Microb Ecol (1995) 30:25-41

MICROBIAL ECOLOGYInc. © 1995Springer-Verlag New York

Diel, Seasonal, and Depth-Related Variability of Viruses and Dissolved DNA in the Northern Adriatic Sea M.G. Weinbauer, ~ D. Fuks] S. Puskaric, 2 R Peduzzi 1 ~Institute of Zoology, University of Vienna, Althanstrasse 14, A-1090, Vienna, Austria 2Rudjer Boskovic Institute, Center for Marine Research Rovinj, 52210, Rovinj, Croatia Received: 24 June 1994; Revised: 31 October 1994

Abstract. The depth-dependent, seasonal, and diel variability of virus numbers, dissolved DNA (D-DNA), and other microbial parameters was investigated in the northern Adriatic Sea. During periods of water stratification, we found higher virus abundances and virus/bacterium ratios (VBRs) as well as a larger variability of D-DNA concentrations at the thermocline, probably as a result of higher microbial biomass. At the two investigated stations, virus densities were highest in summer and autumn (up to 9.5 × 101° 1-l) and lowest in winter (< 109 1 1); D-DNA concentrations were highest in summer and lowest in winter. The VBR as well as an estimated proportion of viral DNA on total D-DNA showed a strong seasonal variability. VBR averaged 15.0 (range, 0.9-89.1), and the percentage of viral DNA in total D-DNA averaged 18.3% (range, 0.1-96.1%). An estimation of the percentage of bacteria lysed by viruses, based on 2-h sample intervals in situ, ranged from 39.6 to 212.2% d -l in 5 m and from 19.9 to 157.2% d -~ in 22 m. The estimated contribution of virus-mediated bacterial DNA release to the D-DNA pool ranged from 32.9 to 161% d -~ in 5 m and from 10.3 to 74.2% d -~ in 22 m. Multiple regression analysis and the diel dynamics of microbial parameters indicate that viral lysis occasionally could be more important in regulating bacterial abundances than grazing by heterotrophic nanoflagellates.

Introduction Recently, it has been demonstrated that virus numbers are high in aquatic systems, usually even exceeding bacterial densities [3]. Furthermore, it has been suggested that viral lysis might be an important cause for the mortality of both bacteria and phytoplankton [3, 6, 34, 40]. Moreover, viruses might play an important role in the dynamics of organic particles [31, 33]. Cell content and cell debris is set free because of viral lysis of the hosts; however, the significance of this mechanism

Correspondence to: M.G. Weinbauer

26

M.G. Weinbauer et al.

for the production of dissolved organic matter remains to be investigated. Attempts have been made to measure viral production and decay [18, 38, 41] and to include viruses into a budget of the carbon transfer within the microbial loop [8]. Attempts have also been made to determine the mechanisms that are responsible for the destruction of viruses and their infectivity as well as the removal of viruses from the water column, including mechanisms such as solar radiation, adsorption to particles, or grazing by heterotrophic nanoflagellates (HNFs) [16, 41]. There are some recent reports on the spatial distribution of viruses in the sea [4, 11, 30, 43, 46]. Although it has been shown that viruses are dynamic partners in microbial food webs [7], the actual role of viruses remains largely unknown. Considerable information has been accumulated on the temporal as well as spatial distribution of dissolved DNA (D-DNA) [12, 22, 27, 36]. D-DNA is actively released by bacteria and phytoplankton [28] and produced by grazing of HNFs on bacteria [42]. Viral lysis of bacteria might release an important amount of dissolved nucleic acids [33, 43]. Dissolved nucleic acids could be an important source of phosphorous in aquatic systems, but dissolved RNA, for example, has been only poorly studied [22, 36]. Because most viruses are smaller than 0.2 Ixm, they can be considered as part of the pool of D-DNA. The estimated contribution of viral DNA to the total D-DNA pool varies considerably [2, 4, 30, 43]. Although there is some information on the seasonal variability of virus abundances [43, 47], in only one study was a systematic sampling performed [21]; viral dynamics on the time scale of hours is known only from mesocosm studies [8, 18, 21]. We investigated the variability of virus numbers in the shallow waters of the northern Adriatic Sea in relation to season, daytime, and water depth, as well as the interdependance of viruses with D-DNA and other microbial parameters. Based on our data, we suggest that viral lysis might be an important source of D-DNA and an important cause for bacterial mortality. Materials and Methods

