Relationship between viral and prokaryotic abundance on the Bajo O ’ Higgins 1 Seamount ( Humboldt Current System off Chile )

There is little known about the ecology of microbial communities living in the water column over seamounts. Here, for the first time, the spatial distribution and abundance of virus-like particles (VLP) are described over a seamount. The association between VLP distribution, prokaryotic abundance, and environmental variables is also analyzed. Sampling was conducted in December 2004 on the Bajo O’Higgins 1 seamount (32°54’S, 73°53’W) located in the Humboldt Current System off Chile. A oxygen minimum layer (OMZ) was clearly present between 130 and 280 m in the water column over the seamount. Water samples were taken with Niskin bottles at 10 oceanographic stations over the seamount at depths of 5, 20, 50, 75, 100, and 150 m and at the benthic boundary layer (BBL; 5-12 m over the sediments). Temperature, salinity, oxygen, chlorophyll a, and phaeopigments were measured at each station. Viral and prokaryotic abundances were determined with fluorochrome SYBR Green I. Viral abundance ranged from 1.53 x 109 VLP L-1 16.48 x 109 VLP L-1, whereas prokaryotic abundance ranged from 1.78 x 108 cell L-1 14.91 x 108 cell L-1. The virus-like particle/prokaryote ratio varied widely among the analyzed layers (i.e. surface, OMZ, and BBL), probably due to the presence of different prokaryotic and viral assemblages in each layer. Our results indicate that the environmental conditions, mainly the concentration of dissolved oxygen in the water column over Bajo O’Higgins 1 seamount, shape the association between viral and prokaryotic abundance.


INTRODUCTION
Viruses are the most abundant members of aquatic microbial communities (Winter et al., 2005) and their density can exceed bacterial abundance by between 5 and 25 times (Fuhrman, 1999).Although viruses were detected in the marine ecosystem several decades ago (Spencer, 1955;Valentine et al., 1966;Torrella and Morita, 1979;Drake et al., 1998), their importance remained unknown until the microbial loop concept was developed and later refined (Azam et al., 1983).Since then, our understanding of the ecological role that viruses play in the food web and in biogeochemical cycles (Fuhrman, 1999;Weinbauer et al., 2003) has expanded.Viruses are now known to play a fundamental role in controlling prokaryote communities (Bratbak et al., 1990;Fuhrman, 1999), generating mortality rates between 8 and 43% (Fuhrman, 1999;Winter et al., 2005).
Numerous studies on the distribution of viruses in marine environments have focused mainly on patterns produced by productivity gradients from coastal to oceanic areas (e.g.Weinbauer et al., 1993;Cochlan et al., 1993;Steward et al., 1996;Corinaldesi et al., 2003).It has been shown that viral abundance increases significantly with system productivity (Weinbauer, 2004) and, in accordance with this, it has been proposed that the trophic state of a system is a controlling factor of viral spatial distribution (Corinaldesi et al., 2003).This general tendency, however, has not been contrasted in many marine environments such as seamounts, which influence local circulation and increase the area's biological productivity (e.g.Sime-Ngando et al., 1992).In fact, no quantitative estimate of viral abundance on a seamount has ever been reported to date.
On the other hand, it has been shown that the presence of low oxygen waters and/or strong oxygen gradients in the water column can affect the abundance of virus-like particles (VLP; Taylor et al., 2003;Weinbauer et al., 2003).The oxygen minimum zone (OMZ) of the Humboldt Current System (HCS) is a strong physical barrier for the distribution of many organisms (Antezana, 1978;Judkins, 1980;González and Quiñones, 2002), and the prokaryote community inhabiting it is quite different to that living in the oxygenated surface waters, as shown for archaea (Levipan, 2006), nitrifying communities (Castro- González, 2004), and photosynthetic cyanobacteria (Prochlorococcus sp. and Synechococcus sp.; Ulloa, 2003).
The water column in the case of the Bajo O'Higgins 1 area is influenced by an OMZ and is located over a seamount.Therefore, the relationship between viral abundance and prokaryote abundance is expected to vary considerably in the water column.
This study analyzes the spatial distribution of VLP in relation to (i) prokaryotic abundance and (ii) environmental factors in the water column on the Bajo O'Higgins 1 seamount.

