Viruses affect picocyanobacterial abundance and biogeography in the North Pacific Ocean

To explore how environmental gradients shape the distribution of cyanophages and picocyanobacteria, we conducted high-resolution surveys in surface waters along five oceanic transects on three cruises covering thousands of kilometres in the North Pacific Ocean in the spring or early summer of 2015, 2016 and 2017 (Fig. 1a–c). These cruises, two of which were out-and-back, passed through distinct regimes from warm, saline and nutrient-poor waters of the North Pacific Subtropical Gyre to cooler, less saline and nutrient-rich waters of higher latitudes influenced by the subpolar gyre (Fig. 1d–i)²⁷. The shift between the two gyres was marked by abrupt changes in trophic indicators such as particulate carbon concentrations (Fig. 1g) and a chlorophyll front (defined as the 0.2 mg m⁻³ chlorophyll contour²⁸; Fig. 1a–c). As such, the inter-gyre transition zone, defined by salinity and temperature thresholds²⁹ (Fig. 1d), was distinct from both the subtropical and subpolar gyre ecosystems²⁸.

Unexpected Prochlorococcus decline

Prochlorococcus concentrations in the oligotrophic waters of the subtropical gyre were 1.5–3.0 × 10⁵ cells ml⁻¹, comprising an average of approximately 29% of the total bacteria (Extended Data Fig. 1) and numerically dominating the phytoplankton community in all three cruises (Extended Data Fig. 2). Prochlorococcus abundance remained high in the southern region of the transition zone in 2015 and 2016, decreasing precipitously to less than 2,000 cells ml⁻¹ north of the chlorophyll front, generally constituting <1% of the total bacteria (Fig. 2 and Extended Data Figs. 1b,f and 2). This decline occurred at temperatures of about 12 °C (Fig. 2) and is consistent with the thermal limits on Prochlorococcus growth determined for cultures and in numerous field observations^11,12,15,30 (see Supplementary Discussion). Conversely, Synechococcus was 10–100-fold more abundant in the transition zone relative to the subtropics and gradually decreased northwards towards the subpolar waters (Fig. 2 and Extended Data Fig. 2).

The picocyanobacterial abundance patterns differed dramatically in June 2017. The decline in Prochlorococcus occurred approximately two latitudinal degrees south (about 220 km) of the chlorophyll front (Extended Data Figs. 1b,f and 2), where the water temperatures were nearly 18 °C (Fig. 2a). This implicated factors other than temperature as responsible for substantially restricting the geographical distribution of Prochlorococcus. In contrast, the geographical range of Synechococcus was broader by approximately three degrees of latitude (about 330 km), and their integrated abundances across the transects were twofold higher than those observed in 2015 and 2016 (Fig. 2b and Extended Data Fig. 2). Thus, a more southern decline in Prochlorococcus and a broader distribution of Synechococcus was observed in 2017 relative to the 2015 and 2016 transects (Fig. 2 and Extended Data Fig. 2) as well as nine previous transects across this transition zone conducted in different years and seasons^31,32,33 (Supplementary Discussion).

Total bacterial abundances were stable in the subtropics and increased 1.4–3.2-fold in the transition zone on all three cruises (Extended Data Fig. 1a,e). This increase occurred north of the chlorophyll front in 2015 and 2016, whereas abundances increased south of this feature in June 2017. Thus, the 2017 increase in total bacteria happened despite the anomalous loss of Prochlorococcus, which made up only 5% of the total bacteria south of the chlorophyll front relative to 20–30% in the equivalent region in 2015 and 2016 (Extended Data Fig. 1b,f).

A suite of abiotic variables beyond temperature—many of which are considered important determinants for the biogeography of Prochlorococcus^{11,12,13,14,15,16,34}—were assessed for their potential role in restricting the geographical distribution of Prochlorococcus on the 2017 transects (Extended Data Fig. 3). Prochlorococcus populations in 2017 were low compared with previous observations at similar macronutrient levels (phosphate and nitrate + nitrite; Extended Data Fig. 3a,c)¹¹. Micronutrient (iron²⁷) concentrations were within the range for optimal Prochlorococcus growth³⁵ and lead concentrations²⁷ were below levels toxic to Prochlorococcus³⁶. Prochlorococcus populations in 2017 declined despite similar salinity (Extended Data Fig. 3e) and particulate carbon values (Extended Data Fig. 3g) to the 2016 transect; furthermore, Prochlorococcus populations thrived across a wider range of values for these variables in 2015. Prochlorococcus maintained high abundances across a range of mixed-layer depths of up to 100 m in 2015 and 2016 but declined at unusually shallow mixing depths in 2017 (Extended Data Fig. 3i). Collectively, the abiotic conditions observed in 2017 in the region of the Prochlorococcus decline supported large populations of Prochlorococcus in 2015 and 2016 (Extended Data Fig. 3). Thus, none of the physical or chemical factors investigated here can alone explain the unexpected decline in Prochlorococcus in 2017. However, we cannot rule out that a unique combination of these factors, or additional abiotic factors, led to the decline in Prochlorococcus.

