Outwelling of reduced porewater drives the biogeochemistry of dissolved organic matter and trace metals in a major mangrove‐fringed estuary in Amazonia

Mangrove‐fringed estuaries are intertidal ecosystems discharging significant amounts of dissolved organic matter (DOM) into coastal oceans. DOM in these ecosystems is derived from autochthonous production, fluvial input, and mangrove porewater outwelling; however, differentiating between these sources remains challenging. Our incomplete understanding of the biogeochemical factors controlling DOM dynamics and its relationship with nutrient and trace metal cycling still hinders the formulation of elemental budgets in coastal environments. Here, we relate the DOM composition in a mangrove‐fringed estuary in North Brazil (Amazonia) to the redox conditions at the formation sites. We combined molecular DOM analyses via ultrahigh‐resolution mass spectrometry (FT‐ICR‐MS) with parallel factor analysis of excitation–emission fluorescence matrices (EEM‐PARAFAC), nutrient and redox‐sensitive trace metal analyses. During low tide, the influx of oxygen‐depleted porewater carried terrigenous DOM, inorganic nutrients, and trace metals into the mangrove‐fringed creeks. Precipitation of metal(hydr)oxides and microbial turnover controlled nutrient and trace metal dynamics in the estuary. The highest inorganic nitrogen concentrations within the upper mangrove‐fringed estuary indicated outwelling from mangrove sediments as an essential source. Phosphate concentrations were highest within the lower mangrove‐fringed estuary, where available phosphate likely exceeded precipitation with iron(hydr)oxides. We tracked the DOM transport to the coastal ocean using a novel molecular index derived from sulfidic porewater (ISuP). Outwelling of recalcitrant DOM from mangrove habitats is relevant in the context of blue carbon storage. Therefore, applying our molecular proxy (ISuP), together with trace‐metal and optical DOM analyses, is a powerful approach for differentiating contributions of diverse DOM sources in highly complex coastal ecosystems.

burial in sediments (Bouillon et al. 2008;Alongi 2014).Recently, fluxes of dissolved inorganic (DIC) and organic (DOC) carbon from coastal ecosystems have regained research attention, as carbon outwelling, i.e., lateral fluxes or horizontal export with subsequent storage in the deep-ocean, could be a critical long-term carbon sequestration mechanism ("blue carbon") (Santos et al. 2021a).
DOM outwelling is driven by tidal porewater discharge in mangrove-fringed areas (Dittmar and Lara 2001a;Bouillon et al. 2007;Rezende et al. 2007;Kristensen et al. 2008;Maher et al. 2013).Around 50% of mangrove porewater DOC is recalcitrant on timescales from weeks to years and is a substantial DOC source to coastal oceans (Jaffé et al. 2004;Dittmar et al. 2006).Organic matter remineralization releases DIC, ammonium, and phosphate that accumulate in porewater.Together with reduced trace metals, nutrients are transported to the adjacent water column through porewater outwelling (Dittmar and Lara, 2001b;Mori et al., 2019;Cabral et al., 2021).However, we still lack a comprehensive understanding of DOM and trace metal cycling in coastal intertidal systems.The export of recalcitrant DOM from coastal habitats is still rarely considered in long-term oceanic carbon storage (Santos et al. 2021a).Hence, comprehending coastal carbon sources and sinks requires an investigation of biogeochemical connections within coastal ecosystems to understand their response to increasing anthropogenic pressure (Santos et al. 2021b).
Mangrove sediments are depleted in oxygen because of high organic matter input and limited oxygen supply, often leading to sulfate-reducing conditions (Dittmar and Lara 2001a;Alongi 2020;N obrega et al. 2022).Under sulfidic conditions, the abiotic incorporation of sulfide into DOM (sulfurization) leads to sulfur-enriched DOM (dissolved organic sulfur, DOS) (Pohlabeln et al. 2017).DOS is potentially recalcitrant because it accumulates in porewater under sulfidic conditions (Schmidt et al. 2009;Seidel et al. 2014).During low tide, DOS-enriched porewater drains into intertidal creeks (Mori et al. 2019) and is potentially further transported to estuaries and across continental shelves and reaches the deep ocean as refractory DOM.
Reducing conditions in the sediments favor Mn-(hydr)oxide and Fe-(hydr)oxide dissolution as well as barium (Ba) and phosphate release from Fe-and Mn-(hydr)oxides (van Raaphorst and Kloosterhuis 1994;Sanders et al. 2015).Therefore, groundwater discharge can be an essential trace metal and inorganic nutrient source to adjacent surface waters (Holloway et al. 2016).
