Covariance of Marine Nucleocytoplasmic Large DNA Viruses with Eukaryotic Plankton Communities in the Sub-Arctic Kongsfjorden Ecosystem: A Metagenomic Analysis of Marine Microbial Ecosystems

Abstract

Nucleocytoplasmic large DNA viruses (NCLDVs) infect various marine eukaryotes. However, little is known about NCLDV diversity and their relationships with eukaryotic hosts in marine environments, the elucidation of which will advance the current understanding of marine ecosystems. This study characterizes the interplay between NCLDVs and the eukaryotic plankton community (EPC) in the sub-Arctic area using metagenomics and metabarcoding to investigate NCLDVs and EPC, respectively, in the Kongsfjorden ecosystem of Svalbard (Norway) in April and June 2018. Gyrodinium helveticum (Dinophyceae) is the most prevalent eukaryotic taxon in the EPC in April, during which time Mimiviridae (31.8%), Poxviridae (25.1%), Phycodnaviridae (14.7%) and Pandoraviridae (13.1%) predominate. However, in June, the predominant taxon is Aureococcus anophagefferens (Pelagophyceae), and the NCLDVs, Poxviridae (32.9%), Mimiviridae (29.1%), and Phycodnaviridae (18.5%) appear in higher proportions with an increase in Pelagophyceae, Bacillariophyceae, and Chlorophyta groups. Thus, differences in NCLDVs may be caused by changes in EPC composition in response to environmental changes, such as increases in water temperature and light intensity. Taken together, these findings are particularly relevant considering the anticipated impact of NCLDV-induced EPC control mechanisms on polar regions and, therefore, improve the understanding of the Sub-Arctic Kongsfjorden ecosystem.

1. Introduction

Viruses are the most abundant organisms on earth, with the highest genetic diversity compared to other biological organisms in marine ecosystems [1,2]. In particular, viruses kill approximately 20% of the oceanic microbial biomass daily, thus, providing significant energy sources, including nutrients and dissolved organic matter [1,3]. Although little is known about viruses that infect eukaryotic plankton communities (EPCs), nucleocytoplasmic large DNA viruses (NCLDVs) are known to infect various eukaryotic organisms [4,5], and to be strongly associated with eukaryotic plankton [3,4,6]. In fact, the diversity and richness of NCLDV families, namely Mimiviridae (i.e., Megaviridae), Phycodnaviridae, Pandoraviridae, Poxviridae, Iridoviridae, Marseilleviridae, Pithoviridae, Ascoviridae, Asfarviridae, and Mininucleoviridae, are higher than those of other viruses in the sea [7,8]. Meanwhile, various marine autotrophic eukaryotes, such as haptophytes, chlorophytes, and dinoflagellates, act as hosts to NCLDVs, which can have fundamental roles in the killing or regulation of phytoplankton blooms in marine coastal waters [9]. Additionally, NCLDVs infect heterotrophic organisms [10,11]. Thus, the diverse groups of NCLDVs highlight the ecological impact exerted by the marine ecosystem in biogeochemical cycling [1,12,13,14]. In fact, several viruses belonging to Phycodnaviridae and Mimivirdae reportedly contribute to biological carbon export from marine surfaces to deep layers, via host-cell death [15,16]. Despite the ecological importance of NCLDVs, data related to the relationship between NCLDVs and eukaryotic hosts remains limited.

Arctic warming, also known as ‘Arctic amplification’, is increasing rapidly compared to global warming in other continents; that is, temperatures have risen faster on the Arctic continent over the past decades [17,18]. Changes caused by ‘Arctic amplification’ include weak stratification, increased vertical mixing, and sea-ice loss. These changes in the Arctic Ocean are similar to those observed in the Southern Ocean, referred to as ‘Atlantification’ [19]. This continued influx of warm Atlantic waters into the Arctic Ocean will likely result in environmental changes caused by ice and snow melting, as well as the development of new ecosystems, including those inhospitable to Arctic endemic species [20,21]. The sub-Arctic Kongsfjorden (Svalbard, Norway) ecosystem is affected by Atlantic Ocean currents characterized by high temperatures and high salinity [22]. Payne and Roesler [22] explained that in Kongsfjorden, the intrusion of Atlantic Waters warmed subsurface waters by 4 °C in 15 years. This further led to sea ice melting, rapid glacial retreat, altered water-column structure, and changes in chlorophyll-a in the region. In addition, the Kongsfjorden Sea exhibits distinct differences in the Arctic Ocean ecosystem when the water temperature drops below 0 °C in April, marking the beginning of the white nights. Meanwhile, the highest primary productivity is observed in June. In this season, the pre-dominant phytoplanktons include dinoflagellates, diatoms, Chrysophyceae, and Dictyochophyceae [23,24]. Furthermore, Coupel, et al. [25] and Fujiwara, et al. [26] observed an increased appearance of nano- and pico-phytoplankton, such as cryptophytes, including Micromonas and cryptophytes species. Until recently, it was believed that the increase in phytoplankton primary production within the ice-covered Arctic Ocean only began after the sea ice receded, when the water mass stratified to provide sufficient light for photosynthesis [27,28]. However, in certain areas of the Arctic, earlier phytoplankton blooms may have been induced by early sea-ice retreat, occurring widely throughout the Arctic [29]. Consequently, the Arctic Ocean experiences weakened ocean stratification with the bottom layers susceptible to surface heat by enhanced vertical mixing and lower salinity caused by ice melting [30]. However, a comprehensive understanding of Arctic phytoplankton indicator dynamics remains elusive. Moreover, the relationship between NCLDVs and phytoplankton dynamics in the Kongsfjorden ecosystem is unknown.