Sampling and hydrographic data
Sampling took place from 24 to 25 December 2004, on board the R/V KOYO MARU (National Fisheries University, Japan).Water samples were taken at 10 stations (bottom depth between 376 and 841 m; see Table 1) over the seamount (Niskin bottles, 2.5 L) at the following sampling depths: 5, 20, 50, 75, 100, and 150 m.Samples from 5 to 12 m over the sediments were also taken at all stations and, for practical purposes, will be referred to here as benthic-boundary layer (BBL) samples.Hydrographic data (temperature, salinity, and oxygen) were collected with a CTD Sea Bird Electronics SBE 9.

Chlorophyll a and phaeopigments
For each depth, 500 ml of water was filtered through a Whatman GF/F filter and then immediately frozen until processing.Pigments were extracted in acetone (90%) for 24 h, in the dark, at -20°C, and then determined using a fluorometer (TD-700 Turner Designs) according to Parson et al. (1984).

Virus-like particles (VLP) and prokaryotic abundance
Water samples for counting VLP and prokaryote cells were fixed immediately on board with glutaraldehyde (2% final concentration) and stored in the dark at 4°C until analyzed.The samples were processed within 10 days.As described by Noble and Fuhrman (1998), 3 ml were filtered (0.02 μm pore diameter; Anodisc) and stained with fluorochrome SYBR Green I (stock 10%).The membranes were incubated in the dark for 10 minutes and then mounted on slides with a drop of buffer solution (50% glycerol, 50% sodium phosphate buffer, 0.1% pphenylenediamine).VLP and prokaryote cell abundances were determined by counting between 10 and 20 randomly selected fields with a Zeiss epifluorescence microscope (model Axioskop 2 plus) at a magnification of 1600x.

Data analysis
Pearson's correlation and partial correlation analysis (surface stratum: 5-50 m; all data) were used to look for possible associations between variables.Lineal regression analysis was also performed between selected variables.An analysis of variance (ANOVA) was done to detect whether the analyzed variables (i.e.VLP abundance, prokaryotic abundance, temperature, oxygen, salinity, chlorophyll a, phaeopigments) varied according to geographic location on the seamount; different stations as well as vertical layers (i.e.5-20 m, 5-50 m, and 50-150 m) were compared.When significant differences were found, the Tukey test was used to identify the stations.All variables were transformed logarithmically (log 10 ) before carrying out any statistical analyses.

Environmental parameters
The water column was stratified on the seamount, and had a mixed layer around 35 m depth.Figure 2 shows the hydrographic parameters for each station.The temperature ranged from 17.7°C (surface) to 5.6°C (bottom) and salinity ranged from 34.31 (surface) to 34.58 (200 m).The oxygen levels varied widely along the water column, averaging 5.2 mL O 2 L -1 at the surface and decaying to <0.5 mL O 2 L -1 between 130 and 280 m depth.These low oxygen conditions reflect the strong influence of the Equatorial Subsurface Waters (ESSW) even at this distance from the shore (Schneider et al., 2003).Below this, the oxygen level increased to a maximum of almost 4 mL O 2 L -1 around 550 m depth.A subsurface chlorophyll a and phaeopigment maximum was observed at 50 m depth in most stations (Fig. 3); the respective concentrations ranged from 0.21 μg L -1 to 0.55 μg L -1 (average 0.41 ± 0.09 μg L -1 ) and from 0.15 μg L -1 to 0.26 μg L -1 (average 0.22 ± 0.03 μg L -1 ).In the BBL samples, the highest concentrations were found in the shallowest bathymetric isolines (400 and 500 m; see Table 1).
In the BBL samples, VLP abundance varied between 0.68 x 10 9 VLP L -1 and 1.41 x 10 9 VLP L -1 and prokaryotic abundance varied between 0.91 x 10 8 cell L -1 and 1.72 x 10 8 cell L -1 (see Table 1).In both cases, the highest concentrations were found over the meseta that constitutes the seamount's peak (~ 400 m).

VLP to prokaryote ratio (VP-ratio)
The VP-ratio in the water column decreased with depth at all stations ranging from an average of 12.56 (surface) to 4.3 (100 m depth).This trend of decreasing with depth reversed upon reaching the OMZ, where the average VP-ratio increased to 7.05 at 150 m (OMZ; Fig. 5a).
On the other hand, no significant spatial variations were found in the VP-ratio of the BBL samples (Fig. 5b).The average VP-ratio of the BBL was 7.87, slightly higher than that of the OMZ.