A virus hotspot affects picocyanobacterial distributions

The lack of an identifiable abiotic variable differentiating the 2017 transects from the other transects and the overall high abundances in total bacteria for all transects (Extended Data Fig. 1a,e) led us to hypothesize that a mortality factor specific to picocyanobacteria, such as infection by viruses, played a role in precipitating the observed shifts in picocyanobacterial geographical ranges. This was investigated through quantification of the abundances of cyanophages and the extent to which they infected Prochlorococcus and Synechococcus using single-virus^37,38 and single-cell-infection³⁹ polony methods in surface waters of the three different regimes across the cruise transects—that is, the subtropical and subpolar gyres as well as the transition zone between them. We targeted the T7-like clade A and clade B cyanopodoviruses and the T4-like cyanomyoviruses—three major cyanophage lineages based on isolation studies^{18,19,20,24,40}, single-cell genomics⁴¹ and global metagenomic surveys^{25,26,42,43,44} (Supplementary Discussion)—as well as a more recently discovered group, the TIM5-like cyanomyoviruses^42,44. Cyanophages from other lineages that are less common in metagenomic surveys^39,42,43 were not investigated.

In the subtropical gyre, cyanophages were major components of the planktonic virus community, averaging 5.7 ± 1.9 × 10⁵ cyanophages ml⁻¹ and ranging between 0.7 and 21% of the total double-stranded DNA viruses (Fig. 2c and Extended Data Fig. 1d,h,2), consistent with earlier observations for this region^38,39. T4-like cyanophages dominated the subtropical cyanophage community and were generally twofold more abundant than the T7-like clade B cyanophages, the second-most abundant group (Fig. 3 and Extended Data Fig. 4). Together, these two clades constituted >80% of cyanophages measured, with the remainder consisting of T7-like clade A and TIM5-like cyanophages (Fig. 3 and Extended Data Fig. 4). Cyanophage abundances correlated positively with total picocyanobacteria in the subtropical gyre (Pearson’s coefficient of multiple correlation (r) = 0.54, P = 0.02, n = 26; Fig. 2d), suggesting that cyanophages were limited by the availability of susceptible hosts in this region and were not regulating picocyanobacterial populations. On average, less than 1% of the cyanobacterial populations were infected (Fig. 4), with higher infection rates by T4-like cyanophages than T7-like cyanophages (Extended Data Figs. 5 and 6). These instantaneous measurements of infection were used to estimate the daily rates of mortality³⁹ (Methods and Supplementary Discussion), which suggests that 0.5–6% of picocyanobacterial populations were lysed by viruses each day (Extended Data Fig. 7). This implicates other factors, such as grazing⁴⁵, as the major causes of cyanobacterial mortality in the North Pacific Subtropical Gyre.

Within the transition zone we observed a steep latitudinal increase in the abundance of cyanophages for every transect, which we define as a cyanophage hotspot (Fig. 2c and Extended Data Figs. 2 and 4). The cyanophage abundances in this hotspot were between three- and tenfold greater than in the subtropical gyre (Fig. 2c). Notably, cyanophages were approximately 25% more abundant (an increase of approximately 5 × 10⁵ viruses ml⁻¹) in the hotspot on the 2017 cruise relative to the other two cruises, reaching a maximum of 2 × 10⁶ viruses ml⁻¹. The hotspot peaked at temperatures of 15–16 °C on all transects, regardless of the geographical location, season or the exact pattern of the Prochlorococcus and Synechococcus distributions (Fig. 2c). Notably, the numbers of T7-like clade B cyanophages increased sharply in the transition zone to become the most abundant lineage, whereas T4-like cyanophages increased more modestly (Fig. 3 and Extended Data Fig. 4). The change in the cyanophage community structure was particularly pronounced in June 2017, when T7-like cyanophages were up to 2.3-fold more abundant than T4-like cyanophages (Fig. 3c). The switch in the relative abundance of T4-like and T7-like clade B cyanophages was diagnostic of the cyanophage hotspot compared with patterns in the subtropical and subpolar gyres.