Sample Collection Surface water samples ( - 0 . 5 m) as well as water samples for selected depth profiles were collected off the Laboratorio di Biologia Marina in Aurisina (Italy) and the Center for Marine Research in Rovinj (Croatia) during different seasons between 1991 and 1993 (Fig. 1). Additionally, during waterstratification periods, water samples were collected directly from the thermocline layer by scuba diving. The exact depth of the thermocline was determined by the use of a thermometer. Water depth at the sampling sites was about 15 m in Aurisina and 20 m in Rovinj. Seasonal water samples (surface water) were taken in 1-3 monthly intervals; between 2 and 7 samples (collected in a period of 2-10 days) were analyzed per monthly sampling, except in August 1992, when only one sample was analysed. A diel study was conducted aboard the R/V "Vila Velebita" at a station on the Rovinj-Po river transect (Fig. 1). To follow a defined water mass during this diel cycle, we used a free-floating buoy, almost entirely submerged, supported by a 2-kg weight and a 160-liter plastic bag filled with water to increase the surface and mass of the buoy. With this setup, the drifting buoy was not affected by wind forces. Water samples were taken from June 24 to 26, 1992, within the thermocline layer (approximately 5 m) and from June 24 to 25 also below the thermocline (22-m depth) at 2-h intervals. The depth of the thermocline was determined by using CTD hydrocasts (10 SBE 25 Sealogger CTD; Sea-Bird Electronics Inc., Bellevue, WA). All water samples in this study were collected either by using acid-rinsed 5-liter Niskin bottles or by scuba divers using acid-rinsed 500-ml syringes. Water samples for determination of viral, bacterial, and HNF densities were preserved immediately

Marine Viruses and Dissolved DNA

27 14" 00"

13"00" ...-.v:-:::!

.~.:-...:..

,.. :':. . -....:.-.

:v...'. ":

45* 31

:

45* (

..,..

....;-..:

o 25km Fig. 1. Location of sampling sites in the northern Adriatic Sea. Circles show the locations of depth profile samples and of the seasonal studies; the triangle indicates the location of the diel cycle study.

after collection with 0.2 i~m-filtered formalin (viruses 2%, bacteria 4%, and HNFs 1% final concentration) and stored at +4°C in the dark until analysis. Chlorophyll a (Chl a) was measured by standard oceanographic methods described in Parsons et al. [25].

Electron Microscopy Virus particle abundance in different depths and during the seasonal study was determined using ultracentrifugation and electron microscopy as described by Mathews and Buthala [24], BCrsheim et al. [5], and Bratbak et ai. [7]. For details, see Weinbauer et al. [43].

Epifluorescence Microscopy Viral and bacterial counts of samples of the diel cycle were obtained by the use of 4,6-diamidino-2phenylindole (DAPI) staining and epifluorescence microscopy, by a modification of the method described by Hara et al. [17]. 2-ml samples were stained with (DAPI) (5 ~g m1-1 final solution) for 15 min and filtered onto Irgalan black stained 0.02-Fm pore-size Anodisc filters. This filter type

28

M.G. Weinbauer et al.

allows a more rapid filtration (usually < 20 rain) compared to the 1- to 1.5-h duration through 0.015txm pore-size Nuclepore polycarbonate filters, as described by Hara et al. [17]. Viruses and bacteria were counted from photographs inspected under a dissecting lens (10×) and identified on the basis of size and intensity of fluorescence. Bacterial numbers along the depth profiles and during the seasonal study were determined by the acridine orange direct counting technique [19]. HNF densities of all samples were obtained by applying Caron's [9] primulin staining technique. Autotrophic nanoflagellates were distinguished from heterotrophs (HNFs) based on the autofluorescence of Chl a [9]. Bacterial and HNF samples were usually processed within 2 weeks after sampling.

Dissolved Nucleic Acids

Seawater samples (300-900 ml) were prefiltered through a 60-1xm net to remove larger particles. Subsequently, the prefiltered seawater was filtered through precombusted (450°C for 4 h) 47-mm diameter Whatman GF/F filters (nominal pore size: 0.7 ixm) and stored frozen in glass flasks until analysis. In samples covering a variability of trophic conditions and D-DNA concentrations typical for the northern Adriatic Sea, the filtration through smaller pore-size filters (i.e. 0.2-p~m polycarbonate filters), reduced D-DNA concentrations on average only by 5.4%, compared to GF/F filters [45]. GF/ F filters were used in the present study, because they can be precombusted and allow rapid filtration of water samples. Thawed samples were diluted with Milli Q water to a final concentration of 25% seawater, and dissolved nucleic acids were determined by the cetyltrimethylammonium bromide-3,5diaminobenzoic acid (CTAB-DABA) method described in Karl and Bailiff [22]. The recovery efficiency was determined by internal standards of DNA (calf thymus DNA, single-stranded, Sigma Chemical Co., St. Louis, Mo., D-8899) [22] and was found to be always >90%. The preparation of the external standards of DNA was slightly modified compared to Karl and Bailiff [22]; in the present study, standard DNA were also precipitated with CTAB, to treat the external standard in the same way as the natural water samples. Standard curves obtained in this way differ slightly from those determined without CTAB precipitation, especially at low concentrations of dissolved nucleic acids or low sample volumes in [45].

