Tide driven microbial dynamics through virus-host interactions in the estuarine ecosystem

Microbes drive ecosystems and their viruses manipulate these processes, yet the importance of tidal functioning on the estuarine viruses and microbes remains poorly elucidated. Here, an integrative investigation on tidal patterns in viral and microbial communities and their inherent interactions over an entire spring-neap tidal cycle was conducted along a macrotidal subtropical estuary. The viral and microbial abundances oscillated significantly over the tidal cycle with relatively higher abundances observed at spring tide compared to neap tide. The distinct tidal dynamic patterns in bacterial production and community composition were tightly associated with the variations in viral infection, production and decay, revealing the tide-driven interactions between viruses and microbes. Concurrent with the higher viral decay but lower bacterial abundance and inhibited bacterial metabolism during the neap tide, lower gross viral production was coupled with a synchronous switching from viral lytic to lysogenic infection induced by the loss of viral infection efficiency and the transition from marine to freshwater bacterial populations triggered by tidal mixing. Our results highlighted the major tidal impact on the microbial dynamics through virus-host interactions, with cascading effects, neglected so far, on estuarine biogeochemical cycles. © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Rhythmicities of environmental conditions generally set the reoccurring life events. Estuarine habitats are vital nodes of landocean interactions, particularly marked by strong periodic tidal fluctuations. The effect of gravitational forces exerted by the motion of celestial bodies generates the diurnal variability in tidal stages (e.g., high tide and low tide) and change in tidal amplitude over spring-neap cycle (Kowalik, 2004;Kvale, 2006). Estuaries, beach intertidal zones, salt marshes, and other nearshore systems suffer multiple forcing mechanisms associated with the tides, such as hydraulic head gradient-driven saltwater circulation, saltwaterfreshwater mixing and tidal pumping (Heiss and Michael, 2014;Kowalik, 2004;Yu et al., 2014). Previous investigations have found that tide-induced mixing and advection greatly contributed to the biological processes such as the global phytoplankton growth-loss balance, the vertical migration of copepods, the semilunar spawning period of nekton and the grazing rate of pelagic and benthic herbivores (Findlay et al., 2013;Kowalik, 2004;Pernthaler, 2005;Sharples, 2007).
Microbes drive biogeochemical processes on a global scale (Falkowski et al., 2008). Lytic and lysogenic viral infections significantly affect the mortality, biomass and composition of microbial communities (Howard-Varona et al., 2017;Weinbauer, 2004;Zhang et al., 2007), fueling viral impact on microbial food web and microbial transformation of dissolved organic matter (Jiao et al., 2010;Suttle, 2005Suttle, , 2007Zhang et al., 2014). The viral lysis-lysogeny switch determines the release of progeny viral particles, which may decrease in infectivity and degrade when exposed to various physico-chemical and biological factors in the environment, such as ultraviolet light, salt stress and enzyme activity (Choi et al., 2010;Mojica and Brussaard, 2014;Motlagh et al., 2015;Suttle and Chen, 1992). It is generally accepted that lytic viral infection tends to follow the well-known density-dependent "Kill-the-Winner" dynamics, wherein abundant and rapidly growing microbes are more affected by lytic viruses (Suttle, 2007;Thingstad and Lignell, 1997), while lysogeny favors lower microbial abundance or activity which is hypothesized as the key to host survival in harsh environments (Paul, 2008;Payet and Suttle, 2013). However, the general validity of this trend is debated and varies across differential environments, which implies that the nature of virus-host interactions remains largely understudied (Maurice et al., 2013;Weinbauer, 2004). The recently proposed "Piggyback-the-Winner" model suggests an alternative scenario for viral infections, wherein lysogenic infection becomes increasingly favored with higher microbial density (Coutinho et al., 2017;Knowles et al., 2016).
In the face of the tidal mixing between riverine water and seawater, estuaries accommodate vital and complex ecological niches occupying the same space, which presumably requires physiological, genetic and ecological adaptations for both viruses and microbes (Maurice et al., 2010b). Patterns in viral life strategies were previously revealed to be driven by the metabolic activities of their hosts that impacted by trophic level, in which lytic production tends to prevailed during the high productivity season while lysogeny usually occurred with a limited resource availability (Brum et al., 2016;Payet and Suttle, 2013). In addition, the changes of other environmental factors (e.g., water flux, salinity and pH) might also affect the physiological status of host microbes, and subsequently regulate the inherent virus-host interactions (e.g., lysis-lysogeny switch) (Jasna et al., 2017;Maurice et al., 2013;Mojica and Brussaard, 2014;Wei et al., 2019). However, information about the tidal impact on estuarine viral and microbial dynamics is scarce (Anderson et al., 2018;Celussi et al., 2008;Hyun et al., 1999). Most studies were conducted in microtidal estuaries harboring a relatively lower tidal amplitude, and often over a single day (Santos et al., 2009;Yoshida et al., 2018). Previous studies have found that it is the spring-neap tidal cycle, rather than the diurnal tidal stirring, that primarily influences the estuarine physical and chemical properties, primary production and other biological processes through vertical stratification and water movement induced by tidal mixing (Cadier et al., 2017;Heiss and Michael, 2014). Additionally, diel cycle studies showed a dramatic change in viral and microbial activities over the course of a single day (Aylward et al., 2017;Bettarel et al., 2002;Winget and Wommack, 2009;Winter et al., 2004;Yoshida et al., 2018). A weaker tidal impact exerted by diurnal tidal fluctuation combined with the influence of diel light-dark cycle might cause the contrasting observations in how viral and microbial communities respond to tidal stirring (Chauhan et al., 2009;Olapade, 2011;Yoshida et al., 2018). Therefore, to precisely elucidate the tidal impact on viruses and microbes excluding completely the distraction of diel influences, tidal samplings at three stations (upper estuary, lower estuary and coastal area) along the macrotidal Jiulong River estuary (JRE) were conducted at fixed time daily during a consecutive half month to follow an entire spring-neap tidal cycle. The objectives were to (i) investigate the spring-neap tidal patterns in viral and microbial abundances, (ii) evaluate the coupling of viral lytic, lysogenic properties and decay associated with bacterial production and community composition in the JRE, and (iii) explore the potential influence of viral life strategies induced by tidal mixing on the estuarine microbial food web. The findings of this study might provide crucial insights into tidally driven estuarine viral and microbial dynamics and may help to predict the responses of virus-microbe interactions to pressures such as increased riverine input projected for global climate change (Bauer et al., 2013).

