Microbiology and atmospheric processes: chemical interactions of Primary Biological Aerosols

This paper discusses the inﬂuence of bioaerosols on atmospheric chemistry and vice versa through microbiological and chemical properties and processes. Several studies have shown that biological matter represents a signiﬁcant fraction of air particulate matter and hence a ﬀ ects the microstructure and water uptake of aerosol particles. 5 Moreover, airborne micro-organisms can transform chemical constituents of the atmosphere by metabolic activity. Recent studies have emphasized the viability of bacteria and metabolic degradation of organic substances in cloud water. On the other hand, the viability and metabolic activity of airborne micro-organisms depend strongly on physical and chemical atmospheric parameters such as temperature, pressure, radia- 10 tion, pH value and nutrient concentrations. In spite of recent advances, however, our knowledge of the microbiological and chemical interactions of primary biological particles in the atmosphere is rather limited. Further targeted investigations combining laboratory experiments, ﬁeld measurements, and modelling studies will be required to characterize the chemical feedbacks, microbiological activities at the air/snow/water 15 interface supplied to the atmosphere.


Printer-friendly Version
Interactive Discussion EGU health-related behaviours of atmospheric aerosols (Seinfeld and Pandis, 1998;Jacobson et al., 2000;Turpin and Saxena, 2000;Seinfeld and Pankow, 2003;Kanakidou et al., 2005;Pöschl, 2005;Fuzzi et al., 2006;and references therein). Traditionally the total carbon (TC) content of air particulate matter is divided into an organic carbon (OC) fraction and a black carbon (BC) or elemental carbon (EC) fraction. The origin 5 of BC and EC is mostly pyrogenic (e.g., fossil fuel combustion, biomass burning, Szidat et al., 2004.). The organic fraction is a complex mixture of thousands of organic compounds. The characterisation, classification, and description of the organic fraction in atmospheric models is a major challenge of current atmospheric and climate research (Kanakidou et al., 2005;Fuzzi et al., 2006;Kalberer et al., 2006;Donahue 10 et al., 2006). The sources of OC are much more diverse and include secondary organic aerosol (SOA) formation in the atmosphere as well as primary biogenic aerosols (PBA).
PBA particles and components are, by definition, directly emitted from the biosphere to the atmosphere. Pollen (>10 µm), bacteria (∼1 µm), fungal, algae, moss and fern 15 spores (∼10 µm), viruses (<0.1 µm) and fragments of animals and plants constitute PBA particles Jaenicke, 1995, 2000;Artaxo and Hansson, 1995;Bauer et al., 2005;Després et al., 2007). PBA components, which are composed of fragments of biological particles (pollen, bacteria, etc.) can be found on other types of aerosol particles such as sea spray, dust etc. (Kellogg and Griffin, 2006). Large PBA particles such as pollen, fern spores and large fungal spores belong to the coarse mode of air particle matter, but PBA particles and components are also found in intermediate and fine fractions of aerosol particles (e.g., fungal spores, small fragments of animals and plants, bacteria, viruses, carbohydrates, proteins, waxes, biopolymers, ions) (Taylor et al., 2004;Zhang and Anastasio, 2003;Franze et al., 2005). Recent 25 investigations suggest that cellular protein-containing particles are a major fraction of atmospheric aerosols with roughly 1000 Tg yr −1 compared to 3300 Tg yr −1 for sea salt and 2000 Tg yr −1 for mineral dust (Jaenicke, 2005). Recent investigations by Bauer 844 the order of 10 4 m −3 and 1 µg m −3 in continental boundary layer air (Bauer et al., 2007;Elbert et al., 2007;and references therein). Based on the average concentrations of spores and molecular tracers, a global emission rate on the order of 17 Tg yr −1 , has been estimated for actively wet discharged spores from Basidiomycota, which is one of the major groups of fungi in the biosphere. The global average atmospheric abundance 10 and emission rate of total fungal spores, including wet and dry discharged species, are estimated to be higher by a factor of about three, i.e. on the order of 50 Tg yr −1 (Elbert et al., 2007). Comparisons with estimated rates of emission and formation of other major types of organic aerosol (47 Tg yr −1 of anthropogenic primary organic aerosol; 12-70 Tg yr −1 of secondary organic aerosol) indicate that emissions from fungi should 15 be taken into account as a significant global source of organic aerosol. The effects of fungal spores and related chemical components might be particularly important in tropical regions, where both physicochemical processes in the atmosphere and biological activity at the Earth's surface are particularly intense, and where the abundance of fungal spores and related chemical compounds are typically higher than in extra-tropical 20 regions.
