Carbonate chemistry seasonality in a tropical mangrove lagoon in La Parguera, Puerto Rico

We investigated the seasonal carbonate chemistry variability within a semi-enclosed tropical mangrove lagoon in southwestern Puerto Rico. Biweekly measurements of seawater temperature, salinity, total alkalinity (TA), and dissolved inorganic carbon (DIC) were conducted from 2014 to 2018. We describe the possible mechanisms driving the observed variability by correlating the DIC/TA ratio with pH and Ωarg, suggesting that the mean pH (7.87 ± 0.09) and aragonite saturation state (Ωarg, 2.96 ± 0.47) of the mangrove lagoon negatively affected calcification. The measured pCO2 and DIC/TA ratios indicate that heterotrophic activity was the primary driver for persistent acidification, which reached its maximum expression during the wet season. We conclude that mangrove lagoons with limited seawater exchange and high carbon input will not mitigate ocean acidification.


Introduction
There is an increasing research interest to identify ecosystems where vulnerable species could persist under foreseen future scenarios, including high atmospheric carbon dioxide (CO 2 ) concentrations and seawater temperatures [1]. Surface oceans have already warmed [2,3] and acidified worldwide at an unprecedented pace [4], and scientists expect such trends to continue based on the Intergovernmental Panel on Climate Change (IPCC) CO 2 emission scenarios [5].
Studies suggest that different coastal ecosystems, such as seagrass beds [6] and mangroves [7,8], can mitigate ocean acidification (OA) driven by oceanic uptake of anthropogenic CO 2 . A buffering effect may occur when "alkalinity increases the CO 2 uptake capacity of seawater by neutralizing H + ions and buffering the pH change associated with CO 2 inputs" [9]. Seagrass ecosystems buffer OA at shallow depths and reduced water mass turnover [10,11] by raising the pH and calcium carbonate (CaCO 3 ) saturation while they are in a state of net autotrophy [10]. The buffering effect may be limited to specific periods throughout the year [12] and could have the opposite effect by exacerbating OA in the future [13]. A Caribbean mangrove system (St. John, United States Virgin Islands) was recently identified as a potential refuge from OA, upper-ocean warming, and solar radiation for non-reef a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 scleractinian coral species [8]. The authors suggested that CaCO 3 sediment dissolution could have increased downstream total alkalinity (TA) to prevent OA from exacerbating [8]. Moreover, a study in an Australian mangrove creek suggested that mangrove sites could generate alkalinity via sulfate reduction and exported to nearby coastal areas, creating a zone buffered against OA [7]. Other studies suggest that mangrove ecosystems can pre-condition corals to low pH's [14] and serve as reservoirs for already tolerant coral species [15,16].
Mangrove communities are known for their high productivity [17,18] and are more productive than most terrestrial and marine communities [19]. However, global budgets still fail to fully quantify the amount of carbon fixated by mangrove vegetation, given that datasets and estimates on wood production and belowground allocation are limited [20]. Estimates note that inorganic carbon produced by mineralization activity could account for the missing carbon in global budgets, given that mineralization rates are severely underestimated [20]. Sulfate reduction and aerobic respiration are commonly the dominant metabolic pathways within tropical mangrove sediments [7,21,22]. The former yields alkalinity, assuming that the resulting sulfide is not re-oxidized to sulfate [23]. Coupled denitrification/nitrification and ammonification are restricted pathways in tropical mangrove environments due to scarce oxygen availability [24][25][26].
Studies published to date provide a good understanding of the carbonate chemistry diel variability at these mangrove sites [7,8,[14][15][16]. However, the buffering role of mangroves at seasonal time scales remains unclear. This study's main objective is to determine if the tropical mangrove lagoon located on the southwest coast of Puerto Rico can buffer OA. We also aim to assess TA's seasonal variability, dissolved inorganic carbon (DIC), and physical dynamics (temperature, salinity, precipitation, tides, and winds). We hypothesize that mangrove lagoons with high mangrove carbon mineralization activity and limited flushing provide conditions that promote OA and reduce the aragonite saturation state (Ω arg ).