Study sites and sampling
The JRE located in the northern South China Sea is a subtropical macrotidal estuary. Tides in the JRE are predominately semi-diurnal with a mean tidal range of 4.08 m, varying from 1.0 m to 6.42 m (Huang and Hong, 1999;Zuo et al., 2016). The estuary continuously receives large quantities of nutrient-rich freshwater input (1.17 Â 10 10 m 3 per year) from the Jiulong River. The river is fed mainly by summer monsoon from April to July and there is no dam in the upstream to regulate the river (Huang and Hong, 1999;Li et al., 2011). Starting in 2012, we conducted long-term monthly observations at 15 stations in the JRE (Fig. 1A) and three stations along a gradient of hydrologic and biological features that were selected according to our previous investigations (Cai et al., 2016;Liu et al., 2017). During the monsoon season, selected stations were sampled every 2 day at a specified time points (10:45, 12:00 and 13:45 at stations S03, S05 and S07, respectively) from April 2e19, 2015 (Fig. 1B). The sampling followed an entire spring-neap tidal cycle, with the tidal range varying from 3.51 m (April 12) to 5.66 m (April 19). Based on the distinct differences in the daily variations of tidal range and salinity by analysis of variance (ANOVA, p < 0.05), April 4e6 (following the full-moon) and April 17e19 (following the new-moon) were classified as the spring tide period, and April 10e12 (following the one-quarter moon) was classified as the neap tide period, with the remaining days classified as the transition tide period (Fig. 1C).
Approximately 5 L of surface water sample (0.5 m depth) was collected into acid-cleaned polycarbonate bottles and was treated immediately for subsequent analysis. The samples for biological analysis were pre-filtered through 20 mm mesh filters to remove large particles and zooplankton. To determine the viral and microbial abundances by flow cytometer, 2 mL subsamples were fixed with glutaraldehyde to a final concentration of 0.5% at room temperature for 15 min in the dark. After the flash freezing in liquid nitrogen, samples were stored at À80 C for later analysis.

Viral and microbial abundances
Viral abundance was determined by flow cytometer (Epics Altra II, Beckman Coulter, USA) with SYBR Green I stain (Molecular Probe, USA) according to the established protocol (Marie et al., 1999).
Fluorescent beads (Molecular Probes) with a diameter of 1 mm were added as an internal standard. Autotrophic Synechococcus and picoeukaryotes were counted directly without staining by the flow cytometer (BD Accuri C6, USA). Heterotrophic bacterial abundance was also determined by flow cytometer (BD Accuri C6) after staining with SYBR Green I (Marie et al., 1999). All the flow cytometric data analyses were performed with the FlowJo vX.0.7 software (Tree Star, USA).

