Tidal variability of nutrients in a coastal coral reef system influenced by groundwater

To investigate variation in nitrite, nitrate, phosphate, and silicate in a spring–neap tide in a coral reef system influenced by groundwater discharge, we carried out a timeseries observation of these nutrients and 228Ra, a tracer of groundwater discharge, in the Luhuitou fringing reef at Sanya Bay in the South China Sea. The maximum 228Ra, 45.3 dpm100L−1, appeared at low tide and the minimum, 14.0 dpm100L−1, appeared during a flood tide in the spring tide. The activity of 228Ra was significantly correlated with water depth and salinity in the spring–neap tide, reflecting the tidal-pumping feature of groundwater discharge. Concentrations of all nutrients exhibited strong diurnal variation, with a maximum in the amplitude of the diel change for nitrite, nitrate, phosphate, and silicate in the spring tide of 0.46, 1.54, 0.12, and 2.68 μM, respectively. Nitrate and phosphate were negatively correlated with water depth during the spring tide but showed no correlation during the neap tide. Nitrite was positively correlated with water depth in the spring and neap tide due to mixing of nitrite-depleted groundwater and nitrite-rich offshore seawater. They were also significantly correlated with salinity (R2≥ 0.9 and P < 0.05) at the ebb flow of the spring tide, negative for nitrate and phosphate and positive for nitrite, indicating the mixing of nitritedepleted, nitrateand phosphate-rich less saline groundwater and nitrite-rich, nitrateand phosphate-depleted saline offshore seawater. We quantified variation in oxidized nitrogen (NOx) and phosphate contributed by biological processes based on deviations from mixing lines of these nutrients. During both the spring and neap tide biologically contributed NOx and phosphate were significantly correlated with regression slopes of 4.60 (R2= 0.16) in the spring tide and 13.4 (R2= 0.75) in the neap tide, similar to the composition of these nutrients in the water column, 5.43 (R2= 0.27) and 14.2 (R2= 0.76), respectively. This similarity indicates that the composition of nutrients in the water column of the reef system was closely related with biological processes during both tidal periods, but the biological influence appeared to be less dominant, as inferred from the less significant correlations (R2= 0.16) during the spring tide when groundwater discharge was more prominent. Thus, the variability of nutrients in the coral reef system was regulated mainly by biological uptake and release in a spring–neap tide and impacted by mixing of tidally driven groundwater and offshore seawater during spring tide.

Abstract.To investigate variation in nitrite, nitrate, phosphate, and silicate in a spring-neap tide in a coral reef system influenced by groundwater discharge, we carried out a timeseries observation of these nutrients and 228 Ra, a tracer of groundwater discharge, in the Luhuitou fringing reef at Sanya Bay in the South China Sea.The maximum 228 Ra, 45.3 dpm 100 L −1 , appeared at low tide and the minimum, 14.0 dpm 100 L −1 , appeared during a flood tide in the spring tide.The activity of 228 Ra was significantly correlated with water depth and salinity in the spring-neap tide, reflecting the tidal-pumping feature of groundwater discharge.Concentrations of all nutrients exhibited strong diurnal variation, with a maximum in the amplitude of the diel change for nitrite, nitrate, phosphate, and silicate in the spring tide of 0.46, 1.54, 0.12, and 2.68 µM, respectively.Nitrate and phosphate were negatively correlated with water depth during the spring tide but showed no correlation during the neap tide.Nitrite was positively correlated with water depth in the spring and neap tide due to mixing of nitrite-depleted groundwater and nitrite-rich offshore seawater.They were also significantly correlated with salinity (R 2 ≥ 0.9 and P < 0.05) at the ebb flow of the spring tide, negative for nitrate and phosphate and positive for nitrite, indicating the mixing of nitritedepleted, nitrate-and phosphate-rich less saline groundwater and nitrite-rich, nitrate-and phosphate-depleted saline offshore seawater.We quantified variation in oxidized nitrogen (NO x ) and phosphate contributed by biological processes based on deviations from mixing lines of these nutrients.During both the spring and neap tide biologically contributed NO x and phosphate were significantly correlated with regres-sion slopes of 4.60 (R 2 = 0.16) in the spring tide and 13.4 (R 2 = 0.75) in the neap tide, similar to the composition of these nutrients in the water column, 5.43 (R 2 = 0.27) and 14.2 (R 2 = 0.76), respectively.This similarity indicates that the composition of nutrients in the water column of the reef system was closely related with biological processes during both tidal periods, but the biological influence appeared to be less dominant, as inferred from the less significant correlations (R 2 = 0.16) during the spring tide when groundwater discharge was more prominent.Thus, the variability of nutrients in the coral reef system was regulated mainly by biological uptake and release in a spring-neap tide and impacted by mixing of tidally driven groundwater and offshore seawater during spring tide.