Recently, a metagenomic tool for analysis of the virus community has been used in various marine ecosystems, including pole-to-pole and deep-sea ecosystems [6,31]. De novo sequencing and bioinformatics technologies have been developed using NCLDV data obtained through sequences extracted from environmental samples collected during various ocean research cruises [32,33,34]. The rapid advancement of metagenomic technology has facilitated the facile acquisition of high-resolution sequences, further enabling the analysis, and more accurate classification, of microorganisms at the genus and species levels, within populations and/or communities in ecological studies. Moreover, owing to advancements in metagenomics, a technology has been developed to analyze microbial diversity using a smaller volume of seawater and DNA [35]. Hence, the current study seeks to examine the NCLDVs and EPC in a sub-Arctic Kongsfjorden ecosystem (glacial fjord) by employing these metagenomic platforms. Although Kongsfjorden, an Arctic glacier fjord area on the western coast of the Spitsbergen Island in Svalbard, is geographically situated in the Arctic, it is affected by various factors, including the Atlantic Ocean currents and Arctic Sea [22,36,37]. In particular, the active physical properties are influenced by changes between the Svalbard coastal waters (glacial outflow) and the western continental shelf coastal waters [38]. Moreover, the surface waters of Kongsfjorden vary with complex factors, including tides, currents, wind, precipitation, water inflow from glacial melt and the coast, and sea ice/glacier drift [39]. The primary aims of this study are to (1) elucidate the spatial distribution and comparison between NCLDVs and EPCs during the early white night (April) and mid-summer (June); (2) identify NCLDVs with a strong association and co-occurrence with the EPCs; and (3) describe potential NCLDV hosts. Collectively, these findings will aid future studies of viral and host community structures in the Arctic seas, particularly within cold waters near the Arctic Ocean, to provide an explicit linkage between viruses, host cells, and water masses and currents.

2. Materials and Methods

2.1. Measurement of Environmental Parameters

The temperature and salinity in the sub-Arctic Kongsfjorden ecosystem were measured using the Sea-Bird 911 plus CTD model (Sea Bird Electronics, Inc., Bellevue, WA, USA). Water samples were collected using a 20-L rosette-type Niskin sampler bottles (Ocean Test Equipment, Inc., Ft. Lauderdale, FL, USA) from the surface (1 m below the sea surface) and bottom layers (1 m above the bottom sediment) at seven sampling sites (St. 1, 2, 4, 5, 7, 8, 9 and 10) on board the RV TEISTEN in April 2018 and six sites (St. 1, 2, 5, 7, 8 and 10) on board the RV MS-FARM in June 2018 (Figure 1 and Table S1). The bottom layers (St. 5, 7, 8, and 10) were investigated in June at a depth of 100 m due to mechanical issues with the Niskin sampler of the research vessel. The size-fractioned chlorophyll-a was analyzed according to three continuous steps: >20 μm (micro-), 2–20 μm (nano-), and 0.2–2 μm (pico-sized chlorophyll-a). Next, 1000 mL of seawater was filtered sequentially through 20, 2, and 0.2 μm pore size and 47 mm diameter Merck Millipore filters (NY2004700, TTTP04700, and GTTP04700, respectively). Gravity filtration was applied for the 20 μm filtrations, while a vacuum of approximately 200 mm Hg was used for the 2 and 0.2 μm filtrations. Each size-fractionated filter was soaked in 4 mL of cold 90% acetone-distilled water (v/v). Subsequently, chlorophyll-a was extracted for 24 h at 4 °C in the dark, and its concentration was measured using a fluorometer (Trilogy; Turner Designs, Sunnyvale, CA, USA).