The numerical density of micro-organisms at a specific site is highly variable, and is dependent on several factors. The most important factor is due to the season with an increase during the warm period (Di Gorgio et al., 1996;Katial et al., 1997). At a shorter time period, a diurnal cycle is also observed with a maximum around midday EGU ally more bacteria are observed carried by urban-influenced air masses than rural and coastal ones (Mancinelli and Shulls, 1978;Shaffer and Lighthart, 1997). This point is developed in detail in Sect. 4.
Several studies have found biopolymers like cellulose and proteins (molecular mass ≫1 kDa) and other compounds with relatively high molecular mass (≫100 Da) such as 5 humic-like substances (HULIS) (analogous to humic substances in soil, surface water and groundwater) in air particulate matter as well as in fog, cloud, and rain water (Zappoli et al., 1999;Puxbaum and Tenze-Kunit, 2003;Pöschl, 2005;Graber and Rudich, 2006;Sanchez-Ochoa et al., 2007). Indeed, biopolymers and other PBA components are likely to account for a large proportion of OC. Proteins, amino acids, and related 10 amino compounds were found to account for up to ∼10% of fine particulate matter and to have a strong influence on the microstructure and water uptake of aerosol particles (Saxena and Hildemann, 1996;Miguel et al., 1999;Anastasio, 2001, 2003;Mace et al., 2003a, b, c;Mikhailov et al., 2004;Kuznetsova et al., 2005;Matsumoto and Uematsu, 2005;Pöschl, 2005;McFiggans et al., 2006). In the atmosphere, they un- 15 dergo chemical reactions such as oxidation, nitration, photolysis and hydrolysis, which can modify their optical, chemical and biological properties and eventually result in the formation of HULIS (Pöschl, 2005;Fuzzi et al., 2006). However, the current understanding of aerosols transformations and the interaction of gases with particles is very limited since they involve multiple physico-chemical processes such as mass transport, 20 phase transition and chemical reactions at the interface or in the bulk of the different compartments: gas, liquid and solid phases (see Sect. 4). Bauer et al. (2002) showed that average mass concentrations of bacteria only amounted to 0.01% of OC in cloud water, snow, rain and aerosol samples. But, although the mass concentrations of bacteria seems low compared to other organic 25 aerosols, their numbers range from ∼10 3 to ∼10 5 cells m −3 (Sattler et al., 2001;Bauer et al., 2002;Amato et al., 2005;Casareto et al., 2006). Bacterial concentrations in cloud water seem to depend mainly on the altitude and location of sampling. Sattler et al. (2001) estimated the average number of bacteria in cloud water 846 EGU to be ∼1.5×10 3 mL −1 at Mt Sonnblick (3106 m a.s.l., Austria) during spring 1997. At Mt Rax (1644 m a.s.l., Austria) during spring 1999 and 2000, the average number of bacteria in cloud water was about 2.0×10 4 cells mL −1 with an air equivalent concentration estimated to be 5.  Fuzzi et al. (1997), analyzing fog droplets sampled in the Pô Valley, identified three bacterial genera among the culturable fraction (Pseudomonas, Bacillus and Acinetobacter ) whereas Amato et al. (2005Amato et al. ( , 2007a) observed a more diversified culturable population consisting of many bacterial genera, distributed among various phyla or sub-phyla (α-, 10 β-and γ-Proteobacteria, Bacteroidetes, Actinobacteria and Firmicutes), as well as an important diversity among the culturable fungal and yeast populations. Some isolates present high genetic similarities with strains isolated from cold regions and aquatic environments as well as with plant pathogens.
