Atmospheric and Oceanographic Forcing Impact Particle Flux Composition and Carbon Sequestration in the Eastern Mediterranean Sea: A Three-Year Time-Series Study in the Deep Ierapetra Basin

Sinking particles are a critical conduit for the export of organic material from surface waters to the deep ocean. Despite their importance in oceanic carbon cycling, little is known about the biotic composition and seasonal variability of sinking particles reaching abyssal depths. Herein, sinking particle flux data, collected in the deep Ierapetra Basin for a three-year period (June 2010 to June 2013), have been examined at the light of atmospheric and oceanographic parameters and main mass components (lithogenic, opal, carbonates, nitrogen, and organic carbon), stable isotopes of particulate organic carbon (POC) and source-specific lipid biomarkers. Our aim is to improve the current understanding of the dynamics of particle fluxes and the linkages between atmospheric dynamics and ocean biogeochemistry shaping the export of organic matter in the deep Eastern Mediterranean Sea. Overall, particle fluxes showed seasonality and interannual variability over the studied period. POC fluxes peaked in spring April–May 2012 (12.2 mg m−2 d−1) related with extreme atmospheric forcing. Summer export was approximately fourfold higher than mean wintertime, fall and springtime (except for the episodic event of spring 2012), fueling efficient organic carbon sequestration. Lipid biomarkers indicate a high relative contribution of natural and anthropogenic, marine- and land-derived POC during both spring (April–May) and summer (June–July) reaching the deep-sea floor. Moreover, our results highlight that both seasonal and episodic pulses are crucial for POC export, while the coupling of extreme weather events and atmospheric deposition can trigger the influx of both marine labile carbon and anthropogenic compounds to the deep Levantine Sea. Finally, the comparison of time series data of sinking particulate flux with the corresponding biogeochemical parameters data previously reported for surface sediment samples from the deep-sea shed light on the benthic–pelagic coupling in the study area. Thus, this study underscores that accounting the seasonal and episodic pulses of organic carbon into the deep sea is critical in modeling the depth and intensity of natural and anthropogenic POC sequestration, and for a better understanding of the global carbon cycle.

Sinking particles are a critical conduit for the export of organic material from surface waters to the deep ocean. Despite their importance in oceanic carbon cycling, little is known about the biotic composition and seasonal variability of sinking particles reaching abyssal depths. Herein, sinking particle flux data, collected in the deep Ierapetra Basin for a three-year period (June 2010 to June 2013), have been examined at the light of atmospheric and oceanographic parameters and main mass components (lithogenic, opal, carbonates, nitrogen, and organic carbon), stable isotopes of particulate organic carbon (POC) and source-specific lipid biomarkers. Our aim is to improve the current understanding of the dynamics of particle fluxes and the linkages between atmospheric dynamics and ocean biogeochemistry shaping the export of organic matter in the deep Eastern Mediterranean Sea. Overall, particle fluxes showed seasonality and interannual variability over the studied period. POC fluxes peaked in spring April-May 2012 (12.2 mg m −2 d −1 ) related with extreme atmospheric forcing. Summer export was approximately fourfold higher than mean wintertime, fall and springtime (except for the episodic event of spring 2012), fueling efficient organic carbon sequestration. Lipid biomarkers indicate a high relative contribution of natural and anthropogenic, marine-and land-derived POC during both spring (April-May) and summer (June-July) reaching the deep-sea floor. Moreover, our results highlight that both seasonal and episodic pulses are crucial for POC export, while the coupling of extreme weather events and atmospheric deposition can trigger the influx of both marine labile carbon and anthropogenic compounds to the deep Levantine Sea. Finally, the comparison of time series data of sinking particulate flux with the corresponding

INTRODUCTION
The export of carbon from the surface to the deep ocean through sinking particulate biogenic material, also known as the biological pump (Volk and Hoffert, 1985), is an essential component of the ocean carbon cycle. This plays an important role in atmospheric CO 2 fluctuations (Archer et al., 2000;Kwon et al., 2009;Lima et al., 2014) and in deep ocean ecosystems functioning (Smith et al., 2009 and references therein). The overall fraction of net primary production (NPP) that is exported from the surface ocean to the mesopelagic and bathypelagic layers and the vertical attenuation of this flux with depth are both good measures of the export efficiency (POC export flux/NPP) of the biological pump (De La Rocha and Passow, 2007;DeVries et al., 2012), which is largely being determined by the structure and functioning of the ocean's ecosystem (e.g., Francois et al., 2002;Honjo et al., 2008;Dagg et al., 2014;Le Moigne et al., 2016). However, the processes controlling the biological pump efficiency are not fully understood (Boyd and Trull, 2007). Since oligotrophic waters cover about 75% of the ocean surface (Lewis et al., 1986) and contribute over 30% of the global marine carbon fixation (Longhurst et al., 1995;Marañón et al., 2003), understanding these processes therein is critical in order to assess their biogeochemical functioning and feedbacks between the ocean carbon cycle and human-induced environmental change.
The oligotrophic regime of the EMS is modulated by sub-basin gyres and mesoscale features. These may result in distinct changes of the water column biogeochemical properties, through the horizontal and vertical mass transfer that play an important role on nutrient and carbon distribution (Siokou-Frangou et al., 2010, and references therein). Moreover, in the EMS atmospheric deposition plays a major role in the supply of bioavailable nutrients and anthropogenic compounds due to the limited inputs from external sources (Krom et al., 2004;Christodoulaki et al., 2013;Theodosi et al., 2019;Kanakidou et al., 2020). A variety of sources at the regions surrounding the Mediterranean Sea, but also distant sources due to long range transport of air masses, affect the quality of atmospheric deposition which could have important implications on the pelagic productivity and, overall, carbon cycle (Kouvarakis et al., 2001;Jickells et al., 2005;Kanakidou et al., 2012).
Understanding the processes that control the quantity and quality of POC flux to the deep EMS is needed to better characterize the strength and efficiency of the biological pump as well as to accurately project the response of this oligotrophic system to climate fluctuations and anthropogenic perturbations. Bulk geochemical proxies, such as the stable isotopic composition of organic carbon (δ 13 C), and lipid biomarkers have been widely used as indices of sources and transformation processes of organic matter in marine systems (Meyers, 1994;Wakeham, 1995;Volkman and Tanoue, 2002;Goñi et al., 2003;Hu et al., 2006;Pedrosa-Pàmies et al., 2013;Quirós-Collazos et al., 2017), as well as to assess the biogeochemical dynamics controlling the export of POC to the deep sea (Goutx et al., 2000;Wakeham et al., 2002;Conte et al., 2003;Pedrosa-Pàmies et al., 2018De Bar et al., 2019). While there have been several investigations on the δ 13 C signature of sinking particles in the Mediterranean Sea (Kerhervé et al., 2001;Turchetto et al., 2012;Pasqual et al., 2015;Theodosi et al., 2019) but also on the molecular composition of POC particulate fluxes in the western Mediterranean Sea, both in the upper mesopelagic layers (100-300 m depth) (e.g., Tolosa et al., 2005;Goutx et al., 2007;Marty et al., 2009;Méjanelle and Dachs, 2009) and the deep sea (Marchand et al., 2005;Wakeham et al., 2009), little is known about the POC molecular composition of the EMS particle fluxes (Tsapakis et al., 2006;Theodosi et al., 2013) and deep-sea floor (Gogou and Stephanou, 2004;Parinos et al., 2013a;Pedrosa-Pàmies et al., 2015).
Herein, we present a 3-year time series of sinking particles flux data in the deep Ierapetra Basin (4,310 m depth). The collected data have been examined at the light of atmospheric and oceanographic parameters and main mass components (lithogenic, calcium carbonate, opal and organic matter), stable isotopic composition of POC and source-specific lipid biomarkers (long chain n-alkanes and the unresolved complex mixture of aliphatic hydrocarbons, long-chain n-alkanols, long-chain alkenones, long-chain diols&keto-ols and selected sterols). The aim of the present study is to improve the current understanding of the dynamics of particle fluxes and the linkages between atmosphere dynamics and ocean biogeochemistry shaping the export of organic matter in this oligotrophic abyssal region of the EMS. Moreover, in order to evaluate the benthic-pelagic coupling in the study area, the time series of sinking particles flux data were compared to the corresponding data of biogeochemical parameters previously reported for surface sediments from the deep-sea in the study area. To the best of our knowledge, this study presents for the first time the lipid biomarker composition of sinking particles reaching the deep EMS.