To begin assessing whether cyanophages negatively affected cyanobacterial populations in the hotspot, we tested the relationship between the abundance of cyanophages and total cyanobacteria. This showed a significant negative correlation between cyanophage and cyanobacterial abundances across all three cruises (Pearson’s r = −0.56, two-sided P = 0.0005, n = 34). This relationship was particularly distinct in 2017, when cyanobacteria were at their overall lowest abundances and cyanophages at their highest (Pearson’s r = −0.65, two-sided P = 0.004, n = 18). This suggests that viruses are one of the key regulators of picocyanobacteria in the region of the hotspot. However, no significant correlation was found across all regimes and all years (Pearson’s r = −0.008, two-sided P = 0.9, n = 87; Fig. 2d), indicating that factors other than viruses are likely to be more important in regulating the abundances of cyanobacteria in other regimes.

Our single-cell infection measurements allowed us to directly evaluate active viral infection and its impact on picocyanobacteria in the transition zone. Viral infection spiked in this region each year with infection levels that were an average of two- to ninefold higher than those in the subtropical gyre (Fig. 4 and Extended Data Figs. 5,6 and 8). Infection peaked within the temperature range of 12–18 °C and was associated with a concomitant dip in Prochlorococcus abundances in all three cruises (Fig. 4 and Extended Data Fig. 5). These findings provide independent support for the strong negative correlation between cell and virus abundances (Fig. 2d) being the result of virus-induced mortality.

Lineage-specific infection was also distinct in the transition zone relative to the subtropical gyre. Infection by T7-like clade B cyanophages generally increased to reach (2015 and 2016) or exceed (2017) those of T4-like cyanophages (Extended Data Figs. 5 and 6). In addition, the ratio of the abundances of T7-like clade B cyanophages to the number of cells they infected was 2.6-fold greater in the hotspot than the subtropics, whereas this ratio was similar in both regions for T4-like cyanophages. Together, these results indicate that, within the hotspot, the T4-like cyanophages displayed increased levels of infection, whereas the T7-like cyanophages displayed both increased levels of infection and produced more viruses per infection, suggesting that T7-like clade B cyanophages are better adapted to conditions in the transition zone (see below).

Of the three cruises, the highest levels of viral infection were observed in June 2017, with up to 9.5% and 8.9% of Prochlorococcus and Synechococcus infected, respectively (Fig. 4e,f). This dramatic increase in infection mirrored the massive decline in Prochlorococcus abundances (Fig. 4e and Extended Data Fig. 5i). We estimate that viruses killed 10–30% of Prochlorococcus and Synechococcus cells daily at these high instantaneous levels of infection (Extended Data Fig. 6) based on the expected number of infection cycles cyanophages were able to complete at the light and temperature conditions in the transition zone (Methods and Supplementary Discussion). Given that Prochlorococcus is estimated to double every 2.8 ± 0.8 d at the low temperatures in this region¹², we estimate that 21–51% of the population was infected and killed in the interval before cell division. Synechococcus is expected to have faster growth rates at these temperatures, doubling every 1.1 ± 0.2 d (refs. ^12,46). Thus, we estimate that less of the Synechococcus population (9–31%) was killed before division.

Under quasi-steady state conditions, abiotic controls on the growth rate of Prochlorococcus are balanced by mortality due to viral lysis, grazing and other mortality agents^39,45,47. Based on the high levels of virus-mediated mortality, the parallel pattern between Prochlorococcus’ death and viral infection, and the negative correlation between cyanophage and picocyanobacterial abundances in the transition zone, we propose that enhanced viral infection in 2017 disrupted this balance, leading to the unexpected decline in Prochlorococcus populations. Grazing and other mortality agents not investigated here could also have contributed to additional mortality beyond the steady state, resulting in further losses of Prochlorococcus. In contrast to Prochlorococcus, Synechococcus maintained large populations despite high levels of infection (Fig. 4f), presumably due to their faster growth rates enabling them to maintain a positive net growth despite enhanced mortality. These findings suggest that virus-mediated mortality in 2017 was an important factor in limiting the geographic range of Prochlorococcus that resulted in a massive loss of habitat of approximately 550 km.

