated (refer to Figure 2.2 for a schematic). The pressure that remains in the system was typically < 0.2 torr. The cryotrap was then cooled with liquid nitrogen. About 100 mL of sample air flowed through the trap and into the end volume, trapping the compounds of interest in the cryotrap. The amount of sample introduced was determined by the pressure reading of the end volume. The GC oven program started as the 6 port valve was turned to the “inject” position, so that the sample remaining on the cryotrap is in line with the carrier gas. A container of hot water replaced the dewar of liquid nitrogen, heating the sample and causing it to vaporize or desorb from the trap and flow with the carrier gas onto the column.

GC/SCD

custom inlet PTFE chamber or system air sample flask 1 L silica-coated end Edwards RV3 6 port valve volume vacuum pump insulated container to GC of boiling column water for liquid nitrogen sample injection cryotrap trap w/ silane- Paroscientific treated helium pressure glass wool meter carrier gas

300 mL liquid nitrogen dewar

Figure 2.2 GC/SCD custom inlet system. The configuration is shown in the “load” position. The 6-port valve is turned manually into the “inject” position so that the helium carrier gas flows through the cryotrap before the GC column.

To quantify signal outputs, a calibration curve was generated by injecting different amounts of a sulfur gas standard diluted to < 1 ppb concentrations. To avoid contaminating the inlet system with repeated reactive sulfur inputs, a whole air standard containing 543 ppt COS was measured every 2 to 10 unknown samples to correct for daily drift. The signal response of the GC/SCD changes by 4 to 20% over the course of the day; more drift occurs after the instrument has been turned on after weeks of being shut down.

The response of the GC/SCD was linear for each individual compound over the range of expected values from environmental samples. The method has a higher sensitivity for the more stable gases (COS, DMS, CS2) than for the more reactive gases (H2S and MeSH). In other words, more H2S is needed to produce the same signal peak area compared to COS or DMS. More details about detector response can be found in Khan et al., (2012).

2.2 Lab incubation chambers

To test the stability of reduced sulfur gases in the 1 L chambers, 10 mL of a 1 ppm standard of COS, CS2, DMS, H2S, and MeSH were injected into jars and sealed, resulting in an overall initial concentration of ≈10 ppb of each compound. Headspace sub samples were then analyzed on the GC/SCD (Figure 2.3). CS2 yielded twice the signal output as COS for a given concentration because CS2 contains two sulfur atoms per molecule, whereas COS only has one. Incubations with 1 ppb starting concentrations resulted in high losses of H2S and MeSH. The lab incubation chambers described here should therefore only be used to measure COS, CS2, and DMS fluxes unless very high concentrations of H2S and MeSH are anticipated. Mold- injected PFA jars (Savillex, Eden Prairie, MN, USA) proved to be the optimal choice for COS, CS2, and DMS fluxes.

x 104 9 CS2 8

7 V µ 6

5 COS

4 DMS

Signal Response, 2

1 H2S MeSH 0 0 10 20 30 40 50 60 Incubation Time (minutes)

Figure 2.3 Stability of sulfur compounds in 1L solid PFA incubation chamber over time. A 10 ppb mixture of the 5 sulfur compounds was injected into the headspace of the chamber at t = 0. 10 mL subsamples of the headspace were measured on the GC/SCD over time.

Lab incubations were performed in two configurations: static and dynamic (Figure 2.4). Both cases used the same PFA incubation chambers, plumbed differently. Air was sampled directly out of the chamber headspace into the GC/SCD inlet system.

In a static measurement, a soil subsample was sealed in the 1 L chamber and at least three subsamples of the chamber headspace were analyzed. The change in concentration over time was measured to calculate the flux of sulfur compounds between the soil and air. One disadvantage to this arrangement is the under- pressurization of soil samples during headspace sampling. A 50 mL aliquot of gas must replace the 50 mL sample being analyzed; here we used a glass and Teflon syringe through a septum port on the lid of the chamber. This dilutes (or amends) the headspace concentrations of compounds of interest by 5% and increases the uncertainty in the flux measurements.

Once steady-state is achieved, dynamic flow-through incubations require concentrations of a compound to be known from two sample streams: entering the chamber and leaving the chamber. Here a micro-diaphragm pump (model UNMP830, KNF Neuberger Inc., Trenton, NJ, USA) was used to move air through the chamber at 0.1 to 0.7 liters per minute; flows were adjusted with a needle valve. Ambient air was used since sulfur-free sweep gas results in spurious COS fluxes (De Mello and Hines, 1994). A second chamber with 100 mL of distilled water was attached to act as a buffer volume and to prevent soils from drying out during the incubation.

Static Lab Incubation Dynamic Lab Inucubation

to pump to GC to GC for pressure “blank” ambient equalization air port to GC

100g soil 100g soil distilled subsample subsample water

Figure 2.4 Two possible configurations of lab incubation chambers.

2.3 Static flux chambers for in situ deployment

All in situ flux measurements reported in this dissertation were produced by deploying two-component static flux chambers in the field and measuring the gas samples in the laboratory. A rectangular frame base was installed several centimeters into the soil, taking care to minimally disturb plant roots and the soil surface within the chamber footprint. The site was then allowed at least an hour to equilibrate. A chamber lid with internal-mixing fans was attached to the base, an event that defined time 0 for a flux measurement. With the chamber closed, subsamples of the headspace were collected at regular intervals; a vent line was opened during sampling to prevent under-pressurization. Pressure and temperature for each chamber was noted and used to calculate fluxes in terms of moles per unit time later on. At least three samples over 31 to 55 minutes were collected for each flux calculation to observe any non-linear feedbacks from artificially exposing soil-plant systems to higher (or lower) concentrations of gases in the chamber headspace. A greater discussion of static, non-steady state flux chamber deployment can be found in Livingston and Hutchinson (1995).

PTFE/PFA-linedPTFE/PFA-lined AluminumAluminum sample sample vent vent collection injection collection injection port motor port PTFE- PTFE Lilm coated fan fan to DC power source fan chamber chamber lid lid clamp channel with d.i. chamber chamber pla nts water base base pla nts soil soil Figure 2.5 PTFE/PFA-lined Lexan and Aluminum-only static flux chambers.

2.3.1 Aluminum-only static flux chambers During the field outing in Texas described in Chapter 3, unlined aluminum chambers were deployed and only COS measurements were reported. We conducted a series of 5 outings to Port Aransas, TX and samples from last two outings (TX4 and TX5, reported in Chapter 3) were analyzed on the GC/MS system with a COS standard. To calculate the uncertainty of field fluxes due to the chamber material, the flux chamber was prepared as it was in the field and a sheet of PTFE film was affixed to bottom of the chamber base. Air samples collected in 3 L SilcoCans were measured on the GC/MS. COS concentrations within the chamber varied by less than 5% (calculated by the standard deviation divided by the mean of measurements) over a 32 minute incubation period, twice the experiment duration in the field.

2.3.2 PTFE and FEP-lined chambers To measure field fluxes for the suite of sulfur gases (H2S, CH3SH, COS, DMS, and CS2), PTFE/FEP-lined two-component chambers (Figure 2.5) were used in the field outside of Santa Cruz, CA, USA (Chapter 4) and Puerto Rico (mentioned in Chapter 1). The chambers were constructed out of Lexan plastic, bolted into a box-shape with space along each vertex. PTFE plastic film was fashioned into a liner using an impulse heat sealer. The seams of the liner were threaded through the vertexes of the Lexan box to keep the film taut and the box volume constant. A flat Lexan lip with weather stripping was added to the bottom of the box to aid in the compression seal between the lid and base (Figure 2.6).

chamber lid Lexan Lexan PTFE Lilm

weather stripping clamp

Aluminum Bytac FEP

chamber Aluminum base Figure 2.6 Detail of the chamber lid and chamber base compression seal and construction for the Aluminum/FEP chamber. PTFE film was pulled taught across the Lexan frame then bolted into a box shape. The aluminum base was lined with FEP in the form of Bytac adhesive FEP/vinyl sheeting.

The aluminum frame base was covered in Bytac adhesive FEP (Saint Gobain Company, Paris, France). The base was installed into the soil-root system with a shovel or hand trowel, and the liner was scratched by soil and plant matter over time. The liner was replaced after every 5 deployments to maintain chamber inertness.

To assess chamber blank flux, the base was wrapped in PTFE film and the chamber lid was prepared for deployment. After clamping the lid to the base and activating the internal mixing fan, the chamber was injected with 10 mL of a sulfur standard containing COS, DMS, and CS2, resulting in the addition of ≈ 200 ppt to the headspace. Since COS has an ambient concentration of ≈ 500 ppt, the concentration in the headspace after injection was about 500 ppt higher than DMS and CS2. The headspace was sub-sampled by linking directly to the GC/SCD inlet system, and samples were taken as quickly as they could be analyzed by this method. The uncertainty, defined as the standard deviation of concentration divided by total concentration, was less than 0.06 for all gases. A representative blank is depicted in Figure 2.7.

800

COS

600

400 DMS

concentration (ppt) 200 CS2

0 0 20 40 60 80 100 120 experiment time (min)

Figure 2.7 The concentration of COS, DMS, and CS2 in a blank chamber over time. DMS and CS2 values were measured simultaneously, but are offset here for ease of reading.

2.4 Limitations

Here a reliable technique for measuring ambient level COS, DMS, and CS2 fluxes with lab and field methods was described; however, a robust method for measuring ambient MeSH and H2S exchange is still needed. The new generation flasks from Restek can capture and store ambient levels of the suite of sulfur compounds which can be directly injected onto the GC/SCD (Khan et al., 2012), but even the lab incubations chambers made from solid PFA are not inert enough to hold < 1 ppb concentrations of MeSH and H2S for more than a few minutes.

Commercially available, compound specific lasers have been recently developed for H2S, though the detection limits are ≈ 10 ppb (Dong et al., 2011). With better mirrors, the detection limit could be reduced to 1 ppb (Manish Gupta, Los Gatos Research, Los Gatos, CA, USA, personal communication), though this still falls short of the parts-per-trillion concentrations found in non-wetland ecosystems.

Since amorphous flasks were successful in holding H2S in whole air samples for a short time, it could be that amorphous silicon-lined stainless steel chambers are the solution to lab-based low-level sulfur gas investigations. However, conventional gaskets (viton, neoprene) caused analyte loss, and the issue of how to seal such chambers remains unsolved.

Works Cited

Aydin, M., De Bruyn, W.J., Saltzman, E.S., 2002. Preindustrial atmospheric carbonyl sulfide (OCS) from an Antarctic ice core. Geophysical Research Letters 29. Benner, R.L., Stedman, D.H., 1989. Universal sulfur detection by chemiluminescence. Analytical Chemistry 61, 1268–1271. De Mello, W.Z., Hines, M.E., 1994. Application of static and dynamic enclosures for determining dimethyl sulfide and carbonyl sulfide exchange in Sphagnum peatlands: Implications for the magnitude and direction of flux. Journal of Geophysical Research 99, 14–601. Dong, F., Junaedi, C., Roychoudhury, S., Gupta, M., 2011. Rapid, online quantification of H2S in JP-8 fuel reformate using near-infrared cavity-enhanced laser absorption spectroscopy. Analytical Chemistry 83, 4132–4136. Khan, M.A.H., Whelan, M.E., Rhew, R.C., 2012. Analysis of low concentration reduced sulfur compounds (RSCs) in air: Storage issues and measurement by gas chromatography with sulfur chemiluminescence detection. Talanta 88, 581– 586. Livingston, G.P., Hutchinson, G.L., 1995. Enclosure-based measurement of trace gas exchange: applications and sources of error, in: Biogenic Trace Gases: Measuring Emissions from Soil and Water. pp. 14–51. Montzka, S.A., Aydin, M., Battle, M., Butler, J.H., Saltzman, E. S., Hall, B.D., Clarke, A.D., Mondeel, D., Elkins, J.W., 2004. A 350-year atmospheric history for carbonyl sulfide inferred from Antarctic firn air and air trapped in ice. Journal of Geophysical Research: Atmospheres 109. Prinn, R.G., 2003. The cleansing capacity of the atmosphere. Annual Review of Environment and Resources 28, 29–57. Rhew, R.C., Teh, Y.A., Abel, T., 2007. Methyl halide and methane fluxes in the northern Alaskan coastal tundra. Journal of Geophysical Research: Biogeosciences 112. Shearer, R.L., O’Neal, D.L., Rios, R., Baker, M.D., 1990. Analysis of sulfur compounds by capillary column gas chromatography with sulfur chemiluminescence detection. Journal of Chromatographic Science 28, 24–28. Whelan, M.E., Min, D., Rhew, R.C., 2013. Salt marshes as a source of atmospheric carbonyl sulfide. Atmospheric Environment 73, 131-137. Yan, X., 2002. Sulfur and nitrogen chemiluminescence detection in gas chromatographic analysis. Journal of Chromatography A 976, 3–10.

3. Salt marsh vegetation: a carbonyl sulfide (COS) source to the atmosphere

The material for this chapter is reproduced by permission of Elsevier Ltd., and was previously published as: Whelan, Mary E.; Min, Dong-Ha; Rhew, Robert C. (2013). Salt marsh vegetation as a carbonyl sulfide (COS) source to the atmosphere. Atmospheric Environment, 73, 131-137. DOI: 10.1016/j.atmosenv.2013.02.048.

3.1 Introduction

The gas carbonyl sulfide (COS) is the most abundant reduced sulfur compound in the atmosphere. During quiescent periods of volcanic activity, COS can act as the sulfur source for the persistent layer of sulfate aerosols in the stratosphere, where they scatter incoming solar radiation and are implicated in ozone depletion (Crutzen, 1976; Pitari et al., 2002; Notholt et al., 2003). Understanding the global budget of COS is important in predicting how Earth’s radiative balance might respond to changing anthropogenic sulfur inputs from biomass burning and shifting consumption patterns of fossil fuels.

Global mean concentrations of COS have been stable since 2000, indicating that sources and sinks are in balance (Montzka et al., 2007). Known sources of COS include outgassing from oceans, biomass burning, and photo-oxidation of dimethyl sulfide and carbon disulfide. The largest sink of COS in the troposphere is uptake by vegetation, followed by consumption in oxic soils and oxidation by the OH radical (Kettle et al., 2002). In our present state of knowledge regarding the global COS budget, there is a large missing source of approximately 200 Tg S, located possibly in the tropics and subtropics (Sandoval-Soto et al., 2005; Suntharalingam et al., 2008).

Atmospheric COS could be used to estimate gross primary productivity (GPP), an ecosystem parameter that cannot easily be measured directly. COS is quickly and irreversibly hydrolyzed into H2S and CO2 via the enzyme carbonic anhydrase in plant leaves (Protoschill-Krebs and Kesselmeier, 1992). Since plants take up COS at a predictable ratio to CO2 (Sandoval-Soto et al., 2005; Geng and Mu, 2006), atmospheric concentrations of COS can act as a proxy for GPP, separating respiration and photosynthesis by a straightforward method. Many groups have proposed that COS vegetative uptake measurements could constrain estimates of GPP (Campbell et al., 2008; Suntharalingam et al., 2008; Montzka et al., 2007; Seibt et al., 2010; Stimler et al., 2010). However, to our knowledge only one published study has actually attempted to do so (Blonquist et al., 2011).

The presence of large terrestrial sources of COS can confound its use as a proxy for the uptake of CO2. It is already known that soils can act as both a source and sink for COS. For example, soils under oxic conditions tend to take up reduced sulfur gases, 20

whereas anoxic conditions lead to production (Watts, 2000), excepting where soil samples were subjected to orders of magnitude higher COS concentrations than are found in ambient air (Lehmann and Conrad, 1996). Higher plants are considered to be the major sink of COS from the troposphere via foliar absorption. However, some plant species can produce more COS than they consume (Kanda et al., 1995; Piluk et al., 2001; Sandoval-Soto et al., 2005; Geng and Mu, 2006) or can act as either a source or a sink of COS (Geng and Mu, 2006).

Rates of transformation and the species of reduced sulfur gases that are emitted from soil depend on factors that affect soil microbial communities, like temperature and redox potential. In general, wetlands and estuarine soils have been reported as sources of COS (de Mello and Hines, 1994; Devai and DeLaune, 1995; Watts, 2000; Li et al., 2006; Yi et al., 2008). The presence of wetland plants can overcome COS production in soils, yielding a net sink (Fried et al., 1993; de Mello and Hines, 1994; Yi et al., 2008), but not always (DeLaune et al., 2002). Most of these studies measured wetland COS exchange in situ, but did not compare intact plots with and without wetland plants to assess their influence on COS net flux. Of these, the largest emission rates were found in a subtropical coastal wetland (DeLaune et al., 2002). Here we broaden the study of subtropical salt marshes to explore the influence of vegetation on COS net fluxes.