Bacterial production
Bacterial production (BP) was estimated by 3 H-Leucine incorporation following the method described with some modifications (Kirchman, 2001). Briefly, triplicate 1.5 mL water samples and one trichloroacetic acid (TCA) killed control (1% final concentration) were inoculated with 3 H-Leucine (specific activity, 16 Ci mmoL À1 ) to a final concentration of 20 nM. After 0.5e2 h incubation in the dark at in situ temperature, incubation was stopped by adding 5% TCA (final concentration) and centrifuged at 16,000Âg for 10 min. The supernatant was extracted twice with 5% ice-cold TCA and rinsed with 80% ethanol. Later, radioactivity was measured with a liquid scintillation counter (280 TR, Waltham, USA) and BP rates were calculated as the average rates of the triplicate leucine incorporation rates corrected by the control.

Viral production and infection rate
To estimate viral production (VP) and infection rate, the viral reduction approach was used ( Fig. S1) (Weinbauer et al., 2002Winget et al., 2005). Approximately 600 mL water sample was filtered using tangential flow filtration (TFF) with a 0.22 mmpore-size polyvinylidene difluoride cartridge (Labscale, Millipore, USA) to obtain 50 mL bacteria concentrate and ca. 500 mL filtrate. The 50 mL bacteria concentrate was diluted with 250 mL virus-free water obtained from the 0.22 mm filtrate that was sequentially TFF ultrafiltered using a 30 kDa polysulfone cartridge (Millipore) (Cai et al., 2015;Weinbauer et al., 2002). The sample was then gently mixed and 200 mL of the mixture was aliquoted into four 50 mL sterile centrifuge tubes and incubated at in situ temperature in dry bath incubators (MK-20, Hangzhou Allsheng, China) under dark condition (Li et al., 2014). In two of these tubes, mitomycin C (the inducing agent of the lytic cycle) was added to a final concentration of 1 mg mL À1 (Sigma-Aldrich), whereas the other two tubes were kept without addition as controls (Weinbauer et al., 2002). Subsamples of 1 mL in replicate for bacterial and viral enumeration were taken at 3 h intervals during the 15 h incubation. VP and viral infection were calculated using the online program VIPCAL (http:// www.univie.ac.at/nuhag-php/vipcal/) (Luef et al., 2009). The lytic VP was assumed to be equal to the rate of viral accumulation in the control tubes with reduced natural viral abundance, and frequency of lytic infection (FIC) can be calculated as established method (Winter et al., 2004). The lysogenic VP and frequency of lysogenic infection (FLC) were calculated as the difference between viral increase in the tubes with mitomycin C treatments and the control tubes (Weinbauer et al., 2002). The lysogeny-to-lysis ratio was calculated as the ratio of FLC to FIC, to determine the switching index between the lysogenic and lytic viral infections.

Viral decay rate
Viral decay rate (VD) was measured according to Noble and Fuhrman (1997). Seawater was filtered by TFF with a 0.22 mmpore-size cartridge to exclude bacteria and particles >0.22 mm, replicates of 50 mL filtrate were incubated at in situ temperature in the dark (Fig. S1). Subsamples of 1 mL in replicate were taken for bacterial and viral enumeration at 3 h intervals up to 15 h. The VD was calculated as the slope of the linear regression fitted to the natural logarithm of viral abundance plotted against time during 15 h incubation.

Bacterial community composition
For bacterial community composition analysis, 500 mL seawater from each station was collected onto 0.2 mm pore-size polycarbonate filters (47 mm, Millipore) by vacuum filtration. Microbial DNA was extracted using the PowerSoil DNA Isolation Kit (Mo Bio Laboratories, USA) according to the manufacturer's protocol. The V3eV4 region of bacterial 16S rRNA gene was amplified with the primer pair 341F (CCTAYGGGRBGCASCAG) and 806R (GGAC-TACNNGGGTATCTAAT) complemented with sample-specific barcodes. Sequencing libraries were generated using MiSeq Reagent Kit v3 for Illumina (Illumina, USA). Sequences that contained more than one ambiguous nucleotide that did not have a complete barcode and primer at one end, or that were shorter than 200 bp after removal of the barcode and primer sequences were eliminated. Chimera sequences were also identified and removed. After quality filtering, denoising and removal of potential chimeras, sequences were used to study the bacterial community composition. For OTU classification, reads were clustered at 97% similarity and taxonomy was assigned in Mothur (v.1.36.1) using the Silva 119 reference database (Quast et al., 2013;Schloss et al., 2009).

