Biosphere-atmosphere interactions

in the p How­ ever, there is a marked similarity obtained from Scotland and from 15% and ]% of N in p u t lost as NO from th s cke with uc t i o n of NO e miss i on s as compared t o an unlimed s p control site. The results indicate that th e in e miss o n was due to an increase in NO consumption within the limed s o il. Liming of a spruce site res'lllted in a s i gnifican t ammonifica ti o n , ni­ trification, and N20emissions as compared with an untreated spruce control s i t e. On the basis of these results was concluded that the importance of temperate and boreal forests for the global Np sourc e strength have been significantly underestimated in the past and that these forests, in which N deposition is high, most likely contribute in excess oho Tg


Introduction
The contemporary atmosphere was created as a result of biological activity some two billion years ago. To this day, its natural composition is supported and modified, mostly through biological processes of trace gas pro duction and destruction, while also involving physical and chemical degradation processes. The biosphere has a major influence on present environmental conditions, both on a regional and global scale. One of the best· documented and most important indicators of global change is the progressive increase of a number of trace gases in the atmosphere, among them carbon dioxide (C02), methane (CH4), and nitrous oxide (N20), all of which are of biospheric origin. There is considerable uncertainty, however, regarding the processes that de termine the concentration and distribution of trace gases and aerosols in the atmosphere and the causes and con· sequences of atmospheric change (Andreae and Schimel i989). To improve our understanding IGAC created an environment for multi-disciplinary collaboration among biologists, chemists, and atmospheric scientists. This was essential to develop analytical methods, to characterise ecosystems, to investigate physiological controls, to de· velop and validate micrometeorological theory, and to design and develop diagnostic and predictive models (Matson and Ojima i990).
Interactions between the biosphere and the atmo· sphere are part of a complex, interconnected system. The emission and uptake of atmospheric constituents by the biota influence chemical and physical climate through interactions with atmospheric photochemistry and Earth's radiation budget. Comparatively small amounts of CH4 and N20 present in the atmosphere make sub stantial contributions to the global greenhouse effect. In addition, emissions of hydrocarbons and nitrogen oxides from biomass burning in the Tropics result in the photochemical production of large amounts of ozone (03) and acidity in the tropical atmosphere. In turn, climate change and atmospheric pollution alter the rates and sometimes even the direction of chemical ex change between the biosphere and atmosphere through influences at both individual organism and ecosystem levels. Recent and expected future changes in land use and land management practices provide further impe tus for closely examining climate-gas flux interactions. Anthropogenic influences, e.g. tropical deforestation and the widespread implementation of agricultural tech nologies, have and will continue to make significant al terations in the sources and sinks for the various trace gases.
Ten years ago, at the beginning of IGAC, researchers sought to establish the source and sink strength of gases in different kinds of ecosystems, in different areas of the world. Specific goals of the programme, related to the biosphere included: • to understand the interactions between atmospheric chemical composition and biological and climatic processes; • to predict the impact of natural and anthropogenic forcings on the chemical composition of the atmo sphere; and • to provide the necessary knowledge for the proper maintenance of the biosphere and climate.
Earlier extrapolations of gas fluxes over space and time were often based on a single, or very small, set of measurements, and researchers sought for "repre sentative" sites at which to make those crucial measure ments. IGAC brought a new focus to the variability among ecosystems and regions of the world, in order to understand better the factors controlling fluxes ( Galbally i989). For example, studies of CH4 flux from wetlands and rice paddies of N20 flux from natural and man aged ecosystems, and of dimetfxsulphide (DMS) efs· sions from oceans, consciously spanned gradients of temperature, hydrological characteristics, soil types, marine systems, management regimes, and nitrogen deposition. One result of this strategy has been the rec ognition that the same basic processes were responsi ble for gas fl uxes across regions, latitudinal zones, and environments. This chapter gives a general overview of the progress that has been made in the field as a whole within the last decade, with emphasis on research ac tivities stimulated, initiated, and/or endorsed by the IGAC community. It is not our intent to provide current assessments of all trace gas source and sink strengths, as those budgets have been compiled and published (with considerable contributions by IGAC researchers) in recent Intergovernmental Panel on Cl i mate Change (IPCC) documents. Examples of research not conducted within the IGAC framework but relevant to the topic are CH4 from landfills, ruminant livestock, and termites; information on these topics can be found in IPCC (1996IPCC ( , 1999.
Exchanges of biogenic trace gases between surfaces and the atmosphere depend on the production and con sumption of gases by microbial and plant processes, on physical transport through soils, sediments, and water, and on flux across the surface-air boundaries. Thus, to understand and predict fluxes, studies of whole ecosys tems are required. The goals of research over the past decade have been to develop an understanding of the factors that control flf£ organise the measurements so that they are useful for regional and global scale budg ets, and use the knowledge to predict how fluxes are likely to change in the future.
The IGAC Project focussed on issues of specific in terest over a number of different geographical regions of Earth. A variety of projects have been conducted over the last ten years, many of which addressed issues re lated to exchange between the biosphere and the atmo sphere. Several field campaigns, using a combination of measurement and modelling techniques, have been conducted very successfully under the IGAC umbrella, e.g. in southern Africa (SAFARI 1992 and2000) and in various oceanic regions (ACE-1, ACE-2, and ACE-Asia) (see A.5).
Why certain trace gases were studied together and why various scientific approaches were adopted to study them is described in this chapter. Research findings spe cifically related to the exchange of trace gases and aero sols between the atmosphere and the terrestrial and marine biospheres will be given. In the terrestrial sec tion, special attention is given to biomass burning and wet deposition in the Tropics, because of the significant contribution made by IGAC to these programmes. We also consider some of the anthropogenic activities that alter biosphere-atmosphere exchange and discuss po tential feedbacks related to climate change, regional level air pollution, and deposition. In the marine section, emphasis is on the biogeochemistry of DMS, given that the greatest advances were made on this topic. The chap ter concludes by summarising the major accomplish ments of the last decade and highlighting some of the remaining research challenges.

Key Biogenic Gases or Families and their
Relevance to Atmospheric Chemistry The study of atmospheric composition has largely focussed on trace compounds that affect either the radiative properties of the atmosphere, or the biosphere as nutrients or toxins, or play a key role in atmospheric chemistry. The trace gases that are important in this regard have been summarised in the preceding chapter. This chapter considers the role of the biosphere in emis sion or removal of such compounds. Although C02 and water (H20) are both greenhouse gases which are strongly affected by the biosphere, stud ies of these compounds have generally been conducted in parallel scientific communities, and IGAC has main tained a focus on the chemically reactive greenhouse gases. Thus no attempt is made here to cover the large body of research on the global carbon cycle and the in teractions of C02 with the biosphere. Methane (CH4) is a greenhouse gas with a lifetime in the atmosphere of about nine years. Its atmospheric concentration is largely controlled by the biosphere, with 70% or more of current emissions and virtually all of pre-industrial emissions being biogenic (Fig. 2.1;Milich 1999). The dominant biogenic production process for CH4 is microbial breakdown of organic compounds in Year anaerobic conditions. This occurs in flooded soils such as natural wetlands and rice paddies, in the rumen of animals such as cattle, in landfills, and in anoxic layers in the marine water column and sediments. Methane is also emitted directly to the atmosphere from burning vegetation as a product of pyrolytic breakdown of or ganic material.
Changes in land use, particularly increases in num bers of domestic ruminants, the extent of rice paddies, and biomass burning, have more than doubled biogenic CH4 emissions since the preindustrial era (Milich 1999;Ehhalt et al. 2001). Fossil fuel related emissions and a decrease in atmospheric oxidation rates (Thompson 1992) have further increased CH4 concentrations but those aspects fall outside the scope of this chapter.
The atmospheric concentration of CH4 is now about 1745 nmol mo1-1, compared to pre-industrial levels of about 700 nmol mo1-1• Growth rates have been observed directly in the atmosphere since the 1950s (Rinsland et al.1985;Zander et al. 1989) and on an increasingly sys tematic basis since 1978 (Blake and Rowland 1988;Dlugokencky et al. 1994). Concentrations were increas i ng at about 20 nmol mo1-1 yr-1 in the 1970s, but that rate has generally declined to an average of 5 nmol mo1-1 yr -1 over the period 1992 to 1998. High growth rates of about 15 nmol mo1-1 yr -1 occurred in 1991 and 1998 ( Fig. 2.2, Dlugokencky et al. 1998;Ehhalt et al. 2001) and appear to be caused by climate related increases in wetland and/ or biomass burning emi ssions (Dlugokencky et al. 2000;Walter and Matthews 2000). IPCC (2001) estimates its rate of increase at 8.4 nmol mo1-1 yr-1• The evolution of the CH4 budget since the pre-in dustrial era provides a good example of interactions between land use and atmospheric change. A schematic of the change in total emissions from the 18th century to the present is shown in Fig. 2.3 (based on Stern and Kaufmann 1996, Lelieveld et al.1998, and Houweling et al. Schematic of the pre-industrial Holocene and current (1990s) atmospheric methane budget. The mean lifetime derived from the ratio of atmospheric burden to removal rate has increa sed by ca. 10%, which i s broadly consi stent with estimates of the relative decrease in OH from atmospheric chemistry models (based on Stern and Kaufmann (1996), Lelieveld et al. (1998) and Houweling et al. (1999) (see also Fig. 7.4.) NMVOC from anthropogenic sources in 1990 Sources: IEA, UN, FAO, misc. ... Research) database (Olivier et al. 1996) of ecosystem disturbance (e.g. Valentini et al. 1997;Kesselmeier et al. 1998;Crutzen et al. 1999). Neglect of VOC and CO terrestrial emi ssions may cause signifi cant errors in estimates of NEP and changes in carbon storage for some ecosystems. While the oceans are supersaturated with CO and surface production of voes is widespread, the ocean atmosphere fluxes are small, but less well studied, com pared with terrestrial emission estimates. voes show a wide range of reactivities in the troposphere, with life times ranging from minutes (e.g. �-caryophyllene) to two weeks (e.g. methanol) (Atkinson and Arey 2998). Many are emitted at very low rates, and in some cases are offset by plant uptake, thus having a negligible impact on atmospheric chemistry; others impact ozone produc tion (see Chap. 3), aerosol production (see Chap. 4), and the global CO budget.
Primary pollutants emitted mainly as a result of hu man activity include hydrocarbons, CO, and nitrogen oxides. About half the terrestrial surface emissions of CO are due to direct emissions from vegetation and biomass burning. In addi tion about 45% of the total CO source to the atmosphere is due to oxidation of meth-ane and other organics in the atmosphere, which them selves are predominantly biogenic compounds. Because CO is the end product i n the methane oxidation chain the two budgets are closely linked; in addition, CO also originates from the breakdown of voes. The concen trations of CO are temporally and spatially highly vari able due to the short lifet i me of CO and the nature of its discontinuous land based sources. Estimates of anthro pogenic CO emissions for 1990 are shown in Fig. 2 .5.

The Nitrogen Family of Gases:
Ammonia (NH3), N20, and NO Despite its importance for particle formation and cli mate, relatively little effort has been spent on un derstanding the sources and removal processes of NH3• Most work on atmospheric ammonia has been per formed with respect to eutrophication and acidification close to the terrestrial sources; large scale transport and chemistry of NH3 and ammonium ( . 1996) 55 Tg N yr-1, which is of similar magnitude to global NOx·N emission (Bouwman et al. 1997).
The most recent estimate for global NH3 emissions (Bouwman et al. 1997) from animals relied on constant emission factors and amounted to 21.7 Tg N yr-1, which is of similar magnitude as fossil fuel related global NOx-N emissions. The second most important emission cat egory is N-containing synthetic fertiliser. Again, huge differences in agricultural practice and environmental conditions cause a large variation of emissions factors. Overall global emission of ammonia derived from ni trogen fertiliser was estimated to be 9 Tg yr -1 , which is 10% of the amount applied. Interestingly, ammonia losses from application of urea fertiliser to rice paddies seem to contribute strongly to this. Other anthropogenic sources, such as biomass burning, cropland, and humans additionally emit about 10 Tg yr-1• Natural sources, such as soils, vegetation, and oceans, emit about 10-20 Tg yr -1 (Bouwman et al.1997;Schlesinger and Hartley 1992) and are highly uncertain.
Nitrous oxide is an important greenhouse gas with a lifetime of about 120 years. The largest production proc ess for N20 is "leakage" during microbial nitrification and dentrification processes in soil and aquatic systems. Significant emissions also occur from decomposition of animal waste, oxidation of ammonia (NH3), and biomass burning. Biogenic sources of N20 have increased with expansion of food production systems, intensification of agriculture, and anthropogenic mod i fication of the global nitrogen cycle.
The concentration of N20 has increased from about 270 nmol mo1-1 in preindustrial times (Kroeze et al. 1999) to 314 nmol mo1-1 today {CMDL 2001). There is some evidence for small variations in growth rates in the early 1990s, but during the period of precise in situ measurements growth rates have remained near con stant at around o.8 nmol mo1-1 yr-1 in both hemispheres (CMDL 2001). The global N20 flux from the ocean to the atmosphere has been calculated based on more than 60 ooo field measurements of the partial pressure of N20 in surface water ( Fig. 2.6). These data were extrapo lated globally and coupled with air-sea gas transfer co efficients estimated on a daily basis (Nevison et al.1995).

-9 -3
Quantifying the wide range of N20 sources has proved difficult and upper and lower bound estimates for specific source types can dif f er by a factor of ten (Ehhalt et al. 2001). However, progress has been made in balancing the sourcesink budget and its recent evo lution has been reviewed by Kroeze et al. (1999).
The radiative forcing of climate due to increases in CH4 (see above) and N20 during the industrial era is about 25% of the total due to all well mixed greenhouse gases (Ramaswamy et al. 2001). In addition both gases play a significant role in atmospheric chemistry. In creases in CH4 tend to decrease atmospheric oxidation rates {e.g. Thompson and Cicerone 1986), but increase 03 and stratospheric H20 levels. The result of these indi rect effects is to amplify the radiative forcing due to CH4 emissions by around 70%. Changes in concentrations of N20 over time have tended to decrease stratospheric 03 (Crutzen 1979) but this effect is small (see Chap. 3).
Nitric oxide has a short lifetime {approximately one day) in the atmosphere and takes part there in a com plex cycle of reactions with CO and hydrocarbons to form tropospheric ozone. Total emissions, both natural and anthropogenic, range from 37 to 59 Tg N yr-1 (Graedel and Crutzen 1993). Estimates in the 1980s of global annual emissions of NO from soils, its largest natural source, were ca. 8 Tg N yr -1; these have been re vised using an extended data set and are now estimated to be as large as 21 Tg NO -N yr·1 with an error term of  1997). Minor sources include lightning, transport from the stratosp!here, biomass burning, and aircraft emis sions.

The Sulphur Family:
Dimethylsulphide and Carbonyl Sulphide Sulphur containing gases are major participants in gas to particle conversion (see Chap. 4). Anthropogenic sul · phur emissions from fossil fuel oxidised to sulphate particles can act, in addition to sea salt particles, as con densation nuclei for marine clouds (see Chap. 4). Natu ral biological emissions of sulphur are predominantly marine in origin, with minor emissions from volcanoes. Dimethylsulphide (DMS), which is produced by micro bial processes in the ocean, is emitted at the rate of 15-30 Tg S yr-1 {Bates et al. 1992). A recent global inven tory of DMS emissions to the atmosphere has been cre ated using the data from more than 16 ooo observations of surface ocean DMS concentrations (Kettle et al.1999) (Fig. 2.7). The estimates ofDMS emitted from the ocean to the atmosphere are constrained largely due to the increased number of field observations and mass bal - ance of the sulphur budget in the marine boundary layer (Chen et al. 1999;Davis et al. 1999 ). Carbonyl sulphide (COS) in the atmosphere origi nates predominantly from the outgassing of the upper ocean (30%), atmospheric oxidation of carbon disul phide (unknown), and biomass burning (20%), with a total emission of about i Tg S yr-1 (Andreae and Crutzen 1997; Chin and Davis 1993). With the longest tropo spheric lifetime of all atmospheric sulphur compounds, COS can reach the stratosphere where it is oxidised to sulphate particles, which may impact the radiation budget of Earth's surface (Crutzen 1976) and influence the stratospheric ozone cycle.

