1. Introduction
Marine viruses, the smallest and most abundant biological entities in the sea, play key roles in biogeochemical cycles, shape microbial communities, and are the largest reservoir of diversity throughout the water column from the tropics to polar systems [
1,
2]. The Southern Ocean (SO), which surrounds Antarctica, plays an essential role in regulating the world’s climate by contributing to the global water circulation system. One of its main traits is the Antarctic Circumpolar Current (ACC), which flows clockwise around Antarctica. The SO is characterized by low temperature and relatively low salinity waters and comprises open sea regions in which high inorganic nutrient availability coexists with low chlorophyll concentrations (HNLC regions), primarily due to a lack of iron supply for phytoplankton to perform photosynthesis [
3] and the limited light during most of year. The phenomenon of lower algal biomass than that expected for the high concentrations of inorganic nutrients present in the system is known as the Antarctic paradox [
4]. However, during the Austral summer, notable phytoplankton blooms occur throughout the area [
5,
6]. These are followed by the proliferation of prokaryotes, heterotrophic protists and viruses, which can reach high levels of abundance, activity, and diversity [
7]. Processes like grazing by protists and/or zooplankton (through sloppy feeding), as well as viral lysis of prokaryotes and phytoplankton, break microbial cells [
8] and promote the leaching of organic matter and micronutrients, such as iron-rich organic compounds, which become available for the growth of prokaryotic and eukaryotic phytoplankton [
9,
10,
11,
12].
Furthermore, viruses, phytoplankton, and prokaryotes are presumably key agents involved in making secondary metabolites, including volatiles that escape to the atmosphere and eventually may evolve into marine secondary aerosols, crucial in the creation of cloud condensation nuclei and, therefore, having consequent effects on climate [
13,
14]. A conspicuous secondary metabolite is dimethyl sulfoniopropionate (DMSP), an algal osmolyte that is produced in high intracellular concentrations by many phytoplankton taxa [
15,
16,
17]. Two of the major aerosol-forming volatiles are isoprene (2-methyl-1,3-butadiene), which is a by-product of algal photosynthesis [
14,
18], and dimethyl sulfide (DMS), which derives from DMSP through the action of enzymatic lyase activity [
15,
16,
19,
20]. Generally, haptophytes, cryptophytic, and small dinoflagellates are the groups recognized to be greater DMSP producers [
16,
21]. DMSP is released from the cell mainly through senescence or exudation in late phases of blooms [
22,
23], but most importantly through grazing [
20,
24] and viral attack [
25,
26,
27] and is partly transformed into DMS. Isoprene is more related to photosynthetic activity as its concentration peaks match those of phytoplankton activity [
28,
29,
30,
31], and chlorophyll-normalized isoprene production rates are known for their different phytoplankton taxa (see Booge [
32] for a review). However, the relationship between phytoplankton, viruses, and isoprene remains poorly explored.
The dominating groups of phytoplankton during austral summer are diatoms, dinoflagellates, haptophytes, cryptophytes, and prasinophytes [
33,
34,
35,
36]. Phytoplankton proliferations are followed by specific prokaryotic taxa [
37] that covary with changes in the phytoplankton community [
38,
39,
40]. Both prokaryotes and phytoplankton are subjected to grazing by protists and lysis by viruses [
41,
42,
43,
44]. Environmental fluctuations could influence the abundance and activity of potential hosts, which will imply changes in the abundance and composition of the viral community. This was observed in several Antarctic lakes ([
45] and refs therein), where changes from a single-stranded DNA to a double-stranded DNA–virus-dominated assemblage appeared to respond to a seasonal shift in host organisms [
46]. At the same time, virus mortality processes are constrained by environmental factors [
47,
48]. For example, it has been reported that virus-host interactions are affected by changes in temperature and salinity [
49,
50], as well as by inorganic and organic nutrient concentrations, as viruses have a high demand for nitrogen and phosphorous during replication [
51]. However, for Antarctic marine waters, there is still little information known on the sources of viral abundance and community variation, except for the study carried out by Miranda [
52] who described seasonal changes of ssRNA viral communities in coastal Antarctic waters, in relation to variations of phytoplankton communities in spring and summer.