Virus-mediated bacterial mortality and carbon releasing
The virus-mediated bacterial mortality (VMM) was calculated by dividing lytic VP by burst sizes of 50 estimated from previous coastal measurement (Winget et al., 2011). The carbon content of bacterial standing stock was calculated by heterotrophic bacterial abundance and the average marine bacterial cell content of carbon (as a factor of 12.4 fg C cell À1 ) (Fukuda et al., 1998). The rate of lysed bacterial carbon (RLC) and the rate of lysed bacterial produced carbon (RLP) were then estimated from dividing VMM by the bacterial abundance and BP, respectively. To estimate the carbon released by viral lysis of bacteria, the VMM was multiplied by the average bacterial cell carbon content. The carbon not released by viral lysogeny was also estimated by a similar method based on lysogenic VP with the assumption that lysogenic VP would be totally transferred into lytic VP and cause the lysis of bacteria.

Statistical analysis
Shapiro-Wilk W tests for data normality were applied before analysis, and the data were logarithmically transformed to meet normality if necessary. Two-way ANOVA was used to determine the significant differences in environmental and biological parameters with respect to stations and spring-neap tidal periods. If significant differences were observed, the post hoc Tukey's test was also performed. All statistical analyses described above were performed with SPSS 24.0 software (SPSS Inc., USA). Linear regression model was also used to characterize the relationship between abiotic and biotic factors using GraphPad Prism 7 (GraphPad, USA). Principal component analysis (PCA) for segregating sample grouping of physicoechemical parameters after normalizing the data, redundancy analysis (RDA) was performed to test whether the abundances and activities of viruses and microbes were related to environmental and biological variables using CANOCO 4.5 software (Microcomputer Power, USA). The significance of the axes was determined by Monte Carlo permutation tests.

Tidal variability of the sampling environments
Three stations along the JRE were sampled over a spring-neap tidal cycle, and the environmental and biological variables are reported in Table S1. PCA biplot based on physicoechemical parameters distinguished the upper estuarine station S03 from the lower estuarine station S05 and coastal station S07, with S03 showed relatively low values for salinity, pH and DO but high nutrients ( Fig. 1C and Table S1). Salinity, pH and DO at all stations showed a significant tidal signature that peaked at spring tide but reached minimum at neap tide, with a drastic variation at station S03 and moderate variation at stations S05 and S07 (ANOVA, p < 0.05; Fig. 2 and Table S2). An inverse tidal pattern in nutrients was observed at all stations with a stronger change amplitude at station S03, although it is not statistically significant ( Fig. 2 and Table S2). Extremely low values of NO À 2 (0.17 mmol L À1 ) and PO 3À 4 (0.05 mmol L À1 ) were found at S03 at April 17 during the spring tide ( Fig. 2E and G). In addition, Chl a concentration at S03 displayed a significant tidal variation with a higher value at spring tide (ANOVA, p < 0.001; Fig. 2I).

Viral and microbial abundances revealed tidal variability
An apparent and similar tidal pattern among viral and microbial abundances was shared by three stations with a compatible amplitude (ANOVA, p < 0.01; Fig. 3 and Table S2). Viral abundance peaked during spring tide (8.48 ± 0.33 Â 10 6 mL À1 , 7.68 ± 0.38 Â 10 6 mL À1 and 8.13 ± 0.10 Â 10 6 mL À1 at S03, S05 and S07, respectively), and then declined with tidal range to reach its minimum value near neap tide (Fig. 3A). The maximum of heterotrophic bacterial abundances was also recorded at spring tide (Fig. 3B). The virus-to-bacteria ratio during spring tide approximately doubled than the neap tide at all stations (Fig. S2A). The abundances of Synechococcus and picoeukaryotes at three stations significantly increased with the tidal range (ANOVA, p < 0.01; Fig. 3C and D). Furthermore, we observed that tidal variability of heterotrophic bacterial abundance was less pronounced than that of viral, Synechococcus and picoeukaryotic abundances, especially at station S03. RDA analysis revealed the concentrations of viruses, heterotrophic bacteria, Synechococcus and picoeukaryotes were positively correlated with DO, pH and salinity but negatively correlated with nutrient levels (pseudo-F statistic, p < 0.05; Fig. 5A and Table S3).