Introduction
Coral reefs are considered to be one of the most sensitive and stressed ecosystems occupying the coastal zone (Ban et al., 2014).Groundwater input to coral reefs has been shown to be globally important and carry a significant amount of terrestrially derived nutrients to the reef systems (D'Elia et al., 1981;Paytan et al., 2006;Houk et al., 2013).Groundwater discharge is usually enriched in N relative to P with an N : P ratio higher than the Redfield ratio, 16 : 1 (Redfield, 1960), because of more efficient immobilization of P than N in coastal aquifers (Slomp and Van Cappellen, 2004).Such groundwater characterized by a high N : P ratio thus could have significant impacts on coastal reef ecosystems, consid-Published by Copernicus Publications on behalf of the European Geosciences Union.
ering that benthic marine plants are much more depleted in P, with an N : P ratio of about 30 : 1 (Atkinson and Smith, 1983).Cuet et al. (2011) have found that the net community production in a coral-dominated fringing reef at La Réunion, France, is sustained by net uptake of new nitrogen from groundwater and net uptake of phosphate from the ocean.
Groundwater flux onto coral reefs was found to fluctuate with the tidal cycle (Lewis, 1987;Santos et al., 2010).The contribution of groundwater discharge to the nutrient budget of adjacent marine waters of coral reefs varies greatly from one site to another around the globe and at each site varies from one tidal state to another (Paytan et al., 2006).However, there is no study to reveal variation in the composition of nutrients from spring to neap tide in reef systems influenced by groundwater.Therefore, questions are posed.(a) In coral reef systems influenced by groundwater, how do the abundance and composition of nutrients vary from spring to neap tide?(b) What contributes to the tidal variation of nutrients in such a system?
To address these questions, this study examined the nutrient variability in a spring-neap tidal cycle in the Luhuitou fringing reef in Sanya Bay, China, during a dry season.Our previous study showed that tidally driven groundwater discharge affected the carbonate system in the Luhuitou fringing reef (Wang et al., 2014).In this reef system, groundwater discharge played a predominant role during the spring tide, and biological activities (including photosynthesis/respiration and calcification/dissolution) dominated during the neap tide in regulating diurnal variation of the carbonate parameters.Time-series observations of nutrients carried out at the same time as for the carbonate parameters in this reef system made this study possible.The naturally occurring radioactive radium isotope, 228 Ra, was utilized as a tracer of groundwater discharge in this study.

Site description
Sanya Bay is a tropical bay situated at the southern tip of Hainan Island, China, in the northern South China Sea under the influence of the Southeast Asian monsoon (Fig. 1).Seasonal monsoons dominate Hainan Island, with northeast winds in November to March and southwest winds in May to September.Rainfall ranges from 961 to 2439 mm yr −1 in 1994-2011, with about 80 % precipitation occurring during May to October (Zhang et al., 2013).The coastal reef timeseries station CT is located at the Luhuitou fringing reef in the southeast of Sanya Bay.There was no rain in the two weeks before our sampling starting on 2 February 2012 and during our 11-day-long sampling period based on data from the nearby meteorological station in the Hainan Tropical Marine Biology Research Station, Chinese Academy of Science.No surface runoff was present during these periods in this area.Surface salinity in Sanya Bay in our sampling period ranged from 33.60 to 33.89 (Wang et al., 2014).Irregular diurnal tides prevail in Sanya Bay, with a mean tidal range of 0.90 m and the largest value of 2.14 m (Zhang, 2001).The Luhuitou fringing reef is a leeward coast with low wave energy in winter (Zhang, 2001).In summer coastal upwelling off the east of Hainan Island mainly induced by the southeast monsoon may extend to this area (Wang et al., 2016).The Holocene deposits of coral debris and biogenic carbonate sands (secondary reef) form the surficial unconfined aquifer around the fringing reef (Zhao et al., 1983), making groundwater a diffuse source of nutrients for the reef system.Macroalgae cover about 60 %, on average, of the bottom hard substrates in the Luhuitou fringing reef (Titlyanov and Titlyanova, 2013).Living scleractinian corals were observed in the lower intertidal zone and subtidal zone with coverage of 5-40 % (Titlyanov and Titlyanova, 2013;Titlyanov et al., 2014Titlyanov et al., , 2015)).Cyanobacteria and Rhodophyta prevailed in the upper intertidal zone, while Rhodophyta and Chlorophyta were the most abundant in the middle and lower intertidal zones (Titlyanov et al., 2014).Rhodophyta dominated the benthic macroalgal community, 54 % in the upper subtidal zone (Titlyanov and Titlyanova, 2013).The number of species in the marine flora has increased by 28 % from 1990 to 2010 with a displacement of slow-growing species likely due to anthropogenic influences and coral bleaching (Titlyanov et al., 2015).The mean coral cover has decreased in the Luhuitou fringing reef from 90 % in the 1960s to 12 % in 2009 (Zhao et al., 2012), likely owing to a combination of regional anthropogenic impacts and climate change (Li et al., 2012).
To the north of the Luhuitou fringing reef, the Sanya River flows into Sanya Bay, with an annual average discharge of 5.86 m 3 s −1 (Wang et al., 2005).The river is fed mainly by southwest monsoons from May to October.There is no dam in the upstream to regulate the river.During our sampling period the Sanya River plume was confined in the northeast of the bay and the coastal reef station CT was outside the influence of the Sanya River plume (Fig. 1) (Wang et al., 2014).Investigations of nutrients, Chl a, and phytoplankton in the bay have been conducted seasonally for several years (Dong et al., 2010;Wu et al., 2011Wu et al., , 2012b, c) , c) and demonstrate that the inner bay is influenced by the discharge of the Sanya River with its relatively high nutrient levels, and the central and outer bay are dominated by oceanic exchange with the South China Sea (Wu et al., 2012a).Nutrients carried by submarine groundwater discharge into Sanya Bay account for at least 79 % of the nutrients into the bay in our sampling period (Wang et al., submitted).The concentrations of nutrients in the saline groundwater near the reef and in the upper stream of the Sanya River estuary during our sampling period were 1.66 ± 0.53 and 8.8 µM phosphate, 142 ± 14 and 36.6 µM oxidized nitrogen (NO x , including nitrate and nitrite), and 237 ± 2 and 271 µM silicate, respectively (Wang et al., 2018).

