SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Viruses saturate the world around us, yet a basic understanding of how viral impacts on microbial host organisms vary over days to hours, which typify the replication cycles of aquatic viruses, remains elusive. Thus, diel patterns of viral production (VP) in Chesapeake Bay surface waters were examined on five sampling dates. Day-to-day variations in VP in the Chesapeake and coastal California surface waters were also investigated. Significant variations in VP were detected over 24 h cycles during four of five studies, but rates did not vary significantly over the course of a few days in either location. Diel patterns of VP displayed seasonality with shorter viral assemblage turnover times and shorter times to maximum viral abundance in summer, implying shorter replication cycles for virus–host systems in warmer months. No correlation was found between VP and time of day, likely due to seasonal changes in the diel patterns of VP. This analysis significantly increases our knowledge of the short-term patterning of in situ VP, and thus viral impacts, and suggests that variations in viral biology in response to changes in host communities or physio-chemical properties affect both diel and seasonal cycles and magnitudes of VP.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Viruses are the most numerous biological entities in the world's oceans, representing about 200 Mt of carbon (Wilhelm and Suttle, 1999; Suttle, 2005) and reaching densities between 104 and 108 ml−1 in marine surface waters (Wommack and Colwell, 2000; Suttle, 2005). Viruses potentially play important roles in microbial food webs as strain- or species-specific parasites of bacterio- and phytoplankton. Indeed, viruses are implicated in the declines and changes in the clonal composition of phytoplankton blooms (Tarutani et al., 2000; Jacquet et al., 2002; Martinez et al., 2007) and bacterioplankton communities (Middelboe et al., 2001; Fuhrman and Schwalbach, 2003; Winter et al., 2004a). Moreover, the process of virus-mediated cell lysis serves as an effective means of creating dissolved organic matter (DOM) and thus limiting the direct transfer of microbial biomass to higher trophic levels (Bratbak et al., 1992; Wilhelm and Suttle, 1999; Middelboe, 2008).

Viruses display a variety of life strategies from lytic to lysogenic replication, but all of these life styles eventually lead to lysis of the host cell and release of viral progeny and DOM into the surrounding environment. Thus, the process of viral lysis is the central means by which viruses exert significant influence on host communities, and an accurate quantification of viral lysis and virus-mediated mortality (VMM) in natural communities is essential to understanding and predicting viral impacts on microbial populations and processes. As no current approaches allow for the direct assessment of VMM of host populations, this parameter is often calculated from measurements of viral production (VP), i.e. the number of viruses produced in a given volume and time span (Steward et al., 1996; Noble and Fuhrman, 2000; Wilhelm et al., 2002).

Viral abundance (VA) can oscillate daily (Wilcox and Fuhrman, 1994; Larsen et al., 2001; Jacquet et al., 2002) and even hourly (Heldal and Bratbak, 1991; Jiang and Paul, 1994; Bratbak et al., 1996) which implies that changes in VP can also occur over short time periods (Heldal and Bratbak, 1991). Likewise, the recognized diel variability in the activity or productivity of bacterioplankton (Shiah, 1999; Winter et al., 2004b; Jugnia et al., 2006) and specific bacterial groups such as Synechococcus (Kondo et al., 1997; Dolan, 1999; Bettarel et al., 2002) suggests that virioplankton production should fluctuate over short time scales. For example, the replication of a dsRNA virus infecting the ubiquitous phytoplankter, Micromonas pusilla, halts when host cultures are deprived of light (Brussaard et al., 2004), indicating that the activity of the virus is influenced by host conditions related to light–dark cycles. Additionally, the knowledge that viruses are susceptible to inactivation by ultra-violet light (Suttle and Chen, 1992; Wommack et al., 1996; Jacquet and Bratbak, 2003) suggests that higher rates of viral lysis and infectivity should occur in the late afternoon and overnight when free viral particles would be exposed to lower levels of UV radiation (Suttle and Chen, 1992; Suttle, 2000; Winter et al., 2004b). However, despite these predictions that VP may vary day to day and hour to hour, limited reports exist of short-term (daily to hourly) variations in VP in marine environments (Bettarel et al., 2002; Winter et al., 2004b; Parada et al., 2008), and none explore such variations in a single environment on numerous occasions. To address this paucity of data and test these theories, diel patterns of VP in Chesapeake Bay surface waters over an annual cycle were recorded as well as patterns in the day-to-day variability of VP in the Chesapeake Bay and coastal California seawater.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Diel variations in viral and bacterial abundances and virus to bacteria ratio