Three-dimensional excitation-emission matrix fluorescence spectroscopy, in combination with parallel factor analysis (EEM-PARAFAC), is used to study DOM optical properties (Murphy et al. 2008;Hansen et al. 2016).Ultrahigh-resolution Fourier-transform ion cyclotron resonance mass spectrometry (FT-ICR-MS) allows for the molecular formula-level characterization of DOM (Hertkorn et al. 2013).Studies applying FT-ICR-MS analyses revealed that mangrove porewater DOM contains highly condensed aromatic (Tremblay et al. 2007) and sulfur-containing molecules (Mori et al. 2019).Combining FT-ICR-MS and FDOM analyses offers insights into the biogeochemical DOM properties, including non-fluorescent molecules (Stubbins et al. 2014).With these techniques, DOM characteristics for ecosystems such as boreal rivers (Stubbins et al. 2014), tropical wetlands (Wagner et al. 2015), and peatlands (Tfaily et al. 2015) have been examined.
We aim to answer how mangrove ecosystems influence the composition of (recalcitrant) DOM and essential nutrients (nitrate, nitrite, ammonium, and phosphate) in a mangrovefringed estuary and on their way to the coastal ocean.We focus particularly on sulfur-containing molecules, hypothesizing that sulfate-reducing mangrove sediments are hotspots of abiotic DOM sulfurization and that sulfurized DOM is carried with tidal fluxes into the estuary and coastal zone.We further hypothesize that dissolved trace metals (iron, manganese, and barium) are introduced from mangrove sediments into the tidal creeks via porewater fluxes and that their oxidation and precipitation in the estuary are connected to DOM and nutrient exports to the coastal ocean.Understanding the outwelling from mangrove habitats is important in the context of blue carbon storage, as export and potential sequestration at locations far from the actual organic matter formation sites have not yet been considered in coastal carbon sequestration scenarios.

Study site
The study area is a macro-tidal mangrove-fringed estuary located at the Caeté River, near Bragança (Par a State, Brazil) in North Brazil.Detailed descriptions of the research area can be found in Call et al. (2019) and Cabral et al. (2021).The sampling transect along the Caeté River and its estuary can be distinguished into a fluvial section (Station (Sta.)23 to 19), upper (Sta.18 to 14) and lower (Sta.13 to 6) mangrove-fringed sections, as well as a marine coastal ocean section (Sta.5 to 1) (Fig. 1).

Sampling
We conducted the sampling campaign during the dry season of 2017 near Bragança (Par a State, Brazil) in the Caeté estuary.Water samples were collected from a boat on the 10 th of October along a 35 km-long transect.Stations 22 and 23 were replicate samples.The different stations covered the salinity gradient of the fluvial and mangrove-fringed estuary to the nearshore ocean.Sampling during high and low tide was conducted between the 27th of September and the 7th of October in the Furo do Meio tidal creek close to the estuarine transect, covering 11 tidal cycles from neap to spring tide.
Water was collected by dispensing a prerinsed bucket over the boat (transect sampling) and a bridge (tidal creek sampling).We took porewater samples at eight locations in the mangroves next to the Furo do Meio tidal creek, close to the tidal sampling site, by digging small holes ($ 50 cm depth).Five liters of accumulating porewater were scooped out from the holes after discarding the first two fillings of accumulating porewater.
Water samples (20-30 mL) for FDOM, nutrient (NO x À and phosphate), total dissolved nitrogen (TDN), and DOC analyses were filtered through Isopore membrane filters (GTTP, 0.2 μm, Merck Millipore) into high-density polyethylene bottles after pre-rinsing each bottle with samples.DOC and TDN samples were acidified to pH 2 with 25% HCl (p.a.).Separate FDOM and nutrient samples were kept frozen without acidification until further analysis.Total humic-like FDOM fluorescence was measured in filtered water samples (without acidification) using an AquaFluor Handheld Fluorometer/Turbidimeter 8000-010 (Turner Designs; wavelength 350 nm excitation, 420 nm detection; noted in relative fluorescence units, rfu) directly after sampling.Salinity, temperature, and dissolved oxygen concentrations were determined using an HQ40D hand device (Hach Lange) with a CDC40101 conductivity sensor (salinity and temperature) and an LDO101 sensor (oxygen).All samples were pre-filtered through a pre-rinsed nylon mesh (100 μm mesh size) and filtered through pre-rinsed 1.0 μm Causapure filter cartridges (CPR-001-09-DOX, PP, Infiltec) with a peristaltic pump.One liter of filtered and acidified samples (HCl, pH 2) was solid-phase extracted (SPE) using styrene-divinylbenzene-polymer-filled cartridges (1 g, Agilent Bond Elut PPL, USA) as described in Dittmar et al. (2008).Due to their high DOC concentrations, porewater samples were extracted with 5 g PPL cartridges.
