Impact of marine biogeochemistry on the chemical mixing state and cloud forming ability of nascent sea spray aerosol


Corresponding author: K. A. Prather, Department of Chemistry and Biochemistry, University of California, San Diego, La Jolla, CA 92093-0314, USA. (


[1] The composition and properties of sea spray aerosol, a major component of the atmosphere, are often controlled by marine biological activity; however, the scope of impacts that ocean chemistry has on the ability for sea spray aerosol to act as cloud condensation nuclei (CCN) is not well understood. In this study, we utilize a mesocosm experiment to investigate the impact of marine biogeochemical processes on the composition and mixing state of sea spray aerosol particles with diameters < 0.2 µm produced by controlled breaking waves in a unique ocean-atmosphere facility. An increase in relative abundance of a distinct, insoluble organic particle type was observed after concentrations of heterotrophic bacteria increased in the seawater, leading to an 86 ± 5% reduction in the hygroscopicity parameter (κ) at 0.2% supersaturation. Aerosol size distributions showed very little change and the submicron organic mass fraction increased by less than 15% throughout the experiment; as such, neither of these typical metrics can explain the observed reduction in hygroscopicity. Predictions of the hygroscopicity parameter that make the common assumption that all particles have the same bulk organic volume fractions lead to overpredictions of CCN concentrations by 25% in these experiments. Importantly, key changes in sea spray aerosol mixing state that ultimately influenced CCN activity were driven by bacteria-mediated alterations to the organic composition of seawater.

1 Introduction

[2] Atmospheric aerosols are known to play a significant role in forcing global climate directly through the scattering and absorption of solar radiation and indirectly by altering cloud microphysics and albedo [Forster et al., 2007]. Aerosol particles induce indirect effects through their ability to act as cloud condensation nuclei (CCN) in air that is supersaturated with respect to water vapor; the ability of an individual particle to do so depends on its size and chemical composition. Wettable particles with a dry diameter (Ddry) > 0.2 µm will activate at climatically relevant supersaturations (< 1%) almost regardless of their composition, while those containing highly soluble material may act as CCN with Ddry as small as 0.05 µm [McFiggans et al., 2006; Andreae and Rosenfeld, 2008]. As such, the composition of aerosol particles, particularly those with Ddry < 0.2 µm, can play an important role in determining the number concentration of subcloud CCN, which, combined with the dynamic nature of cloud supersaturation, is thought to ultimately determine the impact of aerosol particles on cloud drop size and number concentration [Hegg et al., 2012].

[3] A unifying explanation of the production and activity of CCN in the marine boundary layer (MBL) is not well established [Pierce and Adams, 2006; Meskhidze et al., 2011]. Using the hygroscopicity parameter, κ, for CCN activity comparisons [Petters and Kreidenweis, 2007], the global average value for the MBL is 0.72 ± 0.24 based on model simulations [Pringle et al., 2010] with individual in situ reports as low as κ = 0.17–0.3 (off the Southern California coast) [Furutani et al., 2008] and as high as κ = 1.15–1.4 (tropical Atlantic) [Good et al., 2010]. Factors that contribute to this variability include the chemical nature and mixing state of nascent sea spray aerosol (SSA), secondary chemical alterations to SSA during its residence time in the MBL, and the influence of continental/anthropogenic aerosols advected to marine regions [Kaku et al., 2006; Furutani et al., 2008; Sorooshian et al., 2009; Langley et al., 2010]. Anthropogenic and continental impacts on CCN activity in marine regions are particularly evident in the comparison of the hygroscopicity parameter in the North Atlantic (κ = 0.59 ± 0.18) with that over the Southern Ocean (κ = 0.92 ± 0.09) [Pringle et al., 2010], since anthropogenic impacts on atmospheric composition are less pronounced in the Southern Hemisphere on average [Chung and Seinfeld, 2002; Wofsy et al., 2011], indicating that anthropogenic influences tend to reduce the aerosol hygroscopicity in the MBL. Table 1 provides a review of a number of reports of CCN activity measurements from both laboratory and field studies (including this work). The assessment of modern anthropogenic aerosol impacts on cloud properties depends critically on the concentrations of natural aerosol [Menon et al., 2002; Lohmann and Feichter, 2005], although this information has been difficult to assess [Andreae, 2007], stressing the importance of detailed characterization of the sources and properties of natural aerosol particles.

Table 1. Selected CCN-Derived Hygroscopicity Parameter Values for SSA
Author, YearκNotes
  1. aκ calculated from reported Dact/S pairs.
  2. bLarge positive deviations in κ for the highest water activity (aw) are omitted here.
  3. cMobile sampling platforms (e.g., aircraft) are noted where applicable.
Unaltered Natural Seawater and Inorganic Proxies
Collins et al. (this work)1.4 (+0.3, −0.7)Natural seawater, sand filtered
King et al. [2012]1.25, 1.08NaCl and artificial seawater, respectively
Fuentes et al. [2011]v1.3–1.5“Seawater proxy” no organics; κ depends on aw
Niedermeier et al. [2008]1.3a, 1.05–1.25aNaCl(aq), natural seawater (various locations)
Petters and Kreidenweis [2007]1.28Calculated: aerosol inorganic model
Wave Channel Experiments (Collins et al., this work)
Seawater + algae culture0.88 (+0.34, −0.23)Dunaliella tertiolecta
Mesocosm (pregrowth)0.81 (+0.33, −0.40)Alteromonas spp. + ZoBell
Mesocosm (postgrowth)0.14 (±0.04)–0.21 (+0.06, −0.05)Alteromonas spp. + ZoBell
Mesocosm (with algae)0.12 (+0.02, −0.03)–0.20 (+0.06, −0.08)Alteromonas spp. + ZoBell + Dunaliella tertiolecta
Phytoplankton Exudates in Seawater Proxy [Fuentes et al., 2011]
Thalassiosira rotula1.21, 1.11Average κ values: 175 μM and 512 μM DOC, respectively
Chaetoceros sp.1.03, 0.95
Emiliania huxleyi1.14, 0.96
Phaeocystis cf. globosa1.15, 1.03
Phytoplankton Exudate Mixtures [2010bWex et al., 2010b]b
Thalassiosira rotula1.1Mixtures contained exudates from all types listed in varying proportions; entries identified by dominant exudate source; all cited κ values are averaged across various aw conditions
Chaetoceros sp.1.05
Emiliania huxleyi1.19
Phaeocystis cf. globosa1.19
Other Relevant Model and Natural Systems
King et al. [2012]~1.25a60 g NaCl + 1 g sodium laurate, aqueous
~0.2a0.27 g NaCl + 1 g sodium laurate, aqueous
Frosch et al. [2011]0.5750% oxalic acid, 50% NaCl
0.3880% oxalic acid, 20% NaCl
Moore et al. [2011]1.1a>Bilayer oleic acid on NaCl(aq)
1.3a>Bilayer sodium dodecyl sulfate on NaCl(aq)
Schwier et al. [2011]1.187NaCl + 0.001 M sodium oleate
0.869NaCl + 0.01 M sodium oleate
Clean Marine” Field Observationsc
Moore et al. [2012]0.15–0.2Aircraft; coastal Southern California
Martin et al. [2011]0.33–0.50Ship; arctic summer (range: 0.09–0.61)
Mochida et al. [2011]~0.6Ship; subarctic North Pacific
Kammermann et al. [2010]0.15Arctic summer (Sweden), S = 0.2%
Good et al. [2010]1.15–1.40Ship; tropical Atlantic
Bougiatioti et al. [2009]0.24Eastern Mediterranean Sea
Allan et al. [2008]0.6 ± 0.2Puerto Rico
Furutani et al. [2008]0.17–0.3aShip; coastal Southern California
Hudson [2007]0.87 ± 0.24Aircraft; near Antigua Island

