Dust-Associated Airborne Microbes Affect Primary and Bacterial Production Rates, and Eukaryotes Diversity, in the Northern Red Sea: A Mesocosm Approach

The northern Red Sea (NRS) is a low-nutrient, low-chlorophyll (LNLC) ecosystem with high rates of atmospheric deposition due to its proximity to arid regions. Impacts of atmospheric deposition on LNLC ecosystems have been attributed to the chemical constituents of dust, while overlooking bioaerosols. Understanding how these vast areas of the ocean will respond to future climate and anthropogenic change hinges on the response of microbial communities to these changes. We tested the impacts of bioaerosols on the surface water microbial diversity and the primary and bacterial production rates in the NRS, a system representative of other LNLC oceanic regions, using a mesocosm bioassay experiment. By treating NRS surface seawater with dust, which contained nutrients, metals, and viable organisms, and “UV-treated dust” (which contained only nutrients and metals), we were able to assess the impacts of bioaerosols on local natural microbial populations. Following amendments (20 and 44 h) the incubations treated with “live dust” showed different responses than those with UV-treated dust. After 44 h, primary production was suppressed (as much as 50%), and bacterial production increased (as much as 55%) in the live dust treatments relative to incubations amended with UV-treated dust or the control. The diversity of eukaryotes was lower in treatments with airborne microbes. These results suggest that the airborne microorganisms and viruses alter the surface microbial ecology of the NRS. These results may have implications for the carbon cycle in LNLC ecosystems, which are expanding and are especially important since dust storms are predicted to increase in the future due to desertification and expansion of arid regions.


Introduction
Aerosols impact marine ecosystems by delivering macro-and micronutrients to surface seawater upon deposition [1][2][3]. These nutrients typically induce an increase in phytoplankton abundance and bacterial biomass and activity [4][5][6][7]. Atmospheric deposition also supplies a diverse array of microbes to marine ecosystems [8][9][10], of which up to 25% remain viable [11]. Upon deposition, Figure 1. An illustration of the experimental mesocosm used in this study. Treatments included unamended controls, live dust and UV-treated dust run in triplicate 300 L transparent bags.

Bacterial Production (BP) and Primary Production (PP)
To quantify the impacts of airborne microbes on heterotrophic production, BP was measured using the (4,5-3H)-leucine incorporation method [26]. Briefly, seawater samples collected daily from the mesocosms (1.7 mL) were amended with 10 nmol of leucine L −1 (Perkin Elmer, specific activity 156 Ci mmol −1 ) and incubated for 4-6 h in the dark. Incubations were stopped by the addition of 100 µL of cold 100% trichloroacetic acid (TCA). Control samples containing seawater, radioisotope, and TCA added immediately upon collection were also run daily. At the lab, the samples were microcentrifuged twice with TCA 5% and 1 mL of scintillation cocktail (Ultima-Gold) was added to each vial. Disintegrations per minute (DPM) were measured using a TRI-CARB 2100 TR (Packard) liquid counter. A conversion factor of 1.5 kg C mol −1 per mole leucine was used [27].
To assess the impacts of airborne microbes on autotrophic production, the PP was measured following the 14 C incorporation method [28]. Briefly, water samples were analyzed in triplicate with dark and zero time controls. The samples (50 mL) were collected at 8:00 a.m. into transparent polycarbonate bottles (Nalgene) and amended with 5 µCi of NaH 14 CO3 (Perkin Elmer). The bottles were incubated for 4-6 h in the same pool where the mesocosms were placed. The incubations were terminated by filtering the spiked seawater onto GF/F filters under low pressure (<50 mmHg). Excess 14 C-bicarbonate was removed from the filters by acidification with HCl (32%) overnight. After adding 5 mL of scintillation cocktail (Ultima-Gold) to each vial, the radioactivity was measured using a TRI-CARB 2100 TR (Packard) liquid counter.

