Interactive comment on “ Global fungal spore emissions , review and synthesis of literature data ” by T . N . Dallafior and A . Sesartic

and p. 8450, lines 12-13: The referee’s comment should correctly refer to p. 8449, lines 12-13. Exclusion criteria were lack of information about meaC5455 surement sites, biomes, measurement period, only absolute spore counts instead of concentrations, and petri dish samplings. This information was now added to the paper. Technical corrections p. 8447, line 12: corrected p. 8450, lines 14-15: corrected p. 8451, lines 17-20 and Fig. 2: We had JSBACH ecosystem data available and thus used it to create the map. However, JSBACH uses a slightly different division of ecosystems than Olson (2001) does. The observed fungal spore concentrations were attributed to the corresponding ecosystem depending on the ecosystem at the measurement location (see also our remark on the calculation of flux estimates). p. 8453, line 2: corrected p. 8454, line 6: corrected p. 8454, lines 26-27: corrected p. 8455, lines 14-15: Kieft Ruscetti (1990) found that proteins are crucial for the ice nucleation activity of the lichen fungus Rhizoplaca chrysoleuca. Similar observations have been done for bacteria (Kajava Lindow, 1993). Therefore, it is not the viability of a cell that does affect the CCN/IN activity of a fungal spore, but whether the ice nucleation active proteins on its surface are denatured or not. If the protein C5456 conformation is intact, it can trigger ice nucleation, regardless of the cell’s viability. However, it is important that the cell has been alive at one point in order to produce the ice nucleation proteins in the first place. This information has been added to the manuscript. p. 8457, lines 15-18: The mass flux given for Elbert et al. (2007) was corrected in order to account for total global emissions of fungal spores, and not only those of ascomycota, as well as the fact that Elbert et al. (2007) give only emissions over land, and not a global average. The newly calculated fungal spore mass emission flux according to Elbert et al. (2007) accounts to 2.26× 10−11kg m−2 s−1. p. 8458, lines 25-26: corrected Appendix, Table A1: The blank fields denote information which is missing from the original publications. It was now replaced with n/a denoting “not available”. Additionally, it is now noted which articles consider not the total number of spores but the sum of a set of species. Appendix, Table A2: The table was sorted primarily by ecosystem and then by author, according to the referee’s suggestions. Additionally, a column was added noting whether the analysis was done using culture or culture-independent techniques. References Elbert, W. et al. Contribution of fungi to primary biogenic aerosols in the atmosphere: wet and dry discharged spores, carbohydrates, and inorganic ions. Atmos. Chem. C5457 Phys. 7, 4569-4588, (2007). Kajava, A.V., and Lindow, S.E. A model of the three-dimensional structure of ice nucleation proteins. J. Mol. Biol. 232, 709-717 (1993). Kieft, T. L., and Ruscetti, T. Characterization of biological ice nuclei from a lichen. J. Bacterial. 172, 3519-3523 (1990). Winiwarter, W. et al. Quantifying emissions of primary biological aerosol particle mass in Europe, Atmos. Environ., 43, 1403-1409, (2009). Interactive comment on Biogeosciences Discuss., 7, 8445, 2010.