[4] The production of SSA accounts for a large fraction of the total atmospheric burden of natural aerosol due to the vast extent of the Earth's oceans and the widespread strength of SSA production at the sea surface [Andreae and Rosenfeld, 2008; de Leeuw et al., 2011]. Particles directly ejected from the ocean with diameters of approximately 1 µm and larger are composed largely of salts, whereas particles smaller than 1 µm are more numerous [de Leeuw et al., 2011] and have been characterized as increasingly rich in organic matter (OMaero) with decreasing diameter on both mass [O'Dowd et al., 2004; Keene et al., 2007; Facchini et al., 2008] and number fraction basis [Ault et al., 2013; Prather et al., 2013]. Since cloud properties are sensitive to subcloud aerosol number concentrations, understanding the production flux and CCN activity of these small, organic-enriched particles is of critical importance to predictions of cloud albedo and lifetime [Lohmann and Feichter, 2005]. In particular, knowing whether the chemical components of the aerosol are mixed together in the same particle (internally mixed) or exist in separate particles (externally mixed) can play a major role in model predictions of CCN number concentrations [Wex et al., 2010b; Meskhidze et al., 2011]. Some laboratory and field studies of SSA have reported an internally mixed aerosol population [Kammermann et al., 2010; Fuentes et al., 2011], while others provide evidence of distinct, externally mixed subpopulations of particles [Murphy et al., 1998; Bigg and Leck, 2008; Hawkins and Russell, 2010; Hultin et al., 2010; Prather et al., 2013].

[5] Laboratory bubble bursting experiments indicate that nascent SSA from unaltered seawater exhibits κ values between 1.2 and 1.5, which are close to the value for pure NaCl (κ = 1.25–1.3), but well above many field observations (cf. Table 1). The composition of seawater and the resulting SSA are not globally homogeneous, however. Seasonal phytoplankton blooms that lead to regionally elevated dissolved and particulate organic matter (OMsea) concentrations in the ocean [Ducklow et al., 1995] have been associated with reports of strong enhancements in the organic mass fraction of submicron aerosol [O'Dowd et al., 2004; Yoon et al., 2007] at Mace Head, Ireland. Subsequent investigations into the influence of organic material on the CCN activity of sea salt/organic mixtures have mostly utilized model chemical systems (e.g., oleic acid, amino acids, surfactants) to simulate varying concentrations of OMsea that is ejected from the ocean, showing modest reductions in CCN activity with the addition of organics in many cases [Frosch et al., 2011; Moore et al., 2011; Schwier et al., 2011; King et al., 2012]. This weak influence of organics is likely related to the dominant effect of the highly water soluble inorganic components on the CCN activation of internally mixed salt/organic particles [Bilde and Svenningsson, 2004; Broekhuizen et al., 2004], the lack of long-range molecular order in the organic films [Davies et al., 2013], and the relatively low viscosity of the organic proxies [Shiraiwa et al., 2011; Bones et al., 2012].

[6] Recent laboratory studies of the influence of organic matter on the water uptake properties of SSA have begun to utilize more representative chemical proxies for dissolved organic matter (DOM), which is the dominant reservoir for reduced carbon in the ocean [Hansell et al., 2009] and is operationally defined as the organic fraction of seawater that can pass through a filter (pore size ranges 0.2–0.7 µm). Aerosolized DOM from two different marine environments showed different reductions in CCN activity when mixed with salts, likely resulting from their differing chemical compositions [Moore et al., 2008]. Studies that utilize phytoplankton exudates (i.e., phytoplankton culture filtrate) as DOM proxies [e.g., Fuentes et al., 2010] have concluded that the CCN activity of SSA can be reduced by 5–24% with organic carbon concentrations in the parent seawater up to ~7 times mean oceanic levels, depending on which organism produced the exudate material. This observed reduction in CCN activity corresponded to an OMaero fraction of 30–40% by volume [Wex et al., 2010a; Fuentes et al., 2011] (Table 1).

[7] In order to connect field and laboratory investigations of SSA production and physicochemical properties, a prescriptive relationship for including SSA organics in models has been proposed and investigated, linking the total submicron mass fraction of SSA with satellite-observable chlorophyll a (chl a) in the surface ocean [O'Dowd et al., 2008; Vignati et al., 2010; Albert et al., 2012], although this relationship is not well understood. The uncertainties associated with this initially proposed relationship have prompted the suggestion that a sea spray OMaero parameterization may be improved through the inclusion of carbon sources apart from primary production (e.g., heterotrophic bacteria, protozoa) [Quinn and Bates, 2011]. The concentration of heterotrophic bacterial biomass, for instance, is less variable than that of phytoplankton across the spectrum of eutrophic, mesotrophic, and oligotrophic regimes (5 × 105–5 × 10 7 cells mL−1) [Cho and Azam, 1990; Li, 1998]. Marine heterotrophic bacteria play a key role in the structuring of microbial ecosystems [Azam and Malfatti, 2007; Pomeroy et al., 2007] and influence the composition of oceanic DOM [Ogawa et al., 2001; Coble, 2007; Jiao et al., 2010], which could, in turn, exert control over the physicochemical properties of SSA.