Fluctuations in VA and bacterial abundance (BA) and VP in the Chesapeake Bay over the course of a diel cycle were examined on five sampling dates, winter (February 2007), spring (May 2004), summer (July 2004 and July 2006) and autumn (November 2006). Mean BA did not vary significantly with time for any of the sampling dates (Fig. 1). However, BA did differ significantly between sampling dates/seasons (Table 1). The winter and autumn diel studies had the lowest ambient BA while the highest average BA was observed in summer 2007. Bacterial production was measured during the winter and summer 2007 studies (Table 1), providing a limited data set, which was employed only in correlation analysis (below).

image

Figure 1. Mean viral (white bars) and bacterial (shaded bars) abundances for each diel study. The abscissa represents the time of day experimental incubations were started with 0, 24 and 48 representing midnight on the first, second and third days of the experiment, respectively. Error bars are one standard deviation. Letters denote significant differences (P ≤ 0.05).

Download figure to PowerPoint

Table 1.  Microbial parameters at daily and diel study sampling locations.
Sample date, locationWater Temperature (°C)Salinity (psu)Viral abundance (×1011 viruses l−1)Bacterial abundance (×109 cell l−1)VBRVP (×106 viruses ml−1 h−1)VTT (day−1)Cells lysed (% h−1)aTime to maximum viral abundance (h)Minimum time between similar VP rates (h)Bacterial production (×105 cells ml−1 h−1)
  • a.

    Calculated using a burst size of 50.

  • Data reported as mean (standard deviation).

Diel studies           
 February 20071.9141.8 (0.12)2.8 (0.35)62 (5.9)7.3 (3.6)0.99 (0.44)5.1 (2.2)10 (2.2)80.59 (0.11)
 CB 804           
 May 200417110.77 (0.28)5.8 (3.7)18 (11)3.6 (4.0)1.3 (1.5)1.1 (1.1)8.7 (2.8)12 
 CB 804           
 July 200425 (0.2)8.8 (0.6)1.4 (0.32)5.4 (1.4)26 (5.7)0.31 (2.2)0.13 (0.42)0.27 (0.88)6.8 (2.9)18 
 CB 858           
 July 200726132.9 (0.79)20 (5.9)15 (4.6)4.0 (4.6)0.33 (0.36)0.40 (0.54)8.2 (3.5)182.9 (0.45)
 CB 858           
 November 200614161.8 (0.28)3.2 (0.67)60 (15)10 (3.8)1.3 (0.55)6.3 (2.3)9.0 (2.3)24 
 CB 804           
Daily studies           
 July 200425 (0.2)8.9 (0.5)1.2 (0.38)7.4 (4.1)22 (10)2.3 (1.4)0.47 (0.27)0.75 (0.56)   
 CB 858           
 April 200613 (0.6)34 (0.02)0.76 (0.68)1.6 (0.42)38 (23)8.8 (3.9)4.2 (1.6)12 (6.5)   
 Coastal CA           

With the exception of the summer 2004, VA did not significantly vary over any of the diel cycles (Fig. 1). Like BA, mean VA changed significantly between sampling dates, but with a different seasonal pattern (Table 1). Lowest VA occurred in the spring and summer of 2004 with slight increases in the winter and autumn. Like BA, VA was highest in summer 2007. Summer 2007 VA and BA were twofold and 3.7-fold higher, respectively, than those collected at the same location in summer 2004.