3.2 Materials and Methods

3.2.1 Site Description Flux chamber experiments were performed at a coastal salt marsh (27° 38’N, 97° 12’W) on the bay side of Mustang Island, a barrier island on the Gulf of Mexico near Corpus Christi, TX, USA (Figure 3.1). This marsh is part of the Mollie Beattie Coastal Habitat Community, a 1000 acre natural preserve set aside in 1996 to protect intertidal habitats for endangered birds. Measurements focused on sites at a tidally influenced salt marsh vegetated with Batis maritima (saltwort), a common salt marsh plant.

Figure 3.1 Mollie Beattie Coastal Habitat Community near Corpus Christi, Texas, USA. This subtropical salt marsh was inundated with tidal water as well as water percolating vertically through the soil during daily tidal flows.

Avg Soil Chamber Soil Temp Avg COS flux Avg CO2 flux Outing Site Moisture Biomass Air Temp Temp 5 cm ± sd ± sd (μmol m-2 s- (%VWC) (g dry) (°C) (°C) (°C) (pmol m-2 s-1) 1) Jul V1A 38 80.7 30.3 ± 2.6 30.7 ± 2.4 32.8 ± 2.4 85 ± 20 2.4 ± 0.4 2009 V1B 52 76.5 30.0 ± 2.1 30.4 ± 2.3 32.0 ± 2.5 59 ± 9 2.8 ± 0.6 S1 46 0 29.7 30.8 30.8 28 0.6

52- inundate Nov V2A d 35.5 23.2 ± 1.6 23.3 ± 2.8 22.5 ± 1.9 60 ± 20 0.62 ± 0.2 52- inundate 2009 V2B d 34.5 23.4 ± 1.5 23.6 ± 2.4 22.8 ± 1.7 42 ± 26 0.58 ± 0.3 inundate S2 d 0 24.8 28 24.3 26 -0.09 Table 3.1 Descriptions of 6 sites used in this study. %VWC is volumetric water content. Mean values are given with errors as expressed by the standard deviation of flux chamber averages where possible.

3.2.2 Diurnal COS Flux Measurements Two field campaigns were carried out on July 19-20 and November 6-7, 2009 (Table 3.1) to capture the range of annual temperature variation. In both campaigns, two plots containing B. maritima were selected within 5 meters of each other. None of the plots overlapped. Flux chamber experiments were conducted at both plots every 3 to 6 hours over a 24 hour period, for a total of 6 flux measurements per vegetated site. A third site without any vegetation was selected near the other two to measure fluxes from soil experiencing a similar hydrologic regime. For the July outing, the vegetated sites were denoted by V1A and V1B, with the soil-only site labeled S1. For November, the vegetated sites were V2A and V2B, with the soil-only site S2. The soil at the sampling sites was primarily coarse sand.

Air samples were collected using two-component, aluminum, static flux chambers (Livingston and Hutchinson, 1995). Chambers were tested for inertness by wrapping the base with Teflon film and enclosing either ambient or elevated (~1000 parts-per-trillion (ppt)) levels of COS with the chamber base and lid. Subsamples of the chamber headspace were analyzed and COS concentrations within the chamber did not vary more than ± 5% over the course of 32 minutes, twice the field incubation time.

The chamber bases with a footprint of 0.2645 m2 were installed at least 3 hours prior to the first measurement with the lid open to the ambient air, minimizing emissions or uptake of gases that may have occurred from disrupting soil and plant roots during base installation. A small gutter filled with distilled water ran along the top edge of the base. The bottom edge of the chamber top was placed into this water-filled channel to create an airtight, enveloping a volume of 188 L. Since the chamber bases enclosed both soil and plant matter, exchange rates are reported per m2 area of soil, the footprint of the chamber bases. 23

Using dark chambers has specific advantages for determining COS fluxes. The absence of light eliminates the COS contribution from DMS and CS2 photo-oxidation. While interactions within seawater have been known to produce COS, these processes are also primarily photochemical (Von Hobe et al., 1999), though dark reactions have been inferred as non-trivial sources (Von Hobe et al., 2001). Dark chambers are expected to reduce stomatal conductance and therefore reduce CO2 and COS fluxes. However, stomatal conductance is enhanced by high COS concentrations, such as had accumulated in the chamber headspace (Stimler et al., 2010). In addition, the short incubation time may have been too brief and relative humidity too high to allow stomata to fully close, as could be the case with the response time for other dicots (Manzoni et al., 2011).

To minimize complications with feedbacks from high headspace concentrations and the lack of sunlight within the chamber, incubation times were short (16 minutes). The experiment began when the chamber lid was placed on the base. Air within the chamber was mixed with two internal fans, and a vent was opened during sample collection to prevent abrupt changes in chamber pressure. The headspace was sampled using previously evacuated 3 L silica-lined stainless steel canisters (Restek, Bellefonte, PA, USA) or 1 L electropolished stainless steel canisters (LabCommerce, Inc., San Jose, CA, USA) at 1, 8 and 16 minutes after closure. Before sampling, tubing between the flask and chamber was evacuated with a syringe to ensure that only the well-mixed headspace air was collected. In addition, 30 mL air samples were taken from the chamber using gastight syringes and pressurized in previously evacuated 20 mL stoppered glass vials (Wheaton, Millville, New Jersey) for CO2 analyses. Ambient air samples outside the chambers were collected in both canisters and vials between flux chamber experiments.

Measured environmental variables for the 24-hour experiments included ambient air temperature, chamber air temperature, soil temperature, groundwater chemistry, soil moisture, biomass of B. maritima, and photosynthetically active radiation (PAR). Temperature was recorded once a minute with self-contained stainless steel thermocouple data loggers (iButtons, Maxim Inc., Sunnyvale, CA, USA). Several iButtons were placed within the chamber as well as under a solar shield nearby to record the air temperature over the course of the experiment. Soil temperature was logged at 5 cm and 10 cm depths in several locations around the chambers. Volumetric water content (VWC) of soil was measured with a ThetaProbe soil moisture sensor (mineral setting, Delta-T Devices, Cambridge, UK). Water depth was measured when chambers were inundated with tidally influenced groundwater during the last 7 chamber measurements of the November outing. Concentration of H2S in the groundwater was determined with a hydrogen sulfide test kit (Hach Company, Loveland, CO, USA). A geopump (Geotech Environmental Equipment, Inc., Denver, CO, USA) was used to pump groundwater from 60 cm below the surface and ground water salinity, temperature, pH, and dissolved oxygen content were measured with a multi-parameter sonde (YSI Inc., Yellow Springs, OH, USA). PAR was observed at a weather station in nearby Copano Bay as part of the Mission- 24

Aransas National Estuarine Research Reserve (http://lighthouse.tamucc.edu/overview/146). All aboveground vegetation within the chamber was harvested after the final experiment to determine total aboveground biomass and estimate plant volume for each chamber.

3.2.3 Analytical Methods Air samples were analyzed twice on a gas chromatograph - mass spectrometer (Agilent GC 6890N/5973) in selective ion monitoring mode. The custom-built cryotrap inlet system and oven settings are detailed elsewhere (Rhew and Abel, 2007; Rhew et al., 2007). A calibration curve for COS was generated using different volume injections of a whole air gas standard (collected at Trinidad Head, CA, USA and calibrated on the NOAA-SIO provisional scale). In addition, the standard was run three times a day to correct for daily instrumental drift. To make higher concentration standards of COS, a 1 ppm COS standard (Scott Specialty Gases) was diluted with ultra-high purity nitrogen in a 3 L Silcosteel canister using an in-house dilution line and compared against the Trinidad Head standard to confirm the concentration. All calibration curves were linear with a correlation coefficient greater than 0.999.

Vial samples were analyzed for CO2 on a Shimadzu GC-14A gas chromatograph with a thermal conductivity detector (TCD) (Shimadzu Scientific Inc., Columbia, Maryland). The calibration standard (997 ppm CO2, Scott Specialty Gases) was analyzed at the start and end of analyses and in between every 10 samples to correct for instrument drift, which did not exceed ± 6%. The overall precision for the GC-TCD instrument was 2%.

3.2.4 Calculating Fluxes Net fluxes of COS and CO2 were calculated by applying a linear and an exponential fit of concentration versus incubation time, then choosing the model with the better goodness of fit to determine the initial rate of concentration change within the flux chamber. A better exponential fit indicates a significant first order feedback effect on trace gas exchange. This rate was multiplied by the number of moles of air in the chamber and divided by the surface area to yield a flux in pmol m-2 s-1. When the exponential fit was used, the production rate k was determined using the equation C(t)=Cmax-(Cmax-C0)exp(-kt), where t is time from the beginning of the chamber experiment, Cmax is the maximum concentration when chamber air and soil pore space are in equilibrium, and C0 is the concentration of COS at t=0 (de Mello and Hines, 1994). Both Cmax and k are solved iteratively, then the flux rate at the start of the incubation (dC/dt)t=0 is calculated as (dC/dt)t=0=k(Cmax-Cair), where Cair is the average concentration of COS measured from ambient air samples. Most experiments showed better exponential fits of chamber COS concentrations versus time, except for V1A at 20:00, V2A at 2:00, 14:00, and 20:00, and V2B at 2:00 and 20:00. All fits had r2 > 0.99, except the soil-only plots (sites S1 and S2, r2=0.93 for both). Errors reported are ± 1 s.d. unless otherwise noted.

3.3 Results

All chamber sites from this subtropical salt marsh at the Mollie Beattie Habitat showed positive net fluxes of COS (i.e. surface to the atmosphere) during the sampling campaigns, but there were distinct differences associated with temperature, presence/absence of vegetation, and soil moisture (Figure 3.2). Ambient air samples collected during both outings had an average COS concentration of 573 ± 57 ppt.

120 V1A V1B

100 S1 V2A V2B S2

) 80 1 − s 2 −

COS flux (pmol m 40

0 0 5 10 15 20 Hour of Day, Standard Local Time

Figure 3.2 Diurnal fluxes of COS from 2 plots containing B. maritima in July (circles, site V1A in gray and V1B in black) and November (triangles, site V2A in gray and V2B in black). Hour of day is reported in U.S. Central Standard Time (CST = GMT - 6 hours). Soil-only plots (squares: open square for July, filled square for November) contained no vegetation. Solid markers indicate that soil was completely inundated during measurement.

3.3.1 Results from July 2009, Summer In the July outing, the two vegetated sites (V1A and V1B) showed very large COS emission rates, averaging 71 ± 19 pmol m-2 s-1. The largest emission rate of 118 pmol m-2 s-1was measured at 1:30PM from Site V1A. Site V1A had drier soil and more biomass (38% VWC and 80.7 g dry wt) than site V1B (52% VWC and 76.5 g dry wt) (Table 3.1). The soil-only site S1 had a flux less than half the average of sites V1A and V1B (Figure 3.2). Fluxes of CO2 were all positive with an average of 2.4 ± 0.7 μmol m-2s-1.

Groundwater temperature, pH, salinity, and dissolved oxygen content were determined every 1 or 1.5 hours during daylight sampling within a few meters of the chamber sites. Water temperature varied from 30.5 °C at 20:25 to 32.7 °C at 15:30. Dissolved oxygen and salinity measurements showed that the groundwater was consistently hypoxic (< 3 - 7% O2 saturation) and hypersaline (59.8 - 66.5 psu). The ground water pH varied from 7.04 to 7.20.

3.3.2 Results from November 2009, Autumn In the November outing, the overall COS emissions at vegetated sites (V2A and V2B) averaged 51 ± 24 pmol m-2 s-1, lower than the previous summer’s average. Biomass measurements at both sites were similar (35.5 and 34.5 g dry wt), about half as much as the July outings. Soil moisture for sites V2A and V2B were approximately the same, varying from 54% to waterlogged (see Table 3.1). Soil within chambers was inundated with hypersaline water that infiltrated the soil profile during tidal changes in sea level; groundwater inundation at the surface of the site occurred through a vertical fluctuation of water table instead of lateral advection of nearby surface seawater. All three of the November sites were flooded between midnight and midday over the course of the measurement period. The shorter B. maritima plants in V2B were almost entirely submerged for the flux measurements at 8:30 and 11:30. Site S2, which had no vegetation and was measured during inundation, produced net COS emissions half the average of the vegetated sites V2A and V2B -2 -1 (Figure 3.2). CO2 fluxes averaged 0.54 ± 0.3 μmol m s .

Groundwater, measured every few hours during daylight sampling, was consistently hypersaline (67.9 - 72.4 psu) and hypoxic (5 - 10% O2 saturation). Temperature ranged from 22.6 to 23.4 °C and pH between 6.88 and 7.03. The odor of H2S was readily detected when soil was disturbed with a shovel and measured ground water -1 H2S concentrations were ~1 mg L .

3.3.3 Comparison of results from different seasons The July outing took place during a severe drought in 2009, and the November outing occurred just after a series of precipitation events. On average, July plots experienced 7 °C higher chamber temperatures and over 100 μmol m-2 s-1 more PAR compared to November plots. Not surprisingly, the hotter summer outing had larger net CO2 emissions. Despite differences in biomass, temperature, and PAR, the unflooded vegetated sites from both outings yielded comparable emission rates. 27

Net emissions from flooded chambers in November were much lower than non- flooded chamber experiments (Figure 3.2). Several centimeters of standing water effectively inhibited soil fluxes. COS from vegetation was also reduced in the case of plot V2B at 2:30 and 8:30, when vegetation was almost entirely covered with water.

For both outings, the soil-only sites showed net emissions, but of lower magnitude than all but two flooded vegetated plots. However, the soil-only fluxes were not measured simultaneously with fluxes from soil-vegetation systems. To compare the soil-only plots to the vegetated plots, we assumed that fluxes from vegetated plots changed linearly between observations of the vegetated sites. Soil only plots were assumed to have a constant emission over the course of the day, although in reality may have varied slightly with diurnal temperature shifts. Comparing the interpolated results from vegetated plots for July shows that vegetated sites V1A and V1B would have emitted 2.0 to 2.7 times more COS than the soil-only plot S1. In November, when the emissions were much more variable, the vegetated sites would have emitted 2.6 to 2.9 times more COS than S2.

3.4 Discussion

A soil-vegetation system has competing processes that determine the sign of the overall COS net flux. Anoxic soils are generally considered to be the largest terrestrial COS source aside from humans and volcanoes, though an order of magnitude smaller than anthropogenic sources (Kettle et al., 2002; Suntharalingham et al., 2008). Plants are considered to be a major sink of COS due to its destruction by the enzyme carbonic anhydrase in plant leaves (Protoschill-Krebs et al., 1996). Our findings suggest that wetland plants may play a role in the transport and production of atmospheric COS as well.

The net uptake or production of COS is sometimes controlled by the ambient atmospheric concentration: for different soils and plants, there exists a compensation point below which there is net emission of COS by soils or plants, and above which more COS is taken up than produced (Kesselmeier and Merk, 1993; Kesselmeier et al., 1999). However, a more recent study found no compensation point for COS uptake on the leaf scale (Stimler et al., 2010). Despite the high ambient concentration of COS that accrued inside flux chambers (sometimes reaching as high as 6300 ppt), the theoretical compensation point for this ecosystem was not exceeded.

The salt marsh soil in this study exhibited positive fluxes similar to previously examined wetland soils (Figure 3.3). Overall, soils observed alone had a smaller reported range of fluxes than when plants and soils were observed together, aside from the rice paddy soils investigated by Liu et al., 2010. In their field experiments, one rice paddy plot effected an uptake of -57.9 pmol m-2 s-1. However, the ambient concentration of COS during the experiment was 3008 ppt, 5 times the global average concentration and in excess of the lab-measured compensation point. In a separate study in which rice paddy soils remained unflooded, the range of fluxes was small but was the opposite sign (i.e. net COS uptake) (Yi et al., 2008). Aside from these notable exceptions, rice paddy, peatland, and the salt marsh soils discussed here all produced more COS than was consumed (Fried et al., 1993; deMello and Hines, 1994; Yi et al., 2008).

Salt marshes appear to be the only ecosystem with consistent net emissions of COS (Figure 3.3). In a different Gulf of Mexico marsh, DeLaune et al. (2002) reported greater maximum emissions than those reported here by a factor of 3; soils were not explored separately in that study. Since plants are known to consume COS in proportion to CO2, it is not surprising that their inclusion in study plots results in net uptake for most biomes, and that measurements of plant and soil systems show a wider range of COS fluxes than soil only plots. Salt marshes have high concentrations of sulfate from salt water and an anoxic soil environment in which reduced sulfur compounds can be produced. Under these conditions, several mechanisms can explain why it appears that only salt marshes are sources of COS.