3 Influence of micro-organisms on cloud chemistry and multiphase processes 15 The atmospheric liquid phase is a very complex mixture undergoing numerous chemical and photochemical reactions (process 1 in Fig. 1), dissolving soluble molecules of the particulate phase (process 2 in Fig. 1) and exchanging chemical molecules with the gas phase and the condensed phase (process 3 in Fig. 1). Liquid phase chemistry has a potential impact on the oxidizing capacity of the atmosphere and on the greenhouse 20 effect via an indirect effect of aerosol particles on the Earth's radiative budget. Chemical reactions in liquid phase modify the amount of radicals. These radicals drive the oxidizing power of the atmosphere. For example, iron in aerosols plays a major role in the concentration of radicals in cloud droplets (Deguillaume et al., 2005). As many organic compounds are complexing agents with an iron metal centre, the organo-metallic 25 chemistry can be important in controlling the amount of radicals in clouds (Deguillaume et al., 2005;Parazols et al., 2006). During the lifetime of a cloud, cloud chemistry leads 847 to the formation of new chemical molecules with relatively low volatility such as inorganic and organic acids, which can modify the physico-chemical properties of aerosol particles (Hegg, 2001) and lead to the formation of secondary organic aerosols (Gelencsér and Varga, 2005). Once the cloud droplet has evaporated, the resulting aerosol particle presents a different chemical composition compared to the initial CCN. It tends 5 to be even more hygroscopic and, thus, more likely to take part in the formation of a new cloud droplet (Feingold and Kreidenweis, 2002). Several studies have discussed the potential role of living bioaerosols in modifying the multiphase chemistry of the atmosphere via metabolic activity, by using labelled and non-labelled chemical species as nutrients (Ariya et al., 2002;Amato et al., 2005Amato et al., , 10 2007cFuzzi et al., 2006). Micro-organisms interact with nitrogen, sulfur and organic compounds in cloud water potentially modifying organic chemistry which is currently subject to high uncertainties. Organic compounds directly or indirectly interact with the aqueous chemistry of radicals, non metal and metal ions and oxidant pools (Ervens et al., 2003;Tilgner et al., 2005;Deguillaume et al., 2005). Organic chemistry inside clouds may consequently modify the chemical budget of both aqueous and gaseous phases . Finally, organic oxidation inside clouds can also be a potentially important source of SOA in the atmosphere after the evaporation of cloud droplets (Blando and Turpin, 2000;Ervens et al., 2004;Carlton et al., 2006). In particular, recent studies show that chemical reactivity in cloud droplets leads to the formation 20 of dicarboxylic acids (DCA), such as oxalic acid which are commonly found in the particulate phase in the atmosphere (Ervens et al., 2004;Sorooshian et al., 2006). DCAs dominate in atmospheric cloud water representing between 10% and more than 70% of the total dissolved organic carbon (Marinoni et al., 2004).
Micro-organisms are incorporated into cloud droplets and raindrops by nucleation 25 scavenging as they have CCN or IN potential (e.g., Lee et al., 2002;Bauer et al., 2003;Möhler et al., 2007) or by washout processes. Some investigations clearly show that most of these micro-organisms are able to develop at low temperatures (between −5 and 5 • C) encountered in clouds. Furthermore, measurements of concentrations of EGU adenosine triphosphate (ATP) in cloud water indicate that most micro-organisms are still metabolically active (Amato et al., 2007d). Moreover, metabolic pathways involving micro-organisms can show some similarities with chemical reactions common to atmospheric radical chemistry as described below. However, the nature and the extent of such processes at molecular levels still have to be evaluated. Furthermore, 5 biodegradation processes are expected to be sensitive to atmospheric conditions. The following discussion will focus on bacteria and fungi and on their potential influence on the chemistry of the atmosphere for which some studies are now available. Herlihy et al. (1967) first studied the degradation of formic and acetic acid by bacteria in rainwater. Ariya et al. (2002) observed evidence for chemical reactions induced 10 by bioaerosols in solutions containing DCA. They observed that several of the DCA detected in the atmosphere can efficiently be degraded by airborne bacteria and fungi. Estimated degradation lifetimes (few days) are also comparable to major atmospheric oxidants (OH, O 3 and HO 2 ) but highly variable as a function of environmental conditions. They also found that the observed microbiological transformation varied as a 15 function of the organic acid used and of the type of micro-organism. They analysed products of the reaction mixture which are non-or only slightly toxic compounds (acetamide, butanoic and propionic acids, etc.) as well as highly toxic and carcinogenic ones (kojic acid, aflatoxin B1). The isotopic 13 C NMR studies clearly demonstrated that airborne microbes can metabolise and chemically transform the DCA to other organic 20 compounds. During these biodegradation processes, molecules can be transformed to more volatile compounds rendering the products available for recycling back into the atmosphere and also to more toxic/pathogenic chemicals. Amato et al. (2005) show that bacterial strains present in cloud water (Puy de Dôme Mountain) contain the enzymatic equipment required for the degradation of some atmospheric compounds. Cloud EGU to transform numerous atmospheric compounds. From this study based on 60 microbial strains, they established that, depending on atmospheric conditions, the microbial component is likely to be a sink for carboxylic acids like acetate, formate, lactate and succinate, and also of methanol and formaldehyde. They also can represent a source, with example e.g. the transformation of succinate into fumarate, and of lactate into 5 pyruvate. All those compounds can be found in relatively large concentrations in cloud water (Puxbaum et al., 1988;Grosjean et al., 1989;Granby et al., 1997;Suzuki et al., 1998;Löflund et al., 2002;Marinoni et al., 2004;Van Pinxteren et al., 2005) and play a major role in atmospheric chemistry (Ervens et al., 2005;. Their results also indicate preferential metabolic routes for some microbial groups. For instance, Staphylococcus spp. are specialized in the transformation of C 1 compounds. This means that depending on the communities present in a given cloud, microbial participation in chemistry would be different.