The mean virus to bacteria ratio (VBR) was highest in winter and autumn and like microbial abundance, varied significantly between sample dates (Table 1). Virus to bacteria ratio did not vary between sampling times during any of the diel studies except during the spring. In the spring experiments, VBRs at the 10:00 h and 28:00 h sampling times were significantly higher than those at other sampling times (data not shown).

Diel variations in VP

In contrast to the microbial abundances, significant variations in VP, quantified using the tangential flow diafiltration (TFD) method (Winget et al., 2005), were observed over diel cycles in four of the five studies performed (Fig. 2). Averaged over entire diel cycles, VP rates were highest in winter and autumn with a drop of 50% in spring, and VP was more than 10-fold higher in summer 2007 than summer 2004 (Table 1). Despite these differences in magnitude, average VP rates between seasons were not significantly different with the exception of average autumn and winter VP rates being significantly higher than summer 2004 VP estimates (Table 1). Significant diel variations in VP occurred in each season, with the exception of the summer of 2004 experiments. In addition to changes in VP with time of day during individual diel studies, seasonal changes in the magnitude and diel patterns of VP were also detected (Table 1).

image

Figure 2. Estimates of mean viral production (left panel) and mean percentage of cells lysed per hour (right panel) over diel cycles. The abscissa represents time of day experimental incubations were started with 0, 24, and 48 representing midnight on the first, second and third days of the experiment respectively. Error bars are one standard deviation. Different letters denote significant differences (P ≤ 0.05). * denote negative values.

Download figure to PowerPoint

Close examination of hourly variations in VP within each study revealed higher VP rates for winter experiments initiated during the night and afternoon and at midday with 8–17 h between the highest VP rates (Fig. 2, Table 1). In spring, higher VP rates typically occurred in experiments started in the late afternoon and early morning hours, and the highest VP rates recurred at 12–16 h intervals (Fig. 2). Similarly, VP in the summer 2004 study peaked in the late afternoon/early evening. However, this pattern changed in summer 2007 and autumn with VP peaks occurring in the morning (Fig. 2). A pattern of about 18–24 h between the highest VP estimates was observed for the summer and autumn samples (Table 1).

The time to highest VA within the VP incubation experiments is another measure of short-time scale viral replication cycles. In theory, the shorter the time to highest VA, the more rapid is viral replication and release. The time to reach maximum VA averaged 9.5 h for winter and autumn cruises, and was shorter for spring and summer diel studies, averaging 8 h, indicating more rapid VP and lysis during these months. However, the average time to highest VA within VP incubations were not significantly different between sampling dates with the exception of significantly longer times in winter than summer 2004 (Table 1).

Diel variations in viral turnover time and viral impacts

Viral turnover time (VTT), or the number of times the viral population would be completely replaced each day, provides an additional viewpoint on viral replication cycles. Viral turnover time varied significantly between sampling times during all diel studies except summer 2004 (data not shown). Average VTT varied seasonally with virioplankton assemblages turning over in slightly more than a day during autumn and spring and even shorter turnover times in summer (Table 1).

Calculations of the percentage of BA lysed by viruses (%BA) provide an estimate of viral impacts on host organisms. The average %BA lysed h−1 peaked in winter and autumn with significantly lower %BA lysed in spring and summer (Table 1). Like VP and VTT, the %BA lysed varied significantly between sampling times during all diel studies except in summer 2004 (Fig. 2).

Factors correlating with diel VP

Despite similar times of high and low VP rates between various sampling dates, correlation analysis revealed that the time of day VP experiments were initiated was not significantly related to any other variables across all the diel studies (Table S1) or within individual diel studies (data not shown). Viral production was significantly positively correlated to the time to maximum VA within the experimental incubations (Table S1). Thus, the longer the time to reach the maximum observed VA, the higher the VP estimate. Viral production also showed a significant negative correlation with bacterial production, indicating higher VP when bacterial production is lower (Table S1). When examining the diel studies individually, VP was positively correlated to BA only in winter and negatively correlated with bacterial production in summer 2007 and VBR in spring.