Cyanophage abundances and infection levels dropped sharply in the higher-latitude waters north of the hotspot (Figs. 2c, 4 and Extended Data Figs. 1d,h and 2). The abundances of both T7-like clade B and T4-like cyanophages declined precipitously, yet T4-like cyanophages were the dominant cyanophage lineage (Fig. 3). T7-like clade A cyanophages generally increased locally at the northern border of the hotspot and became the dominant T7-like lineage in two samples between 38 and 39.2° N in 2017 (Fig. 3c and Extended Data Fig. 4). In contrast to all other cyanophages, the abundances of TIM5-like cyanophages increased in waters north of the hotspot (Fig. 3 and Extended Data Fig. 4d,i,m) but remained a minor component of the cyanophage community. No relationship was found between cyanophage and cyanobacterial abundances (Fig. 2d), and less than 1.5% of picocyanobacteria were infected by all cyanophage lineages in these waters (Fig. 4).

The cyanophage hotspot in the transition zone is a ridge of high virus activity that separates the subtropical and subpolar gyres. The reproducibility of our observations, which were separated by days to weeks within each cruise (2016 and 2017) and by years among the three cruises (Extended Data Fig. 4), indicates that this virus hotspot is a recurrent feature at the boundary of these two major gyres in the North Pacific Ocean. This suggests that the hotspot forms due to the distinctive environment of the inter-gyre transition zone creating conditions that enhance infection of picocyanobacteria and proliferation of cyanophages. Prochlorococcus in the transition zone may be prone to stress due to being close to the limits of their temperature growth range^5,6, which has the potential to increase susceptibility to viral infection. Alternatively, there may be temperature-dependent trade-offs between virus decay and production that lead to replication optima within a narrow temperature range⁴⁸. Cyanophage infectivity has been observed to decay more slowly at colder temperatures⁴⁹, which may allow for the accumulation of infective viruses, leading to increased infection. In addition, cyanophage infections may be more productive due to enhanced nutrient supply in the transition zone²⁷ (Fig. 1h,i) relative to the subtropics, given that the cyanophages replicate in hosts with presumably greater intracellular nutrient quota and obtain more extracellular nutrients, both of which may increase progeny production^9,10. The environmental factors influencing the production and removal of viruses probably vary in intensity at different times, leading to variability in cyanophage abundance and infection levels. Thus, the putative cyanophage replication optimum in the hotspot may reflect the combined effects of temperature and nutrient conditions that are intrinsically linked to the oceanographic forces that shape the transition zone itself.

Changes in the cyanophage community structure over environmental gradients are likely to reflect differences in host range, infection properties and genomic potential to remodel host metabolism⁹. Our data, together with previous measurements in the North Pacific Subtropical Gyre^38,39, indicate that the T4-like cyanophages are the lineage best adapted to the low-nutrient waters of the subtropics (Fig. 2d–f). As these waters are inhabited by hundreds of genomically diverse subpopulations of Prochlorococcus⁵⁰, the broad host range of many T4-like cyanophages^18,19,22,51 may be advantageous for finding a suitable host. T4-like cyanophages also have a large and diverse repertoire of host-derived genes^21,51—such as nutrient acquisition, photosynthesis and carbon-metabolism genes—that augment host metabolism⁵² and may increase fitness in nutrient-poor conditions in the subtropics⁵¹. In contrast, T7-like clade A and B cyanophages seem to be better adapted to conditions in the transition zone (Fig. 3). T7-like cyanophages have narrow host ranges^19,22,40, with smaller genomes and fewer genes to manipulate the host metabolism²³, which may allow them to replicate and produce more progeny in regions with elevated nutrient concentrations relative to subtropical conditions. The maximal abundances of TIM5-like cyanophages were found in the most productive waters at the northern end of the transects where the cyanobacterial abundances were lowest and Synechococcus was the dominant picocyanobacterium. This may be partially due to the narrow host range of TIM5-like cyanophages and their specificity for Synechococcus^40,44. Our findings of reproducible lineage-specific responses to changing ocean regimes indicate that cyanophage lineages occupy distinct ecological niches.

Temperature and nutrient changes occurring in the transition zone are expected to result in shifts in picocyanobacterial diversity at the sub-genus level (Supplementary Discussion), which we speculate may affect community susceptibility to viral infection. One mechanism for this may be that the picocyanobacteria that thrive in the transition zone are intrinsically more susceptible to viral infection. Another scenario may be related to trade-offs associated with the evolution of resistance to viral infection. The horizontal advection of nutrient-rich waters to the transition zone²⁸ may select for rapidly growing cells adapted for efficient resource utilization. Viral resistance in picocyanobacteria often incurs the cost of reduced growth rates^53,54. Thus, competition for nutrients in this region may favour cells with faster growth rates but increased susceptibility to viral infection. Thus, it is probable that the cyanophage distributions do not always follow the cyanobacterial patterns (Extended Data Fig. 2) because of complex interactions between lineage-specific cyanophage traits, host community structure and environmental variables, which may vary seasonally or annually as a result of interannual variability in environmental conditions (see below).