Chlorophyll-a (Chl-a)
To measure Chl-a, a proxy for total phytoplankton biomass, subsamples of seawater (500 mL) collected from the different mesocosms were passed through a Whatman GF/F filter, and 90% acetone solution was used for overnight extraction. The Chl-a concentrations were quantified using the non-

Bacterial Production (BP) and Primary Production (PP)
To quantify the impacts of airborne microbes on heterotrophic production, BP was measured using the (4,5-3H)-leucine incorporation method [26]. Briefly, seawater samples collected daily from the mesocosms (1.7 mL) were amended with 10 nmol of leucine L −1 (Perkin Elmer, specific activity 156 Ci mmol −1 ) and incubated for 4-6 h in the dark. Incubations were stopped by the addition of 100 µL of cold 100% trichloroacetic acid (TCA). Control samples containing seawater, radioisotope, and TCA added immediately upon collection were also run daily. At the lab, the samples were microcentrifuged twice with TCA 5% and 1 mL of scintillation cocktail (Ultima-Gold) was added to each vial. Disintegrations per minute (DPM) were measured using a TRI-CARB 2100 TR (Packard) liquid counter. A conversion factor of 1.5 kg C mol −1 per mole leucine was used [27].
To assess the impacts of airborne microbes on autotrophic production, the PP was measured following the 14 C incorporation method [28]. Briefly, water samples were analyzed in triplicate with dark and zero time controls. The samples (50 mL) were collected at 8:00 a.m. into transparent polycarbonate bottles (Nalgene) and amended with 5 µCi of NaH 14 CO 3 (Perkin Elmer). The bottles were incubated for 4-6 h in the same pool where the mesocosms were placed. The incubations were terminated by filtering the spiked seawater onto GF/F filters under low pressure (<50 mmHg). Excess 14 C-bicarbonate was removed from the filters by acidification with HCl (32%) overnight. After adding 5 mL of scintillation cocktail (Ultima-Gold) to each vial, the radioactivity was measured using a TRI-CARB 2100 TR (Packard) liquid counter.

Chlorophyll-a (Chl-a)
To measure Chl-a, a proxy for total phytoplankton biomass, subsamples of seawater (500 mL) collected from the different mesocosms were passed through a Whatman GF/F filter, and 90% acetone solution was used for overnight extraction. The Chl-a concentrations were quantified using the non-acidification method [29], using a Trigoly fluorimeter equipped with 436 nm excitation and 680 nm emission filters.

Picophytoplankton and Heterotrophic Bacterial Abundance
To quantify the abundance of picophytoplankton and heterotrophic bacteria, seawater subsamples (1.7 mL) were collected from the different mesocosm bags, fixed with 50% glutaraldehyde (0.15% final concentration, Sigma G7651), incubated for 10 min at room temperature, snap frozen in liquid nitrogen, and stored in −80 • C until analyses within a few weeks. Cell counts were performed by flow cytometry (Attune, Applied Biosystems) equipped with a syringe-based fluidic system and 488 nm and 405 nm lasers. Cyanobacteria (Synechococcus and Prochlorococcus) and picoeukaryotes were detected based on the orange fluorescence of phycoerythrin (585 nm) and the red fluorescence of Chl-a (630 nm), side scattered and forward scattered at a flow rate of 100 µl min −1 . Heterotrophic bacterial cells were first stained with a SYTO9 solution for 10 min in the dark and then run at a low flow rate of 25 µL min −1 using a discrimination threshold of green fluorescence (520 nm). One µm beads (Polysciences) were used as internal reference.
2.6. β-Glucosidase (β-Glu), Aminopeptidase (AMA), and Alkaline Phosphatase (APA) Activity To determine the rate of polysaccharide degradation by bacteria, β-Glu activity was determined by the 4-methylumbelliferyl-β-D-glucopyranoside (Sigma M3633) method [30], and AMA was determined by the L-Leu-7-amido-4-methyl-coumarin method. Substrate was added in triplicate to 1 mL water samples (final concentration of 50 µM) and incubated in the dark at an ambient temperature for 24 h. To assess the rate of scavenging of organic matter due to phosphate limitation, the APA was determined by the 4-methylumbeliferyl phosphate (MUF-P: Sigma M8168) method [31]. Substrate was added in triplicate to 1 mL water samples (final concentration of 50 µM) and incubated in the dark at an ambient temperature for 24 h. The increase in fluorescence of 4-methylumbelliferone (MUF) was measured at 365 nm excitation, 455 nm emissions (GloMax®-Multi Detection System E9032) and calibrated against a MUF standard (Sigma M1508).

DNA Extraction, Library Preperation, and Sequencing
To assess the diversity of microorganisms, seawater (1 L) from each mesocosm was filtered onto 0.2 µm polycarbonate filters. Total DNA was extracted from the filters using the phenol chloroform method [32]. Total DNA was also extracted from the aerosols collected using the same method. Polymerase chain reaction (PCR) using primers 515 (forward) and 806 (reverse) for 16S rRNA and primers EUK7F (forward) and EUK570R (reverse) for 18S rRNA, with barcodes on the forward primer, were carried out using the HotStarTaq Plus Master Mix Kit (Qiagen, Valencia, CA, USA). The samples were pooled together in equal proportions (based on their MW and DNA concentrations), purified using calibrated Ampure XP beads, and used to prepare libraries using a Nextera DNA Sample Preparation Kit (IllFumina). Libraries were loaded to a 600 Cycles v3 Reagent cartridge (Illumina) and sequenced by illumina MiSeq.