/ / reported reported avg range +/− error soil+plants This Study soil only DeLaune et al. 2002 SALTMARSH

Fried et al. 1993 deMello & Hines 1994 PEATLAND unflooded flooded Yi et al. 2008 Wetland RICE PADDY Liu et al. 2010

Oxic Liu et al. 2010 Kesselmeier et al. 1999 ARABLE Geng & Mu 2004 LAWN

Li et al. 2006 MEADOW

Kuhn et al. 1999 SAVANNAH FOREST Xu et al. 2002 Simmons et al. 1999 White et al. 2010 Yi et al. 2007 Castro & Galloway 1991 Steinbacher et al 2004 −150 −125 −100 −75 −50 −25 0 25 50 75 100 / / 275 300 −2 −1 COS fluxes (pmol m sec )

Figure 3.3 Summary of in situ COS flux measurements in the literature compared to this study. Gray bars represent values from soil-only plots; white bars represent soils with plants. Diamonds with bars represent mean fluxes and their reported errors; bars without diamonds are used when flux measurements are reported as a range of values only. The soil-only values for Fried et al. 1993, Geng & Mu 2004 and Yi et al. 2008 (flooded) are from plots that would normally be vegetated, but the vegetation was removed for the experiment. The Yi et al. 2008 mean flux for unflooded rice paddy soil was from a plot that was both unplanted and unflooded. The field measurements of Liu (2010) were made with ambient COS concentrations of up to 3008 ppt, 5 times the average global concentration.

COS production at our site was not likely to be from dark production of COS in seawater. The soil was isolated from the chamber headspace by tidal water during the November 2009 outing. Flux values from emergent vegetation and salt water were still positive and larger than a similarly flooded, nearby soil-only plot (Figure 3.2). Where the emissions from the soil-only plots exceeded those of the flooded, vegetated plot V2B, the plants were almost entirely submerged. In other words, emissions dropped when plants were mostly underwater, thus dark production of COS in sea water would not account for the continued observation of large emissions in flooded chambers.

Salt marsh plants could transport the COS produced in the sulfate-rich, anoxic soils to the atmosphere by diffusion. This would suggest that wetland plants rooted in anoxic soil contribute to the atmospheric sulfur burden by ventilating their roots, a necessary process for root respiration in waterlogged soils. Rice has been shown to transport trace gases via molecular diffusion, eluding oxidation at the soil-water interface (Chanton et al., 1997). However, COS emissions from a flooded rice paddy were not significantly larger than those from unplanted paddy soil (Yi et al., 2008; Liu et al., 2010). Since COS is not readily oxidized in the atmosphere, the longer residence time of COS diffusing through a soil profile versus a plant aerenchyma would not necessarily result in the destruction of half the total COS originally produced in the soil. Escaping through plant shoots may allow COS to bypass consumption by oxic soil near the surface. Barring this, some further interaction with B. maritima or its rhizosphere should account for the larger flux to the atmosphere.

Salt marsh plants have different strategies to thrive in a saline environment. B. maritima collects salts (including sulfates) in its leaves and sheds them to manage internal salt concentrations (Luttge et al., 1989). The high levels of sulfate in plant tissue may create conditions for COS production within the above ground biomass. For the November diurnal measurements site V2A, with taller vegetation, had a consistently higher flux than V2B. This discrepancy between the sites increased when nearly all the vegetation in site V2B was covered with water.

Though COS net emissions varied spatially, fluxes from vegetated sites appeared to respond to the same set of variables. In July, site V1A was consistently higher than site V1B by 26 ± 13 pmol m-2 s-1, though they both exhibited the same diurnal pattern (Figure 3.2). In other words, fluxes co-varied and were correlated (r2 = 0.5). In November, the two vegetated sites showed a consistent offset from each other and a similar diurnal pattern. Although these sites exhibited a larger absolute variation in fluxes than in the July outing, it appears to be the result of tidal influences rather than other factors like sunlight and temperature.

On the other hand, the parameter that correlated best with the COS fluxes for non- flooded chambers was soil temperature at 5 cm depth (Figure 3.4), rather than the temperature of chamber air, of soil at 10 cm depth, or of ambient air. This suggests a shallow subsurface process influenced the magnitude of fluxes in unsaturated soil, though not at a consistent rate between seasons. This hypothesis could be tested in a future study by inserting an inert model of a B. maritima root (e.g. a short length of tubing) into the soil and comparing it to adjacent vegetated plots.

120

110

100 ) 1 − 2 s 90

2 r =0.59 −

80 r2=0.98

60 2

COS flux (pmol m r =0.46 50

40 r2=0.61

30 20 25 30 35 Average Soil Temperature at 5cm(°C)

Figure 3.4 COS flux from July (circles, gray represents site V1A, black V1B) and non-flooded November (triangles, gray represents site V2A, black V2B) chamber experiments versus the soil temperature at 5 cm averaged over the course of the incubation.

Whereas Li et al. (2006) found a positive relationship between COS consumption and vegetation biomass, we did not find a strong connection between COS production and above ground biomass. B. maritima is a halophyte with extensive adventive roots (Johnson, 1935) that we found to reach 20 cm below ground. If the large subsurface biomass component contributes to the COS flux, then this may explain why the above ground biomass here did not appear to explain the variability between plots. 32

Although an interaction with carbonic anhydrase within plant leaves has been known to mediate the exchange of both CO2 and COS (Protoschill-Krebs et al., 1996), the two gases show net emissions to the atmosphere and demonstrate no clear relationship (Figure 3.5). In other words, foliar deposition sink terms for both COS and CO2 were overwhelmed by source terms. This is not surprising, given that the dark chambers would have shut down photosynthesis. However, the lack of a clear production ratio of COS:CO2 in dark chambers, even at nighttime, suggests that the overall COS:CO2 net flux ratios were highly variable. PAR data collected at a nearby weather station followed the summer outing emissions well (co-varying with temperature), but the relationship to autumn fluxes was inconsistent (data not shown). The findings presented here suggest that the COS:CO2 uptake relationship cannot be applied to salt marsh ecosystems, where both COS and CO2 emissions confound measurements of their consumption.

4.5 r2=0.31 )

1 4 − s 2 − 3.5

3 r2=0.05 2.5

2 flux (micro mol m 2

CO 1.5

0.5 0 20 40 60 80 100 120 −2 −1 COS flux (pmol m s )

Figure 3.5 Net CO2 flux versus net COS flux for salt marsh chamber experiments in July (circles, V1A in gray and V1B in black) and November (triangles, V2A in gray and V2B in black). Solid symbols represent flooded sites.

According to Suntharalingham et al. (2008), the sinks of COS totaled 740 Gg S yr-1 while the sources comprised 505 Gg S yr-1. This calculation used the mean estimated source from all anoxic soils and wetlands of 26 Gg S yr-1 (Kettle et al., 2002). In this study, the average flux from vegetated sites was 60 ± 20 pmol COS m-2 s-1 and the average unvegetated flux was 27 ± 1 pmol COS m-2 s-1. If we assume that our sites are representative of all salt marshes and half of the wetland area is vegetated, then we can conduct a scaling exercise to assess the potential of salt marsh contribution to anoxic soils globally. If we only include salt marshes and estuaries at 0.03×106 km2 (Watts, 2000), then the contribution would be 1.3 ± 0.4 Gg S yr-1. If our measurements are representative of all swamp and marsh globally at 2×106 km2 (Sandoval-Soto et al., 2005), the resulting emission would be 87 ± 29 Gg S yr-1, or over 3 times the previous anoxic soil contribution estimate.

3.5 Conclusions

Acknowledging specific plants (B. maritima) and ecosystems (coastal wetlands) can act as large COS sources will help clarify the balance of the global biogeochemical sulfur cycle. Plants are considered to be the largest sink of COS in the troposphere, however, here we show that salt marsh plants may also act as a large net emitter of COS. Fluxes of COS from vegetated plots were twice that of soil-only plots. In addition, when soil systems were separated from the chamber headspace by water, fluxes from emergent vegetation and hypersaline water remained high. The magnitude of fluxes in this study are large compared to those found in other ecosystems. If COS production is biotic and regulated by temperature, emissions from tropical, sulfur-rich ecosystems may be even larger. Thus some of the missing source of COS in the tropics and subtropics (Suntharalingam, et al., 2008) may be found from coastal terrestrial biomes.

Works Cited

Campbell, J.E., Carmichael, G.R., Chai, T., Mena-Carrasco, M., Tang, Y., Blake, D.R., Blake, N.J., Vay, S.A., Collatz, G.J., Baker, I., Berry, J.A., Montzka, S.A., Sweeney, C., Schnoor, J.L., Stanier, C.O., 2008. Photosynthetic control of atmospheric carbonyl sulfide during the growing season. Science 322, 1085–1088. Castro, M.S., Galloway, J.N., 1991. A comparison of sulfur-free and ambient air enclosure techniques for measuring the exchange of reduced sulfur gases between soils and the atmosphere. Journal of Geophysical Research 96, 15427–15. Chanton, J.P., Whiting, G.J., Blair, N.E., Lindau, C.W., Bollich, P.K., 1997. Methane emission from rice: Stable isotopes, diurnal variations, and CO2 exchange. Global Biogeochemical Cycles 11, 15–27. Crutzen, P.J., 1976. The possible importance of CSO for the sulfate layer of the stratosphere. Geophysical Research Letters 3, 73–76. de Mello, W.Z., Hines, M.E., 1994. Application of static and dynamic enclosures for determining dimethyl sulfide and carbonyl sulfide exchange in Sphagnum peatlands: Implications for the magnitude and direction of flux. Journal of Geophysical Research 99, 14–601. DeLaune, R.D., Devai, I., Lindau, C.W., 2002. Flux of reduced sulfur gases along a salinity gradient in Louisiana coastal marshes. Estuarine, Coastal and Shelf Science 54, 1003–1011. Devai, I., DeLaune, R.D., 1995. Formation of volatile sulfur compounds in salt marsh sediment as influenced by soil redox condition. Organic Geochemistry 23, 283–287. Fried, A., Klinger, L.F., Erickson, D.J., 1993. Atmospheric carbonyl sulfide exchange in bog microcosms. Geophysical Research Letters 20, 129–132. Geng, C., Mu, Y., 2004. Carbonyl sulfide and dimethyl sulfide exchange between lawn and the atmosphere. Journal of Geophysical Research 109, D12302. Geng, C., Mu, Y., 2006. Carbonyl sulfide and dimethyl sulfide exchange between trees and the atmosphere. Atmospheric Environment 40, 1373–1383. Johnson, D.S., 1935. The development of the shoot, male flower and seedling of Batis maritima L. Bulletin of the Torrey Botanical Club 62, 19–31. Kanda, K., Tsuruta, H., Minami, K., 1995. Emissions of biogenic sulfur gases from maize and wheat fields. Soil Science and Plant Nutrition 41, 1–8. Kesselmeier, J., Merk, L., 1993. Exchange of carbonyl sulfide (COS) between agricultural plants and the atmosphere: Studies on the deposition of COS to peas, corn and rapeseed. Biogeochemistry 23, 47–59. Kesselmeier, J., Teusch, N., Kuhn, U., 1999. Controlling variables for the uptake of atmospheric carbonyl sulfide by soil. Journal of Geophysical Research 104, 11577–11584. Kettle, A.J., Kuhn, U., Von Hobe, M., Kesselmeier, J., Andreae, M.O., 2002. Global budget of atmospheric carbonyl sulfide: temporal and spatial variations of the dominant sources and sinks. Journal of Geophysical Research 107, 4658- 4674.

Kuhn, U., Ammann, C., Wolf, A., Meixner, F.X., Andreae, M.O., Kesselmeier, J., 1999. Carbonyl sulfide exchange on an ecosystem scale: soil represents a dominant sink for atmospheric COS. Atmospheric Environment 33, 995–1008. Lehmann, S., Conrad, R., 1996. Characteristics of turnover of carbonyl sulfide in four different soils. Journal of Atmospheric Chemistry 23, 193–207. Li, X., Liu, J., Yang, J., 2006. Variation of H2S and COS emission fluxes from Calamagrostis angustifolia wetlands in Sanjiang Plain, Northeast China. Atmospheric Environment 40, 6303–6312. Liu, J., Geng, C., Mu, Y., Zhang, Y., Xu, Z., Wu, H., 2010. Exchange of carbonyl sulfide (COS) between the atmosphere and various soils in China. Biogeosciences 7, 753–762. Livingston, G.P., Hutchinson, G.L., 1995. Enclosure-based measurement of trace gas exchange: applications and sources of error, in: Matson, P.A., Harriss, R.C. (Eds.), Biogenic Trace Gases: Measuring Emissions from Soil and Water, Blackwell Science Ltd, Oxford. pp. 14–51. Luttge, U., Popp, M., Medina, E., Cram, W.J., Diaz, M., Griffiths, H., Lee, H.S.J., Schafer, C., Smith, J.A.C., Stimmel, K.H., 1989. Ecophysiology of xerophytic and halophytic vegetation of a coastal alluvial plain in northern Venezuela. V. The Batis maritima-Seuvium portulacastrum vegetation unit. New Phytologist 111, 283–291. Montzka, S.A., Calvert, P., Hall, B.D., Elkins, J.W., Conway, T.J., Tans, P.P., Sweeney, C., 2007. On the global distribution, seasonality, and budget of atmospheric carbonyl sulfide (COS) and some similarities to CO2. Journal of Geophysical Research 112. Notholt, J., Kuang, Z., Rinsland, C.P., Toon, G.C., Rex, M., Jones, N., Albrecht, T., Deckelmann, H., Krieg, J., Weinzierl, C., Bingemer, H., Weller, R., Schrems, O., 2003. Enhanced upper tropical tropospheric COS: Impact on the stratospheric aerosol layer. Science 300, 307–310. Piluk, J., Hartel, P.G., Haines, B.L., Giannasi, D.E., 2001. Association of carbon disulfide with plants, in the family Fabaceae. Journal of Chemical Ecology 27, 1525– 1534. Pitari, G., Mancini, E., Rizi, V., Shindell, D.T., 2002. Impact of future climate and emission changes on stratospheric aerosols and ozone. Journal of the Atmospheric Sciences 59, 414–440. Protoschill-Krebs, G., Kesselmeier, J., 1992. Enzymatic pathways for the consumption of carbonyl sulphide (COS) by higher plants. Botanica Acta 105, 206–212. Protoschill-Krebs, G., Wilhelm, C., Kesselmeier, J., 1996. Consumption of carbonyl sulphide (COS) by higher plant carbonic anhydrase (CA). Atmospheric Environment 30, 3151–3156. Rhew, R.C., Abel, T., 2007. Measuring simultaneous production and consumption fluxes of methyl chloride and methyl bromide in annual temperate grasslands. Environmental Science & Technology 41, 7837–7843. Rhew, R.C., Teh, Y.A., Abel, T., 2007. Methyl halide and methane fluxes in the northern Alaskan coastal tundra. Journal of Geophysical Research 112, G02009. 36

Sandoval-Soto, L., Stanimirov, M., Von Hobe, M., Schmitt, V., Valdes, J., Wild, A., Kesselmeier, J., 2005. Global uptake of carbonyl sulfide (COS) by terrestrial vegetation: Estimates corrected by deposition velocities normalized to the uptake of carbon dioxide (CO2). Biogeosciences 2, 125–132. Seibt, U., Kesselmeier, J., Sandoval-Soto, L., Kuhn, U., Berry, J.A., 2010. A kinetic analysis of leaf uptake of COS and its relation to transpiration, photosynthesis and carbon isotope fractionation. Biogeosciences 7, 333–341. Simmons, J.S., 1999. Consumption of atmospheric carbonyl sulfide by coniferous boreal forest soils. Journal of Geophysical Research 104, 11569–11576. Steinbacher, M., Bingemer, H.G., Schmidt, U., 2004. Measurements of the exchange of carbonyl sulfide (OCS) and carbon disulfide (CS2) between soil and atmosphere in a spruce forest in central Germany. Atmospheric Environment 38, 6043–6052. Suntharalingam, P., Kettle, A.J., Montzka, S.M., Jacob, D.J., 2008. Global 3-D model analysis of the seasonal cycle of atmospheric carbonyl sulfide: Implications for terrestrial vegetation uptake. Geophysical Research Letters 35, L19801. Von Hobe, M.V., Cutter, G.A., Kettle, A.J., Andreae, M.O., 2001. Dark production: A significant source of oceanic COS. Journal of Geophysical Research 106, 31217–31226. Von Hobe, M., Kettle, A.J., Andreae, M.O., 1999. Carbonyl sulphide in and over seawater: summer data from the northeast Atlantic Ocean. Atmospheric Environment 33, 3503–3514. Watts, S.F., 2000. The mass budgets of carbonyl sulfide, dimethyl sulfide, carbon disulfide and hydrogen sulfide. Atmospheric Environment 34, 761–779. White, M.L., Zhou, Y., Russo, R.S., Mao, H., Talbot, R., Varner, R.K., Sive, B.C., 2010. Carbonyl sulfide exchange in a temperate loblolly pine forest grown under ambient and elevated CO2. Atmospheric Chemistry and Physics 10, 547–561. Xu, X., Bingemer, H.G., Schmidt, U., 2002. The flux of carbonyl sulfide and carbon disulfide between the atmosphere and a spruce forest. Atmospheric Chemistry and Physics 2, 171-181. Yi, Z., Wang, X., Sheng, G., Fu, J., 2008. Exchange of carbonyl sulfide (OCS) and dimethyl sulfide (DMS) between rice paddy fields and the atmosphere in subtropical China. Agriculture, Ecosystems & Environment 123, 116–124. Yi, Z., Wang, X., Sheng, G., Zhang, D., Zhou, G., Fu, J., 2007. Soil uptake of carbonyl sulfide in subtropical forests with different successional stages in south China. Journal of Geophysical Research 112, D08302.

