The effect of biological particles and their ageing processes on aerosol radiative properties: Model sensitivity studies

Abstract. Biological aerosol particles (BAPs) such as bacteria, viruses, fungi and pollen, represent a small fraction of the total aerosol burden. However due to their unique properties, they have been suggested to be important in for radiative forcing by the aerosol direct and indirect effects. By means of process model studies, we compare the sensitivity of these radiative effects to various physicochemical BAP properties (e.g. number concentration, diameter, hygroscopicity, surface tension, contact angle between ice and particles). Exceeding previous sensitivity studies, we explore not only the variability of these properties among different BAP types, but also the extent to which chemical (e.g. nitration), physical (e.g. fragmentation) and biological (e.g. bacteria cell generation) ageing processes of BAPs can modify these properties. Our model results lead to a ranking of the various properties for the radiative effects: (i) Given that BAPs contribute ~ 0.1 % to total cloud condensation nuclei (CCN) number concentration, their effect on total CCN is likely small. (ii) BAPs number fraction of large particles (diameter > ~ 0.5 μm) is much higher, resulting in a relatively more important effect on direct radiative forcing. (iii) In mixed-phase clouds at T > −10 °C, BAPs can contribute ~ 100 % to ice nuclei (IN), which makes their role as IN the most important. Our study highlights the need of implementing ageing processes of different BAPs into models as BAP size, CCN and IN activity and optical properties may be sufficiently altered to affect BAP's residence time and survival in the atmosphere. In particular, we suggest the potential role of biological processes, that are currently not included in aerosol models due to the sparsity of comprehensive data, could affect physicochemical BAP properties.


In numerous recent review articles, it has been suggested that BAPs can affect radiative forcing in multiple ways (Figure 1) (Coluzza et al., 2017;Després et al., 2012;Haddrell and Thomas, 2017;Hu et al., 2018;Šantl-Temkiv et al., 2020;Smets et al., 2016): BAPs might directly interact with radiation by scattering or 55 absorbing light (Figure 1a). While their aerosol direct effect is likely globally small due to low BAP number concentration (Löndahl et al., 2014), it may be of greater interest locally and for specific wavelength ranges due to the large size of BAPs (Myhre et al., 2013). The optical properties of BAPs (Arakawa et al., 2003;Hu et al., 2019;Thrush et al., 2010) resemble those of other organic particles as BAPs are largely composed of proteins and other macromolecules. Accordingly, BAPs' optical properties can be ascribed to specific 60 organic functional entities such as amino groups or aromatic structures (Hill et al., 2015;Hu et al., 2019). https://doi.org/10.5194/acp-2020-781 Preprint. Discussion started: 13 August 2020 c Author(s) 2020. CC BY 4.0 License.
In addition to acting as CCN, some species of plant pathogen bacteria and fungi can nucleate ice at T > -10C (Hoose and Möhler, 2012;Morris et al., 2004Morris et al., , 2008Pouzet et al., 2017), which makes them unique in terms of ice nucleation to affect the evolution of mixed-phase clouds at these temperatures (Figure 1c).
Above vegetated forests (Tobo et al., 2013) and near the surface of the Southern Ocean (Burrows et al., 2013), BAPs have been shown to contribute significantly to the total abundance of IN: In a high altitude 90 mountain region of the United States, ambient measurements suggest that 16-76% of IN at -30 °C consist of primary biological material (Pratt et al., 2009); a similar proportion (33%) was reported at -31 to -34 °C in the Amazon basin (Prenni et al., 2009).
 Biological processes might be initiated by living microorganisms in BAPs, unlike in other aerosol 110 particles in the atmosphere Delort et al., 2017;Joly et al., 2015). Such processes are generally driven by strategies to adapt to the harsh conditions in the atmosphere (e.g., rapid temperature and RH changes, thaw/freeze cycles, humidification/desiccation, UV exposure) (Hamilton and Lenton, 1998;Horneck et al., 1994;Joly et al., 2015;Setlow, 2007) or to limit their atmospheric residence time by initiating precipitation (Hernandez and Lindow, 2019). These processes include nutrient uptake by 115 biodegradation (Khaled et al., 2020), bacteria cell generation that enhances particle size and surface area (Ervens and Amato, 2020), formation of biofilms (extracellular polymeric substances) which enables BAPs to form aggregates Lindow, 2003, 2005;Morris et al., 2008;Sheng et al., 2010), https://doi.org/10.5194/acp-2020-781 Preprint. Discussion started: 13 August 2020 c Author(s) 2020. CC BY 4.0 License. expression of ice-nucleating proteins (Joly et al., 2013;Kjelleberg and Hermansson, 1984), formation of biosurfactants that enhances water uptake (Hernandez and Lindow, 2019;Neu, 1996), desiccation that 120 decreases size of BAPs (Barnard et al., 2013), formation of pigments (Pšenčík et al., 2004;Fong et al., 2001) enhancing light absorption, fungal spore germination (Ayerst, 1969) or formation of bacteria endospores (Enguita et al., 2003) that increases the NBAP, and metabolism of cellular components (membranes, proteins, saccharides, osmolytes, etc) (Fox and Howlett, 2008;Xie et al., 2010). To date, the uncertainties introduced by these BAP ageing processes in the estimate of BAP radiative effects, their 125 atmospheric residence time and distribution can only be assessed qualitatively due to the lack of comprehensive data. However, it may be expected that some of them lead to similar differences in BAP properties than differences between BAP types.
In our study, we give a brief overview of the BAP properties in Figure 1 and summarize which chemical, physical and biological processes may alter these properties (Section 2). By means of process models 130 (Section 3), we explore in a simplistic way the relative importance of these BAP properties and ageing processes for their radiative effects depicted in Figure 1 (Section 4). The results of our sensitivity studies allow a ranking of the importance of the various BAP properties and processes in terms of their radiative impacts. In Section 5, we give some guidance on the need of future laboratory, field and model studies to more accurately describe the radiative effects, distribution and residence time of BAPs in the atmosphere.