Calculations

From the diel dynamics of the viral and bacterial abundances, we estimated the contribution of viral lysis to the mortality of bacteria by using the calculations described by Jiang and Paul [21]. We used the lowest and highest decay rate measured by plaque assay (0.0042 and 0.096 h ') and the lowest and highest decay rate determined by direct counting methods (0.3 and 1.1 h ') as suggested by Jiang and Paul [21]. Using the virus abundances at any one sample time (Vn), the virus abundance at the next sample time without new phage production (V'n+0 was calculated for any of the decay rates by the following equation R = (lnV, - lnV',+I)/(T,+, - T,)

(1)

where R = viral decay rate, T, = time when the nth sample was taken, T,+, = time when the ( n + l ) t h sample was taken, V, = viral abundance at time T,, and V',+I = estimated viral abundance without new phage production at time T,+,. The new production of viruses from lysis of bacteria is the difference between the virus abundances at time T,+, and the abundance of V',+I: V=w = V.+1 - V'.+,

(2)

where V,ew = new viral production, and Vn+, is the viral abundance at time Tn+,. This calculation assumes that the virus community consists of lytic bacteriophages. Bacterial mortality that was due to viral lysis was calculated by the equation

B% = [(V,JBZ)/Bo] x 100

(3)

where B% = percentage of bacteria killed by viruses, B Z = viral burst size and Bn = bacterial densities

Marine Viruses and Dissolved DNA

29

at time T,. The mean burst size of about 50 found in the northern Adriatic Sea [44] was assumed to represent the viral burst size during the diel cycle. A value of V,ew smaller than 0 means that no bacteria were killed by viral lysis. From the percentage of bacteria lysed at one sample time, we estimated the contribution of DNA release by viral lysis of bacteria to the pool of D-DNA at the next sample time. Because bacteria are comparatively large in the northern Adriatic Sea [42] and the bacterial DNA content increases with cell size [10], we assumed a high bacterial DNA content of 9.8 fg per cell [26] to be reasonable for the investigated area. We assumed further that viral DNA is completely derived from bacterial host nucleic acids [46].

Statistical Analysis Because not all data sets for the correlation and multiple regression analyses followed normal distribution, we applied log transformation to meet the requirements for parametric statistics. The variation of parameters between the two depths during the diel cycle was tested for significant differences using Student's t-test.

Results

General Variability of Parameters Virus counts determined in the present study ranged from 1.0 × 108 to 9.5 × 101° 1-1 , thus fluctuating over almost three orders o f magnitude. The virus/bacterium ratio (VBR) varied between 0.9 and 89.1 (mean: 15.0 _+ 10.4). D - D N A concentrations ranged over one order o f magnitude (3.1-26.5 txg 1-1). Assuming a viral D N A content o f 0.099 fg/phage [35], viral D N A concentrations ranged from 0.01 to 9.42 Ixg D N A 1-1 or 0 . 1 - 9 6 . 1 % o f the total dissolved D N A (mean, 18.3 _+ 26.2).

Depth-Dependent Variability of Parameters During the period o f stratification, viral and bacterial numbers as well as the V B R were generally highest in the therm0cline layer (Fig. 2). In months when the whole water column was mixed, viral and bacterial densities were rather uniform within the entire water column, with one exception in N o v e m b e r 1992, when bacterial abundances were slightly higher at the surface than in deeper water. In a similar way, H N F numbers showed the strongest depth-related variation in months that were characterized by a thermocline; H N F densities in the thermocline layer were at least as high as the adjacent water above (Fig. 3). W h e n the water colunm was stratified, Chl a concentrations in the thermocline were higher or at least as high as in above waters, whereas under nonstratified conditions Chl a concentrations decreased slightly with depth (Fig. 3). The depth-related concentrations o f D - D N A showed greater variations during months with water stratification than in the other months (Fig. 4). However, no clear pattern o f the variation o f D - D N A with depth could be observed. During water stratification, the proportion o f viral D N A in total D - D N A was highest in the thermocline (Fig. 4).

30

M.G. Weinbauer et al. VBR

0 0

5-

I' 20 I

viruses, 40 I

60

80

i

i

100

109/I 0

20

40

60

80

I

I

I

I

Jun 1991

~ II

°.o" o.-

,

Jul 1991 ",%

~-i

l"

100

F..- i - - t

~

..-'"

10-

15"

I I

0

I I

I I

/

,:

I,~

}?