Statistical Analysis
The data in figures and tables are means and standard deviations (n = 3). Differences between treatments were tested using analysis of variance (ANOVA) and Tukey's post-hoc testing, and a p-value of 0.05 was used to determine significance unless noted otherwise. All tests were performed using R.
Although we reported results for all time points, we only included results from 20 h and 44 h after dust additions, and not 72 h, in our statistical analysis. These time intervals were selected because this was when the maximum differences in parameters between treatments were observed. Moreover, many of the parameters began to decrease after 44 h (including in the control treatments), suggesting that the changes observed were more likely due to "bottle effects" than to the changes in response to nutrient or microbial additions.

Changes in Phytoplankton and Bacterial Abundance following Dust Additions
Chl-a was measured as a proxy for total phytoplankton biomass ( Figure 2A). For all treatments Chl-a concentrations increased slightly 20 h after amendments, and then steadily decreased throughout the experiment (Figure 2A). Overall, Chl-a concentrations were significantly higher in treatments amended with live dust and UV-treated dust as compared with the control 20-44 h after amendment (Table 1, Figure 2A). However, there were no significant differences between Chl-a concentrations of the live dust and UV-treated dust treatments. Table 1. The net response triggered by airborne microbes in the northern Red Sea (NRS) seawater 24-48 h post addition. Values were calculated as the difference between the "live dust" and "UV-treated dust". Statistically significant differences are highlighted in bold (p < 0.05).

