Nutrients Released by Aerosols
Aerosol composition of both S and A filters was overall within the ranges reported in the literature for the NW Mediterranean (Viana et al., 2005; Pey et al., 2010; Escudero et al., 2011). Nutrient measurements inside the microcosms before and after the aerosol additions (time lag ≈ 3h) showed that S dust released on average 0.25 ± 1.01 μmol TIN mg-1, 0.02 ± 0.02 μmol TIP mg-1, and 1.67 ± 6.91 μmol TOC mg-1. These values are within the reported values of TIN dissolved from S dust in previous experiments carried out in the Mediterranean (Herut et al., 2005, 2016; Lekunberri et al., 2010). TIP concentration is higher than those usually found in the bibliography but within the range reported by Pulido-Villena et al. (2008) after a real deposition event of S dust (2.58 g m-2) in the NW Mediterranean (release of P: 0.03 μmol mg-1), and in microcosm experiments adding similar concentrations of dust than in our experiments (0.02–0.03 μmol mg-1). TOC showed high variability between experiments, especially in the S microcosms. It sometimes increased after adding the dust, but in other occasions it showed no increase or even decrease (Figure 2). TOC concentrations in surface waters of the Catalan coast range from 50 to ca. 150 μmol l-1 (Vila-Reixach et al., 2012; Romera-Castillo et al., 2013; Table 1). Thus, the amount of TOC supplied by aerosols is usually much lower than the actual concentration in the seawater. In situations of OC co-limitation, it may be quickly taken up by heterotrophic bacteria, making difficult the determination of TOC released by aerosols. Although there was no apparent OC co-limitation at the beginning of our experiments, the increase in inorganic nutrients after the aerosol additions may have produced an unbalance in the Redfield ratio in the microcosms, yielding to the quick consumption of TOC. Alternatively, the decrease in TOC determined in some experiments may be associated to the error of the analytical procedure, that is in the same range of the TOC concentration released by the aerosols in the case of the S samples (CV = 1.44 ± 0.76%). A high variability in TOC supplied by aerosols has also been reported in the literature (Duarte et al., 2006; Pulido-Villena et al., 2008; Lekunberri et al., 2010).
Anthropogenic particles released higher amounts of TIN (1.98 ± 1.79 μmol TIN mg-1) and TOC (4.70 ± 3.45 μmol TIN mg-1), but the same concentration of TIP per milligram of aerosol than S dust. TIN released by these aerosols was slightly higher than that reported by Guo et al. (2014) (1 μmol mg-1), but much lower than that determined by Bonnet et al. (2005) and Duarte et al. (2006) (∼24 μmol mg-1). The similar concentration of P released from both types of particles, higher than expected for S particles, must be related to the fact that S particles arriving to the NE of the Iberian Peninsula are mixed with local and regional sources of pollution during their trajectory. Pollution involves increases in NOx and SOx compounds, the main acid precursors species in the atmosphere (Stockdale et al., 2016), and the solubility of P increases with acidity (Nenes et al., 2011; Stockdale et al., 2016). Hence, it seems reasonable to find a higher release of P from dust collected in the Catalan coast than in other experiments performed close to the desert or in more remote areas less exposed to pollution (e.g., Bonnet et al., 2005; Herut et al., 2005; Guo et al., 2016). In fact, Stockdale et al. (2016) suggested that the flux of bioavailable P from acidified dust is expected to increase in the Mediterranean Sea during the coming years.