Samples for dissolved trace metal analysis (Ba, Fe, and Mn) were immediately filtered after sampling through 0.45 μm SFCA (surfactant-free cellulose acetate) syringe filters into 10 mL, acid-cleaned, sample-rinsed, polypropylene vials.Trace metal samples were acidified using concentrated ultrapure HNO 3 to obtain a concentration of 1% (v/v).

Bulk geochemical and FT-ICR-MS analysis
DOC and TDN concentrations were analyzed by hightemperature catalytic combustion with a Shimadzu TOC-V CPH instrument equipped with a TDN unit.Trueness and precision were tested against deep seawater reference material and low- carbon water (provided by D.A. Hansell, University of Miami, FL, USA).Both were better than 5%.
The stable carbon isotopic composition of SPE-DOM was analyzed on an isotope-ratio-monitoring mass spectrometer (FinniganMAT 252, Bremen, Germany) via a Conflo II split interface after pipetting the methanol extract into tin caps and completely drying.Stable carbon isotope ratios are expressed as δ 13 C relative to the PDB reference.
Phosphate concentrations were measured photometrically using the Multiscan GO Microplate Spectrophotometer (Thermo Fisher Scientific, USA) following the protocol from Grashoff et al. (1999).NO x À was determined as described in Schnetger and Lehners (2014).Dissolved Mn, Fe, and total phosphorus were analyzed using inductively coupled plasma optical emission spectrometry (ICP-OES, iCAP 6000, Thermo) as described previously (Seidel et al. 2014).DOM was molecularly characterized by ultrahigh-resolution mass spectrometry using a solariX XR Fourier-transform ion cyclotron resonance mass spectrometer (Bruker Daltonik GmbH, Bremen, Germany) connected to a 15 T superconducting magnet (Bruker Biospin, Wissembourg, France).Details on analytical conditions and molecular formulae assignment are provided in the Supporting Information.The FT-ICR-MS signal intensity of each identified molecular formula was normalized to the sum of the intensities of all identified molecular formulae in each sample.Molecular formulae were assigned to operationally defined compound groups as described in Seidel et al. (2017) according to their H/C and O/C ratios and aromaticity index (AI mod ), which is a parameter developed to identify aromatic and condensed aromatic structures in DOM molecular formulae (Koch and Dittmar 2016), i.e., polycyclic aromatic compounds (PCAs, group 1, AI mod > 0.66), highly aromatic compounds (HACs, group 2, 0.5 < AI mod ≤ 0.66), highly unsaturated compounds (HUCs, group 3, AI mod ≤ 0.5, H/C < 1.5, O/C < 0.9), unsaturated aliphatic compounds (UAs, group 4, 1.5 < H/C ≤ 2, O/C < 0.9, N = 0).We also investigated compound groups containing nitrogen (group 8, N), sulfur (group 9, S), and phosphorous (group 10, P).We emphasize that the assigned compound groups are operational since FT-ICR-MS analysis does not yield explicit structural information.

Optical analysis
After thawing, FDOM samples were filtered again (0.2 μm, GHP, Pall Acrodisc).In a 1 cm quartz cuvette, EEM spectra were measured with a spectrofluorometer (Aqualog ® Horiba Scientific, Germany).Measurements were performed using ultrapure water as a reference, and the cuvettes were prerinsed with ultrapure water and sample.Details on analytical conditions and PARAFAC modeling are provided in the Supporting Information.

Statistical analysis
We used principal coordinate analysis (PCoA) based on a Bray-Curtis dissimilarity matrix of intensity-normalized DOM molecular formulae intensities to determine the variability of the DOM composition between the samples.PCoA is a multidimensional metric scaling method that projects the variability within a dataset onto uncorrelated (PC) axes (Ramette 2007).Centered and scaled environmental data and DOM molecular compound groups were fitted to PCoA ordination post hoc using the function envfit (permutations 10.000, p < 0.05) of the vegan package (Oksanen et al. 2020) in R (R Core Team 2020).