OCEANOGRAPHIC AND ATMOSPHERIC SETTINGS
The northwestern part of the Levantine Basin covers an area extending from south of Crete up to the southeast of Rhodes island. It is dominated by the presence of the permanent Rhodes cyclonic gyre centered south of Rhodes but occasionally expanding westwards toward Crete (Figure 1). A second feature is the recurrent Ierapetra anticyclonic gyre situated to the southeast of Crete (Larnicol et al., 2002;Mkhinini et al., 2014) showing a climatological mean position close to 26-27°E and 34-35°N. The Ierapetra anticyclone is a mesoscale eddy created by the strong northerly Etesian winds and their interaction with the orography of the island of Crete. Its seasonal variability relates to that of the Etesians, hence it starts developing in summer and intensifies in autumn. Its seasonal variability is correlated with an average threemonth lag between maximum negative wind stress curl caused by the Etesians and maximum negative current vorticity linked to the anticyclone (Amitai et al., 2010). The Ierapetra anticyclone entraps water masses exiting from the Cretan straits enhancing the downward fluxes/transport of particulate matter and nutrientdepleted waters with important implications on the biogeochemical functioning of the area (Lampadariou et al., 2009;Pedrosa-Pàmies et al., 2015). The Ierapetra anticyclone also presents interannual variability in intensity and positioning. Nevertheless, during the study period (2010-2013) it had rather typical seasonal behavior (Ioannou et al., 2017).
The Levantine Basin atmosphere is at the crossroad of air masses of various origins that affect atmospheric composition. Air masses coming from the south carry significant amounts of African desert dust, from the northwest/north/northeast carry anthropogenic pollution from Central Europe, Balkans, and Ukraine, and Turkey, respectively from the east air masses are rich in dust mixed with anthropogenic pollution from the Middle-East, while coming from the west air masses are significantly affected by marine emissions (Kanakidou et al., 2011). In addition, when passing over continental regions, air masses are also affected by land sources (i.e., vegetation, forest fires, soil resuspension) (Sciare et al., 2003), and are enriched in air pollutants emitted by land sources (Mihalopoulos et al., 1997;Lelieveld et al., 2002;Kanakidou et al., 2011). Therefore, the quality of atmospheric deposition and its fingerprint in the export of material in the seawater column depend upon the origin of the air masses that are reaching the area during the study period.

Oceanographic and Atmospheric Data
To investigate the effects of various atmospheric and upper ocean processes on particle fluxes, we consider a variety of environmental parameters for the study area during the investigated period, i.e. atmospheric deposition, sea surface temperature (SST), surface Chlorophyll a (Chl a) concentration, and net primary production (NPP). These data were obtained either from in-situ observations or satellite data from one grid cell above the location of the sediment trap: 34.3-34.6°N, 26.0-26.3°E (see below for details).
Total atmospheric deposition data (dry and wet deposition) were obtained from the Finokalia atmospheric monitoring station located at a remote site in the northern coast of the island of Crete, Greece (35°20′N, 25°40′E; Figure 1). The site is not impacted by local human activities and it is a well characterized location for atmospheric composition and deposition, characteristic for the eastern Mediterranean atmosphere, operating since 1993 (Mihalopoulos et al., 1997). Moreover, in order to investigate the potential sources that contributed to the atmospheric deposition and therefore affected the flux of material into the sea water column during the study period, characteristic air mass 5-days backtrajectories have been calculated and analyzed. The backtrajectories have been computed using the HYSPLIT trajectory model of NOAA (Stein et al., 2015;Rolph et al., 2017) and considering as arriving point the Finokalia atmospheric monitoring station at 1,000 and 3,000 m.
Monthly SST and 8-days Chl a concentrations were obtained from the Moderate Resolution Imaging Spectrometer (MODIS), in orbit on the Aqua platform, using 4-km resolution level 3binned data. Both monthly SST and Chl a data are processed and distributed by the NASA Goddard Earth Sciences (GES) Data, Information Services Center (DISC) and Ocean Biology Processing Group (OBPG). Monthly NPP estimates (1/12°r esolution) were obtained from the Vertically Generalized Production Model (VGPM) (Behrenfeld and Falkowski, 1997), and provided by the Oregon State University (OSU).
Additionally, deep-sea daily mean current speed and direction at the deep Ierapetra Basin were evaluated. The moored line deployed in the Ierapetra Basin, included an Aanderaa RCM8 current meter deployed at 23 m above the bottom with a measuring interval of 1 h.

Sample Collection
Sinking particles were collected at 4,285 m water depth with a PPS3 Technicap sequential sampling sediment trap (0.125 m 2 collecting area) from June 2010 to June 2013 at the 4,310 m deep Ierapetra Basin (Figure 1). The trap was equipped with 12 receiving cups with a sampling interval of 1 month. The collecting cups were filled with a 5% (v/v) formaldehyde solution in 0.45 μm filtered seawater buffered with sodium borate. As earlier reported by Altabet (2001), formaldehyde preservative does not add sufficient organic carbon to the sediment trap material to alter organic composition (e.g., δ 13 C of POC), and it has been extensively used for the preservation of sediment trap material (Thunell et al., 2000;Tolosa et al., 2003;Struck et al., 2004;Sanchez-Vidal et al., 2008;Tesi et al., 2010;Henley et al., 2012). After recovering, sediment trap samples were stored in the dark at 2-4°C until they were further processed in the laboratory.
Moreover, in order to evaluate the benthic-pelagic linkages in the study area, the time series data of sinking particles flux are being discussed along with the corresponding biogeochemical parameters and lipid compounds/indices data previously reported by Pedrosa-Pàmies et al. (2015) for surface sediment samples (undisturbed top 0-1 cm) collected from January 2007 to June 2012 in the NW Levantine Sea (Figure 1, Supplementary Table S1). The sedimentation rate in the deep EMS is in the order of 0.003 cm y −1 (Van Santvoort et al., 1996).

