Phytoplankton and Bacterial Response to Desert Dust Deposition in the Coastal Waters of the Southeastern Mediterranean Sea : A Four-Year In Situ Survey

Atmospheric dust/aerosol deposition is an important source of external nutrients for the surface of the ocean. This study shows high-resolution observational data gathered in situ over a period of four years on bacterial and phytoplankton abundance and activity during typical background atmospheric conditions and during intense dust storm events in the low-nutrient, low-chlorophyll (LNLC) coastal waters of the southeastern Mediterranean Sea (SEMS). Chlorophyll a (an estimate for phytoplankton biomass) and bacterial abundance show moderate changes in response to dust deposition/events (−10% and +20%, respectively), while primary production, bacterial production, and N2 fixation rates were all significantly and positively affected by deposition (+25 to +40%; p < 0.05). The rapid changes in bacterial and/or phytoplankton rate parameters suggest that the released micro-/macronutrients from atmospheric deposition are tunneled directly in metabolic processes and, to a lesser extent, for biomass accumulation. The predicted expansion of LNLC areas in oceans in the future, and the projected increase in dust emission due to desertification, may affect the production of marine microbial communities in the surface of the ocean, yet only moderately affect their biomass or standing stock. Such alterations may impact carbon sequestration to the deep ocean.

The coastal southeastern Mediterranean Sea (SEMS) is an LNLC oligotrophic environment [20][21][22][23].The water is dominated by small-size microbial populations (e.g., heterotrophic bacteria, cyanobacteria) throughout the year [22,24,25], while diatom blooms are very rare [26,27].Primary and bacterial production (PP and BP, respectively) as well as N 2 fixation rates are usually low in the SEMS [22,27].Microcosm simulations and local monitoring activities show that external nutrient inputs to SEMS coastal waters can affect bacterial/phytoplankton biomass and/or activity [23,28] and may even trigger a food-web cascade, leading to blooms of a rare cyanobacterial phylotype [27].These data highlight the susceptibility of the ultra-oligotrophic SEMS coastal waters to external nutrient inputs, including those derived from dust/aerosol deposition.
Being enclosed by land/deserts, the SEMS coastal water receives high amounts of dust/aerosols annually from different origins [2,[29][30][31].Indeed, it has been demonstrated that atmospheric deposition is one of the most important external nutrient sources for the SEMS (i.e., NO 3 + NO 2 , PO 4 ) [31][32][33][34].Recent microcosm bioassay studies show, for example, that dust added to the surface microlayer (upper 60 µm layer) resulted in up to fivefold and twofold increases in BP and PP, respectively [18,35].However, such microcosm studies represent conditions during distinct and limited time points and may have some experimental limitations (i.e., bottle effect) [36].Therefore, in situ measurements that capture natural variability in the net effect of external nutrient inputs and associated changes to the microbial community better represent natural conditions.Here, we compare in situ temporal dynamics of phytoplankton and bacteria during "typical" days (background value of Al in aerosols <2000 ng m −3 ) and dust deposition events (high Al concentration in aerosols, >2000 ng m −3 ) in the coastal SEMS between April 2013 and October 2017.The data are used to assess changes in the biomass of autotrophic and heterotrophic microbial communities and their productivity and N 2 fixation rates attributed to aerosol deposition events.