I I

Aug 1991

lit

I I

10

I I

I I

15

I I

0

May 1992

E

v

I

{3. "o

10 ¸

15" 0

U

~.o o °°oo,oo

I

15

I

Nov 1992

**

10

I ).-t

0

,

Feb 1993

"%

5-

b,,I

"ll,'

t'

!

!

10-

li

150

I

I

5

10

\.

\

circles: viruses squares: bacteria triangles: VBR

x

I

5

20

25

bacteria,

0

5

I

I

I

10

15

20

10

15 25

108/I

Fig. 2. Depth profiles of viral and bacterial abundances and virus/baterium ratio (VBR) at various selected seasons. Arrows indicate the depth of a well-established thermocline; arrowhead indicates beginning of water stratification.

Seasonal Variability of Parameters In Aurisina and in Rovinj, viral and bacterial abundances as well as V B R s showed distinct seasonal patterns (Fig. 5A). Virus abundances, calculated as monthly means, were highest in late summer or fall in both areas (more than 5 × 10 ~° viruses 1-~); values o f less than 109 viruses 1 -l were observed in winter or early spring. A moderate m a x i m u m in virus abundance was detected in Rovinj in M a y 1992. A similar pattern was observed for bacterial abundance. VBRs, expressed as monthly means, were higher in Rovinj than in Aurisina (Fig. 5A). In Aurisina HNF, densities

Marine Viruses and Dissolved DNA

31 HNF, 106/I

0

2

4

6

8

2

4

6

8

I

I

I

I

I

I

I

I

..."

,99,

5-

m

: ]

I-

10-

Jul 1991 i

I I

0

I

I

I

I

v

c"-

I

I

I

I

t

I

I

1~_1.4._1

x_

(2) (1) E

10

t

15%--

----i

f

15 0

May 1992

510

10-

d) "t3

15-

! "'''''''''°''''" X I Aug 1991 I

i I

0

i

I

i

i

I

I

_

i

l

f

i

f

15 0

Feb 1993

Nov 1992 / ~ m f

t

f

i

10

lO

/

Ii

15 0

/

,.

circles: HNF squares: Chl a

I

I

I

I

1

3

4

5

0

I

I

I

1

I

1

2

3

4

5

15

Chl a, !~g/I Fig. 3. Depthprofiles of heterotrophicnanoflagellate (HNF) abundances and chlorophylla (Chl a) concentrations in different seasons. Arrows indicate the depth of a well-established thermocline; arrowhead indicates beginning of water stratification. and Chl a concentrations reached maxima in August 1991 (Fig. 5B). In Rovinj, Chl a concentrations were highest in May and November 1992, whereas HNF abundances were highest from June through August. The concentrations of D-DNA showed also a seasonal pattern (Fig. 5C). In Aurisina, D-DNA concentrations, calculated as monthly means, were high between June and October 1991 ( > 1 6 txg DNA 1-1) and low between January and May ( < 1 2 ~g DNA l-l). In Rovinj, D-DNA concentrations were high in June and August 1992 ( > 1 4 Ixg DNA 1 -l) and low in May and November 1992 and in

32

M.G. Weinbauer et al. D-DNA,

10

0 0

15

20

p.gll

25

i

25

I

I

J I

Jun 1991 i

I

I

I

I

I

I

I

I

I

I

10-

I

I I

15

I I

I

May 1992

I

%,

5-

I

;

t

E

10

,d"

15-

(1) "o

20

I

Jul 1991

F--'~---I

t-"

15

I

i

I---

"

10-

v

10

1

,

F-= --4 5-

(1)

5

III

'

t

15-

i

10 Aug 1991

I

I

I

0

I

I

I

f

|

I

liB

15

I

I

I

0

Feb 1993

w"

4

)-I

10

I

I

P

°,

5

I

I

at

%%

10 15-

circles: D-DNA squares: %viral DNA

Nov 1992

2O 0

I

I

I

I

20

40

60

80

100

0

I

I

I

I

20

40

60

80

15

100

%viral DNA

Fig. 4. Depth profiles of dissolved DNA (D-DNA) concentrations and % viral DNA in total DDNA. Arrows indicate the depth of a well-established thermocline; arrowhead indicates begining of water stratification.

February 1993 ( < 1 0 ixg DNA 1-1). The concentrations of D-DNA were generally higher in Aurisina than in Rovinj. In Aurisina, estimated D N A contribution of viruses to total dissolved D N A (expressed as monthly means) were low in January 1992 (0.2%), intermediate in June to August 1991 as well as April and May 1992 (4.3-6.7%), and high in October 1991 (29.9%; Fig. 5C). In Rovinj, the monthly means of the estimated viral D N A within the total pool of D - D N A were low in June 1992 and February 1993 (2.1 and 1.6%, respectively) and high in May, August, and November 1992 (27.5-77.1%).