[8] This study explores the coupled impact of marine heterotrophic bacterial activity and OMaero mixing state on the CCN activity of SSA. Taking advantage of the ability to produce sea spray aerosol using physically realistic breaking waves in a laboratory setting, the impact of seawater biological and chemical conditions were fully isolated over a range of organic matter concentrations and biogeochemical conditions. Rather than utilizing only phytoplankton-produced exudates as a controllable DOM surrogate, these experiments were carried out with the goal of exploring the impact of a broader marine biogeochemical system on SSA properties. Specifically, the mixing state of submicron SSA was investigated in an effort to probe the impact of external mixing on CCN activity and the relative response of the OMaero mass fraction to the observed changes in aerosol water uptake.

2 Experimental Methods

2.1 Aerosol Generation

[9] Measurements of aerosol properties were made during an intensive experiment organized by the Center for Aerosol Impacts on Climate and the Environment (CAICE). A novel ocean-atmosphere chamber was developed utilizing an existing glass-walled, 33 × 0.5 × 0.5 m (length-width-water depth, 8250 L) linear wave channel at the Hydraulics Laboratory at Scripps Institution of Oceanography ( [Prather et al., 2013]; further details can be found in the supporting information. Bubble size distributions generated by controlled breaking waves in this facility have been previously validated against measurements of open ocean waves [Deane and Stokes, 2002]. The wave channel was fitted with a custom clean air handling system and a sealed enclosure for the direct study of aerosol flux from the ocean to the atmosphere under controllable conditions with low-background aerosol [Prather et al., 2013]. Seawater was drawn from approximately 4 m below the low tide line 275 m offshore at Scripps Pier (La Jolla, CA; 32°52.0'N, 117°15.4'W), was passed through two No. 12 crystal sand bed filters to remove macroscopic organisms and debris, and was delivered directly to the wave channel. SSA was generated continuously in the sealed wave channel through the breaking of sinusoidal wave pulses on an artificial shoal with a frequency of 0.6 Hz or through a pulsed plunging waterfall technique, which involves the intermittent gravitational impingement of a waterfall in the wave channel at 6 s intervals to produce sea spray, similar to the system described by Stokes et al. [2013]. The size distributions of sea spray aerosol generated by each of these methods have been determined to agree well with one another [Prather et al., 2013]. A comparison of chemical composition between these different generation methods will be the topic of a separate manuscript.

2.2 Measurement Techniques

[10] Nascent SSA was sampled approximately 2 m downstream of the breaking wave location through 1.3 cm stainless steel tubing to a laminar flow manifold, where the aerosol was diverted to an array of online and off-line measurement techniques. A schematic of the experimental setup is provided in the supporting information.

2.2.1 Chemical Measurements of Sea Spray Aerosol Particles

[11] Single particle measurements of aerosol chemical composition were obtained from both real-time and off-line techniques. Size-resolved chemical composition of individual sea spray aerosol particles with dry (relative humidity (RH) = 15%) aerodynamic diameters (Da) between 0.3 and 1 µm was measured with Ultrafine Aerosol Time-of-Flight Mass Spectrometry (UF-ATOFMS) [Su et al., 2004], which obtains the aerodynamic diameter and laser desorption/ionization (266 nm, ~1.3 mJ, 7 ns pulse) dual polarity mass spectrum for each particle sampled. Analysis of this large data set is performed with the aid of the YAADA toolkit ( for MATLAB (The Math Works, Inc.). Data are reported here as 15 min average mass spectral peak areas normalized to the total area of the mass spectrum from which they were obtained.

[12] Samples were collected using a micro-orifice uniform deposit impactor (MOUDI; MSP Corp. Model 100) sampling air at 30 liters per minute at approximately 60% relative humidity, with 50% aerodynamic cutoff diameters at 1.0, 0.53, 0.30, 0.18, 0.09, and 0.05 µm for off-line microscopy analysis. Aerodynamic diameter bins generated by MOUDI cutoffs have been shown to agree well with independent particle size measurements by electron microscopy [Ault et al., 2013]. Aerodynamic diameters were converted to equivalent spherical diameter in order to calculate size distributions for each particle type. For this conversion, the salt and nonsalt fractions were assigned densities of 1.8 and 1.35 g mL−1 [Zelenyuk et al., 2007; Kuwata et al., 2012] and dynamic shape factors of 1.08 and 1 [Hinds, 1999], respectively. A growth factor of 1.6 was then used to convert the diameter at 60% RH to a dry diameter for comparisons and calculations involving aerosol size distributions (sections 3.5 and 3.6). Aerosol samples were deposited on 400 mesh Carbon Type B/Formvar TEM grids (Ted Pella Inc., part number 01814-F). Transmission electron microsocopy with energy-dispersive X-ray analysis (TEM-EDX) measurements were collected on a JEOL 2100f field emission TEM operated at an accelerating voltage of 200 kV with a Gatan high-angle annular darkfield detector and a Nanotrace EDX detector. In addition to TEM-EDX, scanning transmission X-ray microscopy (STXM) with near-edge X-ray absorption fine structure (NEXAFS) data from the carbon K-edge, sulfur L-edge, and chlorine L-edge were collected at the Advanced Light Source at Lawrence Berkeley National Laboratory [Kilcoyne et al., 2003; Moffet et al., 2010]. Submicron SSA was classified into particle types based on elemental composition (EDX), inorganic/organic ratio and molecular information (STXM-NEXAFS), and morphological analysis of single particles (TEM).

2.2.2 Physical Measurements of Sea Spray Aerosol Particles

[13] Aerosol size distributions were measured at a relative humidity of 15 ± 10% using a Scanning Mobility Particle Sizer (SMPS; TSI Inc. Model 3936) and an Aerodynamic Particle Sizer (APS; TSI Inc. Model 3321). SMPS measured particles with diameters of 0.011–0.6 µm. APS measurements were adjusted from Da to physical diameter (Dp) assuming an effective density of 1.8 g mL−1 [Zelenyuk et al., 2007]. Sizes measured by the APS (after adjustment) that overlap the capabilities of the SMPS (Dp < 0.6 µm) were truncated. After density adjustment, the largest size measured by the APS was 11 µm.