Material and Methods
Surface water samples (1-2 m) were collected every 2-4 weeks between May 2015 and October 2017 at a monitoring station off the SEMS coast (32.28 • N, 34.95 • E).We used additional data collected at the same site between April 2013 and April 2015 [22] to complement this record.Samples were collected using the same protocols and procedures as described in [22], resulting in a relatively long (>4 years) record of phytoplankton and bacterial abundance and production rates in the coastal SEMS.In addition, during dust events, samples were taken more frequently, specifically during and 1-3 days after the events.Water samples were collected in transparent acid-clean (HCl, 10%) and bottle-rinsed 4.6 or 10 L Nalgene carboys and brought to the lab immediately for processing for the different analyses described below.Aerosol samples were usually collected weekly (integrated over 68 h) using a high-volume total suspended particles (TSP) sampler (flow rate of 42 m 3 h −1 ) on Whatman 41 filters (high-quality cotton linters; GE Healthcare).TSP samples were also collected during dust storms over <24 h intervals, defined here by samples in which Al concentration reached >2000 ng m −3 .This classification is based on previous studies from the Eastern Mediterranean Sea [2,37,38] that show: (1) significantly higher concentrations of Al were associated with air masses originating from desert sources/regions, and (2) the background concentrations of particulate Al in air show maximal values of 1670 ng m −3 (median of 717 ng m −3 ).Bulk Al concentrations in the TSP (collected during dust storms and typical non-storm days, n = 126) were measured after total digestion with hydrogen fluoride (HF) following the procedure of ASTM (1983) as described in [13].The concentrations were measured on a Perkin-Elmer 1100B atomic absorption spectrometer [2].We identified 4 main aerosol sources during the study period based on 3-day back trajectory analyses, arriving at 100, 500, and 1000 m altitude levels, commencing at 10:00 UTC using the NOAA HYSPLIT model (https://ready.arl.noaa.gov/HYSPLIT_traj.php)(Figure 1).Chlorophyll a (Chl.a):Triplicate seawater samples (300 mL) were filtered through glass fiber filters (GF/F, pore size 0.7 µm, Whatman, Buckinghamshire, UK) and placed overnight in dark glass vials with 90% acetone solution.Chl.a fluorescence was read with a Turner Designs fluorometer (ex.436 nm, em.680 nm, San Jose, CA, USA) [39].The instrument was calibrated using a Chl.a standard (Sigma, Darmstadt, Germany, CAS Number 479-61-8).
Heterotrophic bacterial abundance (BA): Water samples (1 mL) were fixed with pure glutaraldehyde solution (50% initial concentration; Sigma-Aldrich G7651), frozen in liquid nitrogen, and placed at −80 °C.Before analysis, the samples were thawed in a 37 °C water bath and stained with SYTO9 (1:10 5 vol:vol) for 10 min in the dark.Subsamples (100 µL) were run at 25 µL min −1 using an Attune® Acoustic Focusing Flow Cytometer (Applied Biosystems, Life Technologies, Carlsbad, CA, USA) equipped with 488 and 405 nm lasers.Taxonomic discrimination was based on green fluorescence, side scatter, and forward scatter [40].Cyanobacterial autofluorescence was removed from the total bacterial counts.Beads of 1 µm size were used as reference for bacterial cells.
Primary production (PP): Triplicate seawater samples (50 mL) were placed in transparent polycarbonate Nalgene bottles and amended with a solution of NaH 14 CO3 (1 m Ci mL −1 ; Perkin Elmer, Boston, MA, USA).Additional samples were placed in dark Nalgene bottles as controls.Bottles were incubated for 24 h under ambient natural illumination and temperature.Added activity Chlorophyll a (Chl.a):Triplicate seawater samples (300 mL) were filtered through glass fiber filters (GF/F, pore size 0.7 µm, Whatman, Buckinghamshire, UK) and placed overnight in dark glass vials with 90% acetone solution.Chl.a fluorescence was read with a Turner Designs fluorometer (ex.436 nm, em.680 nm, San Jose, CA, USA) [39].The instrument was calibrated using a Chl.a standard (Sigma, Darmstadt, Germany, CAS Number 479-61-8).
Heterotrophic bacterial abundance (BA): Water samples (1 mL) were fixed with pure glutaraldehyde solution (50% initial concentration; Sigma-Aldrich G7651), frozen in liquid nitrogen, and placed at −80 • C. Before analysis, the samples were thawed in a 37 • C water bath and stained with SYTO9 (1:10 5 vol:vol) for 10 min in the dark.Subsamples (100 µL) were run at 25 µL min −1 using an Attune®Acoustic Focusing Flow Cytometer (Applied Biosystems, Life Technologies, Carlsbad, CA, USA) equipped with 488 and 405 nm lasers.Taxonomic discrimination was based on green fluorescence, side scatter, and forward scatter [40].Cyanobacterial autofluorescence was removed from the total bacterial counts.Beads of 1 µm size were used as reference for bacterial cells.
Primary production (PP): Triplicate seawater samples (50 mL) were placed in transparent polycarbonate Nalgene bottles and amended with a solution of NaH 14 CO 3 (1 m Ci mL −1 ; Perkin Elmer, Boston, MA, USA).Additional samples were placed in dark Nalgene bottles as controls.Bottles were incubated for 24 h under ambient natural illumination and temperature.Added activity measurements were carried out immediately after spiking.The incubations were terminated by filtering the spiked seawater through GF/F filters (pore size 0.7 µm, Whatman, Buckinghamshire, UK) at low pressure.Hydrochloride acid (32%, 50 µL) was then added to remove excess 14 C-bicarbonate (fuming lasted ~8 h).Scintillation cocktail (Ultima-Gold) was added to each vial (4 mL) and the radioactivity of each sample was measured using a TRI-CARB 2100 TR (Packard, Meriden, CT, USA) liquid scintillation counter [41].