A Paleoclimatic Perspective on CH4 and OMS
Information on past concentrations of several trace gases is preserved in air bubbles trapped when snow is pro gressively buried and compacted to form ice in areas of Greenland and the Antarctic where temperatures are cold enough to prevent surface melting. The archived air preserved in this way has provided reliable estimates of changes in atmospheric CH4 and N20 for up to 400 ooo years in the past.
Methane concentration changes are now well de picted in both hemispheres and vary from about 350 nmol mo1 -1 for glacial to about 700 nmol mo1-1 for longitude interglacial climatic conditions (Stauffer et al. 1988;Raynaud et al. 1988;Chappellaz et al. 1990). Significant rapid CH4 changes are associated with nearly all abrupt climatic changes that affected the northern hemisphere over the last ice age ( Chappellaz et al. 1990( Chappellaz et al. , 1993Brook et al. 1996), indicating a very tight response of the natu ral CH4 cycle to climate fluctuations.
The Holocene record (11500 B.P. to present) provides the natural atmospheric CH4 variability in relatively sta ble al] conditions (Blunier et al.1995;Chappellaz et al. 1997). The early Holocene (11500-9 ooo B.P.) i s a period of relatively high concentrations (720 nmol mo1-1), with a lower mean value (570 nmol mol-1 ) centred around 5 ooo B.P. and marked drops of 200 -year duration around 11300, 9 700, and 8 200 B.P. The mean inter-hemispheric difference of concentrations, which is mainly a function of the latitudinal distribution of sources and sinks, has been found to be 45 ±3 nmol mol-1 , i.e. markedly lower than the present-day di fference of ca. 140 nmol mo1-1 (Dlugokencky et al. 1994 For the last 50 years both concentration and isotopic data ( 1 3C/ 1 2C and 1 4 C/ 1 2 C) for CH4 are now becoming available from analyses of firn air samples (e.g. Francey et al. 1999). The concentration data indicate a pause in the increase of anthropogenic emissions during the period 1920-1945, probably due to a stabilisation of fossil fuel emissions at that time, while the isotopic data have placed constraints on the relative role of natural and anthropo genic sources and sinks in the 1978 to 1995 period. Paleo data from ice core studies have had a strong impact on our understanding of the global CH4 cycle, in particular the latitudinal distribution of wetland emissions. Changes in monsoon patterns (Chappellaz et al. 1990) and the distribution of northern mid-and highlatitude wetlands (Chappellaz et al.1993) have been considered. More recently Brook et al. (1996) favoured a boreal control on the CH4 global budget. Changes in methane removal rates must also be taken into account, and model calculations (Thompson 1992;Thompson et al. 1993;Crutzen and Briihl 1993;Martinerie et al. 1995) generally, though not unanimously, suggest that hy droxyl radical (OH) concentrations were higher in gla cial conditions than today. The consequent increased removal rate explains at most 30% of the reduction in concentration, implying that the larger effect is that due to lowered emissions.
Ice core data do not support a sudden release to the atmosphere of large amounts of CH4 from clathrate (hy drate) decomposition at the last deglaciation (Thorpe et al. 1996), as proposed by several authors (e.g. Paull et al. 1991;Nisbet 1992). However, more gradual release of CH4 from clathrates cannot be discounted as a po tentially significant factor and there is some isotopic evi dence for clathrate methane releases synchronous with reorganisation of ocean circulation (Kennett et al. 2000).
Global DMS emissions may be modulated by climatic conditions. Could global warming trigger a change of marine biogenic activity and consequently of DMS emis sions? Human-induced atmospheric changes could also disturb the oxidation processes of DMS and modify the branching ratio between methanesulphonic acid (MSA) and non-sea salt (nss) S04 formation. Ice core studies may help to elucidate these questions, provided that DMS or at least a OMS-related compound is recorded in polar ice. In this regard, MSA has been considered as the most promising parameter to determine in polar ice cores. Over the last decade, a few firn and ice cores have been analysed in detail for MSA and nssS04, in the hope of finding a correlation between concentrations in ice and cliniate fluctuations on various time scales. Some interesting results have been obtained, but glaciological phenomena have been pointed out recently that obscure the interpretation of the data.
At Antarctic locations where accumulation is rela tively high (>20 g cm·• yr·•), MSA concentration records seem to be reliable and decadal variations can be seen in shallow firn cores. In the Weddell Sea area, Pasteur et al. (1995) found from an ice core covering the last three centuries that MSA marine production increases at warmer temperatures, in relation probably to the amount of broken sea ice where phytoplankton can de velop favourably. MSA concentration in coastal Antarc tic snow seems to be linked with sea-ice extent (Welch et al. 1993). On the other hand, the validity of MSA ice records is questionable inland. A marked decreasing trend of MSA concentration was found in upper firn layers (the first 6 m) at Vostok (Wagnon et al. 1999). It is suggested that MSA scavenged in the snow crystals is progressively released from the solid phase by snow metamorphism. Part of the initially deposited MSA probably escapes back to the atmosphere. The profile obtained at Dome F (Dome-F Ice Core Research Group 1998) shows very low MSA concentrations between about 30 and 70 m depth, thereafter a rise from about 70 m up to no m. The effect can be attributed tentatively to the trapping of interstitial gaseous MSA in the air bubbles at the firn-ice transition (pore close-off). These observations, corroborated by MSA measurements at Byrd Station (West Antarctica) (Langway et al. 1994), lead to the conclusion that MSA concentration depth profiles from central Antarctica are most probably strongly aff ected by post-deposition phenomena. Sul phate records are not perturbed.
At Amun. dsen Scott Station (the South Pole), some decreasing trend of MSA concentrations with depth is observable in the firn layers, but it is less steep than at Vostok, probably related to the higher snow accumula tion rate. Interestingly, Legrand and Feniet-Saigne ( 1 991) detected marked spikes of MSA concentration in the upper 12 m of firn (i.e. over the last 60 years) at this site.
These were attributed to the impact of El Nifio events on the production rate of MSA in the sub-Antarctic ma rine areas or on its transport to inner Antarctica. The changes are superimposed on the general decreasing trend of MSA profiles found in the upper firn layers.
MSA records in Greenland firn cores over the last 200 years, on the other hand, show a rise starting from surface layers and lasting several decades (Whung et al. 1994;Legrand et al. 1997). This surprising trend, oppo site to what is found at the South Pole, could be attrib uted to a change in DMS marine productivity during this period or to the marked increase of atmospheric acidity caused by anthropogenic sulphur emissions. In the latter case, the amount of MSA remaining in the snow could depend on the pH of the atmosphere or of the snow.
Long-term changes in OMS-derived compounds can be seen in both Antarctica and Greenland records. The covariance of MSA and nssSO 4 concentrations observed in the Vostok core suggests that both compounds are mainly derived from marine DMS emissions. MSA and nssS04 concentrations are both higher in glacial condi-tions, with higher values of the ratio MSA I nssSO 4 found for ice ages. An increase of marine biogenic productiv ity has been put forward to explain this observation (Legrand et al. 1988(Legrand et al. , 1991(Legrand et al. , 1992, but the glaciological ar tefacts reported above for MSA records in central Ant arctic ffjrn layers cast some doubt on the proposition. Clearly more work has to be done on the understand ing of chemical composition changes of ice on the scale of several glaciations, all the more since Greenland data are contradictory to Antarctic observations. In the Renland ice core (East Greenland), MSA concentration and the MSA I nssS04 ratio are markedly lower for cold than for warm climatic stages (Hansson and Saltzman 1993). For the two deep cores recovered at Summit (GRIP and GISP 2), conclusions are similar (Saltzman et al. 1997;Legrand et al.1997). These observations suggest that, for the sulphur cycle, the cases of the northern and the southern hemispheres have to be discussed differently. In particular, the interaction of the primary aerosol (continental dust, sea salt) with acid sulphur compounds has to be investigated.

Atmospheric Compounds as Nutrients or Toxins
Deposition of atmospheric trace compounds can act as a significant source of nutrients or toxic substances to ecosystems, and their effects on these systems may in turn affect other trace atmospheric constituents. An example is natural fertilisation of the oceans by dust deposition, which leads to increased biological productivity, hence increased uptake of atmospheric C0 2 and release of DMS. The effect of dust deposi tion on community structure in certain marine systems is currently a key research topic among oceanogra phers.
Natural biogenic aerosol particles emitted by plants play an important role in nutrient cycling in tropical ecosystems. Many tropical systems are limited by nitro gen and phosphorus and depend on atmospheric input of certain mineral nutrients to maintain productivity (Vitousek and Sanford 1986). Work conducted in the Okavango Delta in southern Africa showed that in chan nel fringes water is the dominant source of nutrients but that in backswamps aerosols may provide as much as 50% of the phosphorus requirement of the ecosys tem (Garstang et al.1998). Sulphur emissions have been studied since the 1970s when their role in acid rain and forest die-back became key environmental issues (see, e.g. reviews by Sehmel 1980;Hosker and Lindenberg 1982;Voldner et al. 1986). Other acids (e.g. nitric acid) or anhydrides (e.g. sulphur dioffde) can also be depos ited in gaseous form.
Ozone is a significant greenhouse gas and in addi tion plays a major role in the atmospheric chemistry of both the troposphere and stratosphere (see Chap. 3). In the stratosphere its role in removing biologically dam aging UV radiation has received considerable attention. In the troposphere this gas is associated with negative impacts to human health and plant physiology and it can have signiffkcant negative impacts on plant produc tivity in polluted regions. Ozone damage occurs in most crop plants at concentrations of 0.05 to 0.3 µmo! mo1· 1 , with some more sensitive plants being affected at 0.01 µmo! mo1· 1 • Ozone directly affects the photosyn thetic processes, which results in decreases in plant yield (Tingey and Taylor 1982). As 03 has a short lifetime and is produced and consumed in the atmosphere, its con centration is highly variable both spatially and tempo rally. This makes accurate estimates of the total atmo spheric burden difficult and estimates of global scale trends even more so. Surface 03 measurements from background stations have shown both positive and nega tive trends of less than or about 1% yr·1 (e.g. Oltmans et al. 1998;Logan 1999). This complex picture may re fflect real re-distribution of 03 abundance due to changes in the emissions of precursors.

Approaches for Studying Exchange
A basic organising principle for understanding the fluxes of trace gases to and from the atmosphere is that of a source-sink !budget. For each compound, there is a mass balance between the fluxes into the atmosphere (sources), removals ffnom the atmosphere (sinks), includ ing chemical conversions and changes in the atmo spheric burden. Budgets provide the conceptual frame work for bringing together a process-based understand ing of surface exchange fluxes and atmospheric chem istry through demonstration of balanced source-sink budgets.
Exchanges of biogenic trace gases and particles be tween surfaces and the atmosphere are typically driven by the production and consumption of gases by plant, microbial, and chemical processes, and inffmuenced by physical traffsport through soils, sediments, water, or across gas-liquid boundaries.
For many chemical compounds, demonstrating a balanced budget based on process models of these fluxes remains a goal rather than a reality. However, substan tial progress has been made in the last decade through collaborations between a number of disciplines, includ ing atmospheric chemistry, ecology, biogeochemistry, geochemistry, microbiology, soil science, meteorology, hydrology, and oceanography. One of the hallmarks and great successes of IGAC research has been the integra tion of knowledge from such relevant disciplines toward the understanding of trace gas sources and sinks.
Understanding the source-sink budget for a trace gas involves establishing and validating process models across a range of scales. Most terrestrial process stud-ies of trace gas fluxes are carried out at small spatial scales, e.g. of the order of 1 m, in order to control the relevant environmental factors. Va lidation at this scale typically uses flux measurements derived from cham ber studies. However, process models are also increas ingly used as extrapolation tools to derive landscape, reg ional, and even global scale flux estimates. Most mod els can account for short term changes (minutes to hours) of some compounds but are limited in their abil ity to predict longer term (days to years) variations (Ot ter et al. 1999).
This up-scaling provides flux inventories that are rel evant for environmental management, but requires es timation of the key inputs to the process model such as marine plankton speciation, soil or vegetation type, land cover and management, and climatic, radiation, hydro logical, and marine parameters. Validation of these scaled-up inventories requires measurement of average fluxes at the corresponding scale. These may be deter mined by direct flux measurements near the surface, e.g. using eddy-covariance or relaxed eddy accumulation techniques, inferred from vertical gradients in the atmo spheric boundary layer, or derived from regional or glo bal scale transport models used in an inverse mode to calculate the flux di stribution that reproduces observed concentration distributions. Coupled land-ocean-atmo sphere models are only available for C02 and H20 with little attention being paid to other chemical compounds of biogenic origin. A few modelling studies have in cluded the effects of anthropogenic sulphur (Erickson et al. 1995;Meehl et al. 1996;Haywood et al. 1997 • To describe characteristics of soils that influence the area and depth distributions of production-con sumption reactions modulating trace gas emissions.
• To develop mechanistic models that in clude micro biological and physical-chemical processes applica ble at the scale of trace gas exchange experiments and to test these models with field and laboratory experiments.
• To develop ecosystem scale models for biogenic trace gas fluxes.
• To assess what quantitative changes in CH4, N20, and NO fluxes can be expected in response to physical and chemical climate changes.
A large number of studies has been conducted in the last ten years attempting to address these questions. Substantial progress has been made in both expanding the databases by conducting more measurements, and improving markedly the level of sophistication with which these measurements have been carried out (see Chap. 5), together with the way they are li nked to auxil iary data, e.g. isotopic data. Not only do we now have better databases but we also understand better the mechanistic processes and controlling factors regulat ing the fluxes. This has enabled adequate models to be formulated, although many of them are very li mited in their applicability (see Chap. 6). A significant part of the effort has come via the TRAGNET trace gas network, developed in the US with strong European participa tion, and the BATGE trace gas exchange programme centred on the Tropics. The following section sum marises the progress made in the last decade in the quan tification of the terrestrial sources and sinks of meth ane, volatile organic carbon compounds, and nitrous and nitric oxides, and advances in the understanding of the processes controlling their fluxes. Additional sections follow on biomass burning and wet deposition in the Tropics.

Background: Emissions and Deposition
Biogenic emissions from and deposition to vegetation and soils occur in a more or less continuous way over the year with the magnitude of the exchange controlled by a complex interaction of biotic and abiotic factors. On the other hand, biomass burning releases large amounts of emissions in pulses varying in frequency depending on the geographic location, the biome, and the management. The natural biogenic aerosol com prises many different types of particles, including pol len, spores, bacteria, algae, protozoa, fungi, fragments of leaves and insects, and excrement. The mechanisms of particle emission are still not well understood, but probably include mechanical abrasion by wind, biologi cal activity of microorganisms on plant surfaces and forest litter, and plant physiological processes such as transpi ration and guttation. Vegetation has long been recogni sed as an important source of both primary and secondary aerosol particles. Forest vegetation is the principal global source of atmospheric organic parti cles (Cachier et al. 1985) and tropical forests make a major contribution to airborne particle concentrations (Andreae and Crutzen 1997). However, only a few stud-ies of natural biogenic aerosols from vegetation in tropi cal rain forests have been undertaken (Artaxo et al. 1988(Artaxo et al. , 1990(Artaxo et al. , 1994Echalar et al. 1998).
Gaseous or particulate matter may be removed from the atmosphere and transferred to Earth's surface by various mechanisms, known under the generic terms of"dry deposition" and "wet deposition". Research find ings related to the latter in tropical systems are ad dressed specifically in Sect. 2.7.3. Dry deposition is the removal of particles or gases from the atmosphere through the delivery of mass to the surface by non-pre cipitation atmospheric processes and the subsequent chemical reaction with, or physical attachment to, veg etation, soil, or the built environment (Dolske and Gatz 1985). Dry deposition i s best described by the surface flux, F, corresponding to an amount of matter crossing a unit surface area per unit ti me. In most modelling work, an other quantity, called deposition velocity, v d = FI C (flux divided by concentration), is preferred for practical numerical reasons, because its time variations are smoother. Deposition velocities are also easier to para meterise and most data on dry deposition are actually expressed as deposition velocities, usually in cm s-1• A powerful parameterisation of dry deposition is the re sistance analogy (Chamberlain and Chadwick 1953}, where the difference between concentrations in the air and at the surface ( C,) is equal to the product of the flux and a resistance R, an empirical quantity to be para meterised. Through parameterisation of resistances, deposition velocities are readily derived. Further, this scheme may· be extended and adapted to the degree of complexity of the surface, e.g. as in the case of a forest canopy, by u.sing a greater number of resistances, i n se ries or in parallel, according to the rules of an electric circuit. The most powerful mechanism by which depo sition occurs over a canopy is penetration into plant tis sues through the stomata. Although most pollutants undergo deposition only (downward alux), some of them show bidirectional fluxes. An illustration of such behaviour is the case of nitrogen oxides NO and N02, as shown by Delany et al.
(1986) and Wesely et al. (1989). Nitric oxide is emitted by soils (Williams et al. 1992;Wildt et al. 1997). Once emitted, it can readily be oxidised to nitrogen dioxide, with a resulting upward flux of the latter. If the concen tration of nitrogen dioxide is high, as in the case of pol luted air, its flux can be directed downwards. Contrary to ni trogen oxides, ozone undergoes depo sition only, since there is no known process which could produce ozone at the surface. The deposition velocity of ozone depends mostly on the nature of the surface. If vegetation is present, ozone is deposited ("taken up" would be a more appropriate term) by penetration into plant tissues through the stomata! cavities present on leaves. This process is likely to cause damage, and, in extreme cases, decreases i n crop yields. Ozone uptake by vegetation has been put forward to explain ozone downward fluxes by Rich et al. (1970), andTurner et al. (1974), and subsequently by many other authors. Another ozone deposition mechanism occurs on bare soils, where ozone molecules are destroyed by a process similar to wall reactions observed in the laboratory, e.g. in a glass vessel. On mixed surfaces, both processes occur. Dry deposition of ozone has been extensively studied over the last forty years (Regener 1957;Galbally 1971;Galbally and Roy 1980;Wesely et al. 1978Wesely et al. , 1982Delany et al. 1986;Guesten and Heinrich 1996;Labatut 1997;Cieslik 1998). Resistance analysis has been applied to the interpre tation of ozone flux observations (Massman 1993;Padro 1996;Cieslik andLabatut 1996, 1997;Sun and Massman 1999 ), in particular to discriminate between the relative contributions of stomata! uptake and direct deposition on soil to the overall process. Most authors used an ap proach in which stomata! resistance for ozone uptake was deduced from stomata! resistance for evaporation, since both processes depend on stomata! aperture. Com bining direct ozone deposition measurements and the inferred ozone stomata! resistance, its partial resistance for deposition on the soil was deduced as a residual.
The diurnal pattern of ozone deposition is governed by both turbulence and physiological activity of the vegeta tion. At night, ozone deposition is close to zero. It in creases during the morning hours, both because air tur bulence increases, bringing more molecules into contact with the surface, and because stomata are open for tran spiration and carbon assimilation. The noon maximum value of deposition velocity ranges between 0.2 and o.8 cm s-1, depending on the intensity of turbulence and on the state of vegetation: the more active the vegetation, the more ozone is taken up. The daily variation in ozone deposition generally follows the pattern of the surface heat flux. For example, rapid deposition of ozone was observed in the lowest layers of a tropical forest canopy in Brazil, with an average flux of -5.6 ±2.5 x 10 11 molecules cm-2 s-1• This co-occurred with a large NO flux of 5.2 ±1.7 x 1010 molecules cm-2 s-1, which was about three times larger than the flux of N 2 0. The rapid destruction of 03 in the forest environment was also manifested by a pronounced ozone deficit in the atmospheric boundary layer. Rapid removal by the forest clearly plays a role in the regional ozone balance, and, potentially, i n the global ozone bal ance. The location of strong NO sources and sinks in the humid Tropics makes these ecosystems pivotal in the chemistry of the atmosphere (Kaplan et al. 1988).