Physicochemical properties and processes of BAPs
Literature data on physicochemical parameters of BAPs are summarized in Table 1. It is not our goal to repeat exhaustive reviews on these individual properties; for this, we refer to previous overview articles (Bauer et al., 2003;Coluzza et al., 2017;Deguillaume et al., 2008;Després et al., 2012;Fröhlich-Nowoisky et al., 2016;Hoose and Möhler, 2012;Huffman et al., 2020;Šantl-Temkiv et al., 2020). We rather aim at 140 using characteristic orders of magnitude of these properties as input data to our process models (Section 3).
Therefore, we only give a brief overview on the ranges and variability of these properties for different BAP types and due to various ageing processes.

BAP number size distribution parameters (NBAP and DBAP)
The number concentration (NBAP) of most BAP types is in the range of 0.001 ≤ NBAP ≤ 0.1 cm -3 (Table 1). 145 The number concentration of bacteria is higher than that of fungal spores and pollen although the mass concentration of bacteria is lower (Burrows et al., 2009a;Heald and Spracklen, 2009;Hoose et al., 2010).
NBAP can vary by about three orders of magnitude among different ecosystems, locations, seasons, and time of the day (Huffman et al., 2010(Huffman et al., , 2020Matthias-Maser et al., 2000a, 2000bSchumacher et al., 2013). The BAP diameter (DBAP) covers a broad range of 0.01 µm ≤ DBAP ≤ 100 µm. This parameter usually refers to 150 the mass equivalent diameter, which is the diameter of a sphere with the same mass as a non-spherical https://doi.org/10.5194/acp-2020-781 Preprint. Discussion started: 13 August 2020 c Author(s) 2020. CC BY 4.0 License.
BAP. The size depends on the types of BAPs, and on changes due to biological and physical processing.

155
Typical bacterial cell generation rates are in the range of 0.1-0.9 h -1 (Ervens and Amato, 2020). Efficient generation in the atmosphere is assumed to be largely restricted to the time of cell exposure to liquid water (i.e., in-cloud). With an average atmospheric residence time of ~1 week (Burrows et al., 2009b) and an average in-cloud time fraction of ~15% (Lelieveld and Crutzen, 1990), it can be estimated that the generation time scale of bacteria cells in the atmosphere is on the order of ~20 h. Thus, for example, Dbacteria may 160 increase from 1 µm to 2 µm after one week in the atmosphere assuming a generation rate of 0.3 h -1 . Other rates, such as the cell growth, are usually much smaller (Marr, 1991;Middelboe, 2000;Price and Sowers, 2004;Sattler et al., 2001;Vrede et al., 2002), and thus, contribute less efficiently to a change in DBAP. In addition, the formation of extracellular polymeric substances might lead to the formation of biofilms, which increase BAP size by forming agglomerate Lindow, 2003, 2005). Agglomerate formation might 165 be also described as a physical process, when BAPs (e.g. bacteria) attach to other particles (e.g. dust) (Després et al., 2012;Lighthart, 1997), which can result in particle sizes on the order of ~10 µm. Similarly, a biologically-driven physical processes might lead to enhancement of NBAP as it has been observed that pollen ruptures into fragments with diameter of 1-4 µm during thunderstorms (Zhang et al., 2019).