In this study, our goal was to explore the relationships between physicochemical and biological factors, and the abundance and composition of marine viral communities from the surface and the DCM (deep chlorophyll maximum). We studied these relationships in four regions of the Southern Ocean, characterized by distinct hydrographic conditions and dominated by different phytoplankton groups. Specifically, we tested whether (i) viral abundance and community composition were different among the four regions, and (ii) to which degree these dissimilarities were linked to the variability of potential hosts or environmental factors. To achieve our objective, we assessed the abundance and viral community composition in these four regions, and examined their potential relation to an array of (1) physicochemical (temperature, salinity, inorganic nutrients, fluorescent dissolved organic matter -FDOM- as an indicator of organic matter, and DMSP, DMS, and isoprene concentrations), and (2) biological (prokaryote abundance and prokaryotic heterotrophic production, chlorophyll a concentration, and phytoplankton taxonomic composition and biomass) variables. Additionally, we discuss the potential role of the distinct viral communities detected at each site in the production of aerosol precursor compounds.
4. Discussion
Viral communities showed some segregation differences among the sampled regions, both at surface waters and at the DCM (
Figure 4). However, while some viral populations were geographically constrained, others were widely distributed, as shown by the banding pattern of the RAPD (
Figures S2 and S3). Indeed, several studies have reported a high global viral diversity and almost as high local diversity [
74,
75,
76,
77,
78,
79], as well as connectivity along the water column [
80,
81,
82,
83]. This might be based, on one hand, in the migration capability (i.e., transport by oceanic currents and sinking attached to particles) of marine viruses [
84], and on the other hand, on environment selective pressure [
15,
17].
Authors such as Breitbart and Rohwer [
75] suggested the seed-bank theory, where high local viral diversity relies on host availability and diversity. There, viruses would move from the seed-bank state to the active phase whenever their host would be “blooming”. Once the predominant host cells decay, the empty niche would be occupied by another hosting species, and the consequent displacement of viruses from the seed-bank state would occur [
75]. These ecological dynamics may apply in the present study, explaining the obtained grouping of Antarctic viral community structures according to the sampled regions (in particular in the NSO and SSO regions (
Figure 4)). In addition, Brum et al. [
77] proposed temperature as a relevant environmental factor that influences changes in viral activity and diversity. Thus, during the spring–summer transition, they detected a reversion from lysogenic to lytic viruses that increased the viral and microbial diversity. All these processes will favor the geographical differentiation of viral assemblages.
In our study, temperature and salinity substantially varied among regions (
Figure 1B and
Figure S1), and the different degrees of stratification of the water column (with mixed layers from 16 m in SSO to 50 m in NSG) might have affected the light penetration and nutrient supply from deeper layers to the upper mixed layer, as is reported in Nunes et al. [
36], for the studied area. In addition, it is expected that the development of different phytoplankton species [
35] and prokaryotes [
82,
83] may influence the environmental conditions, modifying nutrient concentration and contributing to the production of secondary metabolites, such as DMSP, DMS, and isoprene [
14,
15,
16,
17,
18,
85,
86]. At the same time, viruses could participate in these biogeochemical processes infecting phytoplankton and prokaryotes and releasing to the environment dissolved organic matter (DOM) from the cell-enclosed material [
87,
88]. It makes sense, then, that our results showed a strong significant correlation between fluorescent dissolved organic matter (FDOM, peak T, considered an indicator of proteinaceous compounds [
89], and labile DOM [
57]), and all viral abundance fractions (
Table 2). Then, viruses would intervene in regulating microbial biomass and diversity, which in turn would be reflected in the viral community composition [
25,
26,
27,
84,
88]. Viral abundance related positively with that of their potential hosts (prokaryotes and phytoplankton), but negatively with some inorganic nutrient concentrations (NO
3, SiO
42− and PO
43− (
Table 2)). This could be a consequence of inorganic nutrient uptake by algae at rates which vary among different phytoplankton taxa and depend on micronutrient availability [
36]. Thus, in the iron-rich waters of NSG [
36], where diatoms dominated the phytoplankton biomass (
Figure 3), the inorganic nutrient concentrations, especially silicate, registered the lowest values (
Table 1). In contrast, in the other regions (NSO, SSO, and WA), where iron was limited [
3], inorganic nutrient concentrations were higher (
Table 1).