[14] CCN number concentrations (NCCN) were measured with a miniaturized streamwise thermal gradient cloud condensation nuclei counter (CCNc) [Roberts and Nenes, 2005]. Briefly, this technique exposes particles to an environment supersaturated with respect to water vapor, in which a subset of the particles activate to form cloud drops and are then counted using an optical particle counter. The CCNc sampled aerosol in parallel with a water-based condensation particle counter (CPC; TSI, Inc. Model 3781) to evaluate total aerosol concentrations and was operated at a constant supersaturation (S) of 0.2 ± 0.02%, which is similar to conditions found in marine stratocumulus cloud decks [e.g., Hudson, 1983]. In order to connect CCN measurements with the chemical composition of the particles, the hygroscopicity parameter (κ) was calculated based on the critical activation diameter (Dact) and S using code implemented in the R language ( based on the work of Petters and Kreidenweis [2007]. Calculations assumed the surface tension of water (72 mN m−1). Dact was determined by integrating the aerosol size distribution such that

display math(1)

where Dp is the physical aerosol diameter in micrometers, NCCN is the number concentration of cloud active aerosol at S = 0.2%, n(Dp) is the number size distribution of dry particles, and Dmax is the upper limit diameter measured by the APS. The hygroscopicity parameter is reported as the calculated κ value for the mean Dact based on 30 min averaged size distributions and NCCN. The κ value obtained for unamended seawater using this technique (Table 1) was validated with prior measurements performed by our group by using size-selected monodisperse aerosol populations and directly measuring the fraction of active CCN as a function of dry diameter. A lower limit uncertainty in κ was determined by accounting for systematic biases in the size distribution (due to under-counting). The SMPS-derived size distribution was scaled to the largest ratio of all integrated size distributions and parallel CPC measurements made during these experiments (NSMPS + NAPS/NCPC = 0.85), from which Dact and κ were recalculated. Uncertainty in κ is reported as the value corresponding to ±2σ of the 30 min Dact determinations within each reported data point, or the κ value obtained from aforementioned counting corrections, whichever accounts for a larger deviation from the mean.

2.2.3 Seawater Measurements

[15] Seawater chl a was measured in real time using a WET Labs ECO Triplet customizable fluorometer. Samples for seawater total organic carbon (TOC) analysis were collected from the upper ~5 cm of the tank, transferred (unfiltered) to combusted 40 mL glass vials, and immediately acidified to pH 2 with trace metal-free 12 N HCl. Analysis was later performed by high-temperature combustion (Shimadzu Scientific Instruments). TOC concentrations were linearly interpolated when needed for comparison with other variables (e.g., CCN activity). Water samples for cell counts were obtained using sterile plastic pipettes, immediately fixed with paraformaldehyde, and flash frozen in liquid nitrogen. Heterotrophic prokaryotic cell (bacteria and archaea) abundance was quantified by flow cytometry at the University of Hawai‘i School of Ocean and Earth Science and Technology Flow Cytometry Facility and will be referred to simply as bacteria herein.

2.3 Preparation and Addition of Biological and Organic Matter

[16] The organic matter content of the seawater in these experiments was modulated through two means: addition of preconcentrated algae monoculture and “mesocosm” growth of bacteria. An algae-only experiment utilized a monoculture of a common marine green microalgae species (Dunaliella tertiolecta) grown in Guillard's “f” media diluted by a factor of 2 (i.e., f/2 media) [Guillard and Ryther, 1962]. Direct lighting was not delivered to the culture after being added to the wave channel; therefore, sustained primary production during SSA generation was not expected.

[17] In a separate experiment which began after the wave channel was again filled with fresh unamended seawater, a bacterial mesocosm was initiated through the coincident addition of whole surface seawater from Scripps Pier (collected manually), a polyculture of three bacterial isolates (Alteromonas TW2, TW7, [Bidle and Azam, 2001], and AltSIO (B. E. Pedler and F. Azam, unpublished work, 2013)) grown on ZoBell media (5 g peptone and 1 g yeast extract per liter) [Oppenheimer and ZoBell, 1952], and an aliquot of sterile ZoBell growth media. Further additions of bacteria and ZoBell were performed over the ensuing 2.7 days, maintaining high levels of heterotroph-dominated biological activity with low chl a concentrations. The community of bacteria that grew in the mesocosm was not controlled or speciated. The initial seawater contained a natural assemblage of bacteria and the added cultures of Alteromonas spp. bacteria (Table 2) each contained less than 10% of the total number of bacteria already present in the mesocosm, similar to the proportion of Alteromonas to total heterotrophic bacteria found in a phytoplankton bloom in the North Pacific [Tada et al., 2011]. A final aliquot of ZoBell media, a culture of the heterotrophic bacterium Pseudoalteromonas atlantica (strain 19262), and a culture of Dunaliella tertiolecta (green algae) were added to the wave channel, bringing chl a concentrations up to 5.5 mg m−3. Due to the large volume of the wave channel, the volume of added organic and biological material did not significantly alter the final volume of the mixture. The details of each addition, labeled A1–A4, are shown in Table 2 and Figure 1. The unfettered growth of a natural assemblage of bacteria found in seawater is a key element of the mesocosm experiment, setting it apart from studies involving single biological or chemical components. The highly coupled biogeochemical interactions in this type of experiment allow naturally complex assemblages of biological and chemical species to interact.

Table 2. Mesocosm Experiment Additions
IDTime (days)Material AddedQuantitya
  1. aQuantities are approximate.
 0Seawater8250 L
A10.6Alteromonas spp. bacteria1 × 1010 cells
Whole seawater100 L
ZoBell media18 g C
A21.9Alteromonas spp. bacteria4.5 × 1012 cells
[ZoBell media (in culture)][13.5 g C]
A32.6ZoBell media15 g C
A42.8ZoBell media15 g C
P. atlantica bacteria1 × 1011 cells
Dunaliella tertiolecta algae1 × 1011 cells
Figure 1.

Temporal profile of (a) TOC and chl a in the seawater, (b) the hygroscopicity parameter (κ) of SSA, and (c) 15 min averaged SSA-bound organic nitrogen (UF-ATOFMS) juxtaposed with the concentration of bacteria in seawater for the mesocosm experiment. Error bars on the time axis of Figure 1b represent the averaging time for each determination. See text for discussion of uncertainty in κ. Vertical dashed lines denote additions of biological and/or organic material to the wave channel corresponding to those listed in Table 2. A1 denotes the initial major addition of Alteromonas spp. bacteria, ZoBell media, and unfiltered seawater. Note that the CNO signal increases upon each addition of fresh ZoBell media and/or seed cultures that contain as yet undegraded media. The periods Z1–Z4 were chosen to overlap with samples analyzed by TEM-EDX and STXM-NEXAFS and are described in detail. The phytoplankton-only experiment (section 3.2) is not shown.