Sampling and measurements
The setup of the sampling platform at the time-series station CT is provided in detail in Wang et al. (2014).Briefly, water was collected using a submersible pump, and depth and salinity were measured with a conductivity-temperature-depth system (Citadel, RDI Co., USA) attached on a buoy.Discrete nutrient and radium samples were taken every 3 h during 6-13 February 2012, except on 7-8 February when the maximum tidal range of 1.4 m occurred (Wang et al., 2014), and the samples were collected every 2 h.A mapping cruise was conducted in Sanya Bay during 2-3 February 2012 (Fig. 1) to evaluate the influence of the Sanya River and to constrain the endmember of the offshore water.Nutrient samples for nitrate, nitrite, phosphate, and silicate were collected in Sanya Bay at surface and bottom depths using 5 L Niskin bottles.Temperature and salinity were measured using a multiparameter sonde YSI 6600.The salinity was reported using the Practical Salinity Scale.Nutrient samples were filtered with 0.45 µm cellulose acetate membranes and poisoned with 1-2 ‰ chloroform.One filtrate was preserved at 4 • C for dissolved silicate determination, and one was frozen and kept at −20 • C for nitrate, nitrite, and phosphate measurements.In the laboratory, nutrients were measured with an AA3 Auto-Analyzer (Bran-Luebbe, GmbH), following the same methods in Han et al. (2012).The analytical precision was better than 1 % for nitrate and nitrite, 2 % for phosphate, and 2.8 % for silicate.The detection limit was 0.04 µM for nitrate and nitrite, 0.08 µM for phosphate, and 0.16 µM for silicate.Blanks were directly set up as baselines during the measurements and subtracted.Radium samples of 30 L seawater were passed through a 1 µm cartridge filter before they were passed through an MnO 2 -impregnated acrylic fiber (Mn fiber) column to extract dissolved radium (Rama and Moore, 1996).The Mn fibers were leached with 1 M solutions of hydroxylamine hydrochloride and HCl to release 226 Ra and 228 Ra, which were then co-precipitated with BaSO 4 and measured in a germanium gamma detector (GCW4022, Canberra) (Moore, 1984) with an error of less than 7 %.

Linear regression and contour plotting
To gain insight into factors affecting nutrients from spring to neap tide, linear regressions were conducted between water depth, salinity, and 228 Ra activity, between water depth, salinity, and nutrients' concentration, and between biologically contributed nutrients during the spring and neap tide.A significance level of 0.05 was taken.The data were fit using SigmaPlot (Systat Software, San Jose, California, USA; https://systatsoftware.com/).In plotting contours in Sanya Bay, Surfer 11 was utilized with kriging interpolation due to its good linear unbiased prediction of the intermediate values in spatial analysis (Papritz and Stein, 2002).