Associations between VLP, prokaryotes, and environmental variables
When analyzing all the data (i.e.water column and BBL samples; n = 70), a high correlation (r 2 = 0.821, p <0.05) was found between viral and prokaryotic abundance.Likewise, VLP were highly associated with chlorophyll a, temperature and, to a lesser degree, oxygen (Table 2).Prokaryotic abundance, however, was highly associated with chlorophyll a and temperature and, to a much lesser extent, with oxygen and phaeopigments (Table 2).
An analysis of the surface stratum (between 5 and 50 m depth) revealed that VLP abundance is moderately correlated with prokaryotic abundance, phaeopigments, and salinity but strongly correlated with temperature and oxygen (Table 2).The environmental homogeneity observed during the cruise in the OMZ (below 150 m) and at the BBL does not allow a proper correlation analysis within each of these strata.
When determining the association between VLP and prokaryotic abundance, the co-variance of both variables with environmental variables such as oxygen and temperature must be considered.Furthermore, the environmental variables could be associated with each other.Due to this, a partial correlation analysis was done pooling all the data together (i.e.surface layer, OMZ, and BBL; n = 70), which revealed a significant relationship between VLP and chlorophyll a, phaeopigments, temperature, oxygen, and salinity (Table 3).The partial correlation analysis also showed a significant relation- ship between viral abundance and prokaryotic abundance at the surface (5-50 m) (Table 3).

Between-station comparisons
Significant differences (p <0.05) were detected for the surface layer (5-20 m depth; n = 20) for prokaryotic abundance, chlorophyll a , temperature, oxygen, and salinity.The Tukey test among stations only indicated significant differences for chlorophyll a, temperature, and salinity between some stations (data not shown).No significant differences (p <0.05) were found for the other analyzed layers (i.e.all the data and 50-150 m.).