7 .1 Production and Consumption of CH4
The state of understanding of the CH4 budget in 1990 was well summarised by Fung et al. (1991) who showed that the observed seasonal cycles at sites remote from sources could be reproduced using estimates of sources and removal rates consistent with the literature at that time. However, large uncerta inties in ind ividual com ponents of the budget were evident and the atmospheric global observation network did not provide sufficiently strong constraints to reduce these uncertainties.
Since 1990 considerable progress has been made, particularly through studies of CH4 emissions from wetlands and rice paddies, but also through improved estimates of oxidation rates, better data on animal and landfill emissions, and extension of the observational network. One significant outcome of these studies has been to decrease estimates of rice emissions and in crease estimates of natural wetland emissions. At the regional scale there has been a reduction in the uncer tainty of some type of emissions, e.g. from ruminant ani mals, with some stud ies being prompted by requ ire ments to report national greenhouse gas emission in ventories under the United Nations Framework Conven tion for C li mate Change (UNFCCC). Early successes of IGAC included a systematic char acterisation of CH4 fluxes from wetlands obtained from field programs in the ABLE, BOREAS, and related projects.
This area has received considerable attention by many groups during the last decade; although a comprehen sive literatur·e review i s beyond the scope of this chapter, a partial summary follows. A summary of parallel stud ies of CH4 from rice paddies is given separately (see Box 2.1). Dependencies of CH4 production rates in wetlands and closely related systems on water table depth, tem perature, and precipitation, were examined and used to develop regression-based explanatory models (e.g. Wahlen and Reeburgh 1992;Roulet et al. 1993;Frolking and Crill 1994). Consumption by methanotrophic communities, which may intercept a substantial fraction of below ground production, was also quantified in a variety of situations and related to environmental variables (e.g. Wahlen et al. 1992;Koschorreck and Conrad 1993;Bender and Conrad 1994).
As the available data grew, the value of organising them in terms of latitudinal transects and of using consistent methodologies and reporting formats was recognised. The US Trace Gas network (TRAGNET), a component of the IGAC BATREX project, was established to meet this need (Ojima et al. 2000) and has created a database of flux measurements covering 29 sites ranging largely, but not exclusively, from 10° N to 68° N on the American conti nent (see http://www.nrel.co/ostate.edulprojects/tragnet/). These flux data along with site and climate characteris tics are stimulating the development and validation of more sophisticated models. Recent wetland CH4 mod els have improved their ability to simulate observations by explicit treatment of net primary productivity as an underlying driver of production (Cao et al. 1996;Walter and Heimann 2000). More sophisticated models of soil oxidation processes have also been developed (Del Grosso et al. 2000) and comparison of models across a .------, Box 2.1 . Case study: Methane emissions from rice IGAC researchers have been very active in studying methane em issions from rice paddies and considering mitigation options (RICE Activity). This is particularly relevant given projections that rice production will in crease from 520 million t today up to 1 billion I during this century. R ice agriculture is subdivi d able into dryland, rainfed, deepwater, and wet-paddy produc tion. The latter three categories have land continuously under water at some ti me of the year, creating anoxic conditions. They comprise some 50% of the rice crop area and contribute 70% of total rice production (Minami and Neue 1994).
Emissions from rice fields are influenced by many factors, of which the most i mportant are water management, the amount of decomposable organic matter (e.g. rice straw) inc o r p orated into the soil, and the cullivar of rice grown (Neue i997). Other factors such as temperature, soil redox potential, soil pH, and the type and amount of mineral fertiliser applied also affect the emission, which reflects a net balance between gross pro· duction and microbial oxidation in the rhizosphere.
Substantial CH4 em iss ions occur only during those parts of the cultivation period when rice paddies are flooded, although a delay of typically two weeks occurs after floodi ng. The main control of CH4 production is the availabi lity of degradable or ganic substrates (Yao and Conrad i999). Readily mineralisable carbon, e.g. in rice straw or green manure, produces more CH4 per unit carbon than humified substrates like compost (Van der Gon and Neue 1995). Higher soil temperature also •peeds up the initiat ion of CH, formation but not necessarily the total emitted over a growing season.
Earlier estimates that up to 80-90% of the CH4 produced in a paddy field is oxidised (e.g. Sass and Fisher 1995) may be too high. Th e use of a novel gaseous inhibitor, difluoromethane, which is specific for CH1 oxid ising bacteria in rice fields and which does not affect the CH4 -producing bacteri a, showed that CH4 oxidation was im portant only dur i n g a rather short perio d of time at the beginning of the season, when ca. 40% of the CH4 produced was oxidised before it could enter the atmo sphere. This fraction then decreased rapidly and for most of the season the CH4 ox.idation was only of minor importance (Kriiger et al. 2000). There is now evidence of a nitrogen limitation of the oxidation process (Bodelier et al. 2000). There is also evi dence of systematic changes during the rice-growing season i n the 613C value of emitted CH4 due to changes in production, transport, and oxidation (Tyler et al. 1994;Bergamasch i 1997). This may have an impact on the 613C signal ofatmospheric CH4, which is relevant for inverse mod elling of methane sources. As understanding of the CH4 budget has improved, attention has turned to explaining interannual variabil ity and, in particular, the high growth rates observed in 1991 and 1998, which appear to be associated with anomalous climatic conditions. A key factor in this re spect has been the development of better process mod els for wetland emissions outlined above. Up to 90% of the CH4 emitted from rice fields passes through the rice plant. Well-developed int racell ular air spaces (aerenchyma) in leaf blades, Leaf sheaths, culm, and roots provide a transport system for the conduction of CH4 from the bulk soil into the atmosphere (Nouchi et al. 1990). Modern cultivars emit gener ally less per plant than traditional varieties because the im proved harvest index often results in less unproductive tiller, root biomass, and root exudates (Neue 1997). Work in China (Lin 1993) and the US (Huang et al. 1997) has demonstrated a twofold diff erence in emission rates between rice varieties grown under similar conditio ns. However, under field cond i t i ons, a comparison of cultivars is more complex because farm ers adjust planti n g densities or seed rates to achieve an opti m"m canopy and tiller density.
Existing model approaches are still crude, with low resolu tion, but they provide good regional estimates within the range in soils. Experimental field and laboratory data from five Asian countries participating i n the Inter-regional Research Program were used to devdop, parameterise, and lest the model. !'ield measurements of CH, emissi ons were extrapolated to national levels for various crop management scenarios using spatial databases of required i n puts on a province-distri ct level. Lack of geographic information on required inputs at appropriate scales limits application of this model i n determining current, and pred icting future, source strengths.
Promising candidates for mitigation of rice emi ssions are changes in water management, organic amendments, fertilisa tion, cultural practices, and rice cultivars (Neue et al. i998). How ever, while present knowledge of processes controlling fluxes al lows the development of m it igation technologies, information is still Jacking on trade-offs and socio-econom ic feasibilities. Cli mate change will tend to extend rice produc tion northwards, es pecially in Japan and China. Elevated C02 concentrations will enhance the production of rice yields, but also increase carbon exudation from roots, enhancing CH, emissions. Bre eding of new rice cultivars will be the most effective strategy for dealing with this issue (Milich 1999). However, enhanced temperatures are likely to limit the potential increases. Most CH4 removal occurs through atmospheric oxi dation by OH; however, consumption by methanotrophic bacterial communities in soils is estimated to be respon sible for 3 to 6% of total removals. This process is also important as it i s responsible for reducing the net emis sions from soils, e.g. those from rice paddies, landfills, and natural wetlands, through consumption in aerobic conditions near the surface. Thus, changes in water ta ble can shift the balance between CH4 production and consumption in soils.
At the outset of the IGAC programme, there were few data available on the oxidation of atmospheric CH4 in soils and the total sink was estimated at 30 (range 15-45) Tg CH4 yr-1 (Watson et al. 1990). Now there are many more flux measurements available (including some from studies lasting more than one year), there is more information on the impact of land use change, and the relationships between oxidation rates and soil parameters have been modelled. Flux values for different ecosystems show consistent median values of 10-20 Mg CH4 m-2yr-1 , but with skewed (log-normal) distributions (Smith et al. 2000).A major reason for the similarities between different ecosystems is that the ef fect of temperature on oxidation rate is small, as the organisms responsible are substrate-limited due to di ffusion resistance and low atmospheric concentra tion. This analysis gives a global sink of 29 Tg CH4 yr-1 , and a ±iarange from a 1 4 to 4 times this value (Smith et al. 2000 ). Thus, the best estimate is essenti ally un changed but the uncertainty is increased. Global esti mates from models range from 17-23 Tg CH4 yr-1 (Pot ter et al. 1996a) to 38 (range 20-51) Tg CH4 yr-1 (Ridgwell et al. 1999 ).
Changes in land use between natural grassland, pas ture or arable land, and forestry can produce a large rela tive difference in CH4 removal rates in soils (Smith et al. 2000;Del Grosso et al. 2000). Recent data have shown that methane-oxidisi ng bacteria associated with the roots of rice are sti mu lated by fertilisation rather than inhibited, as had been generally believed; these data will make a re-evaluation of the link between fertiliser use and methane emissions necessary. The impact of dis turbance on oxidation rate can be long lasting, e.g. it may take ioo years or more to recover (Prieme et al. 1997;Fig. 2.8), but nothing is known about the ecological rea sons for this. There is evidence that the microorgani sms principally responsible for CH4 oxidation differ from those responsible for CH4 oxidation in environments such as landfill cover soils, wetland hummocks, termite mounds, and oxidised zones within rice paddy soils, where much higher gas concentrations are the norm (Conrad 1996).
Ongoing climate change i s expected to increase tem peratures and thaw depth in tundra ecosystems, which will tend to increase methane emissions. On the other hand, increased evaporation at the surface may create an oxygenated zone, increasing methanotrophic activ ity. However, climate models also indicate an i n crease in precipitation for northern latitudes (Vourlitis et al. 1993) suggesting that wet tundra soils will continue to be waterlogged and that the temperature effect will dominate. An indication of the potential for climate feedbacks on CH4 wetland emissions is given by recent analyses suggesting that higher CH4 growth rate in the atmo sphere during 1998 can be explained by temperature and precipitation effects on wetlands (Dlugokencky et al. 2000;Walter and Matthews 2000 ). Extrapolation of these results has suggested that a global mean warming of 1 °C would lead to an increase in wetland emissions of 20 to 40 Tg yr-1 (B. Walter, private communication 2000).
Dlugokencky et al. (1998) showed that the overall decline in CH4 growth rates during the 1990s was broadly consistent with a constant total emissions and removal rate and that CH4 concentrations would stabi lise at a level 4% hi gher than observed in 1996 if thi s situation conti nued. Alternati vely, CH4 could be stabi lised at 1996 concentrations if the total emissions were reduced by 4%. However, there is some evidence that re moval rates have been increasing during the last decade (Krol et al. 1998;Karlsdottir and Isaksen 2000) at ca. 0.5% yr-1• Thi s would imply that total sources were in creasing at about the same rate and is consistent with an analysis of trends in the 1 3C / 1 2 C ratios i n CH4 (Francey et al. 1999).
Longer term scenarios for CH4 emissions have not been studied in as much detaal® as those for eo 2 emi s sions. One perspective i s that emissions will generally follow human population because of the connection to agriculture, sewage, and landfill. However, recent trends indicate a decoupling of emissions from population (D. Etheridge, private communication 2001) and several authors have noted that anthropogenic CH4 emissions are generally associated with inadvertent losses of en ergy for both animals and fossil fuel use. These lead to an alternative view that abatement of current eH4 emi s sions may be possible at low or negative cost.

Compounds from Ve getation
Plant growth involves the uptake of C02, H20, and nu trients and the release of particles, water vapour, 0 2 , and reduced carbon compounds to the atmosphere. These reduced carbon compounds are usually described as Steinbrecher and Ziegler 1997). The achievements of IGAC and associated research activities during the last decade on VOC emissions from plants and the curr ently identified research gaps are discussed in the following section.

New Emission Measurements
The advances i n measurement techniques described in Chap. 5 have greatly increased capabilities for investi gating biogenic voe fluxes at multiple spatial and tem poral scales. The resulting data have provided a more complete and ac curate picture of biogenic VOC emis sions. New analytical methods have extended the range of chemical compounds that can be investigated. En closures coupled with env ironmental control systems have been used to characteri se the environmental and genetic controls over emissions, while above-canopy flux measurements provide an integrated measurement of landscape-level trace gas exchange. Tower-based flux measurements are particularly useful for investigating diurnal and seasonal variations without di sturbi ng the emission source. Aircraft and tethered balloon meas urement systems can be used to characterise fluxes over scales similar to those used in regional models. These regional measure me nts are especially useful in tropi cal landscapes with high plant species diversity.
The measurement database that can be used to char acterise biogenic emission processes and distributions has been greatly i n creased by large international field programs including EXPRESSO, LBA, SAFARI, NARSTO/ SOS, EC-BEMA, Ee-BIPHOREP, EC-EUSTAeH (see . Over a thousand plant species have been investigated, for at least a few voes, by these studies.
Equally important has been the large number of land scapes that have been studied. IGAe-endorsed research has been particularly important for advancing meas urements in tropical regions. A number of these studies have investigated emissions on multiple scales result ing in measurements that can be used to evaluate bio genic voe emission model estimates.
Investigations of emission mechanisms, often con ducted under controlled laboratory conditions, have also added to our understanding ofbiogenic voe emissions.
These measurements have been used to relate emissions to both environmental and genetic controls. Although these measurements have not revealed distinct taxo nomic relationships, some patterns have emerged (Harley et al. 1999; esiky and Seufert 1999).

.3 Newly Identified Compounds
A substantial improvement has been achieved in the last ten years in the identification and accurate quantifica tion of VOCs emitted by terrestrial ecosystems. The number of components reported as biogenic voe emis sions has increased from seven (ethylene, isoprene, a-pinene, j3-pinene, limonene, �-3-carene, and p-cymene) to more than 50, belonging to ten different classes. The list of detected compounds is reported in Table 2.1 to gether with information on: • hypothesised biological production pathways occur ring inside and outside the chloroplast; • numerical algorithms adopted for describi n g emis sion vari ations; • relative abundance i n vegetation emission; and • degree of removal by OH and 03 attack under cer tain atmospheric conditions.

1.2
The increased knowledge of voe composition can be attributed to the recent measurement campaigns and analytical advances described in the previous section. Some compounds were not previously observed because of their reactivity with 03 during sampling and analy sis. The use of suitable 03 scrubbers has reduced this problem for some compounds.
Although impressive, the list reported in Table 2.1 cannot be considered exhaustive of the potential voes that could be released by vegetation because of the influence that mechanical injuries, pathogen attack, ozone exposure, and natural decomposition can have on emissions. Such effects can induce the release of or ganic compounds that are not normally emitted by the plant. In addition, some compounds have a li mited dis tribution within the plant kingdom and may be pro duced by plant species that have not yet been investi gated.

Production and Emission
At least seven different biogenic VOe production-emis sion categories have been identified: chloroplast, meta bolic by-products, decaying and drying vegetation, spe cialised defence, unspecialised defence, plant growth hormones, and floral scents. Recent studies have shown that some compounds can be produced by more than one pathway (Table 2.1).There have been significant ad vancements in elucidating the biochemical pathways responsible for voe production in chloroplasts. These include the identification of the enzymes associated with the synthesis of these voes, the chemical precursors, the production sites, and the demonstration that com pounds other than isoprene (e.g. 3-methyl-3-buten-1-ol and a-pinene) are emitted in this manner (Silver and Fall 1995;Loreto et al. 1996). It has also been observed that these voes penetrate the intercellular space of the leaf and exit the pl ant via the stomata; yet emissions are not directly controlled by stomata! conductance. Instead, emissions depend on the rate of synthesis in the plant, which is coupled to the availability of precur sors.
Some oxygenated voes (e.g. acetone, methanol, formaldehyde) could be produced as metabolic by products and, although there has been little experi mental investigation, plausible pathways have been proposed (Fall 1999). The production pathways asso ciated with the remaining categories (floral scents, growth hormones, and specialised and unspecialised defence) have been studied primarily due to their bio logical importance. The production mechanisms for voe emissions associated with these four categories have been described and are summarised by Guenther et al. ( 2000 ).

.1.3.2 Environmental Influences on Emissions
A quantitative description of the environmental factors controlling biogenic emissions is needed for predicting regional emissions and how they might change. The progress made for each individual voe is shown in Ta ble 2.1. However, since each voe can be emitted by more than one process, it is more convenient to discuss the environmental controls associated with the seven emis sion categories mentioned above.
IGAe-related research has resulted in significant ad vancements in descriptions of the factors controlling voe emi ssion, particularly isoprene, from chloroplasts. This has included the development of numerical algo rithms that accurately describe short-term variations  flux density (P PFD) on isoprene emission activity factors predic ted by the algorithm described by Guenther et al (2000). Tempera ture during the past 15 mi n (TM) and tempe rature dur i ng the past 15 days (TD) both influence isoprene em ission. LAI is the cumula tive leaf area index above a point in the canopy an improved physiological and biochemical understand ing is needed (Schnitzler et al. 1997;Guenther et al. 2000 ). It is known that the light response of shade leaves differs considerably from sun leaves and that the tem perature that the plant has been exposed to in the past can influence its temperature dependence. This general understanding is illustrated by the response curves shown in Fig. 2.9. Emission of stored monoterpenes from specialised defence structures results from the diffusion of com pounds through the cell barrier around the resin ves sels or ducts. The amount released by this process at a given temperature is dependent on the nature of the compound, resistance properties of the cell layers, and other transport resistances within and outside of the leaf. As cell resistances and vapour pressure of the com pound are temperature dependent, the emission source strength is strongly dependent on the temperature of the leaf. Recent advancements have been made in de scribing how the exponential increase in emissions with temperature is dependent on voe compounds and the resistance properties of the plant.
The environmental controls over metabolic by-prod ucts, decaying and drying vegetation, plant hormones, floral scents, and unspecialised defence have not been characterised in a manner that is useful for emission modelling. Some of the primary controlling factors are known but there are no quantitative algorithms for simulating emission variations (Fall 1999;Kirstine et al. 1998;Guenther 1999). A significant obstacle to regional scale extrapolation of these emissions is the need for databases of driving variables, e.g. temperature and ir radiance intensity, which are currently unavailable.