2.2. Metavirome Analysis of NCLDVs

To analyze the dsDNA virus community, 30 L of seawater was collected for duplicated experiments of 15 L each from the surface and bottom layers in each sampling site and stored at approximately 4 °C for transportation to the DASAN Arctic Research Center of Korea Polar Research Institute, the Republic of Korea (located in Kongsfjorden, Svalbard, Norway). However, some water samples (St. 1, 5, 7, 8, and 10) collected in April were not sufficient for duplicate experiments, owing to a disturbance caused by the sea-ice movement and large waves during sampling.

dsDNA virus communities were analyzed according to a previously described protocol [34] with certain modifications. To extract genomic DNA (gDNA) from the dsDNA virus community, 15 L of seawater were passed through a 3-μm filter (TSTP04700, 47 mm, Merck Millipore, Bedford, MA, USA) to remove inorganic and organic particles. DNA viruses were harvested from pre-filtered seawater by flocculation, filtration, and resuspension methods using FeCl₃, as previously described [40]. To aggregate DNA viruses onto Fe₃⁺ ions, the reservoir (cylindrical with black-colored acrylic materials) was maintained at 20 °C for 1 h. The seawater in the reservoir was mixed with three impellers, each 6 cm wide and 15 cm long, at 10 rpm [41]. The aggregated DNA viruses flocculated using Fe₃⁺ ions were harvested using a 0.2 μm polycarbonate membrane (111106; 47 mm; Whatman, Buckinghamshire, UK) and stored at −80 °C until total gDNA extraction. Although the temperature for resuspending the virus community aggregated in Fe ions on the filter should be maintained at 4 °C [40], despite the losses of viral DNA, they were kept at −80 °C due to sample transportation from Norway to South Korea. Subsequently, the membrane was cut into eight pieces and placed in a 50-mL conical tube with a suspension buffer (10 mL of 0.1 M EDTA, 0.2 M MgCl₂, 0.2 M Ascorbate). Flocculated viruses attached to the membrane were suspended and mixed in the buffer (pH 6, adjusted using approximately 5 mL of 10 M NaOH). Total gDNA was extracted using a Viral Gene-spin Viral DNA/RNA Extraction Kit (iNtRON Biotechnology, Seoul, Republic of Korea). The gDNA library for metagenomics was prepared using the NEBNext Ultra II DNA Library Prep Kit (Illumina, San Diego, CA, USA). The sequencing library was prepared through random fragmentation of the DNA sample, followed by 5′ and 3′ adapter ligation. The adapter-ligated fragments were amplified using polymerase chain reaction (PCR) over 12 cycles (denaturation: 98 °C 30 s [1 cycle]; annealing: 98 °C 10 s, extension: 65 °C 75 s [12 cycles], final extension: 65 °C 5 min [1 cycle]). Illumina tagged Universal primer (5′-AAT GAT ACG GCG ACC ACC GAG ATC TAC ACT CTT TCC CTA CAC GAC GCT CTT CCG ATC-3′) and Indexed primers using the NEBNext Multiplex Oligos (Illumina, San Diego, CA, USA). The DNA library was then sequenced via paired-end sequencing using the Illumina HiSeq 2500 platform.

Subsequently, the modified bioinformatics analysis methods were performed in accordance with the protocol described by Kim et al. Kim, Jung, Park, Kim, Lee, Ha, and Lee [34]. Briefly, the raw sequencing data (Fastq file) was sorted by trimming using the Qiagen CLC genomics workbench (v. 20.0.4 Qiagen, Hilden, Germany), which removed low-quality adapters according to the criteria of two ambiguous nucleotides and a minimum length of 50 nucleotides. Next, an assembly analysis was performed using metaSPAdes v. 3.13.0 [42] and five k-mer sizes (K-mer: 21, 33, 55, 77, 99, 128) were selected. A quality check of viral contigs was then performed using Check V v.1.0 to remove contiguous host regions and non-viral regions from the assembled proviruses [43]. Through the Check V quality check, only virus contigs of medium completeness or higher, Read mapping was applied to BBMap v38.51 [44] using 95% minimum alignment identity (Table S2). The virus contigs quality-checked through Check V and Read mapping were subjected to virus taxonomy analysis using Basic local alignment search tool (BLASTn) analysis. BLASTn was performed using the Microbial Genomic Module in the CLC Genomics Workbench settings with the following option: NCBI Viral RefSeq database. The results were sorted according to virus taxon (eukaryote viruses and bacteriophages) and NCLDVs in eukaryote viral communities using the customized taxonomic profiling assignment coding developed by Kim et al. Kim, Jung, Park, Kim, Lee, Ha, and Lee [34] in the open-source software R (version 3.6.3, R Core Team, 2020) within R studio (version 1.3.959).