Time-series observations of nutrients and radium at the coastal coral reef station
Time-series observations of salinity, 226 Ra, and water depth at station CT were reported in Wang et al. (2014), which demonstrated that the water depth at station CT varied from 0.7 to 2.1 m and the salinity ranged from 33.43 to 33.67 during 6-13 February 2012 (Table S1).The greatest tidal range occurred on 7 February 2012 (Wang et al., 2014), the 16th of the lunar month.To separate neap tide from spring tide days, the daily variance of water depth and salinity were plotted (Fig. 2).The daily variance of a variable was calculated as where x is the average of the variable in a day and n is the number of samples of the variable in that day.A sharp decrease in the variance of salinity occurred on 10 February 2012 and the variance remained low (< 0.001) afterwards.Thus, two distinctive groups stood out, with one group in the period of 6-9 February 2012 having greater variance of water depth and salinity and the other in the period of 10-13 February 2012 having less variance.Therefore, we took 6-9 February 2012 as the spring tide period and 10-13 February 2012 as the neap tide period in this work.
The concentration of nutrients varied with different patterns from spring to neap tide (Fig. 3).Nitrite varied from 0.11 to 0.71 µM during the spring tide and from 0.12 to 0.74 µM in the neap tide, with the maximum diel variation of 0.46 µM present during the spring tide (Fig. 3a).The diurnal variation was 0.24-0.46µM during the spring tide and 0.34-0.45µM in the neap tide.Daily peaks of nitrite usually appeared at high tide from the spring to neap tide.The concentration was positively correlated with water depth (P < 0.05) during both the spring and neap tide, but the correlation was less significant during the neap tide (Fig. 4a).Nitrate and phosphate, however, showed an opposite pattern.During the spring tide, nitrate and phosphate were negatively correlated with water depth (P < 0.05) (Fig. 4b and c).They reached their peak concentrations of 1.91 and 0.22 µM, respectively, in the late afternoon and their minima of 0.37 and 0.10 µM, respectively, at night on 7 February 2012 (Fig. 3b and c).The diurnal variation fell in the range of 0.44-1.54µM for nitrate and 0.04-0.12µM for phosphate.During the neap tide, the concentrations varied from 0.27 to 1.32 µM for nitrate and 0.084 to 0.18 µM for phosphate, with less diurnal variation in the range of 0.35-0.52µM for nitrate and 0.04-0.05µM for phosphate.The correlation with water depth was not significant for both nutrients (P > 0.15).Nitrate is the dominant species (> 50 %) of NO x during the spring-neap tidal period except at 02:00 LT on 12 February 2012 when the concentrations of nitrite and nitrate were almost equal.The NO x : P ratio varied from 4.78 to 12.9 in the spring-neap tide (Fig. 3c).Silicate showed a trend different from either nitrite or nitrate and phosphate (Fig. 3d).It was not significantly correlated with water depth during either spring or neap tide (P > 0.2).The concentration of silicate, in general, decreased from spring to neap tide.During the spring tide, the concentration of silicate fell in the range of 4.57-7.25 µM.The daily peak concentration of silicate appeared almost at the daily lowest salinity.The diurnal variation in silicate was 1.91-2.68µM.During the neap tide, however, silicate ranged from 2.89 to 5.59 µM and showed less diurnal variability, 1.44-2.09µM.
The diurnal variation in the activity of 228 Ra at station CT was 16.5-27.4dpm 100 L −1 (i.e., 2.75-4.56Bq m −3 ) during the spring tide, the maximum of which appeared on 7 February, and 5.31-10.6dpm 100 L −1 around the neap tide (Fig. 3e).The maximum 228 Ra, 45.3 dpm 100 L −1 , appeared at low tide on 8 February during the spring tide and the minimum, 14.0 dpm 100 L −1 , appeared during the flood tide of the spring tide on 7 February.The activity of 228 Ra was significantly correlated with water depth in the spring-neap tidal period (P = 0.002) (Fig. 5a).This pattern reflected the variation in the groundwater discharge induced by tidal pumping in this coral reef system (Wang et al., 2014), which is also observed in other coastal regions (Burnett and Dulaiova, 2003;Santos et al., 2010).