.1.3.3 Production and Loss Mechanisms in the Plant Canopy
Field experiments carried out within the frame of the EC-BEMA project (Valentini et al. 1997;Ciccioli et al. 1999) have shown that within-canopy losses are signifi cant for voes that have atmospheric lifetimes compa rable to the transport time from the canopy to the at mospheric boundary layer (ABL). Compounds with at mospheric lifetimes ranging between one and three minutes (ranked as -2 in Table 2.1) never reach the ABL whereas severe losses (>50%) are observed for com pounds characterised by atmospheric lifetimes ranging between three and ten minutes (ranked as -1 in Table 2.1) (Ciccioli et al. 1999).
In addition to gas-phase reactions, adsorption and partition processes can also play an important role i n removing emitted components inside the forest canopy. These effects can be particularly important in the case of polar compounds (such as alcohols and carboxylic acids) that are three orders of magnitude more soluble than isoprenoids in water droplets and stick on parti-des and surfaces. They have been invoked to explain the reduced flux of linalool from pine-oak forests and orange orchards.
Degradation of VO Cs in the canopy may lead to the formation of secondary organic aerosols (SOA), as men tioned earlier (see also Chap. 4), or gaseous products (mainly very volatile carbonyls) that can diffuse in the ABL. Photochemical degradation of VO Cs has been sus pected to be the main source of the huge fluxes of acetal dehyde, formaldehyde, and, partly, acetone from orange orchards. A substantial contribution to carbonyl fluxes can also arise from heterogeneous ozonolysis of lipids coveri ng the leaf surface (Fruekilde et al. 1998), which can produce acetone, 6-methyl-5-hepten-2-one, and geranyl acetone as a function of the levels of ozone i n air. Mesoscale modelling studies applied to a region north of Valencia, Spain (Thunis and Cuvelier 2000) have shown that secondary products formed by withi n canopy reactions accounted for more than 70% of the ozone formed by biogenic emission from orange or chards. The complexity of within-canopy processes oc curring in certain ecosystems can only be assessed by incorporating chemical processes into models describ ing the transport of VO Cs into the ABL. At the present time, development of such models is made difficult by the fact that degradation pathways of primary products formed by photochemical reactions of mono-and sesqu iterpenes are still unknown and it is not clear to what extent and in which conditions they can possibly nucleate to form SOA.

.3.4 Models of Emissions
The IGAC-GEIA natural VOC project has compiled and synthesised the available information on biogenic voe emissions and their driving variables into a global model that has been used to generate inputs for regional and global chemistry and transport models. The initial ef fort described by Guenther et al. (1995) provided monthly emissions of isoprene and three voe catego ries (monoterpenes, other reactive voe, less reactive VOC) with a spatial resolution of 0.5 degree of latitude/ longitude. The global distribution of isoprene emissions predicted by this model for the month of July is illus trated i n Fig. 2.10. The model was constructed using the following in formation. Emission factors were based on the results of 20 studies that were primarily located at temperate forest field sites. Two emission types were utilised: iso prene emissions were estimated using current light and temperature conditions while all other emissions were assumed to be dependent on current temperature. Landcover characteristics were primarily based on val ues assigned to landscape types and a global database of current landcover distributions. Satellite (AVHRR) measurements were used to estimate monthly foliar Global distribution of iso prene emi ssion rate estimates (g C m·2month-1) for July (Guenther et al. 1995) July lsoprene Em issions (g C m-2 month-1) The direction of this change will be controlled by the phenology of the vegetation and the geographic region.

30'
Savannah areas that may undergo bush encroachment would produce higher voe emissions with a different chemical signature.
Changes in air temperature, the length of the grow ing season, precipitation, and atmospheric C02 concen trations could lead to large changes in the voe emis sions from temperate latitude ecosystems. In tropical regions, little change may take place because plant leaves already may be near their optimal temperature for iso prene emissions. There are some suggestions that plants adapt to changing temperature regimes and that voe emissions would also rise in the Tropics. It is also un clear if cultivating increased areas of genetically modi fied plants could alter the nature ofVOC emissions sig nificantly.
Changes in vegetation type can lead to large changes in VOC emi ssions. For instance, replacing C3 grass spe cies with C4 types could change the direct emissions and emissions from the plant decay process. Current knowl edge is inadequate, however, to quantify these changes. Woody plants (shrubs and sun tolerant trees) tend to have much higher i s oprene and monoterpene emissions rates, compared to annual crops and grasses; therefore deforestation involving conversion of closed forest to grassland could greatly reduce biogenic voe emissions. However, there is a tendency for higher emissions from the woody plants that replace a closed canopy forest i997). The microbial breakdown of urea and uric acid present in animal waste produces ammonium, which subsequently partly volatilises as NH3• The overall emis sion of NH3 from waste is dependent on the specific N excretion per animal, and the NH3 loss during housing, storage of waste outside the stable, grazing, and appli cation of manure on grassland or arable land. Further important properties in fluencing NH3 volatilisation in volve soil pH and moisture, and temperature. There is a pH-dependent equilibrium between NH3 and NH4, with NH3 being emitted from soils when they are alkaline.
Some northern European countries have measured and calculated country-and animal-specific emission fac tors. The applications of such emission factors to calcu late animal related emissions elsewhere may be quite problematic, since agricultural practice and climatic factors may differ substantially from those in northern Europe. In addition, the number of animals per country may fluctuate strongly from year to year.
Dry deposition and reactions with acidic particles and particle precursor gases are the main removal mechanisms for NH3 (see also Sect. 2.7.3). Oxidation chemistry of NH3 is thought to play a relatively minor role (Dentener and Crutzen i994). Because NH3 emis sions occur almost exclusively close to Earth's surface, and because plants utilise nitrogen in their metabolism, dry depositj.on is a very efficient process that may re move 40-60% of all emissions. Ammonia that has re acted with sulphuric or nitric acid to form NHt i s re moved mainly by wet deposition and much less effi ciently by dry deposition. It has an average residence time in the atmosphere of up to one week, in contrast to the much shorter �esidence time of gas phase ammo nia, which is less than one day.
The understanding of the atmospheric NH3 cycle is still limited, because: • NH3 emissions are estimated for most countries, rather than measured because of the difficulty of making such measurements (Fehsenfeld i995).
• Emissions of NH3 show a large spatial and temporal variability (e.g. farm-scale, winter-summer). Trans port models ofNH3 and NHt utilise larger grid scales and the temporal variability of NH3 emissions is not accounted for in such models.
• Most models of NH3 chemistry and transport are highly si mplified and parameterised and may there fore produce results that may be spurious.
• Measurements of particulate NHt are common, al though their quality is frequently suspect; gas phase NH3 data are scarcely available.
The comparison of models with such measurements is difficult si nce the measurements may not be repre sentative for the model grid scale. Thi s is especially the case for gas phase NH3, which may have an atmospheric lifetime of hours. In addition, there may be substantial instrumental problems, e.g. the evaporation of ammo nium nitrate from filter packs, which makes it difficult to interpret routi nely performed aerosol measurements.
Wet deposition measurements of NHt are relatively straightforward, and there are, at least in Europe and the US, substantial data sets available. However, these meas urements alÖ again not be representative for a larger model region. Also, discrepancies of model results and measurements may be due to a host of reasons, such as a poor representation of dry deposition, vertical mixing, and in-and below-cloud scavenging or emissions. An evaluation of global model deposition with measure ments in Europe and the US by Holland et al. (1999) in deed indicated substantial discrepancies but no single process could be identified as the cause of the problem.
Despite all uncertainties involved, several studies (e.g. Galloway et al1 1995) have indicated the significance of terrestrial NH3 emissions for the global nitrogen cycle. Recent model li ng studies (Adams et al. 1999;Metzger et al. 1999) indicated the potential significance of NHt and nitrate (N03) for aerosol burden and composition. These studies were performed using thermodynamic equili brium models, developed originally for urban smog condi tions. Substantial effo rt has been spent on extending these schemes to global modelling (e.g. Nenes et al. 1998).
A major challenge i s presented by the need to i n crease the resolution of the models and develop sub-grid para meterisations that represent the variability of ammo nia emissions and the resulting effect on particle com position. Long-term, representative, and reliable meas urements of NH3, NH4, SO�-, and N03 are needed in conjunction with deposition measurements to constrain the NHt budget further.

.1 .5 Production and Consumption of N10 and NO in Soils
Nitrogen oxides are produced in soils as obligate inter mediates or by-products of the microbially mediated processes of nitrification and denitrification (Conrad 1996). The same environmental factors of soil tempera ture, nitrogen availability, and soil moisture affect the production of both nitrogen oxides. The pathways and enzymatic mechanisms of these processes were not well understood in the 1980s. The development of chemilu minescence instruments for NO measurement and re ports in the late 1970s that increased use of nitrogen fertilisers could be one of the main causes of accumula tion of nitrous oxide in the atmosphere, thus contribut ing to global warming, stimulated scientists to research the mechanisms involved (Firestone and Davidson 1989; Davidson and Kingerlee 1997). The conceptual model offered by Firestone and Davidson in 1989, which has since become known as the "hole-in-the-pipe" (HIP) model, synthesised the info rmation known at that time about the microbiological and ecological factors influ encing soil emissions of NO and N20. The HIP model linked the two gases through their common processes of microbial production and consumption. It was a breakthrough in understanding factors controlling emissions and the model has stood the test of time.
Ongoi ng testing of the model over the last decade with numerous data sets from temperate and tropical agro ecosystems shows that it provides a sound ecological and mechanistic basi s for interpreti ng temporal and spatial variation at all scales of study by neatly encapsulati ng into two functions -nitrogen availability and soil water content -a large fraction of the variability caused by numerous environmental factors that i n fluence the pro duction and consumption of NO and N20 by nitri fying and denitrifying bacteria (Davidson et al. 2ooob). Matson et al. (1989) stated that empirical models that are based on correlation analysis involving easily meas ured soil variables (e.g. temperature, moi sture, texture, and organic carbon) often predict trace gas fluxes quite well. As data sets became more available, this set of vari ables has been further reduced to moisture and tem perature with some corrections needed to take account of texture differences. Empirical relationships have been established for a number of different ecosystems around the world for both NO and N20 emissions with water filled pore space values of approximately 35% being the switch from NO to N20 emissions. The magnitude of the emissions varies with substrate availability; the use of 15N labelling techniques to measure the turnover of the soil ammonium and nitrate pools has greatly en hanced our capacity to partition nitrogen gas produc tion among NO, N20, and N2 • Trace gases are produced and consumed by defined reactions in individual micro organisms and control must be exerted at this level ini tially. To date, empirical models based on various physi cal and chemical parameters have been successful with out considering the structure of the microbial commu nity; even those models that differentiate between nitri fication and denitrification neglect microbi al commu nity structure. It is still unknown whether microbi al spe cies di versity is an important factor especially if one con siders changes associated with land use (Conrad 1996).
Soil mineral N, resulting from additions of synthetic N fertilisers and N from animal manure, crop residues, etc., and the mineralisation of soil organic matter and deposition from the atmosphere, is recognised as a ma jor driver of these emissions. Much work has gone into establishing the relationships between the fl uxes of N 2 0 and NO and ·the other key drivers, soil moisture and tern· perature. Although some questions remain to be an swered, significant developments in this di rection have been achieved. The logarithmic relationship between N20 flux and soil water-filled pore space (WFPS) in a tropical forest soil is illustrated in Fig. 2  Continuous flux measurements are few and far between, with many more studies of N20 than NO, mostly driven by the global warming community rather than the broader atmospheric community. Among the significant observations that have emerged from continuous N20 flux studies over extended periods are: (a) the great variation in annual flux (from 20 to >200 Mg N20-N m-2 yr-1) that can occur from fertilised temperate grassland as a result of variations in the tim i n g and amount of rainfall ( Fig. 2.12); (b) the large pro portion of the annual N20 that can be derived from soil freezing/thawing events in winter, both in agricultural soils (Flessa et al. 1995;Kaiser et al. 1998) and in forest soils (Papen and Butterbach-Bahl 1999); and (c) that ni trification can be as important as dentrification in pro ducing N20 (Davidson et al. 1993;Panek et al. 2000). In hot dry environments, variations in the time elapsed be tween fertilisation, sowing, and irrigation of cereal crops have been shown to be almost as important (Matson et al. 1998).   1999), and this input is likely to increase in the future. The expected effect of the add iti onal N use is a further major increase in N20 from agricultural sources. These increases in N fertiliser use are also expected to raise the agricultural con tribution to soil NO emissio ns to over 50% (Yienger and Levy 1995). In the IPCC (1999) assessment, direct emissions of N20 from agricultural soils were taken to be 1.25 ±1 % of the N ap plied (Bouwman 1996), but a re-evaluation in dicates that the observed emission factors are strongly skewed, giving an un certainty range from one-fifth to five times the mean value, i.e., from 0.25% to about 6% of the N applied, and suggesting that the mean flux from this source may be even higher or lower than presently accept ed. Recent data ( Fig. 2.13) on N20 and NO emis sions from N fertilised and N-saturated systems in tempente regions give indicat ions as to bow global change and changing land management practic es may be en h ancing emiss ions. How ever, Hall and Matson (1999), raise the possibility that increasing nitrogen deposition in tropical regions is likely to have very dif ferent effects than nitrogen depositi on in the temperate zone, with much greater feedbacks to the atmosphere (see Sect 2.7.3).

20
(a) Bahl 1999; Gasche and Papen 1999). At the bee ch site, 10% of the acrual N input was released from the soil in the form ofN20 whereas at the spruce site the fraction was 0.5%,indicating that forest type itself is an important modulator of N20 release from soil How ever, there is a marked simi l arity in the N20 data obtained from Scotland and from Germany. Approximately 15% and ]% of the actual N in put was lost as NO from the German soils stocke d with spruce and beech, respectively. Liming resulted in 49% reduction of NO emiss ions as compared to an unlimed spruce control site.
The results indicate that the reduction in NO emiss io n was due to an increase in NO consumption within the limed soil. Liming of a spruce site res'lllte d in a s ignificant increase in ammonifi cation, ni trification, and N20emissio ns as al with an untreated alġalĘalX control site. On the basis of these results it was concluded that the importance of temperate and boreal forests for the global Np sourc e strength may have been si gnificantly underestimated in the past and that these forests, in which N deposition is high, most likely contribute in excess oho Tg N20 -N and 0.3 Tg NOx-N yr-1 (Papen and Bunerbacll-Bahl 1999; Gasche and Papen 1999). in upland areas (grey-shaded circles), together with the IPCC default emissi on factor (1% of the N applied) (solid line) (Skiba and Smith 1999); c linear regression analysis between in situ N input by wet deposition and in situ N20 emission rates from soil of spruce and beech control sites. For correlation analysis, da ta for mean weekly N20 emission rates and mean weekly N input (data from Huben997) by wet deposition     years. An ecosystem approach to evaluating N20 fluxes is useful for regional and global modelling and for com putation of national N20 flux inventories for regulatory purposes, but only if measurement programmes are comprehensive and continuous. The investigation of the role of biomass burning in at mospheric chemistry was therefore seen as a high prior ity when the objectives of IGAC were formulated in 1988.
Since the WAC Biomass Burning Experiment (BIBEX) became active in 1990, research activity in this field has increased rapidly, and, over the last decade, fire has been recognised widely as a major source of important trace gases and aerosol particles to the global atmosphere. America, the tropical Atlantic, the Indian Ocean, and the Pacific Ocean. There is now also strong evidence that smoke aerosols perturb climate by scattering and ab sorbing sunlight and by influencing cloud microphysi cal processes.
We have also learned that the effects of burning are not limited to the emissions from the fires themselves, but that vegetation fires have pronounced effects on trace gas emissions from plants and soils. In the case of C02, NO, and N20, post-fire emissions may be more sig nificant than the immediate pyrogenic release. Fire also alters the long-term dynam ics of the cycling and stor age of elements within terrestrial ecosystems, thereby changing their potential as sources or sinks of various trace gases. Finally, deposition of pyrogenic compounds onto tropical ecosystems may affect their composition and dynamics. In the following sections, we review some of the results and attempt to put them into the larger context of global change research.