2.3. Metabarcoding Analysis of Eukaryotic Plankton Communities (EPCs)

Our laboratory’s methodology was employed to perform the metabarcoding analysis of EPCs [45,46]. A 0.2 μm polycarbonate filter (GTTP04700; 47 mm) was used to analyze EPC from 2 L of seawater collected from sample sites consistent with NCLDV samples. Distilled water (50 mL, approximately 50–60 °C) was used to wash the harvested filter three times to remove environmental DNA debris [45]. The filters of each fraction were cut into several pieces using sterilized scissors before gDNA extraction. The gDNA was extracted using the DNeasy PowerSoil Kit (Qiagen, Valencia, CA, USA). Amplicon sequencing was performed using Illumina tagged forward (TAReuk454FWD1) and reverse (TAReukREV3) primers of the 18S rRNA gene (V4–V5 regions) (Table S3). The products from the first PCR were amplified using tags of the Nextera XT 96 index kit v2 (Illumina). The final PCR products were purified using a QIAquick PCR purification kit (Qiagen, Hilden, Germany), and DNA concentration was measured using a Bioanalyzer 2100 (Agilent Technologies, Palo Alto, CA, USA). All samples with equal concentrations of the amplified products were mixed, and the pooled samples were analyzed using a Mi-Seq platform (Illumina).

Bioinformatics analysis from raw data (FASTQ files) using Mi-Seq equipment was performed according to the Bioinformatic process [46] and pre-processed using the Mi-Seq Control Software (MCS) v2.4.1 and FastQC [47]. Further, the quality score distribution per base and poor-quality sequences were flagged. Ambiguous and chimeric reads were removed, and noised sequences (denoising), containing operational taxonomic units (OTUs) with 1, 2, or 3 reads, were removed at a cut-off of 97%. Fast length adjustment of short reads (FLASH) was used to merge paired-end reads [48]. To remove short sequence reads (<150 bp), low-quality sequences with a score < 33, singletons, and non-target sequences, we performed a quality check using BLAST [49]. Subsequently, the sequences were compared with reference sequences from the NCBI database. Based on E-values < 0.001, a similarity score of >97% was used to classify the sequences. The CD-HIT-OTU software [50] was used for clustering and analyzing the OTUs.

2.4. Statistical Interpretation of the Data

To explain relationships between common taxa of NCLDVs and EPCs (each displaying a relative abundance > mean 1%), Spearman’s correlation analysis was performed using SPSS v.12 (SAS Institute Inc., Cary, NC, USA), and the significant correlation coefficient was expressed as a circular flow chart. A cluster analysis was performed in the heatmap to express differences between common OTUs of NCLDVs and EPCs using the Bray–Curtis dissimilarity method and the group average. The circular flow chart and heatmap were plotted using the ‘ggplot2′ package [51] with a combination of custom R and R Studio (v. 1.2.5042). Using the ranked similarity matrix, an ordination plot was produced through non-metric multidimensional scaling (NMDS) using PRIMER 6. Hierarchical agglomerative clustering using the group average method was performed on the most abundant OTUs based on the groups selected from the NMDS analysis. A redundancy analysis (RDA) was performed using the CANOCO program version 4.55 to estimate the common NCLDVs, EPCs, and environmental relationships [52]. All axes were tested for significance using 999 unrestricted Monte Carlo permutations. The extended local similarity analysis (eLSA) was performed using analyzed parameters (common taxa of NCLDVs and EPC, and environmental factors, such as water temperature, salinity, and size-fractionated chlorophyll-a concentration) [53]. The p-value was calculated using permutation testing while ensuring accuracy, and the Q-value was determined to estimate the likelihood of false positives [54]. Network visualization was performed using Cytoscape v3.9.2 [55]. Here, the p- and Q-values were <0.05, and the Pearson correlation coefficients (PCC) were set to >0.5. The time lag was set to zero as the analyzed data did not include time series data [56].

3. Results

3.1. Changes in Environmental Factors in the Sub-Arctic Kongsfjorden

We found the mean salinity and water temperature in April to be 30 and 0 °C, respectively, in their vertical distributions (Figure 2a,b). The chlorophyll-a concentration in April was uniformly distributed at <0.1 μg L⁻¹ across all measured data (Figure 2c). In June, the mean water temperature increased to 2 °C, while that of salinity decreased to 28 in the surface layer, implying that the salinity of the surface layer was affected by sea ice or glacial meltwater. The chlorophyll-a concentration in the surface layer adjacent to the inner bay (St. 1) was relatively high at 1.2 μg L⁻¹, however, at the other sampling sites was <0.3 μg L⁻¹.

3.2. Changes in NCLDVs in the Sub-Arctic Kongsfjorden

Taxonomic analysis of contigs in the genome fragments confirmed their identity as dsDNA viruses. A total of 15,727 DNA virus contigs and 2138 contigs affiliated with NCLDVs were detected via metavirome analysis. More specifically, in April and June, 837 and 967 mean NCLDVs contigs were identified, respectively. The NCLDVs were classified into three groups at 64% similarity using NMDS analysis (Figure 3). Virus(v)Group 1, associated with the ‘NCLDVs in April’, comprised Mimiviridae (31.8%), Poxviridae (25.1%), Phycodnaviridae (14.7%), Pandoraviridae (13.1%), and Iridoviridae (7.4%). The predominant virus was virus Operational Taxonomic Unit (vOTU) #075 (Mimivirus terra2, Mimiviridae) at a relative abundance of 7.11%, followed by Acanthamoeba polyphaga moumouvirus (vOTU#002, Mimiviridae) at 5.99% (Figure 4a and Table S4). vGroup 2 was associated with the ‘NCLDVs in April collected from certain surface layers (St. 1, and 5)’; Phycodnaviridae accounted for 20.5% of the NCLDV population. vGroup 3 was associated with the ‘NCLDVs in June’, wherein Poxviridae, a pre-dominant group, contributed 31.0%, which was higher than that in April. Meanwhile, Pandoraviridae rapidly decreased to 62% compared to April. Meanwhile, in June, most viruses were evenly distributed between the surface and bottom layers.