Introduction
Fungal spores are part of the bioaerosol population in our atmosphere which also comprises components such as pollen, bacteria or viruses.Interest in bioaerosols is mainly related to their health effects, agriculture, ice nucleation and cloud droplet activation or atmospheric chemistry (Ariya et al., 2009).In the present study, the focus lies on fungal aerosols.
Measurements of fungal aerosols report average ground level concentrations of around 10 000-50 000 spores m −3 , sometimes even exceeding 200 000 spores m −3 (Levetin, 1995).This is two orders of magnitude higher than observed Correspondence to: A. Sesartic (ana.sesartic@env.ethz.ch)peak pollen concentrations (1000-2000 grains m −3 ) ( Mandrioli, 1998).Froehlich-Nowoisky et al. (2009) also state that up to 45 % of the coarse particle mass in tropical rainforest air consists of fungal spores.Elbert et al. (2007) found that fungal spores are the main contributor to the bioaerosol mass in the Amazon basin.Simulations conducted by Heald and Spracklen (2009) came to the conclusion that 23 % of all primary emissions of organic aerosol are of fungal origin.Investigations of Bauer et al. (2002b) have shown that "5.8 % of the organic carbon in the coarse aerosol mode" was due to fungal spores and bacteria.When sampling bioaerosols at the Rothampsted Experimental Station in the south of England, Gregory (1978) found that they mainly consisted of fungal spores; nearly half of a fair weather sample consisted of Cladosporium spores (a mould) and only one percent of all bioaerosols were plant pollen.These findings show that fungal spores are an non-negligible part of the atmospheric aerosol.Fungi are even found in marine habitats where they are important decomposers of plant substrates (Hyde et al., 1998).As marine fungal spore emission observations are lacking so far, this biome has not been further taken account of in this review.
The primary source of fungal aerosols are plants (Burgess, 2002), soil, litter and decaying organic matter (Heald and Spracklen, 2009).Release mechanisms of fungal spores are numerous and vary from species to species (Elbert et al., 2007;Gregory, 1967Gregory, , 1973;;Levetin, 1995;Jones and Harrison, 2004;Madelin, 1994;Hirst, 1953).Generally, release of spores is highly dependent on meteorological factors.Some require rather humid conditions whereas others favour dry and windy conditions for spore release.Several studies have been conducted on the relationship between meteorological factors and spore concentrations.Significant correlations between spore counts and wind speeds could be found (Glikson et al., 1995) as well as a positive correlation of Alternaria spore counts with temperature (Burch and Levetin, 2002).
A. Sesartic and T. N. Dallafior: Global fungal spore emissions Stepalska and Wolek (2009) on the other hand could not find a significant correlation of spore concentrations with weather conditions for most species investigated in their study.It is hence difficult to predict which and how many spores are released according to weather conditions.
As for their transport behaviour, most of the spores do not travel very long distances.As calculations by Gregory (1962) have shown only a fraction of about 10 % of all released fungal spores is transported farther away than 100m.This fraction is called the "escape fraction".The measured concentrations are a blend of local emissions and advected spores.It is difficult to distinguish between those two groups.A possible distinction criterion might be the size or shape since larger particles are deposited more easily than smaller ones.However, Heald and Spracklen (2009) note that the larger size fraction is less well investigated due to measurement device constraints.The farther away the sampling device is from the spore source, the more is the measured concentration influenced by deposition and other processes.This can lead to devices in immediate proximity to the ground measuring the actually emitted spore numbers, whereas other devices on higher levels might measure the escape fraction only.As most observational data was taken further away from the ground, we assume our estimate to represent the escape fraction.
Wind speed, temperature, atmospheric pressure or precipitation are important conditions determining transport and deposition of the dispersed aerosols (Hirst et al., 1967).There is evidence that fungal spores can also be transported over long distances (Griffin et al., 2006(Griffin et al., , 2001;;Prospero et al., 2005) before they are deposited either due to gravity, washout by rain or impaction (Gregory, 1967).Among others, Prospero et al. (2005) found fungal spores originating from the African desert to influence the prevailing fungal spore concentrations on the Virgin Islands in the Caribbean.
Some fungal spores can act as very effective ice nucleators.Lichen were found to nucleate ice at temperatures higher than −8 • C and some even at temperatures higher than −5 • C (Kieft, 1988).The lichen fungus Rhizoplaca chrysoleuca was even found to be an active ice nucleus at temperatures as high as −2 • C (Kieft, 1988;Kieft and Ruscetti, 1990).To date, only a few fungus species have been found to be active ice nucleators: besides the above mentioned lichen these are Fusarium avenaceum and Fusarium acuminatum (Pouleur et al., 1992).The ice nucleating activity of F. avenaceum is comparable to that of the bacterium Pseudomonas sp.(Pouleur et al., 1992).In contrast to those findings is the recent research by Iannone et al. (2011) that showed poor ice nucleation ability of Cladosporium spores, with immersion freezing starting at −28.5 • C.This might be due to the spores being coated with hydrophobic proteins that are widespread in filamentous fungi such as Cladosporium sp.Additionally, fungal aerosols are likely to be effective cloud condensation nuclei, but data on behalf of this is still scarce.
Recent field measurements have highlighted the importance of bioaerosols as ice nucleators in the atmosphere, e.g.Pratt et al. (2009) and Prenni (2009).Relying on the above evidence, there is probably a link between meteorological conditions and fungal spores as well.On the one hand, fungal spores acting as ice nuclei might influence cloud and precipitation formation process, as has already been proposed by Morris et al. (2004) in general for biological ice nuclei.On the other hand, changes in climatic conditions also alter the meteorological situation on a smaller time scale which in turn might influence fungal spore release as well as transport according to the respective release mechanism.These possible interactions with the weather and climate system as well as the fact that fungi are one of the major contributors to global bioaerosols makes it crucial to gain more knowledge about the circumstances and amounts in which they are emitted as well as their transport behaviour.Many studies have already focused on sampling of fungal spores in order to estimate their concentration in the atmosphere.However, a standardised procedure in order to do so is still missing which leads to very heterogeneous and hardly comparable results.
The goal of this paper is to review the available literature data on fungal spores and estimate global fungal spore emissions by biome area.Available literature is reviewed and data provided used to derive the respective fungal spore fluxes for major ecoregions.Moreover, measurement methods are reviewed and discussed.