Site description
The study area is within La Parguera Marine Reserve, an estuary in which evaporation exceeds freshwater influx [27], located on the southwestern coast of Puerto Rico (Fig 1). The Reserve's cays, submerged reefs, and mangroves ecosystems surround and protect La Parguera town (17.974 (Fig 2), 3.2 km east of La Parguera town (Fig 1). The Bioluminescent Bay has a 0.19 km 2 surface area, an average depth of 3.5 m [31], and a maximum of 4.5 m [32]. A 25-meter-wide red mangrove (Rhizophora mangle L.) fringe surrounds the bay and is a significant organic carbon source. The total estimated primary productivity for R. mangle in La Parguera was 3.85 g dry wt m -2 day -1 , while the litterfall fraction was in the 0.15 to 0.91 g dry wt m -2 day -1 range with a marked seasonal variability [33]. The BB water column's net community production has been reported to range  from -5.99 to -2.49 g C m -2 day -1 [34,35]. The bathymetry has an irregular shape with a narrow and shallow outlet to nearby shelf waters and a sill (1 m depth) at the entrance, limiting exchange with outer waters and shelters from wave energy [32,36]. Most of the bottom comprises muddy sediments (ca. 7.5% grams of organic carbon content per gram of sediment, down to a 20 cm depth [37]). While coral reef systems are not known to exist within the bay, small coral communities live 200 meters beyond its entrance. Runoff from salt flats to the north of the bay flows into the lagoon via three channels during infrequent heavy rain periods. The lagoon is partially sheltered from the prevailing trade winds by mangroves and hills on its outlet's east and west sides.

Fieldwork
Surveys were conducted every two weeks at SE (June 2014 to May 2018) and at BB (June 2014 to May 2015, October 2016 to May 2018) stations at approximately 8:00 and 9:00 AM local time, respectively. At times, we altered sampling frequency due to unfavorable weather conditions or instrument failure. Sampling depth at the SE station was 4 m, while at BB, it was 3 m. No permits were required to conduct this fieldwork.
Our data set includes DIC and TA measurements from discrete seawater samples, drawn from a 2.2 L Van Dorn sampler into 250 (or 125) mL borosilicate glass bottles and secured using greased stoppers, following the guide for best practices for CO 2 measurements [38]. Each sample was fixed immediately with 0.02% of the total bottle volume using a saturated solution of mercury chloride (HgCl 2 ) or immediately after returning to the laboratory (same day) to prevent biological alteration and stored at room temperature for later analysis. Each bottle was tightly sealed with about 1% of headspace to prevent atmospheric gas exchange. Analyses were performed within six months after sampling. Profile measurements of temperature and salinity were taken using an SBE251 conductivity, temperature, and depth recorder (CTD).

Meteorological and tidal observations
Wind observations ( Fig 3A) were obtained from a meteorological station [39] at the UPRM's Magueyes Island (Fig 1). The station is 3.1 km west of the BB site. We obtained daily accumulated precipitation observations ( Fig 3B) collected at Magueyes Island from NOAA's National Center of Environmental Information [39]. Seasonality at La Parguera is clearly defined by precipitation patterns on a climatological scale and throughout the sampling period (2014-2018). Sea-level data for La Parguera is available from a NOAA tide gauge (station number 9759110) located in the west-end of Magueyes Island [40].