Variable T20 h T48 h
Chl-a (µg L −1 ) −0.13 −0.06 Prochlorococcus (Cells mL −1 ) −2140 −1920 Synechococcus (Cells mL −1 ) −7300 −9100 Picoeukaryotes (Cells mL −1 ) −120 −10 Heterotrophic bacteria (Cells mL −1 ) −82,000 −3000 Prochlorococcus ( Figure 2B), Synechococcus ( Figure 2C) and picoeukaryotes ( Figure 2D) dominate autotrophic communities in the NRS during the summer time [41]. All three autotrophs' abundances were different in the dust amendments as compared with the control treatments ( Figure 2B-D). Prochlorococcus abundances decreased dramatically immediately after amendment in both dust treatments, while in the control we saw a decrease at only 20 h after amendment ( Figure 2B). There was a slight increase in all treatments at 44 h, followed by another decrease. Prochlorococcus abundances were significantly lower in the live dust treatments (mean 3.7 × 10 3 cells/mL) and UV-treated dust (mean 5.8 × 10 3 cells/mL) than in the control treatments (mean 9.0 × 10 3 cells/mL) at 20 h post addition ( Figure 2B, Supplementary Table S3). Prochlorococcus abundance in the dust treatments remained significantly lower than the control throughout the remainder of the experiment ( Figure 2B), with no significant differences between live dust and UV-treated dust at both 20 h and 44 h after amendment. Synechococcus abundances initially decreased in all treatments and increased again at 20 h ( Figure 2C). Synechococcus abundances in the dust treatments decreased again at 44 h after addition and were significantly different between the control (mean 6.62 × 10 4 cells/mL), live dust (mean 4.73 × 10 4 cells/mL), and UV-treated dust (mean 5.64 × 10 4 cells/mL), treatments ( Figure 2C, Supplementary Table S3). Synechococcus abundance was significantly higher in the control than in the live dust (28%) and UV-treated dust (14%) (p-value < 0.05) treatments, and live dust treatments had significantly lower Synechococcus abundance (−9.0 × 10 3 cells/mL) than the UV-treated dust treatments ( Table 1, Supplementary Table S4). Picoeukaryote abundance was significantly increased in the live dust at 20 h and at 44 h (mean 4.4 × 10 3 cells/mL, mean 5.2 × 10 3 cells/mL, respectively) and UV-treated dust (mean 4.6 × 10 3 cells/mL, 5.2 × 10 3 cells/mL, respectively) treatments than in the control (mean 1.9 × 10 3 cells/mL, mean 1.9 × 10 3 cells/mL, respectively) ( Figure 2D, Supplementary Table S3), after which abundances declined ( Figure 2D) paralleling the dynamic of total Chl-a ( Figure 2A). There was no significant difference in picoeukaryote abundance between the live dust and UV-treated dust (Supplementary Table S4).
Atmosphere 2019, 10, x FOR PEER REVIEW 6 of 21 treatments, while in the control we saw a decrease at only 20 h after amendment ( Figure 2B). There was a slight increase in all treatments at 44 h, followed by another decrease. Prochlorococcus abundances were significantly lower in the live dust treatments (mean 3.7 × 10 3 cells/mL) and UVtreated dust (mean 5.8 × 10 3 cells/mL) than in the control treatments (mean 9.0 × 10 3 cells/mL) at 20 h post addition ( Figure 2B, Supplementary Table S3). Prochlorococcus abundance in the dust treatments remained significantly lower than the control throughout the remainder of the experiment ( Figure  2B), with no significant differences between live dust and UV-treated dust at both 20 h and 44 h after amendment. Synechococcus abundances initially decreased in all treatments and increased again at 20 h ( Figure 2C). Synechococcus abundances in the dust treatments decreased again at 44 h after addition and were significantly different between the control (mean 6.62 × 10 4 cells/mL), live dust (mean 4.73 × 10 4 cells/mL), and UV-treated dust (mean 5.64 × 10 4 cells/mL), treatments ( Figure 2C, Supplementary Table S3). Synechococcus abundance was significantly higher in the control than in the live dust (28%) and UV-treated dust (14%) (p-value < 0.05) treatments, and live dust treatments had significantly lower Synechococcus abundance (−9.0 × 10 3 cells/mL) than the UV-treated dust treatments ( Table 1, Supplementary Table S4). Picoeukaryote abundance was significantly increased in the live dust at 20 h and at 44 h (mean 4.4 × 10 3 cells/mL, mean 5.2 × 10 3 cells/mL, respectively) and UV-treated dust (mean 4.6 × 10 3 cells/mL, 5.2 × 10 3 cells/mL, respectively) treatments than in the control (mean 1.9 × 10 3 cells/mL, mean 1.9 × 10 3 cells/mL, respectively) ( Figure 2D, Supplementary Table S3), after which abundances declined ( Figure 2D) paralleling the dynamic of total Chl-a ( Figure 2A). There was no significant difference in picoeukaryote abundance between the live dust and UV-treated dust (Supplementary Table S4). Heterotrophic bacterial abundance (BA) 20 h post-amendment in both dust treatments increased up to ~150% relative to control levels, after which BA remained constant ( Figure 2E). The BA in the control remained relatively stable throughout the experiment ( Figure 2E). There was a significant difference (p-value < 0.05) in heterotrophic bacterial abundance between the control (mean 3.68 × 10 5 cell/mL) and the live dust (mean 9.09 × 10 5 cell/mL) and UV-treated dust (mean 9.91 × 10 5 cell/mL) treatments at 20 h, and between the control (mean 3.96 × 10 5 cell/mL) and the live dust (mean 9.13 × 10 5 cell/mL), and UV-treated dust (mean 9.17 × 10 5 cell/mL) treatments at 44 h ( Figure 2E, Supplementary Tables S3 and S4). There were no significant differences in BA between the live dust and UV-treated dust (Supplemental Table S4). Heterotrophic bacterial abundance (BA) 20 h post-amendment in both dust treatments increased up to~150% relative to control levels, after which BA remained constant ( Figure 2E). The BA in the control remained relatively stable throughout the experiment ( Figure 2E). There was a significant difference (p-value < 0.05) in heterotrophic bacterial abundance between the control (mean 3.68 × 10 5 cell/mL) and the live dust (mean 9.09 × 10 5 cell/mL) and UV-treated dust (mean 9.91 × 10 5 cell/mL) treatments at 20 h, and between the control (mean 3.96 × 10 5 cell/mL) and the live dust (mean 9.13 × 10 5 cell/mL), and UV-treated dust (mean 9.17 × 10 5 cell/mL) treatments at 44 h ( Figure 2E, Supplementary Tables  S3 and S4). There were no significant differences in BA between the live dust and UV-treated dust (Supplemental Table S4).

Changes in Autotrophic and Heterotrophic Production following Dust Additions
Rates of primary production (PP) in all the treatments were relatively constant throughout the experiment with small differences between dust and control treatments, aside from 44 h after amendment ( Figure 3A Table S3). At 44 h, PP rates were significantly lower in the live dust treatments than in UV-treated dust treatments, with a net difference of 0.52 µg C L −1 h −1 ( Table 1,  Supplementary Table S4).