Heterotrophic bacterial production (BP): Triplicate seawater samples (1.7 mL) were spiked with 3 H-leucine (123 Ci mmol −1 ; Perkin Elmer, Boston, MA, USA) and incubated for 4-9 h under dark conditions at ambient temperature.Incubation was terminated by adding 100% trichloroacetic acid, followed by microcentrifugation [42].Killed samples (trichloroacetic acid was added at the beginning of the incubation) were also carried out in triplicate.After adding 1 mL of scintillation cocktail (Ultima-Gold) to each vial, the samples were counted using a TRI-CARB 2100 TR (Packard, Meriden, CT, USA) liquid scintillation counter.A conversion factor of 3 kg C per mole of leucine incorporated and an isotopic dilution of 2.0 were used to calculate the C incorporated [43].
Triplicate seawater samples were collected in 4.6 L transparent Nalgene bottles and 225 mL of 15 N-enriched seawater was added (5% of the incubation volume) without headspace.The samples were incubated for 24 h under ambient light and temperature.Natural isotope abundance measurements were also carried out.Incubations were terminated by filtering the seawater onto precombusted (450 • C, 4.5 h) GF/F (nominal pore size 0.7 µm) and dried overnight in an oven at 60 • C. The samples were analyzed on a NC2500 elemental analyzer (CE Instruments, Manchester, UK) interfaced to a Finnigan Delta Plus XP isotope ratio mass spectrometer (IRMS, Thermo Fisher Scientific, MT, USA).A standard curve was generated to determine N mass for each sample run.Detection limit for 15 N uptake was 0.02 nmol N L −1 d −1 .
Statistical analysis: A comparison between the values of Chl.a, BA, BP, PP, and N 2 fixation during "dusty" days vs. "typical background" days was performed by student t-test.For regression analysis, the data were first log-transformed.All statistical analyses were performed using XLSTAT 2016 (Microsoft, New-York, USA) with a confidence level of 95%.
Atmosphere 2018, 9, x FOR PEER REVIEW 5 of 15 (Figure 2).Heterotrophic BA ranged from 4.2 × 10 5 to 1.6 × 10 6 cells mL −1 (n = 89) and BP rates were 0.2-1.8µg C L −1 d −1 (n = 89) (Figure 3).Rates of N2 fixation were 0.1-0.5 nmol N L −1 d −1 (n = 40) (Figure 4).These values are in the range of previously published data from the SEMS [22,23,62], as well as from other oligotrophic marine environments [17,[63][64][65].Measurements were taken during "typical" days (white, Supplementary Table S1, n = 89) and dust storms (orange, Supplementary Table S2, n = 29).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.Measurements were taken during "typical" days (white, Supplementary Table S1, n = 89) and dust storms (orange, Supplementary Table S2, n = 29).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.S1, n = 89) and during dust storms (orange, Supplementary Table S2, n = 29).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the box represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.Measurements were taken during "typical" days (white, Supplementary Table S1, n = 89) and during dust storms (orange, Supplementary Table S2, n = 29).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the box represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.S1, n = 40) and during dust storms (orange, Supplementary Table S2, n = 18).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.
Here we show that dust storm events caused larger changes in metabolic rates (PP, BP, N2 fixation) than in abundance of phytoplankton or heterotrophic bacteria (Figures 2-4, Supplementary Tables S1 and S2).Thus, during dust events, in situ Chl.a levels did not differ significantly from the background values (−6% change, p = 0.444) and bacterial abundance increased only moderately (+19%, p = 0.01), while PP (+28%, p = 0.006), BP (+25%, p = 0.048), and N2 fixation (+35%, p = 0.002) were all significantly higher (Figure 5).We therefore suggest that metabolic rate measurements (i.e., PP, BP, and N2 fixation) are more "sensitive" to atmospheric deposition events and related environmental changes in surface water than phytoplankton/bacterial standing stocks.Indeed, several previous studies based on bioassays reported that standing stocks of phytoplankton and bacterial abundance showed much smaller change to dust addition than activity measurements such as PP and BP (e.g., [13]).Similarly, Astrahan et al. [18] reported that dust added to the surface microlayer resulted in a 4-to 7-fold increase in BP, whereas bacterial abundance was only enhanced by 1.5-to 2-fold.We argue that these differences are attributed to the usually low growth rate of marine phytoplankton and heterotrophic bacteria (usually >1 d −1 , [66,67]), suggesting that measurements made <24 h post deposition may not be sufficient to detect significant changes in biomass or diversity.On the contrary, alterations in rate measurements are usually much faster and therefore can be detected on short time scales.For example, Guo et al. [68] showed that PP and BP were enhanced within 6-24 h post aerosol addition, whereas the prokaryotic community composition (determined by 16S rRNA gene diversity analysis) remained unchanged in that short time scale.S1, n = 40) and during dust storms (orange, Supplementary Table S2, n = 18).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.
Here we show that dust storm events caused larger changes in metabolic rates (PP, BP, N 2 fixation) than in abundance of phytoplankton or heterotrophic bacteria (Figures 2-4, Supplementary Tables S1 and S2).Thus, during dust events, in situ Chl.a levels did not differ significantly from the background values (−6% change, p = 0.444) and bacterial abundance increased only moderately (+19%, p = 0.01), while PP (+28%, p = 0.006), BP (+25%, p = 0.048), and N 2 fixation (+35%, p = 0.002) were all significantly higher (Figure 5).We therefore suggest that metabolic rate measurements (i.e., PP, BP, and N 2 fixation) are more "sensitive" to atmospheric deposition events and related environmental changes in surface water than phytoplankton/bacterial standing stocks.Indeed, several previous studies based on bioassays reported that standing stocks of phytoplankton and bacterial abundance showed much smaller change to dust addition than activity measurements such as PP and BP (e.g., [13]).Similarly, Astrahan et al. [18] reported that dust