4. Carbonyl sulfide produced by abiotic thermal and photo-degradation of soil organic matter from wheat field substrate

Carbonyl sulfide (COS) is a reduced sulfur gas that is taken up irreversibly in plant leaves proportionally with CO2, allowing its potential use as a tracer for gross primary production. Recently, wheat field soil at the Southern Great Plains Atmospheric Radiation Measurement (SGP-ARM) site in Lamont, Oklahoma was found to be a measureable source of COS to the atmosphere. To understand the mechanism of COS production, soil and root samples were collected from the site and incubated in the laboratory over a range of temperatures (15-34 °C) and light conditions (light and dark). Soil exhibited mostly COS net uptake from the atmosphere in dark and cool (<22-25 °C) trials. COS emission was observed from dark incubations at high temperatures (> 25 °C), consistent with field observations, and at all temperatures when a full spectrum lamp (max wavelength 600nm) was applied. Sterilized soils exhibited only COS production that increased with temperature, supporting the hypothesis that a) COS production in these soils is abiotic, b) production is directly influenced by temperature and light, and c) some COS consumption in soils is biotic.

4.1 Introduction

Over the last decade, the relationship between concentrations of carbonyl sulfide (COS) gas and gross primary production has been explored as a tool for constraining estimates of fluxes in both the carbon and sulfur cycle (Asaf et al., 2013; Berry et al., 2013; Billesbach et al., 2014; Blonquist et al., 2011; Campbell et al., 2008; Sandoval- Soto et al., 2005; Seibt et al., 2010; Suntharalingam et al., 2008). CO2 and COS diffuse through plant stomata; while CO2 is simultaneously released through autotrophic respiration, generally COS is not produced by leaves (reviewed in Wohlfahrt et al., 2012). When the ratio of CO2 to COS uptake by ecosystem vegetation is known, gross primary production could be calculated from measurements of COS and CO2 concentrations (Seibt et al., 2010). This approach becomes complicated if there are other large sources or sinks of atmospheric COS in the ecosystem.

COS is well-mixed in the atmosphere with an average concentration of 476±4 ppt in the Northern Hemisphere (Montzka et al., 2007). The primary source of COS is thought to be from the oceans, whereas the largest tropospheric sink is terrestrial vegetation (Chin and Davis, 1993; Kettle et al., 2002, 2001; Montzka et al., 2007). COS is also destroyed in the stratosphere, contributing to the persistent sulfate aerosol layer; however, this indirect cooling effect of COS is approximately canceled out by its direct global warming potential (Brühl et al., 2012). In the global budget, the anthropogenic source fraction has been estimated to be 34-43%, based on ice core records which show an increase in atmospheric COS since 1850, perhaps from a combination of industrial emissions and deforestation (Montzka et al., 2004).

Soils were originally treated as a source of COS because early observations were often made in chambers with an initially sulfur-free headspace. This led to high COS emissions compared to measurements in ambient air (Castro and Galloway, 1991; de Mello and Hines, 1994). Using accurate but sparse data, Watts (2000) divided soils into two categories: anoxic soils that produce COS and oxic soils that are a sink for COS, suggesting that redox potential alone drives net fluxes of COS. There are few in situ soil-only field experiments, many involving plots that simply had plants removed by cutting or whole plant extraction before measurements commenced (e.g., Fried et al., 1993; Geng and Mu, 2004; Yi et al., 2008). The parameterization of the soil sink in the modeling effort by Kettle et al., (2002) relied on a laboratory- based study by Kesselmeier et al. (1999), using sieved soils in a climate controlled chamber. The soil uptake mechanism is thought to be enzymatic (Conrad and Meuser, 2000; Kesselmeier et al., 1999; Lehmann and Conrad, 1996; Van Diest and Kesselmeier, 2008), probably due to carbonic anhydrase, the same enzyme that destroys COS in plant leaves (Protoschill-Krebs and J. Kesselmeier, 1992; Protoschill-Krebs et al., 1996).

The production of COS in soils is partially understood. In a laboratory-based redox manipulation experiment, under lower redox conditions more COS evolved from salt marsh sediments, attributed to microbially-mediated degradation of soil organic matter (Devai and DeLaune, 1995). Field experiments in similarly anoxic wetlands showed that these soils tend to act as a net COS source to the atmosphere (De Mello and Hines, 1994; DeLaune et al., 2002; Fried et al., 1993; Yi et al., 2008). However, COS production was stimulated when rice paddy soils were incubated under aerobic conditions with organic sulfur (e.g. cysteine) (Minami and Fukushi, 1981).

In a recent field outing investigating COS exchange over a wheat field, aerobic agricultural soil was shown to be a quantifiable source of COS to the atmosphere (Billesbach et al., 2014). Using soils from this field site, here we investigate further the mechanism of COS production in aerobic soils.

4.2 Methods

4.2.1 Sample collection Soil samples were obtained from a wheat field in the Southern Great Plains Atmospheric Radiation Measurement site (36° 36' 18.0" N, 97° 29' 6.0" W), 3 m from where soil exchange measurements were previously performed (Billesbach et al., 2014). Two soil-only samples plus two samples containing wheat plants, roots and surrounding soils were collected from the top 15cm of the soil column and sealed in polyethylene bags. The samples were then mailed overnight for analysis at UC Berkeley.

4.2.2 Quantification of COS COS was quantified using an Agilent 7890A gas chromatograph oven outfitted with an Agilent 355 sulfur chemiluminescence detector (GC-SCD, Agilent Technologies, Santa Clara, CA, USA) with a custom inlet (described in Khan, et al. 2012). A five point calibration curve was generated by injecting different volumes of a whole air standard containing 543 ppt COS, calibrated to the NOAA-SIO provisional scale. An aliquot of the standard was injected between every 4 unknown samples to correct for instrument drift. The inlet and outlet of the incubation chamber was directly plumbed to the GC (Figure 4.1) with stainless steel tubing coated in amorphous silica (Sulfinert coating, Restek, Bellefonte, PA, USA).

4.2.3 Flux measurements Flux measurements were made with a dynamic, flow-through chamber method similar to the cuvette system used by Kuhn and Kesselmeier (2000). Soil and root matter subsamples of 100 g were enclosed in individual 1 L solid PFA incubation chambers (Savillex, Eden Prairie, MN, USA). Another 1 L PFA chamber containing 100 mL of distilled water was attached in series to the inlet of each chamber before analysis (Figure 4.1). Both chambers were then partially submerged in a constant temperature water bath maintaining a temperature between 15 and 34°C. For the temperature manipulation experiments, the chambers were covered in foil to prevent complications from ambient light reacting with soil samples. The second chamber was used both to mitigate moisture loss in soil samples and to act as a buffer volume allowing incoming air to equilibrate with the temperature of the water bath. Air was pumped downstream of the chambers with a micro-diaphragm pump (model UNMP830, KNF Neuberger Inc., Trenton, NJ).

For the light/dark experiment, the sample chamber lid was replaced by transparent PTFE film, affixed in place with a hose clamp. The edges of the film were further affixed to the chamber by tightly wrapping with clear adhesive tape. The flux measurement was performed either with the sample chamber covered in aluminum foil for the dark condition or illuminated with an incandescent light bulb (maximum intensity wavelength 600 nm). To correct for the additional warming caused during the light condition, self-contained thermocouple data loggers (iButtons, Maxim Inc., Sunnyvale, CA, USA) covered in PTFE were placed in the soil subsamples during initial trials. The constant temperature water bath was then adjusted so that both light and dark experiments proceeded with soils at 19 °C.

Fluxes were calculated using the formula F=V/m(Cf-C0) (Livingston and Hutchinson, 1995), where V is the moles of air flushing through the chamber per second, m is the mass of the soil sample, C0 is the initial concentration of COS entering the soil chamber in parts-per-trillion (ppt), Cf is the concentration of COS leaving the chamber in ppt, and F is the flux of COS reported in pmol min-1 per 100g of soil. At least two flux measurements were made for each sample and treatment. To assess method uncertainty, COS fluxes from an empty chamber were quantified along side chambers containing soil and root matter subsamples.

to pump to GC for “initial concentration”

ambient flow meter to GC for “final air concentration”

100 g soil distilled subsample water

Figure 4.1 Soil incubation setup. Two PFA chambers arranged in series and partially submerged in a constant temperature water bath were flushed with ambient air at 0.1 to 0.5 LPM. Air subsamples were analyzed from the outlet of the blank jar to establish the initial concentration and the outlet of the jar containing the soil sample to quantify the final concentration of COS.

4.2.4 Soil sample processing Six 100 g soil and root matter subsamples were placed in incubation chambers: 2 denoted “soil” collected from between wheat rows with little root matter, 1 “soil, fine roots” from soil adjacent to wheat plants, and 3 “wheat roots” containing a higher fraction of roots with some interspersed soil. See Table 4.1 for soil sample abbreviations. These soil samples were allowed to equilibrate with the chamber for 2 days before analysis. Field moisture was maintained during the course of experiment by daily weighing and the addition of distilled water approximately once per week. To investigate the influence of abiotic processes versus microbial processes, 3 subsamples were sterilized at 121 °C in a Steris Amsco Century autoclave (Steris Corporation, Mentor, OH, USA). The sterilized subsamples were then incubated as before. After flux experiments were completed, root matter was separated from soils using a 2mm sieve and weighed. Volumetric water content (VWC), water holding capacity (WHC), and bulk density were determined using standard gravimetric methods.

4.3 Results

4.3.1 Temperature response soil incubation experiments Live soil and root samples consistently demonstrated net uptake at 15 °C and 19 °C, transitioning to COS production at higher temperatures (Figure 4.2). Each soil showed small variability in repeated measurements with much larger variation between subsamples. The wheat root fluxes at 15 °C were the first performed and showed the highest variability, suggesting that the subsamples had not yet equilibrated after two days. Incubations of sterilized “dead” soils resulted in solely net COS production for the temperature range investigated (15 to 34 °C). Every soil exhibited a positive linear relationship with correlation coefficients greater than 0.6, regressing the COS mean flux for each sample on incubation temperature (Table 4.1). Sterilizing soil samples produced a reduction in slope for all three trials. For the WRA and SF samples, sterilization yielded a reduction in maximum COS flux at 34 °C; however, SBd (0.4 pmol min-1 per 100 g soil) had an increase in net COS production compared to SB (0.2 pmol min-1 per 100 g soil).

Sample name r2 slope wheat roots a (WRA) 0.90 0.24 wheat roots b (WRB) 0.68 0.13 wheat roots c (WRC) 0.87 0.088 wheat roots a dead (WRAd) 0.99 0.070 soil, fine roots (SF) 0.94 0.063 soil, fine roots dead (SFd) 0.99 0.045 soil a (SA) 0.68 0.25 soil b (SB) 0.90 0.050 soil b dead (SBd) 0.99 0.018 Table 4.1 The r2 values and slope of least squares linear regressions of mean soil COS flux and soil incubation temperature for each soil sample. Fluxes from sterilized “dead” samples were only measured at 15, 21, and 34 °C.

4.3.2 Light and dark soil incubation experiments Exposing soil samples to light induced net COS production and/or inhibited COS consumption in all cases except for the sterilized SBd. The largest production rate with light exposure was from the wheat root samples (WRB and WRC) at 0.7 ± 0.03 pmol min-1 per 100 g soil (Figure 4.3). The empty “blank” incubation chamber had a measureable flux of COS of 0.09 ± 0.006 pmol min-1 soil under light conditions

Comparing fluxes measured with a clear PTFE versus the opaque PFA lid used in the temperature manipulation experiments (Figure 4.3 to Figure 4.2), most dark flux measurements were within the bounds of uncertainty of the corresponding temperature-manipulation flux measurements at 19 °C. The exception is sample SF, which exhibited a dark uptake value closer to the COS flux estimate at 15°C.

−1 −1 COS (pmol 100 g soil min ) COS (pmol 100 g soil−1 min−1) 0.2 0.4 0.6 0.8 1.2 1.4 1.6 − − − 0 1 0 1 2 3 4 5 3 2 1 15 15 wheat roots DEAD SOILS r wheat roots LIVE SOILS

2 =0.86 r

incubation temperature (C) 2 =0.90

20 20

25 25 r

2 =0.68

30 30

35 35

0.2 0.4 0.6 0.8 1.2 1.4 1.6 − − − 0 1 0 1 2 3 4 5 3 2 1 15 15

soil, fine roots soil, fine roots

incubation temperature (C)

20 20

25 25

30 30

r 2 =0.94 35 35

0.2 0.4 0.6 0.8 1.2 1.4 1.6 − − − 0 1 0 1 2 3 4 5 3 2 1 15 15

soil soil

incubation temperature (C)

20 20

25 25

30 30 r 2 =0.67 r

2 =0.90 35 35

Figure 4.2 Net COS fluxes versus soil incubation temperature, with least square linear regressions for each separate soil sample. Individual points represent (n) flux measurements at a specific temperature; error bars represent the standard deviation (1 sigma, some error bars are smaller than the symbols). 44

Live Soil ) 1 − 0.5 min 1 − LIGHT 0 DARK

−0.5 COS (pmol 100 g soil

blank soilwheat root soil, fine roots

Dead Soil ) 1 − 0.5 min 1 −

−0.5 COS (pmol 100 g soil

blank soilwheat root soil, fine roots

Figure 4.3 COS fluxes from soil incubation experiments under light and dark conditions. Error bars indicate the standard deviation of repeated measurements. All measurements were performed with soil temperature at 19 °C.

4.4 Discussion

Here we have demonstrated that abiotic thermal and photo degradation of soil matter is sufficient to overcome biotic COS consumption in aerobic soil. This could explain two surprising patterns in COS fluxes that Billesbach et al. (2014) found over the SGP-ARM site, where the soil samples for this study were collected. Firstly, during the wheat growing season there was nighttime net COS uptake which could not be accounted for by a small nighttime stomatal conductance. Secondly, after the wheat harvest, Billesbach et al. (2014) observed large COS fluxes to the atmosphere with maximum net production during the day. These observations could be explained by variations in soil temperature and light.

4.4.1 Temperature and soil COS fluxes Prior work identified the importance of temperature in regulating COS fluxes in soils; however, these studies report soil COS uptake exclusively. (Kesselmeier et al., 1999; Steinbacher et al., 2004) modeled soil COS uptake in terms of temperature and soil water content, with the former study also relating COS uptake to ambient COS concentration. Van Diest and Kesselmeier (2008) and Kesselmeier et al. (1999) found that COS uptake had a distinct temperature maximum followed by declining uptake rates at higher temperatures, corresponding to presumed enzyme destruction. In contrast, soil COS fluxes in this study had a linear relationship to temperature, sustaining net COS uptake when incubated below 19 °C and net COS production above 25 °C.

To determine if the COS exchange was primarily biological or abiotic, we used two approaches: first, we autoclaved the soils to see if production would cease and second, we incubated the soils at increasing temperatures to see if results resembled abiotic or biological temperature response curves (Conrad, 1996). This study found that autoclaved soils still generate COS, indicating abiotic production, confirming results by Kato et al., (2008). Compared to live soils, autoclaved soils had a shallower slope when regressed with temperature, resulting in a small net production of COS at the minimum experiment temperature of 15 °C. The change in slope could be attributed to the changes in functional groups in soil organic matter brought on by autoclaving as shown by Berns et al., (2008). We further investigated incubation temperatures from 15 to 34 °C and did not find evidence of the temperature optimum found within this range by Van Diest and Kesselmeier (2008) and Kesselmeier et al. (1999) for other agricultural soils. Instead, we found emissions increasing with temperature, consistent with an abiotic production mechanism. This is consistent with some of the results of Kesselmeier et al. (1999) who used a chemical inhibitor to remove the influence of carbonic anhydrase in soils, but found only a 50% reduction in COS uptake. It is possible that adsorption on soil surfaces could explain some of the uptake and subsequent release under higher temperatures. The additional nighttime soil uptake during the Billesbach et al. (2014) study may be accounted for by this abiotic interaction. 46

4.4.2 Light and soil COS fluxes Earlier terrestrial field studies which considered the influence of light on COS fluxes used the relationship between PAR and photosynthesis to infer a relationship between the COS uptake by plants and available light (Geng and Mu, 2006; J. Kesselmeier and Merk, 1993; Kuhn et al., 1999). To our knowledge, no study has considered the influence of PAR on soil fluxes outside of the indirect effect on temperature.

In this study we observed photo-production of COS in live soils under light conditions compared to incubations without light at the same temperature. This may explain the high daytime COS production after the wheat was harvested at the SGP-ARM site (Billesbach et al., 2014). Harvesting wheat served to remove the proportionally large COS sink and caused remaining soil and litter to be subjected to direct sunlight. These circumstances could lead to the variable COS production rates and roughly diel behavior.