3 Results and Discussion

3.1 Sea Spray From Unaltered Seawater

[18] Oceanic dissolved organic carbon (DOC; a standard metric for seawater DOM content) typically has a concentration < 80 μM in vast regions of the ocean during nonphytoplankton bloom conditions [Hansell et al., 2009]. The physicochemical properties of aerosol generated within this condition are particularly relevant for regions where biological activity is not elevated, and also act as a baseline control for experiments conducted with higher biological activity and organic carbon concentrations. Measurements of the CCN activity of SSA derived from breaking waves in unaltered coastal seawater (TOC = 69 ± 2 μM, chl a = 0.18 ± 0.03 mg m−3) yielded κ = 1.4 (+0.3, −0.7). It is notable that this hygroscopicity parameter value is similar to published values for artificial seawater and other inorganic proxies [Niedermeier et al., 2008; Fuentes et al., 2011], but is higher than many field observations made in the marine boundary layer (κ ~ 0.72; Table 1) [Hudson, 2007; Allan et al., 2008; Furutani et al., 2008; Bougiatioti et al., 2009; Good et al., 2010; Pringle et al., 2010; Martin et al., 2011; Mochida et al., 2011; Moore et al., 2012]. This difference is attributed to the inability of ambient aerosol measurements in the MBL to fully deconvolute influence from secondary marine, continental, and/or anthropogenic sources [Sorooshian et al., 2009; Langley et al., 2010; Shank et al., 2012] in addition to possible contributions from oceanic organic matter in SSA.

3.2 Sea Spray From Phytoplankton-Rich Seawater

[19] In order to draw initial parallels with studies that utilized phytoplankton exudates to simulate marine DOM [Wex et al., 2010a; Fuentes et al., 2011], the seawater was spiked with a culture of Dunaliella tertiolecta. As a result of the phytoplankton addition to the seawater, the chl a concentration in the seawater increased from 0.1 to 0.78 mg m−3, which is comparable to a spring phytoplankton bloom in the North Atlantic [Martinez et al., 2011], while the concentration of bacteria remained stable (1 ± 0.2 × 106 cells mL−1) at levels commonly found in the coastal ocean [Li, 1998]. We found a 37% reduction from the values of unaltered coastal seawater in CCN-derived hygroscopicity with the unfiltered phytoplankton culture used in this study (κ = 0.88 (+0.34, −0.23), TOC = 120–127 μM). This value indicates a comparable suppression in CCN activity to that found using filtered phytoplankton cultures as a proxy for DOM enrichment [Wex et al., 2010a; Fuentes et al., 2011]. For a given concentration of phytoplankton and/or organic matter, deviations in κ values obtained in these and other similar experiments are likely attributable to differences in the chemical composition of OMsea produced by each species [e.g., Kujawinski, 2011] and/or differences in the culture growth conditions. It is unlikely that the unfiltered nature of this algae culture significantly affected CCN-derived hygroscopicity, since the Dunaliella tertiolecta cells are far larger than Dact at S = 0.2% (cellular equivalent spherical diameter ~ 8 µm) [Stramski et al., 1993] and thus would not influence the determination of the hygroscopicity parameter.

3.3 Impact of Bacteria on Sea Spray Aerosol

[20] Marine bacteria are known to chemically alter the organic matter composition of seawater [Ogawa et al., 2001; Gruber et al., 2006; Coble, 2007; Jiao et al., 2010]; therefore, an evaluation of the impact of bacterial processes on SSA composition and CCN activity must be made by continuously monitoring changes in aerosol properties, along with changes in biological metrics in the seawater. Coincident time-resolved measurements of single particle chemical composition, CCN, chlorophyll concentration, and bacterial abundance in the seawater during the mesocosm experiment are presented in Figures 1a–c, demonstrating the interplay between marine biogeochemistry, aerosol composition, and CCN activity. Using the CNO ion (m/z = −42) as a marker for organic nitrogen species in the UF-ATOFMS mass spectra, inclusion of the peptide-rich ZoBell media in SSA is observed immediately following each addition of this material to the seawater in the wave channel (Figure 1c). Figure 2 summarizes the hygrosocopicity parameters (κ) of all experiments presented herein with respect to seawater TOC, compared with the results of Fuentes et al. [2011], who utilized phytoplankton exudates as a DOM proxy. SSA hygroscopicity was reduced by 42% to κ = 0.81 (+0.33, −0.40) (Figure 2, period Z1) in association with the initial addition of bacteria culture, ZoBell media, and unfiltered seawater (A1) at t = 0.6 days. The TOC concentration in the seawater after this initial addition averaged 89 μM, approximately 20 μM higher than the unaltered seawater control, which is slightly higher than typical springtime regional averages [Sohrin and Sempere, 2005], but is well below values observed at the ocean surface in regions of high biological activity [Engel et al., 2012]. This initial suppression of aerosol hygroscopicity is attributed primarily to the influence of just the ZoBell media on the OMaero content. Similar to the case with the phytoplankton-only experiment described above, all bacteria added at t = 0.6 days are large enough to be CCN active at S = 0.2% regardless of any chemical effect (average diameter of 0.9 µm) should they be ejected as whole cells, and as such would be insignificant in CCN-derived hygroscopicity parameter measurements.

Figure 2.

Scatter plot of κ values related to the total organic carbon content of the seawater from which the SSA was derived, compared with a study utilizing phytoplankton culture exudates as a DOM proxy. Labels Z1–Z4 correspond to the shaded time periods in Figure 1.

[21] CCN-derived hygroscopicity was further reduced to κ = 0.14 (±0.04)–0.21 (+0.06, −0.05) (Figures 1 and 2, triangles) at t = 1.25 days (Δt = 0.65 days after A1), coincident with an increase in seawater bacteria concentration and a reduction in aerosol-phase organic nitrogen, measured by UF-ATOFMS (cf. Figure 1 periods Z1–Z2). The change in organic nitrogen could indicate bacterial assimilation of peptidic or amino acid components of ZoBell media. This 86 ± 5% reduction in κ from the base seawater case with only 110 μM TOC starkly contrasts prior laboratory studies which indicate relatively modest reductions in CCN activity even with >500 μM dissolved organic carbon in the seawater [Wex et al., 2010a; Frosch et al., 2011; Fuentes et al., 2011; Schwier et al., 2011]. The reduced CCN-derived hygroscopicity can be explained by an increase in the abundance of organic carbon-rich particles with D < 0.2 µm, as shown by Ault et al. [2013]; the impact of aerosol mixing state on CCN activity will be shown in detail in section 3.5. STXM-NEXAFS measurements of a collection of individual, internally mixed salt/organic particles also indicate an increase in the organic volume fraction of approximately 5% (Figure S5). To date, this represents the largest observed change in CCN activity for ocean-relevant TOC (50–120 μM) [Sohrin and Sempere, 2005; Engel et al., 2012] and heterotrophic bacteria concentrations (approximately 5 × 105–5 × 107 cells/mL) [Cho and Azam, 1990; Li, 1998]. Teeling et al. [2012] recently observed heterotrophic bacteria concentrations of ~3.6 × 106 cells mL−1 in the North Sea following periods of high photosynthetic biological activity (high chl a), which acted as the carbon source for the observed surge in bacterial growth in that locale. While the labile organic carbon source in this study was ZoBell growth media, the change in bacterial abundance in this mesocosm experiment was similar to that observed by Teeling et al. [2012] and may have produced SSA with similar properties.