EGU
This preliminary result indicates that common radical chemistry considered in a numerical cloud model could compete with kinetic biological degradation as already shown by Ariya et al. (2002), but as they noted, there are substantial differences in the microbiological activity for different environmental conditions, and thus the direct observation without adjustments to relevant environmental conditions (such as T, pH, 5 irradiation, nutrients and other existing chemicals) can not yield meaningful conclusions. Therefore, the contribution of biodegradation processes has to be evaluated in a cloud system where microphysical processes, photochemical reactivity, dissolution processes and biodegradation will compete and modify the general chemistry of the troposphere. adopting several survival mechanisms; for instance, becoming dormant, changing size and forming spores (Price and Sowers, 2004). In cold temperatures they can adapt by reducing their cell size and the thickness of their capsular polysaccharide, by changing their fatty acid and phospholipids composition, or through energy releasing catalytic redox reactions in ice or in permafrost thereby leading to the creation of habitable mi- 25 croenvironments in ice crystals (Thomas and Dieckmann, 2002;Kos and Ariya, 2008 2 ). In addition to temperature, other environmental limiting factors for cell activity in the 851 metabolism can be reduced and cell development can be stopped. During cell division, irradiation, especially UV, can cause significant damage such as mutation of DNA which can lead to cell death (Sommagura et al., 1997). In response to this stressor, bacteria can also produce a diverse suite of pigments (scytonemins, carotenoids, phycobiliproteins, chlorophylls etc.) that absorb from the near UV-B to red wavelengths 10 (Mueller et al., 2005). Sensitivity to irradiation strongly depends on the type of strains and also on altitude since intensity of the UV radiation increases with the altitude. Micro-organisms in the atmosphere can suffer from dehydration and desiccation leading to the modification of vital structure and ultimately to death (Zentner, 1966). Cell aggregation can diminish sensitivity to desiccation as cells in aggregates protect 15 each other. When incorporated into cloud or fog droplets, microbial cells would not suffer from desiccation and would be transported over greater distances than under clear sky conditions. Thus, clouds and fog represent relatively favourable environments for micro-organisms thereby sustaining their activity and viability. Consequently, fog and clouds represent media in which microbial cells can potentially divide as proposed by In the atmospheric aqueous phase, the pH is a crucial parameter (Ariya et al., 2002) for cell viability, partially as cells have to adjust progressively to an acidic medium (Koutsoumanis and Sofos, 2004). The effect of pH is also linked to other parameters such 25 as oxygenation; hence, the responses to oxidative and acidic stress are similar. The concentration of micro-organisms is also clearly linked to the chemical composition of cloud water. Recent investigations led by Amato et al. (2007d) show that the number of micro-organisms in cloud water increases with increasing oceanic contribution and 852 EGU decreases with increasing anthropic influence. It suggests that micro-organisms originating from the ocean are preferably integrated in cloud droplets compared to those from continental sources. They also suggest that an eventual multiplication of cells in cloud droplets can be disturbed due to the toxicity of certain polluted clouds.
Thus, we have to consider positive and negative feed-backs amongst various envi-5 ronmental factors with regard to the viability and the metabolic state of airborne microorganisms.