The NCLDVs in April and June comprised 413 taxa and 437 taxa, respectively, of which, 78.1% (411 taxa) were common to both sampling time points. However, only 28 NCLDV taxa with a mean relative abundance > 1% were common in all samples and overlapped by 10% between April and June (Figure 4b). Of these common NCLDV taxa, 17 were relatively increased in April compared to June, while 11 taxa decreased in June. In particular, Mimivirus terra2 (vOTU#75, Mimiviridae), Pandoravirus neocaledonia (vOTU #102, Pandoraviridae), Pandoravirus macleodensis (vOTU #101, Pandoraviridae), and Cotia virus SPAn232 (vOTU #29, Poxviridae) rapidly decreased in June, while Acanthamoeba polyphaga moumouvirus (vOTU#17, Mimiviridae), Orpheovirus IHUMI-LCC2 (vOTU#90, Phycodnaviridae), Megavirus chiliensis (vOTU#2, Mimiviridae), Chrysochromulina ericina virus (vOTU#26, Mimiviridae), and Emiliania huxleyi virus (vOTU#39, Phycodnaviridae) rapidly increased in June.

3.3. Dominant Eukaryotic Plankton Operational Taxonomic Units

The mNGS results for the EPCs are summarized in Table S5. Sequences and read counts (97% cut-off) were generated across the EPCs at means of 73,258,358 and 174,478 (61,793), respectively. A total of 218,599 (87,698) reads were detected in June, which was approximately twice that in April. Moreover, the eukaryotic protist OTUs (pOTUs) of the EPCs in April and June comprised 448 taxa and 144 taxa, respectively. Of these, 18.9% (94 taxa) of the pOTUs were common to both sampling time points. The unique pOTUs in April and June represented 71.1% (354 taxa) and 10.0% (50 taxa), respectively. In particular, the common pOTUs (with a mean relative abundance > 1% in all samples) overlapped in 30 taxa (60.0%) over the two months (Figure 4b). The EPCs were classified into two groups at 64% similarity using NMDS analysis (Figure 3b). protist(p)Group 1 was associated with the ‘EPC in April’, comprising Dinophyceae (40.7%), Arthropoda (8.7%), Pelagophyceae (8.1%), Chlorophyta (5.9%), and Bacillariophyceae (5.2%), and appeared to be evenly distributed at all sites. Moreover, 11 pOTUs were common species; Gyrodinium helveticum (pOTU#7; 37.2%), Aureococcus anophagefferens (pOTU#0; 8.1%), uncultured copepod (pOTU#3; 5.7%), and Pyramimonas australis (pOTU#5; 3.7%) predominating, with an accumulated proportion of 53.7% (Figure 4a and Table S4). The relative abundance of Aureococcus anophagefferens (pOTU #0) was high in the surface layer of St. 1 (30.7%). Gyrodinium helveticum was the dominant species in April, but not in June. Thus, the relative abundance of Dinophyceae appeared to be the highest in April. pGroup 2 was associated with the ‘EPC in June’, comprising Pelagophyceae (43.2%), Bacillariophyceae (15.3%), Dinophyceae (15.1%), Chlorophyta (11.4%), and Arthropoda (9.8%). Compared to those in April, EPCs in June exhibited a highly variable distribution. Moreover, the abundance of Chlorophyta and Pelagophyceae were 7.6% and 13.5% in the surface layer and 4.30% and 2.64% in the bottom layer, respectively, indicating a relatively high abundance in the surface layer. In this group, Aureococcus anophagefferens was a pre-dominant taxon, contributing 43.24%, which was a rapid increase from that observed in April and was evenly distributed regardless of the layer. Thalassiosira sp. (pOTU#22; 14.6%, Bacillariophyceae), Heterocapsa rotundata (pOTU#26; 11.7%, Dinophyceae), Pyramimonas australis (pOTU#5; 11.1%, Chlorophyta), and Calanus finmarchicus (pOTU#1; 6.9%, Arthropoda) were dominant with an accumulated proportion of 87.5%.