The additions of TOC released with both aerosols (from 0 to 7 μmol l-1) are negligible compared to ambient concentration, as pointed above. However, P released by the atmospheric particles (0.02 ± 0.02 μmol l-1) is in the same range of P annual concentration in the offshore station (Lazzari et al., 2016), and within the range of concentrations observed during the summer stratification period in BCN (Arin et al., 2013; Romero et al., 2014) and BLA (Guadayol et al., 2009) coastal waters. Thus, both aerosols have the potential to be an important source of P during the stratification period in coastal waters, and during the whole year in offshore waters of the NW Mediterranean Sea. TIN released by the S dust was overall small compared to ambient concentrations in western Mediterranean coastal waters, but it might also be an important source of inorganic N during the stratification period, when TIN concentration in the seawater can be <0.1 μM (Guadayol et al., 2009; Arin et al., 2013; Romero et al., 2014). The average TIN concentration leached by the A particles (1.67 μmol l-1) is equivalent to the annual average concentration in BLA and in the offshore station, and about three times higher than that during the stratification period (Gasol et al., 2012; Lazzari et al., 2016). This concentration is also close to the TIN summer average in BCN (Arin et al., 2013; Romero et al., 2014). A aerosols have therefore the potential to constitute an important source of bioavailable N during the whole year in coastal locations of the NW Mediterranean, but especially during the stratification period.
Aerosol Effect on Bacterial Abundance and Metabolism
Our results show that HBP was significantly enhanced with both types of atmospheric particles (in the SU experiment of BCN), or only with the A particles (when all the experiments were considered together), whereas aerosols did not yield significant increases in bacterial abundance overall. In a previous work where only results from the BCN experiments were considered, we found a significantly higher enhancement of bacterial abundance with S dust (Marín et al., 2017). This evidences the importance to perform several experiments, considering different initial conditions, to be able to draw general patterns regarding the effect of aerosols in marine surface waters, especially in the more variable coastal environment. In other previous studies carried out with S dust, a great stimulation of bacterial production was observed, in comparison with no, or lower, response on bacterial abundance (e.g., Duarte et al., 2006; Lekunberri et al., 2010; Marañón et al., 2010; Pulido-Villena et al., 2014). Studies that evaluated the effect of A aerosols on bacterial abundance did not observe significant increases either (Bonnet et al., 2005; Ternon et al., 2011), while Herut et al. (2016) noticed a larger increase in bacterial production than in bacterial abundance after the addition of A aerosols in eastern Mediterranean surface waters. Teira et al. (2013) found a higher effect on bacterial production than on abundance in experiments with rainwater added to coastal eutrophic waters of the Ría de Vigo (NW Spain). We did observe a significant increase of HBA in the filtered samples compared to those not filtered, though (Figure 3D and Table 3), pointing to the importance of competition and grazing processes controlling bacterial growth and masking effects on biomass. We discuss the results from this particular experiment further below.
We also looked at the activity of three extracellular enzymatic activities: AMA, APA, and β-Gl. Extracellular enzymes can be synthetized by several organisms, including bacteria, phytoplankton, heterotrophic nanoflagellates, and zooplankton (Karner et al., 1994; Jamet and Boge, 1998). Nonetheless, APA is mainly synthesized by phytoplankton, whereas AMA and β-Gl are mainly produced by bacteria (Vives-Rego et al., 1985; Münster et al., 1992). Thus, the enzymatic activities related with bacteria were the ones more stimulated with A aerosols (AMA, spAMA, β-Gl, and spβ-Gl). An increase in AMA activity is indicative of either TIN limitation or a supply of dissolved organic N, while an increase in β-Gl points to a higher hydrolysis of carbohydrates, mainly derived from cellulose (Rath et al., 1993; Kamer and Rassoulzadegan, 1995; Sala et al., 2001, 2016; Misic et al., 2008). As A particles produced a significant increase in inorganic N (and in the N:P ratio), we can discard the stimulation of AMA as indicative of TIN limitation. Hence, the observed enhancement of AMA and β-Gl activities determined in these experiments in the A microcosms must be attributed to either an addition of organic sources of N and C, or the hydrolysis of polymeric substrates already present in the seawater that can be mineralized thanks to the supply of inorganic nutrients that could be limiting bacterial metabolic processes.