Distributions of nutrients in Sanya Bay
In Sanya Bay the highest concentration of nutrients appeared near the Sanya River estuary, and the concentration, in gen-eral, decreased from the northeast coast, where the influence of the Sanya River plume is apparent in winter (Wang et al., 2014), to the south and west, where the South China Sea water intrudes (Fig. 6).At stations far offshore (stations J4-5 and W3-4), the concentrations of nitrite, nitrate, and phosphate were all below the detection limit and the concentration of silicate was about 4.00 µM.At other stations, the concentration of all the nutrients remained low, but was nonetheless detectable.For example, the maximum concentrations of only 0.43 µM for nitrite, 0.70 µM for nitrate, 0.18 µM for phosphate, and 7.92 µM for silicate were recorded at station P1, the station closest to the Sanya River estuary.The small islands in Sanya Bay did not show apparent influence on the nutrients in the bay since nutrients were below their detection limits or remained low around these islands (Fig. 6).The water depth at these mapping stations was no less than 5 m and the concentration of nutrients at the bottom depth differed little from that at the surface at most of these offshore stations (Table 1).This vertical distribution confirms that the water in Sanya Bay is relatively homogenous in February (Wang et al., 2014).The NO x : P ratio was less than 7 in Sanya Bay, except at stations P2 and L6 where the NO x : P ratio was around 9. The time-series observation of salinity at station CT suggests that more freshwater input into the reef system occurred during the ebb flow of the spring tide as inferred from lower salinity than during that of the neap tide (Wang et al., 2014).
The distribution of salinity in Sanya Bay demonstrated that the surface salinity was slightly lower in the northeast off the Sanya River mouth than in southern Sanya Bay (Fig. 1a).At stations P1 and P2 the surface salinity was less than 33.70, while at stations around station CT, i.e., stations L1, L2, L3, and P3, the surface salinity was greater than 33.80 (Table 1).
The vertical profiles of salinity at these stations also showed a similar difference, with salinity below 33.80 throughout the water column at river-influenced stations (Fig. 1b).This indicates that the Sanya River plume affected the northeast of the bay with little impact on station CT and that the only source of freshwater at station CT in February would be groundwater discharge (Wang et al., 2014) since in the two weeks before our sampling and during our sampling period there were no rainfall and consequent surface runoff in this area.The coincidence of the daily minimum salinity with the highest activity of 228 Ra during the ebb flow of the spring tide (Fig. 3e) and the significant correlation between the activity of 228 Ra and salinity during the spring-neap tidal period (P < 0.0001) (Fig. 5b) confirms that the tidally driven groundwater discharge occurred at the coral reef station CT.Greater ground-  water discharge appeared during the ebb flow in the spring tide than in the neap tide as indicated by the higher activity of 228 Ra, bringing more groundwater into the reef system.Under the influence of tidally driven groundwater discharge, variation in nitrite, nitrate, phosphate, and silicate during the spring tide followed a tidal pattern.Inferred from the significant correlation between nutrients and water depth during the spring tide (Fig. 4), the groundwater discharge was characterized by higher nitrate and phosphate and lower nitrite than the offshore seawater.Since nitrate dominated NO x during the time-series observation (Fig. 3), groundwater discharge was characterized by higher NO x and phosphate than the offshore seawater.Because groundwater discharge was greater at low tide than at high tide due to its tidal pump-ing feature, higher NO x and phosphate would appear at low tide.From Fig. 4, the daily maximum concentration of NO x , phosphate, and silicate appeared in the daytime at relatively low tide, while the minimum appeared mostly at night at high tide, indicating the mixing of tidally driven groundwater and offshore seawater.During the neap tide, however, NO x and phosphate showed less diurnal variation.The daily maximum concentration of NO x and phosphate appeared around midnight, when a flood tide appeared.This pattern of daily maximum in a flood tide at night reflected dominance of biological processes because in a flood tide there were fewer groundwater-associated nutrients and nutrients were released the most at night by biological processes.This pattern is consistent with the time-series observation of dissolved oxygen Note that BDL refers to below the detection limit.
at this site (Wang et al., 2014).The daily minimum appeared for NO x and phosphate in the afternoon or between midnight and dawn at high tide, reflecting the dominance of nutrientdepleted offshore seawater.Adsorption/desorption from particles might be a factor influencing the phosphate concentration, as proposed for estuaries (e.g., Froelich et al., 1982;van der Zee et al., 2007).At the reef station the salinity was close to the seawater (> 33) and the water was clear (the total suspended matter was low, about 15 mg L −1 ), which makes adsorption/desorption negligible.The clear water, as well as low wave energy in the reef in winter (Zhang, 2001), also limits the possibility of sediment resuspension being a source of radium and nutrients.
Under the controls of tidally driven groundwater discharge and biological processes, the composition of nutrients in the reef system also differed from the spring tide to the neap tide.During the spring tide when groundwater discharge played a predominant role in regulating the concentration of nutrients in the reef system, the concentration of NO x was positively correlated with the concentration of phosphate, with a regression slope of 5.43 and R 2 of 0.27 (Fig. 7a).The concentration of silicate was not significantly correlated with the concentration of NO x (Fig. 7b).During the neap tide when groundwater discharge was less prominent, the correlation between the concentrations of NO x and phosphate was more significant, with a regression slope of 14.2 and R 2 of 0.76.The NO x : P ratio was closer to the Redfield ratio than during the spring tide.The concentration of silicate showed significant correlation with the concentration of NO x in the water column, with a regression slope of 1.24 and R 2 of 0.58.Diatoms dominate the phytoplankton community in Sanya Bay (Zhou et al., 2009).The elemental ratio of Si : N is 0.80 ± 0.35 for nanoplankton and 1.20 ± 0.37 for net plankton (Brzezinski, 1985).The similarity of the composition of silicate and NO x in the water column to the elemental ratio of diatoms implies a biological control.Unfortunately, no information is available on particular reef primary producers and sponges that may take up/release silicate in this reef system to further the discussion.The activity of 228 Ra, however, was not significantly correlated with the NO x : P ratio in the water column from spring to neap tide (P > 0.05) (Fig. 5c), indicating that the composition of nutrients in the water column was not predominantly controlled by groundwater discharge.Therefore, we propose that biological processes predominantly controlled the composition of nutrients in the reef system but that there was less of an impact in the spring tide due to groundwater discharge.