170
The scattering and absorption of particles are commonly described by the refractive index mBAP with real part (nBAP) and imaginary parts (kBAP) that depend on the chemical composition and wavelength of irradiation. Arakawa et al. (2003) reported 1.5 ≤ nBAP ≤ 1.56 and 3 · 10 -5 ≤ kBAP ≤ 6 · 10 -4 for bacteria (Erwinia herbicola) in the wavelength range of 0.3-2.5 µm. Other groups found a broader range of n and k (Table 1) for different types of BAPs and irradiation wavelengths (Hu et al., 2019;Thrush et al., 2010). The 175 imaginary part can vary by three orders of magnitude for different BAP types (Hu et al., 2019). Hill et al. (2015) showed that the refractive index of BAPs can be estimated based on the chemical composition. They reported 1.59 + i0.045 for Bacillus vegetative cells at 0.266 µm. Also BAP shape (e.g. core-shell structure, hexagonal grids, and barbs), as it has been demonstrated for pollen, influences the optical properties (Liu and Yin, 2016). Due to the similarity of the molecular structure of organic macromolecules (e.g. proteins) 180 and secondary organic aerosols (SOA), it can be likely assumed that might alter the BAP refractive index similar to that of SOA. Experimental results show 1.516 ≤ n ≤ 1.576 and 0 ≤ k ≤ 0.013 for fresh SOA; after nitration, the real part changed to 1.534 ≤ n ≤ 1.594 and the imaginary part increased to 0.001≤ k ≤ 0.035 (Liu et al., 2015;Moise et al., 2015). https://doi.org/10.5194/acp-2020-781 Preprint. Discussion started: 13 August 2020 c Author(s) 2020. CC BY 4.0 License.

Hygroscopicity (κBAP) of BAPs
The hygroscopicity determines BAP's hygroscopic growth factor (gf, as the ratio of wet to dry particle

Surface tension (σBAP) of BAPs
In most model studies that explore CCN activation, it is assumed that particles have a surface tension close to that of water (σwater = 72 mN m -1 ). This assumption is likely justified under many conditions due to the 210 strong dilution of internally mixed aerosol particles near droplet activation. There are numerous studies that postulate that surfactants in aerosol particles might influence the surface tension sufficiently to significantly change their CCN activity (Bzdek et al., 2020;Facchini et al., 1999;Lowe et al., 2019;Nozière et al., 2014).
These surfactants are usually assumed to have natural sources such as the ocean surface (Gérard et al., 2019;Ovadnevaite et al., 2017). Another source of surfactants might be living microorganisms that produce 215 biosurfactants which enhance surface hygroscopocity and decrease surface tension (Akbari et al., 2018).
These biosurfactants might not only be associated with BAP themselves as they might be deposited on surfaces (e.g. leaves) where they can be taken up by other particles. Renard et al. (2016) reported that 41% of tested strains actively produced surfactant with σBAP < 55 mN m -1 and 7% of tested strains can produce extremely efficient biosurfactants with σBAP < 30 mN m -1 . All of these tested strains were collected and 220 isolated in cloud water samples. The efficient biosurfactants (σBAP < 45 mN m -1 ) are mostly produced by Pseudomonas and Xanthomonas of bacteria (78%) and Udeniomyces fungi (11%). For the most efficient biosurfactants, we fit the following linear approximation based on their experiments: where σBAP is BAP surface tension in (mN m -1 ) and Cbiosurf is the biosurfactant concentration in (mg L -1 ). and on the mass fraction of biosurfactants in the particle. The mass fraction has not been determined for biosurfactants; however, other surfactants have been shown to contribute ~0.1% to the total particle mass (Gérard et al., 2019).

IN active particle number fraction
In freezing experiments of pollen, it has been demonstrated that all particles freeze at sufficiently low temperatures, i.e. the IN active number fraction can be assumed as ~100%. Both condensation and immersion/contact freezing led to frozen fractions of 100% at T ~-18 °C (Diehl et al., 2001) and T ~-20 °C (Diehl et al., 2002), respectively. However, for bacteria such as Pseudomonas syringae, the maximum frozen 240 fraction only reaches values of 0.1-10% at T ~-10 °C (Joly et al., 2013). This might be explained by the fact that not all of the bacteria cells express the same proteins even if they belong to the same species and the same population. It was observed that bacteria express more IN proteins under stress conditions (Kjelleberg and Hermansson, 1984), as a strategy to reach nutrients after destroying the cells of plants by freezing.
However, to date it is not fully understood why in lab experiments some of the bacteria cells show freezing 245 behaviour while others from the same population do not and why individual cells show stochastic behaviour in repeated experiments (Lukas et al., 2020).

Contact angle between substrate and ice (θBAP)
In agreement with previous studies, we base our discussion on the contact angle as a fit parameter in the classical nucleation theory (CNT) to parametrize the frozen fraction observed in experiments. Chen et al. deterministically as opposed to stochastic freezing described by CNT. As the sensitivity of ice nucleation to time is generally small compared to other parameters (Ervens and Feingold, 2013), we fitted their data using CNT and obtained a range of 32° ≤ θbacteria ≤ 34°, consistent with other bacteria (Attard et al., 2012).
Chemical processes (e.g. nitration) can change the molecular surface of BAPs by e.g., adding nitro groups 260 to tyrosine residues of proteins (Estillore et al., 2016), which can alter the IN activity. Attard et al. (2012) https://doi.org/10.5194/acp-2020-781 Preprint. In order to study the oxidation effect, Gute and Abbatt (2018) exposed pollen to OH radicals and measured the cumulative frozen fraction of pollen in terms of deposition freezing. We calculated that the contact angle increased by ~0.5° ≤ Δθpollen ≤ 0.8° after oxidation. While experimental conditions are often such that a large fraction of particles is nitrated or oxidized, respectively, only a small fraction of ambient proteins 270 (~0.1%) have been found to be only nitrated (Franze et al., 2005). Attard et al., (2012) showed that a decrease of pH from 7.0 to 4.1, led to a decrease of the cumulative fraction of IN of P. syringae (32b-74) from 10 -2 to 10 -8 at -4 °C. This change can be described by an increase of θ from 28.7° to 30.3° (Δθbacteria ~ 1.6°). P.