Scientific Approach
Since the early 1990s, BIBEX has designed and carried out a number of biomass burning experiments in vari ous ecosystems throughout the world, often in collabo ration with other international programmes, particu larly with other IGBP Core Projects. These experiments have produced extensive local-scale data on vegetation fire characteristics, emissions, and ecology, while simul taneous regional-scale measurements, using remote sensing and aircraft sampling platforms, have provided a capability to scale results up. Typically, these experi ments have involved ground measurements on indi vidual fires, airborne sampling and analysis of smoke plumes, and remote sensing of regional and global fire activity. Emphasis to date has been placed on tropical ecosystems, but an increasi ng number of experiments are now being organised in the boreal zone in response to climate change concerns. STARE (Southern Tropical Atlantic Regional Experi ment), with its two components SAFARI (Southern Af rica Fire-Atmosphere Research Initiative) (see Box 2 .3), and TRACE-A (Transport and Chemistry near the Equa tor) was the first large experiment coordinated by BIBEX. Conducted in 1992, STARE brought together sci entists from many countries to investigate the chemical composition, transport, and fate of fire emissions origi nating from South America and southern Africa. Tropical evergreen forest. Deforestation statistics by the FAO and others i n many studies have provided the base line data for calculation of pyrogenic emissions due to land use change. While these numbers are useful for es timating the net release of carbon to the atmosphere, they do not reflect the entire spectrum of fire activities. Re current fires following the initial deforestation burns not only present additional emission pulses but also lead to impoverishment of forest ecosystems resulting in re duced above-and below-ground phytomass (Goldammer 1999a;Nepstad et al. 1999). Extreme climate variability such as the ENSO-related droughts of 1982-1983 and 1997-1998  Tropical savannas and open seasonal forests. Assess ments made in the early 1990s on the average annual amount of savannah phytomass burned were in the range of 3-4 Pg yr-1 (Andreae 1993). Model predictions on the savannah area annually burned ranged between 750 x 10 1 0 m· 2 yr -1 (Hao et al. 1990) and 1500 x 1010 m·2yr -1 (Goldammer 1993). More detailed studies on fire re gimes and fuel loads in Africa poi nt towards lesser amounts of regional and global combustion of sa vannah phytomass (Menaut et al. 1991;Scholes et al0 1996). Recent and ongoing growth of rural populations and i n tensity of land use involves landscape fragmen tation and competitive utilisation of phytomass for grazing and domestic burning (biofuel use) and may represent a reason for a decrease of fire activities in tropical savannas and open forests; desertification in the sub -Saharan Sahel zone of Africa and other regions leads to a reduction and discontinuity of fuel loads and wild fire occurrence. The following example from the SAFARl-92 campaign (Lindesay et al. 1996) highlights the scientific approaches used to test by· potheses and validate models related to biogeni c and biomass burning emissions and depositions. It is approaches like these that have allowed for an integrated understanding of the magni· tude and controllers of sources, sinks, and exchange processes.
During SAFARI-92, experimental vegetation fires were set and studied in the Kruger National Park, South Africa, and at some si tes i n Zambia and Swa2iland. These experiments provided a broad set of data on trace gases and aerosol emissions, from which emission factors for fires in ald savannas and related biomes could be derived. The relationships between fuel characteristics, burning conditions, and fire behaviour were elucidated.
Regional studies on atmospheric chemistry and air mass trans port showed that savannah fires in southern Africa account for a substantial amount of photochemical oxidants and haze over the subcontinenL These studies also showed that the export of smoke· laden air masses contributes strongly to the burden of orone and other trace gases and aerosols over the tropical ocean surround· ing Africa. However, results also showed that biogenic soil emis si ons severely i m pacted atmospheri c cheinUtry. Investigations on the relationships among fire, soil moisture status, and soil trace gas emissions showed that soil moisture played a crucial role but that fire history also had an important influence on the emission of several trace gases. Remote sensing studies confirmed that Advanced Ve ry High Resolution Radiometery/Land Aeri al Cover (AVHRR/LAC, J km) imagery was a useful tool for fire monitoring in the region. In combination with biomass models, the remote sensing data could be used for the estimation of the seasonal and geographical dis· tribution of pyrogenic emissions. The results from SAFARI -92 confirmed that it is justified to consider biomass burning as a si gnificant contribut,or to the overall increase in greenhouse gases that has occurred over the last t50 years, accounting for some 10-25% of current estimates (Andreae 1993).
In order to establish accur ately the global budgets of trace gases, reliable source strength and distribution estimates are needed-At present, the uncertainties associated with budget cal culations arc necessarily large, owing to the often-inadequate quantification of individual sources and the problems associ ated with extrapolating from a number of poorly known sources to achieve a global estimate. The contribution of vegetation fires in the savannah regions of southern Africa bas been such a poorly quantified source, despite the fact that savanna s are recognised as one of the most significant biomes in terms of global bio mass burning emissions (Andreae 1993) and that a large portion of the savannah burns each year. It will now be possible to refine these estimate s on the basis of results obtained from SAFARI ·92.
Modelling studies incorpora ting the emission data, meteorologi cal information, and the chemical measurement data obtained during these campaigns indicate that the fires on the African and South American continents are indeed a major source of the gas-Areas of Mediterranean and temperate vegetation.

Mediterranean forest and shrub vegetation, including
Californian chaparral and South African fynbos, are in creasingly converted to suburban residential use-The consequent suppression of natural and human-caused wildfires results in a buildup of fuels that often cannot eous and particulate pollutants, particularly ozone. found in the troposphere OYer the srudy regions (Thompson et al. 1996a). Data from airborne observations (Fig. 2.15) aboard a DC-3 using a combination of spectrometers and chemiluminescence instru ments, showed! that episodic pyrogenic emissions were not ad equate to account for the buildup of tropospheri c ozone in the region but that the continuous productio n ofbi ogenic NO, emis sions and especially the amounts produced a1 the s1art of the rainy seasons 'have important consequences for regional scale owne formati on (Harri s et al. 1996). The vertical distribution of N02 and NO as well as that of C02 showed markedly diffe rent characteristics_ All three compounds have a strong gradient to ward hig her values near the ground, and the C02 and NO, mix· ing ratios correlated linearly. The anticorrelati on of the profiles of these compounds with that of CO rules out biomass burning as a source of the observed NO, and C02 near the ground, sup porting the field evidence of no active fires in the region. It was concluded that the source of the ele\ll' ted NO, mixing ratios near the surface was biogenic emission from the soil (Harris et al.1996).
SAFARl-92 was an innovative project in many ways. In addi· tion to being the largest internati onal, interdi sciplinary investi gation of biomass burning and its atmospheric emissions, it also represented the first time that a large-scale fire emission meas urement campaign included, as integral components: the char acteristics of the biomass, the fire ecology, the fire dynamics in the area, the biogenic emissions, and the long-range transport of the aerosols and part iculates.
As a follow-up to SAFARl-92, a much smaller experiment, SAFARI-94, was organised by BIBEX to investigate the composi tion of trace gases in the troposphere over Africa outside the burn al° sea son. EXPRESSO (Experiment for Regional Sources and Sinks of Oxidants). designed prim arily to investigate the excha nge fluxes of trace gases between the tropical biosphere and atmosphere, took the 1980s and 1990s. This is due to a growing popula tion and increased forest use, but also reflects an ex panded fire detection capability. During the 1981-1996 period an average of 9 246 fires annually burned over an average of 2.5 x io 1 0 m -2 in Canada, with the annual area burned fluctuating by an order of magnitude (0.76-7.28 x 1010 m-2). Lightning accounts for 35% of Canada's fires, yet these fires result in 85% of the total area burned, due to the fact that lightni ng fires occur randomly and therefore present access problems usually not associ ated wi th human-cau sed fires, with the end resu lt that light ning fires generally grow larger, and detection and con trol efforts are often delayed. In addition, the practice of "modified" or "selective" protection in remote reg ions of Canada results in many large fires in low-priority ar eas being allowed to perform their natural function. Summary assessment of trends in global vegetation fire occurrence. The trends of changing fire occurrence and fire regimes are not uniform. Qualitative and quan titative data on fire occurrence and fire effects are still insufficient to determine reliably the amount of phyto mass burned in all eco-and land-use systems world wide. However, improved remote sensing capabilities and rigorous fire detection algorithms now provide re gional fuel load and burning estimates within a much narrower range of uncertainty. Fire in boreal and tropi cal forests and the resulting ecolog ical effects play a po tentially critical role in determining the rate of global climate change (Goldammer and Price 1998;Stocks et al. 2000;Nepstad et al. 1999). Changes in the carbon bal ance of these two forest biomes could strongly influ ence global warming through impacts on atmospheric C02• The implications of regional circumpolar changes of climate and fi re regimes on boreal ecosystem prop erties, permafrost changes, and the release of gas and carbon stored in organic terrain and ice must be fur ther addressed by research.

Characterisation of Emissions
A central objective of BIBEX was to characterise and quantify the production of chemically and radiatively important gases and aerosol compounds from biomass burning. To meet this goal, the BIBEX scientific com munity has produced a large set of measurements that describe qualitatively and quantitatively the pyrogenic emission of gases and aerosols. The results show that the composition of fire emissions is mainly determined by two factors: the elemental composition (carbon, ni trogen, sulphur, halogen, minerals, etc.) of the biomass fuel, and the relative contribution of flaming and smoul dering combustion in the vegetation fires.
Heating of vegetation fuels produces combustible gases by pyrolysis and volatilisation of waxes, oils, etc.

Diffe rent Chemical Compounds from Fires in Various Ecological Systems or Ve getation Ty pes
To express the emission of trace gases and aerosols from fires quantitatively, we use the concept of emission ra tios and emission factors. These parameters relate the emission of a particular compound of in terest to that of a reference compound, such as C02 or CO (emission ratio), or to the amount of fuel burned (emission fac tor). Emission ratios are obtained by dividing the ex cess trace compound concentrations measured in a fire plume by the excess concentration of a simultaneously measured reference gas, such as C02 or CO. To obtain "excess" concentrations, the ambient background con centrations must be subtracted from the values meas-ured in the smoke. For example the emi ssion ratio of methyl chloride, CH3CI, relative to CO is:

�CO (CO)s moke -(C O) Ambient
The various techniques for these calculations and the associated errors are discussed in Le Canut et al. (1996). On the other hand, the emi ssion ratio relative to C02 permits the estimation of trace gas emission from fires based on the amount of biomass burned, because most of the biomass carbon is released as C02• Therefore this ratio is the most suitable for regional or global estima tions; however, it is worth noting that when multiple ratios are used to estimate emissions, errors are propa gated making overall estimates quite uncertain.
Another parameter frequently used to characterise emissions from fires is the emission factor, which is de fi ned as the amount of a compound released per amount of fuel consumed (g kg-1 dm; dm: dry matter ties persist for regional and global fire emissions because of the difficulties inherent in estimating the amount of biomass burned. In particular, there are differences of as much as an order of magnitude in regional estimates based on estimates of typical fire frequencies in the vari ous vegetation types, and those based on actual fire counts obtained from remote sensing. These issues will be discussed in more detail in Sect. 2.6.2.7.     All currently available estimates thus agree that biofuel use is a significant source for many a�mospheric trace The available database, i n particular the biofuel con sumption figures, for biofuel emissions should be im proved further, even if extrapolations from spot assess ments will remain necessary. Most measurements have been made i n Africa. However, cooking and heating hab its vary considerably between different developing re gions of the world. Add itionally, agricultural waste burn ing and, even more, smoulderi ng dump sites are not yet characteri sed adequately, but are expected to contrib ute significantly to global emi ssions.

.Detection of Fires and Burned Area by Remote Sensing
Reporting of national estimates of anthropogenic trace gas emissions, including those from biomass burning, are a requirement of the Framework Convention on Cli mate Change, and the IPCC provides guidelines for these emissions calculations (Callander i995). For many parts of the world !however, national emissions estimates from biomass burning are based largely on expert opinion or summary statistics, and the resulting accuracies are largely unknown. Synoptic fire information derived from satellites provides a source of information for aug menting available national fi re statistics. Satellite detec tion of active fire occurrence has been used to identify the timing and location of fires, and has been used i n emission product transport studies, for example. Polar orbiting and geostationary satellite systems have been used to provide fire information (Elvidge et al. 1996;French et al. 1996;Prins et al. 1998). The first global data set of annual satellite fire distributions was developed directly as a contribution to BIBEX (Stroppiana et al. 2000).
Automated algorithms for direct estimation of burned area are currently under development with the intent of providing direct input to emissions modelling (Roy et al. 1999 ). Satellite based techniques for direct estimation of emitted energy, fire intensity, atmospheric aerosol loading, and vegetation recovery are also bei ng developed. Since in most cases the data products are to be used in numerical modelling, there is a need to pro vide a quantitative assessment of their accuracy. For satellite products, validation using independent data sources needs to be undertaken to determine product accuracy.
New satellite systems are planned that will improve our current fire monitoring capability (e.g. Kaufman et al. i998b ). The requirements for these systems come in part from the experience gained from BIBEX. The satellite fire research community is working to secure the necessary long-term fi re observations from the next generation of operational satellite systems, such as the US National Polar Orbiting Envi ronmental Satellite Sys tem (NPOESS).
With the operational availability of satellite-derived information on the location and timing of fires and on the area burned, it will be feasible to run an improved class of models to estimate emissions on an annual ba sis. These improved models will requi re ground-based estimates of emission factors and modelled estimates of fuel load and fuel consumed for a given year, rather than representative values for a given vegetation type. As new satellite i n formation becomes available on fire intensity, emitted energy, and fuel moisture content, these first order emissions estimates can be improved. Providing robust models that can be used for opera tional generation of annual emissions estimates and developing approaches to validate them provide the next challenge for the fire and global change research com munity.

Impacts of Burning on Trace Gas Exchange from Soils
The process of biomass burning represents a vast real location of nutrients in cleared tropical forest and sa-vannah systems. Large proportions of system carbon, nitrogen, and sulphur are volatilised. Soils are affected by changes in nutrient levels, pH, and temperature, with associated changes in microbial communities. Studies conducted during the SAFARI 92 campaign showed that the mean NO emissions increased after burning, reaching 15 ng N m -2 s -1 from dry sites and exceeding 60 ng N m·2 s-1 from the wetted sites (Levine et al. 1996). The long-term eff ect of excluding fire from a savannah is to increase the soil nitrogen content through increased litter inputs, which in turn increases nitrification rates and soil NO emissions (Parsons et al. 1996). Soil emissions of C02 and CO were increased by an order of magnitude after burning, whereas exchange of CH4 was not affected. In all cases the increases were short lived and dropped back to pre-burn levels within a few days (Zepp et al. 1996). Studies on the impact of burning on soil carbon pools showed that annual burn ing in a semi-arid savannah reduced the light-fraction carbon markedly but did not impact the intermediate or passive carbon pools. This has implications for the amount of soil carbon that can be readily metabolised by the soil microorganisms. Burning the savannas at longer time intervals had no effect on the pool size or the turnover rates of the various soil carbon pools (Ot ter 1992).

Importance to Atmospheric Chemistry and Climate
We have already pointed out that biomass burning is a significant source of several greenhouse gases, among them C02, CH4, and, to a much lesser extent, N20. It also makes important contributions to the budget of several gases of stratospheric significance, such as methyl chlo ride and methyl bromide, N 2 0, and COS. Of particular importance to the chemistry and radiative characteri s tics of the atmosphere are the emissions of ozone pre cursors, particularly NOx, voe, CO, and CH4• Because vegetation fl.res in tropical regions can occur only when the vegetation is dry enough to burn, fires are most abun dant in the dry season, when the trade wind inversion with its large-scale subsidence and suppression of rain forming convection prevails over the region. Because this inversion prevents convection to heights of more than a few kilometres, it was initially thought that the linkage between dry conditions and subsidence more or less precluded the transport of pyrogenic ozone precursors to the middle and upper troposphere. Recent work has shown, however, that large amounts of smoke can get swept by low-level circulation, e.g. the trade winds, to wards convergent regions over the continents or the In ter-Tropical Convergence Zone, and there become sub ject to deep convection (Andreae et al. 2000;Chatfield et al. 1996;Thompson et al. 1996a). This transport pat-tern can explain the abundance of fire-related 03 and 03-precursors in the middle and upper troposphere as observed by remote sensing and in situ measurements (Browell et al. i996a;Connors et al. i996;Olson et al. i996).
The aerosols from biomass fires, the most obvious and visible sign of pyrogenic air pollution, may have an important impact on climate. Biomass burning is the second largest source of anthropogenic sub-micrometer aerosol (after sulphates from fossil fuel combustion), and possibly the largest source of black carbon parti cles. These aerosols influence climate and the hydrologi cal cycle by scattering and absorbing solar radi ation, and by changing the properties of clouds in ways that are just now being elucidated (Hobbs et al. i997;Kaufman et al. i998a;Ramanathan et al. 2000). Further charac terisation of the radiative and cloud-nucleating proper ties of pyrogenic aerosols and their effect on regional and global climate remains a major challenge to the sc i entific community.
Whether the impact of biomass burning will grow in the future depends both on climate change and on hu man factors. The amount of fuel available for burning at a given place and time is a function of ecological fac tors, e.g. soil fertility, precipitation, and temperature. It also depends on land use, i.e. if the area has been burned previously, is used for grazing or agriculture, and so on. If climatic vari ations become more extreme, as climate models have suggested, we can expect a more frequent occurrence of drought years fol lowing very wet years. This would result in large amounts of fuel ready to burn in the fire season. Furthermore, in a warmer and drier climate, fire frequency is likely to increase, which would reduce biomass carbon storage by changing the age class structure of vegetation, as well as causing increased emi ssions of ozone precursors. To monitor the regional and global evolution of pyrogenic emissions, it would be very useful to develop unique tracers for biomass burning, and to set up continuous measurements of these tracers at selected sites.
Human activities are of central importance to the frequency and severity of biomass fires. If large parts of the humid Tropics are deforested further, they will be transformed from a biome essentially free of fires (the tropical rainforest) to biomes with much more fre quent fires (grazi ng lands, agricultural lands, and waste lands). With a higher human population density, the fre quency of ignition will go up as well. And fi n ally, the amount of biomass burned for cooking and domestic heating, already a major source of emissions in tropical countries, will increase further. To follow these changes, we will need to develop further and validate techniques to determine the spatial and temporal distribution of biomass burn ing and the amounts of biomass burned in the various fire regimes.

Wet Deposition in the Tropics
Wet deposition plays an essential role in controlling the concentrations of trace gases and aerosol particles i n the atmosphere and in providing the essential nutrients for the biological functioning of ecosystems. Wet and dry deposition affect the budgets of key nutrients and trace gases ii n both terrestri al and marine ecosystems, as described in other sections of this chapter.
The Tropics are a particularly important region re garding global atmospheric chemistry. Due to intense ultraviolet radiation and high water vapour concentra tions, high OH concentrations oxidise inorganic and organic gases, and induce an efficient removal from the atmosphere of the oxidised products. Strong convection in the tropical regions results in huge volumes of ai r being drawn out of the sub-cloud layer with the result ant chemical composition of the precipitation coming from the capture of gases and small particles by the liq uid phases of cloud and rain. Knowledge of the chemi alU composition of wet deposition allows one to track seasonal emi ssions from various ecosystems.
In the i99os, due to the Jack of information on wet deposition in the Tropics, a cooperative programme was undertaken, involving the Global Atmosphere Watch (GAW) of the World Meteorological Organisation and the Deposition of Biogeochemically Important Trace Species (DEBITS) Activity of IGAC (see A.5), mostly in Asia. It was followed by the Composition and Acidity of the Asian Precipitation (CAAP) programme, later expanded into Africa and South America (Lacaux 1999).
In some tropical areas, however, dry deposition is at least as important as wet deposition and must be con sidered in the calculation of total deposition. Dry depo sition of acidic gases impacts soil and plants, as indi cated in Sect. 2.4 and 2.6.i. High concentrations of sul phur and nitrogen oxides and nitric and sulphuric ac ids may increase acidification processes. In many arid or semi-arid regions, transport and deposition of alka line soil particles to adjacent ecosystems are also im portant. The deposition of alkaline particles partly miti gates the effects of wet and dry deposition of acidic com pounds.
In addition to acidity, the N content of wet deposi tion may strongly affect ecosystem properties such as C storage, trace gas exchange, cation leaching, biodiversity, and estuarine eutrophication. This has been shown for temperate regions with altered N inputs (e.g. Howarth et al. 1996;Mansfield et al. 1998 and papers therein). Now, however, 40% of global applications of industrial N fertiliser takes place in the Tropics and subtropics, and over two-thirds is expected to occur i n now-developing regions by 2020 (Matthews i993; Bouwman 1998). Simi-larly, fossil fuel combustion is increasing dramaticall y in less economically developed regions, includi ng much of the tropical and subtro pi cal regions. Galloway et al. (1994) estimated that by 2020 nearly two -thirds of Earth's energy-related N inputs would take place in the Tropics and subtropics. In addition, N emissions asso ciated with biomass burn ing are heavily concentrated in the Tropics, and will likely remain so for decades (Andreae 1993) (see Sect. 2.6.2).