Data and methods
A review of available fungal spore concentration data has been undertaken.Fungal spore concentration data have been assigned to an ecosystem and converted to surface number and mass fluxes.More than 150 studies have been reviewed of which 35 have been found to contain data relevant for this study, and thus were taken into account for flux calculations.Exclusion criteria were a lack of information about measurement sites, biomes, measurement period, only absolute spore counts considered instead of concentrations, and petri dish samplings.
The biome areas by Olson et al. (2001) have been used for ecosystem classification.But since data points were not sufficiently dense, broader definitions of the respective biomes had to be taken.These ecoregions covered tropical and nontropical forests, shrub, grass, crop, tundra and land ice (data based on (World Resource Institute, 2003a,b)).Attribution of the respective ecoregions by Olson et al. (2001) to the biome areas used here has been done according to Table 1.The biome "crop" was used when the studies mentioned close proximity to agriculture land.Effects of urban environments were not taken into account.Biome area data has been derived from World Resource Institute (2003a,b).Figure 1 shows the global distribution of available measurement data.(Olson et al., 2001)   Fluxes have been calculated based on fungal spore concentrations, the height ( z) at which the spore concentration has been measured and the time ( t) necessary for an uniform mixing of the atmospheric layer between ground and the height of the measurement: z has been assumed to be 10 m.This is the average sampling height at which fungal spore concentrations have been measured, considering that sometimes samples have been taken at ground level, sometimes above tree level or on top of buildings.The height of 10 m for z is justified because it is at the top of the Prandtl layer in which fluxes between the ground and the atmosphere are constant with height (Colombe Siegenthaler-LeDrian and Tanja Stanelle, personal communication, 2011).We assume the concentration measured at this level to be in a steady state between spore emissions and dry deposition.
As the density of fungal spores is slightly lower than that of air, they are lifted with thermals and their dry deposition can be neglected for this offline calculation.This is supported by a terminal velocity of v = 6.12 × 10 −7 ms −1 which can be derived assuming a fungal spore diameter of 10 µm and a density of 1kgm −3 .
The timescale for turbulent diffusion in the 10 m layer can be obtained from with z = 10 m and turbulence diffusion coefficient K turb = 10 5 cm −2 s −1 (Jacob, 1999).
Alternative approaches to calculate fungus fluxes have been derived for instance by Heald and Spracklen (2009) and used by Hoose et al. (2010): where LAI is the leaf area index, a measure for the leaf area per surface area and q is the specific humidity.The second of the two flux calculation methods has not been used since the LAI is not necessarily a good measure of spore emissions, as it would create a bias towards too low fungal spore fluxes in regions where the LAI is very low, such as grasslands.Spore masses have been derived by Winiwarter et al. (2009): Assuming an average carbon content of 13 pg C per spore (Bauer et al., 2002a,b), a water content of 20 % per spore and 50 % C per dry mass (Sedlbauer and Krus, 2001) they determined a fungal spore mass of 33 pg.Calculations by Elbert et al. (2007) resulted in remarkably higher values: assuming a mass density of about 1 gcm −3 and a volume equivalent diameter of about 7 µm, the average mass of wet spore discharged by Ascomycota would be around 200 pg.Assuming the same density for Basidiomycota, Elbert et al. (2007) derived a mass of 65 pg.
For calculation of mass fluxes in this paper, the average mass of 33 pg per spore has been assumed.The number and mass fluxes are listed once per second and once per second and square meter of the specific biome.Global averages are derived by taking the mean of the respective values for each biome area.
The following assumptions have been made for the flux calculations: -Seasonal or daily cycles do not influence the measured spore concentrations.
-There is no difference between colony forming units (CFU) and total counts.
-Due to data scarcity the measurements taken at specific ecosystems were summarised in broader categories (cf.Table 1).
-Similar assumptions have been made for the other biome regions.
Best estimates have been calculated from all average spore concentration data available.Lau et al. (2006) are the only researchers that have expressed their data with the geometric mean instead of the arithmetic mean.Since the difference between the geometric mean and the arithmetic mean are small, they have been treated in the same manner.It should be noted that for the low and high estimates of crop emissions, only one measurement result was available.Where no average concentrations were provided, the average between the maximum and minimum spore concentrations (if available) has been taken instead.The high and low estimates are the averages of all minimum and maximum spore counts (where available).In the optimum case, total spore counts were provided covering all identifiable species.However, in some studies, only certain genera were investigated.Where spore counts had been split according to species, the sum over all has been taken in order to get as close as possible to the number of total counts.Table 2 shows the aggregated number fluxes of fungal spores per square metre of biome and second, while Table 3 shows the respective mass fluxes in kg per square metre and second.The biome area with the largest fungal spore flux is crop followed by shrub, tropical and non-tropical forest, grassland and tundra in descending order.This can also be  seen in Fig. 2, showing the fungal number flux, which was produced by combining the best estimates of reviewed fungal spore data with the plant functional types from the JSBACH dynamic vegetation model (Raddatz et al., 2007).
As for the prevalence of different spore genera, most of the studies agree that Cladosporium make up a very dominant part of the fungal spore air spora, e.g.Sakiyan and Inceoglu (2003); Mallo et al. (2010).Other species such as Alternaria, Aspergillus, Ganoderma, Agaricus, Coprinus, Leptosphaeria or smuts and rusts have been considered as important constituents, but usually showed much lower concentrations than Cladosporium (e.g.Sakiyan and Inceoglu, 2003;Mallo et al., 2010).Based on literature studies, Goncalvez et al. (2010) concluded that Asperigllus, Alternaria and Penicillium were predominant in hot climates, whereas Cladosporium spores were found to be most abundant in temperate climatic regimes.