Day-to-day variations in VP and VA and BA

Viral production rates at Chesapeake Bay station CB858 in July 2004 did not vary significantly over the course of four consecutive days (Fig. 3A). Viral abundance did not change, but BAs were significantly higher during the second half of the study (Fig. 3B). The lowest %BA lysed occurred on the day of highest BA (25 July, Fig. 3A), indicating a potential correlation between the two variables. However, although VP and BA both increased on the last 2 days, no significant correlations were found between VP, microbial abundances and the %BA lysed.

image

Figure 3. Daily variations in viral production, percentage of cells lysed and microbial abundance at Chesapeake Bay station CB 858, July 2004. A. Mean viral production rates (white bars) and percentages of bacterial abundance lysed h−1 (shaded bars). B. Mean ambient viral (white bars) and bacterial (shaded bars) abundances. Error bars are one standard deviation. Bars denoted with different letters are significantly different (P ≤ 0.05).

Download figure to PowerPoint

Viral production estimates and the %BA lysed were higher in coastal California surface waters than in the Chesapeake Bay (Table 1, Fig. 4A). In contrast, VAs and BAs were lower than in the Chesapeake Bay (Table 1). While VA did not change over the 4 day sampling period, BA declined steadily (Fig. 4B). The %BA lysed h−1 also varied significantly in this study with the lowest percentage of cells lysed occurring on 19 April , the day of lowest VP (Fig. 4A). However, as in the Chesapeake Bay study, no significant correlations were found between VP, microbial abundances and the %BA lysed.

image

Figure 4. Daily variations in viral production, percentage of cells lysed, and microbial abundance in coastal California, April 2006. A. Mean viral production rates (white bars) and percentages of bacterial abundance lysed h−1 (shaded bars). B. Mean ambient viral (white bars) and bacterial (shaded bars) abundances. Error bars are one standard deviation. Bars denoted with different letters are significantly different (P ≤ 0.05).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

VP estimates

Viral production rates are the principal measures used in estimating viral impacts on bacterioplankton communities (i.e. VMM and the %BA lysed h−1) and the amounts of C, N and other cellular constituents recycled by viruses into the DOM pool (Weinbauer, 2004). Understanding viral impacts on marine food webs is vital for accurately modelling C fluxes in the marine environment and variations in phytoplankton and BA, production and community composition. This study represents the first examination of day-to-day fluctuations in VP and of seasonal influences on diel VP cycles.

Average VP estimates in the Chesapeake Bay and coastal California using the TFD approach ranged between 0.31 and 10 × 106 viruses ml−1 h−1, and were within the scope of previously reported values for the Chesapeake Bay and similar coastal environments (Noble and Fuhrman, 2000; Wilhelm et al., 2002; Poorvin et al., 2004; Winter et al., 2004b; Winget et al., 2005). Likewise, microbial abundances were similar to previous observations in the Chesapeake (Wommack et al., 1992; Winget et al., 2005; Kan et al., 2007). While BAs in this study mirrored earlier reports for California waters, VAs were about a half a log higher than previous reports (Fuhrman and Noble, 1995; Noble and Fuhrman, 1999; 2000).

Application of the TFD technique to measure VP and estimate VMM is founded on a few notable assumptions and caveats. First, VP within the experimental incubations is assumed to be representative of production in the natural environment. Because over half the bacterial community was retained after TFD, it is likely that the observed VP within experimental incubations is similar to the natural rate, and a correction factor was applied to account for bacterial loss during sample processing (Table S2).