Box model: Scattering/absorption of wet particles at RH < 100% calculated by Mie theory
A box model was used to simulate total scattering/absorption based on Mie theory (Bohren, 1983) for a constant aerosol distribution at different RH. Water uptake by particles is calculated based on Köhler theory.
Mie theory is applied to calculate total scattering and absorption of the wet aerosol population as a function 280 of D, N, and m at different wavelengths (λ). The input aerosol size distribution is based on ambient measurements by an ultraviolet aerodynamic particle sizer (UV-APS) in central Europe (Zhang et al., 2019) that cannot detect particles with D < 0.5 µm. At  ≥ 300 nm, the particles with D > 3 µm interact with light by geometric scattering, rather than Mie scattering. Therefore, we intentionally only consider particles with diameters of 0.5 µm < D < 2.8 µm in 24 size classes to represent ambient aerosol particles relevant for our 285 study with a concentration of Nother = 1.4 cm -3 . We consider one additional BAP size class which has specific parameters (NBAP, DBAP, mBAP, κBAP, σBAP) that are varied in the sensitivity studies.
Calculations are performed for RH of 10% and 90%, i.e. for different BAP growth factors. In a series of sensitivity studies (Sopt1 -Sopt11; Table 2), we explore the sensitivity of scattering and absorption to the NBAP, DBAP, κBAP, and mBAP (mBAP = n + ik). We not only compare model results for properties representing 290 different BAP types (e.g. Dbacteria vs Dfungal), but also explore the ranges of property variation due to ageing processes of individual BAP types (e.g. ΔDbacteria).

CCN activation in warm clouds
An adiabatic parcel model was applied to simulate the formation of warm clouds (Ervens et al., 2005;Feingold and Heymsfield, 1992). The activation of an aerosol population to cloud droplets is described as a function of N, D, κ, and σ. The dry aerosol size distribution covers a size range of 5 nm < Dother < 7.7 µm 300 with Nother = 902 cm -3 . Similar to the studies on optical properties (Section 3.1), we assume that one aerosol size class is composed of biological material for which we vary DBAP, κBAP, and σBAP to explore the role of differences in BAP type and ageing processes on cloud droplet activation (SCCN1-SCCN9, Table 2).

Ice nucleation in mixed-phase clouds
The adiabatic parcel model as used for the CCN calculations was extended by the description of immersion  Table 2).

Influence of concentration (NBAP) and diameter (DBAP) on scattering and absorption
As explained in Section 3.1, in the sensitivity studies of optical properties, we consider only particles with D in the same range as  so that scattering and absorption can be calculated by Mie theory. As ambient aerosol particles also include smaller and larger particles, our conclusions on BAP direct radiative effects 320 should be regarded as the upper limit on total scattering and absorption.
In Figure 2, we compare the total scattering coefficient for a case without BAP (NBAP = 0, Sopt1) to that predicted for NBAP = 0.01 cm -3 (Sopt2), NBAP = 0.1 cm -3 (Sopt3) and NBAP = 1 cm -3 (Sopt4). At a typical low number concentration of NBAP = 0.01 cm -3 , the effect on total scattering coefficient is negligible. At NBAP = 0.1 cm -3 the total scattering coefficient increases by 15-18% at λ = 0.3-1.5 µm although the number fraction 325 of BAP is only 6%. We also use a higher concentration (NBAP = 1 cm -3 ) to explore the maximum effect as BAP concentrations of this order of magnitude have been observed under haze conditions (Wei et al., 2016).   Table 2. The black line, red line, blue line, and brown line correspond to Sopt1, Sopt2, Sopt3, and Sopt4 in Table 2, respectively.