Apart from the viability of micro-organisms, multiphase chemical reactions in the atmosphere can also have influences on the effects of PBA on climate and public health. For example, protein molecules can be efficiently nitrated in polluted urban air under 10 summer smog conditions or in mixtures of NO 2 and O 3 (Franze et al., 2005). The nitration reaction leads to the formation of 3-nitrotyrosine residues in the polypeptide chain of the protein molecule, i.e. to the introduction of an NO 2 group next to the OH group on the phenyl ring of the aromatic amino acid tyrosine (2-amino-3-(4-hydroxyphenyl)propanoic acid). This posttranslational modification of proteins is likely to trigger im- 15 mune reactions and provides a molecular rationale for the promotion of allergies by traffic-related air pollution, which is confirmed by recent and ongoing biomedical investigations (Gruijthuijsen et al., 2006).
Nitration and other chemical modifications may not only influence the biological properties (allergenicity, toxicity, viability) of PBA. Chemical transformations can also influ-20 ence the PBA physico-chemical properties which determine their effects on climate (optical parameters; hygroscopicity, surfactant activity, and CCN activity; crystallinity and INA (Ice Nucleation Active) activity). For example, nitration renders colourless proteins yellow (nitrotyrosine acts as a chromophore) thereby potentially modifying reflectivity of PBA. It may also change the molecular and crystalline structures of proteins 25 which serve as active sites for ice nucleation on INA bacteria. The amino acid sequence at the water binding site on the bacterial IFN active protein contains tyrosine. Folding of the protein leading to the formation of β-sheets occurs at the gly-tyr-gly site in the amino acid sequence in the octapeptide repeated zone of this protein (Gazit, 853 EGU 2002). Formation of 3-nitrotyrosine residues as described above might reduce the water-binding capacity of this site or disrupt the protein tertiary structure necessary for INA.

Research perspectives and suggestions for future studies
Current knowledge regarding the relationship between atmospheric chemistry and 5 bioaerosols is not very advanced. Bioaerosol chemistry and the reciprocal influence of environmental conditions on bioaerosols with regard to the effect on the atmosphere are very complex subjects at the early stages of their development. This young scientific issue gives rise to many questions.
Firstly, regarding the influence of bioaerosols in dry air, the major uncertainty concerns to the modification of the physico-chemical properties of aerosols: (1) Does the modification of chemical composition of organic aerosols due to contact/collision with bioaerosols along with particle size and density significantly influence their CCN efficiency?; (2) Do biopolymers present on carbonaceous aerosols undergo heterogeneous chemistry leading to possible formation of macromolecules such as HULIS 15 substances? Secondly, recent studies have clearly demonstrated that viable micro-organisms present in the atmosphere can contribute to atmospheric chemistry through degradation processes, as well as chemical change due to the release or desorption of molecules from microbiological entities (Cote et al., 2008). Clouds, fog and rain seem 20 to represent media where the biological activity is significant due to the protection offered from desiccation. Micro-organisms present in fog and clouds use chemical compounds in the aqueous solution as nutrients and therefore perturb the radical chemistry of clouds.
Different approaches are available for studying the effect of bioaerosols on atmo-EGU species. In-situ measurements aim at better understanding the chemical processes within clouds through biological, chemical, microphysical, and meteorological measurements and complement laboratory studies in providing more general information on the environmental conditions. The main difficulty is to coordinate the types of measurements in time and space. Process models are good tools to integrate laboratory 5 measurements while transport models allow studying various reaction pathways, complex interactions between biological, microphysical, chemical, radiative, and dynamic processes in the framework of the whole cloud system. Then, models are also helpful in running sensitivity tests to study the influence of environmental parameters on multiphase chemistry. These approaches need to be used and developed in synergy in 10 order to quantify the effect of bioaerosols in the frame of atmospheric chemistry: (1) Further fundamental laboratory chemical-biological research is mandatory to provide answers concerning the kinetics and mechanisms of (bio)chemical transformation involving enzymatic and non-enzymatic processes.