This method also assumes that VP results only from host cells infected at the time of sampling and not from reinfection and lysis events during incubation. The removal of free viruses during TFD prior to incubation greatly reduces the chance of virus–host contacts and new infections (Weinbauer et al., 2002; Winter et al., 2004b), and incubation times were intentionally limited to 12 h to reduce containment effects and reinfection opportunities (Wilhelm et al., 2002). However, VA often dropped and recovered during the incubation periods, indicating potential reinfection events. As observed previously (Winter et al., 2004b), peaks in VA within a collection of incubation bottles often occurred at the same time of day despite initiation of the experiments at different times of day (data not shown). Thus, peaks in VA likely reflect lysis of host cells by viruses with specific latent periods or infection times rather than new rounds of infection and lysis. Other methods of measuring VP (e.g. FVIC) may have allowed for more instantaneous sampling. However, the reliance of such methods on several poorly constrained conversion factors for the calculation of VP (Binder, 1999; Weinbauer et al., 2002), as well as the lack of direct observation of VP, made these approaches unsuited to the goals of this study.

Third, conversion of VP estimates to virus-mediated mortality relies on an assumed burst size of 50 viral particles released per cell lysed. This burst size has been widely used in previous VP studies (Heldal and Bratbak, 1991; Steward et al., 1996; Noble and Fuhrman, 2000; Wilhelm et al., 2002). However, recent research on cyanophage isolates from the Chesapeake Bay found an average burst size of 83 (Wang and Chen, 2008), indicating that a burst size of 50 may overestimate VMM. Regardless, the diel and daily trends in VMM will remain the same so long as a constant burst size is applied for each experiment.

Diel variability in VP

Few studies have previously examined diel variation in VP or viral impacts on host abundance, and none have compared diel patterns of VP rates between seasons. In addition to diel variations in VP, VTT and the %BA lysed (Fig. 2), VP, the %BA lysed and the time interval between similar VP rates all generally increase in winter and autumn compared with spring and summer. The variability of VP on both short (hourly) and longer time scales (seasonal and yearly) are intriguing in light of known seasonal variations in abiotic factors and host communities in the Chesapeake Bay (discussed below). The interaction of these two time scales likely confounded one another, and contributed to the observed lack of a correlation between VP and time of day.

Viral abundance and BA were not significantly correlated with VP rates across all or individual diel and daily studies, except in winter. Significant variations in VP were often observed without corresponding changes in VA or BA (Figs 1 and 2) and vice versa (Figs 3 and 4). Therefore, total abundance does not appear to be a very influential factor in VP rates in this study. Still, the significantly higher VBRs and %BA lysed in winter and autumn and the positive correlation between these parameters suggest that changes in the ratio of viruses to hosts affect relative mortality rates, perhaps due to increased virus–host contacts at higher VBRs.

Viral production has been hypothesized to decrease in the afternoon due to either the direct degradation of viruses and/or loss of infectivity; or the preferential infection of cells to avoid UV damage and utilize light-induced DNA repair systems within cells (Jacquet et al., 2002; Winter et al., 2004b). This mid-day infection scenario would lead to increases in the %BA lysed h−1 and VP during overnight hours as viral particles would be released when the risk of UV damage is minimal. While this scenario was true for the winter and autumn diel studies, the spring and summer studies ran counter to this hypothesis with generally higher values of VP during the day, suggesting that this hypothesis may only be valid for winter and autumn seasons in the Chesapeake.

Like this study, a previous study in the North Sea observed significant diel fluctuations in VP, but an increasing frequency of visibly infected cells – a proxy measure of VMM – occurred during the night (Winter et al., 2004b). In contrast, estimates of VMM and %BA lysed were not consistently higher during night-time sampling points (Fig. 2) for any of our Chesapeake experiments. Interestingly, Bratbak and colleagues (1992) observed higher percentages of phage-producing bacteria between about noon and 18:00 h, similar to our higher %BA lysed in the afternoon.

The large interannual difference in the two summer samplings (2004 versus 2007) reflects a general trend of low VP rates in 2004 across the Chesapeake Bay and very high rates in 2006–2007 (D. Winget, unpublished). However, despite differences in the absolute values of VP between the two summer samplings, the %BA lysed, VTT and the time to maximum VA during both summer studies were similar (Table 1), indicating recurring summer trends in diel VP. These comparisons reveal that the diel cycles of VP and VMM vary between geographic locations as well as between seasons and years.