3.4. Potential Association between NCLDVs and the EPC

Bivariate Pearson correlations were performed to assess the correlations between common NCLDVs and EPC lineages with a mean relative abundances of >1%. Using this threshold (p < 0.05), 48 common NCLDV–EPCs lineage pairs were identified with strong significant positive correlation coefficients (r > 0.5, Figure 5, Table S6). The NCLDV lineages were significantly correlated with a known host group of Dinophyceae (r = 0.65, p = 0.01). Among the NCLDVs, five Mimiviridae vOTUs (Mimivirus terra2 [vOTU#75], Megavirus chiliensis, [vOTU#69], Aureococcus anophagefferens virus, [vOTU#12], Cafeteria roenbergensis virus BV-PW1 [vOTU#19], and Chrysochromulina ericina virus [vOTU#26]) was significantly correlated with nine EPC taxa (Aureococcus anophagefferens [pOTU#0], Calanus finmarchicus [pOTU#1], Nephtys pente [pOTU#11], Chaetoceros sp. [pOTU#13], Polarella glacialis [pOTU#15], Gyrodinium helveticum [pOTU#7], Uncultured Prasinophyceae [pOTU#20], Copepoda environmental [pOTU#3], Uncultured eukaryotes [pOTU#6]). One vOTU of Phycodnaviridae (Emiliania huxleyi virus 86 [vOTU#39]) and five vOTUs of Pandoraviridae (Pandoravirus inopinatum [vOTU#100], Pandoravirus macleodensis [vOTU#101], Pandoravirus salinus [vOTU#104], Pandoravirus neocaledonia [vOTU#102], and Pandoravirus dulcis [vOTU#99]) were significantly correlated with two (Thalassiosira sp. [pOTU#13] and Pyramimonas australis [pOTU#5]) and seven EPC pOTUs (Chaetoceros sp. [pOTU#13], Navicula cari [pOTU#9], Gyrodinium helveticum [pOTU#7], Polarella glacialis [pOTU#15], uncultured Prasinophyceae [pOTU#20], uncultured copepod [pOTU#3], and uncultured eukaryote [pOTU#6]), respectively. Two vOTUs of Poxviridae (Melanoplus sanguinipes entomopoxvirus [vOTU#70] and Choristoneura biennis entomopoxvirus [vOTU#240)] were significantly correlated with three pOTUs (Heterocapsa rotundata [pOTU#26], Karlodinium venefium [pOTU#23], and Calanus finmarchicus [pOTU#1]). Other families, including one vOTU of Pithoviridae (Pithovirus sibericum [vOTU#117]) and one vOTU of Asfaviridae (African swine fever virus [vOTU#6]) were each significantly correlated with one pOTU (uncultured Prasinophyceae [pOTU#20] and Oithona similis [pOTU#4]), respectively.

The RDA biplot in Figure 6 and Table S4 was analyzed to explain the relative importance of the variables between the NCLDVs and EPCs (F value: 5.73, p = 0.0001). The NCLDV families-common EPC, including environmental relationships represented in the plot, included six common families and 14 common NCLDV taxa and 21 common variables, including EPC phylum (or class level) and environmental factors that accounted for 43.4% (first axis) and 18.9% (second axis). More specifically, Asfarviridae, Mimiviridae, Pandoraviridae, and Pithoviridae were primarily distributed with salinity in April. In particular, most Pandoraviruses were strongly associated with Gyrodinium helveticum and Dinophyceae. In June, Phycodnaviridae and Poxviridae formed a cluster of sampling sites (Figure 6). Moreover, specific NCLDVs (particularly Poxviridae and Phycodnaviridae) were associated with each size-fractionated chlorophyll-a concentration, temperature, and Pelagophyceae. The two clusters reflect the succession of other phytoplankton communities that respond to the distinguished NCLDVs.

Network analyses between NCLDVs and EPCs revealed specifically associated co-occurrences. The 32 nodes and 57 edges exhibited a significant correlation between NCLDVs, the EPC, and environmental factors (Figure 7, Table S7). A total of six vOTUs (one Mimiviridae, four Pandoraviridae, and one Poxviridae) were highly correlated with four specific pOTUs (two Chlorophyta, one Annelida, and one Dinophyceae) and four environmental factors (salinity, dissolved oxygen, and pico- and nano-sized chlorophyll-a concentrations). More specifically, Pandoravirus inopinatum (vOTU#100, Pandoraviridae) and Pandoravirus salines (vOTU#104, Pandoraviridae) exhibited co-occurrence with Uncultured eukaryotes (pOTU#6) and salinity, respectively (Figure 7). Meanwhile, Mimivirus terra2 (vOTU#075, Mimiviridae) was significantly and positively correlated with Polarella glacialis (pOTU#15, Dinophyceae) and Uncultured Prasinophyceae (pOTU#20, Chlorophyta).