Spearman's rank correlations (Spearman's ρ, α-level 0.05) were calculated to find significant relationships between the DOM molecular compound groups, FDOM data, bulk geochemical parameters, and dissolved trace metals and between the relative intensities of all detected molecular formulae and relative fluorescence intensities of the modeled PARAFAC components (Spearman's ρ, α-level 0.01, Supporting Information).False-positive p-values were corrected after Benjamini and Hochberg (1995).
Relative FDOM intensities of the PARAFAC components were not normally distributed (Shapiro-Wilk test).Therefore, a Wilcoxon Rank-Sum test (p < 0.05) was used to test for significant differences between porewater, low tide, and high tide samples and when comparing the fluvial, mangrove-fringed, and coastal marine sections for the relative fluorescence of PARAFAC components as well as the obtained DOM compound groups.Additionally, a Wilcoxon Rank-Sum test was conducted to test for differences in bulk geochemical parameters between spring and neap tide (Supporting Information Table S1).

Molecular formulae indicator of mangrove porewater inputs (I SuP )
DOM molecular formulae from normalized FT-ICR-MS data have been previously applied to develop indices assessing the degradation state of DOM (I DEG ) (Flerus et al. 2012) and the presence of terrigenous DOM (I Terr ) (Medeiros et al. 2016).We used a similar approach to trace mangrove-derived molecular formulae with normalized intensities from FT-ICR-MS analysis.As a first step, we selected terrigenous molecular formulae containing S, resulting in 147 molecular formulae (s-Peaks) (Supporting Information Table S2).These s-Peaks represent potentially sulfurized mangrove porewater-derived DOM and were selected according to the criteria of terrigenous molecular formulae (t-Peaks) identified by Medeiros et al. (2016).As a second step, we further identified 40 S-containing and porewater-derived molecular formulae (termed SuP) that were part of the s-Peaks and that were particularly enriched in mangrove porewater samples and strongly correlated with Ba concentrations ( p < 0.01, R > 0.7) as a tracer for groundwater input (Shaw et al. 1998).We additionally used the 40 molecular formulae identified by Medeiros et al. (2016) as marine molecular formulae (Mar), which are also part of recalcitrant oceanic DOM ("island of stability") (Lechtenfeld et al. 2014), to calculate the ratio of the sums of the normalized intensities (sum magnitudes SuP/sum magnitudes (SuP + Mar), which we refer to as I SuP .Accordingly, the I SuP increases with the abundance of S-containing DOM derived from mangrove porewater.

Spectral properties of PARAFAC components
Four FDOM components were identified by PARAFAC analysis (Fig. 2, Supporting Information Fig. S1), and they were comparable to previously reported peaks with multiple matches in the OpenFluor database (TCC exÂem > 0.95) (Supporting Information Table S3).Components C1, C2, and C3 were assigned to classical A (UV humic-like) and C (visible humic-like) peaks, which are typical terrestrial-derived FDOM components (Supporting Information Table S3), and C4 to protein-like components (Em < 350 nm), resembling tryptophanlike fluorescence.
The highest relative abundances of terrestrial humic-like components C1, C2, and C3 were observed at low tide ($ 47%, 27%, and 14%, respectively, Fig. 2a.B-c.B).High tide samples differed significantly from low tide samples during spring tide, as did high tide samples between spring and neap tide conditions across multiple parameters (DOC, FDOM, TDN, δ 13 C SPE-DOM, and Ba).Also, low-tide salinity, DOC, and phosphate values significantly differed between spring and neap tide, indicating distinct porewater sources.Furthermore, during neap tide conditions, dissolved Fe, Mn, DOC, and FDOM values significantly varied between low and high tide samples (Supporting Information Table S1).
The SPE-DOC extraction efficiencies were 67 AE 5% for surface water samples (n = 37) and 30 AE 4% for porewater samples (n = 8).Salinity and extraction efficiency were not correlated (Supporting Information Fig. S2).In 45 samples, more than 22,000 molecular formulae were detected.The intensity-weighted average H/C ratios were lowest in the porewater samples (H/C 1.12, Fig. 4a, Supporting Information Table S5).The relative abundances of HAC (group 2) in porewater were comparable to those observed at low tide.However, they were significantly lower at high tide (HAC, 17%, p < 0.05, Fig. 4c).The average relative abundances of highly unsaturated (HUC, group 3) molecular formulae were $ 58% in porewater and low tide samples, with a slightly higher percentage ($ 60%) in high tide samples (Supporting Information Figs.S3 and S4). ), (i) dissolved iron (Fe), and (j) dissolved manganese (Mn).Water levels reflect high and low tide gauges highlighted in gray.Neap tide is highlighted with the black line, and spring tide occurred after the last sampling point ($day 11; see Asp et al. 2018).