Day-to-day variability in VP

Viral production rates at a single location recorded at similar times of day over a 4–6 day period did not vary (Figs. 3A and 4A). This result was observed in two marine environments that differed substantially in absolute values of VP, VA and BA. Moreover, VP estimates repeated at the same time of day on consecutive days during the diel studies also showed no daily change in VP (Fig. 2).

Similar to these findings for the Chesapeake Bay and coastal California, only minor variations in VA were observed on consecutive days in surface water samples from the North Sea (Winter et al., 2004b). Viral production estimates on successive days in a mesocosm of un-manipulated water also did not vary (Fuhrman and Noble, 1995). In contrast to the stability in VA and VP seen in these studies, significant day-to-day changes in both of these parameters occurred in a Norwegian bay (Larsen et al., 2001) and the Mediterranean Sea (Guixa-Boixereu et al., 1999). Water samples in these two experiments were manipulated by nutrient additions (Guixa-Boixereu et al., 1999; Larsen et al., 2001), which certainly induced changes in the productivity or composition of host community that would not necessarily have occurred in un-enriched seawater used in this study.

Mechanisms of diel, daily and seasonal VP variability

The observed patterns of diel, daily and seasonal variations in VP could result from several processes, including: (i) changes in virus–host contact and infection rates, (ii) changes in viral diversity and (iii) changes in viral characteristics related to changes in host growth rates and activity. These processes are not mutually exclusive and likely work in tandem to produce the observed patterns of diel, daily, seasonal and interannual variations in the bulk rates of VP, which represent the sum of many individual virus–host interactions. Ultimately, changes in viral assemblage diversity or the activity of viral strains over a diel cycle and from season to season are linked to seasonal and diel changes in host abundance, diversity and activity which are in turn influenced by environmental parameters.

For example, alterations in the biological features of virioplankton assemblages or populations, such as latent period or burst size, change with variations in host physiology, and in turn alter the magnitude and patterns of VP. In particular, viral latent periods decrease as host density (Abedon et al., 2001) and productivity increase (Schrader et al., 1997). Thus, a shorter average latent period for viruses in summer and spring when BA and production are highest, likely produced the shortened times to maximum VA within experimental incubations and times between similar VP rates in spring and summer and contributed to the seasonal variations in diel cycles of VP.

Burst sizes of some bacteriophages increase with host growth rates (Schrader et al., 1997; Middelboe, 2000), and in natural bacterioplankton communities, viral burst sizes increase with temperature (Mathias et al., 1995) and bacterial production rates (Parada et al., 2006). Thus, higher bacterial production should be correlated with higher burst size and VP. This was not observed in this study as the highest magnitudes of VP occurred in cooler months, which were typified by lower bacterial production rates, and bacterial production and VP were negatively correlated. This negative relationship is likely an artefact of sampling for the winter and autumn studies during 2006–2007 when VP rates were significantly higher than during other sampling years (D. Winget, unpublished).

In addition to changes in host physiology with sampling date, BA and cyanobacterial abundance vary seasonally in the Chesapeake Bay (Shiah and Ducklow, 1994; Wang and Chen, 2004) along with the composition and relative proportions of the major bacterial clades (Kan et al., 2007) and bacterial production (this study; Shiah and Ducklow, 1994). Thus, as temperature and nutrient levels change, host abundances and productivity vary, likely leading to changes in both viral assemblage diversity as well as the productivity of individual viral strains (Schrader et al., 1997; Wang and Chen, 2004). Indeed, the composition of the whole virioplankton and cyanophage assemblages do change monthly in the Chesapeake Bay (Wommack et al., 1999a,b; Wang and Chen, 2004) with higher abundances of cyanophages in summer (Wang and Chen, 2004). Cyanophages that carry functional photosystem II genes (Lindell et al., 2005) may find it advantageous to complete their replication and lysis cycles during daylight hours rather than at night – a timing that corresponds with our observations of higher daytime VP rates during summer and previous observations in culture of higher cyanophage production during light periods (Sherman, 1976). The diel pattern of VP then changes in winter and autumn when cyanophages and other algal viruses no longer dominate the viral assemblage, producing the observed night-time peaks in VP in these seasons. Thus, different virus–host pairs with different diel cycles of productivity and lysis occur on different sampling dates, producing the observed seasonal variations in diel patterns of VP and virus-mediated mortality.