[22] The reduction in CCN activity upon the increase in bacterial abundance and depletion of organic nitrogen was attributed to the bacteria-mediated chemical degradation of ZoBell media, which consists predominantly of peptides. In addition, the chemical composition of DOM in the wave channel was also likely impacted by senescent phytoplankton cells, based on the observed reduction in chl a (Figure 1a). The correlation between the observed changes in UF-ATOFMS and CCN-derived hygroscopicity measurements with changes in the concentration of bacterial and phytoplankton biomass in the seawater suggests that the combined effect of the complex biogeochemical interactions occurring in this simulated post-phytoplankton bloom regime influences the physicochemical properties of SSA. Other similar mesocosm experiments have found that bacterial abundance tends to increase as phytoplankton bloom activity begins to subside [Smith et al., 1995], which was also found to be associated with an increase in bacterial surface enzyme activity (which would lead to chemical transformation of OMsea), colonization of phytoplankton particulates, and changes in bacterial community composition [Riemann et al., 2000]. A recent field study also indicated that sequential changes in bacterial community composition occur as a result of growth substrate changes in the ocean [Teeling et al., 2012]. The results presented in this work, coupled with an understanding of biogeochemical dynamics in the mesocosm, demonstrate that the physicochemical properties of SSA, at least in this study, are strongly affected by the chemical composition of OMsea and that biologically mediated changes in organic matter composition can significantly influence the mixing state and overall CCN activity of SSA.

3.4 Phytoplankton Addition After Bacterial Growth

[23] The hygroscopicity parameter did not show significant changes after the addition of phytoplankton to the mesocosm (Figure 1, after A4). Curiously, the relatively weak dependence of κ on TOC after A4 was qualitatively similar to experiments where phytoplankton exudates dominate the composition of DOM described in section 3.2 and in prior studies [Wex et al., 2010a; Fuentes et al., 2011]. Since the chemical composition of DOM that has been influenced by marine bacteria differs from that of freshly produced DOM by photoautotrophs [e.g., Shimotori et al., 2012], the observed differences in κ with respect to TOC between heterotrophic (bacteria-dominated; mesocosm pre-A4) and autotrophic (phytoplankton-dominated; algae-only experiment and mesocosm post-A4) regimes suggest that the chemical composition of OMsea is driving the enrichment and/or physicochemical nature of OMaero. Bacterial processing of DOM with little-to-no change in TOC in these experiments (TOCZ2–TOCZ1 = 0 μM) shows a stronger effect on the CCN activity of SSA (κZ2κZ1 = −0.6) than large changes in TOC when nondegraded phytoplankton exudates are added to the seawater (TOCZ4–TOCZ2 = 324 μM, κZ4κZ2 = −0.09). Two physical possibilities could have forced the observed reduction in κ : (1) the influence of bacteria-processed OMsea could be enhancing the formation of a distinct population of insoluble organic particles, similar to those observed by Facchini et al. [2008], or (2) the OMsea associated with bacterial activity is more hydrophobic than OMsea derived from phytoplankton exudates, which have been shown to be rich in carbohydrates [e.g., Aluwihare and Repeta, 1999], in which case this chemical change to OMsea would then be simply transferred to OMaero upon SSA production. These options are investigated through a closure study of CCN activity below.

3.5 CCN Activity and Mixing State

[24] Since CCN activity is driven by both size and chemical composition, closure between theoretical and observed CCN can be achieved if the role of size and chemical composition is accurately represented in the theoretical prediction. Despite induced changes in the organic and biological content of the seawater in these experiments, the measured size distributions of nascent SSA, shown in Figure 3, did not indicate significant systematic variability such that the reduction in CCN activity could be explained on the basis of particle size. Prior studies have reported changes in the size distribution of nascent SSA as a function of OMsea concentration [e.g., Sellegri et al., 2006; Fuentes et al., 2010; King et al., 2012]. Our experiments are possibly limited by low concentrations, especially for Dp < 0.1 µm. The lack of significant change in size distributions during these experiments demonstrates the impact of aerosol chemical composition and mixing state on the CCN activity measurements presented herein.

Figure 3.

Aerosol size distributions used for CCN activity determinations in this study. Note that neither the magnitude nor the shape of the SMPS-derived size distribution changed significantly throughout this set of experiments. Since all particles sampled by the APS would theoretically act as CCN based on their large size alone [McFiggans et al., 2006], total particle concentration for Dp = 0.6–11 µm is shown at the right for simplicity. The top axis shows the scale converted to the r80 particle size metric used in many sea spray flux studies [de Leeuw et al., 2011].

[25] Most studies that strive for CCN closure utilize bulk chemical analysis to drive predictive calculations based on the known hygroscopicity of various substances [e.g., Broekhuizen et al., 2006; Martin et al., 2011]. Often, the incorporation of external mixing assumptions are required to gain closure due to the importance of the number concentration-weighted mixing state of particles rather than widely used bulk aerosol mass concentration measurement techniques [Wex et al., 2010b]. In this set of experiments, two major, distinct particle types occur in the submicron size range [Prather et al., 2013]. Each of these distinct particle types was identified using TEM-EDX and STXM-NEXAFS analysis for diameters < 0.3 µm: an internally mixed salt/organic particle type (SS-OC) and an insoluble organic type that contains small amounts of inorganic sulfur (OC), along with a fraction labeled “Other” which contains particles identified as contaminants [Ault et al., 2013]. Prather et al. [2013] show that the OC type in these experiments was correlated with a population of particles that had a subsaturated hygroscopic growth factor < 1.2 (RH = 93%) and is thus descriptively labeled “insoluble” here. After stimulated bacterial growth, the OC type began to dominate at Dp < 0.2 µm and down to 0.03 µm. The bars in Figure 4a show the size-resolved submicron chemical composition as observed by TEM-EDX for each of the four periods, Z1–Z4.

Figure 4.

(a) TEM-EDX size resolved number fractions of particles superimposed with a fit of the size-resolved SS-OC fraction and the size-resolved fraction of mode 2 + mode 3 particles. The sigmoid fits to the TEM-EDX data were used for interpolation in ensemble volume and mass fraction calculations. All four of the sigmoid fits are shown superimposed on one another in Figure S3. (b) The size distributions for each of the four periods are fitted with three lognormal component distributions. Mode 1 represents the insoluble organic particle type, while modes 2 and 3 represent the SS-OC particles. The height of mode 1 was constrained by the mixing state observations made by TEM-EDX, as described in the text.