The potential role of viruses in producing secondary metabolites as DMSP (DMS precursor) by lysing phytoplankton host cells, was experimentally reported by Hill et al. [
26]. Then, we expected that viral abundance and DMSP concentration would correlate; however, that was not reflected in our results (
Table 2). Bratbak et al. [
25] also observed no correlation between viral abundance and DMSP concentration and posed that bacteria degradation represents the major sink for DMSP and DMS, which is also supported by Kiene and Service [
90]. Furthermore, other DMSP-freeing pathways (i.e., grazing, senescence, apoptosis, etc.) may also play a role in increasing its release [
20,
22,
23,
91,
92]. In the case of DMS, its concentration was negatively correlated with viruses. DMS has been shown to be strongly dependent on microbial community composition (e.g., presence of DMS consumers [
15] or demethylating bacteria [
17]) and to environmental conditions (e.g., wind-related ventilation and photolysis), which modify the structure of the microbial food web [
15]. This may be the case of the area south of South Orkney (SSO) area, where the highest irradiance was registered, and thus, DMSP to DMS transformation could be enhanced. The formation of isoprene is mainly attributed to the activity and physiology of phytoplankton [
28,
29,
30,
31], while the effect of viral lysis is still unclear [
14,
18]. In our study, although isoprene concentration and viral abundance strongly covaried (
Table 2), we believe that it is not a cause–effect. We are aware that diatom viruses are mainly ssRNA and ssDNA [
93], and viral abundances reported here, by FCM, refer to dsDNA. Then, through multiple regression analyses, we obtained that the model that better predicted the isoprene concentration variability highlighted diatom biomass as the only explanatory factor and excluded the dsDNA viral abundance (log isoprene conc. = −3.53 (±1.43) + 0.37(±0.07) × log diatom conc.; n = 18, R
2 = 0.78,
p < 0.0001). Hence, further investigation is needed in studying the interaction between diatoms and their specific viruses in the formation of isoprene. Indeed, some studies with other photosynthetic microorganisms reported an increase of isoprene production after viral infection, as is the case of
Prochlorococcus infected by phages [
14,
18].
In summary, the observed regional segregation of the different viral communities (
Figure 4 and
Figure 5) in this study, could be a result—to a greater or lesser extent—of the presence of different potential prokaryote and phytoplankton hosts, as observed by several authors [
35,
50,
51,
77,
82,
83], and to the variability of some physicochemical parameters [
15,
17]. Viral communities at the NSO and SSO regions were associated with the coldest waters and with high inorganic nutrient concentrations (
Table 1,
Figure 5). In particular, the group of viruses dominating the NSO region also coexisted with high DMSP concentration (
Table 1) and a high proportion of haptophyte and cryptophyte biomass (
Figure 2B,C), which according to Stefels et al. [
16], are the main phytoplankton taxa responsible of the DMSP production. However, this was not reflected in the increase of DMS, perhaps due to low bacterial lyase activity, important in the DMSP degradation [
91,
92,
94], or to environmental factors (i.e., temperature and sunlight etc. [
15]). In the NSG and WA regions, although the number of samples was low, viral communities were more heterogeneous. Indeed, high temperatures and the availability of different dominant potential prokaryotic and phytoplanktonic hosts may influence the viral community composition. The high prokaryotic heterotrophic production and prokaryote abundance recorded at the NSG could be the results of a high growth rate of different successful prokaryote communities, derived from the carbon released by the various diatom taxa blooming during the sampling time [
36]. High isoprene concentrations can be also associated with this bloom of diatoms, according to several studies [
29,
95,
96]. Finally, regarding the variability of viral assemblages in the WA, relatively high temperatures could be associated with an advanced state of phytoplankton succession [
36], which may have been driven by the high abundance and diversity of potential hosts such as haptophytes, cryptophytes, and prokaryotes [
75].