Preci pitation Chemistry in Equatorial Forests
To illustrate work unde rtaken by DEBITS, data for pre cipitation chemistry al5 associated wet deposit ion from several sampling sites located in equatorial forests ar e presented in Table 2.5. Hydrogen ion is abundant at all sites on an annual mean basis, indicating the general ly acidic character of equatorial precipitation. This acid ity is due to a mixture of mineral acids (HNO,. H2S04, etc.) and organic acids (formic, acetic, propionic, and others) (Andreae et al. 19 9 0; Ayers and Gillet 1988; Gal loway et al. 1982;Lacaux et al. 1991;Williams et al. 1997 ).
In equatorial African forest precipitation, the acidity contributed by organic acids (40-60%) is equ ivalent to that contributed by mineral acids (ca. 40%). In Amazonia the co mpo sition of precipitation is very dif fe rent, with organic acids accounting for 80-90% of the total acidity. In the rainwater collected at several remote locations in the Northern Territory of Australia, Gillet et al. (1990) found a volume-weighted mean (vwm) pH for all samples of 4.89, with organic acids contributing about 50% of the free acidity, the remainder bei ng H 2 SO 4 and HNO).
Duri ng the dry season, biomass burning has a dras tic influence on rainwater composition. The chemical content of rainwater from Amazonia (ABLE-2A, wet sea son) and African equato rial sites (Dimonika, Congo and Zoetele, Cameroon) can be compared to get a rough es timate of the cont ribution of some of the chemical com pounds from the vegetation fires (Table 2.6). In the case of Amazonia, it was assumed that the precipitation chemistry reflected the bio genic emissions of soils and vegetation, with little influence of biomass burning emissions. Therefore, the mean contribution of the veg etation-burning source in the African sites was esti mated to be about 60 to 70% of the N03, NHt, and acid ity contents. On the other hand, the African sites, lo cated on opposite sides of the Equator, are alternately affected by savannah burning sources from the South ern (June to October) and Northern (November to Feb ruary) Hemispheres, as shown by the ubiquitous pres ence of high concentrations of N03, NH;, and H+ in rain water samples. The gases and particles produced by sa vannah burning in the Northern and Southern Hemi spheres are transported by the north-east and south- 1.9 6.4 70 9.6 80 east Trade Winds, respectively, to the equatorial forests and progressively scavenged by convective clouds. The wet deposition measured i n the semi-arid and humid savannas surrounding the forested ecosystems presents a source of high potential acidity, which may not result in final strong acidity of the deposition. For example, Galy and Modi (1998) have shown that the precipitation from arid savannas is characterised by a weak acidity (H+ = 2µ eq1 -1) in spite of a high potential acidity (nitrate + dissociated formate+ dissociated acetate = 22 µeq 1 -1 ). This result is explained by hetero geneous interactions occurring between alkali ne soil dust particles and acidic gases. Many of these mineral dust particles are able to entirely neutralise gaseous ni tric acid. These gas-particle interactions occur before the incorporation of these particles into cloud droplets or raindrops. Furthermore, high concentrations of or ganic aerosols (from biomass burning and condensa tion ofbiogenic hyd rocarbons) and mineral dust (from deserts and arid areas) could also promote intense het erogeneous atmospheric chemistry (e.g. Dentener et al. 2996;Carmichael et al. 1996Carmichael et al. , 1997 (also see Chap. 4).
These processes may affect the cycles of nitrogen, sul phur, and atmospheric oxidants significantly.

Te rrestrial Ecosystems
Acidification effects are mainly due to deposition of mineral sulphur and nitrogen compounds. In tropical regions, organic acid deposition may contribute as much as 80%; however, these acids are oxidised in soils and will not participate directly in soil acid ification. Soil particles exchange alkaline cations with H+ and the con centration of alkaline ions determines the soil base satu ration. When base saturation is low, acids may release aluminium ions from soil particles. In spite of its limi tations, the "critical load" concept, characterising eco system sensitivi ty to acidic deposition, has been adopted as a tool for estimati ng potential impacts on ecosystems. In order to facilitate the development of strategies to control pollut ion in tropical countries, the Stockholm Environment Institute (SEI) has recently proposed a global assessment of terrestrial ecosystem sensitivity to acidic deposition that uses soil buffering capacity as a key indicator (Cinderby et al. 1 998). This assessment depends on two factors: the buffering capacity of the base layer to identify soils that have high weathering rate, and the cation exchange capacity to quantify the capacity of a soil to buffer acidity.
A global map prepared by SEI (see Fig. 2. 1 6), shows five classes of sensitivity to acidic deposition, from a critical load of 200 meq m -2 yr-1 for the insensitive class to a critical load of 25 meq m -2yr -1 for the most sensi ti ve class. Some selected wet deposition measurements of non-sea salt sulphate, nitrate, and organic acids, mainly obtained by the DEBITS Activity, are also in cluded. These combined measurements provide an over all view of tropical regions where the potential risk of acidification i s important. All the equatorial rainforests of South America, Africa, and Asia are classi fied in the most sensitive classes. For South American soils, which have a level of mineral acidity deposition of about 10-20 meq m-2 yr-1, future acidification problems may become severe if further land use change and indus trial activities occur in these reg ions. For tropical Af rica, due to the high contribution of mineral acidity from wet deposition from biomass burni ng sources, the criti cal load is nearly exceeded in many parts of western Africa. For Asia, in some parts of China, Japan, and other i n dustrialised and populated zones, the critical load has already been exceeded (Fig. 2.16).
Work by IGBP researchers suggests that there is sub stantial, although mostly indirect, evidence that the sup ply of N may not limit plant production in some tropi cal forests (Hall and Matson 2999 ). Thus, additions of N may have little di rect effect on plant production and carbon storage, but may substantially affect rate and tim ing of N losses. As i n dicated above, tropical forest soils are highly acidic; additions of anthropogenic N may increase that acidity, leading to increased losses of cati ons and decreased availability of phosphorus and other nutrients, ultimately limiting plant production and other ecosystem functions. Moreover, N add itions to tropical soils may result in immedi ate and relatively large pro portional losses of N in trace gas forms, as discussed above. On-going work strives to identify the direct and indirect effects of wet deposition on tropical agro-eco systems, and to determine its implications for ecosys tem functioning and feed-backs to the atmosphere lo cally, regionally, and globally.

Marine Highlights
From its inception, IGAC stimulated and sponsored research on marine aerosol and gas exchange of com pounds of biological origin through the Marine Aero sol and Gas Exchange (MAGE) Activity (see A.5), exam ples of which are given below. Similar to terrestrial biosphere-atmosphere research, integrated field cam paigns in marine regions, such as the ACE-1, ACE-2 and ASGA/MAGE experiments (see A.5) have been an IGAC hallmark. These field research efforts have linked stud ies of emissions, transformations, and transport in the marine boundary layer. The study of pertinent marine biogeochemical cycles resulting in sea-air fluxes, however, has yet to be fully integrated into these field campaigns. While research on biosphere-atmosphere interactions in marine regions has progressed signifi cantly in the last decade, it remains less advanced than   active gases, as for $02, the high transfer resistance in the water is shortcut by its very fast hydrat ion reaction, and the transfer of S02 is controlled, as water vapour, on the ai r side (compare S02 only physically dissolved at a low pH with S02 at pH= 6 in Fig. 2.17).
The intensity of turbulence determines the transfer resistance: the more intense the turbulence, the thinner the boundary layers. At the scales of the viscous bound al7 layer, turbulence is strongly attenuated by viscous forces. Thus, the turbulent diffusivity must decrease much faster to zero at the interface than the linear de crease found in the turbulent layer. A free water surface is, however, not solid, nor is it smooth as soon as short wind waves are generated. On a free water surface, ve locity fluctuations are possi ble that make convergence or divergence zones at the surface possible. A film on the water surface, however, creates pressure that works against the contraction of surface elements. This is the point at which the physicochemical structure of the sur- A collection of field data i s shown i n Fig. 2.18. Al though the data show a clear increase of the transfer velocity with wind speed, there is signi ficant scatter in the data that can only partly be attributed to uncertain ties and systematic errors in the measurements. The gas transfer velocity is not simply a function of the wind speed. The scatter mainly reflects the additional influ ence of the wind wave field, which may vary with all parameters that modify the microturbulence in the boundary layer such as the viscoelastic properties of the surface films present and the wind wave field.
Part of the data shown in Fig. 2.18 is based on geo chemical tracers such as the 1 4 C, 3He IT, or 222Rn t 226Ra methods. The transfer velocities obtained in this way are only mean values. Thus a parameterisation is only possible under steady state conditions over extended periods; it is questionable under changing conditions. The changes of the parameters (e.g. wind speed) are some orders of magnitude faster. Thus mass balance methods are not suitable for a study of the mechanisms of air-water gas transfer. This is also true for the tracer injection techniques pioneered by Wanninkhof et al. (1985, 1987) in lakes, and Watson et al. (1991a and Night ingale et al. (2oooa) in oceans. Progress in better un derstand ing the mechanisms of air-water gas exchange has been hindered by inadequate measuring technol ogy. Promisi ng new techniques are now available (Jiihne and HauBecke 1998;McGillis et al. 1999) but there are currently too few measurements using them for defi nite conclusions to be drawn. Thus, empirical gas ex change-wind speed relationships (see Fig. 2.17) must still be applied with caution since they have an uncertainty of up to a factor of two.

Methane
The ocean is a small source of methane to the atmos phere. Open Pacific Ocean saturation ratios (ratio of seawater CH4 partial pressure to the overlying atmo spheric CH4 partial pressure) range from 0.95 to 1.17 . Large areas of the Pacific Ocean are undersaturated with respect to atmospheric CH4 partial pressures duri ng the fall and winter. On a seasonal time scale, the driving force controlling saturation ratios outside the Tropics appears to be the change in sea surface temperature. Saturation ratios in the equatorial region have always been posi tive and appear to be driven by the strength of the equa torial upwelling. Extrapolating the Pacific data globally and regionally into ten zones, the calculated average flux of CH4 to the atmosphere is 0.4 Tg yr -1 (0.2 -0.6 Tg yr-1 ) (Bates et al/ 1996). This is approximately an order of magnitude less than previous estimates, which lacked all and winter data. Thus the open ocean is a very mi nor source of methane to the atmosphere ( <0.1%) com pared with other sources (IPCC 1996). However, the coastal ocean and marginal seas appear to be a much larger source (Owens et al. 1991;Kv envolden et al. 1993;Bange et al. 1994;Lammers et al. 1995;Scranton and McShane 1991) due to CH4 emissions from bottom sediments; this definitely warrants further investigation.

Carbon Monoxide
The ocean is ubiquitously supersaturated with CO with respect to the atmosphere, resulting in a net flux to the atmosphere ranging seasonally and regionally from 0.25 to 13 µmo! m-2 d-1. However, the total annual emission to the atmosphere (13 Tg; see Table 3.1) is small com pared to current estimates from both terrestrial natu ral and anthropogenic sources (1 150 Tg yr-1 ) (Bates et aal¿ 1995;WMO 1999). Even in the Southern Hemisphere, which accounts alr two-thirds of the oceanic emissions, the ocean source is relatively small ( <1%), since both methane oxidation and biomass burning are large sources of CO (Bates et al. 1995).

Volatile Organic Carbon Compounds
Volatile organic carbon (VOC) compounds, or non-meth ane hydrocarbons, are produced in surface seawater pos sibly by photochemical mechanisms, phytoplankton ac tivity, and/or microbial breakdown of organic matter (Plass-Diilmer et al. 1995;Ratte et al. 1995;Broadgate et al. i997). Oceanic concentrations show a strong seasonal cycle (Broadgate et al.1997). The ocean-atmosphere flux is dominated by alkenes and i s small compared to ter restrial emission estimates ( <1%). However, the emis sions may be significant on local scales considering the short lifetimes of the unsaturated compounds (Donahue and Prinn 1993;Pszenny et al. 1999 ). Additional seasonal measurements of isoprene, ethene, and propene in par ticular are needed in different oceanic regions.

Ammonia
Ammon ia is the dominant gas phase basic compound in the marine atmosphere and, as such, has a unique influ ence on marine multi-phase atmospheric chemistry. Ammonia exists in seawater as both ionised ammonium, NH4(s), and dissolved amm onia, NH3(s). Dissolved am moni a makes up about ten percent of the total seawater ammonium concentration, NHx(s), at a pH of 8.2 and a temperature of 25 •c (Quinn et al. 1996a) and is the com pound that is emitted across the air-sea interface. NHx(s) i s produced in the upper ocean from the degradation of organic nitrogen-containing compounds and excretion from zooplankton. It also i s released from bottom sedi ments to overlying waters. Loss processes for NHx(s) in clude bacteri al nitrification, uptake by phy topla nkton and bacteria, and emission across the air-sea interface. There will be a net flux of ammonia from the ocean to the atmo sphere if the atmospheric NH3 (g) concentrat i o n is less than the gas phase concentration in equilibrium with NH3(s).
Alternatively, there will be a net flux into the ocean if the atmospheric NH3(g) concentration is greater. In either direction, the magnitude of the flux depends on the con centration difference and the transfer velocity.
Attempts to estimate the air-sea flux of amm oni a have been hindered by a lack of teclmiques with sufficient sensitivity and by difficulties in avoiding sample con tamination (Williams et al. 1992). As a result, the contri bution of NH3 to the oceanic biogeochemical cycling of N is poorly understood. The few estimates of the air sea flux of NH3 that have been reported and that are based on measurements of ammonia in the gas, parti cle, and/or seawater phases are summ arised below.
The first estimates of the flux for the Pacific Ocean were based on filter collection of NH3(g) and NH3(s) (Quinn et al.1988(Quinn et al. , 1990. These measurements indicated a net flux of ammonia from the ocean to the atmosphere in the northeastern and central Pacific rangi ng between t.8 and 16 µmo! m-2d-1• Clarke and Porter (1993) used measurements of aerosol volatility (which i ndi cate the degree of neutralisation of sulphate aerosol by ammonia) to infer an efflux of ammonia from the ocean to the at mosphere of about to µmo! m-2 d-1 over the equatorial Pacific. Similar results have been reported for the Atlan tic Ocean and the Arabian Sea. Based on aircraft meas urements of aerosol ammonium during a Lagrangian experiment near the Azores, Zhuang and Huebert (1996) estimated a flux of NH3 from the ocean to the atmo sphere of 26 ±20 µmo! m-2 d-1• Simultaneous measure ments ofNHx(s) and NH3(g) were made in the Arabian Sea (Gibb et al. 1999). It was found that in both coastal and remote oligotrophic regions there was a flux ofNH3 from the ocean to the atmosphere. Hence, to date, meas urements over portions of the Pacific and Atlantic Oceans, and the Arabian Sea indicate that the remote ocean serves as a source ofNH3 to the atmosphere even in regions of low nutrient concentrations.
Given the importance of NHx(s) as an oceanic mi cronutrient, the loss of ammonia through venting to the atmosphere may seem surpri sing. However, only a small percentage ofNHx(s) exists as NH3(s) so that this efflux most likely represents a relatively minor loss of NH3 (Gibb et al. 1999). In addition, this loss can be episodi cally compensated for through the wet and dry deposi tion of ammonium-containing aerosol particles. For example, Quinn et al. (1988) estimated that, over the northeastern Pacific, the transfer of NH3(g) from the ocean to the atmosphere was balanced by wet and dry deposition processes. In certain regions, such as the Southern Bight of the North Sea, there is a flux of am monia from the atmosphere to the ocean due to the advection of high concentrations of ammonia from ad jacent land (Asman et al. 1994). The extent and impact of the deposition of continentally derived ammonia to marine regions is unknown but may be significant.
Model results suggest that about six percent of the glo bal continental emissions of ammoni a are deposited to the North Atlantic and Caribbean (Prospero et al. 1996). The deposition would be greatest in coastal waters.
It is clear that ammonia, as an oceanic micronutrient and the dominant atmospheric al phase compound, plays a unique role in both the ocean and the atmosphere. The fl ux of ammonia from the ocean to the atmosphere affects aerosol chemical composition, pH, and hygroscopicity. The reverse f11 ux, of amm onia plus amm onium in particles and rain from the atmosphere, to the ocean, may affect biological productivity. Simultaneous measurements of ammonia in the atmospheric gas and particle phases, in seawater, and i n rainwater are needed to improve our understanding of the multi-phase marine amm onia sys tem in general and the ai r-sea exchange of ammonia in particular. It i s i n teresting to note that ammonia, due to its strong partitioning into the water phase, is the only gas discussed in thi s chapter whose transfer velocity is under the control of air-side transfer processes.