In aerosol-amended experiments where the whole planktonic community is assessed, bacteria might take up the organic compounds needed for growth either from the atmospheric particles or from those released by phytoplankton cells. The concomitant increase of HBP and the three enzymatic activities observed toward the end of the experiment carried out in BCN in SP (Figures 4B, 5B, 6B, 7B) may indicate a release of organic compounds from phytoplankton cells, as we observed a peak of Chl 1 day after the aerosol additions, and then phytoplankton concentration quickly decreased toward the end of the experiment, and so decreased TIN and TIP (Marín et al., 2017). However, the enhancement of HBP and enzymatic activities observed in BLA during SP and in the three SU experiments is unlikely an indirect effect from phytoplankton released compounds, since HBP and enzymatic activities started to increase earlier or at the same time as Chl in these experiments (data not shown). Aerosols, especially of A origin, are also a source of OC (Viana et al., 2005; Pateraki et al., 2012), as reflected in the increase in humic-like and protein-like substances (Supplementary Figure S1). However, these compounds may be of recalcitrant nature, as bacteria did not take them up through the incubation time (Sánchez-Pérez et al., 2016). More specifically, the enhancement of β-Gl suggests that A particles may be a source of carbohydrates. The main source of these compounds to the atmosphere is biomass burning (Pöschl, 2005), probably coming from local waste incineration. The stimulation of AMA also suggests a supply of organic N. Thus, although, we did not find a correlation between the TOC released by the aerosols and the biological response (Figures 8, 9), the stimulation of AMA and β-Gl points to the exhaustion of the labile fraction of organic compounds in the seawater (Chróst, 1990) and the hydrolysis of polymeric substrates that either were already present or came from the aerosols.
Regarding the inorganic nutrients, we found that A aerosols released significantly more nitrate and TIN than S dust. Hence, the higher stimulation of HBP and enzymatic activities observed in the A samples points to inorganic N (and more specifically nitrate) as the major driver in the observed response. In fact, considering all the results from the six experiments, most of the biological variables showed a high correlation with nitrate in the A samples (Figure 8A). The biological response to S dust instead was not significant when considering all the experiments, but the SU experiment of BCN (SUBCN) (Table 3). Interestingly, considering both the A and S samples in SUBCN, we found a significant positive correlation between TIP, HBP, and APA, and it was also higher between TIP and AMA and β-Gl, than between NO3– and these biological variables (Figure 9). One could expect an inverse correlation between APA and TIP (Sala et al., 2001, and references therein). However, the positive correlation between TIP and APA found in this experiment can be explained because the AIR value for TIP was calculated considering the moment just after the aerosol additions (section “Statistical Analysis”), while maxima APA values were found toward the end of the experiment (Figure 7D), when TIP was starting to decrease. In this experiment, aerosols released a higher concentration of phosphate than in other experiments, and the concentration of N related to that of P at the beginning of the SUBCN experiment (N:P ratio = 46) was much larger (Table 1), being P the main limiting nutrient in this case. In all, the increase observed in HBP and enzymatic activities with the atmospheric – especially the A – particles should be a combination of two different factors: (1) the release of inorganic nutrients (i.e., TIN and TIP) and some organic compounds from the aerosols, and (2) the re-mineralization of particulate organic matter already present in the seawater. The final impact will be determined by the chemical composition of the aerosols and the initial status of the seawater. In agreement with our results, Herut et al. (2016) pointed to N as the main driver of the higher response observed in the microbial community with A particles compared to S dust in the eastern Mediterranean, whereas Pitta et al. (2017) pointed to both N and P as the main components released by dust triggering a response in the bacterial community when seawater was N and P co-limited (N:P = 10). Therefore, as A aerosols release more inorganic N than P, these particles have a huge potential to stimulate bacterial activity when both N and P are in low concentrations, a common situation in coastal (Guadayol et al., 2009) and open waters (Moutin et al., 2012) of the western Mediterranean during the stratification period. N has been considered the main limiting nutrient in most of the oligotrophic ocean (e.g., Moore et al., 2013). However, N2 fixation may be important in some of these areas (Bonnet et al., 2011), where phosphorous and iron could become the main limiting nutrients. In any case, as oligotrophic marine areas represent more than 50% of the global ocean (Antoine et al., 1996), it is important to address the effect of an increasingly polluted and acidified atmosphere (IPCC, 2014; Stockdale et al., 2016) in these regions.