added to the surface microlayer resulted in a 4-to 7-fold increase in BP, whereas bacterial abundance was only enhanced by 1.5-to 2-fold.We argue that these differences are attributed to the usually low growth rate of marine phytoplankton and heterotrophic bacteria (usually >1 d −1 , [66,67]), suggesting that measurements made <24 h post deposition may not be sufficient to detect significant changes in biomass or diversity.On the contrary, alterations in rate measurements are usually much faster and therefore can be detected on short time scales.For example, Guo et al. [68] showed that PP and BP were enhanced within 6-24 h post aerosol addition, whereas the prokaryotic community composition (determined by 16S rRNA gene diversity analysis) remained unchanged in that short time scale.Our results are consistent with the literature data compilation presented in Guieu et al. [19] and demonstrate that the response of heterotrophic microbes is usually stronger to dust deposition than that of autotrophic-dominated processes.Thus, the changes recorded in the BA, BP, and N2 fixation (which is governed mostly by heterotrophic diazotrophy in our study site, [28]) are usually higher compared to changes in Chl.a or PP (Figure 5).We argue that these differences may be attributed to the overall faster heterotrophic vs. autotrophic metabolism rate (i.e., respiration, bacterial growth efficiency), which may be especially evident in oligotrophic realms [67,69] such as the SEMS.Specifically, similar observations have been reported in other incubation studies from the Mediterranean Sea.For example, Herut et al. [12] showed that dust addition triggered a four-fold stronger response for BP than PP in the eastern Mediterranean Sea.Similarly, Pulido-Villena et al. [70] reported that Saharan dust inputs increased BA significantly, while Chl.a remained overall unchanged.The different responses between autotrophic and heterotrophic microbes may be explained by competition for resources.Heterotrophic bacteria are usually smaller than phytoplankton, therefore they have a higher surface area to volume ratio and may utilize dust-borne nutrients more efficiently [71].Another possible explanation for the higher impact on heterotrophicdominated processes is a nutrient "metabolic bypass."A similar observation was reported during the CYCLOPS campaign [71] showing that addition of PO4 3+ caused a doubling of BP, while Chl.a significantly decreased.The authors suggested that the added PO4 3+ was utilized directly by heterotrophic bacteria, bypassing the primary producers [72,73].
The overall changes in the standing stock of heterotrophic bacteria and autotrophic phytoplankton of the SEMS's coastal waters (−6% to +19%) and in the metabolic rates (+25% to +35%) reported in this study are within the range of values reported in the compilation presented in Guieu et al. [19], although for autotrophic components they were somewhat lower (Figure 5).The range of values reported for the various responses may be attributed to the amount of dust added/deposited (and its corresponding leached nutrients/trace metals), seasonality, the ambient surface microbial populations (cyanobacteria, diatoms, etc.), airborne bacteria [7,35,[74][75][76], experimental limitation due to "bottle effect" [36], and the initial trophic state of the receiving waters (Figure 6, and see discussion below).Our results are consistent with the literature data compilation presented in Guieu et al. [19] and demonstrate that the response of heterotrophic microbes is usually stronger to dust deposition than that of autotrophic-dominated processes.Thus, the changes recorded in the BA, BP, and N 2 fixation (which is governed mostly by heterotrophic diazotrophy in our study site, [28]) are usually higher compared to changes in Chl.a or PP (Figure 5).We argue that these differences may be attributed to the overall faster heterotrophic vs. autotrophic metabolism rate (i.e., respiration, bacterial growth efficiency), which may be especially evident in oligotrophic realms [67,69] such as the SEMS.Specifically, similar observations have been reported in other incubation studies from the Mediterranean Sea.For example, Herut et al. [12] showed that dust addition triggered a four-fold stronger response for BP than PP in the eastern Mediterranean Sea.Similarly, Pulido-Villena et al. [70] reported that Saharan dust inputs increased BA significantly, while Chl.a remained overall unchanged.The different responses between autotrophic and heterotrophic microbes may be explained by competition for resources.Heterotrophic bacteria are usually smaller than phytoplankton, therefore they have a higher surface area to volume ratio and may utilize dust-borne nutrients more efficiently [71].Another possible explanation for the higher impact on heterotrophic-dominated processes is a nutrient "metabolic bypass."A similar observation was reported during the CYCLOPS campaign [71] showing that addition of PO 4 3+ caused a doubling of BP, while Chl.a significantly decreased.The authors suggested that the added PO 4 3+ was utilized directly by heterotrophic bacteria, bypassing the primary producers [72,73].The overall changes in the standing stock of heterotrophic bacteria and autotrophic phytoplankton of the SEMS's coastal waters (−6% to +19%) and in the metabolic rates (+25% to +35%) reported in this study are within the range of values reported in the compilation presented in Guieu et al. [19], although for autotrophic components they were somewhat lower (Figure 5).The range of values reported for the various responses may be attributed to the amount of dust added/deposited (and its corresponding leached nutrients/trace metals), seasonality, the ambient surface microbial populations (cyanobacteria, diatoms, etc.), airborne bacteria [7,35,[74][75][76], experimental limitation due to "bottle effect" [36], and the initial trophic state of the receiving waters (Figure 6, and see discussion below).[12,13,[16][17][18][77][78][79][80][81][82][83] and this study.