Carbonate chemistry analyses and calculations
From June 2014 to May 2017, seawater samples for TA were analyzed at the University of New Hampshire (UNH) Ocean Processes Analysis Laboratory. TA analyses followed the Gran titration method [41], using an Apollo SciTech automated titrator and 0.1 N hydrochloric acid (HCl) titrant. The Orion 3-Star pH electrode (Thermo Fisher Inc.) was calibrated using low ionic strength pH buffers (certified on the U.S. National Bureau of Standards Scale to ± 0.01). Dickson's Certified Reference Materials (CRM) [42] were used to standardize the HCl titrant and validate the results. The precision and accuracy of the TA method as implemented at UNH were ± 2 μmol kg -1 and ± 3-4 μmol kg -1 , respectively. A CRM was analyzed at the beginning and end of each run to correct for any potential drift. From June 2017 to May 2018, TA samples were analyzed at the University of Puerto Rico Mayagüez Campus (UPRM), Department of Marine Sciences, following the method described above and using similar instrumentation. The precision and accuracy of the TA method as implemented at UPRM were ± 2 μmol kg -1 and ± 3-4 μmol kg -1 , respectively. We note that the TA samples from BB may include contributions from organic compounds [43], which we did not account for in this experiment.
All DIC samples were analyzed at UNH by measuring the infrared absorption of the CO 2 released upon the sample's acidification with 10% phosphoric acid using an Apollo SciTech AS-C2 automated analyzer. CRM's were also used to calibrate the instrument. The precision and accuracy of the DIC method as implemented were ± 2 μmol kg -1 and ± 5 μmol kg -1 , respectively.
We calculated the seawater CO 2 partial pressure (pCO 2 ), Ω arg, and pH (total scale) from TA and DIC using the CO2SYS v2.  [48], given that it is more suitable for higher pH values. The solubility constant used to derive the O arg is from Mucci [49].

Air-sea CO 2 flux
The air-sea CO 2 flux was estimated according to F = k S ΔpCO 2 , where F is the flux in mmol m -2 day -1 , k is the gas transfer velocity (m day -1 ), S is the CO 2 solubility coefficient according to Weiss [50], and ΔpCO 2 is the ocean surface and atmosphere pCO 2 difference. We obtained the atmospheric pCO 2 values from the MapCO 2 buoy in the Enrique Cay forereef (Fig 1). We considered the recent parametrization from Ho [52] that integrates horizontal current velocity, depth, and wind speed to calculate the gas transfer velocity. This parametrization works well for this type of study since it considers any bottom-generated turbulence. We assumed a horizontal current velocity of 5 cm s -1 for the latter calculation, given that such magnitude is approximate from previous observations (unpublished results). Wind speed was measured at a 15-meter height and adjusted to a 10-meter wind [51].

pH drivers and seasonality analysis
To assess the pH seasonality and its controlling mechanisms in BB, we calculated (using CO2SYS) the effects of temperature, salinity, TA, DIC, and air-sea DIC exchange on pH. The following equation describes observed monthly changes between time t 2 and t 1 (e.g., February and January) in pH (ΔpH), assuming that each process modified the carbonate system independently and that the water column is well mixed: where the terms on the right-hand side of the equation correspond to changes in temperature (T), salinity (S), air-sea DIC exchange (F) in CO 2 form, TA, DIC, and a residual (R), which represents the remainder of the difference between the terms of the right-hand side of Eq (1) and the observed pH changes (ΔpH). Differences in pH were between the initial (pH calc at t 1 ) and modified (pH 2 at t 2 ) conditions according to the following equations: DpHðSÞ ¼ pH 2 ðT 1 ; S 2 ; TA 1 ; DIC 1 Þ À pH calc ð3Þ We refer to pH 1 (T 1 ,S 1 ,DIC 1 ,TA 1 ) at t 1 as pH calc . Changes in DIC drive both ΔpH(F) and ΔpH(DIC): the former term refers to DIC changes resulting from air-sea CO 2 exchange, while the latter term includes DIC changes caused by both water-column and benthic mechanisms. To determine ΔpH(F), we first calculated the DIC change (ΔDIC air-sea ) using the CO 2 air-sea exchange (F, mmol m -2 d -1 ), seawater density (d, kg m -3 ), change in time (t 2 -t 1 , days), and sampled depth (H, m) [52]: Given that changes in pH due to DIC include the effect of CO 2 air-sea exchange, we subtracted ΔpH(F) from ΔpH(DIC) to separate both signals:

Carbonate chemistry spatio-temporal variability
We averaged all the interannual data from BB and SE to monthly values (Fig 4). Differences in salinity and temperature variabilities between BB and SE support the contention of limited exchange between BB and outer coastal waters [27,53]. The exchange regime results from the semi-enclosed physiography and the small tidal range typical for the region. The larger magnitudes of temperature and salinity observed at BB are likely due to higher daytime evaporation rates [54]. Water temperature remained above 29 o C at BB for most of the year, from May to November. We detected the salinity minimums at both stations in November due to large amounts of accumulated precipitation (Fig 3B). The salinity decrease at SE after April was due to the influence of riverine water from the Amazon and Orinoco river plumes, later magnified by the maximum outflow of the Orinoco in September [29, 30] and the start of the wet season in August (Fig 3B). June was the driest month during the sampling period and anomalously dry relative to climatological data. BB's higher salinities are due to its enclosed nature and the year-round dominance of evaporation [27] and evapotranspiration [54] over precipitation. The data presented in Table 1 and Fig 4 show that BB was consistently more acidified (lower pH and Ω arg ) than SE throughout the year. The larger amplitudes observed in the BB data evidence a marked seasonality relative to the SE data. BB's pH values are near the projected 7.8 mean for 2100 under business-as-usual CO 2 scenarios (RCP 8.5) during September and November, while we often observed O arg below or near the 3.0 projected value [55].
BB's lower pH values coincided with higher TA concentrations (relative to SE), suggesting that the excess TA was insufficient to buffer the declines in pH and O arg . Thus, we observed an average excess (relative to the oceanic reference station, SE) of mean TA and DIC of + 36 and + 94 μmol kg -1 , respectively. The mean seawater pCO 2 (Fig 4G) at BB station was above the mean atmospheric pCO 2 (400 μatm) year-round by approximately 270 μatm.
As indicated by pH and Ω arg data, the acidification reaches its maximum at BB during the wet season when the mean pH decreased by 0.06 units, and the mean Ω arg decreased by 0.19. During that time (Aug-Nov), we observed an excess of mean TA and DIC of + 81 and + 152 μmol kg -1 , respectively, relative to SE. Thus, mean TA and DIC in BB increased during the wet season by 15 and 48 μmol kg -1 , which is not enough to induce a net buffering effect [7]. Also, we observed pCO 2 values above 600 μatm, thus 200 μatm above the 400 μatm atmospheric global means. This study provides the first characterization of the seasonal carbonate chemistry within a mangrove system in Puerto Rico. These results show that not all mangrove lagoons can buffer OA at seasonal time scales and serve to advance our understanding of the diverse roles mangroves may play in modulating acidification in tropical coastal environments.

Influence of TA and DIC variability on pH and Ω arg
The relationship between DIC and TA can provide insights into the mechanisms that control OA [56]. Both Ω arg and pH calc showed significant (p < 0.00001; n = 47) and strong correlations (R 2 = 0.97 and 0.91, respectively) with the DIC/TA ratio (Fig 5). This ratio evidences the sensitivity of the system to perturbations in dissolved seawater CO 2 concentration [57]. If the DIC