Tidal variability in bacterial production, viral infection and decay
The rates of BP, VD, VP, FIC and FLC were measured to investigate the dynamics of virus-host interactions. The highest BP at station S03 (6.86 ± 2.07 mg C L À1 d À1 ) occurred when spring tide appeared and the lowest value (3.11 ± 0.22 mg C L À1 d À1 ) was recorded during neap tide, showing a more distinct tide-associated change than the lower estuarine station S05 and the coastal station S07 (Fig. 3E). Estuarine stations S03 and S05 shared a similar spring-neap tidal pattern for VD with the lowest values occurred during spring tide, significantly increasing with the decrease in tidal range (Figs. 3F and 6A). Station S03 showed a more intense tidal change in both lytic and lysogenic viral infection rates. A FIC peak value (22.42 ± 4.77%) was recorded at April 17 during spring tide, and two FLC peak values were observed at April 4 (18.82 ± 3.89%) and April 17 (16.46 ± 2.66%), following spring tide ( Fig. 3G and H). The dynamics of lytic and lysogenic VP at station S03 were similar to the changes in FIC and FLC, resulting in higher total viral production occurred with the higher tidal range (Fig. S2). The lysogeny-to-lysis ratio at three stations shared a similar tidal pattern that a significant higher value was usually recorded near neap tide (ANOVA, p < 0.05; Fig. 3I), showing a significantly negative correlation with tidal range at S03 and S05 (Fig. 6B). RDA analysis revealed the salinity, pH, DO, NO À 3 , PO 3À 4 and heterotrophic bacterial abundance were significantly related to the dynamics of BP, VD, VP, FIC, FLC and the lysogeny-to-lysis ratio during spring-neap cycle (pseudo-F statistic, p < 0.05; Fig. 5B and Table S3). The tide-associated lysogeny-to-lysis ratio had a strongly negative relationship with BP but was positively coupled to VD.

Tidal trends in bacterial community composition
Bacterial community composition was represented by 13 most abundant populations along with other bacterial members at class level according to their relative sequence abundances (Fig. 4A). Different bacterial groups displayed distinct distributions and tidal variations across three stations. For example, Betaproteobacteria, the typical freshwater population (Paver et al., 2018;Riemann et al., 2008), was more abundant at the upper estuarine station S03, coupled with the maximum relative abundance occurring during neap tide, but Bacteroidetes showed the inverse trend (Fig. 4A). Spearman's rank correlation indicated that variations in certain bacterial populations were tightly associated with tidal range, as well as viral, heterotrophic bacterial abundances, BP, lytic and lysogenic viral infections, particularly at station S03 (Fig. 4B). For instance, when the high tidal range turned the hydrologic environment of S03 into a coastal-like feature during spring tide, the relative abundances of Bacteroidetes, Cyanobacteria and Actinobacteria were increased compared to the decreases in Betaproteobacteria and Verrucomicrobia, coupled with increases in viral and heterotrophic bacterial abundances but decrease in the lysogenyto-lysis ratio.

Virus-mediated bacterial mortality and carbon releasing showed tidal specialization
At station S03 during spring tide, VMM was 26.52 ± 17.37 Â 10 7 cells L À1 d À1 together with approximately 30% d À1 of RLC and 70% d À1 of RLP, compared to ca. fourfold lower values of VMM, RLC and RLP at neap tide (Table 1). Hence, more carbon was released by viral lysis during spring tide (3.29 ± 2.15 mg L À1 d À1 ) relative to the neap tide (0.74 ± 0.002 mg L À1 d À1 ) at S03. The amount of carbon not released by viral lysogeny referred to the reduction in lysate carbon due to the establishment of lysogeny rather than lysis. At S03, the amount of carbon not released by viral lysogeny during spring tide (3.24 ± 1.83 mg L À1 d À1 ) was comparable with the carbon released by lysis, but this amount during neap tide (1.88 ± 0.43 mg L À1 d À1 ) was ca. 2e3 times higher than lysis-released carbon. In fact, the ratio of the amount of carbon released by viral lysis and not released by lysogeny was significantly positively correlated with tidal range at all stations, indicating that organic carbon tended to be released into the environment from viral lysis rather than stored as bacterial biomass from neap to spring tide shifting (Fig. 6C).