Changes in Autotrophic and Heterotrophic Production following Dust Additions
Rates of primary production (PP) in all the treatments were relatively constant throughout the experiment with small differences between dust and control treatments, aside from 44 h after amendment ( Figure 3A   Rates of heterotrophic bacterial production (BP) increased in the live dust (mean 2.70 µg C L −1 h −1 ) and UV-treated dust (mean 3.02 µg C L −1 h −1 ), treatments 20 h after amendment, and were significantly higher than rates of BP in the control (mean 1.56 µg C L −1 h −1 ) ( Figure 3B, Supplementary  Table S3), respectively. At 44 h after amendment, BP rates decreased in the UV-treated dust treatments (mean 1.53 µg C L −1 h −1 ) drastically while the live dust (mean 2.36 µg C L −1 h −1 ) remained relatively stable ( Figure 3B). Live dust treatments with airborne microbes had significantly higher rates of BP than both control and UV-treated dust treatments (Supplementary Table S4). The net BP rates of airborne microbes (difference between average BP rates in live and UV-treated treatments) was 0.83 µg C L −1 h −1 (Table 1).
Extracellular enzymatic activity rates were used as additional measures of the activity of different groups of organisms. Beta-glucosidase (β-Glu) ( Figure 3C) and leu-aminopeptidase (AMA) Rates of heterotrophic bacterial production (BP) increased in the live dust (mean 2.70 µg C L −1 h −1 ) and UV-treated dust (mean 3.02 µg C L −1 h −1 ), treatments 20 h after amendment, and were significantly higher than rates of BP in the control (mean 1.56 µg C L −1 h −1 ) ( Figure 3B, Supplementary Table S3), respectively. At 44 h after amendment, BP rates decreased in the UV-treated dust treatments (mean 1.53 µg C L −1 h −1 ) drastically while the live dust (mean 2.36 µg C L −1 h −1 ) remained relatively stable ( Figure 3B). Live dust treatments with airborne microbes had significantly higher rates of BP than both control and UV-treated dust treatments (Supplementary Table S4). The net BP rates of airborne  (Table 1). Extracellular enzymatic activity rates were used as additional measures of the activity of different groups of organisms. Beta-glucosidase (β-Glu) ( Figure 3C) and leu-aminopeptidase (AMA) ( Figure 3D) activities were used to measure the extracellular enzymatic activity of heterotrophic prokaryotes, whereas alkaline phosphatase activity (APA) ( Figure 3E) rates were used to measure extracellular enzymatic activity of algae.
While  Figure 3C). The ß-Gl activity rates were significantly higher in live dust treatments than UV-treated dust treatments (Supplementary Table S4), with airborne microbes contributing up to 1.24 nM L −1 h −1 ( Table 1). The AMA rates in the live dust'(mean 8.08 nM L −1 h −1 ) and UV-treated dust (mean 7.59 nM L −1 h −1 ) treatments increased 20 h after amendment, whereas AMA rates in the control decreased (mean 5.09 nM L −1 h −1 ) ( Figure 3D, Supplementary Table S4). At both timepoints, both UV-treated dust and live dust treatments had higher rates of AMA (64% and 45%, respectively at 44 h) than the control (p-value < 0.05), but no significant differences between the live and UV-treated treatments were observed ( Figure 3D, Supplementary Table S4).
The APA rates decreased for both live dust (mean 1.3 nM L −1 h −1 ) and UV-treated dust (mean 1.097 nM L −1 h −1 ) treatments while rates in the control (mean 3.45 nM L −1 h −1 ) increased 20 h after amendment ( Figure 3E). There were no significant differences in the APA rates between the live and UV-treated dust throughout the experiment (Supplementary Table S4).

Dust-Associated Prokaryotes and Eukaryotes
The prokaryotic community in the dust sample collected and used in this experiment was comprised of bacteria from the Firmicutes (15%), Gemmatimonadetes (15%), Actinobacteria (13%) and Bacteroidetes (12%) phyla ( Figure 4A). A large portion of prokaryotes (30%) were bacteria phyla that individually made up less than 1% of the relative abundance ( Figure 4A). The eukaryotic community in the aerosols was dominated by Dikarya (55%), a subkingdom of Fungi containing the Ascomycota and Basidiomycota phyla, and Phragmoplastophyta (29%), a subclade of Charophyta ( Figure 4B). All the organisms in the Phragmoplastophyta subclade were land plants (Supplementary Table S5). Of the eukaryotes in the dust, at least 8% were marine organisms (algae, protists) (Supplementary Table S5).