Marine Viruses and Dissolved DNA Aurisina

33

I

Rovinj

100

75-

% 50o~

20

T

circles:

'i

tsriquaglee~: bBteria ' .........i ~

viruses

T./;".,',.

! ..~

/

/

-15

-",'.,

-10

d

5

~

2

250 8-

%

6-

Zq.-

4'

circles: HNF squares: Chl

T a

~

/

..c 0

2 0 2520-

0.7; P < 0.0001) between several of the investigated parameters. Virus abundance is most highly correlated with bacterial numbers, followed by Chl a concentrations. Bacteria were also significantly correlated to temperature and HNFs. Some of the other parameters, for example, HNF abundances and D-DNA concentrations (r -- 0.656), were also correlated significantly (P < 0.0001), but correlation coefficients were almost always comparatively low (r < 0.6). Because some of the correlations might have been influenced by depth effects or by the higher number of samples analysed per month during summer, we performed multiple regression analysis on only the data from surface values given as monthly means. Virus densities were explained only by bacterial numbers (r = 0.947; P = 0.0065). For D-DNA concentrations, no significant multiple regression could be established (r = 0.849; P = 0.0999). Bacterial densities could be explained by virus abundances, Chl a, and D-DNA concentrations (r = 0.955; P = 0.0039). Discussion

Depth-Dependent Variability During water stratification, bacterial and HNF abundances, as well as Chl a concentrations within the thermocline layer, were generally as high or higher than those

Marine Viruses and DissolvedDNA

37

in the water above (Figs. 2, 3). This may result from the capacity of the thermocline layer to retain cells. During the diel cycle, bacterial and HNF abundances and Chl a concentrations were higher within the thermocline layer than in the water body below (Table 1). One reason for the higher virus numbers and VBR at the thermocline (Fig. 2, Table 1) could be a higher number of host cells present in this layer. This is supported by the finding that phage production and the frequency of bacteria containing mature phages increases with bacterial abundances [39, 43] and that the viral burst size increases with the frequency of infected bacteria [44], thus probably causing a higher VBR. Similar depth profiles of virus densities were also reported by Cochlan et al. [11] and Boehme et al. [4]. However, no such relationship of virus densities with water depth was found in the Chesapeake Bay [46]. Along the depth profiles, concentrations of D-DNA showed a stronger variability in months that were characterized by a thermocline than in other periods (Fig. 4). Moreover, during the diel cycle, D-DNA concentrations were higher in the thermocline than in deeper waters (Table 1). A similar depth-related distribution of D-DNA during stratified water conditions is given by Sakano and Kamatani [36]. Therefore, the development of a thermocline seems to affect the depth distribution of the D-DNA concentrations. However, the generally high bacterial and HNF densities and CH1 a concentrations in the thermocline layer are not reflected by higher D-DNA concentrations.

Seasonal Variability In the northern Adriatic Sea, phytoplankton blooms frequently develop not only in spring, but also in the autumn [20], as in November 1992 (Fig. 5B). Because the extracellular release of carbon by phytoplankton cells can stimulate bacterial secondary production [e.g., 1], high primary productivity in the autumn could have caused the observed high bacterial abundance. Therefore, the high viral numbers found in this period are probably a result of high bacterial host densities. The VBR also showed a strong seasonal variability, with highest VBR values in October and November (Fig. 5A). High VBRs in the autumn were also reported from other environments [21, 47]. Thus, viral numbers and viral production may vary not only along a spatial trophic gradient [4, 11, 17, 43, 46] but also with a temporal, that is, seasonal, trophic gradient. D-DNA is a heterogeneous pool consisting of soluble DNA, viral DNA, and uncharacterized bound DNA [23, 29]. Viral DNA is thought to contribute in a varying amount to the total D-DNA concentrations in seawater [2, 4, 29, 43]. In the present study, it was found that the estimated viral DNA contribution to the total pool of D-DNA varied strongly on a seasonal scale (Fig. 5C). The high percentage of viral DNA in total D-DNA found in November 1992 was due to high viral abundances at comparatively low total D-DNA concentrations. Because bacteria probably use predominantly soluble DNA, use of DNA may depend on the relative amounts of single D-DNA compounds, rather than total D-DNA concentrations. However, the molecular weight distribution of D-DNA is poorly studied [2, 12, 23], and it is not known how much single-stranded D-DNA exists in seawater. More information on the structure of the pool of D-DNA is required

38

M.G. Weinbaueret al.

to assess the role of D-DNA as a source of dissolved organic matter and organic phosphorous for the activity of microorganisms in the sea.