[26] In order to bring closure to the changes in relative abundance of each type and illustrate their changes in number concentration constrained by the size distributions (which remained largely unchanged throughout all of the experiments), three lognormal modes were fitted to the size distribution for each of the four periods, Z1–Z4 (Figure 4b). Since the two main particle types (SS-OC, OC) were distinct in composition, it is expected that each should exist with its own lognormal distribution of sizes. Mode 1, with the smallest modal diameter, is considered to consist entirely of the OC type, while the two larger modes (modes 2 and 3) were considered to be composed of SS-OC particles based on the size-resolved trend in the TEM-EDX analysis (Figure 4a). The height of mode 1 was constrained such that the reconstructed fraction of SS-OC (Figure 4a, red line) came into agreement with the TEM-EDX observations for each of the four cases. The change in magnitude of mode 1 in relation to modes 2 and 3 illustrate that underlying changes in the relative abundances of the particle populations occurred despite the lack of significant alterations in the measured size distribution magnitude and/or shape. Figure 4 shows that the insoluble organic particle type (mode 1) increases relative to the SS-OC type upon the growth of bacteria (between periods Z1 and Z2). Due to the fact that the shape and magnitude of the size distributions did not change significantly, the SS-OC particle type appears to have been replaced by OC particles at smaller sizes with changing bacterial abundance in the seawater.

[27] In order to illustrate the importance of external mixing to CCN activity in these experiments, the predicted hygroscopicity parameter (κpred) values are calculated using the simplifying assumption that the population is completely internally mixed and that the salt and organic components do not interact with one another in the mixture. The size-resolved chemical mixing state provided by TEM-EDX and STXM-NEXAFS observations allowed for the calculation of the ensemble submicron organic volume fraction assuming that the OC type is entirely organic and volume fractions for the SS-OC type were derived from the STXM-NEXAFS analysis (see supporting information). The Zdanovskii-Stokes-Robinson mixing rule [Stokes and Robinson, 1966] was used to generate κpred using equation (2):

display math(2)

where εsalt and εorg are the bulk volume fractions of salt and organic material, respectively, κsalt is assigned a value of 1.25 (see Table 1), and κorg is assigned a value of 0.006 [Moore et al., 2008]. The modal organic volume fraction based on STXM-NEXAFS measurements was used for εorg, with the remainder of the volume assigned to εsalt. For period Z3, insufficient statistics precluded construction of a histogram; the value of εorg is estimated to be 0.45. These data were not available for period Z4 and are omitted. Figure 5 shows the results of this calculation for periods Z1–Z3, compared with the measured values of κ during each of these time frames. The value of κpred deviates positively from the measurements after the growth of bacteria, when the OC type dominates the size distribution at Dp < 0.2 µm (Figure 4, periods Z2–Z4), stressing that accurate characterization of mixing state in this case is critical to predicting CCN activity. The use of κpred based on the assumption of internal mixing in this case would overpredict CCN concentrations (S = 0.2%) by approximately 25% for periods Z2 and Z3. The OC particle type, while not contributing significant aerosol volume to the submicron particle population, has a large impact on κ due to the sensitivity of this parameter to the number concentration and hygroscopicity of each particle type. The number concentration of less hygroscopic OC particles is higher than that of salt-containing aerosol in the smallest sizes, forcing Dact to become larger (and κ to become smaller) as bacterial abundance increases and OMsea composition is influenced. Hence, these results indicate that bacterial growth in the seawater altered the physicochemical properties of OMsea in a way that influenced the mixing state and CCN activity of nascent SSA.

Figure 5.

Predicted κ based on the assumption of volume-weighted internal mixing (equation (2)) compared with measured values. Note that the predicted κ only overestimates hygroscopicity when the less hygroscopic mode 1 particles begin to outnumber the more hygroscopic mode 2 particles (Figure 4) through the influence of bacterial activity on the seawater organic composition. Data for Z4 are omitted here since organic volume fractions derived from STXM-NEXAFS data were unavailable.

3.6 Organic Mass Fraction

[28] In order to compare these results with prior studies of the variability of OMaero with seawater composition, the bulk submicron organic mass fraction of SSA was calculated. Figure 6 shows the calculated bulk organic mass fractions for each sample time frame (Z1–Z3) as a function of chl a in comparison with parameterized data sets from the literature. Aerosol number size distributions were determined for each of the two main particle types based on sigmoid curves fit to the size-resolved TEM-EDX and STXM-NEXAFS number fractions of SS-OC particles (Figure 4a). Aerosol mass fractions are determined using an organic matter density of 1.35 g mL−1 [Kuwata et al., 2012] and a salt density of 1.8 g mL−1 [Zelenyuk et al., 2007]. The organic volume fraction of the mixed salt/organic particle type was determined by calculating the organic and salt thickness based on STXM-NEXAFS measurements of single particle composition of approximately 300 particles in each time period (see supporting information). The OC type was taken to be 100% organic for the purposes of this analysis and, due to the presence of some inorganic material, may produce an upper limit estimate. For samples collected during Z1 and Z2, frequency distributions of individual particle organic volume fraction values were obtained (Figure S5). Insufficient data were available to generate a reliable histogram for Z3 in this manner; the estimated range reported herein was based on a limited data set.

Figure 6.

Submicron OMaero mass fraction is calculated considering two particle types observed by microscopy: a mixed salt/organic type and a separate insoluble organic type. The vertical bars indicate the range of submicron OMaero mass fraction values expected based on the distribution of organic volume fractions in the SS-OC particle type observed by STXM-NEXAFS (Figure S5). The mode of the distribution of organic volume fractions is shown by the filled circles. A statistically valid histogram could not be generated for Z3, so only the estimated range is indicated. Filled triangles show the calculated OMaero mass fraction when neglecting OM in the SS-OC particle type, indicating the organic mass contained in only the insoluble organic type. Published parameterizations of the OMaero-chl a relationship are superimposed for direct comparison.

[29] It is important to note that changes in the organic mass fraction are small (<15%; Figure 6), despite more prominent changes in the fractional contribution of each particle type by number, since the insoluble organic subpopulation of particles exists mainly in the smaller size range (<0.2 µm; Figure 4). The relative invariance of the OMaero mass fraction between periods Z1 and Z2, which corresponds to a large reduction in κ, emphasizes that the enhancement of a subpopulation of small, less hygroscopic particles can exert a strong influence on CCN activity, but may not be clearly resolved by aerosol mass measurements. As shown in Figures 4 and 5, changes in κ are driven in this case by the relative abundances of two different particle types, each containing OMaero, but with drastically different hygroscopicity, despite only minor changes in the sheer quantity of organic material present in the SSA particle population.