Elemental and Stable Isotope Analysis of Carbon
Sediment trap samples were processed in the laboratory according to a modified version of Heussner et al. (1990). Large swimming organisms were removed by wet sieving through a 1 mm nylon mesh, while organisms <1 mm were handpicked under a microscope with fine tweezers. Samples were repeatedly split into aliquots using a high precision peristaltic pump robot to obtain 10-20 mg subsamples, and then they were stirred three times with ultrapure water, centrifuged and the supernatant removed to eliminate salt and formaldehyde. Samples were finally freeze-dried and weighted for total mass determination.
Total carbon (TC), POC and total nitrogen (TN) contents and stable isotopic composition of POC were analyzed using a Flash 1112 EA elemental analyzer interfaced to a Delta C Finnigan MAT isotope ratio mass spectrometer. Samples analyzed for %OC and δ 13 C were initially de-carbonated using repetitive additions of a 25% HCl (v/v) solution, separated by 60°C drying steps, until no effervescence was observed (Nieuwenhuize et al., 1994). Uncertainties for elemental composition were lower than 0.1%, while uncertainty for δ 13 C was lower than 0.05‰. In consistency with published data in the Mediterranean Sea we assumed OM as twice the OC content (e.g., Heussner et al., 1996;Masqué et al., 2003). The inorganic carbon content was calculated from the difference between TC and OC measurements. Assuming all inorganic carbon is contained within calcium carbonate, CaCO 3 content was determined using the molecular mass ratio of 100/12 [(TC%−OC%)×8.33].
Molar TN/POC ratios were also calculated. TN/POC is plotted in order to constrain the elemental ratios of N-depleted samples (i.e. TN/POC ≈0 rather than POC/TN ∞ 0) following Goñi et al. (2006), and to avoid the underestimation of the terrestrialderived carbon fraction (Perdue and Koprivnjak, 2007).

Biogenic Opal and Lithogenic Fraction Analysis
The biogenic silica content was analyzed using a two-step 2.5 h extraction with a 0.5 M Na 2 CO 3 solution, separated by centrifugation of the leachates. Si and Al contents of both leachates were analyzed with a Perkin-Elmer Optima 3200RL Inductive Coupled Plasma Optical Emission Spectrometer (ICP-OES), correcting the Si content of the first leachate by the Si/Al ratio of the second one (Kamatani and Oku, 2000). All values are reported as opal (SiO 2 .0.4H 2 O), a parameter defined by 2.4 times the weight percentage of biogenic Si content determined for each sample (Mortlock and Froelich, 1989). The opal detection limit, associated to the detection limit of the ICP-OES system, is approximately 0.2%. Analytical precision of opal measurements was 4.5%.

Lipid Biomarkers
A range of selected lipid biomarkers are considered in this study, namely long chain n-alkanes and the unresolved complex mixture of aliphatic hydrocarbons, long-chain n-alkanols, long-chain di-and tri-unsaturated C 37 and C 38 methyl ketones and C 38 ethyl ketones, commonly referred to as long-chain alkenones, long-chain diols and keto-ols and a suite of sterols.
The analytical procedure followed for the determination of lipid biomarkers has been previously presented in detail (Gogou et al., 1998(Gogou et al., , 2000(Gogou et al., , 2007. Briefly, freeze-dried samples were initially solventextracted three times by sonication with dichloromethane. Combined extracts were subsequently separated into different compound classes by column chromatography using silica gel that had been activated for 1 h at 150°C. The following solvent systems were used to elute the different compound classes of the considered lipid compounds: (1) n-hexane (fraction F1; aliphatic hydrocarbons), (2) dichloromethane/n-hexane (fraction F2; longchain alkenones) and (3) ethyl acetate/n-hexane (fraction F3; n-alkanols, sterols, diols and keto-ols).
The individual lipids were identified by a combination of comparison of GC-retention times to authentic standards and comparison of their mass spectral data to those in the literature. Quantification was based on the GC-MS or GC-FID response and comparison of peak areas with those of known quantities of standards added prior to the extraction of the samples ([ 2 H 50 ] n-tetracosane for n-alkanes, n-hexatriacontane for long-chain alkenones, 5α-androstan-3β-ol for sterols, diols and keto-ols, and n-heneicosanol for n-alkanols).

Source Indicators of Sinking Particulate Organic Carbon
The study of the natural isotopic variations of organic carbon (δ 13 C) is a valuable tool to trace sources and transformation processes of the particulate organic matter in the marine environment. The isotopic variations of carbon in the organic particles initially present depletion in 13 C, which derives from the formation procedures of the organic matter during primary production (Kerhervé et al., 2001). Then, a stepwise enrichment of 13 C occurs in the particles through the food chain and by the processes that are performed at the successive trophic levels (Altabet, 1996). According to this, the isotopic ratios of carbon, should be essentially affected both by the biological sources of the particles and the transformation processes that the latter have to undergo during their transfer through the water column (Kerhervé et al., 2001).
The abundance of the Unresolved Complex Mixture (UCM) of aliphatic hydrocarbons, a commonly observed persistent contaminant mixture in environmental samples consisting of branched alicyclic hydrocarbons (Gough and Rowland, 1990), is used as an indicator of the contribution of anthropogenic organic matter from degraded petroleum hydrocarbons and/or apolar products deriving from combustion processes, such as grass/ wood/coal combustion and/or the incomplete combustion of fossil fuels (e.g., Simoneit, 1984;Wang et al., 1999;Hays, 2004 and references therein).
The carbon preference index of long-chain n-alkanes (Cn with n ≥25; CPI NA ) refers to the ratio of the concentrations of the long straight-chain homologues with an odd number of carbon atoms over those with an even number of carbon atoms. Higher (terrestrial) plants produce a higher proportion of odd than even hydrocarbons with a CPI NA of ≥4 (Eglinton and Hamilton, 1967;Collister et al., 1994). Fossil fuels have a CPI NA of ∼1 (Wang et al., 1999). For phytoplankton, data are not conclusive, though they generally tend to produce shorter n-alkanes (Cn with n <20) (Volkman et al., 1998). The index is calculated as: The sum of the concentrations of the most abundant high molecular weight odd n-alkanes (TerNA) and even n-alkanols (TerN-OH), which are major components of epicuticular higher plant waxes (Eglinton and Hamilton, 1967;Ohkouchi et al., 1997), are defined, respectively, as: TerNA n−C 27,29,31,33 TerN-OH n−C 24,26,28,30 With their total sum referred to hereafter as Ter: Since the CPI NA values are indicative of a mixture of both fossil and land-derived long-chain n-alkanes in most of the cases (see discussion below), the n-alkane contribution by fossil fuels (CPI NA 1) was subtracted in order to obtain the distribution signatures of the plant wax n-alkanes. To this, each individual terrestrial higher plant wax n-alkane signature considered in TerNA sum was calculated by subtracting the average of the next higher and lower even carbon numbered homolog after Schneider et al. (1983), Simoneit et al. (1990) and Aboul-Kassim and Simoneit (1995) as follows:
Prior to PCA, the data were standardized using the mean values and standard deviations to bring all of the variables into proportion with one another. A subroutine, the Varimax rotation, was applied to the first three factors in order to maximize or minimize loadings within each factor and thus simplify the visual interpretation of PCA projections. Correlation analysis was also performed using the same variables. PCA was carried out by using STATGRAPHICS v. 18.
Deep-sea currents at the sediment trap site at in the deep Ierapetra Basin were extremely slow with a predominant S/SW direction of the flow ( Figure 2C). Intensified current speeds were recorded in January 2011 (up to 8 cm s −1 ) and March 2012 (up to 9.8 cm s −1 ) with a SW direction (i.e., topographically constrained). The magnitude of the current speeds and the strong topographic control registered in the deep Ierapetra Basin are of the same magnitude as currents registered at 3,542 m in the Samaria Canyon at the southern Cretan margin (Karageorgis et al., 2018) and at 4,300 m at NESTOR site in the deep Ionian Sea .
The opal/CaCO 3 mole ratio ranged from 0.01 to 1.60, with an average value of 0.34 ± 0.36, and the POC/CaCO 3 mole ratio from 0.38 to 3.58, with an average of 1.06 ± 0.74 (Table 1). These two ratios visibly exhibit similar trends, which are positively correlated (r 0.51, p 0.004), with the highest opal/CaCO 3 and POC/CaCO 3 ratios during summer-fall 2011 and spring 2012. The POC/opal ratio ranged from 1.04 to 83.9, with an average of 7.23 ± 15.3, and exhibited exponential negative correlation with opal/CaCO 3 (r 0.73, p < 0.0001).