Observations of COS photo-production in ocean and rain water have been linked to the intensity of UV light (Mu et al., 2004; Weiss et al., 1995; Zepp and Andreae, 1994). Rates of COS production were further correlated with concentrations of chromophoric dissolved organic matter (Uher and Andreae, 1997), organosulfur compounds (Zepp and Andreae, 1994), and humic acid with cysteine (Flöck et al., 1997). Considering the soil organic matter content in the wheat field soils in this study, COS photo-production could be related to a comparable process suggested by Flöck et al., (1997) for ocean water, where humus photosensitizes COS precursors, e.g. cysteine. In the study by Minami and Fukushi (1981), adding cysteine and a number of other organic sulfur compounds to soil samples resulted in aerobic COS production; light controls during their experiment were unspecified.

4.5 Conclusion

Although the soils studied here exhibited similar overall patterns in COS exchange and temperature, there remains unexplained variability in the magnitude of fluxes between soil samples. This may be accounted for by the overall quality and form of soil organic matter. In a further experiment, subsets of litter (roots, leaves) and soil could be incubated separately to assess the COS contribution of various types of litter, previously found to exhibit COS uptake (Kesselmeier and Hubert, 2002). Additionally, soil samples from SGP-ARM could be sieved and dried, then re-wetted to determine if this sample preparation contributes to the disparity between our results and Van Diest and Kesselmeier, (2008) and Kesselmeier et al., (1999).

The results here offer some hope for creating a model that would explain COS soil fluxes in terms of temperature and PAR, easily measured environmental variables. In future eddy flux covariance measurements of COS, as demonstrated in Billesbach et al. (2014) and Asaf et al., (2013), we hope to better understand variables controlling COS fluxes accurately in order to separate COS soil and plant fluxes.

Works Cited

Asaf, D., Rotenberg, E., Tatarinov, F., Dicken, U., Montzka, S.A., Yakir, D., 2013. Ecosystem photosynthesis inferred from measurements of carbonyl sulphide flux. Nature Geosciences 6, 186–190. Berns, A.E., Philipp, H., Narres, H.D., Burauel, P., Vereecken, H., Tappe, W., 2008. Effect of gamma-sterilization and autoclaving on soil organic matter structure as studied by solid state NMR, UV and fluorescence spectroscopy. European Journal of Soil Science 59, 540–550. Berry, J., Wolf, A., Campbell, J. E., Baker, I., Blake, N., Blake, D., Denning, A.S., Kawa, S.R., Montzka, S.A., Seibt, U., Stimler, K., Yakir, D., Zhu, Z., 2013. A coupled model of the global cycles of carbonyl sulfide and CO2: A possible new window on the carbon cycle. Journal of Geophysical Research: Biogeosciences 118, 842–852. Billesbach, D.P., Berry, J.A., Seibt, U., Maseyk, K., Torn, M.S., Fischer, M.L., Abu-Naser, M., Campbell, J.E., 2014. Growing season eddy covariance measurements of carbonyl sulfide and CO2 fluxes: COS and CO2 relationships in Southern Great Plains winter wheat. Agricultural and Forest Meteorology 184, 48–55. Blonquist, J.M., Montzka, S.A., Munger, J.W., Yakir, D., Desai, A.R., Dragoni, D., Griffis, T.J., Monson, R.K., Scott, R.L., Bowling, D.R., 2011. The potential of carbonyl sulfide as a proxy for gross primary production at flux tower sites. Journal of Geophysical Research: Biogeosciences 116, G04019. Brühl, C., Lelieveld, J., Crutzen, P.J., Tost, H., 2012. The role of carbonyl sulphide as a source of stratospheric sulphate aerosol and its impact on climate. Atmospheric Chemistry and Physics 12, 1239–1253. Campbell, J. E, Carmichael, G.R., Chai, T., Mena-Carrasco, M., Tang, Y., Blake, D.R., Blake, N.J., Vay, S.A., Collatz, G.J., Baker, I., Berry, J.A., Montzka, S.A, Sweeney, C., Schnoor, J.L., Stanier, C.O., 2008. Photosynthetic control of atmospheric carbonyl sulfide during the growing season. Science 322, 1085–1088. Castro, M.S., Galloway, J.N., 1991. A comparison of sulfur-free and ambient air enclosure techniques for measuring the exchange of reduced sulfur gases between soils and the atmosphere. Journal of Geophysical Research 96, 15427–15. Chin, M., Davis, D.D., 1993. Global sources and sinks of OCS and CS2 and their distributions. Global Biogeochemical Cycles 7, 321–337. Conrad, R., 1996. Soil microorganisms as controllers of atmospheric trace gases (H2, CO, CH4, OCS, N2O, and NO). Microbiological Reviews 60, 609–640. Conrad, R., Meuser, K., 2000. Soils contain more than one activity consuming carbonyl sulfide. Atmospheric Environment 34, 3635–3639. De Mello, W.Z., Hines, M.E., 1994. Application of static and dynamic enclosures for determining dimethyl sulfide and carbonyl sulfide exchange in Sphagnum peatlands: Implications for the magnitude and direction of flux. Journal of Geophysical Research 99, 14–601.

DeLaune, R.D., Devai, I., Lindau, C.W., 2002. Flux of reduced sulfur gases along a salinity gradient in Louisiana coastal marshes. Estuarine, Coastal and Shelf Science 54, 1003–1011. Devai, I., DeLaune, R.D., 1995. Formation of volatile sulfur compounds in salt marsh sediment as influenced by soil redox condition. Organic Geochemistry 23, 283–287. Fried, A., Klinger, L.F., Erickson, D.J., 1993. Atmospheric carbonyl sulfide exchange in bog microcosms. Geophysical Research Letters 20, 129–132. Geng, C., Mu, Y., 2004. Carbonyl sulfide and dimethyl sulfide exchange between lawn and the atmosphere. Journal of Geophysical Research: Atmospheres 109. Geng, C., Mu, Y., 2006. Carbonyl sulfide and dimethyl sulfide exchange between trees and the atmosphere. Atmospheric Environment 40, 1373–1383. Flöck, O.R., Andreae, M.O., Dräger, M., 1997. Environmentally Relevant Precursors of Carbonyl Sulfide in Aquatic Systems. Marine Chemistry 59, 71–85. Kato, H., Saito, M., Nagahata, Y., Katayama, Y., 2008. Degradation of ambient carbonyl sulfide by Mycobacterium spp. in soil. Microbiology 154, 249–255. Kesselmeier, J., Merk, L., 1993. Exchange of carbonyl sulfide (COS) between agricultural plants and the atmosphere: Studies on the deposition of COS to peas, corn and rapeseed. Biogeochemistry 23, 47–59. Kesselmeier, J., Teusch, N., Kuhn, U., 1999. Controlling variables for the uptake of atmospheric carbonyl sulfide by soil. Journal of Geophysical Research 104, 11577–11584. Kesselmeier, J., Hubert, A., 2002. Exchange of reduced volatile sulfur compounds between leaf litter and the atmosphere. Atmospheric Environment 36, 4679– 4686. Kettle, A.J., Kuhn, U., Von Hobe, M., Kesselmeier, J., Andreae, M.O., 2002. Global budget of atmospheric carbonyl sulfide: Temporal and spatial variations of the dominant sources and sinks. Journal of Geophysical Research: Atmospheres 107. Kettle, A.J., Rhee, T.S., Von Hobe, M., Poulton, A., Aiken, J., Andreae, M.O., 2001. Assessing the flux of different volatile sulfur gases from the ocean to the atmosphere. Journal of Geophysical Research: Atmospheres 106, 12193– 12209. Kuhn, U., Ammann, C., Wolf, A., Meixner, F.X., Andreae, M.O., Kesselmeier, J., 1999. Carbonyl sulfide exchange on an ecosystem scale: soil represents a dominant sink for atmospheric COS. Atmospheric Environment 33, 995–1008. Kuhn, U., Kesselmeier, J., 2000. Environmental variables controlling the uptake of carbonyl sulfide by lichens. Journal of Geophysical Research: Atmospheres 105, 26783–26792. Lehmann, S., Conrad, R., 1996. Characteristics of turnover of carbonyl sulfide in four different soils. Journal of Atmospheric Chemistry 23, 193–207. Livingston, G.P., Hutchinson, G.L., 1995. Enclosure-based measurement of trace gas exchange: applications and sources of error, in: Biogenic Trace Gases: Measuring Emissions from Soil and Water. pp. 14–51.

Minami, K., Fukushi, S., 1981. Volatilization of carbonyl sulfide from paddy soils treated with sulfur-containing substances. Soil Science and Plant Nutrition 27, 339–345. Montzka, S.A., Aydin, M., Battle, M., Butler, J.H., Saltzman, E.S., Hall, B.D., Clarke, A.D., Mondeel, D., Elkins, J.W., 2004. A 350-year atmospheric history for carbonyl sulfide inferred from Antarctic firn air and air trapped in ice. Journal of Geophysical Research: Atmospheres 109. Montzka, S.A., Calvert, P., Hall, B.D., Elkins, J.W., Conway, T.J., Tans, P.P., Sweeney, C., 2007. On the global distribution, seasonality, and budget of atmospheric carbonyl sulfide (COS) and some similarities to CO2. Journal of Geophysical Research 112. Mu, Y., Geng, C., Wang, M., Wu, H., Zhang, X., and Jiang, G., 2004. Photochemical Production of Carbonyl Sulfide in Precipitation. Journal of Geophysical Research: Atmospheres 109. Protoschill-Krebs, G., Kesselmeier, J., 1992. Enzymatic pathways for the consumption of carbonyl sulphide (COS) by higher plants. Botanica Acta 105, 206–212. Protoschill-Krebs, G., Wilhelm, C., Kesselmeier, J., 1996. Consumption of carbonyl sulphide (COS) by higher plant carbonic anhydrase (CA). Atmospheric Environment 30, 3151–3156. Sandoval-Soto, L., Stanimirov, M., Von Hobe, M., Schmitt, V., Valdes, J., Wild, A., Kesselmeier, J., 2005. Global uptake of carbonyl sulfide (COS) by terrestrial vegetation: Estimates corrected by deposition velocities normalized to the uptake of carbon dioxide (CO2). Biogeosciences 2, 125–132. Seibt, U., Kesselmeier, J., Sandoval-Soto, L., Kuhn, U., Berry, J.A., 2010. A kinetic analysis of leaf uptake of COS and its relation to transpiration, photosynthesis and carbon isotope fractionation. Biogeosciences 7, 333–341. Steinbacher, M., Bingemer, H.G., Schmidt, U., 2004. Measurements of the exchange of carbonyl sulfide (OCS) and carbon disulfide (CS2) between soil and atmosphere in a spruce forest in central Germany. Atmospheric Environment 38, 6043–6052. Suntharalingam, P., Kettle, A.J., Montzka, S.M., Jacob, D.J., 2008. Global 3-D model analysis of the seasonal cycle of atmospheric carbonyl sulfide: Implications for terrestrial vegetation uptake. Geophysical Research Letters 35. Uher, G., and Andreae, M.O., 1997. Photochemical Production of Carbonyl Sulfide in North Sea Water: A Process Study. Limnology and Oceanography 42, 432– 442. Van Diest, H., Kesselmeier, J., 2008. Soil atmosphere exchange of carbonyl sulfide (COS) regulated by diffusivity depending on water-filled pore space. Biogeosciences 5, 475–483. Watts, S.F., 2000. The mass budgets of carbonyl sulfide, dimethyl sulfide, carbon disulfide and hydrogen sulfide. Atmospheric Environment 34, 761–779.

Weiss, P.S., Johnson, J.E., Gammon, R.H.,Bates, T.S., 1995. Reevaluation of the Open Ocean Source of Carbonyl Sulfide to the Atmosphere. Journal of Geophysical Research: Atmospheres 100, 23083–23092. Wohlfahrt, G., Brilli, F., Hörtnagl, L., Xu, X., Bingemer, H., Hansel, A., Loreto, F., 2012. Carbonyl sulfide (COS) as a tracer for canopy photosynthesis, transpiration and stomatal conductance: potential and limitations. Plant, Cell & Environment 35, 657–667. Yi, Z., Wang, X., Sheng, G., Fu, J., 2008. Exchange of carbonyl sulfide (OCS) and dimethyl sulfide (DMS) between rice paddy fields and the atmosphere in subtropical China. Agriculture, Ecosystems & Environment 123, 116–124. Zepp, R.G., Andreae, M.O., 1994. Factors Affecting the Photochemical Production of Carbonyl Sulfide in Seawater. Geophysical Research Letters 21, 2813–2816.

5. Exchange of carbonyl sulfide (COS) between a grassland and the atmosphere

Simultaneous measurement of CO2 and carbonyl sulfide (COS) fluxes have recently been used to approximate gross primary production (GPP) from net ecosystem carbon exchange measurements. Plants act simultaneously as a CO2 source and sink, through autotrophic respiration and photosynthesis. Their relationship to COS is much simpler, as COS is irreversibly destroyed by enzymes involved in photosynthetic pathways. If both the gross COS leaf uptake and the CO2:COS leaf uptake ratio is known, the gross uptake of CO2 can be calculated. When this approach is applied at larger spatial scales, COS exchange from soils can confound GPP estimates. Soil and litter exchange may be quantified separately in grassland ecosystems using dry season COS measurements, when green plants are largely absent. Here we report in situ fluxes of sulfur gases and CO2 from a grassland outside of Santa Cruz, CA, U.S.A (37.0°N, 122°W). Monthly measurements were made using static flux chambers from March 2012 to March 2013. During the non- growing dry season, we found small but significant COS exchange rates that were correlated with soil moisture and temperature. Growing season COS flux measurements were then corrected for soil COS uptake and used to calculate GPP. Estimates of GPP without the soil correction resulted in a 4 to 13% overestimation of carbon uptake. When soil moisture was artificially increased, we observed an increase in GPP estimated with the COS proxy. This approach should be applied to terrestrial ecosystems where COS exchanges from plants and soils can be determined independently, especially in forest and savannah ecosystems with large observed soil COS uptake. Furthermore, a large net emission of dimethyl sulfide (DMS), an important precursor to COS, was observed during the growing season only.

5.1 Introduction

As anthropogenic CO2 emissions continue increasing, it is necessary to characterize the partitioning of carbon exchange between atmospheric and terrestrial ecosystem reservoirs to predict future CO2 concentrations in the atmosphere (Wofsy, 2001). Large uncertainties remain in estimates of the amount of carbon removed from the atmosphere by photosynthesis (Beer et al., 2010), called gross primary productivity (GPP). This quantity is essential for describing carbon-climate feedbacks and assessing ecosystem-based CO2 capture and storage projects. GPP is also an important input parameter in models of ecosystem carbon exchange; errors in GPP propagate into estimates of other variables (Campbell et al., 2008; Schaefer et al., 2012). Global GPP appraisals rely on data-oriented or process-oriented models, the former lacking in predictive ability and the latter requiring accurate atmospheric and soil parameter inputs (Koffi et al., 2012).

Measuring fluxes of carbonyl sulfide (formula: COS) can provide an additional constraint on estimates of GPP. With a globally averaged tropospheric concentration of 500 ± 100 parts-per-trillion (ppt) (Montzka et al., 2007), COS is the most abundant sulfur containing gas in Earth’s atmosphere. Both COS and CO2 enter a plant through leaf stomata. Whereas some CO2 is released again in back diffusion or in respiration, COS is irreversibly destroyed by the enzyme carbonic anhydrase (Protoschill-Krebs et al., 1996). The regeneration of COS is energetically unfavorable in this reaction (Schenk et al., 2004).

In order to characterize atmospheric sulfur exchange, GPP or NPP estimates were at first used to calculate the COS vegetation sink (e.g. Sandoval-Soto et al., 2005; Montzka et al., 2007). Several groups have already suggested using COS vegetative uptake to constrain GPP instead (Campbell et al., 2008; Montzka et al., 2007; Seibt et al., 2010; Stimler et al., 2011; Suntharalingam et al., 2008). However, to date only two published studies have actually attempted to do so (Blonquist et al., 2011, Asaf et al., 2013).

Many of the assumptions involved in using COS fluxes as a GPP proxy have been empirically investigated. Stimler et al. (2010) confirmed the assumptions about plant physiology and COS/CO2 exchange that need to be met to use COS as a tracer for GPP: that COS co-diffuses with CO2 via the same pathway in plant leaves , that COS and CO2 do not inhibit one another at reaction sites with carbonic anhydrase, and that emission of COS by leaves is negligible. Using COS to predict GPP on the leaf-level was comparable to other methods like C18OO exchange (Seibt et al., 2010; Stimler et al., 2011).

A problem arises when the COS/CO2 model is applied to an ecosystem beyond the leaf scale. Empirical measurements of the ratio of COS to CO2 uptake deviate from the average value of 3 (Sandoval-Soto et al., 2005) when processes other than photosynthesis dominate trace gas exchange (Seibt et al., 2010). In these cases, it is 54 assumed that a missing source or sink of COS is present in the system. In most ecosystems investigated, the obvious culprit is COS exchange by soils.