The generation and consumption of NO
x and phosphate at the reef station CT N and P are the general limiting nutrients for the abundance of phytoplankton in coastal ecosystems (Jickells et al., 1998).
To quantify the contribution of biological processes to the variation in the NO x and phosphate at station CT, a closer look was taken at the behaviors of nitrite, nitrate, and phosphate with salinity during the falling and rising phases on February 7, the day with the greatest tidal range in the spring tide period.Figure 8 shows that these nutrients behaved differently during the two phases.During the ebb flow, with a fast falling speed as indicated by the sharp slope of water depth (Fig. 3), nitrite, nitrate, and phosphate behaved conservatively; i.e., their concentrations were significantly correlated with salinity (P < 0.05).Nitrite was positively correlated with salinity (R 2 = 0.94), while nitrate and phosphate were negatively correlated with salinity (R 2 = 0.91 and 0.90, respectively) (Fig. 8).These conservative behaviors indicated mixing between the groundwater discharge and the offshore seawater.During the flood tide, with a relatively slow speed as indicated by a smaller slope of water depth (Fig. 3), however, nitrite showed an apparent removal signal relative to the conservative mixing line while additions of nitrate and phosphate appeared.This consumption of nitrite and generation of nitrate and phosphate were due to biological processes in this period.Based on the conservative mixing lines shown in Fig. 8, we could estimate nitrite, nitrate, and phosphate owing to mixing of the offshore seawater and groundwater discharge using the salinity measured at station CT (S CT ), designated as NO 2mix , NO 3mix , and P mix : Two assumptions were made before setting up these equations: (a) there was no other water mass flowing into the reef system besides offshore seawater and groundwater, and (b) mixing of offshore seawater and groundwater from spring to neap tide followed the relation derived from data on the day with the greatest tidal range.The differences between the measured concentrations of nutrients and the nutrient concentrations resulting from mixing represented nutrients contributed by biological processes, designated as NO 2bio , NO 3bio and P bio : where the subscript "CT" represents the measured value at station CT.The oxidized nitrogen contributed by biological processes, NO xbio , is the sum of NO 2bio and NO 3bio .
Positive values represent regeneration and release of nutrients in the water column, and negative values reflect uptake of nutrients by marine flora (including phytoplankton and benthic flora in this system).Benthic release due to remineralization of organic matter contributes to the positive values.The nutrients contributed by biological processes showed the greatest diurnal variation in nitrate and phosphate on 7 February 2012, which is in the spring tide, while the  maximum of biologically contributed nitrite appeared on 12 February 2012, which is in the neap tide (Fig. 9).Nitrite contributed by biological processes ranged from −0.15 to 0.39 µM during the spring tide and from −0.20 to 0.40 µM during the neap tide (Fig. 9a).From 18:00 on 8 February to 18:00 on 11 February 2012, biologically contributed nitrite was positive throughout the period, indicating production of nitrite.For nitrate it was produced throughout the period from 04:00 on 8 February to midnight on 11 February 2012.During the spring tide biologically contributed nitrate varied from −0.24 to 1.25 µM and during the neap tide it fell in the range of −0.38 to 0.70 µM.Net NO x production oc- Water depth was reported in Wang et al. (2014).
curred from 18:00 on 8 February to 08:00 on 12 February 2012 and NO xbio was negative afterwards on 12-13 February 2012, indicating net consumption (Fig. 9b).The biological contribution of phosphate had greater diurnal variation during the spring tide than during the neap tide (Fig. 9c).The greatest diel variation during the spring tide in P bio appeared on 7 February 2012 when P bio varied from −0.027 to 0.088 µM, while during the neap tide the greatest variation occurred on 10 February 2012 when P bio ranged from 0.009 to 0.056 µM.Net phosphate consumption occurred throughout the period of 12-13 February 2012.
The relationship between NO xbio and P bio during the spring tide differed from that during the neap tide.Note that NO x is not equivalent to dissolved inorganic nitrogen (DIN).Ammonium data are not available in this study for us to discuss the relationship between DIN and phosphate, which limited our discussion to NO x and makes the picture of microbial processes that control dissolved inorganic nitrogen in the coral reef system incomplete.However, tidal variation in ammonium is expected at this site and may have been related to the change in the relationship between NO xbio and P bio .During the spring tide there was significant correlation between NO xbio and P bio , with a regression slope of 4.60 and R 2 of 0.16 (Fig. 10).During the neap tide, however, the correlation was much more significant with a regression slope of 13.4 and R 2 of 0.75.The regression slope of the regression between biologically contributed NO xbio and phosphate was similar to that of the significant regression between NO xbio and phosphate in the water column, which was 5.43 during the spring tide and 14.18 during the neap tide.This similarity indicates that the composition of nutrients in the water column was closely related with biological processes during both tidal periods, but the biological effect appeared to be less evident during the spring tide as inferred from the less significant correlations.The net release of nutrients during the neap tide with a very Redfield-like ratio suggests that the net nutrient fluxes in this system were likely to be dominated by the uptake and remineralization of plankton/oceanic organic particles by benthic filter feeders as observed in other reefs (e.g., Ayukai, 1995;Ribes et al., 2005;Southwell et al., 2008;Genin et al., 2009;Monismith et al., 2010).The net uptake of nitrate and phosphate was mainly done by reef primary producers.Thus, the composition of nutrients in the water column seemed to be directly related with biological contributions from the spring to neap tide.The biological influence was less dominant during the spring tide, most likely due to groundwater discharge.This confirms our proposal that biological processes predominantly controlled the composition of nutrients in the reef system but that there was less of an impact due to groundwater discharge.
Successive uptake rates of NO x were approximated by the depth integration of the biologically contributed NO x divided by the sampling time interval from the spring to neap tide.The uptake rate ranged from −9.04 to 19.1 mmol m −2 d −1 , which compares well with the sum of nitrate and nitrite fluxes over Ningaloo Reef, a fringing reef in Australia, −24 to 15 mmol m −2 d −1 (Wyatt et al., 2012).It is significantly correlated with the concentration of NO x in the water column (Fig. 11), with a slope of 14.5 and R 2 of 0.94 (P < 0.0001), indicating the mass-transfer limitation of NO x uptake.The slope (in m d −1 ) falls in the range of the typical uptake rate coefficient for dissolved inorganic nitrogen reported in Falter et al. (2004).Corals may be capable of adaptive changes in uptake kinetics depending on nutrient availability.However, the rate of nitrogen acquisition appeared to be influenced on a diel cycle in the coral reef system, presumably due to depletion of photosynthetic products during the night.