Review of measurement methods
The simplest method to measure airborne fungal spores is to expose sticky surfaces or petri dishes and then count either all detectable fungal spores or, in the case of petri dishes, the colony forming units (CFU).This method has for instance been used by Bhati and Gaur (1979) or Abu-Dieyeh et al. (2010).The obtained sample results are useful in order to get a qualitative impression of the composition of the prevailing air spora.However, Gregory (1952) pointed out that these data imply the "tacit assumption that the relation between the number of particles suspended in the air flowing over the surface and the number deposited on the surface is known".Gregory (1952) further argue that these simple rules would only apply for still air.According to the authors of the study, the efficiency of a sticky surface to collect fungal spores rather varies with wind speed and subsequently number concentrations per unit volume are difficult to obtain.Gregory (1952) note, that a good sampling device draws in "a known volume of air without altering its spore content, removes all particles over the 2-100 µm size range, and leaves them in a form in which they can be examined, counted and classified".Hirst (1952) described the features a measurement device should have in order to give useful results: Besides the ability of assessing the spore concentration per unit volume of air, it should also be possible to measure in distinct time intervals to better correlate concentrations with meteorological conditions.He designed a spore trap that was able to suck in air and subsequently impact the contained spores on a sticky surface.The principle of suction increases, as Hirst (1952) argues, the efficiency of filtration and impaction.However, he also found that this does not guarantee for the air masses entering the trap being representative of the actual spore load (collection efficiency).To counteract this problem, Hirst (1952) proposed that air should be sampled isokinetically and that the orifice should always be directed into the airstream.In order to get a distinction in time, a sticky slide is moved slowly past the orifice (Hirst, 1952).As suction rates are constant, trapping efficiencies change with wind speeds (Hirst, 1953).
Recent methods take advantage of biochemical properties of fungal spores.Elbert et al. (2007) for instance use mannitol (a component of fungal cell membranes) as a fungal biomarker to determine their prevalence in air masses.Bauer et al. (2008) rely on other biochemical tracers such as mannitol or arabitol to measure the fungal spore content in the precipitation.
A review of the available studies of spore concentrations showed that the measurement methods applied were diverse.As outlined above, data from measurements with petri dishes have not been included in the flux calculations for this paper due to the above named reasons.The trapping efficiency of petri dishes is no higher than 5 % which is remarkably lower than the 80 % trapping efficiency of a Hirst trap (Davies et al., 1963, citing Gregory andStedman, 1953).
Among the data sets used for calculations, many of the researchers relied on the Hirst-type spore trap as described above, e.g.(Davies et al., 1963;Hamilton, 1959;Rodríguez-Rajo et al., 2005;Oliveira et al., 2009;Mallo et al., 2010;Levetin and Dorsey, 2006;Herrero et al., 2006;Wu et al., 2004;Stepalska and Wolek, 2009;Kasprzyk and Worek, 2006;Sakiyan and Inceoglu, 2003).Among these, mostly models from Lanzoni (VPPS 2000) (Lanzoni, 2010) or Burkard (Burkard Scientific, 2000) were used.Others relied on Filterhousings containing 2 µm filter membranes, e.g.Prospero et al. (2005); Griffin et al. (2001Griffin et al. ( , 2003Griffin et al. ( , 2007)), for their measurement.Another device sometimes used was the May Cascade Impactor (May, 1945).However, this device was found to have a considerably lower trapping efficiency for small spores than the Volumetric Spore Trap (Hirst, 1953).Only a few of the studies monitored the fungal spore content in cloud droplets or precipitation (Amato et al., 2007;Bauer et al., 2002a) and some other measurement devices and methods apart from those named above have been used as well (for detailed information on measurement methods see Table A1 in the appendix).
The measurement duration varied from a few minutes to continuous measurements seven days a week (especially applied for the Hirst-type spore traps).Also the time span over the year varied from study to study: some only measured on one single day, e.g.Côté et al. (2008), others even over several years continuously, e.g.Mallo et al. (2010).
Besides the heterogeneity of the used measurement devices also the airflow varied from 1.9 lmin −1 (Griffin et al., 2001) up to 1.13 m 3 min −1 (Lau et al., 2006).An airflow of 10 lmin −1 seems to be the most commonly used value for such measurements.Moreover it was difficult to retrieve the airflow rates from all studies.The correct choice of airflow is an important factor for spore measurements since measurements that are not conducted isokinetically can lead to remarkable biases in spore counts.
Furthermore the height at which the devices had been installed was not the same in the respective studies.Some used air samples in heights up to several kilometres, whereas other conducted ground-based measurements in heights from 2 m up to 50 m.When considering spore dispersal from a boundary layer meteorological point of view, the choice of the measurement height might be crucial since it decides on whether measurements are taken within or outside the turbulent layer.
The situation is alike for the different impaction media chosen: whereas those using the Lanzoni Hirst-type spore trap rely on so-called Melinex tape which is mounted after spore collection with glycerol jelly, others use different surfaces.The same situation can be found when it comes to the use of nutrient media for growth of the trapped fungal spores (viable counts).Some rely on R2A agar, others on YM (yeast-morphology) agar, others on Sabouraud's medium for fungi or malt-agar extract, etc.The possibilities for fungus incubation for viable counts are numerous.Incubation times vary from two days up to two weeks and longer and so do also the incubation temperatures applied.Moreover, counting methods are not the same in all studies.Some use optical methods such as microscopy or macroscopy with different magnifications (ranging from 40x−1000x) while others use Polymerase Chain Reaction (PCR) to determine genera and species from their genome.As for the nutrient medium chosen, both Abdel-Hafez et al. (1985) and Abdel-Hafez et al. (1986) conclude that the choice of nutrient medium had an influence on the prevalence of respective species counted.
To sum up, the measurement methods differ at many stages of the measurement and counting process: The counting device, the airflow chosen, the placement of the sampling device, the sticky medium, the nutrient medium (for viable counts), the incubation duration and the counting method.It is crucial to introduce a convention on how to exactly measure fungal spores in the atmosphere in order to prevent biases due to different measurement methods and to increase comparability of results.