Discussion
Spring-neap tidal dynamics of physico-chemical variables were observed at all stations, with the amplitude of the change decreased along the estuary (from the upper estuarine station S03 to coastal station S07), and perfectly matched the dilution effect pattern, wherein eutrophic fresh water was diluted by oligotrophic offshore seawater (Fig. 2). Although the samples among three stations were collected at different diurnal tidal phase (e.g., high tide and low tide; Fig. 1B), viral and microbial abundances at all stations shared similar tidal patterns and amplitudes which oscillated synchronously more in relation to the spring-neap tide than diurnal tide (Fig. 3). These results suggested the sensitivity of viruses and microbes to spring-neap tidal dynamic, potentially confirming the primary tidal influence exerted by spring-neap cycle over diurnal tide (Cadier et al., 2017;Heiss and Michael, 2014). Future field work covers more than one entire spring-neap tide along the estuarine habitat, coupled with the corresponding diurnal sampling within each day, might help to distinguish the impacts of spring-neap tide and diurnal tide on microbial ecosystem.
The tidal trend in Synechococcus and picoeukaryotic abundances that increased with tidal range can be predicted to match the previous results that the abundances of estuarine cyanobacteria and picoeukaryote generally increase with salinity (Jochem, 2003;Zhang et al., 2013). However, the observed viral and heterotrophic bacterial abundances (Fig. 3A and B) did not follow the dilution theory (De Brauwere et al., 2011;Koh et al., 1994), that the viral and bacterial abundances may decrease from neap tide to spring tide, since the eutrophic riverine water containing more microorganisms would be diluted at spring tide (Abu-Bakar et al., 2017;Almeida et al., 2001). This tidal pattern implied that, over the spring-neap cycle, the viral and microbial standing stocks and their controlling processes, such as production, mortality and decay, were not exclusively driven by dilution between riverine water and seawater, but rather by the adaption and internal feed-back mechanisms among viral and microbial communities induced by niche partitioning from tidal mixing. The dynamics of microbial community are regarded to under bottom-up control (resource availability) and top-down control (viral lysis or protozoan grazing) (Fuhrman et al., 2015;Menon et al., 2003;Winter et al., 2010). Our results showed that both the concentrations and productions of viruses and heterotrophic bacteria were not positively coupled with nutrients and Chl a during tidal cycle (Fig. 5). In fact, their highest values at S03 occurred during spring tide with the relatively low nutrients (Figs. 2 and 3 and Fig. S2). These results validated that, in eutrophic estuarine-coastal environments where are rich in terrigenous input, the nutrients and primary production play a relatively weaker regulatory role in bacterial activities and the subsequent viral dynamics (Findlay et al., 1991;Iriarte et al., 2003;Mojica and Brussaard, 2014;Weinbauer, 2004).
Our results revealed that lysogenic viral infection and the lysogeny-to-lysis ratio were negatively related to bacterial abundance and production (Fig. 5B). This is consistent with previous finding that lytic viral infection prefers the metabolically active bacteria to produce more viral progeny, whereas lysogeny favors lower bacterial abundance or activity (Maurice et al., 2013;Payet and Suttle, 2013). Despite the presence of more riverine runoffderived nutrients during neap tide at S03, lower BP significantly contributed to lower bacterial abundance (R 2 ¼ 0.695, p ¼ 0.005; linear regression), suggesting the potential inhibition effect of bacterial metabolism at neap tide (Shen et al., 2018). These results can be explained by that bacteria are sensitive to changes in ionic strength, and the drastic variation in salinity during tidal cycle can trigger the shift in physiological stress on bacterial metabolic processes (Bettarel et al., 2011b;Kukkaro and Bamford, 2009;Wei et al., 2019). Freshwater bacteria were previously revealed to be greatly inhibited by seawater (Troussellier et al., 2002), whereas marine bacteria showed relatively high adaptability to variation in salinity (Forsyth et al., 1971). The freshwater-bacteria-dominated microbial community at neap tide suffered from high osmotic pressure, which restricted their metabolic activity to use energyconsuming mechanisms to maintain the osmotic balance in their cells instead of using this energy for cell division (Bettarel et al., 2011b). During spring tide, metabolism in the marine-bacteriadominated microbial community was stimulated by efficient substrate utilization, increasing the bacterial abundance and subsequent lytic viral production and viral abundance (Fig. 3).
Viral production and the counterbalancing loss from viral decay jointly determine the net balance of viral abundance over tidal cycle. Similar to their host, the realized niche of viruses would be constrained by salinity since salt stress may directly affect their survival and successful infection (Wei et al., 2019), potentially through decrease in capsid pressure and consequent reduction in DNA injection efficiency (Cordova et al., 2003;Evilevitch et al., 2008). Viruses have been found to require adequate ionic strength (e.g., Na þ and Mg 2þ ) to maintain structural stability and remain successful infectivity to host (Keynan et al., 1974;Mojica and Brussaard, 2014), demonstrating that a higher salt concentration usually promotes viral adsorption and subsequent infection (Bettarel et al., 2011a;Mei et al., 2015). These can explain the significantly negative relationship between viral decay rate and tidal range at estuarine stations S03 and S05 (Fig. 6A), which reflected higher viral losses induced by the infective inefficiency and high particle decomposition resulting from low salinity at neap tide (Mojica and Brussaard, 2014;Wei et al., 2019). In turn, lower viral decay coupled with higher lytic viral activity led to higher net production of viral particles at spring tide ( Fig. 3 and Fig. S2), Fig. 4. Tidal patterns in (A) the relative abundance of bacterial community at three stations in the Jiulong River estuary at class level and (B) Spearman's rank correlation coefficient between the relative abundance of bacterial community and tidal range as well as salinity, viral and microbial parameters. Class below 0.1% of total abundance are represented as others. TR, tidal range; VA, viral abundance; HBA, heterotrophic bacterial abundance; VBR, virus-to-bacteria ratio; BP, bacterial production; FIC, frequency of lytic infection; FLC, frequency of lysogenic infection and LysoLysis, the lysogeny-to-lysis ratio.