Bulk Geochemical and Molecular Proxies
TN/POC and δ 13 C Total nitrogen (TN) vs POC content presents a significant positive correlation (Figure 4). The TN/POC molar ratio of the sinking particles ranged from 0.11 to 0.25, with an average of 0.15 ± 0.04 ( Figure 3D). This ratio was highest during fall 2010 and lowest during fall 2012. The δ 13 C values of the sinking POC ranged from −26.53 to −21.33‰, with an average of −24.24 ± 0.96‰ ( Figure 3D). The heaviest δ 13 C was observed during the high-flux period of spring 2012 and the lightest during the relatively low-flux period in winter 2010.
Highest fluxes and concentrations are observed during spring. The fluxes of Phyto, marine phytoplankton-derived lipids, ranged from 0.02 to 5.52 μg m −2 d −1 and POC-normalized concentrations ranged from 1.07 to 8.99 μg mg OC −1 ( Table 2). Total aliphatic hydrocarbons (AH) fluxes ranged from 0.44 to 40.6 μg m −2 d −1 , averaging 3.91 μg m −2 d −1 . Highest AH fluxes are observed during spring 2012, and in early summers 2010, 2011 and 2012. Total AH concentrations ranged from 23.2 to 492 μg g −1 (average 124 μg g −1 ). POC normalized AH concentrations are 1.07-8.99 μg mg OC −1 , with the highest concentrations during fall ( Table 2). The molecular profile of sinking particle-associated AH is dominated in all cases by a UCM and a series of resolved compounds. UCM, appearing as a unimodal hump centered around n-C 30 , was the major component of sinking particle-associated AH, accounting for up to 91% of their total sum (67% on average). UCM fluxes ranged from 0.29 to 28.7 μg m −2 day −1 , averaging 3.04 μg m −2 day −1 ( Figure 7A). The highest UCM flux was recorded in May 2012, a month after the maximum TMF. UCM concentrations ranged from 15.6 to 372 μg g −1 , and from 0.82 to 7.93 μg mg OC −1 . n-alkanes were the main AH resolved compounds accounting for up to 72% (average 31%) of their total sum and for up to 18% of total AH (average 3.1%). Total n-alkanes ( n-C 24-33 ) fluxes were 0.04-4.12 μg m −2 d −1 , and concentrations 0.74-8.45 μg g −1 and 0.11-1.31 μg mg OC −1 . During the study period, their molecular profile is dominated by long-chain homologues (n-C 20 to n-C 35 ) with a CPI NA value from 1.10 to 4.73. Highest CPI NA were observed for summer 2011, and lowest for spring 2012 coupled with the highest UCM flux (Figures 7A,D). The flux of the most abundant high molecular weight plant wax n-alkanes corrected for the contribution of oil-derived long chain oddnumbered homologues ( TerNA) ranged from 0.01 to 0.72 μg m −2 d −1 , and concentrations ranged from 0.74 to 8.45 μg g −1 (from 0.06 to 0.27 μg mg OC −1 ) ( Figure 7B).
The aliphatic alcohol fraction in the sinking particle samples is dominated in all cases by n-C 24 to n-C 30 alkanols with a maximum at n-C 26 . The plant wax n-alkanols ( TerN-OH) fluxes ranged from 0.03 to 1.65 μg m −2 d −1 , with the highest

Multivariate Analysis
PCA was conducted to assess the compositional and temporal evolution of particle fluxes during the studied period. The first two principal components are responsible for 68.8% of the variance ( Figure 9A). Factor 1 (PC1) explains 57.1% of the total variance and is characterized by positive loadings, above 0.7, for all bulk parameters, brassicasterol, dinosterol and TerNA lipid fluxes, δ 13 C and CPI NA , and negative loadings, below −0.15, for SST and atmospheric deposition. PC1 separates samples according to particle fluxes, with the highest factor scores for the extreme episodic event on spring 2012, and summer months ( Figure 9B). Thus, PC1 reflects the high particle flux periods and the impact of episodic events on the deep particle flux in the EMS. Factor 2 (PC2) explains another 11.7% of the variance and has positive loadings, above 0.7, for UCM, β-sitosterol, TerN-OH fluxes, and negative loadings, below -0.15, for TN/ POC ratio and CPI NA . Therefore, PC2 separates samples by relative contribution of land-derived POC sources. Spring and summer samples showed the highest PC2 factor scores, and fall had the lowest. Factor 3 (PC3) explains 7.9% of the total variance FIGURE 4 | Pearson correlation heatmap matrix with significance level expressed by asterisks (***p-value ≤ 0.001, **p-value ≤ 0.05, *p-value ≤ 0.1) of A) fluxes (F) and percentages (%) of bulk parameters, fluxes of lipid biomarkers, CPI NA , total atmospheric deposition, and net primary production, and (B) POC-normalized concentrations of lipid biomarkers and CPI NA , total atmospheric deposition, and net primary production. Positive correlations are displayed in blue and negative correlations in red color. Color intensity are proportional to the correlation coefficients. and is characterized by positive loadings, above 0.5, for NPP, δ 13 C, and negative loadings, below -0.15, for SST, TN/POC ratio and CPI NA . Highest PC3 factor scores were observed for spring samples. Therefore, PC3 separates samples by the relative contribution of phytoplankton-derived POC.