Diel Variability Despite the use of CTD technology, it is difficult to get all the samples from the same depth within the thermocline by using Niskin bottles. Thus, a part of the variability of the data, for example, the Chl a peak at 20:00 h on June 25 consisting of only one data point (Fig. 6B), might have been caused by problems during sampling. However, peaks in other parameters, such as bacterial and viral abundances, showed a gradual increase (Fig. 6A), which was probably due to the tight sampling interval of 2 h. Thus, although the data might have been biased by the sampling procedure, the overall dynamics of the investigated parameters should not have been affected. Moreover, sampling in the water body below the thermocline (22 m) should not have been influenced by the problems discussed above. By using the highest decay rates, we estimated that the average percentage of bacterial mortality by viral lysis is 212.2% d -1 in 5 m and 157.2% d -1 in 22 m (Table 2), which is much higher than the 53.3% d -1 reported by Jiang and Paul [21] and the 72% d -~ found by Bratbak et al. [8]. Jiang and Paul [21] argue that any average decay rate used throughout a day may overestimate viral decay at one time and underestimate it at another. These authors found, for example that at a decay rate of 0.0042 h -1, in almost 50% of the samples, no mortality of bacteria resulted from viral lysis. At the same decay rate, we found a similar percentage of samples characterized by a lack of virus-mediated mortality of bacteria (data not shown). Thus, the decay rates of 0.096 h -~ and 0.3 h -~, corresponding to a percentage of bacteria lysed per day in the range of 43.2 to 104.9% in 5 m and 36.5 to 83.0% in 22 m (Table 2) may more closely represent actual decay rates. Assuming that these estimations are true, viral lysis of bacteria can be assumed as an important cause for bacterial mortality. The diel investigation at 5 and 22 m showed that the major maxima in bacterial abundance were followed by viral peaks, whereas an HNF maximum followed a bacterial peak only at 22 m (Figs. 6A, B, 7A, B). This is a further indication that viral-mediated mortality could have been an important cause of the observed decrease in bacterial abundance. Similar oscillations of viral and bacterial numbers were found during a diurnal study in seawater enclosures [8, 18, 21]. Interestingly, the HNF peak at 22 m followed not only the bacterial, but also the virus, peak, indicating that HNFs may have grazed on viruses [16] as well as on bacteria. However, we cannot exclude the possibility that grazing on bacteria by HNFs was not reflected by HNF abundances. We did not find a significant multiple regression coefficient for D-DNA as a dependent variable. A combination of mechanisms of D-DNA production acting simultaneously may explain why we could not relate the oscillation pattern of DDNA to the distribution pattern of bacterial abundances or to Chl a concentrations alone. Moreover, the demand of bacteria for phosphorous and therefore the bacterial use of D-DNA might vary in the investigated area, because it has been shown that in the northern Adriatic Sea, the concentration of dissolved organic phosphorous varies strongly among seasons [13]. We estimated that, at decay rates of 0.096 h -~

Marine Viruses and Dissolved DNA

39

and 0.3 h -1, between 35.8 and 81.6% of the D-DNA pool might have been derived from DNA released from lysed bacteria at 5 m, and between 17.9 and 39.4% at 22 m (Table 2). Thus, the release of D-DNA from bacteria by viral lysis (including viral DNA) might be an important source for the D-DNA and also for the dissolved organic phosphorous pool in the northern Adriatic Sea.

Implications It is generally believed that bacterial mortality is mainly controlled by protozoan grazing [14, 37]. However, recently it has been speculated that viral lysis might be an additional cause for bacterial mortality [32, 34]. In the present study, multiple regression analysis showed that the variation of bacterial abundances could be better explained (in a statistical sense) by virus densities than by HNF numbers. Similarly, bacterial abundances were better correlated to virus than to HNF densities along atrophic gradient [43]. Moreoever, during this diel cycle investigation, bacterial peaks were followed by viral, but not by HNF, maxima at 5 m (Figs. 6, 7), and estimations of bacterial mortality that are due to viral lysis are high (Table 2). We can therefore speculate that the impact of viruses in controlling bacterial densities occasionally could be more important than HNF grazing. Because viral production and the frequency of phage-infected cells increase along with bacterial densities [39, 43], viral mortality of bacteria should be strongest at high host densities. This is supported by recent findings that variations in HNF abundance generally do not allow the prediction of bacterial densities in aquatic systems, especially at high bacterial densities [15]. Consequently, in such a situation, Ctransfer from bacteria to predators should be less important than previously thought. Thus, it might well be that within the "viral loop" [8] a large portion of the bacterial production is cycled back to the pool of dissolved organic matter by the release of cell constituents, which is due to viral lysis. Acknowledgments. We thank the Laboratorio di Biologia Marina at Aurisina-Trieste (Italy) and the Ruder Boskovic Center for Marine Research at Rovinj (Croatia) for hospitality and laboratory space. We are grateful to the Institute of Zoology for laboratory space, W. Klepal for providing TEM equipment, G.J. Hemdl for discussions and support, and J.A. Ott for valuable comments on the manuscript. We also appreciate all members of the working group of G.J. Hemdl for assistance during field and lab work. This research is in partial fulfillment for doctoral requirements for M.G.W. Special thanks are due to E Starmiihlner for making the study possible. This work was supported by the Austrian Science Foundation, grant P 8335 to EE