335
DBAP also affects the scattering coefficient of the aerosol population significantly (Figure 3). DBAP = 1 µm (Sopt3) and DBAP = 2 µm (Sopt5) can be considered to represent different BAP types such as bacteria and fungi, respectively, or an aged bacteria cell that has undergone processing by cell generation (Ervens and Amato, 2020). For these assumptions, the scattering coefficient increases depending on λ, with the largest changes 340 of 73-100% at λ > l.5 µm when DBAP increases from 1 µm to 2 µm (Sopt5). Larger BAP (DBAP = 3 µm, Sopt6) such as pollen fragments show lead to an increase in the scattering coefficient by a factor of 1.4-4.7 depending on λ. The absorption coefficient of the aerosol population remains nearly the same.
The results in Figure 3 clearly show that the size of BAP needs to be known in order to assess their optical properties. Even a relatively small variation in particle diameter from 1 to 2 µm due to different BAP types 345 or to cell size changes (ΔDBAP) might lead to change in scattering coefficient by 8-100 % depending on λ.
Given that the diameter (DBAP) might vary by four orders of magnitude among different BAP types, our analysis shows that different sizes for the various BAP types need to be taken into account if their optical properties are evaluated.  Figure 3. Influence of BAP diameter on (a) scattering coefficient and (b) absorption coefficient of total particles. The detailed input parameters can be found in Table 2. The black line, red line, blue line, and brown line correspond to Sopt1, Sopt3, Sopt5, and Sopt6, respectively.
In our model studies, we make the simplistic assumption of spherical BAP particles. Electron scanning microscopic imaging has shown that BAP are not spherical but exhibit a variety of different shapes (Valsan by a factor of one to three for small pollen with D < 4 µm. For larger pollen with D > 5 µm, the extinction efficiency varies by ~25% (Liu and Yin, 2016). While we do not explore sensitivities of BAP geometry, it 360 may be postulated that under atmospheric conditions, i.e. when BAP are wet, they are more spherical than under the experimental dry conditions, and thus effects due to non-sphericity may be reduced.

Influence of hygroscopicity (κBAP) and surface tension (σBAP) on scattering and absorption
As discussed in Section 2.3, gf BAP might vary depending on BAP hygroscopicity (κBAP) and surface tension (σBAP). Figure 4 shows the influence of κ on scattering and absorption at RH of 10% (Sopt7, Sopt8) and 90% 365 (Sopt9, Sopt10). At RH = 10% (Sopt7, Sopt8), the influence of BAP on scattering coefficient of total particles is small (< 19%) and the influence on absorption coefficient is negligible. Assuming κ = 0.25 (Sopt10) instead of κ = 0.03 (Sopt9), leads to an increase of the scattering coefficient by 17-90% at RH = 90%. Also the absorption coefficient increases by ~40% at λ > 2 µm. It can be concluded that the importance of Δκ increases for higher RH as under these conditions BAP hygroscopic growth is most efficient.

375
In addition to hygroscopicity (κBAP), we explore the importance of biosurfactants which decrease surface tension of particles (σBAP). A lower surface tension leads to a reduced particle curvature which, in turn, enhances the water uptake. Numerically, this is expressed in the Köhler equation, which can be generally expressed as Equation 2: where s is the equilibrium water vapor saturation ratio, Dwet the wet particle diameter, the first term in the parentheses is the Kelvin (curvature) term which is a function of surface tension (σBAP) following Equation 3 and the second term is the Raoult (solute) term which can be parameterized by κBAP (Rose et al., 2008) following where σsol is surface tension of solution droplet (72 mN m -1 ); Mω is molar mass of water (18 g mol -1 ); ρω is density of water (1 g cm -3 ); R is the universal gas constant (8.31 · 10 7 g cm 2 s -2 K -1 mol -1 ); T is the absolute temperature (K); Dwet is droplet diameter (cm); and Ds is the diameter of dry solute particle (cm).

390
The comparison of the two dimensionless terms shows that in most of the cases, the Raoult term exceeds the Kelvin term by at least one order of magnitude. Only for very small BAP, i.e. representative for viruses or bacteria fragments (Section 2.1), the curvature term significantly influences s (Figure 5). Based on this analysis, we can conclude that (bio)surfactants likely do not have a significant impact on the hygroscopic growth of BAP. A coating with surfactants might slow down the kinetics of the water uptake by particles 395 (Davidovits et al., 2006). However, since the growth time scales of particles at RH < 100% are usually relatively long, the impact of surfactants on the time scale to reach equilibrium sizes is likely small, leading to a small importance of the effect of surfactant on water uptake and the corresponding optical properties.

Influence of complex refractive index (mBAP = n + ik) on scattering and absorption
The complex refractive index of BAP can be explained by their building blocks of various functional groups 405 (Hill et al., 2015). Hu  five, depending on the wavelength with the largest effects at λ > 2 µm (Figure 6).

415
In addition to the variability in refractive index due to BAP type, chemical processing of the macromolecules at the BAP surface might modify the refractive index. It has been shown that nitration of SOA, i.e. the addition of a nitro group, leads to the formation of brown carbon (Moise et al., 2015). Qualitatively, it has been demonstrated that proteins can be nitrated, similar to SOA compounds (Shiraiwa et al., 2012). Due to 420 the lack of data on Δm for nitrated proteins in BAP, we assume a similar change in the refractive index as in SOA (Sopt13 and Sopt14). The scattering coefficient can change by up to 20% and the absorption coefficient by a factor of three at λ = 0.42 µm (Figure 7). Thus, the variability in scattering/absorption properties of BAP due to Δm caused by nitration is likely smaller than due to Δm caused by different BAP types. The assumptions on Δm made for the simulations shown in Figure 7 are likely an overestimate of the chemical 425 processing of proteins as BAP constituents since (1) experimental conditions are often optimized so that a large fraction of particles is nitrated (Liu et al., 2015) as opposed to ~0.1% of nitrated proteins observed in the atmosphere (Franze et al., 2005), (2) we assume nitration to occur over the whole residence time of particles in the atmosphere while proteins can be nitrated only under conditions of sufficiently high NOx levels (Shiraiwa et al., 2012), and (3) a rather high concentration of NBAP = 1 cm -3 is considered.