In contrast, consistency in the day-to-day estimates of VP could result from stability in the species composition and physiological activity of viral and host assemblages over the weekly time scale. The lack of short-term changes in environmental and host conditions at a single location, such as during the daily VP experiments, most likely produced stable day-to-day levels of viral diversity, abundance and production.

These explanations all imply that one, or a few, dominant host populations synchronously release viruses throughout the course of a day, producing the observed dips and peaks in VP. In natural environments, not all virus–host systems exhibit cell cycle-dependent VP (Thyrhaug et al., 2002; Juneau et al., 2003) or changes in burst size or latent period length with changes in host abundance and quality (Schrader et al., 1997). However, coinciding peaks in VA at a particular time of day were often observed in experiments with overlapping incubation periods, but different start times (data not shown; Winter et al., 2004b). Thus, the synchronized lysis of host populations that were infected by viruses with a variety of latent periods, which vary seasonally with the types of viruses and hosts present or host physiology, seems to be the most likely cause of the observed diel oscillations in VP. Prophage induction events could also serve to synchronize the release of viral particles. Intriguingly, Bratbak and colleagues (1996) predicted that marine phage–host systems should oscillate at about 15 h intervals based on analysis of Lotka–Volterra equations. This nearly matches the 18 h periodicity of VP in the Chesapeake Bay in summer, suggesting that diel fluctuations in viral activity are functions of synchronous predator–prey interactions.

In conclusion, diel changes in VP indicate that estimates of virus-mediated bacterioplankton mortality and DOM release also vary during the day and thus single-time point measurements must be carefully interpreted. New techniques to simultaneously monitor changes in the production of several un-related virus–host pairs are needed to clarify the mechanisms behind the stability of aggregate VP over daily time scales and to verify the underlying mechanisms behind diel, seasonal and yearly variations in VP rates.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Sampling locations

Diel changes in VP were recorded at Chesapeake Bay station CB804 (38°04′N, 76°13′W) in May 2004 (spring), November 2006 (autumn) and February 2007 (winter) and at station CB858 (38°58′N, 76°23′W) in July 2004 and July 2007 (summer). For all diel studies except July 2004, a 50 l surface water sample from 1 m depth was collected from 10 l Niskin bottles mounted on a conductivity-temperature-depth rosette (CTD) (Seabird Electronics 911 Plus CTD) and transferred into a triple-rinsed polypropylene carboy. The carboy was then placed in an on-board flow-through incubation table. A 2 l sample for estimation of VP was siphoned from the carboy every 4 h for 24 h. During the July 2004 study, 2 l of surface water for VP assays was freshly collected via bucket every 6 h for 24 h from the sample location. From the 2 l sample, a 4.5 ml unfiltered, ambient water subsample was collected, preserved by addition of formaldehyde (1% final concentration) and snap-frozen in liquid nitrogen. The remaining surface water was filtered through a 50 μm Nitex mesh screen and stored in the dark at ambient temperature in an on-board flow-through incubation table until processing for VP.

Daily studies of variations in VP were performed at Chesapeake Bay station CB858 from 25 to 28 July 2004 and at a coastal California site (34°16′N, 119°55′W) from 17 to 22 April 2006. Two litres of sea water from approximately 1 m depth was collected each day during the sampling period within 1–2 h of the same time of day via bucket or 10 l Niskin bottles mounted on a CTD. Samples were then treated as described above.