Efficiency of the Biological Pump in the Deep Ierapetra Basin
The mean annual TMF measured in the deep Ierapetra Basin ( Figure 3A) is fairly comparable with those reported in previous studies that investigated particle fluxes in the bathypelagic EMS (e.g., Kerhervé et al., 1999;Stavrakakis et al., 2000;Patara et al., 2009;Karageorgis et al., 2018;Theodosi et al., 2019). For comparison, in the western Mediterranean Sea, TMF above 500 mg m −2 d −1 has been reported in the Alboran Sea (e.g., Fabres et al., 2002;Sanchez-Vidal et al., 2005) and submarine canyons of the NW Mediterranean Sea (e.g., Zúñiga et al., 2009;Stabholz et al., 2013). POC fluxes recorded in this study are also of the same order of magnitude as those reported in previous studies that investigate POC fluxes in the deep EMS. The low POC fluxes during winter and spring are in line with the POC fluxes measured at 2,700 m in the Ierapetra Basin by Koppelmann et al. (2004) during winter/ early spring 1999 (0.25-0.57 mg m −2 d −1 , from January to March) and late spring 1999 (1.16 mg m −2 d −1 during April). Karageorgis    November 1994to November 1995 in the southern Cretan Sea at 200 m and 1,515 m water depth, respectively. The low POC fluxes reported herein match the very low NPP observed in the Ierapetra Basin during the study period (∼118 g C m −2 y −1 ) and reported from other studies in the open waters of the Levantine Basin (97-113 g C m −2 y −1 ) (Antoine et al., 1995;Napolitano et al., 2000;Bosc et al., 2004), which is lower than in the center of large oceanic gyres, such as the Sargasso Sea, where values around 150 g C m −2 y −1 have been measured (Steinberg et al., 2001;Lomas et al., 2013). For comparison, the primary productivity of coastal and upwelling zones, in some of the most productive areas of the ocean, typically exhibit values close to 250 g C m −2 y −1 (Longhurst et al., 1995). The fraction of primary production exported to the deep Ierapetra Basin averaged 0.35%, which is comparable to what has been observed in the EMS in the southern Ionian Sea at depths from 2,000 to 2,800 m Gogou et al., 2014), but two-fold higher than what was reported at 4,300 m in the sourtheastern Ionian Sea .
Over the study period, the export efficiency was low during 66% (<0.02, twentieth percentile of data) of the time, and moderate during 30.6% of the time (between 0.2 and 0.02) ( Figure 3C). This underscores the strong oligotrophic conditions of the study area. To further investigate the efficiency of the biological carbon pump we tested the correlation between the export efficiency (POC flux/NPP) and NPP ( Figure 11). Herein we cannot conclude that in the EMS there is an inverse relationship between PP and export efficiency, as previously identified for the Southern Ocean (Maiti et al., 2013;Cavan et al., 2015;Le Moigne et al., 2016) and globally (Henson et al., 2019). We observed that, overall, there is no clear relationship between the export efficiency and NPP, except when observing seasonal trends (see discussion in section "Seasonal Export Pulses to the Deep Eastern Mediterranean Sea" below).
The POC content of sinking particles of the deep Ierapetra Basin (3.00 ± 1.60% POC) was 6-fold higher than the reported for deep-sea surface sediments in the study area (0.49 ± 0.09% POC). Thus, our data indicate that, on average, just about 16% of the POC that reaches the bathypelagic Ierapetra Basin gets settled and preserved to the seafloor sediments. Frontiers in Earth Science | www.frontiersin.org February 2021 | Volume 9 | Article 591948

Drivers of Natural and Anthropogenic Particulate Organic Carbon Export
The positive correlation of POC fluxes with the different mineral fractions, especially with the biogenic minerals ( Figure 5), observed throughout the study period suggests that the ballasting effect of the biogenic minerals is an important factor promoting the export of organic carbon to the deep EMS by increasing the density of settling particles, and/or providing POC protection against remineralization (Armstrong et al., 2002;Francois et al., 2002;Klaas and Archer, 2002). These correlations agree with the study from Klaas and Archer (2002) where they evaluated data from 52 sediment trap experiments worldwide. Low slope of opal vs POC flux suggest that in the EMS opal is more effective than carbonates at facilitating the POC flux. Plankton and microbial community structure in the upper ocean also play a significant role in the export and transfer efficiency of carbon to the ocean interior (Herndl et al., 2008;Guidi et al., 2009;Henson et al., 2012). The main functional groups of phytoplankton contributing to export production through direct sedimentation are nitrogen fixers (e.g., diazotrophic cyanobacteria), silicifiers (e.g., diatoms), calcifiers (e.g., coccolithophores) and other marine plankton (e.g., dinoflagellates) (Agusti et al., 2015;Le Moigne et al., 2015;Turner, 2015;Basu et al., 2018). Generally, in the EMS there is a predominance of dinoflagellates and coccolithophores over diatoms (Malinverno et al., 2003;Ignatiades et al., 2009;Skampa et al., 2020 and references therein), except for the Rhodes gyre area under strong upwelling conditions where diatoms can be dominant (Siokou-Frangou et al., 1999). Therefore, the more stable POC/CaCO 3 ratio compared to the POC/opal ratio (Table 1) can be associated with the phytoplankton community dynamics and morphological and ecological differences between diatoms and coccolithophorids (De La Rocha and Passow, 2007). Moreover, the relatively high fluxes and POC-normalized concentrations of cholesterol (C 27 Δ 5 ) indicate that secondary production has a critical role exporting POC to the deep EMS.
Phytoplankton growth rate, and availability of CO 2 , light and nutrients affect isotopic fractionation and thus the δ 13 C of POC (Rau et al., 1992;Burkhardt et al., 1999). Winter and spring are the only two seasons when sinking particles reaching the deep Ierapetra Basin show a significant positive correlation (spring: r 0.98, p < 0.0001; winter: r 0.90, p 0.003) with sinking POC flux and δ 13 C values (Figure 11). Because primary production affects the CO 2 concentration near the producers, there is a positive correlation between δ 13 C of the phytoplankton and primary production, and consequently, also with sinking POC fluxes (Deuser et al., 1968;Rau et al., 1992). This suggests that during summer and fall, either CO 2 was not depleted enough to cause detectable enrichment of 13 C, or δ 13 C values were shaped by other processes than primary production. Factors contributing to the decline in δ 13 C of POC values can be microbial activity (Lehmann et al., 2002) and inputs from the overlying surface waters of terrestrial or anthropogenic POC. δ 13 C values in marine algae from low-to mid-latitude temperate seas vary from -18‰ to -22‰ (Goericke and Fry, 1994;Meyers, 1994;Harmelin-Vivien et al., 2008), whereas most terrestrial POC inputs from C3 plants are -25‰ to -28‰ (Hedges et al., 1997) and POC from crude oil and petroleum products has values between −28.5 and −28.9‰ (Rumolo et al., 2011, and references therein). Therefore, a significant incorporation of terrestrial and anthropogenic POC may be a cause of the more depleted δ 13 C values in the summer and fall samples. The δ 13 C of the sinking particles was significantly lighter (average isotopic change -1.9‰) than those of surface sediments ( Figure 11). This isotopic change indicates isotopic discrimination during sinking processes and potential incorporation of allochthonous material from the deep EMS. The isotopic fractionation between sinking particles from the deep ocean and surface sediments has been previously reported (e.g., Fischer and Wefer, 1996;Nakanishi and Minagawa, 2003 and references therein), but this is the first study to document it for the EMS.
The deviation of the TN/POC ratios of the sinking particles from the classical Redfield ratio (Redfield et al., 1963), further indicates POC degradation processes of marine labile compounds during transport from surface waters to the deep Ierapetra Basin ( Figure 11A). The overall higher TN/POC ratios from the sinking particles vs the seafloor sediments ( Figure 11) further reflect the preferential degradation of labile nitrogen-enriched organic matter and the re-working of marine algal organic matter by zooplankton and benthic invertebrates during transport and deposition (Gogou and Stephanou, 2004;Lampadariou et al., 2009). Additionally, by comparing the lipid biomarker composition of sinking particles vs surface sediments, we can provide insight about the pre-depositional processing of the Frontiers in Earth Science | www.frontiersin.org February 2021 | Volume 9 | Article 591948 organic matter affecting the POC content and quality in the deep EMS before it gets buried. POC-normalized concentrations of phytoplankton-( Phyto) vs land-derived ( Ter) lipid biomarkers of sinking particles compared with surface sediments (Figure 8) show that sinking particles are an order of magnitude more enriched in phytoplankton-derived POC, indicating the higher preservation efficiency of terrestrial vs marine POC (e.g., Hoefs et al., 2002;Burdige, 2007;Blair and Aller, 2012). The weak negative correlation between molar TN/POC ratios and δ 13 C values of the sinking particles (Figures 4 and 11b) indicates that the POC composition of the sinking particles was rather a mixture of marine (high TN/OC and δ 13 C) and terrestrial (low TN/OC and δ 13 C) particulate organic matter, but may also suggest significant contribution of other sources (e.g., anthropogenic POC). Indeed, the profile of AHs is indicative of the contribution of anthropogenic POC in the considered samples. Specifically, the CPI NA values of long-chain n-alkanes ranging from 1.10 to 4.73 (Figure 7d), being, <2 in 61% of the cases, are indicative of a mixture of long straight-chain aliphatic compounds from both fossil (unburned fossil fuels) and land-derived (higher plant waxes) sources (Wang et al., 1999). Moreover, the presence of a UCM in all cases is indicative of anthropogenic POC deriving from degraded petroleum hydrocarbons and/or apolar products deriving from combustion processes (Wang et al., 1999). The entire EMS is a marine setting under intense anthropogenic pressure receiving substantial amounts of petroleum hydrocarbons through direct discharges from merchant shipping and oil transportation, mainly along shipping routes (Gogou et al., 2000;Parinos et al., 2013a;Parinos and Gogou, 2016;Hatzianestis et al., 2020). Furthermore, UCM hydrocarbons can also be derived from combustion processes, such as grass/wood/coal combustion and/ or the incomplete combustion of fossil fuels, and are found in atmospheric aerosols collected from marine and urban areas of the EMS constituting a major source of combustion-related AHs through long-range atmospheric transport and subsequent dry/ wet deposition (e.g., Tsapakis et al., 2006;Theodosi et al., 2013;Gogou et al., 1998 and references therein).
The strong positive correlation of UCM vs. POC fluxes ( Figure 4A) is in agreement with the widely documented fact that the biogeochemical cycling of anthropogenic hydrocarbons Frontiers in Earth Science | www.frontiersin.org February 2021 | Volume 9 | Article 591948 in marine systems is largely controlled by the biological pump and physico-chemical properties of individual compounds (Farrington and Takada, 2014, and references therein). Fossil hydrocarbons introduced into surface waters via the direct release of petroleum products by ships are more available to the dissolved phase, and are efficiently scavenged by organic rich particles (e.g., phytoplankton and fecal pellets), resulting in their sinking in open seas. Direct and indirect evidence of biodegradation has been shown as an important factor affecting their abundance during their downward transport in the water column. On the other hand, hydrocarbons derived from combustion sources, which are mainly introduced into surface EMS waters through long-range atmospheric transport and subsequent dry/wet deposition, are strongly associated to fine combustion particles that protect them from degradation during their transport through the water column. Sorption of combustion particles onto biogenic aggregates has been highlighted as the important mechanism regulating their cycling in marine systems (Deyme et al., 2011;Parinos et al., 2013b;Theodosi et al., 2013 and references therein)