Estimate of change of radiative forcing introduced by BAP
In order to give an estimate of the local radiative forcing due to BAPs, we applied the same approach as 435 Dinar et al. (2007). The radiative forcing efficiency (RFE, i.e. radiative forcing per unit optical depth) at 390 nm can be calculated as: where Scon is the solar constant (1370 W m -2 ); Dlen is the fractional day length (0.5); Acld is the fractional cloud cover (0.6); Tatm is the solar atmospheric transmittance (0.76), and Rsfc is surface albedo (0.15); ω is 440 the single scattering albedo (SSA), which is the ratio of scattering coefficient to extinction coefficient; β is average upscatter fraction, which can be calculated as: where b is the ratio of backscattering to scattering coefficient, g is the asymmetry factor which is assumed as 0.65 as an average of ambient measurements (~0.59-0.72 (Andrews et al., 2006)). The calculated RFE 445 are listed in Table 3 for some of the model results of the simulations listed in  Note that the RFE values in Table 3 only represent radiative forcing of a small range of particle sizes and a 455 constant composition and number concentration of other particles; however, the relative differences (ΔRFE) are meaningful and allow evaluating the relative importance of the various BAP parameters (NBAP, DBAP, ΔmBAP) in terms of the direct radiative effect. A decrease in RFE implies less absorption, and thus more cooling of atmosphere (Dinar et al., 2007). ΔNBAP (Sopt3) and ΔDBAP (Sopt4) have a significant influence on ΔRFE. In addition, ΔRFE at λ = 390 nm is higher than that at λ = 532 nm, implying the increasing importance 460 of BAP in the UV range.
When fungal spores are considered instead of bacteria (i.e., m = n +ik is changed), SSA decreases and RFE even changes from a negative to a positive value (Sopt13), resulting in predicted RFE of 7.56 W m -2 at λ = 390 nm and 5.69 W m -2 at λ = 532 nm, respectively. This might be explained by the strong light absorption (very high k) of Aspergillus oryzae fungal spores. Generally, the imaginary part k can vary by three orders 465 of magnitude between different types of BAP (Table 1)

Influence of BAP concentration (NBAP) and diameter (DBAP) on CCN activation
NBAP is low compared to the total CCN concentration (Chow et al., 2015;Sun and Ariya, 2006). The upper limit NBAP is on the order of ~1 cm -3 (Table 1)  properties related to the CCN activation of BAP should be considered being more important for biological reasons, i.e. for BAP to be surrounded by water and the significant modification of the atmospheric residence time of BAP that is consequently changed by the transport and precipitation in clouds.
The critical saturation sc can be used as a measure to estimate whether a particle will be activated into a cloud droplet (Rose et al., 2008): where A can be found at Equation 5, κ is hygroscopicity, and Ds (cm) is mass equivalent diameter of dry solute particle. Applying this equation, one finds that for particles with DBAP = 0.01-10 µm, the critical supersaturations (Sc = (sc-1) · 100%) are in a broad range of 0.0007%-24% (assuming κ = 0.03; σ = 72mN m -1 ). For large BAP with DBAP > 0.5 µm, the critical supersaturations Sc is smaller than 0.062%. Typical 490 supersaturations (Senv) in stratocumulus and convective cumulus clouds are in the range of ~0.1-0.5% and ~0.5-1%, respectively (Pruppacher and Klett, 1997). Comparison to Sc,BAP shows that most BAP (DBAP > 0.5 µm) are likely activated in clouds as their Sc are significantly smaller than Senv in clouds. Figure 8a shows the range of Sc for the κ values shown in Table 2 for the smallest BAP with DBAP = 500 only for fairly small BAP, the hygroscopicity κBAP may impact their CCN activation.

500
Overlaid on the vertical lines for Sc in Figure 8a are Senv in the cloud as calculated in our parcel model for different updraft velocities (w = 10 cm s -1 , 100 cm s -1 , and 300 cm s -1 ). The sensitivity of CCN properties to https://doi.org/10.5194/acp-2020-781 Preprint. Discussion started: 13 August 2020 c Author(s) 2020. CC BY 4.0 License.