Nitrous Oxide
The world oceans represent a significant natural source of N 2 0 to the atmosphere (e.g. Seitzinger et al. 2000). The surface waters of many oceanic regions are super saturated in N 2 0 with respect to solubility equilibrium During the past decade there have been significant improvements in our understanding of oceanic proc esses of N20 production, of the distribution of N20 in the surface and subsurface ocean, and of the magnitude of the oceanic N20 source to the global atmosphere. The relative roles of nitrification and denitri alcation proc esses have been addressed by measuring nitrogen sta ble isotopes and their fractionation between N20 and other dissolved nitrogen-beari ng compounds. The in terpretation of these di fficult measurements is compli cated by the likelihood that both nitrification and denitrification are coupled in many oceanic systems, and no clear picture has yet emerged. There have also been recent advances in the study of air-sea gas exchange processes, as indicated in Sect. 2.8.1, which will lead to improvements in the quantification of exchange coeffi cients as a function of wind speed.
Fi n all y, our understanding of the large-scale distri bution ofN20 in the oceans has been improved through a number of shipboard measurement programs, such as those associated with the World Ocean Circulation Experiment (WOCE) and Joi nt Global Ocean Flux Study (JGOFS) programs. These have generally reinforced our vi ew that open ocean upwelling regions along eastern ocean boundaries and in equatorial and coastal regions, represent major sources of atmospheric N20. By con trast, the great subtropical gyres, which represent a large portion of the surface area of the oceans, are relatively close to atmospheric equilibrium for N20. In recent years, some extremely high N20 concentrations have been found in the eastern Arabian Sea, in suboxic wa ters over the Indian Shelf (Naqvi et al. 2000). Since an thropogenic impingements on the coastal ocean may cause an increase in hypoxia, suboxia, and anoxia in some areas, these recent observations from the Arabian Sea are provocative. By modelling these distributions together with the wind field (e.g. Nevison et al. 1995), we have come to believe that the global oceans consti tute a net source to the atmosphere of about 4-5 Tg of N 20, or about one third of the global natural source strength. This value may increase as more is learned about the diverse distribution of N20 in coastal waters (e.g. Seitzinger and Kroeze 1998). In the years since publication of the DMS-CCN-cli mate hypothesis, almost i ooo papers have been pub lished discussing the distribution and biogeochemistry of DMS (and its precursors) and its link to climate. Sev eral IGAC-i nspired studies have addressed aspects of the OMS-aerosol-climate connection, most prominently among them ASTEX/MAGE (e.g. Huebert et al. 1996), ACE-1 (e.g. Bates et al. 1998), and AOE-91, 96 (e.g. Leck et al. 1996, 2001. As a result of these projects, and the large number of independe ntly conducted studies re lated to the OMS-climate hypothesis, we now understand many of the details of DMS production in the oceans, its transfer to the atmosphere, and the atmospheric oxi dation processes (see Chap. 3) that lead to the forma tion of aerosols (see Chap. 4) that can act as CCN. How ever, in spite of this progress, fundamental gaps remain in our understanding of key issues in this biosphere climate interaction, in particular with regard to the proc esses that regulate the concentration ofDMS in seawater.
While the basic processes have been identified, and even quantified in specific locations (e.g. Bates et al. 1994; Si mo and Pedro -Alio 1999), generally applicable mod els of OMS-plankton relationships are still in thei r in fancy (e.g. G:abric et al. 1993;Jodwalis et al. 2000 ). There fore, we are still not able to represent the DMS-CCN climate hypothesis in the form of a process-based, quan titative, and predictive model. Even the overall sign of the feedback cannot be deduced with certainty, since it is not known yet if a warming climate would result in an increase or decrease of DMS emi ssions. Glacial-to interglacial changes i n the amounts of DMS oxidation products in polar ice cores cannot answer this question unambiguously, as they may reflect variations in atmo spheric transport patterns as much as differences in DMS production (e.g. Whung et al. 1994), as is discussed in detail in Sect. 2.3.
Early, li mited data sets had suggested a possi ble cor relation between DMS and phytoplankton concentra tion (e.g. Andreae and Barnard 1984). This correlation is part icularly evident in vertical profiles of DMS and chlorophyll a, which in most instances show a sharp drop of both parameters around the depth correspond ing to a light penetration of one percent of the surface light flux. Close correlations between DMS and phyto plankton densities were also found in situations where a single species accounted for much of the DMS pro duction or phytoplankton biomass (e.g. Barnard et al. 1984;Matrai and Keller 1994). These find ings led to the hope that global DMS distributions could be estimated from chlorophyll concentrations obtained by remote sensing, but experimental investigations of this proposal were not encouraging (e.g. Matrai et al. 1993), except in frontal regions (e.g. Holligan et al. 1993;Belviso et al. 2000 ). Furthermore, a statistical analysis of almost 16 ooo measurements of DMS in surface seawater fai led to show any useful correlations between DMS and chlo rophyll or other chemical or physical parameters (Ket tle et al. 1999). One reason for the absence of a general correlation between plankton biomass and DMS is that the i n tracellular concentration of its metabolic precur sor, dimethylsulphoniopropionate (DMSP), varies be tween different phytoplankton species over a range of five orders of magnitude. While it is clear that some taxo nomic groups typically contain higher amounts of DMSP, these relationships are by no means clear-cut (e.g. Keller et al. 1989). At least as important, however, are the complexities of DMS cycling by biological and abiotic processes in the surface ocean, which will be addressed below.

Physiological and Ecological Controls of DMS Production
The pathways of DMSP biosynthesis in phytoplankton have been studied and have shed light on potential regu lating mechani sms such as nitrogen nutrition (e.g. Grone and Kirst 1992;Keller et al.1999a,b), temperature (Baumann et al. 1994), and light (Vetter and Sharp 1993Matrai et al. 1995). While DMSP, and sometimes DMS, is directly re leased by phytoplankton, zooplankton also play a role by grazing, or avoiding, DMSP-rich cells (e.g. Dacey and Wakeham 1986;Wolfe et al. 1997;Tang 2000).
Very high concentrations of DMS and dissolved DMSP have been reported from several coastal and/or high latitude areas, especially where blooms of DMSP producing phytoplankton such as the coccolithophore Emiliania huxleyi and the prymnesiophyte Phaeocystis pouchetii occur (e.g. Malin et al. 1993;Barnard et al. 1984). In this context, it is interesting to note that Kettle et al.'s (1999) DMS database revealed that high DMS re gions corresponded roughly to the coccolithophorid bloom areas derived by Brown and Yoder (1994) al om remotely sensed ocean colour data. Prymnesiophytes (including coccolithophores) and dinoflagellates are phytoplankton groups that tend to have high DMSP cell quotas (Keller et al. 1989) and, not surprisingly, DMS is often relativ·ely high when these groups dominate the phytoplankton assemblage. Diatoms, on the other hand, tend to have low intracellular DMSP concentrations and it is generally observed that diatoms are less important DMSP producers in the field (e.g. Keller et al. 1989 ). Pre dicting DMS concentrations from the algal assemblage is not straightforward, however. For example, Matrai and Vernet (1997) reported that DMS concentrations were as high in diatom-dominated, Arctic waters as they were in those dominated by Phaeocystis sp. It is now recog nised that some phytoplankton species not only pro duce high intracellular concentrations ofDMSP, but they also have cell-surface (Stefels and Dijkhuizen 1996) and intracellular (Steinke et al. 1996) DMSP lyase enzymes that may be involved actively in DMS production, thereby contributi ng further to the elevated DMS con centrations associated with these organisms. The eco logical roles of these lyase enzymes are not well under stood but several recent studies have pointed to very interesting functions such as in grazing deterrence, car bon acquisition, and bacterial inhibition (Noordkamp et al. 1998;Wolfe and Steinke 1996;Wolfe et al. 1997).
Blooms of marine phytoplankton provide convenient natural "laboratories" for investigating the production of DMS in relation to phytoplankton community dy namics and species succession and associated processes, including grazing and bacterial turnover. However, this apparent focus on "hotspots" of DMS production i n rela tively nutri.ent rich areas can be criticised in that oligotrophic areas of the oceans, which generally have relatively low levels of DMS and DMSP throughout the year, make up a large fraction of the total ocean area and so must contribute significantly to the total global flux of OMS (Bates et al. 1992). These pioneering stud ies established the link between phytoplankton and DMS levels, but failed to account for a large part of the natu ral variability in DMS concentrations. There have been rather few actual DMS time-series studies (Leal[ et al. 1990;Turner et al. 1996a;Dacey et al. 1998), all of which noted seasonal periods of elevated DMS concentrations.
We now realise that bacterial processes are also very important i n the overall DMS cycle. More isolates of bac teria are available with which to study biochemical path ways and physiology of DMSP and DMS metabolism (e.g. Ledyard andDacey 1994i Yoch et al. 1997). New meth ods, including use of 35S tracers, improved inhibitors, and molecular genetics techniques have allowed ever more sensitive analyses of DMSP-DMS cycling rates and fates, and have permitted more detailed examination of the complex microbial communities involved (e.g. Gonzalez et al. 1999;Wolfe and K.iene 1993). The potential for DMS production from dissolved DMSP is quite large (e.g. K.iene 1996b;van Duy! et al. 1998), but recent studies indicate that most of the DMSP in the sea is not converted to DMS. A demethylation-demethiolation pathway leading to production of methanethiol (MeSH) can account for 70-95% of DMSP metabolism in situ thereby diverting sulphur away from DMS (K.iene 1996a). The predomi nance of this non-DMS producing demethylation demethiolation pathway is explained by the fact that bacteria use it to assimilate the sulphur from DMSP into protein amino acids (Kiene et al. 1999). Further under standing of this DMSP-DMS-MeSH-bacteria interaction is critical because a relatively small change in the yield of DMS from DMSP could have a major impact on the gross production of DMS, which would then be avail able for sea-air exchange.
Removal of DMS from the water column by biologi cal and photochemical mechanisms also exerts a great i n fluence on the net accumulations of DMS in surface waters. Slow biological degradation of OMS may par tially explain the rise in DMS concentrations observed at the peak and initial decline phases of phytoplankton blooms (e.g. Matrai and Keller 1993;Nguyen et al. 1988). Net consumption of DMS appears to occur in the later stages of blooms after OMS-consuming bacteria have had time to develop (Kwint et al. 1996;van Duy! et al. 1998). The photochemistry of OMS in seawater remains poorly understood, despite the fact that it has been iden tified as a major removal mechanism under some cir cumstances (e.g. Kieber et al. 1996;Sakka et al. 1997;Brugger et al. i998). OMS photooxidation appears to depend on photosensitisers i n seawater, which are most likely part of the coloured dissolved organic matter (CDOM) (Dacey et al. 1998). In the open ocean CDOM originates from autochthonous primary productivity and food web processes so the interaction with OMS i s probably complex. Add to this the fact that OMS pro ducing and consuming bacterial populations are likely to be strongly influenced by UV-B in surface waters, and one can easily see the importance of understanding photophysical effects on the OMS cycle. Recently, it has been shown that viruses are significant agents in the con trol of bacteria and phytoplankton. Viral infections can cause a total release of intracellular OMSP (Hill et al. 1998) and viruses are known to infect DMSP-containing bloom organisms such as Emiliania huxleyi (Brussaard et al. 1996) and Phaeocystis sp. (Malin et al. 1998). It seems clear from studies such as these that the overall food web dynamics, including macro-and microzooplankton graz-ing, bacterial, and viral activities, as well as the physico chemical dynamics of the upper ocean (e.g. incoming solar radiation, mixing, temperature, ai r-sea exchange) are important factors governing OMS accumulation.
Modelling efforts have expanded our understanding of OMS production, both for field situations (e.g. Gabric et al.1999) and laboratory systems (Laroche et al.1999). However, our current knowledge base is not sufficient to develop and constrain predictive OMS production models for diverse biogeographic regions, in order to allow interpretation of the role of OMS in climate change, for example. Future research will need to focus on (1) gaining a full understanding of the processes that control OMS production and allow the prediction of OMS emissions, and (2) obtaining much more data concern ing spati al, temporal, and interannual variation in the concentration of DMS and related compounds. Empha sis on undersampled areas and seasons would be valu able. For process studies, there is an i n creasing need to cross disciplinary and international boundaries to bring together experts on different aspects of DMS and related compounds for integrated field campaigns. For analysis of variability, "remote" sampling systems could be con sidered (such as attempted in ACE-1). It mi ght be possible to develop a buoy-mounted monitori ng system whereby samples were stored on a carousel for later analysis. Alter nati vely, we might follow the example of the pC0 2 meas uring community, who have demonstrated that it is fea sible to employ unmanned instruments on merchant ships (Cooper et al. 2998a). This would enable the col lection oflarge data sets during long passage routes, cov ering diverse biogeographic areas, and different seasons, and the chance to investi gate i n terannual variability at relatively low cost. New techniques will be needed to cir cumvent the present lack of a reliable storage method for DMS samples. In the first instance, it might be more realistic to concentrate on OMSP analyses.

Carbonyl Sulphide
The oceans represent approximately 30% of the total atmospheric source of COS, and much of the oceano graphic work on COS over the last decade has focussed on assessing the spatial and temporal di stributions of COS concentration and understanding the processes that control its temporal and spatial distribution. The photochemical source of COS was first recognised by Ferek and Andreae (1984), who demonstrated a clear diurnal cycle in the sea surface concentration of the compound. A mechanism of formation of COS was pro posed by Pos et al. (1998) who suggested that the photo chemical production of COS and carbon monoxide pro ceeds along a coupled pathway which first i n volves the photochemical formation of an acyl radical from col oured dissolved organic matter (CDOM). Flock et al. (1997) andUlshOfer et al. (1996) suggested that cysteine is probably i m plicated in the reaction mechanism of COS formation as the result of its reactivity and abun dance in the oceans. The photochemical COS produc tion in natural seawater is probably not limited by the concentration of a precursor sulphur compound but rather by the concentration of CDOM represented by its ultraviolet attenuation coefficient (Ulshafer et al. 1996;Uher and Andreae 1997). Zepp and Andreae (1994) and Weiss et al. (1995) quantified the wavelength de pendence of COS photoproduction from CDOM and found that quantum efficiency of photoproduction de creases monotonically with increasing wavelength. The dark (or non-photochemical) production of COS has been proposed on the basis of the non-zero COS con centration observed at ocean depths where there is no photochemical production and where there is no mix ing from the surface (Radford-Knoery and Cutter 1994;Flock and Andreae 1996) and also on the basis of care ful interpretation of sea surface COS concentration measurements using inverse models (UlshOfer 1995). COS hydrolysis varies as a function of temperature and pH and has been evaluated several times over the last decade (Elliott et al. 1989;Radford-Knoery and Cutter 1994;Uher and Andreae 1997).
Recent models have used laboratory results for the photoproduction and hydrolysis rate constants to ex plain COS sea surface measurements obtained during expeditions made in the 1980s and 1990s (see UlshOfer (1995) for a review of recent sea surface COS concentra tion measurements). Andreae and Ferek (1992) devel oped the first chemical box model to explain the diur nal variation of COS in terms of photochemical forma tion and hydrolysis destruction. UlshOfer (1995) adopted an optimisation scheme based on the coupled photo chemical-mixed layer used by Kettle (1994) to calculate the photoproduction and dark production constants for COS from a series of sea surface measurements made between 1992 and 1994 in the North Atlantic Ocean. von Hobe (1999) extended this work for other models and expedition measurements. Najjar et al. (1995) general ised a simplified coupled physical-chemical model on a global scale to investigate the sensitivity of COS sea sur face concentration on ozone reduction and tropospheric increases of carbon dioxide. Kettle and Andreae (1998) and Preiswerk and Najjar (1998) have used existing measurements of the CDOM absorption coefficient of seawater to predict a seasonal variation in the absolute COS concentration, with maximum values at high lati tudes i n the summer of either hemisphere.
Future work on COS should aim to quantify more accurately the role of the oceans as a source or sink of the gas to the atmosphere. The global application of the photochemical production model for COS is currently limited by the absence of an algorithm to predict the global CDOM absorption coefficient and by the sugges-tion that the apparent quantum yield of COS formation may vary by more than an order of magnitude i n differ ent regions of the ocean. The scarcity of profile meas urements of COS concentration has been problematic for modelling efforts which have so far been developed to explain only the surface COS concentration di stribu tions. Finally, the precise quantification of the sea-air flux of all gases produced i n the upper ocean (i ncluding COS) is currently limited by the absence of an effective gas exchange parameterisation based on wind speed, average wave slope, or other measure of upper ocean turbulence, as already ind icated.  et al. 1996) and that the open ocean is a net sink, rather than a source, for CH3Br (see below). Methyl bromide (CH3Br) in the environment began to receive considerable attention in the early 1990s when it was being evaluated as an ozone-depleting gas, along with chlorofluorocarbons, chlorocarbons, and halons. First-order calculations indicated that methyl bromide was likely to be a significant contributor to stratospheric ozone depletion. Before then, only a few studies of CH3Br in the ocean and atmosphere had been conducted. Lovelock (1975) detected CH3Br in coastal waters of England and suggested that this gas could have a large natural source. Singh et al. (1983) later reported wide spread supersaturations greater than 200% off the west coast of North America, lending support to the ocean as a large natural source of CH3Br. Khalil et al. (1993) suggested that the open ocean was supersaturated in methyl bromide by 40-80%. However, prompted i n part by calculations showing that the ocean simultaneously had to be a large sink for CH3Br because of reaction with c1-i n seawater (Elliott and Rowland 1995;Jeffers and Wolfe 1996;King and Saltzman 1997), numerous investigations, using in situ mass spectrometry-gas chromatography, demonstrated that the ocean on aver age was a net sink for atmospheric CH3Br, with tropical and subtropical waters of the open ocean highly undersaturated and coastal waters often supersaturated in this gas (Lobert et al. 1995(Lobert et al. , 1996(Lobert et al. , 1997Moore and Webb 1996;Groszko and Moore 1998). Certain species of phytoplankton produce CH3Br, but apparently not at rates sufficient to explain the observed saturation lev els (Saemundsdottir and Matrai 1998;Moore et al. 1995;Scarratt and Moore 1996). Most recently, there have been suggestions that CH3Br in temperate and coastal waters might undergo a seasonal cycle, with higher concentra tions or supersaturations in the spring and early sum mer and undersaturations the rest of the year (Balcer et al. 1999;King et al. 2000 ). About the same time, it also became clear that chemical and biological removal of CH3Br i n seawater constituted such a large sink for thi s gas that it would have a profound effect on the lifetime of CH3Br in the atmosphere, even if the ocean were eve rywhere a net source (Butler 1994;Yvon et al. 1996b;Yvon -Lewis and Butler 1997). ln the latest budget calcu lations, irreversible loss of atmospheric CH3Br to the ocean accounts for one-quarter to one-third of the total removal (Kurylo et al. 1999).
These two findings -that the oceanic source was outweighed by its sinks and that the lifetime of atmo spheric CH3Br depended strongly upon its reaction in seawater -necessitated a re-evaluation of the global budget of this gas in the atmosphere. Once the appar ently large soil sink was discovered and confirmed (Sen;:a et al. 1998;Shorter et al. 1995;Varner et al. 1999), the calculated budget of atmospheric CH3Br was no longer in balance. The latest calculations have sinks outwe ighi ng sources by 80 Gg yr·1, out of a budget of 205 Gg yr· 1 (Kurylo et al. 1999). It is unlikely that this Arctic 20.04.1997 0 2 3 4 5 6 7 additional source will come from the ocean, as the cur rent global coverage of surface measurements, although not complete, is representative of the various oceanic regimes, although with reduced coverage of coastal wa ters. Currently, a small net sink is calculated for the ocean (3-30 Gg yr-1) which is unlikely to change much, unless, of course, there is some significant global change driv ing it. Furthermore, recent studies are identifying ter restrial sources from plants and salt marshes that are making the budget gap smaller (Gan et al. 1998;Rhew et al. 2000;Dimmer et al. 1999).
Perhaps one of the most significant things to come out of these intensified studies of methyl bromide in the ocean is that other halogen gases may behave in simi lar, quantifiable ways. Many of these gases, which may include CH31, CHBr3, CH2Br2, CH2BrCl, and C2H5Br, among others, also have climatic implications through their chemistry or radiative effects; however, specific studies of them in the past have been limited (e.g. Large amounts of reactive bromine (and smaller amounts of chlorine) are also found in polar regions and near salt pans likely due to oxidation of halides by inor ganic reactions (see also As a result of these polar di scoveries as well as mod elling studies (e.g. Vogt et al. 1996;Sander and Crutzen 1996) (see also Chap. 3), researchers have begun to seek and confirm the occurrence of reactive halogen com pounds (10, BrO, ClO) from air-surface exchange proc esses in other regions (e.g. remote mid-and lowlatitude marine sites, midlatitudes coastal sites, Dead Sea basin, and the free troposphere).