Seasonal Export Pulses to the Deep Eastern Mediterranean Sea
Export during summer periods in the deep Ierapetra Basin was approximately four-fold greater than the export in wintertime, fall and springtime (except for the episodic event in spring 2012). In summer, the highest relative contribution of lithogenic to the TMF and relatively high POC fluxes coincided with the lowest NPP and high atmospheric dry deposition (Figures 2 and 3). It is welldocumented that in the Levantine Basin the lowest Chl a and primary production levels and the highest transparency are found during the thermally stratified summer period Bosc, 2004;Ignatiades et al., 2009). Our results indicate that the export efficiency is relatively high during summer in the Levantine Basin (below Figure 3C). This rather high export efficiency in summertime has been previously reported in a wide range of open oceanic regions (Buesseler, 1998;Henson et al., 2012;Karl et al., 2012;Puigcorbé et al., 2015;Smith et al., 2018), and highlighted by biogeochemical models . The export efficiency can be highly variable depending on the degree of (de-)coupling between phytoplankton community structure (Buesseler, 1998;Francois et al., 2002), and upper ocean remineralization by zooplankton (Steinberg et al., 2002;Dagg et al., 2014;Cavan et al., 2015) and bacteria (Buchan et al., 2014;Belcher et al., 2016;Le Moigne et al., 2016;Henson et al., 2019). Sinking velocities in the EMS have been estimated to be 100-200 m d −1 , and are accelerated by pulses of primary production and grazing and by Saharan dust events that fertilize the upper ocean (Patara et al., 2009) The fact that the relatively high POC fluxes in summer were enriched with land-derived natural AHs (higher plant waxes) as inferred from both TerNA and TerN-OH concentrations and elevated CPI NA ratio values), anthropogenic AHs (UCM) (relatively high PC1 and PC2 scores), but also from phytoplankton-and zooplankton-derived lipids suggests that POC export to the deep EMS in summer is controlled by the interactions between the biological pump and different types of ballast minerals (dust/-induced POC export by fertilization and aggregation/adsorption processes) (Marañón et al., 2010;Bressac et al., 2014;van der Jagt et al., 2018).
Moreover, the relatively high fluxes of brassicasterol, dinosterol and C 30 diols&keto-ols, and low POC-normalized concentrations of these phytoplankton-derived lipids coupled with high cholesterol fluxes and relatively high POCnormalized concentration of cholesterol (Figure 6), suggest that zooplankton has a significant role in the particle flux and POC degradation in summertime through grazing and repackaging in fecal pellets. This is consistent with previous studies that have reported relatively high export efficiencies during summer, when grazing is able to keep pace with NPP so that the POC flux via fecal pellets increases . Summertime nutrient limitation in the EMS leads to lower biological production and a plankton community structure dominated by small pico-autotrophic cells (<3 μm) (Psarra et al., 2000;Meador et al., 2010;Siokou-Frangou et al., 2010;Varkitzi et al., 2020). Raveh et al. (2015), revealed that autotrophic cyanobacteria such as Synechococcus and Prochlorococcus generally dominate the phytoplankton biomass during summer in the eastern Levantine Basin, and are also the main contributors to primary production. Small phytoplankton cells have a higher surface area-to-volume ratio, nutrient assimilation efficiency, photosynthetic activity, faster division rates and lower sinking rates than larger phytoplankton (Raven, 1998;Finkel et al., 2010). Therefore, picoautotrophdominated communities during summer in the EMS may represent food webs with a high degree of aggregate repackaging with potential for accelerated sinking (Buesseler and Boyd, 2009;Bach et al., 2016).
The strong positive correlation of POC-normalized concentrations of Ter vs. Phyto of summer sinking particles (Figure 8), but also fluxes of UCM vs. TerNA, Supplementary Figure S2), suggest that coupled with the plankton dynamics, the mineral ballast plays an important role in the export of natural and anthropogenic POC to the deep EMS during summer. For the EMS the periods of increased atmospheric dust are in spring and early summer, followed by autumn, and the dust sources are from the coasts of Africa, Arabian Peninsula and Red Sea (Moulin et al., 1998;Israelevich et al., 2012). Strict dry deposition is a highly effective mechanism of dust deposition in the Mediterranean Sea, and in summer it can contribute up to 93% to the total dust deposition due to the reduction or even absence of wet deposition (Kubilay et al., 2000). The highest summertime atmospheric deposition values were recorded in 2011. During this period, high aerosol dry deposition was recorded, associated with air masses from the south, carrying significant amounts of desert dust, and also air masses marked by local influence (Supplementary Figure S3).
In winter-spring (November to May) Chl a concentration increased ( Figure 2B) but compared to summer, POC fluxes to the deep Ierapetra Basin were relatively low ( Figure 3B Figure S4).
Fall and winter had the lowest particle and POC fluxes. These low fluxes could be explained by the microbial communities in the mesopelagic zone reducing sinking velocity of particles during this period and thus low export to the deep Ierapetra Basin (Sempéré et al., 2000;Herndl and Reinthaler, 2013). This hypothesis of POC reworking by microbial communities is supported by the very low POC-normalized concentrations of labile phytoplankton-derived lipids (i.e., brassicasterol and dinosterol), and cholesterol (lowest PC3 factor scores) (Figure 9). Fall is the only season where we observed an apparent relationship between NPP and export efficiency ( Figure 10). This could imply that NPP become decoupled from the processes acting to remineralize POC in the upper ocean during fall periods.
Additionally, during fall period, it is particularly noteworthy to highlight September 2011, which had high total atmospheric deposition ( Figure 2A). Air masses had significant continental and regional influence, coming from the northeast and east (sweeping the Balkans and Turkey), but also from the west (African continent) (Supplementary Figure S5). Therefore, during this period atmospheric deposition provided terrestrial material enriched in soil dust and vegetation, as imprinted in elevated land-derived terrestrial lipids ( TerNA and TerN-OH) POC-normalized concentrations and CPI NA values (Figure 7). Interestingly, the highest POC-normalized concentrations of UCM (but also TerNA, TerN-OH and β-sitosterol) are recorded during late fall and winter periods. This could be attributed to the fact that fall/winter periods have the highest atmospheric wet deposition, especially in February 2011 (Figure 2A). During this month, air masses were significantly influenced by land, both coming from the Atlantic close-by the northwest Africa and reaching the EMS after having swept Central and Eastern Europe, and also directly coming from the African continent (Supplementary Figure S6). Atmospheric scavenging by rain drops of hydrocarbons emitted from anthropogenic (mainly air masses originating from the NW/N/NE sector) and terrestrial sources (from the SW/S sector), and subsequent efficient sorption of these hydrocarbons on organic-rich particles is likely the main mechanism driving their cycling during this period.