4. Discussion

NCLDVs infect numerous autotrophs, including haptophytes, chlorophytes, and pelagophytes [5,57]. The current study aimed to elucidate how the NCLDV community changes with the increasing diversity and abundance of the EPC. We found that a variety of NCLDVs had co-occurrence with a ubiquitous host or that an NCLDV was significantly correlated with various hosts. The findings of the current study provide novel insights to help elucidate the significant relationships and ecological processes that exist between NCLDVs and EPCs in the context of ecological emergence, and how NCLDVs contribute to ecological functions in the sub-Arctic Kongsfjorden.

The NMDS analysis showed apparent differences in each NCLDV and EPC composition between early white night and mid-summer. These changes occurred according to environmental changes, namely warm water temperature influx and increased light intensity. Endo et al. [9] reported that NCLDV families correlate with specific eukaryotes and infect many eukaryotic lineages. Herein, Mimiviridae viruses were the most commonly detected compared to other NCLDV families, which agrees with the results of a previous study [57]. This is likely due to the diversity of Mimiviridae structure and, a subsequent, broad range of hosts and environments compared to the reference sequences of known viral genomes [58]. In particular, Mimiviridae species are frequently found in low-temperature seawater, such as in the Arctic [59,60], which is likely related to the frequency and types of hosts. Indeed, our results showed that Mimiviridae and Pandoraviridae were significantly and positively correlated with specific EPC taxa, namely, Aureococcus anophagefferens (pOTU#0) increased rapidly in June, as did Megavirus chiliensis (vOTU#69) and Chrysochromulina ericina virus (vOTU#26) also increased. Chrysochromulina ericina virus was predicted to be an abundant viral group as it readily infects microalga (the haptophyceae Haptolina ericina, formerly Chrysochromulina ericina) that are widespread marine plankton species [61,62]. Although no significant correlation was observed among Aureococcus anophagefferens (pOTU#0), Pyramimonas australis (pOTU#5) and Aureococcus anophagefferens virus (vOTU#12), Orpheovirus IHUMI-LCC2 (vOTU#90), and Emiliania huxleyi virus 86 (vOTU#39), as the hosts increased, these viruses also increased, becoming the common taxa in June. Meanwhile, when Gyrodinium helveticum increased rapidly in April, specific Pandoraviuses (Pandoravirus neocaledonia [vOTU#102], Pandoravirus macleodensis [vOTU#101], Pandoravirus dulcis [vOTU#99]) also significantly increased. These viruses showed strong correlations with Gyrodinium helveticum (pOTU#7, Dinophyceae), which is commonly found in cold seawater, such as the Barents Sea, Canadian Arctic, and Greenland Sea [62,63,64]. Moreover, Poxviridae of Choristoneura rosaceana entomopoxvirus ‘L’ (vOTU#151) and Melanoplus sanguinipes entomopoxvirus (vOTU#70) increased with the increase in Thalassiosira sp. (pOTU#22), Heterocapsa rotundata (pOTU#26), and some copepods indicating that NCLDVs are involved in regulating the plankton community. Members of the family Poxviridae can infect a diverse range of vertebrates and invertebrates, although some poxviruses have narrow host ranges and others have very broad host ranges [65]. Copepods are among the most common invertebrates in Arctic sea ice [66]. Although, infected activities between Poxviridae and copepods have been few studies, our results in this study (e.g., the correlation between Calanus finamchicus and members of Poxviridae) will serve as a basis for future infection experiments.

The nano- and pico-sized fractioned chlorophyll-a concentration were also significantly correlated with the NCLDVs, which was consistent with the change in pico-sized Aureococcus and Micromonas. Milligan and Cosper [67] isolated the species-specific infection phage-like virus during a Aureococcus anophagefferens bloom, and Garry, et al. [68] identify its lytic characteristics of Aureococcus anophagefferens virus-1 (AaV-1) and strong infectivity against A. anophagefferens. Vardi, et al. [69] demonstrated that a viral infection controls Emiliania huxleyi blooms in artificial ecosystems. When light intensity and blooming density increase, virus activity also increases, potentially leading to bloom collapse [70]. Hence, in an extended interpretation, Mimiviridae may be an essential regulator of biochemical processes derived from phytoplankton.