PLOS ONE
Carbonate chemistry seasonality in a tropical mangrove lagoon to TA ratio equals 1, any addition/removal of CO 2 (in the form of DIC) will result in a maximum decrease/increase of seawater pH and Ω arg [58], with soluble CO 2 being more sensitive to temperature changes. Thus, TA and DIC changes controlled the pH decrease and Ω arg we observed in BB during the wet season (Fig 4). When transitioning from the peak of the dry season (June-July) to the wet season (August-November), the mean DIC/TA ratio increased significantly (p < 0.001, n = 6 and 16, respectively) from 0.87 ± 0.00 to 0.91 ± 0.02. At this time, we observed increases in DIC and TA of 100 and 6 μmol kg -1 , respectively, decreasing pH and Ω arg by 0.12 and 0.75. Thus, the observed changes in water chemistry evidence that the increase of CO 2 production (presumably from benthic and water column aerobic respiration) surpassed the increase of TA production from anaerobic processes, suggesting aerobic dominance over anaerobic remineralization. Sulfate reduction is an important TA producing reaction in tropical mangrove sediments [7,21,59,60], while denitrification is not [7,24,61,62]. Previously measured water column nitrate concentrations at BB were low (0.12 μmol L -1 , annual mean) and showed a limited seasonal variability [35]. Another study [24] in La Parguera reported low benthic denitrification (0.03-0.35 μmol m -2 h -1 ) and nitrification (0.44-3.15 μmol m -2 h -1 ) rates. However, we cannot rule out denitrification from occurring in the water column or the sediment-seawater interface. During the wet season, the salt flats to the north get flooded, and water flows into BB, possibly exporting nitrate and further enhancing denitrification [63].
Previous research (S3 Table) at BB reported that water column respiration exceeds primary production year-round. Aerobic respiration of organic matter releases essentially unbuffered DIC [9], given that it increases DIC by 1 mol and decreases TA by 0.2 moles for each mol of carbon oxidized (assuming a Redfield ratio). Previous work in mangrove environments has documented an increase in heterotrophic activity during the wet season [58,61,64], possibly fueled by carbon derived from allochthonous sources (i.e., phytoplankton, seagrass, terrestrial non-mangrove, or direct input of TA and DIC from runoff) or local sources (i.e., water column mechanisms or benthic mechanisms) [22,59]. [32] and in porewaters (0.024-3.2 mmol L -1 ) [37] throughout the year. High sulfate reduction rates in BB (34.06 mmol L -1 h -1 ) relative to an oceanic station located ca. 10 km from shore (9.33 mmol L -1 h -1 ) were reported as well [37]. The higher sulfate reduction rates observed (see [37]) in BB indicate its potential role as a TA source, assuming minimal re-oxidation of reduced sulfur species [65,66]. Sulfate reduction produces one mole of TA per mole of organic carbon oxidized, assuming that the sulfur species are not re-oxidized to sulfate [67].

The drivers of the pH seasonality
The pH decomposition (Fig 6A) shows that seasonal changes were influenced the most by biological processes, given that ΔpH(DIC) and ΔpH(TA) exerted the most considerable changes to PLOS ONE monthly pH. The DIC/TA ratio (Fig 6B) closely follows the observed pH changes (ΔpH). Thus, the balance between changes in DIC and TA mainly modulated the pH variability. Increases in DIC decrease pH, while increases in TA increase pH. The mean annual magnitudes of ΔpH(T) and ΔpH(S) together equal 0.04, while the mean magnitudes of ΔpH(DIC) and ΔpH(TA) equal 0.08 and 0.04, respectively. Therefore, changes in pH due to mixing, dilution, and evaporation were the least significant. With a mean annual magnitude of 0.05, air-sea CO 2 exchange increased pH (ΔpH(F)) year-round as expected [21]. When comparing the TA at BB with the TA at SE from March to May, we observe that the TA depleted in BB. Thus, some local processes must have consumed alkalinity. During the wet season, we observed the largest decrease in ΔpH(DIC) and increase in ΔpH(TA), confirming that CO 2 production from (presumably) aerobic respiration and TA production from anaerobic respiration increased at this time, while aerobic respiration dominated.

Summary and conclusions
Results from this study evidence that BB could not buffer OA over a seasonal time frame. Although TA increased throughout the year, it was not enough to buffer the observed declines in pH and Ω arg resulting from increases in DIC. This acidification effect exacerbated during the wet season due to increased precipitation coinciding with more extensive tidal excursions (range), presumably enhancing the transport of labile organic matter into BB. Such an increase in heterotrophic activity during the wet season seems to be a general feature at La Parguera [68] enhanced in BB.
Given the insight gained in this study, future research should integrate temporal and diel surveys to better assess carbonate chemistry variability and trends at various time scales. It is unknown if the observed acidic conditions export to nearby environments (and to what extent) or if such conditions are detrimental to organisms other than corals. Future studies should couple the rates of benthic and water column remineralization processes with in-situ carbonate chemistry measurements. The chemistry and nutrient concentrations of runoff water should be measured as well. It is also essential to assess how the acidification discussed can impact organisms that live in the mangrove root systems and sediments.