Figure 8a
corroborates the conclusions from these previous studies that the variation of the κ over wide ranges only introduces a small change in the CCN activity and in cloud properties (e.g., drop number 505 concentration, LWC) and that particle composition is most important in clouds with low updraft velocities.  Table 2.
Similar to Sc ranges due to different κBAP values, we compare in Figure 8b predicted Sc ranges due to particles. Our sensitivity studies show once more that under dynamic conditions in clouds buffering reduces the feedbacks of particle composition on supersaturation (Ervens et al., 2005;Stevens and Feingold, 2009).
Therefore, previous estimates of surfactant effects on cloud properties that are based on a simplified assumption of equilibrium conditions in clouds (Facchini et al., 1999), led to an overestimate of the role of surfactants on CCN. 535 We conclude that the mass concentration of biosurfactants needs to be quantified in order to better explore the biosurfactant effect on CCN activation of small particles. Given that the surface concentration of surfactants is likely higher than the bulk concentration (Bzdek et al., 2020;Lowe et al., 2019;Ruehl et al., 2016) as assumed here, even a smaller mass fraction of biosurfactants than calculated by Equation 1 might be sufficient to decrease the surface tension of small aqueous BAP and the corresponding critical 540 supersaturation. However, also for the concept of surface partitioning of biosurfactants, rather than for a bulk concentration, our conclusions hold true on the limited impact of surface tension suppression on CCN activation of supermicron BAPs.

545
NBAP is on the same order of magnitude as that of total IN in some regions and at temperatures T > ~-10C (Pratt et al., 2009;Prenni et al., 2009), which makes BAP play an important role in mixed-phase clouds.
Especially, at these relatively high temperatures, some bacteria and fungi species can nucleate ice while other particles cannot, and therefore NBAP, IN / NIN is ~100%. Figure 9a shows the change of ice water content (IWC, solid lines) and liquid water content (LWC, dashed lines) in a mixed-phase cloud (SIN1, SIN2, SIN3).

550
Above an IWC of ~3%, ice particles start growing at the expense of liquid water (Bergeron-Findeisen-Process) (SIN1). At lower NBAP ~0.01 cm -3 (SIN2), the onset of the Bergeron-Findeisen-Process starts slightly later. With NBAP ~0.001 cm -3 (SIN3), both IWC and LWC are predicted to increase simultaneously throughout the whole cloud, i.e. the Bergeron-Findeisen-Process is not initiated and cloud glaciation does not take place.
In Figure 9b, we compare model results for simulations SIN4 and SIN5 in order to explore the effect of DBAP.

555
With larger BAP size such as DBAP = 2 µm (SIN4) or DBAP = 5 µm (SIN5), ice formation starts earlier in the cloud, but the onset of the Bergeron-Findeisen process occurs at approximately the same temperature as for smaller DBAP because of the feedbacks of IWC and LWC on the supersaturations in the cloud and vice versa.
In agreement with previous sensitivity studies (Ervens et al., 2011;Ervens and Feingold, 2013), these results confirm that the influence of D on the IN activity is relatively small (Figure 9). Based on these trends, it 560 can be also concluded that processes that change the BAP size (e.g. ΔDBAP by cell generation) are not critical to be included in models to represent the variability of IN property effect on mixed-phase clouds.

Influence of the contact angle (θBAP) on ice nucleation
Different types of BAPs exhibit a wide range of contact angles of 4° < θBAP < 44° (Table 1 and Section 2). Figure 10, different BAP types that have θBAP of 4° or 20°, respectively, lead to a difference in temperature, at which the Bergeron-Findeisen process occurs, by ΔT ~0.6 °C. For BAP with even higher 570 θBAP (40°), the Bergeron-Findeisen process occurs even at a lower temperature (ΔT ~3.3°C).

As shown in
As discussed in Section 2, chemical (e.g., nitration, oxidation, adjustments due to pH) or physical processing of IN surfaces might lead to ΔθBAP ~1. In Figure 10d, we show the resulting change in IWC and LWC evolution by comparing SIN2 and SIN9. It is clear that even such a small change in θ can cause a significant difference in the IWC and LWC evolutions. The temperature, at which the Bergeron-Findeisen process 575 occurs differs by ΔT ~1.3 °C. These results suggest that a small change of contact angle due to different types of BAP or due to processing (Δθ) might affect the Bergeron-Findeisen process significantly. We only exemplarily explore Δθ for nitration based on the experiments by Attard et al. (2012). In the same study, it https://doi.org/10.5194/acp-2020-781 Preprint. Discussion started: 13 August 2020 c Author(s) 2020. CC BY 4.0 License.
was found that Δθ is ~1.5° for bacteria such as Pseudomonas syringae when the cells were exposed to solutions of pH 7.0 and 4.1 at temperatures of T > -10 °C. 580 Figure 10. Ice water content (IWC, solid lines) and liquid water content (LWC, dashed lines) as a function of θBAP. Even when the contact angle increases by 1°, the initiation of Bergeron-Findeisen process might be influenced significantly.
Similar exercises could be done for differences in θ due to other processes, such as the oxidation of pollen 585 that lead to Δθ ~1.5° at T ~ -39 °C (Gute and Abbatt, 2018). However, at this much lower temperature, the sensitivity of the frozen fraction to Δθ decreases (Ervens and Feingold, 2013

Conclusions
Based on our model sensitivity studies, we can rank the importance of the various parameters and processes of BAPs shown in Figure 1 in terms of their radiative effects: The increasing importance and sensitivity are summarized in Figure 11. 595 Figure 11. Schematic of the importance of BAP in the climate system and the sensitivity of radiative effect to BAP properties. The bottom arrow shows the increasing importance of BAP in CCN, scattering/absorption, and IN. The left arrow indicates the increasing sensitivity to BAP properties, which depend on the type of BAP and ageing processes.