Despite consistent features in cyanophage distributions across the North Pacific Ocean, cyanophage infection was higher (Fig. 4 and Extended Data Fig. 7), whereas Prochlorococcus abundances were consistently lower (Fig. 2a), across the June 2017 transects relative to the March 2015 and April 2016 transects. Seasonality and/or climate variability could explain this interannual variability, although the data currently available to assess this are sparse. Viral infection of picocyanobacteria in the subtropical gyre increased from early spring to summer, suggesting a potential seasonal pattern that may extend across the transect (Extended Data Fig. 9a). In addition, the June 2017 transect occurred during a neutral-to-negative El Niño phase with lower sea-surface temperatures relative to the 2015 and 2016 transects, which were in years of a record marine heatwave, followed by a strong El Niño⁵⁵ (Extended Data Fig. 9b). In 2015 and 2016, the Prochlorococcus abundances were found to be higher than usual in the North Pacific Ocean in this (Fig. 2a) and other studies^56,57. Irrespective of the underlying drivers for the observed interannual variability, we speculate that an ecosystem tipping point was reached in the hotspot under the prevailing conditions in June 2017, aided by the higher cyanophage abundances yet smaller Prochlorococcus population sizes. In this scenario, picocyanobacterial populations were subjected to high infection levels that resulted in an accumulation of cyanophages, initiating a stronger than usual positive-feedback loop between infection and virus production, and precipitating the unexpected Prochlorococcus decline. Continued observations in the North Pacific Ocean are needed to evaluate the potential link between seasonality and/or large-scale climate forcing as ultimate drivers affecting virus–host interactions.

Predicting basin-scale virus dynamics

Measurements of cyanobacterial and cyanophage abundances rely on discrete sample collection from shipboard oceanographic expeditions, which limits the geographical and seasonal extent of available data. Therefore, we developed a multiple regression model based on high-resolution satellite data of temperature and chlorophyll to predict cyanophage abundances, a key proxy of cyanobacterial infection (Pearson’s r = 0.61, two-sided P = 1.7 × 10⁻⁸, degrees of freedom = 68, n = 70). We used the model to estimate the geographical extent of the virus hotspot. The model accurately predicted the location of the hotspot and cyanophage abundances along a fourth transect in April 2019 (Supplementary Table 1), with the majority of observations falling within the 95% confidence intervals of the model predictions (Fig. 5a–c). Application of the model to the larger region predicted that the virus hotspot formed a boundary extending across the North Pacific Ocean, with lower cyanophage abundances on both sides (Fig. 5d,e and Supplementary Fig. 1). This boundary had the hallmarks of the hotspot with a core that was dominated by T7-like cyanophages and the flanking gyre regions dominated by T4-like cyanophages. Thus, this feature may be more appropriately termed a ‘hot-zone’ due to its substantial projected aerial extent. Assuming the infection levels observed in the hotspot in June 2017 were similar throughout the hot-zone, the potential habitat loss for Prochlorococcus would be about 3.2 × 10⁶ km², approximately half of the cumulative area loss of the Amazonian rainforest to date⁵⁸.

Virus hotspot biogeochemistry

With the ability to predict biogeographic patterns of cyanophages, we evaluated the potential biogeochemical implications of virus-mediated picocyanobacterial lysis and release of organic material in sustaining the bacterial community^6,7,8,9. The aerial extent of the hot-zone (approximately 4 × 10⁶ km²) is only 14% of the size of the subtropical gyre (2.9 × 10⁷ km²), and yet the total virus-mediated organic matter released from picocyanobacteria in the hot-zone in June 2017 was estimated to be on par with that for the entire North Pacific Subtropical Gyre (Methods and Supplementary Discussion). We estimate that viral lysate released from picocyanobacteria in the subtropical gyre could sustain 4.4 ± 0.8% of the calculated bacterial carbon demand there (Extended Data Fig. 10). In contrast, viral lysate released in the transition zone could sustain an average of 21 ± 12% of the bacterial carbon demand, reaching 33% in some regions (Extended Data Fig. 10), assuming that the bacterial assimilation and growth efficiencies were similar between the subtropical gyre and the hotspot. Thus, local generation of cyanobacterial viral lysate in the transition zone is likely to be an important source of carbon for the heterotrophic bacterial community that can rapidly utilize large molecular weight dissolved organic matter⁵⁹ and may have contributed to the increase in their abundances south of the chlorophyll front in 2017 (Extended Data Fig. 1a,e).

Source: Ecology - nature.com