References 1. Baines SB, Pace ML (1991) The production of dissolved organic matter by phyt0Plankton and its importance to bacteria: patterns across marine and freshwater system. Limnol Oceanogr 36:1078-1090 2. Beebee TJC (1991) Analysis, purification and quantification of extracellular DNA from aquatic environments. Freshwater Biol 25:525-532 3. Bergh O, BCrsheim KY, Bratbak G, Heldal M (1989) High abundance of viruses found in aquatic environments. Nature 340:467-468 4. Boehme J, Frischer ME, Jiang SC, Kellogg CA, Pichard S, Rose JB, Steinway C, Paul JH (1993)

40

5. 6. 7. 8. 9.

10. 11.

12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

28. 29.

M.G. Weinbauer et al. Viruses, bacterioplankton, and phytoplankton in the southeastern Gulf of Mexico: distribution and contribution to oceanic DNA pools. Mar Ecol Prog Ser 91:1-10 Bcrsheim KY, Bratbak G, Heldal M (1990) Enumeration and biomass estimation of planktonic bacteria and viruses by transmission electron microscopy. Appl Environ Microbiol 56:352-356 Bratbak G, Egge JK, Heldal M (1993) Viral mortality of the marine alga Emiliania huxleyi (Haptophyceae) and termination of algal blooms. Mar Ecol Prog Ser 93:39-48 Bratbak G, Heldal M, Norland S, Thingstad TF (1990) Viruses as partners in spring bloom microbiol trophodynamics. Appl Environ Microbiol 56:1400-1405 Bratbak G, Heldal M, Thingstad TF, Riemann B, Haslund OH (1992) Incorporation of viruses in the budget of microbial C-transfer. A first approach. Mar Ecol Prog Ser 83:273-280 Caron DA (1983) Technique for enumeration of heterotrophic and phototrophic nanoplankton, using epifluorescence microscopy, and comparison with other procedures. Appl Environ Microbiol 46:491-498 Cho BC, Azam F (1988) Heterotrophic bacterioplankton production measurement by the tritiated thymidine incorporation method. Arch Hydrobiol B eih 31:153-162 Cochlan WP, Wikner J, Steward GF, Smith DC, Azam F (1993) Spatial distribution of viruses, bacteria and chlorophyll a in neritic, oceanic and estuarine environments. Mar Ecol Prog Ser 92:77-87 DeFlaun ME Paul JH, Jeffrey WH (1987) Distribution and molecular weight of dissolved DNA in subtropical estuarine and oceanic environments. Mar Ecol Prog Ser 38:65-73 Faganelli J, Herndl G (1991) Dissolved organic matter in the waters of the Gulf of Trieste (Northern Adriatic). Thalassia Jugoslavica 23:51-63 Fenchel T (1982) Ecology of heterotrophic microflagellates. IV. Quantitative occurrence and importance as bacterial consumers. Mar Ecol Prog Ser 9:35-42 Gasol MJ, Vaqu6 D (1993) Lack of coupling between heterotropic nanoflagellates and bacteria: a general phenomenon across aquatic systems? Limnol Oceanogr 38:657-665 Gonzfilez JM, Suttle CA (1993) Grazing by marine nanoflagellates on viruses and viral-sized particles: ingestion and digestion. Mar Ecol Prog Ser 94:1-10 Hara S, Terauchi K, Koike I (1991) Abundance of viruses in marine waters: assessment by epiflourescence and transmission electron microscopy. Appl Environ Microbiol 57:2731-2734 Heldal M, Bratbak G (1991) Production and decay of viruses in aquatic environments. Mar Ecol Prog Ser 72:205-212 Hobble JE, Daley RJ, Jasper S (1977) Use of Nuclepore filters for counting bacteria by epifluorescence microscopy. Appl Environ Microbiol 33:1225-1228 lvancic I, Degobbis D (1987) Mechanisms of production and fate of organic phosphorous in the northern Adriatic Sea. Mar Biol 94:117-125 Jiang SC, Paul JH (1994) Seasonal and diel abundances of viruses and occurrence of lysogeny/ bacteriocinogeny in the marine environment. Mar Ecol Prog Ser 104:163-172 Karl DM, Bailiff MD (1989) The measurement and distribution of dissolved nucleic acids in aquatic environments. Limnol Oceanogr 34:543-558 Maruyama A, Oda M, Higashihara T (1993) Abundance of virus-sized non-DNase-digestible (coated DNA) in eutrophic seawater. Appl Environ Microbiol 59:712-717 Mathews J, Buthala DA (1970) Centrifugal sedimentation of virus particles for electron microscopic counting. J Virol 5:598-603 Parsons T, Malta Y, Lalli C (1984) A Manual of chemical and biological methods for seawater analysis. Pergamon Press, Oxford, UK Paul JH, Carlson DJ (1984) Genetic material in the marine environment: implication for bacterial DNA. Limnol Oceanogr 29:1091-1097 Paul JH, DeFlaun ME Jeffrey WH, David AW (1988) Seasonal and diel variability in dissolved DNA and in microbial biomass and activity in a subtropical estuary. Appl Environ Microbiol 54:718-727 Paul JH, Jeffrey WH, Cannon JP (1990) Production of dissolved DNA, RNA, and protein by microbiol populations in a Florida reservoir. Appl Environ Microbiol 56:2957-2962 Paul JH, Jiang SC, Rose JB (1991) Concentration of viruses and dissolved DNA from aquatic environments by vortex flow filtration. Appl Environ Microbiol 57:2197-2204