Relationship between Degree of Oligotrophy and Magnitude of the Metabolic Response
We suggest that the intensity of the metabolic rate change following dust deposition is a function of the degree of oligotrophy, defined here by the initial (predeposition) concentrations of Chl.a.When ultra-oligotrophic conditions prevailed (usually during summertime, Chl.a < 0.07 µg L −1 ; Figure 2A and see [77]), stronger increases were recorded in PP (maximum 100%), BP (maximum 256%), and N2 fixation (maximum 101%) (Figures 2-4).This observation is strengthened by additional data from other studies across different marine provinces (Figure 6, and see [12,13,[16][17][18]35,[77][78][79][80][81][82][83]).We argue that the degree of the ecosystem's oligotrophy can explain the variable responses recorded in metabolic rates following dust/aerosol deposition events worldwide (Figure 6, and reviewed by Guieu et al. [19]).For example, several studies have shown that nutrient release from dust may lead to an increase in phytoplankton biomass and PP [13,84], and thus potentially stimulate C export production [85,86].Other studies, however, showed that dust addition may have deleterious effects on some cyanobacterial species/phytoplankton biomass [12,15,87,88], or may have an insignificant effect on PP rates and export production [88,89].Thus, in marine areas where nutrients (i.e., N and P) are available for primary producers, one would expect a weak response to dust deposition.In contrary, when ultra-oligotrophic conditions prevail, leached nutrients from dust/aerosol may constitute a significant portion of the nutrient resource stimulating biological responses.Another possible explanation for the higher metabolic response in ultra-oligotrophic regimes may be related to the diversity and activity of airborne microbes deposited with the dust/aerosol.Several studies have demonstrated that dust/aerosols contain a wide array of airborne microbes [6][7][8][9][10]90,91], with some remaining viable after deposition in seawater [7,35] or lake water [74,75].This subject merits further examination in future studies.