Changes in Prokaryotic and Eukaryotic Diversity following Dust Addition
The prokaryotic community in the NRS surface water was dominated by Proteobacteria constituting ~60% of taxa before the experiment, and ~70% to 75% during the experiment ( Figure 5A). Alphaproteobacteria and Gammaproteobacteria were also abundant classes, making up 40-48% and 17-25% of the community, respectively ( Figure 5A). Shannon's diversity index (H) and Faith's phylogenetic diversity (FPD) index were used to characterize the diversity of prokaryotes (Figure 6). At both timepoints (20 h, 44 h), the diversity (H and FPD) was significantly different between the control and the dust treatments ( Figure 6, Supplementary Figure S1

Changes in Prokaryotic and Eukaryotic Diversity following Dust Addition
The prokaryotic community in the NRS surface water was dominated by Proteobacteria constituting~60% of taxa before the experiment, and~70% to 75% during the experiment ( Figure 5A). Alphaproteobacteria and Gammaproteobacteria were also abundant classes, making up 40-48% and 17-25% of the community, respectively ( Figure 5A). Shannon's diversity index (H) and Faith's phylogenetic diversity (FPD) index were used to characterize the diversity of prokaryotes (Figure 6). At both timepoints (20 h, 44 h), the diversity (H and FPD) was significantly different between the control and the dust treatments ( Figure 6, Supplementary Figure S1   The eukaryotic community of the NRS was dominated by Alveolata (Dinoflagellata and Protalveolata) before the experiment (33%) and by Stramenopiles (Marine Stramenopiles and Ochrophyta) (33%) during the experiment ( Figure 5B). The diversity of eukaryotes was also measured using Shannon's diversity (H) and Faith's phylogenetic diversity (FPD) indices ( Figure 6). The control (mean H = 7.3, mean FPD = 38.2) had higher diversity than that in the live dust (mean H = 6.7) and UV-treated dust (mean H = 6.6 and mean FPD = 27.3) treatments 20 h after amendment (Kruskal-Wallis test: H = 3.857, df = 2, p < 0.05), with no differences between the live dust and UV-treated dust (Figure 6, Supplementary Figure S1). At 44 h after amendment, the live dust treatments (mean H = 6.9, mean FPD = 29.7) had lower diversity than both the UV-treated dust (mean H = 7.3, mean FPD = 37.8) and the control (mean H = 7.3, mean FPD = 34.6) treatments ( Figure 6). The beta diversity (Bray-Curtis) showed no differences between the live dust and UV-treated dust treatments at 20 h and 44 h after amendment.

Impact of Airbone Microbes (or Dust Deposition) on Phytoplankton Biomass and Primary Production
The NRS receives high amounts of dust deposition every year, reaching dust loads of~30 µg m −3 in normal non-storm conditions and~700 µg m −3 during dust events [16][17][18]. Dust particles shelter and serve as a temporary habitat for microorganisms, protecting them from direct exposure to UV radiation [42,43]. Microorganisms survive long-range transport, and it has been reported that up to 25% of airborne microorganisms remain viable upon deposition [11,44] and subsequently thrive in novel marine ecosystems and impact native microbial populations and processes [13,14]. Low-nutrient, low-chlorophyll (LNLC) marine environments are strongly impacted by dust addition during high deposition events, because atmosphere is a crucial source of limiting nutrients and trace metals [19]. However, the potential effects of airborne microorganisms delivered by atmospheric deposition in these areas has been often overlooked. Specifically, several studies in LNLC systems have shown an increase in primary production rates following aerosol addition and have solely attributed this outcome to the delivery of limiting nutrients (P, N, Fe) for photosynthesis [6,13,[45][46][47][48][49][50][51][52][53]. Picophytoplankton, including cyanobacteria (Prochlorococcus and Synechococcus) and picoeukaryotes, account for a large portion of algal biomass and primary productivity in LNLC systems [41,[54][55][56][57]. Two factors that drive picophytoplankton abundance in LNLC regions are (1) nutrient availability [58,59], and (2) population decline through viral infections and lysis [60]. Previous studies have shown that atmospheric deposition affects picophytoplankton variably, both increasing abundance due to a supply of nutrients [45][46][47] and decreasing abundance due to a supply of toxic metals [20]. Comparing the effect of live dust potentially carrying live microorganisms and viruses ( Figure 4) and UV-treated dust where microorganisms (and other biotic entities as viruses) were inactivated, we provide evidence that airborne microorganisms also have an impact on surface ecosystem.
Both live dust and UV-treated dust amendments led to an initial (0 to 44 h) increase of phytoplankton biomass as indicated by Chl-a (Figure 2A), mostly explained by the increase in picoeukaryotes ( Figure 2D). This response likely resulted from the released nutrients from the dust particles that boosted cell growth (Supplementary Table S2). In contrast, cyanobacteria declined or showed small variations as compared with the control samples ( Figure 2B,C). We note however that Chl-a and PP at 44 h were lower in the live dust samples, suggesting that living constituents present in the live dust negatively impacted phytoplankton growth and primary productivity rates (Figures 2A and 3A).
This effect could have resulted from the (1) introduction of competitors for the same food resources, for example heterotrophic bacteria [61,62] and/or (2) introduction of predatory and/or pathogenic entities, for example viruses [12,63]. Indeed, the dust used in the experiment contained bacteria as well as other marine organisms (Figure 4, Supplementary Table S5), which could compete with ambient phytoplankton for resources and may have served as vectors for viruses.
Viruses are abundant in desert soils, reaching concentrations of 2.2 × 10 3 to 1.1 × 10 7 virus-like particles per gram [64]. Upon aerosolization, viruses travel long distances via dust storms. The number of airborne viruses has been reported to increase by an order of magnitude during dust storms as compared with normal atmospheric conditions [65]. Dust storms also pick up marine viruses, which become aerosolized through wind-induced bubble bursting on ocean surface [12,63,66] while traveling over oceans. Some airborne viruses have been shown to infect phytoplankton in numbers large enough to terminate entire blooms [12]. Furthermore, there are viruses known to infect cyanobacteria that are very host-specific [67], potentially explaining why the three autotrophic populations which were measured here (picoeukaryotes, Prochlorococcus and Synechococcus) responded differently during the experiment.