Marine Viruses and Dissolved DNA

41

30. Paul JH, Rose JB, Jiang SC, Kellogg CA, Dickson L (1993) Distribution of viral abundance in the reef environment of Key Largo, Florida. Appl Environ Microbiol 59:718-724 31. Peduzzi P, Weinbauer MG (1993) Effect of concentrating the virus-rich 2-200nm size fraction for the formulation of algal flocs (marine snow). Limnol Oceanogr 38:1562-1565 32. Proctor LM, Fuhrman JA (1990) Viral mortality of marine bacteria and cyanobacteria. Nature 343:60-62 33. Proctor LM, Fuhrman JA (1991) Roles of viral infection in organic particle flux. Mar Ecol Prog Ser 69:133-142 34. Proctor LM, Fuhrman JA (1992) Mortality of marine bacteria in response to enrichments of the virus size fraction from seawater. Mar Ecol Prog Set 87:283-293 35. Reanney DC, Ackermann H-W (1982) Comparative biology and evolution of bacteriophages. Adv Viral Res 27:205-280 36. Sakano S, Kamatani A. (1992) Determination of dissolved nucleic acids in seawater by the fluorescence dye, ethidium bromide. Mar Chem 37:239-255 37. Sherr EB, Sherr BF (1987) High rates of consumption of bacteria by pelagic ciliates. Nature 325:710-711 38. Steward GF, Wikner J, Cochlan WE Smith DC, Azam F (1992) Estimation of virus production in the sea. I. Method development. Mar Microb Food Webs 6:57-78 39. Steward GF, Wikner J, Cochlan WE Smith DC, Azam F (1992) Estimation of virus production in the sea. II. Field results. Mar Microb Food Webs 6:79-90 40. Suttle CA, Chan AM, Cottrell MT (1990) Infection of phytoplankton by viruses and reduction of primary productivity. Nature 347:467-469 41. Suttle CA, Chen F (1992) Mechanisms and rates of decay of marine viruses in seawater. Appl Environ Microbiol 58:3721-3729 42. Turk V, Rehnstam A-S, Lundberg E, Hagstr6m A (1992) Release of bacterial DNA by marine nanoflagellates, an intermediate step in phosphorus regeneration, Appl Environ Microbiol 58:3744-3750 43. Weinbauer MG, Fuks D, Peduzzi P (1993) Distribution of viruses and dissolved DNA along a coastal trophic gradient in the northern Adriatic Sea. Appl Environ Microbiol 59:4074-4082 44. Weinbauer MG, Peduzzi P (1994) Distribution, size and frequency of bacteriophages in differentbacterial morphotypes. Mar Ecol Prog Ser 108:11-20 45. Weinbauer MG, Peduzzi, P (in press) Comments on the determination o f nucleic acids in natural waters by the CTAB-DABA-orcinol method. Sci. Total Environ. 46. Wikner J, Vallino JJ, Steward GF, Smith DC, Azam F (1993) Nucleic acids from the host bacterium as a major source of nucleotides for three marine bacteriophages. FEMS Microbiol Ecol 12:237-248 47. Wommack KE, Hill RT, Kessel M, Russek-Cohen E, Colwell RR (1992) Distribution of viruses in the Chesapeake Bay. Appl Environ Microbiol 58:2965-2970