Notably, during low tide and high tide conditions, the H/C ratios were significantly higher (1.15 and 1.18, p < 0.05, Supporting Information Table S6).The intensity-weighted average O/C ratios were significantly lower in low tide samples (O/C = 0.39, p < 0.05), compared to 0.41 in porewater and 0.43 at high tide (Fig. 4b).The relative abundances of PCA molecular formulae (group 1) were highest in porewater and at low tide ($ 4%), compared to high tide (3.75%, p > 0.05).The proportion of unsaturated aliphatic molecular formulae (UA, group 4) was lowest in porewater (9%) and slightly higher ($ 10%) in both low and high tide samples ( p < 0.05).Among the molecular formulae occurring exclusively in porewater, 32% and 31% included at least one N or S atom, respectively.The relative abundance of molecular formulae containing N was lowest at low tide (29%, p < 0.05, group 8, Fig. 4d) compared to porewater and high tide samples.In contrast, the relative abundance of molecular formulae containing S was lowest during high tide (28%, p < 0.05, Fig. 4e) compared to porewater and low tide samples.On average, only 2.5% of the molecular formulae across all samples contained P, and there was no consistent trend between sampling sites.(group 10, Supporting Information Figs.S3  and S4).

Geochemical parameters along the river to coastal ocean transect
Surface salinity increased from 0 (fluvial Sta.23) to 35.8 (coastal marine Sta. 1) (Fig. 1, Supporting Information Fig. S5).The highest dissolved oxygen concentrations were observed in the nearshore coastal section (8 mg L À1 ), and they decreased toward the mangrove-fringed section (5 mg L À1 , Sta. 15, Fig. 5a).The dissolved oxygen concentration at fluvial station 23 was 6 mg L À1 .DOC concentrations were lowest in the coastal marine section (112 μM), increasing toward the mangrove-fringed section to 260 μM before decreasing again to 186 μM in the fluvial section (Fig. 5b).
Humic-like FDOM values were highest in the mangrovefringed section (96 rfu, Sta. 13, Fig. 5c, Supporting Information Fig. S5), compared to the coastal marine (7 rfu) and the fluvial (65 rfu) sections.On average, TDN concentrations were 7 μM in the coastal marine, 17 μM in the fluvial, and 32 μM in the mangrove-fringed area (Fig. 5d).Values of δ 13 C SPE-DOM increased from the riverine (À 29‰) to the coastal marine section (À 22‰), which was not consistent with values expected from conservative mixing in the mangrove-fringed area (note the non-linear relationship of δ 13 C SPE-DOM values with salinity, Fig. 5e).
Dissolved Ba concentrations were highest at mangrove station 13 (1.2 μM, Fig. 5f).Maximum NO x À concentrations were observed at salinities 8 to 10 (Sta.16 and 17, Fig. 5g).NO x À concentrations approached the detection limit at the coastal marine section.Phosphate concentrations were 0.3 μM at the fluvial station 23 and increased to 0.8 μM at the outer mangrove station 8 (Fig. 5h).Dissolved Fe and Mn concentrations were highest (60 and 48 μM) in the fluvial station 23 and 21 (salinity 0 and 1) and decreased with increasing salinity to undetectable values (< 0.06 μM) (Fig. 5i,j, Supporting Information Fig. S5).The humification index (HIX) was highest in porewater and low tide samples (HIX > 0.9) and slightly lower in the mangrove-fringed section (HIX > 0.87, Supporting Information Table S7).The biological index reached the highest levels in the coastal marine section (BIX = 1.54,Sta. 1, Fig. 1) and decreased with decreasing salinity to < 0.66.
Porewater and low tide samples also had low biological index levels (Supporting Information Table S7).