Impact of Airborne Microbes (or Dust Deposition) on Heterotrophic Prokaryotes Production
Atmospheric dust deposition has also been shown to increase heterotrophic prokaryotes production rates in LNLC systems [5,14,46,48,68]. While these previous studies suggested that the input of nutrients from aerosols has increased in-situ surface water BP (i.e., that of resident heterotrophic bacteria in the marine surface layer), the differences in BP between the live dust and UV-treated dust treatments show that viable airborne microbes also rapidly contribute to increasing marine BP rates (by 50%, Figure 3B). Similar results, showing contributions of airborne microbes to BP, as well as to N 2 fixation, have been previously reported for southeastern Mediterranean waters [13,14]. The BP rates were previously tested in microcosm bioassay experiments where aerosols collected during a dust storm were added to sterile southeastern Mediterranean water and showed that BP increased by fourfold [13], corresponding to 20-50% of the typical BP rates measured in the open and coastal southeastern Mediterranean, respectively [5,69]. Our study furthers their findings by showing that BP rates increase with the addition of airborne microbes in non-sterile water with resident microbial community from the Gulf of Aqaba in the NRS.
Our conclusion is also supported by the enzyme activity data ( Figure 3C-E). Extracellular enzymes are synthesized by microorganisms to hydrolyze polymeric substances into bioavailable monomers [70]. Measuring rates of extracellular enzyme activity therefore provide insight into productivity of marine microbes. We show that AMA and β-Gl activities increased significantly with dust addition, while APA rates decreased significantly ( Figure 3A-E). Synthesis of extracellular enzymes are dependent on nutrient availability (APA dependent on P, AMA and β-Gl dependent on organic carbon) and their activity rates change with the input of nutrients via atmospheric deposition [71,72]. The β-Gl activity is attributed mostly to heterotrophic bacteria [30,73], while AMA activity is attributed to heterotrophic bacteria as well as cyanobacteria, phytoplankton, and zooplankton [30,[73][74][75] and its activities have been shown to increase after aerosol addition in a microcosm study in the Mediterranean Sea [72]. Specifically, β-Gl is utilized in hydrolysis of cellobiose found in polymers, such as cellulose and mucopolysaccharides, and can be related to Chl-a [76,77], whereas AMA is utilized in the decay of particulate matter composed of biotic and abiotic material [74,75]. APA, synthesized by phytoplankton [73], has also been shown to vary in response to dust additions to the Mediterranean [71,72] and northern Red Sea [78] seawater during incubation experiments. Changes in enzymatic activity were observed in both the UV-treated dust and live dust treatments as compared with the control, showing that chemical components of atmospheric deposition are enhanced in-situ microbial activities, as previously suggested. However, we also measured differences in β-Gl activity rates between the UV-treated dust and live dust treatments, indicating that viable airborne microorganisms specifically contribute to increasing β-Gl activity rates (20-25%, Figure 3C). A significant difference between UV-treated dust and live dust treatments was not observed in the AMA activity rates ( Figure 3D), which is also synthesized by heterotrophic bacteria. These results suggest that deposited airborne microbes preferentially synthesized β-Gl for hydrolysis of carbohydrates, and since the chemical constituents were the same in both treatments, the preferential use of this enzyme in the live dust treatment likely corresponds to synthesis by the microbial assemblage of the dust ( Figure 4A).
Interestingly, although we saw differences in the BP between treatments, the heterotrophic bacteria abundance significantly increased with the addition of dust in both treatments relative to the control but was not different between the live dust and UV-treated dust treatments ( Figure 2B). This difference in cell specific activity (bacterial production per bacterial abundance) was previously observed in a mesocosm study in the southeastern MS [13]. We attribute this increase solely to the chemical constituents of aerosols, which provide nutrients and organic C to increase heterotrophic bacteria abundance, in both treatments regardless of their origin [5,14,46,48,68]. Alternatively, there may be a higher removal of heterotrophic bacteria in the live dust treatments by grazing or viral lysis, preventing an increase in abundance.