Primary Marine Aerosols
Primary aerosols are also emitted directly from tlte oceans.
The work of Blanchard and colleagues (Blanchard 1983) has shown that bubble bursti ng at the air-water inter face injects aerosols into the atmosphere from two sources. One is from fragments of the bubble al (al drops), the otlter from a jet of water that follows tlte bubble burst. Bubbles selectively scavenge high molecu lar weight surface-active compounds (Gershey 1983) and vi able particulate material from the water such as bac teri a and viruses (Blanchard 1983), leading to a consid erable enrichment of these organic components in the aerosol relative to the water. As a result, primary parti cles in the marine environment will usually contain a wide range of biogenic compounds. Long-chain fatty acids, alcohols, esters, and soluble proteins have al3l been found in marine aerosols. Proteinaceous material and free amino acids are present in marine rain (Mopper and Zika 1987 It is now recognised that a primary transport path for iron found in the ocean is through the atmosphere. Among the first papers to address the importance of atmospherically derived iron were those of Moore et al. (1984) and IDuce (1986). These autltors calculated the aeolian transport of mineral matter into many areas of the ocean, and pointed out that some fraction of the iron from the mineral matter dissolved into seawater after the dust was deposited to the ocean surface. Duce and Tindale (1991) and, more recently, Jickells and Spokes (2001) have rev iewed this topic.
The major reason why atmospheric dust transport has received considerable research effort over the last decade is because of the role ir on has been hypothesised to play in controlling marine primary product ivity over large areas of the oceans remote from land. Because of their distance from riverine and shelf inputs in these regions (e.g. Southern Oceans, North and Equatorial Pacific) one of tlte primary ways in which "new" iron gets in to the system is via deposition from the atmo sphere of terrestrially derived material. The idea of iron being a major control on ocean production is not new.
In the early decades of the 20 th century it was hypoth esised that the reason why large areas of the Southern Oceans contained significant amounts of residual con any resulti ng change in productiv ity in the receiving water. Several dust deposition events appeared to be correlated with increases in primary productivity meas ured i n on-deck incubators, but with a four day lag be tween the dust input and the peak in productivi ty. Al though suggestive of a relationship, the results were too few and insufficiently clear-cut to be totally convincing.
In addition, interpretation was complicated because productivity change was measured in a deck i n cubator, not in the ocean itself. Also, when deposition occurred, meteorological conditions changed, with greater stirring of near-surface water, which itself may have changed the productivity. However, this experiment represents a novel and potentially powerful tool since it uses the natural atmospheric input and examines the response of the real oceanic system.
A diffe rent approach to testi ng the iron hypothesis is that of adding inorganic iron (FeS04) directly to a small patch (of the order of 100 km 2 ) of the oceans. In order to be able to track the iron enriched patch as it moves  (2000) ). On all three occasions, rais ing the iron level in the water by a few nanomoles per litre produced a significant enhancement in phyto plankton activ ity, as measured by chlorophyll concen-  Days since beginning of experiment tration increase, consistent with the iron fertilisation hypothesis. In the case of lronEx II, the increase was at least an order of magnitude. Smaller organisms were the first to utilise the iron supplement, with the larger plankton (mainly diatoms) benefitting later. Trace gases measured in these experiments were C02 and DMS. The former was drawn down due the en hanced primary production. The extent of C02 removal roughly mirrors the increase in chlorophyll, except for IronEx I where it was very small, probably due to rapid recycling of the fixed carbon by grazers. For D MS, three to five-fold increases occurred in all three studies, with much less variation than for C02• Carbon dioxide changes between inside and outside the enriched patch during the course of SOIREE are shown in Fig. 2.21, and the time evolution of DMS and its precursor DMSP i n tegrated over a vertical column are shown in Fig. 2.22. Such a fertilisation experiment is akin to a batch cul ture perturbation and it is not clear whether long-term Fe enrichment and sustained higher productivity would lead to higher steady-state DMS concentrations. Given  enough time, shifts in the population structure, or sim ply growth of the microbial populations, might re-es tablish low steady-state DMS concentrations with per haps higher turnover of both DMSP and DMS. Our present understanding of the response of microbial populations to changes in DMSP and DMS supply is i n sufficient to make confident predictions in this regard.
To put these results in a broader time context, a com pendium of results from ice cores for iron, C02, MSA (an atmospheric oxidation product of DMS}, and sev eral other parameters is provided in Fig. 2.23. lt is note-worthy that the elevated iron and MSA and lowered C02 levels during the last glacial period are consistent with a scenario wherein ocean productivity was higher then due to enhanced atmospheric inputs of iron. For fur ther discussion of the use of ice core records to exam ine the overall sulphur cycle see Sect. 2.3. There is now widespread evidence that atmospheric fixed nitrogen compounds contribute to enrichment and in some areas probably to coastal and estuari ne eutrophication (Jaworski et al. 1997;Howarth et al. 1996). Current estimates of the percentage of total (natural + anthrop ogenic) new N loading attributed to direct atmospheric deposition at a number of North American and European locations range from 5% to over 50% (Duce 1991;Fisher and Oppenheimer 1991;Valigura et al. 1996;Dennis 1997;Holland et al. 1999). Inputs of N to estuarine systems that result from di rect atmospheric deposition by-pass much of the estuarine N "filter" (Kennedy 1983;Paerl 1995Paerl , 1997. Thus, atmospheric deposition assumes an increasingly important role as a new N source in lower estuarine and coastal waters below the biological N fil tering zone (Fig. 2.24). Dry and wet atmospheric deposition introduces into estuaries a variety of biologically available inorganic (N03, NHt, DON) compounds, most of which result from human activities (Likens et al. 1974;Galloway et al. 1994). In addition, organic nitrogen (ON) comprises a significant fraction (from 15 to over 30%) of wet and dry atmospheric deposition in coastal watersheds (Correll and Ford 1982;Scudlark and Church 1993

Ad�ect1on. mixing
Upper and lower estuarine processes Paerl 1997). Although the composition of atmospheric ON is poorly known, recent work (Peierls and Paerl 1997;Seitzinger and Sanders 1999) indicates that constituents of this pool are biologically utilised and, hence, should be included in eutrophication assessments.
In situ bioassays and field surveys show that enrich ment with the major deposition constituents NHt and NO) at natural di lu tions and atmospherically derived dissolved organic nitrogen (DON) results in enhanced phytoplankton primary production and increased biomass (Paerl 1985;Willey and Paerl 1993;Paerl and Fogel 1994;Peierls and Paerl 1997). Atmospheric DON may selectively stimulate growth of specific types of marine phytoplankton (Neilson and Lewin 1974;Antia et al. 1991). These selective phytoplankton responses to specific nitrogen inputs, and changes in stoichiometric C: N ratios resulting from these inputs may induce changes at the zooplankton, invertebrate, herbivorous fish, and higher trophic levels.  1977 1980 1983 1986 1989 1992 1995 Year patterns, and proximity of atmospheric sources, an im portant fraction of nitrogen from atmospheric deposi tion is di rectly deposited on the estuary. In the case of the APSS, recent estimates are on the order of 20% (for its estuari ne tributaries) to 40% (for the downstream waters of Pamlico Sound) (Paerl and Fogel 1994;Paerl 1995Paerl , 1997 of the N being directly deposited. Atmospheric N generated from expanding intensive animal farming is of particular concern. Examination of the long-term record of atmospheric NHt and N03 deposition in Sampson County, eastern North Caro lina, shows a nearly three-fold increase in annual NH! depo sition (also relative to N03) since i977, with a particu larly precipitous rise since the late 1980s (Fig. 2.25). The reason for this may be that unlike human waste, swine waste is stored in open lagoons and remains largely untreated, and substantial amounts (30 to >80%) of N are lost via NH3 volatilisation alone ( O'Halloran 1993).

The Open Ocean
There is also growing concealē about the increasing in put of human-derived nitrogen compounds to the open ocean. This is especially important i n parts of the open ocean where nitrogen is the nutrient that limits biologi cal growth. This is the case in the nutrient-poor waters of the large central oceanic gyres in the North and South Pacific and Atlantic Oceans and tlie soutliern Indian Ocean.
Current estimates suggest that, at present, atmospheric nitrogen accounts for only a few percent of the total new nitrogen deli vered to surface waters in tliese regions, with upwelling from deep waters being the primary source of new surface nitro gen. It is recognised, however, that the atmospheric in put to the ocean is highly episodic, often coming in large pulses extending over a few days. At such times, atmospheric input plays a much more important role as a source for nitrogen in surface waters. A recent estimate of the current input of fixed nitrogen to the glo bal ocean from rivers, the atmosphere, and nitrogen fixa tion indicates that all three sources are important (CoalĔell et al. 1995). Paerl and Whitall (1999) estimate that 46-5;>o/o of the total human-mobilised nitrogen enteri ng the North Atlantic Ocean is coming via the atmosphere.
In addition, the atmospheric organic nitrogen flux may be equal to or perhaps greater tlian tlie inorganic (i.e. ammonium and nitrate) nitrogen flux in open ocean re gions. The source of tlie organic nitrogen is not known, but a large fraction of it is likely to be human-derived. This form of atmospheric nitrogen input to the open ocean had not been considered in detail until very recently. Not only will the input of atmospheric fixed nitro gen to the open ocean increase significantly in the fu ture as a result of in creasing human activities, but the geographical locations of much of thi s input will prob ably change as well. Galloway et al. (1994Galloway et al. ( , 1995 have evaluated pre-industri al nitrogen fi xation (formation of the so-called reactive nitrogen) on the continents; the near-current (1990) reactive nitrogen generated from human activities such as energy production (primarily as nitrogen oxides), fertiliser use, and legume growth; and the estimated reactive nitrogen that will be pro duced i n 2020 as a result of human activities. The most highly developed regions in the world are predicted to show relatively little increase in the forma tion of reactive nitrogen, with none of these areas con tri buting more tlian a few per cent to the overall global increase. However, other areas will contribute very sig nificantly to increased human-derived reactive nitro gen formation in 2020. For example, it is predicted tliat Asia will account for -40% of the global increase in energy-derived reactive nitrogen, while Africa will have a six-fold increase accounting for 15% of the total glo bal increase. It is predicted that production of reactive nitrogen from the use of fertilisers in Asia will account for -87'Yo of the global in crease from this source. Both energy sources (nitrogen oxides and ulti mately nitrate) and fe rtiliser (ammonia, urea) result in the extensive release of reactive nitrogen to the atmosphere. Thus, these predictions indicate very significant potential in creases i n the atmospheric deposition of nutrient ni trogen compounds to the ocean downwind of such re gions as Asia, Central and South America, Africa, and the former Soviet Union (see Fig. 2.26).
The potential problem outlined above was high lighted by a computer modelling study undertaken by Galloway et al. (1994), who generated maps of the re cent (1980) and expected (2020) annual deposition of reactive nitrogen compounds from the atmosphere to the global ocean. Figure 2.26 is a map of the projected ratio of the estimated deposition of oxidised forms of nitrogen in 2020 to the values for 1980. It appears that from one and a half to three, and in some limited areas up to four, ti mes the i980 rate will occur over large areas of the oceans. Thi s i n creased nitrogen deposition will provide new sources of nutrient nitrogen to some regions of the ocean where biological palĕduction is currently lim ited by nitrogen. There is thus the possibility of i m por tant impacts on regional biological production and the marine carbon cycle in these regions of the open ocean.  1994, andWatson 1997) Debate continues about the relative importance of iron, nitrogen, or other compounds as prime determi nants of oceanic phytoplankton productivity and, con sequently, potential controls of mari ne gas emissions.
On geological time scales, phosphorus is accepted to be the ultimate control. Silica, another component of the same Fe and N-containing dust but with a significantly longer residence time, has also been examined as an al tering agent of the species composition of marine phytoplankton in oceanic regions, favouring siliceous organisms (e.g. diatoms) (e.g. Harri son 2000; Treguer and Pondaven 2000). Such organisms would differen ti ally affect total gas emissions but not total primary production. Such a silica hypothesis reinforces the link between marine biogeochemi stry and resulting sea-air gas emissions.

Research Challenges
Much progress has been achieved over the last decade through technological advances and appropriate scien tific approaches. The advances include: remote sensing in strumentation to provide detailed spatial and tempo ral data; micrometeorological and isotopic techniques for estimati ng the flux of matter and energy withi n eco systems and between ecosystems and the atmosphere, geosphere, and biosphere; techniques for manipulations of local scale selected environmental factors; and, sta tistical and numerical modelling techniques capable of analysing multi-variate, nonlinear problems.
A developing system-based approach, including Lagrangian studies of air and water masses, compris ing all components, e.g. soils, vegetation, and atmo sphere, has led to an understanding of biogeochemical cycles of individual chemical compounds and interac tions among chemical compounds. The campaign mode of carrying out field measurements has enhanced the understanding of the interconnectedness of systems and the importance of scaling issues.
Highli ghts of the research include: • Reduced uncertainties in N20, NO, CH4, DMS, and certain organohalogen emissions, and a better char acterisation of local and regional distribution pat terns of fluxes together with a mechanistic, but not necessarily integrated, understanding of the surface factors which control these emissions and exchange.
• Effective mitigation strategies have been developed for some CH4 emissions and a better understanding of how land management practices influence N20 and CH4 emissions has been gained.
• Mechanisms and pathways of production and envi ronmental controls have been identi fied for a large number ofVOC compounds emitted from vegetation, including canopy transfer processes. Emission mod els estimating global emissions have been developed.
• voe emissions can account for a loss of two to four percent of C taken up by photosynthesis, which has implications for understanding and quantifying the C cycle.
• Improved understanding of atmospheric input of inorganic N and Fe, mainly of anthropogenic and soil origin, into coastal and open ocean environments representing 5-70% of total input in the case of ni trogen and the subsequent impacts on the C uptake of the oceans and the C and S cycles.

•
The acidic nature of wet deposition, which differs by source and region, has been characterised.
• Improved understanding of the partitioning of dry deposition (particularly 03 and N02) on leaves and soil surfaces and related physiological mechanisms has been developed.
• Emission ratios for biomass burning are well de scribed for savannas, but less well described for hu mid forests and biofuels. Broad databases are avail able of emi ssion factors for a large number of sub stances.
To achieve more plausible and quantitatively reliable answers, several key issues remain: • To investigate mechanisms (chemical and biological) responsible for trace gas cycling (emission and depo sition) in oceans, soils, and plants, and to establish long-term sites/studies to provide that information, undertaking field experiments to determine, quan tify, and discriminate among driving vari ables.
• To study the exchange of voes between vegetation, oceans, and the atmosphere along with the exchange of other trace gases.
• To understand and quanti fy the effect of soil-released NO and its oxidation product N02, under different management practices, to the atmosphere including interaction of these gases both within and above the canopy.
• To determine whether changes in the marine emis sions of trace gases and particles are likely to have a significant influence on atmospheric chemistry and vice-versa, resulting from climatic (e.g. rainfall, tem perature; perhaps small}, elevated C02 (perhaps large), and/or land use changes. Key areas may in clude the greenhouse effect (tropospheric 03), strato spheric ozone (CH3Br), radiation and clouds (DMS), VOCs, tropospheric chemistry (dust-FE-DMS-C02), and other unpredicted impacts (e.g. the change in marine phytoplankton communities coupled with changes in N and Fe deposition), especially in the Tropics and high latitudes.
• To understand how the hydrological cycle will be af fected in various regions with climate change and the subsequent impacts on emissions.
• To improve the parameterisation of air-sea exchange and its links to biogeochem ical cycling in surface waters as well as improve Lagrangian studies in wa ter, air, and the combined ocean-atmospheric front, i n cluding international participation i n order to over come the intrinsic organisational and logistical dif ficulties.
• To promote the establishment, wherever possible, of long-term sites for flux measurements, to investigate the magnitude of interannual variation and thus achieve more robust estimates of mean annual fluxes and global budgets.
• To design experiments that will bring synthesis from emission-type studies, regional means of fire detec tion and prediction, spatially and temporally re solved, and chemical transport models (see Chap. 6) in order to determine the impact of burning on at mospheric chemistry.

•
To develop more reali stic biological and deposition process-oriented models with interaction and feed back among process-oriented, regional models and global models in order to provide improved estimates of emission and deposition fluxes.