600
As the number concentration of BAPs only contribute ~0.1% to the total CCN concentration, even under conditions of high NBAP, their role in CCN activation in warm clouds is negligible as they do not lead to any significant change in cloud properties. Since BAPs have usually supermicron sizes, they will act as CCN and even small changes in their chemical composition do not affect their CCN activity. The CCN activation of smaller BAPs such as bacteria fragments or viruses might be influenced by their hygroscopicity (κBAP) 605 and surface tension (σBAP). κBAP might be modified by chemical (e.g., nitration, oxidation), physical (e.g., condensation of gases), and biological processes (e.g., formation of metabolic products, biosurfactants).
Biosurfactants decrease the surface tension of BAPs (σBAP) and possibly even of other particles (σother) to should be considered more representative than the absolute numbers.
The complex refractive index mBAP can be modified due to chemical or biological processing. For example, nitration could lead to an enhancement of the imaginary part (absorptive properties), but the difference of scattering and absorption coefficient induced by nitration is much smaller compared to the differences caused by the refractive indices of different BAP types. Biological processing such as pigment formation 625 (Pšenčík et al., 2004;Fong et al., 2001) might also lead to ΔmBAP to a significant extent, but we cannot quantify the role of this process in our model framework due to the lack of corresponding data. The second ranked parameter is ΔDBAP, which also differs among different types of BAP or might change for one BAP type due cell generation or desiccation in the atmosphere. Obviously, the total number of BAP (NBAP) is of importance for all effects discussed here. However, as it has been shown that at many location NBAP / Ntotal 630 is approximately constant, the relative role of BAP likely does not change due to differences in absolute BAP concentration. Hygroscopicity κ might have an effect under high RH conditions. The effect of surface tension σ on direct radiative property is negligible.
The most important role of BAP is to act as IN because NBAP, IN / NIN can reach up to ~100% at T > -10 °C.
Given the high sensitivity of BAPs that initiate freezing, it is clear that not only the total NBAP but also the 635 fraction that can freeze needs to be constrained. While this fraction is usually ~100% for pollen, it can be as small as 0.01%-10% for bacteria. As identified in previous sensitivity studies, the surface composition properties, often expressed in terms of a contact angle θBAP, shows the highest importance to IN activity and therefore to the evolution of mixed-phase clouds (Bergeron-Findeisen process). The variability of θBAP between different types of BAP (4° < θBAP < 44°) determines the onset temperature of freezing and the 640 temperature interval in which the Bergeron Findeisen process may occur. Even a small change of ΔθBAP ~1° https://doi.org/10.5194/acp-2020-781 Preprint. Discussion started: 13 August 2020 c Author(s) 2020. CC BY 4.0 License.
as caused by chemical processing on BAP surfaces or pH change might affect the onset of the Bergeron-Findeisen process significantly. Thus, not only various BAP types should be parameterized with different θBAP in models but also ΔθBAP due to modification by chemical and possibly biological processes.
The trends discussed above are summarized in Figure 11 and show the relative importance of BAPs in the 645 atmosphere, increasing from their roles in CCN activation, to the aerosol direct effect and to mixed-phase cloud evolution. The arrows on the left and on the bottom point to the most sensitive and most important parameters, respectively, which are placed in the upper right corner of the table.
Our study highlights the possible importance of BAP processing as not only chemical and physical processes but also biological ageing processes can modify the chemical composition and physical properties of BAPs.

650
While the former two process types commonly occur on/in many other ambient particles as well (e.g. Δm due to nitration of SOA, or ΔD due to condensation of low volatility material), biological processing is unique to BAPs and currently not comprehensively included and explored in atmospheric models. For example, we suggest that cell generation or the expression of specific proteins might significantly affect BAP's IN ability. While the role of biosurfactant production (ΔσBAP) is limited in modulating warm cloud 655 properties and the aerosol direct effect, the biological aspects of this process might be of much larger importance: Enhanced water uptake by BAPs may extend lifetime of the microorganisms by improving their living conditions, i.e. reduce stress due to harsh ambient conditions (e.g. high ionic strength, low pH, desiccation). In addition, their inclusion in clouds as IN or CCN will lead to a more efficient transport and distribution across the atmosphere.

660
In addition to the few biological processes discussed in our study, additional biological processes (e.g., pigment formation, carotenoid accumulation, formation of metabolic products, biofilm formation) are included in Figure 11 to give a more complete picture of ageing processes of BAP that may affect their radiative properties. Several of our results repeat findings from previous sensitivity studies of aerosol properties on the direct and indirect radiative effects. However, our study should be considered as guidance 665 to future field, lab and model studies to further characterize the role of biological particles in the atmosphere as their emissions, budgets and processing are currently poorly constrained (Khaled et al., 2020) compared to more abundant aerosol types, despite their unique characteristics of living organisms that may affect not only climate but also public health.