Extreme Episodic Event in Spring 2012: Coupling of Atmospheric Conditions Funneled Labile and Recalcitrant POC to the Deep EMS
In spring 2012, TMF and POC fluxes were one and two orders of magnitude, respectively, higher than in other spring periods over the study period ( Figure 3C). As highlighted by Pedrosa-Pàmies et al. (2016), winter and spring of 2012 were characterized by exceptional atmospheric conditions. During winter 2011-12, particularly strong cold and dry northerly winds triggered intense convection not only in the EMS but also in the NW Mediterranean Sea (Durrieu de Palanques and Puig, 2018), and in the Adriatic Sea (Bensi et al., 2013;Janeković et al., 2014). The intense convection in winter 2012 caused an upwelling of relatively nutrient-rich cold water masses into the euphotic zone, triggering an exceptional phytoplankton FIGURE 10 | Net primary productivity (NPP) plotted against export efficiency (POC flux/NPP). Horizontal dashed lines indicate threshold for export efficiency regimes: high export efficiency is defined as >0.2 (eightieth percentile of data), low export efficiency is <0.02 (twentieth percentile), and moderate export efficiency lies between. Significative regression lines are shown for summer (Export efficiency 0.00006NPP-0.0082, r 0.824, p < 0.05), and fall (Export efficiency −0.00001NPP+0.0043, r 0.661, p 0.053).
Frontiers in Earth Science | www.frontiersin.org February 2021 | Volume 9 | Article 591948 bloom on March 2012 ( Figure 2B). As emphasized by Pedrosa-Pàmies et al. (2016), the exceptional phytoplankton bloom was most probably stimulated by the arrival of Si and Fe from airborne Etna volcano ash during March-April 2012. Volcanic eruptions have the potential to cause regional or even global-scale nutrient (mainly Fe)-fertilization (Olgun et al., 2013;Browning et al., 2015;Weinbauer et al., 2017). Mount Etna experienced high paroxysmal volcanic activity from January 2011 to April 2012 (Scollo et al., 2014;Giacomoni et al., 2018). In February 2012, the air masses that arrived at Finokalia station from the west had passed over south Italy/Sicily, thus collecting pollutants emitted from Etna (Supplementary Figure  S7). Similarly, during March air masses remained of west sector and were influenced by volcanic activities in the Etna region together with marine emissions from the Mediterranean Sea, but also from land emissions over Africa (first part of the month) and Central Europe and long-range transport (second part of the month) (Supplementary Figure S8). Therefore, atmospheric deposition during this month integrates all potential sources surrounding the East Mediterranean, including Etna's emissions, except the East sector source (sources in the Middle-East).
In spring 2012 there was the highest export efficiency of the study period ( Figure 3C), indicating that there was a decoupling between primary producers and the grazing population, allowing a strong episodic export pulse to reach the deep Ierapetra Basin (highest PC1, PC2 and PC3 factor scores) (Figure 9). The results of this study highlights that POC fluxes during this reported extreme episodic event were enriched in natural and anthropogenic, marine-and land-derived, lipid biomarkers (as inferred from UCM fluxes and CPI NA ratio values ∼1) (Figure 7). In April 2012, the POC exported during this event shows the highest δ 13 C values (−21.33‰) over the study period, suggesting a high contribution of POC derived from marine relatively to land-derived natural sources (Goericke and Fry, 1994;Harmelin-Vivien et al., 2008). It is particularly noteworthy that surface sediment samples collected in the deep Ierapetra Basin (i.e., BF1MC22, Ier01 and Red 11) showed similar δ 13 C and TN/POC ratio as sinking particles from the episodic event of spring 2012 ( Figure 11B), suggesting that pulses of sinking particles during extreme episodic events can have a major role in modulating the seafloor POC composition in the deep Ierapetra Basin.
During the same month, the highest fluxes of brassicasterol of the study period and the highest opal fluxes, indicate the enhanced sedimentation of relatively fresh diatoms reaching bathypelagic depths. Although dinosterol, C 30 diol and alkenones fluxes also had their maxima in April 2012, the high mole ratios of opal/ CaCO 3 and POC/CaCO 3 in the sinking particles ( Table 2) suggest that the upwelling event preferentially exported siliceous plankton relative to calcareous plankton (although the calcareous plankton benefitted too). This conclusion agrees with the phytoplankton community dynamics of the Rhodes' Gyre, where high abundances of large diatoms have been found under extreme meteorological conditions resulting in nutrient enrichment (Siokou-Frangou et al., 1999;Siokou-Frangou et al., 2010). In upwelling regions, the phytoplankton communities are often dominated by large diatoms (Malviya et al., 2016).
Supportive to the above, is that the settling particles collected during this period were embedded in a mucilaginous matrix with a high abundance of ash spherules and highly diverse, wellpreserved, medium-sized (5-300 μm) long-chain-forming  (Meyers, 1994;Goñi and Hedges, 1995;Hedges et al., 1997;Harmelin-Vivien et al., 2008;Sanchez-Vidal et al., 2013). Symbol shapes correspond to season and colors correspond to sampling year.
Frontiers in Earth Science | www.frontiersin.org February 2021 | Volume 9 | Article 591948 centric diatoms such as Chaetoceros spp., Thalassiosira spp., and Rhizosolenia spp (Figure 12). This mucilaginous matrix could be attributed to the production of transparent exopolymer particles (TEP) from diatom exudates rich in acidic polysaccharides (Gogou and Repeta, 2010). TEP increase the collision rates between particles, mostly in the form of large, rapidly sinking aggregates, playing an important role in carbon cycling in pelagic/ oceanic ecosystems in general (Passow, 2002b and references therein) and the oligotrophic EMS in particular (Bar-Zeev et al., 2011;Parinos et al., 2017;Ortega-Retuerta et al., 2019). Diatoms have been reported to produce large amounts of TEP precursors or TEP directly via sloughing and lysis of senescent colonies under exponentially growing conditions (Passow, 2002a), such as in April of 2012, which are likely to have enhanced post-bloom flocculation and massive sinking of particulate matter in the study area. The ash particles, which typically have a density of twice that of seawater, could have also played an important role for rapid settling of biogenic particles (Passow and De La Rocha, 2006). Furthermore, the exceptional diatom-dominated bloom fueled higher trophic levels, which further enhanced POC export. The covariation of the heterotrophic activity with the labile phytodetritus flux, producing a pulse of zooplankton-derived POC to the deep waters, is evidenced by the increase of cholesterol fluxes in April and May 2012 (3.20 and 7.23 μg m −2 d −1 , respectively) ( Figure 6A). In the oligotrophic Sargasso Sea (N Atlantic), a similar ecosystem response has been observed as a result of the impact of mesoscale ocean features (fronts and eddies) (Conte et al., 1998(Conte et al., , 2003 and extreme weather events (Pedrosa-Pàmies et al., 2019), locally enhancing vertical advection and stimulating production and sinking export of labile marine carbon to the deep ocean (e.g., increased cholesterol fluxes of 4.95 μg m −2 d −1 at 3,200 m depth after a hurricane passage). Sinking particles during this event were relatively depleted in land-derived natural POC and anthropogenic compounds due to dilution from biogenic material. Nevertheless, in April 2012, the highest fluxes of TerNA and TerN-OH, a decrease of CPI NA and an increase in UCM fluxes (Figure 7) indicate that there was an efficient transfer of anthropogenic and higher plant waxderived hydrocarbons to the deep Ierapetra Basin.