Tangential flow diafiltration

Samples for VP measurements were processed by the TFD method (Winget et al., 2005). Briefly, 300 ml of 50 μm pre-filtered ambient water was filtered through a 0.22 μm Pellicon XL (Millipore) tangential flow cartridge at a rate of 40 ± 2 ml min−1. Water volume lost through filtration was replaced with 30 kDa filtered virus-free water collected at the same time and location as the original sample (Wang and Chen, 2004). Diafiltration continued until 4× the original sample volume (1200 ml) had been collected from the permeate port. An average of 56% of the bacterial community was retained after TFD.

The 300 ml diafiltered sample was split into triplicate 100 ml subsamples and incubated in the dark at ambient temperature in 250 ml polycarbonate bottles. Incubations were subsampled for VA and BA immediately after aliquoting and at 3 h intervals for 12 h as described above. All samples for viral and bacterial enumeration were snap-frozen in liquid nitrogen and stored at −70°C until analysis.

Viral and bacterial enumeration

Viruses and bacteria were enumerated via epifluorescence microscopy as described elsewhere (Helton et al., 2005; Winget et al., 2005) after staining with SYBR Gold (Molecular Probes) diluted to 2.5× the commercial stock. Viruses and bacteria were visualized at 1000× magnification using an Olympus BX61 microscope (100× UPlanFI oil objective) under FITC excitation. Fields of view were photographed using a Q Imaging Retiga EX-i digital camera. For each sample, at least 200 virus-like particles were enumerated using IPLab software (Biovision v3.4r4), which discriminated between virus particles and bacterial cells based on a combination of size and staining intensity.

Viral and bacterial production

Viral production rates were estimated from the slope of the first-order regression line of VA versus incubation time (Wilhelm et al., 2002) using Prism (GraphPad Software, v. 4.0). The slopes of replicate incubations were averaged and corrected for bacterial loss during diafiltration (Table S2) (Wilhelm et al., 2002; Winget et al., 2005). The VMM, the percentage of cells lysed h−1, VTT and VBR were calculated as reported in Table S2 (Wilhelm et al., 2002; Winget et al., 2005). The time to maximum VA within each incubation was noted as well as the time interval between statistically similar peaks in VP rates across a given diel series. Fifty-millilitre samples for bacterial production were collected synchronously with samples for VP during the February and July 2007 diel studies. Immediately after sample collection, bacterial production was assessed via the microcentrifuge 3H leucine method (Kirchman, 2001).

Statistical analyses

Statistical comparisons were performed in spss (SPSS, v 11.0.0). Outliers in triplicate incubations for VP estimation, defined as greater or less than two standard deviations of the mean of the other two replicates, were removed prior to statistical analysis. Log 10 transformations were performed as necessary to achieve normal distribution. Significant variations were identified by one-way anova followed by Tukey's HSD post hoc tests or the Kruskal–Wallis H-test as appropriate. Differences between the two summer studies were detected by Student's t-tests. For the day-to-day studies, two-tailed Pearson's correlations were employed while two-tailed Spearman's rank correlations were used in the diel studies. For all statistical analyses, the significance level was set at P ≤ 0.05.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

This work was supported by NSF grants MCB-0132070 and OCE-0221825 awarded to K.E.W. and a NSF Graduate Research Fellowship and the Preston C. Townsend Biotechnology Fellowship awarded to D.M.W. The authors would like to thank the crews of the R/V Cape Henlopen and the R/V Point Sur and M. Simon, T. Mills, S. Srinivasiah, S. Bench and Drs K. E. Williamson, R. R. Helton, D. Bronk, K. Wang and D. W. Coats for countless hours of tireless assistance.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Table S1. Spearman's rank correlations between microbial parameters in diel viral production studies.

Table S2. Formulas for calculations of viral production, virus-mediated mortality, percentage of bacterial lysed by viruses and viral turnover time.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
EMI_2038_sm_Tab.S1-S2.doc1067KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.