determining the budget for viral abundance over tidal cycle with higher values recorded at spring tide but lower during neap tide. There was a strong negative association between the lysogeny-to-lysis ratio and tidal range at S03 and S05, with a significant higher lysogeny-to-lysis ratio in response to lower levels of bacterial productivity and abundance but higher viral decay at neap tide  Table S3). Sample numbers 1 to 9, 10 to 18 and 19 to 27 are the samples from stations S03, S05 and S07, respectively. The symbols with different color and shapes mean the sampled station and the tidal periods. DO, dissolved oxygen; Sal, salinity; NO2, NO À 2 ; NO3, NO À 3 ; Si, SiO 2À 3 ; VA, viral abundance; HBA, heterotrophic bacterial abundance; VBR, virus-to-bacteria ratio; Syn, Synechococcus abundance; Peuk, picoeukaryotic abundance; BP, bacterial production; VP, viral production; VD, viral decay rate; FIC, frequency of lytic infection; FLC, frequency of lysogenic infection and LysoLysis, the lysogeny-to-lysis ratio. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 6. Linear relationships between tidal range and (A) the viral decay rate, (B) the lysogeny-to-lysis ratio and (C) the ratio of carbon released by lysis and not released by lysogeny. The colored best-fit lines with 95% confidence intervals from linear regression are shown and the lines with a significance level of p < 0.05 are bold on the graph.  (Mojica and Brussaard, 2014;Weinbauer, 2004). Additionally, viral lyticlysogenic switch was generally constrained by bacterial metabolism that greatly impacted by surrounding environmental conditions (Maurice et al., 2010b), owing to viruses rely heavily on the hosts being healthy enough to complete their lytic cycle (e.g., the synthesis of offspring particle constituents) (Mei et al., 2015). For example, the marine temperate phage fHSIC was found to enter the lysogenic life cycle due to a reduction in its host's growth rate at low salinity (Williamson and Paul, 2006). Indeed, host HSIC-1a cells growing at low salinity were found to produced ca. two orders of magnitude fewer phages than those growing at high salinity (Long et al., 2007). The low metabolic activities of microbes under high osmotic stress from spring to neap tide shifting may limit their susceptibility to lytic viral infection, causing the viral infection switching from lytic to lysogenic cycle and thus the higher values of the lysogeny-to-lysis ratio at neap tide. Lysogeny is generally considered as the favorable viral survival strategy during episodes of low energy resources (Paul, 2008;Weinbauer, 2004). Indeed, the lysogeny-to-lysis ratio was significantly negative with heterotrophic bacterial abundance over spring-neap tidal cycle in the JRE (R 2 ¼ 0.308, p < 0.01; linear regression), suggesting that the high density of microbes at spring tide did not drive viral lytic to lysogenic switching (Knowles et al., 2016). Ultimately, the harsh environmental conditions facing drastic tidal forcing resulted in lower bacterial activities and tended to drive the viral infection switching towards lysogeny. Unlike seasonal patterns in viral lysis and lysogeny were usually driven by the trophic status (Brum et al., 2016;Payet and Suttle, 2013), tidal stirring generating fast changing in surrounding habitat would trigger a highly variable amplitude in the host bacteria's ecophysiological parameters, which might greatly influence the patterns in lysis-lysogeny switch over tidal cycle (Maurice et al., 2013(Maurice et al., , 2010a. The tidal shift in viral life strategies in conjunction with change in high viral decay, led to the lower viral lytic activities and bacterial mortality, which would help both virus and host survive at neap tide (Paul, 2008).
The tidal transition in bacterial community composition may also affect the viral infection pattern since host phylogeny has been found to be tightly related to the viral life strategies (Keshri et al., 2017;Kim and Bae, 2018;Maurice et al., 2010b). Tidal disturbances might act as a filter for bacterial community adaptions, resulting in the "loss" of some species but the "activation" of others among freshwater and seawater bacterial populations (Chauhan et al., 2009). Spearman rank correlation revealed that lysogenic viral infection and the lysogeny-to-lysis ratio at station S03 generally increased along with the increases in certain bacterial populations such as Betaproteobacteria and Verrucomicrobia that were previously confirmed as the typical freshwater species (Paver et al., 2018;Riemann et al., 2008), but showed negative correlations with Bacteroidetes and Cyanobacteria (Fig. 4B). These results suggested that viral lytic and lysogenic properties may be contributed partially by the change in viral infection ability induced by tidal shift in freshwater and marine microbial communities. The coexistence of freshwater and marine viruses has been found in the JRE (Cai et al., 2016;Liu et al., 2017). Previous studies have also Fig. 7. Schematic overview depicting the tidally driven viral and microbial dynamics. When the riverine regime dominates the estuarine ecosystem during neap tide, the low metabolic activities of host bacteria together with high level of viral decay jointly lead to the viral infection inefficiency and the lysis to lysogeny switching, resulting in lower viral particle production and subsequent lower efficiency of viral shunt. During the tide transition stage, the moderate pressure for viral infection, viral decay and microbial metabolism help to maintain a trade-off between viral lysis and lysogeny. The promotion of host bacterial metabolism and lower viral decay would be induced owing to the lower salt pressure when marine water dominates the ecosystem during spring tide, resulting in the lysogeny to lysis switching, high viral production and the higher efficiency of viral shunt. Marine viruses and microbes were represented by blue color, whereas freshwater viruses and microbes were represented by green color. DOM, dissolved organic matter. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) revealed that the freshwater viral community might be able to efficiently infect marine hosts, whereas the freshwater microbes did not act as good hosts for the marine viruses (Bonilla-Findji et al., 2009;Xu et al., 2014). Consequently, the reduced infection efficiency due to the marine viruses to freshwater bacterial members would also be a potential factor causing the higher level of lysogeny during neap tide than during spring tide.