Although it is well known that algal species are susceptible to infection by a specific NCLDV [32], recent studies suggest that NCLDVs are regulators of algal populations [60,71]. Here, most NCLDVs were found to be ubiquitously distributed throughout the study area. For example, Mimiviridae, Poxviridae, Phycodnaviridae, and Pandoraviridae accounted for a significant proportion of NCLDVs in the Kongsfjorden ecosystem. In particular, Mimiviridae and Pandoraviridae were abundant and were significantly correlated with pOTUs of Chlorophyta and Dinophyceae. However, other studies have reported that representative viruses exhibit host infection specificity. For example, Heterocapsa circularisquama DNA virus (HcDNAV) and Emiliania huxleyi virus infect the blooms of H. circularisquama and E. huxleyi, respectively [69,72]. In contrast, specific viruses have lineages of widespread host infection [73]. Our results indicate that common NCLDVs may regulate common host groups. The RDA results showed that NCLDVs were broadly associated with EPC. Endo et al. [9] explained that the distribution patterns of NCLDVs highlight their close interplay with various EPCs. The network analysis potentially represents relationships such as viral lysis, grazing, or other interactions. Indeed, some studies have demonstrated a significant association through network analysis between the host–NCLDV [46,74]. In particular, the network results in our study revealed that Pyramimonas australis (pOTU#5, Chlorophyta) significantly co-occurred with Pandoravirus dulcis (vOTU#99, Pandoraviridae), Uncultured Prasinophyceae (pOTU#20, Chlorophyta) co-occurred with Mimivirus terra2 (vOTU#75, Mimiviridae). Meanwhile, previous network analysis revealed that Pandoravirus and the harmful algal blooms, Akashiwo sanguinea (Dinophyceae), exhibit significant co-occurrence [46]. Moreover, P. macleodensis and P. salinus primarily infect amoebas [75]. However, additional studies are needed to evaluate these hosts as substantial evidence has not been provided to conclude whether Pandoraviruses can infect Gyrodinium sp. or whether the host infectivities of Pandoraviruses are diverse because of given the high genetic similarity between Pandoraviruses and Phycodnaviruses [76]. Additionally, in co-occurrence and host prediction analysis performed by Kaneko, et al. [58], specific Mimiviridae viruses were found to be associated with Prymnesiales and Phaeocystales hosts (Haptophyta), while prasinoviruses and putative prymnesioviruses were strongly associated with the Prymnesiales host lineages.

Although the study of NCLDVs in marine ecosystems differs considerably from laboratory-based viral infection and isolation assays, co-occurrence analyses, such as network analysis, will provide important insights regarding inter-relationships in ecosystems. Therefore, we believe that the emergence of any viral community will influence host distribution depending on the infectious activity. Furthermore, we suggest that the viral community structure and infectious activity reflect host distribution [77]. The observation that NCLDV communities were mostly similar in the bottom and surface layers, suggests the presence of vertical NCLDV connectivity, implying that the absence of stratification may connect the surface and bottom NCLDV communities due to low water temperatures. Moreover, surface viruses and energy sources may be transferred by gravitational export to the bottom layers as sinking aggregated organic particles, including phytoplankton cells [78,79]. When organic matter in the surface layer is transported to the bottom layer, it is implied that viruses also interact with their hosts in the lower ecosystem [80]. Indeed, a recent study revealed that various Mimiviridae and Phycodnaviridae viruses could potentially infect their host species in the bottom layers through supplementation with biological carbon pumps [58,81]. A “viral shunt” refers to the lysis, and subsequent transformation, of the host cell into organic matter. According to the viral shunt pathway [82], viruses infect auto- and heterotrophs producing released substrates, such as dissolved organic matter and particulate organic matter, by host cell destruction. Viruses can then be transplanted into the deep layers of the ocean either by adsorption by host cells or infection of host cells [83,84]. Therefore, the vertical connectivity of various NCLDVs may be associated with the infected hosts in the bottom layer via vertical sinking.

5. Conclusions

This study describes the diversity of NCLDVs and EPC during the early white night (April) and mid-summer (June) in the Sub-Arctic Kongsfjorden ecosystem. In addition, the potential host lineages were described including the co-occurrence of various NCLDVs with a ubiquitous host, or the correlation between a single NCLDV and multiple hosts. NCLDVs and EPC populations were found to be similar between surface and bottom layers, however, differed between April and June sampling points. In particular, three Phycodnaviridae, two Poxviridae, three Pandoraviridae, and two Mimiviridae viruses accounted for an essential proportion of NCLDVs diversity. Furthermore, Pandoraviridae and Mimiviridae were strongly associated with Dinophyceae and Chlorophyta hosts, respectively. Additionally, A. anophagefferens (pOTU#0) increased as Chrysochromulina ericina virus (v#OTU26) increased, while Gyrodinium helveticum (pOTU#7) increased as Pandoravirus neocaledonia(v#OTU102), Pandoravirus macleodensis (v#OTU102), and Pandoravirus dulcis(v#OTU101) increased. Although a correlation was not detected between A. aophagefferens and Mimiviridae, this family of viruses may contribute to the regulation of A. ophagefferens blooms as Mimiviridae is abundant in the Sub-Arctic Kongsfjorden ecosystem. Given the wide range of viral host species, not all marine viral hosts are defined. Nevertheless, our findings underscore the importance of NCLDVs as a component of marine microbial communities and refine our knowledge related to marine ecosystems. In conclusion, our findings, which are particularly relevant considering the anticipated impact of biological EPC control mechanisms by NCLDVs on polar regions, can contribute to a better understanding of the Sub-Arctic Kongsfjorden ecosystem. Future studies of both viral and host community structures in the Arctic Seas, particularly from the cold waters near the Arctic Ocean, could provide a more explicit linkage between viruses, host cells, and water masses and currents.