The sum of the relative intensities of the 147 s-Peaks molecular formulae (Supporting Information Table S9) was lowest at fluvial station 22 (sum = 24) (Fig. 5l).Highest values occurred at station 9 (sum = 50, salinity 31) as well in porewater and low tide samples, respectively (sum > 75).The sum of the relative intensities of t-Peaks was highest at fluvial station 23 (sum = 266, Fig. 5k).S-and t-Peaks sums decreased toward the coastal marine section and were low during high tide (s-Peaks < 50, t-Peaks < 191, Fig. 5k,l, Supplementary Information S9).The I Terr values were at maximum in the fluvial section (I Terr > 0.6, Fig. 6a).Values of I Terr decreased with increasing salinity (Fig. 6b).In contrast, I SuP values were highest in the mangrove-fringed section (salinity $ 20) and lowest at the coastal marine and fluvial stations (Sta. 1 and 23, Fig. 1, I SuP < 0.12, Fig. 6a).
Overall, 358 (C1), 1836 (C2), 1066 (C3), and 920 (C4) DOM molecular formulae from FT-ICR-MS analysis were significantly correlated with the respective PARAFAC ), (i) dissolved Fe (μM), and (j) dissolved Mn (μM) concentrations, (k) the sum of the relative intensity of all t-peaks, and (l) the sum of the relative intensity of all s-peaks (FT-ICR-MS data).The shaded area indicates the mangrove-fringed area of the sampling transect.The dotted lines represent the theoretical conservative mixing of the fluvial with the coastal marine endmember.Note that conservative mixing results in a non-linear relationship with salinity for δ 13 C SPE-DOM and relative abundances of s-and t-peaks.
components ( p ≤ 0.01).The individual positive correlations of the PARAFAC components with FT-ICR-MS data are presented in Table 1 and Supporting Information Fig. S6.
In PCoA, the first two axes explained 58% of the DOM molecular variability (Fig. 7).The first axis separated samples into fluvial and marine, while the second axis distinguished porewater and tidal cycle samples from the estuarine transect.The first axis and fluvial transect samples were linked to HAC (group 2).Fluvial and mangrove-fringed transect samples were associated with PCA compounds (group 1) and NO x À concentrations.The mangrove-fringed section also correlated with P-containing formulae (group 10), unsaturated aliphatic compounds (UA, group 4), and dissolved oxygen levels in intertidal creek surface water.Coastal marine samples in the river-to-ocean transect exhibited higher H/C ratios, proteinlike C4, N-containing formulae (N, group 8), BIX, high salinity, and enriched δ 13 C SPE-DOM values.Intertidal creek high tide samples resembled marine transect samples, while lowtide samples were more similar to mangrove porewater samples.Porewater and low tide samples exhibited high DOC, TDN concentrations, elevated I SuP values, C2 abundance, Scontaining formulae (S, group 9), and dissolved Ba, Mn, and Fe concentrations.Porewater and low tide samples shared high O/C ratios and an abundance of highly unsaturated molecular compounds (HUC, group 3) with high tide samples.
Fluvial samples and porewater were associated with high HIX and I Terr values and elevated C1 and C3 relative abundances.

Discussion
Biogeochemical dynamics of DOC, nutrients, and trace metals in the Caeté estuary Tidal pumping drives groundwater exchange in the Caeté estuary (Cabral et al. 2021).Since evaporation during the dry season causes elevated salinities in the upper porewater horizon (Dittmar and Lara 2001b), surface runoff after inundation and porewater outflow into the mangrove creeks explain the high salinity values in the tidal creek at low tide (Fig. 3a).In the Caeté estuary, outflowing porewater is a nutrient source for the water column and nearshore ocean (Dittmar and Lara 2001b).Organic matter degradation and the microbial reduction of Fe and Mn (hydr)oxides in the sediments results in the Ba and phosphate dissolution from the metal oxides (Coffey et al. 1997) and the subsequent enrichment of these species in the porewater.Previous observations of pCO 2 , CH 4 , and 222 Rn confirmed tidally driven porewater exchange at our study site, with elevated concentrations during low tide (Call et al. 2019).Similarly, porewater outwelling explains the elevated Fe, Mn, Ba, and phosphate concentrations in the tidal creek during low tide in our study (Fig. 3).
Porewater-derived nutrients and trace metals are transported from the mangrove creeks to the adjacent estuary, Table 1.Intensity-weighted averages and relative abundances of DOM molecular compound groups in molecular formulae (from FT-ICR-MS analysis) that were positively correlated (p ≤ 0.01) to the PARAFAC components.

Component
Number  explaining, for example, elevated dissolved Ba concentrations in the mangrove-fringed section (Fig. 5f).In previous studies, Mn concentrations exceeded conservative mixing values in mangrove-fringed estuaries (Holloway et al. 2016;Mori et al. 2019).However, in our study, Mn concentrations were generally lower than expected from conservative mixing in the mangrove section, even though dissolved Mn concentrations were high during low tide due to recent input from porewaters (Fig. 3).This suggests efficient dissolved Mn removal, for example, by precipitation of Mn oxides in the water column.