CONCLUSION
In the Ierapetra basin, particle flux data have been examined in combination with atmospheric and oceanographic parameters and main mass flux components (lithogenic, calcium carbonate, opal and organic matter), stable isotopes of POC (δ 13 C) and sourcespecific lipid biomarkers, aiming to improve the current understanding of the dynamics of particle fluxes and the impact of external forcing on the deep Eastern Mediterranean Sea. This study highlights that both seasonal and episodic pulses are crucial for POC export to the deep Ierapetra Basin, and likely most of the deep EMS. Overall, summer particle export to the deep EMS fuels more efficient carbon sequestration than during the other seasons. Our results also show that the combination of extreme weather events and aerosol deposition can trigger an influx of marine organic matter to the deep EMS. This influx is a critical factor determining food supply for deep ocean ecosystems, but also induces an increased influx of anthropogenic organic carbon to the deep EMS. Therefore, this study underscores the importance of accounting both seasonal and episodic pulses of POC to the deep sea when modeling the sequestration of natural and anthropogenic POC, and for a better understanding of the global carbon cycle. Finally, the comparison of biogeochemical parameters of the sinking particles flux data with previously reported surface sediments from the deep-sea in the study area revealed an isotopic discrimination, as also as a preferential degradation of labile organic matter during deposition and burial, along with higher preservation of land-derived POC in the underlying sediments. This line of research provides key knowledge to better understand the export, fate and preservation vs. degradation of marine and land-derived organic carbon and for modeling the organic carbon burial rates in the Mediterranean Sea.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
RP-P performed the sample analyses and data processing and lead the interpretation of the results and wrote the manuscript. CP aided in interpreting the results and worked on the manuscript. RP-P and AS-V participated in the sample collection. AS-V and AC were involved in the design of the study, and supervised RP-P work during the early stage of the study. AG and CP supervised RP-P during the lipid biomarkers analyses at the Organic Chemistry Laboratory at HCMR. NL and MC were the REDECO's project co-ordinator and the lead of the University of Barcelona research team, respectively. DV contributed to the interpretation of the oceanographic settings of the study area. NM and MK contributed with Finokalia atmospheric observatory data and atmospheric data interpretation. All authors provided feedback on the manuscript.

FUNDING
This research was supported by the REDECO (CTM2008-04973-E/ MAR) and PERSEUS (GA 287600) projects. We further acknowledge support by the projects PANACEA-'PANhellenic infrastructure for Atmospheric Composition and climatE chAnge' (MIS 5021516) and ENIRISST-'Intelligent Research Infrastructure for Shipping, Supply Chain, Transport and Logistics' (MIS 5027930), which are implemented under the Action "Reinforcement of the Research and Innovation Infrastructure," funded by the Operational Program "Competitiveness, Entrepreneurship and Innovation" (NSRF 2014-2020) and co-financed by Greece and EU; and by the Action "National Νetwork on Climate Change and its Impacts -Climpact" which is implemented under the sub-project 3 of the project "Infrastructure of national research networks in the fields of Precision Medicine, Quantum Technology and Climate Change," funded by the Public Investment Program of Greece, General Secretary of Research and Technology/Ministry of Development and Investments." Researchers from GRC Geociències Marines benefited from a Grups de Recerca Consolidats grant (2017 SGR 315) by Generalitat de Catalunya autonomous government.