DISCUSSION
This study describes, for the first time, the distribution and abundance of VLP on a seamount and the relationship of this distribution with prokaryotic abundance as well as with environmental factors.
The spatial distribution of chlorophyll a on the seamount did not vary much between sampling stations (Fig. 3a).Nevertheless, this observed homogeneity should be taken with caution considering the short time scale involved in our sampling.The chlorophyll a and phaeopigment levels measured on the seamount are low (Fig 3), but within the ranges reported for adjacent areas (Pizarro et al., 2001).Unfortunately, no information is available on primary production from this area, making it impossible to determine the level of biological productivity for Bajo O'Higgins 1.However, increased levels of phytoplanktonic biomass and productivity were found in areas influenced by seamounts (e.g.Comeau et al., 1995) and therefore it is possible that primary productivity could be enhanced on the Bajo O'Higgins 1 seamount in some periods of the year.In fact, the high fishery activity on species such as Orange Roughy (Hoplostethus atlanticus) and Alfonsino (Beryx splendens), which concentrate on this seamount (Young et al., 2000), suggests that the seamount's topography may influence biological and oceanographic processes, therefore enhancing biological productivity.Seamounts can generate physical disturbances that cause water masses to rise from the sea bottom to the photic zone, increasing primary production (Boehlert and Genin, 1987;Sime-Ngando et al., 1992;Pusch et al., 2004) and creating conditions capable of sustaining a high diversity and abundance of organisms (Rowden and Clark, 2004).
With the increase in research in the field of marine virology, it is becoming more evident that virioplankton play a significant role in aquatic communities (e.g.Suttle et al., 1990;Eissler and Quiñones, 2003;Winter et al., 2004).Accordingly, the precision and accuracy of the methods for estimating viral abundance are crucial.However, the discussion regarding the best methodology to be used is on-going.The methodology utilized in this research (i.e.fixation with glutaraldehyde and storing at 4ºC) is the most common method used to preserve and store viruses for determining their abundance in aquatic ecosys- tems (e.g.Noble and Fuhrman, 1998;Chen et al., 2001;Hewson et al., 2001;Culley and Welschmeyer, 2002;Taylor et al., 2003;Corinaldesi et al., 2003;Glud and Middelboe, 2004;Lunau et al., 2005;Vanucci et al., 2005).Nevertheless, this methodology is becoming controversial because there is evidence showing that it can underestimate viral abundance (Brussaard, 2004;Wen et al., 2004).It is highly likely that there has been a certain level of underestimation with the methodology we used here.We recognize the importance of the role of fixing and storing in estimating viral abundance and, for that matter, bacterial abundance.In fact, Wen et al. (2004) suggest that it is much more reliable to store the samples at -80°C.Wen's conclusions are coherent with those of Brussaard (2004), who uses flow cytometry combined with liquid nitrogen, deep freezing (-80ºC), and glutaraldehyde.In contrast, other authors have reached the conclusion that storing at -80ºC is more effective only when the objective is to keep the samples for long periods of time (more than 40 days) but that for shorter periods storing at 4ºC is better (Olson et al., 2004).It is important to note that methodological controversies are frequent in the fields of marine microbiology and biological oceanography.For instance, the controversies concerning the real efficiency of filters of different materials and pore sizes (e.g.Stockner et al., 1990;Lee et al., 1995), the most appropriate methodology for measuring DOC (e.g.Sugimura and Suzuki, 1988;Suzuki et al., 1992;Wangersky, 1993), or even the most adequate methodology for obtaining an optimum measurement of chlorophyll in seawater (e.g.Holm-Hansen and Riemann, 1978;Baker et al., 1983;Stramski, 1990).The history of these controversies has shown that it is necessary to accumulate a large body of evidence before discarding one method and the knowledge obtained from it.The high VLP abundance observed on the Bajo O'Higgins 1 seamount is consistent with previously published values for other oceanic areas (see Table 4), although it is slightly lower than that observed in the water column over the continental shelf off central Chile (average 2.05 x 10 10 VLP L -1 ; unpublished results).Viral abundance decreases with depth on the seamount (Fig. 4a), which is in accordance with trends previously reported in other oceanic ecosystems (e.g.Hara et al., 1996;Steward et al., 1996).Similarly, the prokaryotic abundance is within the ranges registered off central Chile, especially in oceanic zones of the HCS (e.g.Cuevas et al., 2004).The decreasing prokaryotic vertical distribution observed on Bajo O'Higgins 1 (Fig. 4b) is also consistent with the pattern described for other areas, such as the Adriatic Sea (Corinaldesi et al., 2003) and the North Pacific Ocean (Culley and Welschmeyer, 2002).
The significant correlation found between VLP abundance and prokaryote cells (Table 2) suggests that the members of the latter group are the main hosts for viruses, which is a typical trend in marine systems (e.g.Hara et al., 1996;Weinbauer and Suttle, 1997;Culley and Welschmeyer, 2002).Nevertheless the correlation found between VLP and chlorophyll a (Table 2) suggests that phytoplankton may also play a relevant role as hosts.