Microbial Population in the Dust and their Impact on Biodiversity
Most of the bacteria found in the dust (>1%) were not found in the seawater from the mesocosm experiments (>1%) at the family level, aside from organisms belonging to the family Rhodobacteraceae ( Figures 4A and 5A). However, there was a large number of eukaryotes (Alveolates and Stramenopiles) present in the dust, as well as in all the seawater samples ( Figures 4B and 5B). These organisms made up 8% of the dust ( Figure 4B, Supplementary Table S5), and up to 52% of the seawater samples. These marine organisms likely become aerosolized from ocean surfaces through bubble bursting and sea spray [12,63,66].
We found similar taxonomic relative abundances in all treatments at 44 h after amendment ( Figure 5), however this visual representation is not sensitive enough to detect differences in diversity. Therefore, to quantitatively assess community differences we calculated diversity of both prokaryotes and eukaryotes of treatments (live dust, UV-treated dust, and control) ( Figure 6, Supplementary Figure S1) using alpha diversity metrics. Alpha diversity metrics are used quantitatively (Shannon's diversity index), qualitatively (Faith's phylogenetic diversity), and phylogenetically (Faith's phylogenetic diversity) to show how many unique taxa are present in a sample. Our experiment showed that eukaryote diversity (both measures) was different between live dust and UV-treated dust treatments at 44 h, with significantly lower diversity when airborne microbes were present ( Figure 6). The lower diversity in the live dust treatments indicate that the number of unique eukaryotes in the NRS surface water decreased with the addition of airborne microbes. The difference in diversity may be a result of competitive relationships between airborne microbes and eukaryotes in the NRS surface water. Alternatively, airborne viruses or fungi in our sample may have specifically infected certain marine eukaryote groups. However, the diversity of prokaryotes did not change between treatments, indicating that there was no antagonistic relationship between the bioaerosols and bacterial communities of the NRS. These processes need to be further studied.

Conclusions
Our results show that microbial diversity is altered by bioaerosols and that while the rates of primary productivity decline, the rates of bacterial production increase in response to the deposition of bioaerosols. Since our experiment lasted 72 h, with most changes occurring in the first 48 h, the observed effects of bioaerosols might be transient. Although our study was conducted in mesocosms and lasted only~72 h, surface water with local microbial assemblages was used and the amount of aerosol added was representative of conditions during dust storm events. The results are consistent with previous reports indicating that atmospheric deposition is a crucial source of nutrients and trace metals in LNLC systems [2,3,79,80]. However, we also clearly show unique effects of airborne microorganisms impacting biogeochemical processes indicating that the impact of bioaerosols is also relevant, particularly in LNLC regions where dust storm events occur. Importantly, since dust deposition is predicted to increase with climate change, the impact of airborne microorganisms found in our study may also increase. Additional studies that improve our understanding of how these geographically vast areas of ocean will be impacted by bioaerosols will be of great value.  Table S1. Chemical and biological properties of the NRS water used in the experiment (before amendments); Table S2. Nutrients and trace metals concentrations added from the aerosols to each mesocosm; Table S3. ANOVA test results between control, "UV-treated" and "live-dust' treatments at 20 h or 44 h; Table S4. Tukey post-hoc test results between "UV-treated" and control, "live-dust" and control and "live-dust" and "UV-treated' treatments at 20 h or 44 h; Table S5. Eukaryotes, relative abundances (abundance), and general information on taxa (details).