Seasonal and regional extrapolations
This study was carried out in winter.Seasonal variation is present in the river discharge as inferred from precipitation (Wang et al., 2005) and there might be an increase in the groundwater discharge and associated nutrient fluxes in summer as in other coastal systems (e.g., Lewis, 1987;Costa et al., 2006;Kelly and Moran, 2002;Wang et al., 2015).However, the relative changes in the groundwater discharge and associated nutrient fluxes would be much smaller than those of the river.The tidally driven feature of the groundwater discharge in this reef system might make our conclusions applicable to other seasons.But it is likely that what we observed in a dry season might be different from what would happen in a wet season due to the involvement of other forces, e.g., upwelling in summer (Wu et al., 2012b;Wang et al., 2016), which merits further studies.
In relatively oligotrophic coastal systems with coral reefs, such groundwater-associated nutrient fluxes may sustain the reef community production (Cuet et al., 2011), result in increases in diversity and occurrence of algae and sponge where relatively low salinity is present (Houk and Starmer, 2010), or induce the proliferation of diatom and cyanobac-Biogeosciences, 15, 997-1009, 2018 teria (Blanco et al., 2011).In addition, groundwater tidally driven into nearshore ecosystems was found to be negatively correlated with seagrass habitat conditions (Houk et al., 2013).Nutrient loads via groundwater discharge may affect the community structure to move towards macroalgal blooms via bottom-up control (Lapointe, 1997) and likely play a role in the displacement of slow-growing benthic flora with fast-growing species observed in Sanya Bay in the last two decades (Titlyanov et al., 2015).Future changes in these fluxes, likely caused by climate change and human activities, might make the situation worse and need to be monitored in reef protection programs and be considered in assessing the environmental health of coral reef systems, especially in regions with expected higher inputs of anthropogenic nutrients into the groundwater.