Quantifying the terrestrial-atmospheric exchanges of sulfur has been challenging as only some of the controlling factors are known. In general, oxic soils are observed to act as a net sink of reduced sulfur compounds, while anoxic soils appear to act as a net source (Watts, 2000). Changes in soil moisture are expected to influence the direction of net gas fluxes of reduced sulfur compounds, like COS and its atmospheric precursor dimethyl sulfide (DMS). DMS, known to be produced by phytoplankton and bacteria, is oxidized in the atmosphere by the hydroxyl radical, forming a considerable portion of the total COS budget (Barnes et al., 1994).

This study observes the changes in net COS and DMS exchange in a grassland during the wet and dry season and the effect of simulated rainfall on these fluxes. Since there are few green plants present during the dry season, the soil and litter layer sulfur gas exchange can be quantified separately from the stomatal uptake by plants. To take into account the difference in soil moisture between seasons, water was added to plots to assess the effect of soil moisture with and without gross primary production underway. Here, fluxes of COS were corrected for non-stomatal component fluxes, then used to calculate gross primary production in a grassland ecosystem.

5.2 Site Description

Coastal California experiences a Mediterranean climate with warm, dry summers followed by cool, wet winters (Figure 5.1); grasses have distinct growing and dormancy seasons. Soil profiles on uncultivated marine terraces near Santa Cruz, CA, USA (37.0 °N, 122°W) have been used as a “natural laboratory” for the study of weathering processes and marine terrace formation through well-instrumented research sites. Using the sites characterized in White, et al. (2008), here we chose plots located on SCT2 and SCT3 as annual grassland sites (Schulz et al., 2011). The site SCT5, perennial grassland site, was also used in this study.

20 200

15 100 Mean Total Precipitation (mm) Mean Monthly Temperature (C)

10 0 0 2 4 6 8 10 12 Month Figure 5.1 Mean monthly temperature and mean monthly precipitation totals for Santa Cruz, CA from 1981 to 2010. Data from NOAA National Climatic Data Center (http://www.ncdc.noaa.gov).

5.3 Methods

5.3.1 Static flux chamber deployment Static flux chambers (Livingston and Hutchinson, 1995) were deployed in 52 instances to quantify grassland-atmospheric fluxes of sulfur-containing gases. First, a 0.4 m × 0.4 m × 0.23 m PTFE-lined aluminum frame was installed at least 3 cm into the soil profile, taking care to minimize severing plant roots. After an hour to allow the plot to recover from the disturbance, a chamber lid was clamped to the chamber base to mark the beginning of the flux measurement, when t=0 (Figure 5.2). The lid was a Lexan plastic rectangular box 0.4 m × 0.4 m × 0.43 m, lined internally with a 0.254 mm thick PTFE film. To avoid complications from photo-oxidation and rising temperatures within the chamber, the chamber lid was covered with reflective insulation. Weather stripping was installed along the box lip between the Lexan and PTFE liner to aid in sealing the chamber lid to the base. The chamber lid and base enclosed a total of 109 L, and the headspace was mixed by a PTFE-coated aluminum fan. Before deployment in the field, the chamber equipment was tested for inertness under both ambient and elevated concentrations of COS and DMS in the chamber headspace; measured concentrations had an RSD of 3.5% over a 32 minute incubation period, translating to a minimum detectable flux of 0.6 pmol m-2 sec-1.

For each flux measurement, at least 3 whole air samples were collected at regular intervals through a silica-coated steel 2 μm filter into previously-evacuated 1 L canisters lined with amorphous silicon (Restek, Bellefonte, PA, USA) that have been shown to be appropriate for collection and storage of ambient-level sulfur- containing compounds (Khan et al., 2012). During collection, a short vent line was opened to the atmosphere to avoid under-pressurizing the chamber headspace. Along with canister sampling, 30 mL aliquot subsamples of the chamber headspace were collected in glass vials (Wheaton, Millville, NJ, USA) for analysis of CO2.

sample vent collection injection port motor

PTFE !lm fan

ambient COS COS, DMS usu. 500ppt production

chamber lid clamp

chamber base pla nts soil

Figure 5.2 A schematic of the PTFE-lined static flux chamber base and lid installed at a grassland site. Arrows show gross production and consumption of sulfur-containing trace gases.

Flux chamber measurements were performed between noon and 4:30 PM local time for each outing. The prevailing wind was from the ocean in the south, traveling over approximately 1.5 km of sparsely populated coastal farm and a two lane state highway. Temperatures were recorded outside the chambers and in approximately the middle of the chambers using stainless steel thermocouple dataloggers (iButtons, Maxim Inc., Sunnyvale, CA, USA) wrapped in Teflon tape and suspended from the Teflon sampling line. Barometric pressure was measured throughout the experiment. Soil volumetric water content (%VWC) at 5 cm depth was measured outside and within the experimental plot after air sampling for each outing (ThetaProbe, organic setting, Delta-T Devices, Cambridge, UK). To quantify soil bulk density, two soil cores (0 – 5 cm) were collected from each plot for analysis in the lab. The volume of the chamber headspace was determined by estimating plant volume and measuring soil volume from 16 depth measurements in the trough left by the chamber base after removal, then subtracting these volumes from the known volume of the chamber lid and base.

5.3.2 Rainfall manipulation sampling design Flux measurements were performed with static flux chambers every month except December from 21 March 2012 to 10 March 2013 (Table 5.1). The number of flux measurements that could be made in a single outing was constrained by the number of canisters appropriate for sulfur gas sampling (see Khan et al. 2012 for a discussion of sampling canisters). To optimize the amount of useful information generated from a limited number of flux measurements, three sampling strategies were used to characterize (1) natural variation in situ of sulfur and carbon gas fluxes, (2) the change in fluxes over time from a single plot when rainfall was added and (3) the spatial variation of fluxes when rainfall was added by comparing 2 nearby plots in either an annual or perennial grassland site.

On May 30, November 27, and January 24th, a single chamber base was installed, no water was added and 3 flux measurements were made on each date to see the variation in fluxes from individual plots, without further manipulations. There was a precipitation event during the January 24 outing, and the chamber lid excluded rainfall from the plot, causing the plot soil to have an average VWC 11% less than adjacent soil that was not shielded from the rain.

A single chamber base was installed for the outings on July 26, August 24, September 20, October 28, 2012 and March 4, 2013 to investigate temporal variation from individual plots before and after artificial rainfall addition. The outings consisted of a first flux measurement at field soil moisture. Using a piston- pressurized hand water sprayer, 0.7 L of water was added to the plot, equivalent to 4 mm rain from a typical wet season storm. Then 2 or 3 subsequent flux measurements were performed with an elevated soil moisture condition.

A third design was employed for 3 outings in the dry season, (June 26, June 28, and July 3, 2012), and 2 outings in the wet season, (February 27 and March 10, 2013) to assess spatial variation. The July 3 and March 10 outings took place in a perennial grassland, the rest from annual grassland sites. 2 separate chamber bases were installed less than 5 m apart at plots distinct from any past chamber installations. A single flux measurement was performed at field moisture from the first plot. After the addition of 0.7 L of artificial rain water to both plots, 2 wetted-condition flux measurements were performed for the first plot and 1 wetted-condition flux measurement for the second plot.

Date Sites Total # Flux Notes Average Field Additional soil measuremen Soil Moisture moisture from ts (VWC %) added rainfall 2012 March 21 2 3 Method evaluation 26.1 ± 2.8 0 April 26 1 2 Method evaluation 18.2 ± 3.3 0 May 30 1 3 No water added 6.4 ± 3.5 0 June 26 2 4 Water added 4.4 ± 1.4 7.9 ± 4.8 June 28 2 4 Water added 2.4 ± 1.8 5.7 ± 3.9 July 3 2 4 Water added, 3.9 ± 1.9 7.0 ± 4.1 perennial grass site July 26 1 4 Water added 2.2 ± 0.28 6.9 ± 1.3 August 24 1 4 Water added 3.2 ± 0.1 6.5 ± 6.5 September 1 4 Water added 4.8 ± 1.0 6.6 ± 5.4 20 October 28 1 4 Water added 7.1 ± 2.3 2.8 ± 4.9 November 1 3 Rained recently; no 20.9 ± 3.1 0 27 water added 2013 January 24 1 3 Experiment 34.8 ± 8.4* -11.9 ± 8.8* performed during rain storm February 27 2 4 Water added 13.7 ± 4.9 2.5 ± 5.7 March 4 1 3 Water added 13.5 ± 1.9 6.1 ± 4.4 March 10 2 4 Water added, 21.9 ± 4.9 5.7 ± 4.8 perennial grass site

Table 5.1 Summary of field outings. All outings took place in an annual grassland unless otherwise indicated. *The January 24 outing took place during a rainstorm. The field soil moisture after several hours of rain was on average 11.9 %VWC higher than the experimental plot soil moisture due to rainfall exclusion by the chamber lid.

5.3.3 Quantification of gas samples To quantify COS and DMS, 50 mL subsamples of air from the canisters were injected onto a GC column after cryo-focusing on a narrow bore silica-coated tube trap immersed in liquid nitrogen. Air samples were analyzed on an Agilent 7890A gas chromatograph attached to an Agilent 355 sulfur chemiluminescence detector (GC/SCD). The SCD is ideal for analysis of reduced sulfur compounds because of its high sensitivity and linear, equimolar response. All wetted surfaces were either PFA/PTFE or Siltek-treated stainless steel to minimize analyte loss. A five point calibration curve was generated before each outing using a whole air standard (COS concentration = 543 parts-per-trillion) calibrated to the provisional scale of NOAA- SIO for COS. A calibration curve for DMS was created by diluting a 1 ppm gas standard (Matheson Tri-Gas, Newark, CA, USA) on a custom built dilution line and comparing it to the whole air calibration curve to confirm the equimolar detector response. The analytical detection limit was 120 parts-per-trillion. More details on this method can be found in Khan et al. (2012).

CO2 in the Wheaton vial air samples was analyzed on a Shimadzu GC-14A with a thermal conductivity detector (TCD) (Shimadzu Scientific Inc., Columbia, MD, USA). A 997 ppm CO2 standard (Scott Specialty Gases) was analyzed between every 10 unknown samples to determine concentration and correct for instrument drift.

5.3.4 Calculation of fluxes The concentration of COS, DMS, and CO2 in the chamber headspace over time was used to calculate the overall flux from experimental field plots. A linear and an exponential model were both applied to data from each chamber, and the approach with better goodness of fit was assigned as the flux. Only experiments yielding fits with r2 > 0.9 were included in this study. The exponential flux model was adapted from DeMello and Hines (1994): the concentration at time t, C(t) is a function of the concentration in the headspace at time 0 (C0) , the concentration when soil pore space and chamber headspace are in equilibrium (Cmax) , and a first order uptake rate constant k: C(t) = Cmax – (Cmax – C0)exp(-kt). Cmax and k are solved iteratively, and the rate of concentration change is calculated for time 0, (dC/dt)t=0 = k(Cmax – C0).

-1 The trace gas flux F is then calculated by F = [nchamber(dC/dt)t=0]N , where nchamber is the number of moles in the chamber during sampling and N is the surface area of the 2 -2 -1 chamber plot in m . CO2 fluxes are reported in μmol m sec . COS and DMS fluxes are reported in pmol m-2 sec-1.

5.4 Results

5.4.1 Annual variability of sulfur gas fluxes With few exception, fluxes of DMS and COS from grasslands at field moisture were of similar magnitude and opposite sign: DMS fluxes ranged from +39 pmol m-2 sec-1 to slight uptake of -2 pmol m-2 sec-1 and COS fluxes exhibited a maximum uptake of -75 pmol m-2 sec-1 to emissions at +7 pmol m-2 sec-1.

In the wet growing season, COS was taken up (-26 ± 27 pmol m-2 sec-1) and DMS was emitted to the atmosphere (5.9 ± 8 pmol m-2 sec-1). During the dry season, uptake of COS was smaller and less variable (-6.1 ± 10 pmol m-2 sec-1 ) and production of DMS was reduced (4.2 ± 4.5 pmol m-2 sec-1, excluding the high late season emission point), with a few instances of DMS uptake and COS production (Figure 5.3). Of the 24 flux measurements made at field soil moisture, 18 chamber experiments yielded r2 > 0.9 when the flux calculation model (Section 3.5) was applied to concentrations of DMS and COS over time. Measurements of DMS uptake were limited by the ambient concentration of DMS, which was often below the detection limit of 120 ppt. H2S, CS2, and dimethyl disulfide (DMDS) were also measured, but were present in such low concentrations that no flux measurements could be reported. The total number of measurements in the study was limited by the number of available flasks for sampling.

40 ) 1 − sec 2 −

0 DMS fluxes (pmol m

0 ) 1 − −20 sec 2 −

−40 COS fluxes (pmol m

−60 Dry Wet Season Season

−80 0 60 120 180 240 300 360 Day of Year

Figure 5.3 Natural fluxes of COS and DMS from monthly grassland plots. Field soil moisture (Table 5.1) in the dry season varied from 2.2 to 4.8 %VWC. These values may be higher than the actual soil moisture: measurements that fell below the probe detection limit were not included in the plot average. Soil moisture in the wet season varied between 13.4 and 25.6 %VWC. Air temperature averaged over the course of each flux measurement ranged from 11.7 to 30.0°C.

5.4.2 Rainfall manipulation experiments Adding water to chamber plots in the dry season, when plant matter was dormant or senescing, resulted in a reduction of COS uptake and/or an increase of COS production by the soil system. In contrast, during wet season manipulation experiments with green plants, COS uptake increased significantly (Figure 5.4).

DRY SEASON WET SEASON

40 40 ) ) 1 1 − −

sec 0 sec 0 2 2 − −

−40 −40

−80 −80 COS (pmol m COS (pmol m

−120 −120

80 80

) Jun 26 Oct 28 ) 1 1 − 60 Jun 28 − 60 Feb 27 Jul 03 Mar 4 sec sec 2 2 −

Aug 24 − Mar 10 40 40 Sep 20 water added water added mol m mol m µ 20 µ 20 CO2 CO2 ( 0 0

80 80 ) ) 1 1 − 60 − 60 sec sec 2 2 − 40 − 40

20 20 DMS (pmol m DMS (pmol m 0 0

0 40 80 120 0 40 80 120 time since water addition (min) time since water addition (min)

Figure 5.4 COS, CO2, and DMS fluxes from in situ grassland plots before and after water additions. The gray symbols are fluxes from a perennial grassland site; black symbols represent fluxes from a annual grassland sites.

In most cases, artificial rain on plots in both seasons resulted in an increase of net DMS and CO2 release to the atmosphere, usually increasing within 2 minutes of adding water and decreasing within 2 hours. A notable exception is the September outing (inverted triangles, Figure 5.4): not only did DMS net emissions start out high and then decrease, but CO2 emissions exhibited an increase over the next 2 hours. In the March 4th outing (upright triangles, Figure 5.4), a site 2 m from the February 27th outing (black squares), CO2 fluxes started high relative to previous measurements at 24.7 μmol m-2 sec-1, then emission decreased after water addition.

5.5 Discussion

5.5.1 Growing season COS fluxes as a proxy for GPP Most of the ecosystem exchange of carbon in a Mediterranean grassland happens during the wet, growing season (Xu and Baldocchi, 2004). Not surprisingly, in this study the grassland took up COS during the wet season when plants are green and photosynthetic rates are high (Figure 5.4). Measurements of COS exchange could be used to partition GPP during the growing season if (1) the ratio of COS to CO2 uptake for the plants of interest was known and (2) dark measurements of COS exchange can reliably be extrapolated to COS exchange rate in light conditions.

As an exercise to investigate the potential of this method, we perform the simple -1 calculation FCOS = GPP[COS][CO2] vCOS/CO2. FCOS is the one-way flux of COS into plant leaves, [COS] and [CO2] are ambient concentrations, and the factor vCOS/CO2 is the experimentally determined ratio of deposition velocities for COS and CO2. For C3 plants, like the dominant grasses at our annual grassland sites, Stimler et al. (2010) found a consistent vCOS/CO2 of 1.82 ± 0.18. Ambient concentrations of COS and CO2 were determined with flask samples. Our measurements of FCOS included soil and plants as opposed to only live leaves.

To perform this calculation, we would need to make a few further assumptions. First, the factor vCOS/CO2 in our grassland system is represented by the ratio determined by Stimler et al. (2010) with tobacco, sage, and hibiscus leaves. Second, FCOS into plant leaves can be estimated by subtracting FCOS,dry measurements from plots without green plants from FCOS,wet measurements from plots with green plants. Third, measurements of FCOS using dark chambers can be extrapolated to t=0 (see Section 3.5), and are representative of FCOS in light conditions. This last assumption is built into the model from DeMello and Hines (1994), and is further bolstered by the evidence of a long delay (~12 minutes) in stomatal closing for graminoids in dry climates (Vico, et al. 2011).