CFU vs. total spore count
Not all studies have investigated both the total count of all spores available and the counts of viable spores called colony forming units (CFU) after incubation.However, this would be an important source of information on what share of the atmospheric fungal spora is viable and which is not (Gregory, 1967).In very few of the studies, both the viable and total counts of spores have been investigated, e.g.Lau et al. (2006); Pady and Kapica (1955); Griffin et al. (2001); Bauer et al. (2002a).All other studies either published total spores or only CFUs.In order to assess their activity as CCN or IN it is not relevant whether they are alive or not.The important feature is the shape and presence of the active proteins which actually act as nuclei (Kieft and Ruscetti, 1990).Therefore, it is not the viability of the fungal spore that affects its IN activity, but whether the ice nucleation active proteins on its surface are denatured or not.If the protein conformation is intact, it can trigger the ice nucleation, regardless of the cell's viability.However, it is important that the cell has been alive at one point in order to produce the ice nucleation proteins in the first place.Therefore it is all the more important to evaluate total counts instead of only viable counts in order to include all possible IN present on the spore surface.Pady and Kapica (1955) found that silicone slides exposed in a slit sampler revealed spore counts a manifold higher than those of plates exposed at the same time where CFUs had been counted.This could be explained by a remarkable bias between viable and total spore counts.Gregory (1967) found that viability of Alternaria spores averaged at 80 % and that of Cladosporium spores at 42 % with viability decreasing at midday.Hence, viability varies from species to species.Other sources report Cladosporium viability to be on average at 62 % (Pady and Gregory, 1963).Experiments by Harvey (1967) investigated viability of Cladosporium and came to the conclusion that single spores germinate more readily than clumps of spores.Moreover they also found a diurnal cycle in germination that reached maximum values between 10:00 UTC and 18:00 UTC and minimum values at 02:00 UTC.However, the authors also state that this cyclicality of viability is in disagreement with the findings by Pathak and Pady (1965).Bauer et al. (2002a) estimated the total vi- ability of the sampled bacteria and fungi to be around 87 % using the condition of the cell wall as a criterion to determine viability.Analysis of snow samples then showed that the cultivable part of fungi amounted only up to 0.7 % (Bauer et al., 2002a).This seems a very low number, but it is sensible considering the fact that these spores already had to survive very harsh conditions within the ice crystals or even during precipitation formation processes.Fisar et al. ( 1990) compared two counting methods, CFUs and direct counts (Fluorescence technique; see respective paper for detailed description of this method) and came to the conclusion that the difference between CFUs and direct counts for both, bacteria and yeasts, is not only considerable but also highly variable.
The authors added that seasonal trends in these discrepancies were not detected.The information provided in the above section is strong evidence for the fact that CFU counts and total spore counts are not equal at all.Due to the scarcity of studies including both viable and total counts, values for CFUs and total spore counts have been treated equally in this paper knowing that the two counting methods show remarkable differences in resulting counts.