Conclusions
Atmospheric dust/aerosol concentrations have increased in the last century as a result of landuse changes and desertification [19,92].Model simulations predict that LNLC marine environments and arid areas will expand [93,94].Numerous biochemical models suggest that atmospheric deposition can fertilize the surface water with micro-/macronutrients, resulting in a proportional increase in bacterial/phytoplankton biomass and production [3,19].Our in situ measurements from the coastal SEMS show, however, that this fertilization effect is not straightforward, and that other environmental factors and interactions should be taken into account.Specifically, we show the following: (1) Dust addition affects metabolic rate measurements (PP, BP, and N2 fixation) more than phytoplankton or heterotrophic bacterial abundance/biomass.(2) Heterotrophic bacteria (biomass and activity) show a faster and stronger response to dust deposition than autotrophic-related variables (Chl.a,PP).We conjecture that this can cause a food-web cascade where heterotrophic metabolism will become more significant than autotrophic processes.(3) The more oligotrophic the ecosystem, the more pronounced the recorded production rate.

Relationship between Degree of Oligotrophy and Magnitude of the Metabolic Response
We suggest that the intensity of the metabolic rate change following dust deposition is a function of the degree of oligotrophy, defined here by the initial (predeposition) concentrations of Chl.a.When ultra-oligotrophic conditions prevailed (usually during summertime, Chl.a < 0.07 µg L −1 ; Figure 2A and see [77]), stronger increases were recorded in PP (maximum 100%), BP (maximum 256%), and N 2 fixation (maximum 101%) (Figures 2-4).This observation is strengthened by additional data from other studies across different marine provinces (Figure 6, and see [12,13,[16][17][18]35,[77][78][79][80][81][82][83]).We argue that the degree of the ecosystem's oligotrophy can explain the variable responses recorded in metabolic rates following dust/aerosol deposition events worldwide (Figure 6, and reviewed by Guieu et al. [19]).For example, several studies have shown that nutrient release from dust may lead to an increase in phytoplankton biomass and PP [13,84], and thus potentially stimulate C export production [85,86].Other studies, however, showed that dust addition may have deleterious effects on some cyanobacterial species/phytoplankton biomass [12,15,87,88], or may have an insignificant effect on PP rates and export production [88,89].Thus, in marine areas where nutrients (i.e., N and P) are available for primary producers, one would expect a weak response to dust deposition.In contrary, when ultra-oligotrophic conditions prevail, leached nutrients from dust/aerosol may constitute a significant portion of the nutrient resource stimulating biological responses.Another possible explanation for the higher metabolic response in ultra-oligotrophic regimes may be related to the diversity and activity of airborne microbes deposited with the dust/aerosol.Several studies have demonstrated that dust/aerosols contain a wide array of airborne microbes [6][7][8][9][10]90,91], with some remaining viable after deposition in seawater [7,35] or lake water [74,75].This subject merits further examination in future studies.