When evaluating the effect of the location in SU, a larger stimulation of HBP, spHBP, and APA was observed offshore than in the coastal locations. Marañón et al. (2010) showed that the effect of aerosols on HBP increased with increasing oligotrophy. Hence, the larger increase of these variables observed offshore can be attributed to the higher oligotrophy of the open ocean waters offshore Balearic Islands (Chl annual average of 0.25 ± 0.09 μg l-1; Gallisai et al., 2014) in comparison with the coastal waters of BCN (Chl average of 1.58 ± 1.09 μg l-1; Romero et al., 2014) and BLA (Chl average of 0.63 ± 0.05 μg l-1; Guadayol et al., 2009). In fact, inorganic nutrients and Chl concentration were much lower at the beginning of the offshore experiment than in the others (Table 1), and inorganic N and P were significantly higher after the aerosols addition offshore than in the other experiments (Table 3). It must be noted that offshore waters were amended with aerosols from BCN, which are probably richer in N, P, and TOC than actual aerosols over the atmosphere of the OFF station. In contrast, HBA, AMA, spAMA, and β-Gl were more stimulated in BCN. TOC and TOP were significantly higher in BCN, but it is difficult to establish causality, as no clear correlation was observed between TOC and TOP and the biological variables, and both remained at similar concentrations in all microcosms through the incubation time. With respect to the season, HBA and HBP were more enhanced in SU and SP than in WI. In SU, this was attributed to the high oligotrophy of the initial water, with P being the main limiting nutrient, and aerosols leaching more P than in other experiments, as previously said. While in SP, although the initial water was rich in both N and P (Table 1), both started to decrease quickly, the extra nitrate added with the aerosols rapidly taken up by phytoplankton, yielding a subsequent increase in HBP (Marín et al., 2017). Enzymatic activity instead was higher in WI than in the other seasons, but without apparent differences between treatments (Figures 5A, 6A, 7A). In the WI experiment, an increase in Chl was observed the day after the addition in all the microcosms, followed by a decrease toward the end of the experiment (Marín et al., 2017). Thus, the major stimulation of enzymatic activities observed in WI in all the microcosms is likely due to the remineralization of organic compounds already present in the seawater.
Finally, we have shown that bacterial abundance and activity can be directly stimulated by atmospheric deposition, and that the effect is higher without the presence of competitors (i.e., phytoplankton) and predators (i.e., heterotrophic nanoflagellates). These results explain why previous studies where only bacteria were considered (e.g., Pulido-Villena et al., 2008) showed a higher effect of S dust on bacterial abundance with respect to other experiments with the whole planktonic community, at similar dust additions (e.g., Herut et al., 2005; Pulido-Villena et al., 2014). More experiments testing the effect of aerosols on the heterotrophic prokaryote community with and without the presence of competitors and predators need to be carried out to confirm our results, though. No significant differences were found in nutrient concentration between the filtered and not filtered samples, except, surprisingly, for TIN (Table 3). TIN tended to decrease toward the end of the experiment in all the samples, but one of the S containers showed a second peak on days 3 and 4, probably associated to the remineralization of organic matter already present in the seawater. Although this peak was of not enough intensity to represent a difference in the biological activity, much higher in the F samples, this might be the reason of the higher content in TIN found in the NF samples. It must also be pointed out that the effect of both types of aerosols in the F samples was similar for almost all the biological variables, except β-Gl and spβ-Gl, which were higher in the A microcosms (Figures 3D, 4D, 5D, 6D, 7D; SUBCN). These observations suggest that, in the absence of competitors and predators, bacteria show a similar preference to both types of aerosols. Bacterial losses later in the experiment may be attributed to virus or residual nanoflagellates passing through the 0.8 μm filter and grazing, unchecked themselves. In any case, these losses are also similar in both A and S. We did not found significant differences in the concentration of heterotrophic nanoflagellates in the unfiltered A and S samples in any of the experiments (data not shown). Hence, the higher bacterial production and activity of AMA observed with the A particles when the whole community was evaluated must then be attributed to either the indirect stimulation after phytoplankton growth with these particles (which may be the case of BCN during SP) or to the presence of other competitors for S dust. For instance, cyanobacteria were significantly more stimulated with S dust than with the other treatments in the experiment carried out in BLA during SU (Marín et al., in preparation).