Discussion
The fungal spore fluxes for the respective biome areas are in the expected orders of magnitude.The more a biome region is vegetated, the higher are the fungal spore emissions (cf.Table 2 and Fig. 2).This makes sense considering the fact that plants are the largest source of fungal bioaerosol.This reasoning is also supported by Heald and Spracklen (2009), although their fungal spore flux calculations are based on the leaf area index (LAI) which might result in a bias towards too low emissions for biomes with lower LAI as already pointed out in Sect. 2.
Unfortunately, no useful data in land ice areas were found so that emissions are not available for this ecoregion.It would nonetheless be worth investigating these regions as well.Lichen soredia (i.e. the reproductive structure of lichen) have been reported to be most abundant in maritime Antarctica (Henderson-Begg et al., 2009, citing Marshall 1996).Considering the fact that lichen have been found to be effective ice nuclei, it is probable that lichen in these ecoregions do have an influence on cloud microphysical processes as well as precipitation formation.A number of studies mainly focused on species such as Cladosporium, Aspergillus or Alternaria.This is mainly because these fungi can cause allergies (Goncalvez et al., 2010, citing Vijay 2005and Shen 2007).Heald and Spracklen (2009) conducted an estimate of fungal spore emissions based on mannitol concentrations (a biotracer for fungi) using the GEOS-chem chemical transport model (see respective paper for detailed information).Two major differences between the respective results can be seen in the tropical regions.Whereas Heald and Spracklen (2009) found highest values in the Amazon and tropical African region, the present study does not reveal maximum values in these regions.Moreover, mass fluxes found by Heald and Spracklen (2009) are with 5.96 × 10 −12 kgm −2 s −1 an order of magnitude lower than those presented here (cf.Table 4).(Winiwarter et al., 2009), who calculated global average fungal spore emissions of 5.71 × 10 −13 kgm −2 s −1 came to values two orders of magnitude smaller.Nevertheless, the values presented here (1.69 × 10 −11 kgm −2 s −1 , for best estimate cf.Table 4) are in agreement with the study by Elbert et al. (2007) that presents a fungal spore mass flux of 2.26 × 10 −11 kgm −2 s −1 .
The lower values found in tropical regions can partly be explained by the fact that fungal spore data for tropical forests have been aggregated up to one biome area irrespective of forest type.Both methods are prone to biases and errors.The downsides of the present method will be outlined in Sect.3.1.However, it is also likely that the results by Heald and Spracklen (2009) are based on different asumptions.As outlined in Sect. 1 fungal spore composition and size can be very heterogeneous and variable.Nevertheless, the approach of Heald and Spracklen (2009) is promising as seasonality and meteorological influences have been taken into account.
Probably the measurements of actual spore concentrations as well as the measurements of biological tracer substances to derive fungal spore fluxes have to be seen in a complementary way.Concentration measurements on the one hand yield absolute counts of all fungal spores, but comprise measurement issues outlined in previous sections as well as some problems with the identification of fungal spores.As for the biochemical tracer method, these problems are not encountered, but more knowledge needs to be gained in terms of chemical composition of primary biological aerosols as to better describe them and hence yield more accurate results.Therefore it is useful to intercompare the two approaches in order to determine possible sources of error and strengthen hypotheses where agreement is achieved.