Conclusions
Atmospheric dust/aerosol concentrations have increased in the last century as a result of land-use changes and desertification [19,92].Model simulations predict that LNLC marine environments and arid areas will expand [93,94].Numerous biochemical models suggest that atmospheric deposition can fertilize the surface water with micro-/macronutrients, resulting in a proportional increase in bacterial/phytoplankton biomass and production [3,19].Our in situ measurements from the coastal SEMS show, however, that this fertilization effect is not straightforward, and that other environmental factors and interactions should be taken into account.Specifically, we show the following: (1) Dust addition affects metabolic rate measurements (PP, BP, and N 2 fixation) more than phytoplankton or heterotrophic bacterial abundance/biomass.(2) Heterotrophic bacteria (biomass and activity) show a faster and stronger response to dust deposition than autotrophic-related variables (Chl.a,PP).We conjecture that this can cause a food-web cascade where heterotrophic metabolism

Figure 1 .
Figure 1.Back-trajectory analyses showing the main dust sources that arrived to the southeastern Mediterranean Sea (SEMS) from 2013 to 2017: (A) Sahara Desert, (B) Eastern Europe, (C) Iraq/Syria, and (D) the Middle East; and (E) Al concentration measured during non-storm "typical" days (white, median Al = 381 ng m −3 , n = 89) and during dust storm events (orange, median Al = 3844 ng m −3 , n = 29) using a high-volume sampler.Trajectory analyses were done for 100 m (red), 500 m (green), and 1000 m (blue) altitude levels.Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.

Figure 1 .
Figure 1.Back-trajectory analyses showing the main dust sources that arrived to the southeastern Mediterranean Sea (SEMS) from 2013 to 2017: (A) Sahara Desert, (B) Eastern Europe, (C) Iraq/Syria, and (D) the Middle East; and (E) Al concentration measured during non-storm "typical" days (white, median Al = 381 ng m −3 , n = 89) and during dust storm events (orange, median Al = 3844 ng m −3 , n = 29) using a high-volume sampler.Trajectory analyses were done for 100 m (red), 500 m (green), and 1000 m (blue) altitude levels.Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.

Figure 2 .
Figure 2. Surface (1-2 m) (A,B) Chl.a temporal variability and distribution and (C,D) primary production (PP) variability and distribution, respectively, at the coastal SEMS, 2013-2017.Measurements were taken during "typical" days (white, Supplementary TableS1, n = 89) and dust storms (orange, Supplementary TableS2, n = 29).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.

Figure 2 .
Figure 2. Surface (1-2 m) (A,B) Chl.a temporal variability and distribution and (C,D) primary production (PP) variability and distribution, respectively, at the coastal SEMS, 2013-2017.Measurements were taken during "typical" days (white, Supplementary TableS1, n = 89) and dust storms (orange, Supplementary TableS2, n = 29).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.

Figure 3 .
Figure 3. Surface (1-2 m) (A,B) bacterial abundance (BA) temporal variability and distribution and (C,D) bacterial production (BP) variability and distribution, respectively, at the coastal SEMS, 2013-2017.Measurements were taken during "typical" days (white, Supplementary TableS1, n = 89) and during dust storms (orange, Supplementary TableS2, n = 29).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the box represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.

Figure 3 .
Figure 3. Surface (1-2 m) (A,B) bacterial abundance (BA) temporal variability and distribution and (C,D) bacterial production (BP) variability and distribution, respectively, at the coastal SEMS, 2013-2017.Measurements were taken during "typical" days (white, Supplementary TableS1, n = 89) and during dust storms (orange, Supplementary TableS2, n = 29).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the box represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.

Figure 4 .
Figure 4. Surface (1-2 m) N 2 fixation (A) temporal variability and (B) distribution at the coastal SEMS, 2013-2017.Measurements were taken during "typical" days (white, Supplementary TableS1, n = 40) and during dust storms (orange, Supplementary TableS2, n = 18).Box-whisker plots show interquartile range (25th to 75th percentile) of the dataset.Horizontal lines within the boxes represent median value.Letters above the box-plots represent significant differences (t-test, p < 0.05) for mean values between background and dust measurements.

Figure 5 .
Figure 5. Summary of relative change in different variables (ratio of median value during dust events to median background conditions) in response to desert dust deposition at the coastal SEMS, 2013-2017 (in situ, black squares) and the ratio compiled by Guieu et al. [19] (field and laboratory aerosol addition bioassay and mesocosm experiments, red triangles).

Figure 5 .
Figure 5. Summary of relative change in different variables (ratio of median value during dust events to median background conditions) in response to desert dust deposition at the coastal SEMS, 2013-2017 (in situ, black squares) and the ratio compiled by Guieu et al. [19] (field and laboratory aerosol addition bioassay and mesocosm experiments, red triangles).