To estimate FCOS,dry,we make use of the averaged COS fluxes from the April 26 outing: the coolest and wettest of the dry season measurements at ≈ 20.2 °C and field VWC at ≈ 18 %, COS fluxes were -3.8 ± 4.4 pmol COS m-2 s-1. These dry season conditions (with few green plants) most closely match the soil conditions for growing season chambers in early 2013 (Table 5.2). Lab-based measurements of incubated soil show similarly small rates of uptake for wet season conditions (Whelan, unpublished data). To correct for the non-leaf COS exchange, here we simply subtract the April 26th COS fluxes from wet season COS fluxes to yield FCOS,adjusted. Calculating GPP without correcting for soil uptake overestimates the carbon uptake flux by 4 to 13%.

Wet season plot with green plants Calculated Day of FCOS Rd VWC GPPadj -2 -1 -2 -1 -2 -1 measurement pmol m s μmol m s % μmol m s Jan 24 -75 +4.5 23 -35 -51 +6.6 23 -23 Feb 27 -37 +11 17 -13 Mar 4 -32 +25 13 -11 -50 +9.6 20 -18 -70 +3.6 20 -26 Mar 10 -109 +16 29 -42 -99 +8.1 26 -38 Table 5.2 Calculated GPP from measurements of FCO2, FCOS, ambient measurements of COS and CO2, and using Stimler et al. (2010) estimate of vCOS/CO2 for C3 plants.

By comparing the calculated GPP to the measured fluxes of net CO2 exchange, we find that adding water to chambers stimulated both assimilation of carbon via photosynthesis, as estimated with COS-based GPP calculation, and loss of carbon via respiration from plants and the soil microbial community (Figure 5.4). The change in CO2 flux after water addition indicates respiration increases after soil moisture was increased (Orchard and Cook, 1983); however, we can infer the increase of carbon uptake outside of respiration processes by examining the simultaneous COS increase in uptake (Figure 5.4). However, looking at all the dry season fluxes of COS, we see that COS exchange is not always negligible, ranging from -26 to +7.3 pmol COS m-2 s-1 for plots at field moisture (Figure 5.3).

To confirm that these calculations are reasonable, a further experiment must be performed to estimate GPP using a more standard method. The net ecosystem exchange (NEE) of carbon would be measured using light chambers followed by measurements of ecosystem respiration (Rd) using dark chambers on the same plots. GPP can be calculated as GPP = NEE – Rd. Simultaneous observations of COS uptake will allow the GPP calculation presented here to be compared to conventional methods.

5.5.2 Seasonal COS fluxes and temperature During the dry season, grass dries out and organic matter senesces in place; in the wet season, new growth sprouts through the dense litter layer. In the dry season temperature appeared to have had a greater influence on COS exchange. Taken together with soil VWC, a multiple linear regression of temperature and moisture on COS exchange yielded an r2 of 0.67 (n = 8). After green up in the wet season, the relationship to temperature became more complicated (Figure 5.5), depending more on whether living grasses were present. To untangle ecosystem COS exchange from leaf-level COS exchange, COS fluxes of non-photosynthesizing components of the ecosystem have been inferred from chambers with similar temperature and soil moisture, but containing no green plants. The relationship of COS fluxes to soil conditions may be based on physical constraints to soil/litter-atmosphere trace gas exchange and the availability of COS precursors.

20 ) 1

− 0 sec 2 − −20 r2=.39 r2=.55 −40

−60 COS fluxes (pmol m

−80 10 15 20 25 Temperature (C) Figure 5.5 Fluxes of COS and average chamber temperature. Asterisk symbols are wet season measurements; cross symbols are dry season measurements. The r2 values given are for least squares linear regression.

Soil structure and water content are additional factors regulating atmospheric COS uptake. In a study of 4 soil types, the uptake of COS was controlled by the diffusivity of the soil as related to water-filled pore space (Van Diest and Kesselmeier 2008). For this particular soil, COS uptake could be hindered by decreased diffusivity, preventing COS from entering into the soil profile where it was consumed or adsorbed. When water was added to already moist plots, uptake appeared to decrease in all cases (Figure 5.4); however, it is impossible to separate COS uptake from production without using an additional tracer method.

Sulfur-containing amino acids can act as the precursors of COS (Minami and Fukushi 1981), so it may be expected that COS production should increase with soil moisture, as increased microbial activity may liberate more organic matter. After wetting a plot during the Aug 24 outing, apparent production far exceeds COS uptake from the atmosphere (Figure 5.4). On the other hand, this could have been the result of displacing the soil atmosphere by the addition of water, leading to emission of COS otherwise trapped in the soil profile. This second interpretation is supported by a subsequent, much lower observation 41 minutes later.

This suggests that an increase in COS production caused by a change in available water may be initially dampened by the increase in water-filled pore space, acting as a barrier between in-soil production of COS and the atmosphere. The relationship between soil moisture and COS production was complicated further by the release of DMS, a precursor to COS that is produced by different biochemical pathways, in a similar order of magnitude to COS uptake.

5.5.3 DMS, CO2, and Soil Moisture The field moisture flux measurements indicate DMS production may be specific to live grasses. DMS fluxes from dead grass and soil during the bulk of the dry season were generally low (Figure 5.6). However, the highest flux of DMS was found in September before first precipitation event, when grass litter was matted down (Figure 5.3). Overall, DMS emissions to the atmosphere were on average higher during the wet season, when field soil moisture had a much stronger influence on DMS production compared to dry season fluxes (Figure 5.6). Wet season plots yielded much larger increases in DMS production after water addition, perhaps from an already active microbial population and from interactions with live plants. DMS production has been found over several important agricultural grasses (Kanda, et al. 1995). The DMS fluxes here fall within the expected range found in the literature (Table 5.3).

Citation Ecosystem Range (pmol m-2 sec-1) Geng & Mu (2004) grass lawn 0 – 3 DeMello & Hines (1994) Sphagnum peatlands 1 – 118 DeLaune, et al. (2002) coastal marsh 12 – 1247 This Study coastal grassland 0 – 67 Table 5.3 DMS fluxes reported in the literature converted to pmol m-2 sec-1 and compared to this study.

)

1 40 − sec

2 30 −

DMS fluxes (pmol m 0 5 10 15 20 25 Soil Moisture (VWC %) Figure 5.6 Field (unmanipulated) soil moisture (VWC %) and DMS fluxes (pmol DMS m-2 s-1) in the dry season (crosses) and the wet season (asterisks).

The water manipulation experiments show that increasing soil water content stimulated DMS fluxes to the atmosphere in the 6 of 9 cases where DMS fluxes at field moisture were initially near zero (Figure 5.4). DMS production can be microbially-mediated: microbes produce the osmoprotectant dimethylsulfoniopropionate (DMSP) which decomposes, producing DMS. The production of DMSP and subsequent generation of DMS is well known in marine ecosystems (Schafer et al., 2010).

However, observations of CO2 do not reveal a simple link between microbial activity and terrestrial DMS production. Basal respiration, observing the CO2 flux from soil, is a well-established method for measuring soil microbial activity. Soils that experience lengthy dry periods have been found to exhibit increases in carbon mineralization after rain events due to a combination of released soil organic matter and increased microbial activity (Fierer and Schimel, 2003). This effect was observed in the Sep 20 outing, when a sustained increase in CO2 net release to the atmosphere was observed over the course of two hours. In this outing, DMS production was already high and was suppressed slightly by the water addition. Only one plot on Jun 28 saw a persistent order of magnitude increase of DMS emissions.

5.6 Conclusions and Implications

Here we have demonstrated a different approach for measuring gross primary production of ecosystems during manipulation experiments by measuring CO2 and fluxes of COS. This approach can be applied to any ecosystem where the fluxes of COS from plants and soils can be distinguished, as evaluated by chamber measurements. In a Mediterranean grassland, this determination was more straightforward because the landscape had few green plants during the dry season, allowing for COS fluxes to be quantified without the complication of COS uptake through stomata.

Using estimates of temperate grassland extent (1.78 ×109 ha) and GPP (8.5 Pg C yr-1) from Beer et al. (2010), we can evaluate the impact of soil COS flux corrections on GPP estimates. Assuming a 100 day growing season with 12 h of carbon assimilation per day, the carbon flux from these ecosystems would average 9 μmol C -2 -1 m sec . We can then anticipate the flux of COS with the equation FCOS = -1 GPP[COS][CO2] vCOS/CO2, with CO2 at 400 ppm, COS at 500 ppt and vCOS/CO2 at 1.66, the average of leaf relative uptakes from C3 and C4 plants (Stimler et al., 2010) weighted by their contribution to GPP (Still et al., 2003). This approach yields a COS flux estimate of -19 pmol m-2 sec-1. Dry season COS flux observations ranged from -25 to +7.3 pmol m-2 sec-1. This translates to a GPP of 5 to 21 μmol C m-2 sec-1 or 5 to 20 Pg C yr-1, an additional error of +138% to -38%. This uncertainty for temperate grasslands and shrublands is comparable to recent GPP uncertainty estimates for the entire globe (Koffi et al., 2012), underscoring the importance of correcting for soil COS fluxing when estimating GPP.

The advantage of the estimating GPP with COS lies in generating a real-time, daylight estimate of GPP, contrasted with annual estimates of NPP by harvesting whole plants or problems with using eddy flux covariance towers to measure ecosystem respiration at night when turbulent mixing is low. This technique may be relevant to future studies focused on carbon sequestration efforts, assessing the effects of drought stress, and large scale manipulation experiments where high frequency GPP estimates are needed.

Works Cited

Asaf, D., Rotenberg, E., Tatarinov, F., Dicken, U., Montzka, S.A., Yakir, D., 2013. Ecosystem photosynthesis inferred from measurements of carbonyl sulphide flux. Nature Geosciences 6, 186–190. Barnes, I., Becker, K.H., Patroescu, I., 1994. The tropospheric oxidation of dimethyl sulfide: A new source of carbonyl sulfide. Geophysical Research Letters 21, 2389–2392.

Beer, C., Reichstein, M., Tomelleri, E., Ciais, P., Jung, M., Carvalhais, N., Rödenbeck, C., Arain, M.A., Baldocchi, D., Bonan, G.B., 2010. Terrestrial gross carbon dioxide uptake: global distribution and covariation with climate. Science 329, 834– 838. Blonquist, J.M., Montzka, S.A., Munger, J.W., Yakir, D., Desai, A.R., Dragoni, D., Griffis, T.J., Monson, R.K., Scott, R.L., Bowling, D.R., 2011. The potential of carbonyl sulfide as a proxy for gross primary production at flux tower sites. Journal of Geophysical Research: Biogeosciences 116. Campbell, J.E., Carmichael, G.R., Chai, T., Mena-Carrasco, M., Tang, Y., Blake, D.R., Blake, N.J., Vay, S.A., Collatz, G.J., Baker, I., Berry, J. A, Montzka, S. A, Sweeney, C., Schnoor, J.L., Stanier, C.O., 2008. Photosynthetic control of atmospheric carbonyl sulfide during the growing season. Science 322, 1085–1088. De Mello, W.Z., Hines, M.E., 1994. Application of static and dynamic enclosures for determining dimethyl sulfide and carbonyl sulfide exchange in Sphagnum peatlands: Implications for the magnitude and direction of flux. Journal of Geophysical Research 99, 14–601. Kanda, K., Tsuruta, H., Minami, K., 1995. Emissions of biogenic sulfur gases from maize and wheat fields. Soil Science and Plant Nutrition 41, 1–8. Khan, M.A.H., Whelan, M.E., Rhew, R.C., 2012. Analysis of low concentration reduced sulfur compounds (RSCs) in air: Storage issues and measurement by gas chromatography with sulfur chemiluminescence detection. Talanta 88, 581– 586. Koffi, E.N., Rayner, P.J., Scholze, M., Beer, C., 2012. Atmospheric constraints on gross primary productivity and net ecosystem productivity: Results from a carbon- cycle data assimilation system. Global Biogeochemical Cycles 26, GB1024. Livingston, G.P., Hutchinson, G.L., 1995. Enclosure-based measurement of trace gas exchange: applications and sources of error, in: Biogenic Trace Gases: Measuring Emissions from Soil and Water. pp. 14–51. Minami, K., Fukushi, S., 1981. Volatilization of carbonyl sulfide from paddy soils treated with sulfur-containing substances. Soil Science and Plant Nutrition 27, 339–345. Montzka, S. A., Calvert, P., Hall, B.D., Elkins, J.W., Conway, T.J., Tans, P.P., Sweeney, C., 2007. On the global distribution, seasonality, and budget of atmospheric carbonyl sulfide (COS) and some similarities to CO2. Journal of Geophysical Research 112. Orchard, V.A., Cook, F.J., 1983. Relationship Between Soil Respiration and Soil Moisture. Soil Biology and Biochemistry 15, 447–453. Protoschill-Krebs, G., Wilhelm, C., Kesselmeier, J., 1996. Consumption of carbonyl sulphide (COS) by higher plant carbonic anhydrase (CA). Atmospheric Environment 30, 3151–3156. Sandoval-Soto, L., Stanimirov, M., Von Hobe, M., Schmitt, V., Valdes, J., Wild, A., Kesselmeier, J., 2005. Global uptake of carbonyl sulfide (COS) by terrestrial vegetation: Estimates corrected by deposition velocities normalized to the uptake of carbon dioxide (CO2). Biogeosciences 2, 125–132.

Schaefer, K., Schwalm, C.R., Williams, C., Arain, M. Altaf, Barr, A., et al., 2012. A model-data comparison of gross primary productivity: Results from the North American Carbon Program site synthesis. Journal of Geophysical Research 117. Schäfer, H., Myronova, N., Boden, R., 2010. Microbial degradation of dimethylsulphide and related C1-sulphur compounds: organisms and pathways controlling fluxes of sulphur in the biosphere. Journal of Experimental Botany 61, 315–334. Schenk, S., Kesselmeier, Jürgen, Anders, E., 2004. How does the exchange of one oxygen atom with sulfur affect the catalytic cycle of carbonic anhydrase? Chemistry–A European Journal 10, 3091–3105. Schulz, M., Stonestrom, D., Von Kiparski, G., Lawrence, C., Masiello, C., White, A., Fitzpatrick, J. “Seasonal Dynamics of CO2 Profiles Across a Soil Chronosequence, Santa Cruz, California.” Applied Geochemistry 26, Supplement (June 2011): S132–S134. Seibt, U., Kesselmeier, J., Sandoval-Soto, L., Kuhn, U., Berry, J.A., 2010. A kinetic analysis of leaf uptake of COS and its relation to transpiration, photosynthesis and carbon isotope fractionation. Biogeosciences 7, 333–341. Still, C.J., Berry, J.A., Collatz, G.J., and DeFries, R.S., 2003. Global Distribution of C3 and C4 Vegetation: Carbon Cycle Implications. Global Biogeochemical Cycles 17. Stimler, K., Berry, J.A., Montzka, S.A., Yakir, D., 2011. Association between carbonyl sulfide uptake and 18Δ during gas exchange in C3 and C4 leaves. Plant Physiology 157, 509–517. Stimler, K., Montzka, S.A., Berry, J.A., Rudich, Y., Yakir, D., 2010. Relationships between carbonyl sulfide (COS) and CO2 during leaf gas exchange. New Phytologist 186, 869–878. Suntharalingam, P., Kettle, A.J., Montzka, S.M., Jacob, D.J., 2008. Global 3-D model analysis of the seasonal cycle of atmospheric carbonyl sulfide: Implications for terrestrial vegetation uptake. Geophysical Research Letters 35. Van Diest, H., Kesselmeier, J., 2008. Soil atmosphere exchange of carbonyl sulfide (COS) regulated by diffusivity depending on water-filled pore space. Biogeosciences 5, 475–483. Vico, G., Manzoni, S., Palmroth, S., Katul, G., 2011. Effects of stomatal delays on the economics of leaf gas exchange under intermittent light regimes. New Phytologist 192, 640–652. Watts, S.F., 2000. The mass budgets of carbonyl sulfide, dimethyl sulfide, carbon disulfide and hydrogen sulfide. Atmospheric Environment 34, 761–779. White, A.F., Schulz, M.S., Vivit, D.V., Blum, A.E., Stonestrom, D.A., Anderson, S.P., 2008. Chemical weathering of a marine terrace chronosequence, Santa Cruz, California I: Interpreting rates and controls based on soil concentration– depth profiles. Geochimica et Cosmochimica Acta 72, 36–68. Wofsy, S.C., 2001. Where Has All the Carbon Gone? Science 292, 2261–2263. Xu, L., Baldocchi, D.D., 2004. Seasonal variation in carbon dioxide exchange over a Mediterranean annual grassland in California. Agricultural and Forest Meteorology 123, 79–96.

6. Conclusion

In this dissertation we have explored three cases which challenge current thinking about terrestrial-atmospheric sulfur gas exchange. Although plants are the major sink of atmospheric COS on the continents, in Chapter 3 we showed that some salt marsh plants may actively produce large amounts of COS. Similarly, aerobic soils are considered to be another significant sink of atmospheric COS in natural ecosystems, but in Chapter 4 we found net COS production when a wheat field soil was exposed to light and high temperatures. Finally, we investigated separating plant and soil COS fluxes by conducting a year long experiment at a site where green plants were present for only part of the year. This last study emphasizes the importance of correcting for COS soil exchange when estimating gross primary production (GPP) from ecosystem COS flux measurements.