Increasing dissolved phosphate concentrations within the upper mangrove-fringed section (Sta.6-13) can be related to porewater discharge, i.e., mobilized phosphate from organic matter degradation and Fe and Mn(hydr)oxides dissolution under anoxic conditions (van Raaphorst and Kloosterhuis 1994).In the parts of the mangrove-fringed estuary where dissolved Fe and Mn precipitated in the water column, dissolved phosphate was probably removed by co-precipitation with metal oxides (salinity < 10 in Fig. 5h).This explains why the increase in dissolved phosphate concentrations was less pronounced in the fluvial and within the upper mangrove-fringed section (salinity < 10, Sta.17-23) than within the lower mangrove-fringed section (salinity > 10, , where most of the dissolved Fe and Mn from fluvial sources had already been removed (Fig. 5h-j, summarized in Fig. 8).
The concentrations of inorganic nutrients were significantly different between low and high tide samples during spring tide ( p < 0.05, Supporting Information Table S1, Fig. 3) when Atlantic seawater was pushed further upstream into the estuary (Call et al. 2019), resulting in greater dilution with seawater during high tide and, consequently, a greater dilution of the porewater-derived inorganic nutrients (Call et al. 2019).The higher-level mangrove areas are only flooded during spring tides (Asp et al. 2018).Thus, the flushing of the mangrove forests during spring tide in the dry season results in the outwelling of aged porewater from the upper sediments into ) were not significant (light gray vectors).Numbers refer to the respective stations along the river to coastal ocean sampling transect.
the mangrove creeks (Call et al. 2019).Aged porewater input results in enhanced pCO 2 and CH 4 concentrations in the water column during spring compared to neap tide in the Caeté estuary (Call et al. 2019).We found lower dissolved phosphate concentrations in the water column during spring than during neap tide conditions (Fig. 3h,i, p < 0.05, Supporting Information Table S1).Dissolved phosphate and Fe were likely depleted in the upper, semi-flushed mangrove sediments due to biological uptake and Fe removal as monosulfides and disulfides (Canfield 1989;Seidel et al. 2014), possibly resulting in the observed phosphate concentration below the detection limit.Percolating estuarine water during spring tide can also lead to Fe(hydr)oxides formation with phosphate co-precipitation.The outwelling of "aged" phosphate and Fe-depleted porewater from the higher mangrove sediments results in lower Fe and phosphate concentrations during spring tide.
TDN excess was most prominent in the mangrove-fringed area, confirming that porewater outwelling is a TDN source in mangrove estuaries (Dittmar and Lara 2001a;Mori et al. 2019).While ammonium and dissolved organic nitrogen (DON) are significant constituents of the TDN pool in mangrove-fringed areas, NO x À is often elevated in the fluvial section of mangrove-fringed estuaries (Reading et al. 2017;Mori et al. 2019), as was also observed in our sampling transect (Supporting Information Fig. S5).In addition to the discharge from the mangroves, high NO x À concentrations near Bragança may result from sewage and untreated wastewater input (Pereira et al. 2010).NO x À decreased at salinities > 10 and was depleted at salinities > 30, while at the same time, TDN was still detectable (Fig. 5), demonstrating a shift from nitrate in the fluvial endmember to ammonium and DON toward the coastal marine endmember (Reading et al. 2017).N obrega et al. (2022) observed elevated abundances of microbes involved with the nitrogen cycle in the low salinity (fluvial) and low oxygen (mangrove-fringed) sections in our sampling transect.Furthermore, nitrate from fluvial sources can be removed through uptake by (planktonic) primary producers in the estuary's less turbid parts and/or by nitrate removal in the oxygen-deficient mangrove sediments (Nedwell 1975).DOC concentrations were above values expected from conservative mixing in the mangrove-fringed section (Fig. 5b).This is likely due to porewater discharge from the mangroves and DOC desorption from resuspended particles in the turbidity maximum zone in this sector (Asp et al. 2018).The average annual DOC export from the mangrove estuary comprises 2.2 Â 10 12 mol C yr À1 (similar to the annual Amazon river discharge) (Dittmar et al. 2006).More precisely, the DOC export during the dry season consists of 0.3 AE 10 mmol DOC m À2 mangrove d À1 in our sampling transect (Cabral et al. 2021).Thus, it takes 22.3 AE 7.2 d for mangrove-derived DOC to cross the shelf toward the open ocean (Cabral et al. 2021).Terrigenous mangrove-derived particulate organic matter, especially lignin, is degraded in the sulfidic sediments and released as DOM to the adjacent waters (Dittmar and Lara 2001b).Thereby, the δ 13 C SPE-DOM signal is relatively reduced by leaching of mangrove litter and possible preferential uptake of enriched δ 13 C-DOC from, e.g., benthic algae (Maher et al. 2013;Mori et al. 2019).The observed major contribution of mangrove-derived DOC to the overall DOC pool within the lower mangrove-fringed section is consistent with previous observations of the transport of mangrove-derived DOC to the coastal ocean (Dittmar et al. 2006).