[30] Published parameterizations of the submicron OMaero mass fraction based on oceanic chl a concentrations [Vignati et al., 2010; Fuentes et al., 2011; Gantt et al., 2011] are superimposed on the mass fraction calculations given in Figure 6. The mixing ratio of organic matter within the SS-OC type, due to its dominance in particles with D > 0.2 µm, exerts most of the organic mass variability in SSA overall. Prior studies have reported bulk organic mass fractions as high as 77–80% [Keene et al., 2007; Facchini et al., 2008] and as low as ~4% [Modini et al., 2010] in SSA generated from natural seawater samples. The ensemble submicron organic mass fraction in these experiments was approximately 35–40% for periods Z1 and Z2 (Figure 6). The OMaero mass fractions found in this work deviate positively from published parameterizations at low values of chl a. This is likely due to the fact that the addition of organic matter to the seawater at A1 did not include the addition of phytoplankton (and hence chl a), and in fact, significant phytoplankton senescence was observed (shown as a decay in chl a; Figure 1). The modal organic volume fraction within the SS-OC particles increased between Z1 and Z2 (Figure S5) as chl a decreased, opposite the expected trend based on current parameterizations [Vignati et al., 2010; Fuentes et al., 2011; Gantt et al., 2011], which do not consider possible differences in the SSA composition between increasing and decreasing chl a trends. In a natural phytoplankton bloom ecosystem, such as those studied biologically by Teeling et al. [2012] or Smith et al. [1995], autotrophic activity imparts DOM to the seawater, followed by increases in the populations of heterotrophic organisms (e.g., bacteria). The experiments reported herein suggest that marine biogeochemical interactions in the postbloom period induce changes in the chemical mixing state and CCN activity of SSA. We emphasize that in these experiments, CCN activity changed without increasing chl a or the bulk submicron OMaero fraction. Instead, the importance of the relative abundances of distinct particle types with significantly different compositions and hygroscopicity were observed.

4 Implications for Clouds and Climate in Marine Environments

[31] Biological processes actively and dynamically influence the chemical composition of oceanic DOM. Associated with organic matter-producing autotrophic microbes (e.g., phytoplankton) through the microbial loop [Pomeroy et al., 2007], heterotrophic bacteria and their associated enzymes serve as the primary reactive surface responsible for chemical transformation of DOM [Azam and Malfatti, 2007; Jiao et al., 2010]. While the molecular-level details have not been thoroughly characterized to date, bacterial activity is known to alter the chemical composition of DOM [Gruber et al., 2006; Coble, 2007; Kujawinski et al., 2009; Flerus et al., 2012]. In stark contrast with laboratory sea spray experiments that use phytoplankton exudates as an OMsea chemical proxy [Wex et al., 2010a; Fuentes et al., 2011], the results presented in this study demonstrate the importance of including a more holistic representation of the biological complexity of marine microbial assemblages. Since the ratio of bacterial production to phytoplankton production in the ocean varies based on environmental conditions [e.g., Hoppe et al., 2002], the localized biogeochemical state of the ocean will likely influence the CCN properties of nascent SSA through differing compositions of OMsea. The ratio of bacterial carbon to phytoplankton carbon increases almost asymptotically as chl a concentration approaches zero in the ocean, owing to the relatively stable abundance of marine heterotrophic bacterial biomass (5 × 105–5 × 10 7 cells mL−1) [Cho and Azam, 1990]. This emphasizes the significant influence that heterotrophic bacteria can exert on the composition of OMsea through their ability to consume, utilize, and produce DOM, thus impacting the composition, mixing state, and cloud activity of SSA independently of phytoplankton abundance as estimated by chl a concentrations.

[32] Recent global modeling efforts have used the OMaero-chl a relationship to derive aerosol composition (with hygroscopicity assumptions) [Meskhidze et al., 2011; Westervelt et al., 2012] due to the near global coverage of satellite-derived oceanic chl a measurements. While this approach may closely approximate OMaero mass fractions in some regions, the results of this study indicate that more detailed information may be necessary for the derivation of CCN concentrations. Still, aerosol mixing state information has been shown to be crucial in constraining marine CCN concentrations [Meskhidze et al., 2011]; experimental evidence for the importance of external mixing to the CCN properties of nascent SSA is provided herein. Due to the lack of significant changes in the size distribution despite the increasing relative abundance of a distinct insoluble organic particle type (Figure 4), enrichment of OMaero resulting from elevated oceanic biological activity appears to replace salt-containing particles in this experiment with concomitant changes in OMsea. This type of behavior has been shown in model simulations to reduce globally averaged CCN concentrations compared to cases where the SSA source function disregards organic matter [Westervelt et al., 2012]. The size dependence of the insoluble organic particle type supports observations of decreasing hygroscopicity with decreasing particle size [Kaku et al., 2006] and likely plays a strong role in buffering aerosol-mediated changes in cloud microphysics due to the dynamic nature of water vapor supersaturation within marine stratocumulus clouds [Hegg et al., 2012]. The observed close connection of the CCN activity of SSA to heterotrophic bacterial activity within a dynamic microbial ecosystem stresses the climatic influence of biosphere/atmosphere coupling. Investigations into the physicochemical properties of SSA that are integrated with detailed marine microbiological and chemical analyses will further elucidate the interactions between marine biogeochemical processes and climate-relevant aerosol properties.


[33] This work was funded through the Center for Aerosol Impacts on Climate and the Environment (CAICE), a National Science Foundation Phase I Center for Chemical Innovation (CHE-1038028). The Advanced Light Source is supported by the Director, Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under contract DE-AC02-05CH11231. F. Azam was supported by grants from NSF OCE0962721 and the Gordon and Betty Moore Foundation Marine Microbial Initiative. The authors would like to thank all investigators involved in the CAICE intensive campaign, with special thanks to Christopher Cappa, Grant Deane, Dale Stokes, John Seinfeld, Scott Hersey, Wilton Mui, and Michelle Kim for their valuable contributions. Nathan Schoepp and William Lambert provided algae cultures through the San Diego Center for Algae Biotechnology (SD-CAB) greenhouse facility. The Bartlett Lab at Scripps Institution of Oceanography supplied the culture of Pseudoalteromonas atlantica. Dave Aglietti, John Lyons, Paul Harvey, and Charles Coughran of the Scripps Institution of Oceanography Hydraulics Laboratory provided valuable assistance and expertise. The authors also thank the anonymous reviewers for their highly constructive comments. Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.