6.1 COS exchange between plants and the atmosphere It has been known for decades that COS is taken up by enzymes in plants (Protoschill-Krebs and Kesselmeier, 1992; Protoschill-Krebs et al., 1996), although it was not until 2007 that Montzka et al. pointed out COS terrestrial uptake scaled with GPP. Several studies sought to use estimates of carbon uptake by plants to balance the atmospheric COS budget (Campbell et al., 2008; Kettle et al., 2002; Suntharalingam et al., 2008). Only recently have observations of COS been used to answer questions about the carbon cycle instead (Asaf et al., 2013; Billesbach et al., 2014; Blonquist et al., 2011).

In order to relate COS to CO2 fluxes, the concept of a leaf relative uptake (LRU) was -1 introduced, defined as LRU = (ACOS/[COS])(ACO2/[CO2]) , where [COS] and [CO2] are gas concentrations and ACOS and ACO2 are the assimilation rates of COS and CO2, respectively (Sandoval-Soto et al., 2005). The leaf relative uptake of COS to CO2 has been found to be consistent for plants of the same photosynthetic pathway: 1.16 ± 0.2 for C4 plants and 1.82 ± 0.18 for C3 plants (Stimler et al., 2010).

A plant that uses the C3 photosynthetic pathway (Doliner and Jolliffe, 1979), Batis martima is an important exception to the consistent LRU estimate. The presence of B. maritima in wetland study plots increased the emission of COS to the atmosphere compared to soil-only plots (Chapter 3). Surprisingly, this indicates that COS production can be mediated by plants. If our wetland study is representative of global wetlands, the COS wetland source that has been used in global COS models (Kettle et al., 2002; Suntharalingam et al., 2008) is underestimated by a factor of 3. More field studies are needed to assess the spatial heterogeneity of the terrestrial COS source.

6.2 COS exchange between soils and the atmosphere Soils were thought to have a straightforward relationship with COS fluxes, where anoxic soils exhibited net COS production and oxic soils net COS consumption (Watts, 2000). As demonstrated in Yi et al., (2008) (Figure 6.1), the same soil that produced COS when flooded could effect net consumption when not flooded. This suggests that redox potential, as influenced by soil moisture, controls the overall direction of net COS exchange in soils.

The relationship becomes more complicated with shifting soil moisture and redox potential. We conducted a pilot study in the Luquillo Experimental Forest, a rainforest in Puerto Rico, using the equipment deployed in the Chapter 5 study. Most of the experiment was conducted under rainy conditions. COS fluxes over rainforest soil and litter were too variable to generate a static flux chamber measurement using a 30 minute incubation time. At the end of the field campaign we had 4 useable flux measurements, represented in Figure 6.1.

Contrary to the conception that redox potential drives COS production, we showed in Chapter 4 that net consumption or production of COS in an aerobic wheat field soil was controlled by temperature and light. Photoproduction of COS may be associated with organosulfur molecules photosensitized by soil organic matter, a process analogous to COS production in seawater (Flöck et al., 1997). The soil samples were taken from a site where eddy flux covariance measurements of CO2 and COS were performed in order to estimate GPP; variable soil COS production interfered with GPP calculations (Billesbach et al., 2014). In order to use COS observations as a proxy for GPP, soil COS exchange must be taken into account.

6.3 Soil COS exchange corrections for using COS as a proxy for GPP Investigating the Mediterranean grassland in Chapter 5, with its distinct dry senescence and wet growing seasons, allowed us to separately assess soil and plant COS exchange. Wet season measurements resulted in net COS uptake when green plants were present. We were able to manipulate the soil moisture content during the dry season with artificial rain, assessing the variability of soil COS fluxes with soil moisture outside of the influence of live grasses. Using this information, we calculated estimates of gross CO2 exchange using a model developed by Campbell et al., (2008) and furthered by Seibt et al., (2010) and Stimler et al., (2010). We showed that calculating GPP from COS measurements could lead to a +138% to - 38% error when soil COS exchange was not accounted for.

In biomes where plants do not conveniently lay dormant for months, a chamber- based approach could be sufficient to quantify the production or non-leaf consumption of COS. For example, observations of temperate and subtropical forest COS soil fluxes range between -8 and +1.45 pmol COS m-2 s-1.(Castro and Galloway, 1991; Steinbacher et al., 2004; White et al., 2010; Z. Yi et al., 2007). Compared to ecosystem scale forest measurements (Xu et al., 2002), the soil term represents 8% of the total COS exchange.

/ /

reported reported avg range +/− error soil+plants Whelan et al. 2013

soil only DeLaune et al. 2002 SALTMARSH Fried et al. 1993 deMello & Hines 1994 PEATLAND unflooded Yi et al. 2008 Wetland RICE PADDY Liu et al. 2010

water added artificially Whelan, Chapter 5

Variable Redox Whelan Chapter 5 MEDITERRANEAN

Whelan, unpublished data RAINFOREST Oxic OXIC SYSTEMS Geng & Mu 2004 Li et al. 2006 Xu et al. 2002 Asaf et al. 2013 wheat field post−harvest Billesbach et al. 2014 All Oxic Soils Reported / / −125 − 100 −75 −50 −25 0 25 50 75 100 275 300 COS fluxes (pmol m−2 sec−1)

Figure 6.1 A summary of COS fluxes from field campaigns across various ecosystem types, divided into soil-only sites and soil + plants sites.

To quantify the error introduced by calculating GPP from ecosystem COS exchange without correcting for soil fluxes, we use the biome GPP estimates from Beer et al. (2010) and back calculate anticipated COS fluxes using the equation from Campbell -1 et al. (2008), FCOS = GPP[COS][CO2] vCOS/CO2, where FCOS is the uptake of COS into plant leaves, [CO2] and [COS] are concentrations assumed to be 400 ppm and 500 ppt, respectively, and vCOS/CO2 is the leaf relative uptake, found by Stimler et al. (2010) to be 1.16 ± 0.2 for C4 plants and 1.82 ± 0.18 for C3 plants. We further assume a 100 day growing season with 12h of light per day for the purposes of converting between annual estimates of GPP and field measurements calculated in sec-1 units, though this is obviously does not represent the diversity of biomes’ carbon assimilation patterns. Additionally, we assume that plants in tropical and desert biomes photosynthesize using the C4 pathway. The results are presented in Table 6.1. 74

Biome GPP Biome area Anticipated FCOS,soil % estimated (10^9 ha) FCOS, plants in from field overestimated by Beer et pmol m-2 studies in GPP by al. (2010) sec-1 pmol m-2 neglecting soil in Pg C yr-1 sec-1 COS Tropical -124 to - +120 to +60 forests 40.8 1.75 -102 61a Temperate -8 to 1.45b +20 to -3 forests 9.9 1.04 -42 Boreal 1.2 to 3.8c -5 to -14 forests 8.3 1.37 -27 Tropical No data savannas and grasslands 31.3 2.76 -32 Temperate -25 to 7.3d +119 to -34 grasslands and shrublands 8.5 1.78 -21 Deserts 6.4 2.77 -6 No data Tundra 5.27 to -42 to -220 1.6 0.56 -13 27.6e Croplands 14.8 1.35 -48 -18 to 40f +37 to -83 Total 121.7 0 -33 Table 6.1 The error introduced to GPP estimates when COS soil fluxes are held negligible. The last column describes how much GPP would be overestimated, as a percentage of GPP estimated by Beer et al. (2010), if soil COS uptake determined from chamber measurements was included in the

FCOS,plants term. Negative values indicate underestimated GPP. aSee previous section for measurements of tropical rainforest soil fluxes. bRange of values from Castro and Galloway, 1991, Steinbacher et al., 2004, White et al., 2010, and Yi et al., 2007. cThe average reported here is the average and one standard deviation from non-vegetated plots in a boreal forest, defined as plots having less than 10% vegetation cover (Simmons et al., 1999) dRange reported in Chapter 5. The error estimate here is different than the one calculated in Chapter 5 because a different LRU was used. eThe lower production is from De Mello and Hines (1994). The larger production is an average estimate from Fried et al. (1993) fPost-harvest soil exchange estimate from the wheat field (Billesbach et al., 2014) investigated further in Chapter 4.

The importance of taking COS soil fluxes into account when estimating GPP is clear from the results in Table 6.1. However, observations are still scarce: despite a plea for data from desert soils in 2002 by Kettle et al., we were not able to find such a study in the literature ten years later. Boreal forests soil COS exchange estimates are represented by a single study performed at a single site in Sweden over the course of two months in 1993 (Simmons, 1999). Modeling efforts suggest large COS fluxes in the tropics (J. Berry et al., 2013; Suntharalingam et al., 2008) and tropical forests and savannas are associated with 60% of global terrestrial GPP (Beer et al., 2010) However, there remains a dearth of observations in tropical latitudes.

Measurements of COS have been used recently to estimate GPP on the ecosystem scale (Asaf et al., 2013; Billesbach et al., 2014; Blonquist et al., 2011). We need to know more about COS soil fluxes in specific ecosystems in order to use this approach in good faith. This dissertation describes quantification of sulfur gas fluxes from soils and plants independent of each other in several ecosystems. With further application of this approach, particularly in tropical forests and savannas, COS fluxes can be used as a viable measure of GPP in terrestrial ecosystems globally.

Works Cited

Asaf, D., Rotenberg, E., Tatarinov, F., Dicken, U., Montzka, S.A., Yakir, D., 2013. Ecosystem photosynthesis inferred from measurements of carbonyl sulphide flux. Nature Geosciences 6, 186–190. Beer, C., Reichstein, M., Tomelleri, E., Ciais, P., Jung, M., Carvalhais, N., Rödenbeck, C., Arain, M.A., Baldocchi, D., Bonan, G.B., 2010. Terrestrial gross carbon dioxide uptake: global distribution and covariation with climate. Science 329, 834– 838. Berry, J., Wolf, A., Campbell, J.E., Baker, I., Blake, N., Blake, D., Denning, A.S., Kawa, S.R., Montzka, S.A., Seibt, U., Stimler, K., Yakir, D., Zhu, Z., 2013. A coupled model of the global cycles of carbonyl sulfide and CO2: A possible new window on the carbon cycle. Journal of Geophysical Research: Biogeosciences 118, 842–852. Billesbach, D.P., Berry, J.A., Seibt, U., Maseyk, K., Torn, M.S., Fischer, M.L., Abu-Naser, M., Campbell, J.E., 2014. Growing season eddy covariance measurements of carbonyl sulfide and CO2 fluxes: COS and CO2 relationships in Southern Great Plains winter wheat. Agricultural and Forest Meteorology 184, 48–55. Blonquist, J.M., Montzka, S.A., Munger, J.W., Yakir, D., Desai, A.R., Dragoni, D., Griffis, T.J., Monson, R.K., Scott, R.L., Bowling, D.R., 2011. The potential of carbonyl sulfide as a proxy for gross primary production at flux tower sites. Journal of Geophysical Research: Biogeosciences 116. Campbell, J. E, Carmichael, G.R., Chai, T., Mena-Carrasco, M., Tang, Y., Blake, D.R., Blake, N.J., Vay, S.A., Collatz, G.J., Baker, I., Berry, J. A, Montzka, S. A, Sweeney, C., Schnoor, J.L., Stanier, C.O., 2008. Photosynthetic control of atmospheric carbonyl sulfide during the growing season. Science 322, 1085–1088.

Castro, M.S., Galloway, J.N., 1991. A comparison of sulfur-free and ambient air enclosure techniques for measuring the exchange of reduced sulfur gases between soils and the atmosphere. Journal of Geophysical Research 96, 15427–15. DeLaune, R.D., Devai, I., Lindau, C.W., 2002. Flux of reduced sulfur gases along a salinity gradient in Louisiana coastal marshes. Estuarine, Coastal and Shelf Science 54, 1003–1011. De Mello, W.Z., Hines, M.E., 1994. Application of static and dynamic enclosures for determining dimethyl sulfide and carbonyl sulfide exchange in Sphagnum peatlands: Implications for the magnitude and direction of flux. Journal of Geophysical Research 99, 14–601. Doliner, L.H., Jolliffe, P.A., 1979. Ecological evidence concerning the adaptive significance of the C4 dicarboxylic acid pathway of photosynthesis. Oecologia 38, 23–34. Flöck, O.R., Andreae, Meinrat O., Dräger, M., 1997. Environmentally relevant precursors of carbonyl sulfide in aquatic systems. Marine Chemistry 59, 71– 85. Fried, A., Klinger, L.F., Erickson, D.J., 1993. Atmospheric carbonyl sulfide exchange in bog microcosms. Geophysical Research Letters. 20, 129–132. Geng, C., Mu, Y., 2004. Carbonyl sulfide and dimethyl sulfide exchange between lawn and the atmosphere. Journal of Geophysical Research: Atmospheres 109. Kettle, A.J., Kuhn, U., Von Hobe, M., Kesselmeier, J., Andreae, M.O., 2002. Global budget of atmospheric carbonyl sulfide: Temporal and spatial variations of the dominant sources and sinks. Journal of Geophysical Research: Atmospheres 107. Li, X., Liu, J., Yang, J., 2006. Variation of H2S and COS emission fluxes from Calamagrostis angustifolia wetlands in Sanjiang Plain, Northeast China. Atmospheric Environment 40, 6303–6312. Liu, J., Geng, C., Mu, Y., Zhang, Y., Xu, Z., Wu, H., 2010. Exchange of carbonyl sulfide (COS) between the atmosphere and various soils in China. Biogeosciences 7, 753–762. Protoschill-Krebs, G., Kesselmeier, J., 1992. Enzymatic pathways for the consumption of carbonyl sulphide (COS) by higher plants. Botanica Acta 105, 206–212. Protoschill-Krebs, G., Wilhelm, C., Kesselmeier, J., 1996. Consumption of carbonyl sulphide (COS) by higher plant carbonic anhydrase (CA). Atmospheric Environment 30, 3151–3156. Sandoval-Soto, L., Stanimirov, M., Von Hobe, M., Schmitt, V., Valdes, J., Wild, A., Kesselmeier, J., 2005. Global uptake of carbonyl sulfide (COS) by terrestrial vegetation: Estimates corrected by deposition velocities normalized to the uptake of carbon dioxide (CO2). Biogeosciences 2, 125–132. Seibt, U., Kesselmeier, J., Sandoval-Soto, L., Kuhn, U., Berry, J. A., 2010. A kinetic analysis of leaf uptake of COS and its relation to transpiration, photosynthesis and carbon isotope fractionation. Biogeosciences 7, 333–341.

Simmons, J.S., Klemedtsson, L., Hultberg, H., Hines, M.E., 1999. Consumption of atmospheric carbonyl sulfide by coniferous boreal forest soils. Journal of Geophysical Research 104, 11569–11576. Steinbacher, M., Bingemer, H.G., Schmidt, U., 2004. Measurements of the exchange of carbonyl sulfide (OCS) and carbon disulfide (CS2) between soil and atmosphere in a spruce forest in central Germany. Atmospheric Environment 38, 6043–6052. Stimler, K., Montzka, S.A., Berry, J.A., Rudich, Y., Yakir, D., 2010. Relationships between carbonyl sulfide (COS) and CO2 during leaf gas exchange. New Phytologist 186, 869–878. Suntharalingam, P., Kettle, A.J., Montzka, S.M., Jacob, D.J., 2008. Global 3-D model analysis of the seasonal cycle of atmospheric carbonyl sulfide: Implications for terrestrial vegetation uptake. Geophysical Research Letters 35. Watts, S.F., 2000. The mass budgets of carbonyl sulfide, dimethyl sulfide, carbon disulfide and hydrogen sulfide. Atmospheric Environment 34, 761–779. Whelan, M.E., Min, D., Rhew, R.C., 2013. Salt marshes as a source of atmospheric carbonyl sulfide. Atmospheric Environment 73, 131-137. White, M.L., Zhou, Y., Russo, R.S., Mao, H., Talbot, R., Varner, R.K., Sive, B.C., 2010. Carbonyl sulfide exchange in a temperate loblolly pine forest grown under ambient and elevated CO2. Atmospheric Chemistry & Physics 10, 547–561. Xu, X., Bingemer, H.G., Schmidt, U., 2002. The flux of carbonyl sulfide and carbon disulfide between the atmosphere and a spruce forest. Atmospheric Chemistry and Physics Discussions 2, 181–212. Yi, Z., Wang, X., Sheng, G., Zhang, D., Zhou, G., Fu, J., 2007. Soil uptake of carbonyl sulfide in subtropical forests with different successional stages in south China. Journal of Geophysical Research 112. Yi, Zhigang, Wang, Xinming, Sheng, Guoying, Fu, Jiamo, 2008. Exchange of carbonyl sulfide (OCS) and dimethyl sulfide (DMS) between rice paddy fields and the atmosphere in subtropical China. Agriculture, Ecosystems & Environment 123, 116–124.