Implicates to estuarine carbon cycle
Tidal dynamics of viral lysis and lysogeny were highly variable at the upper estuarine station S03, resulting in higher carbon released by lysis during spring tide than neap tide (Table 1). This high viral lysis at spring tide enhanced the transfer of microbial biomass into the pool of dissolved organic matter, and diverted organic matter flow away from higher trophic levels, thus accelerating the transformation of carbon from particulate (e.g., living organisms) to dissolved states (Fuhrman, 1999;Jiao et al., 2010). The ratio of carbon released by viral lysis and not released by lysogeny significantly increased with tidal range at all stations (Fig. 6C), indicating lysogeny would lower the efficiency of viral shunt and keep more carbon flux within microbial biomass at neap tide, channeling most of the carbon to higher trophic levels in the food web (Wommack and Colwell, 2000). Totally, we found lysogeny reduce 1.64 ± 1.21 mg L À1 d À1 carbon into environmental carbon pool in JRE (Table 1), which was higher than the amount released by lysis (1.36 ± 1.22 mg L À1 d À1 ). Overall, our findings indicated that the differential virus-mediated carbon flux induced by lysis-lysogeny switch and by tidal shifting over spring-neap cycle, neglected so far, must be accounted in the modelling of the microbial food web and biogeochemical processes in estuarine-coastal systems (Sheyn et al., 2018).

Conclusions
Taken together, this study presented a comprehensive assessment of estuarine viral and microbial dynamics over an entire spring-neap tidal cycle (as summarized in Fig. 7). We demonstrated that: An apparent spring-neap tidal pattern in viral and microbial abundances was recorded, with relatively higher abundances observed at spring tide compared to neap tide. Tidal dynamics in viral and microbial communities were shaped by viral lysis-lysogeny switch induced by tidal mixing, with higher level of lysogeny at neap tide compared to the higher level of lysis at spring tide. The switched viral life strategies during spring-neap tidal cycle would cause an intense influence on microbial food web and carbon cycle in the estuarine ecosystems.

Author contributions
RZ and WW initiated and planned the work, XC, WW, JW, HL, JS and RM carried out sampling and performed research, XC and WW analyzed data and XC and RZ wrote the manuscript with input from all authors.

Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.