Firstly, many future laboratory investigations are needed for studying the contribu- 15 tion of bioaerosols to organic aerosols and its role on climate. Most of the techniques for monitoring bioaerosols are offline and highly labour intensive (for example, cultivation, staining, fluorescence and electron microscopy, enzyme and immunoassays, DNA analysis etc.) (Maron et al., 2005). New challenging techniques such as aerosol mass spectroscopy (Noble et al., 2000;van Wuijckhuijse et al., 2005;Kleefsman et al., 2007) 20 or aerodynamic sizing and fluorescence spectroscopy (Ho, 2002) allow online monitoring of bioaerosols but still improvements in capturing detailed chemical characteristics at sufficient detection limits are needed. Secondly, further studies should also include understanding of kinetics of biological and chemical processes. Laboratory studies are needed to determine the kinet- 25 ics and mechanisms of the chemical reactions involving organics and molecules such as nitrogenous compounds and trace metals that can potentially interact with microorganisms. Micro-organism strains isolated in the aqueous phase of clouds need to be selectively isolated or identified in terms of their ability to degrade chemical compounds EGU as well as for their ability to be active under extreme conditions (bacteria with pigments, psychrotrophic bacteria, spore-discharging micro-organisms etc.). The degradation ability of these strains will be followed by the laboratory with the objective of determining kinetic constants which will be incorporated in numerical models. The enzymatic activity of micro-organisms has also to be evaluated as a function of numerous factors 5 such as pH, concentrations of oxidizing agents (H 2 O 2 , iron) and actinic flux. Therefore, parameterization of the relationships between these sensitive parameters and the metabolic activity of micro-organisms has to be developed.
(2) As mentioned before, the biological activities of micro-organisms are very dependent on environmental conditions. Long-term and detailed observations of the micro-10 bial communities (total number, characterization of strains, metabolic activity etc.) are necessary to evaluate the diurnal and seasonal variations of structure and activity of cells as a function of environmental conditions such as humidity, light, temperature etc. Moreover, this climatology of the population of micro-organisms needs to be correlated with the chemical characterization of the sampled air mass. 15 More specifically, in atmospheric water, similar long-term observations need to be performed in order to characterize the living microbial biota as a function of encountered physico-chemical conditions. The laboratory investigations detailed above combined with these in-situ measurements of chemical composition of water will provide a real overall picture of the capacity of micro-organisms to develop under the nutritional 20 conditions offered by atmospheric water.
(3) Finally, modelling will allow us to evaluate the relative contribution of bioaerosols to the chemistry and physics of the atmosphere, and their possible impact on climate. A complete treatment of this interaction requires a rather complex model concerning cloud microphysics (cloud droplets activation, ice activation and multiplication), cloud 25 chemistry, tropospheric chemistry including transport, etc, involving the representation of different scales. For instance, cloud microphysics and chemistry are treated locally in a cloud while tropospheric chemistry and transport will be addressed by the way of 3-D meso-scale simulations. A possible approach to consider the effect of micro-organisms 856 EGU on atmospheric chemistry, modelling could be the following: (a) Numerical models describing multiphase cloud chemistry need to consider biological processes in order to evaluate the role of micro-organisms in cloud chemistry. Even though theoretical, models allow us to study -within the cloud system -the chemical pathways, the complex interactions between microphysics, chemistry 5 and dynamics (i.e. transport) and the influence of environmental parameters on cloud chemistry. At the same time, the structure of numerical models offers the possibility to perform sensitivity analysis on uncertain parameters. Biological degradation processes have to be implemented in models in order to compare explicit photochemistry with biological degradation processes. The whole complexity of biological processes 10 including all the metabolic pathways cannot be implemented in these models. The objective will be rather to evaluate the global potential of micro-organisms present in cloud water on the degradation of relevant chemical compounds. Laboratory studies, as indicated above, need to develop biodegradation parameterizations as a function of environmental conditions. Finally, simulations of various cloud events under different 15 environmental conditions will allow us to generalize the effect of micro-organisms on cloud chemistry.
(b) Sensitivity analysis performed with models will allow development of parameterizations of the biological degradation of chemical compounds under specific environmental conditions. In 3-dimensionnal models, micro-organisms will be considered as 20 one category of aerosol particles and kinetic constants of biological degradation will be introduced as a function of environmental conditions. These formulations will be first implemented in regional models such as the RAMS model (Cotton et al., 2003) or the Meso-NH model (Lafore et al., 1998) and afterwards in more global models such as the MOCAGE model (Josse et al., 2004) or the ECHAM5/MESSy1 model (Tost et al., 25 2007).