Sources of error
As mentioned above, the results of this study have to be taken with caution.The density of measurement points was not sufficient to take into account all ecosystems and therefore only a very coarse distinction into the here presented biome areas could be made.This also implies that for example a deciduous broadleaf forest is considered to emit the same amount and kind of spores as a coniferous forest in higher latitudes.This of course does not make sense from a biological point of view.Moreover, the seasonal and diurnal cycles which clearly have been detected (see Sect. 2) could not be extended to all biomes and fungal species due to scarcity of data.The fact that the difference between CFUs and total counts is not considered clearly must bias the results especially since only very little information is available on what share of fungal spores are viable.
The present data have been calculated on the basis of many simplifications and assumptions that had to be made due to the scarcity of data.As outlined in Sect.2, also measurement methods showed high heterogeneity which can clearly be regarded as a source of bias in the present dataset.Nevertheless it is a step towards enumerating fungal spore concentrations and fluxes on a global basis.We hope that our review will motivate further observations of fungal spore emissions, which might ultimately lead to an improved global database.

Outlook
In order to minimise the above named sources of error, measurement methods need to be standardised in order to allow for better comparability.Furthermore, results should always comprise both CFU and viable count data.To allow for a finer resolution of flux data, data points should be distributed more densely and represent a larger set of investigated ecosystems.In order to account for seasonal and daily cycles, measurements should be continuous over periods at least lasting a full year.
Since investigation of bioaerosols in general is highly interesting due to their possible influence on cloud microphysical processes, further research would be needed to learn more about the potential of fungal spores to act as CCN or IN.Moreover, lichen which have been reported to be effective IN should be investigated better since they are more abundant in the atmosphere than bacteria (Henderson-Begg et al., 2009).Additionally, data should not only focus on allergy causing fungi, but also on those spores which presumably influence atmospheric processes.It would therefore also make sense to investigate concentrations of IN active spores such as Fusarium in order to get a broader picture of what share of all fungal spores is actually involved in ice crystal formation processes.

Overview of measurement methods and observational data
The following tables comprise information on measurement methods employed by the respective studies (see Table A1) as well as all the values used for flux calculations (see Table A2).Multiple values for the same source can occur in case multiple measurements over different intervals of time had been undertaken or different sites had been chosen for measurement.These cases were treated as single measurement points.

Fig. 1 .
Fig. 1.Global distribution of locations where fungal spore emissions have been measured.The transect over the ocean denotes ship measurements.

Table 1 .
Attribution of ecosystems by to the respective biome areas

Table 2 .
Number fluxes of fungal spores per square metre of biome and second [m −2 s −1 ].

Table 3 .
Mass fluxes of fungal spores per square metre of biome and second [kgm −2 s −1 ].

Table A1 .
Sources used for fungal spore flux calculations and details about their measurement procedure.Most samplings considered the total spore counts, those that considered only certain genera are indicated by the footnotes.

Table A2 .
Data used as input for fungal spore flux calculations; average, minimum and maximum number concentrations and number fluxes of fungal spores.The data were always rounded to a whole number.The column CFU (Culture Forming Units) counts denotes if a culture based technique was used for the analysis (yes), if it was a culture-independent technique (no), or if both methods were used.