Conclusions
The variability of nutrients in a spring-neap tidal cycle in a coral reef system in winter was revealed for the first time under the synergistic control of tidally driven groundwater discharge and biological processes.The activity of 228 Ra was significantly correlated with water depth and salinity, indicating tidally driven groundwater discharge at this site.Nitrate and phosphate were negatively correlated with salinity at the ebb flow of the spring tide, indicating that groundwater discharge was enriched in nitrate and phosphate.Nitrate, phosphate, and silicate in the water column showed greater diurnal variation during the spring tide than during the neap tide, while the diel change in the concentration of nitrite demonstrated no consistent pattern.The nutrient composition in the water column seemed to differ between the spring tide and neap tide but was similar to their biological uptake/release in either tidal period for oxidized nitrogen (NO x ) and phosphate.This similarity indicates that variation in nutrients in the water column in the reef system was mainly regulated by biological processes.However, correlations between NO x and phosphate in the water column and between biologically contributed NO x and phosphate were less significant during the spring tide when groundwater discharge was more prominent.The concentration of silicate in the water column was significantly correlated with that of NO x during the neap tide, but they were not significantly correlated during the spring tide.This indicates that the composition of nutrients in the water column was also affected by tidally driven groundwater discharge, especially during the spring tide.Therefore, biological processes predominantly controlled the composition of nutrients in the reef system but that there was less of an impact in the spring tide due to groundwater discharge.
The stoichiometric relationship of NO x and phosphate from the spring to neap tide in this reef system is important in understanding how biological processes predominantly affected these nutrients' variation under the influence of tidally driven groundwater discharge.The composition of silicate and NO x during the neap tide when groundwater discharge was less prominent was comparable to the elemental ratio of diatoms.The release/consumption ratio of NO x : P by biological processes followed a Redfield-like ratio during the neap tide but was about one-third as much during the spring tide.Whether this change in the biological release/uptake ratio of NO x : P is associated with a change in the community structure needs further study.

Figure 1 .
Figure 1.Study area, sampling stations and salinity distribution (a) and vertical profiles of salinity (b) in February 2012 in Sanya Bay, Hainan Island (HI) in the South China Sea.HK represents Hong Kong.CT is the coastal reef time-series station.Salinity data were reported in Wang et al. (2014).

Figure 3 .
Figure 3. Time-series observations of nutrients and 228 Ra at station CT in the Luhuitou reef of Sanya Bay, China, during 6-13 February 2012.(a) Nitrite, (b) nitrate, (c) phosphate and NO x : P ratio, (d) silicate, and (e) 228 Ra.Lines connecting the symbols are to show trends.Water depth and salinity were reported in Wang et al. (2014).

Figure 4 .
Figure 4. Concentrations of nutrients in the water column against water depth during the spring tide and neap tide at station CT in the Luhuitou reef during 6-13 February 2012.(a) Nitrite, (b) nitrate, (c) phosphate, and (d) silicate.
affects tidal variation in nutrients at the reef station CT?

Figure 5 .
Figure 5.The activity of 228 Ra against (a) water depth, (b) salinity, and (c) the NO x : P ratio in the water column at station CT during 6-13 February 2012.

Figure 6 .
Figure 6.Surface distributions of nutrients in Sanya Bay in February 2012.(a) Nitrite, (b) nitrate, (c) phosphate, and (d) silicate.The units are in µM.BDL refers to below the detection limit, which is 0.04 µM for nitrate and nitrite and 0.08 µM for phosphate.

Figure 7 .
Figure 7. Concentrations of (a) NO x against phosphate and (b) silicate against NO x in the water column during the spring tide and neap tide at station CT during 6-13 February 2012.

Figure 8 .
Figure 8. Behaviors of nutrients with salinity during the ebb flow and flood tide of the spring tide at station CT.(a) Nitrite, (b) nitrate, and (c) phosphate.

Figure 9 .
Figure 9. Variations of nutrients contributed by biological processes in a spring-neap tide during 6-13 February 2012 at the coastal reef station CT.(a) Nitrite and nitrate, (b) NO x , and (c) phosphate (P).Water depth was reported inWang et al. (2014).

Figure 10 .
Figure 10.Relationship between biologically contributed NO x and phosphate during the spring tide and neap tide at station CT in the Luhuitou fringing reef in 6-13 February 2012.

Figure 11 .
Figure 11.Uptake rate of NO x against the concentration of NO x in the water column at reef station CT in a spring-neap tide during 6-13 February 2012.

Table 1 .
Sampling stations and data collected in Sanya Bay in February 2012.