Viral distribution and abundance in the ocean can be controlled or modified by environmental factors such as light, temperature, oxygen (e.g.Gantzer et al., 1998;Weinbauer and Höfle, 1998;Bettarel et al., 2003, Corinaldesi et al., 2003), and turbidity (Winter et al., 2005).Likewise, abiotic factors like nutrient availability and temperature affect prokary- ote distribution and abundance (e.g.Brett et al., 1999;Heidelberg et al., 2002).Thus, environmental gradients in the ocean that are able to modify viral or prokaryotic abundance can potentially influence the VLP/prokaryote relationship.Similarly, biotic factors could also modulate the relationship between VLP and prokaryotes such as grazing by protists (i.e.ciliates and flagellates) over prokaryotes (e.g.Fuhrman and Noble, 1995;Cuevas et al., 2004).
The HCS is characterized by sharp vertical gradients, especially of oxygen, generated by the presence of several water masses, namely SubAntarctic Waters (SAAW), ESSW, and Antarctic Intermediate Waters (AAIW) (Strub et al., 1998).The ESSW brought by the Gunther Current is particularly relevant to the present study, since these waters are very poor in dissolved oxygen (Brandhorst, 1971;Sievers and Silva, 1982;Silva, 1983).The environment generated by the OMZ associated with the ESSW represents an important biological barrier for organisms relying on aerobic metabolism (Eissler and Quiñones, 1999).Based on the oxygen vertical distribution (Fig. 2b), three systems were identified in the water column on the Bajo O'Higgins 1 seamount.The surface system (i.e.0 to 100 m) is characterized by high oxygen concentrations (>2 mL O 2 L -1 ) and influenced by light.The second system is below 100 m and corresponds largely to ESSW, with extremely low oxygen conditions (<0.5 mL O 2 L -1 ).Finally, at the BBL, oxygen concentrations are higher than at the OMZ due to the influence of the more oxygenated waters of the AAIW.
Although the microbial assemblages inhabiting the OMZ of the HCS remain largely unidentified, some evidence indicates that they are very different from those dwelling in oxic waters.For example, the abundance of archaea (Levipan, 2006) and assemblages of denitrifying microorganisms (Molina, 2004;Castro-González et al., 2005) are important in the OMZ of the HCS.Consequently, the prokaryote community of the OMZ is expected to be very different from the microbial communities that live at the surface and bottom of the ocean, which is reflected in the different VP-ratios obtained (Fig. 5).A similar phenomenon was described by Taylor et al. (2003) for variations of viral particles and prokaryotic abundance between three layers (oxic, transitional, and hypoxic) of the water column in Cariaco Basin.These authors found viral distribution to be marked by an environmental oxygen gradient, which was clearly reflected in VP-ratios; the higher proportions corresponded to the hypoxic zone.Similarly, Weinbauer et al. (2003) showed an incre-ment in the abundances of both VLP and prokaryotes in anoxic waters of the Baltic Sea.In fact, the same trend was observed in the present study, where the VPratios decreased with depth in the oxic zone and then drastically increased under hypoxic conditions.
The differential VLP/prokaryote association with respect to environmental gradients on the Bajo O'Higgins 1 seamount could be due to the fact that the microbial assemblage compositions, particularly for prokaryotes, are very different in these layers (i.e.surface, OMZ, BBL).A low VPratio may indicate that the observed relationship between VLP and host prokaryotic cells could be explained by the presence of a low infecting rate, a low number of viruses per available host cell, or a high virus decay rate (Alonso et al., 2001).In addition, the difference in the VP-ratios could be caused by virus-host specificity (Middelboe et al., 2000) or the different viral assemblages found in the oxic and hypoxic layer.Another alternative hypothesis for the lack of correlation between VLP and prokaryotes and the high VP-ratio in the OMZ is a relative increase in virioplankton not able to infect the existing prokaryotes, as Bettarel et al. (2003) observed in some lakes.
In contrast to the rest of the water column, no change in the VP-ratio associated with the dissolved oxygen content was observed in the BBL, probably due to the spatial variability of the community living in the sediment.It is known that the BBL is strongly influenced by the resuspension of organic matter (Schallenberg and Burns, 2004), sinking of POM from surface layers (Turley, 2000), and degradation processes acting on the organic matter found at the sea bottom (Suess, 1980;Peterson et al., 1988;Graco et al., 2001).
Given these results, we conclude that the association between viral and prokaryotic abundance is significantly shaped by the environmental conditions, mainly dissolved oxygen concentration in the water column associated with the Bajo O'Higgins 1 seamount.Austral de Chile; Undersecretariat of Fisheries, Chilean Government) and Mr. Alejandro Zuleta (CEPES, Universidad Austral de Chile) for the opportunity to participate in the R/V Koyo Maru cruise.We extend our appreciation to the captain and crew of the R/V Koyo Maru, and particularly to First Officer Tadashi Kamano.Furthermore, we are grateful to Raul Gili, Gerdhard Jessen, and Karol Espejo, who contributed with logistic support for the expedition; and to Pedro Inostroza, Cristian Chandía, and María Inés Muñoz, who were of great assistance during on board sampling.We would also like to recognize Eduardo Navarro and Rodrigo Montes for providing support in data processing and analysis.This research was funded by the COPAS Center (FONDAP, CONICYT).
FIG. 4. -Vertical profiles of (a) virus-like particles and (b) prokaryotes at the stations on Bajo O'Higgins 1 seamount.

TABLE 1 .
-Station location, pigment concentrations, and the abundance of virus-like particles and prokaryotes in the benthic-boundary layer (BBL) on the Bajo O'Higgins 1 seamount.n.a.= data not available.