The geochemical parameter distribution in our study revealed terrestrial and mangrove-derived DOC input to the estuary.Still, a marine DOM influence from the nearshore ocean was evident from the tide-dependent changes of DOC concentrations and δ 13 C SPE-DOM values (Fig. 3).However, changes in bulk DOC concentrations and δ 13 C SPE-DOM values may not necessarily reflect compositional DOM changes.Therefore, in the following, we will explore the optical and molecular DOM properties to decipher the unique mangrove-derived molecular DOM and FDOM signatures that are carried into the estuary and adjacent coastal shelf by mangrove-derived porewater outwelling.

Transformations of FDOM along the river-to-coastal ocean transect
Sources and relative intensities of FDOM in the Caeté estuary varied between the different endmembers from the fluvial, mangrove-fringed, and coastal marine sections (summarized in Fig. 8).The relative increase of the humic-like FDOM signal at low tide corresponded to the observed high DOC and nutrient concentrations in the tidal creek, demonstrating that mangrove outwelling is a terrestrial FDOM source to the estuary.
C1 resembled the fluorescence of ubiquitous peak C, commonly found in terrestrially influenced ecosystems (Stedmon et al. 2011).Components C2 and C3 were red-shifted, indicating a higher degree of aromatic substitution, polycondensation, and higher levels of conjugated chromophores (Senesi 1990).C2 and C3 resemble peaks A and C, representing biogeochemically processed terrestrial organic matter (Supplementary Table S3).Terrestrial humic-like FDOM is derived from vascular plants, degraded peats, and litter leaching (Moran et al. 1991), making it difficult to differentiate between these sources (Amaral et al. 2016).
Component C4 resembled protein-like FDOM (e.g., Stedmon et al. 2011), indicating autochthonous and microbial-derived DOM sources (Murphy et al. 2008;Yamashita et al. 2011).The relative fluorescence intensities of component C4 were highest in the coastal marine section and during high tide in the mangrove creek, demonstrating its high abundance in marine FDOM.However, this fluorescence signal was present in the whole terrestrial-marine transect, indicating autochthonous DOM in situ production along the estuary (Fig. 2).
The HIX describes the degree to which FDOM is humified (Ohno, 2002).Higher values indicate more aromatic, redshifted FDOM.The HIX was highest in the porewater samples where components C1 and C2 were enriched (Supporting Information Table S7) since microbial alteration increases humic-like DOM in deeper aquifers of mangrove ecosystems (Xiao et al., 2023).Thus, increased HIX levels during low tide confirm the outwelling of a terrestrial DOM from mangrove porewater to the estuary.During high tide and in the coastal marine section, the HIX decreased (< 0.85, Supplementary Table S7) when terrestrial FDOM was diluted with less aromatic algal-derived FDOM (Hansen et al., 2016).
The BIX represents biological modification and autochthonous FDOM sources (Stedmon et al., 2011a).Higher BIX values in the coastal marine section compared to the fluvial and mangrove sections, as well as during high tide compared to low tide sampling (Fig. 7, Supplementary Table S7), demonstrated a higher contribution of autochthonous (algal and microbial) FDOM (Huguet et al., 2009).The PARAFAC components and fluorescence indices reflected compositional (F) DOM changes in the Caeté estuary.Overall, terrestrial FDOM dominated the fluvial and mangrove-derived sections (C1, C2, C3, enriched HIX) and microbial-derived FDOM in the coastal marine section (C4 and enriched BIX).

Molecular elucidation of terrestrial and mangrovederived DOM
The molecular DOM composition in the Caeté estuary changed from the fluvial to the coastal marine section (summarized in Fig. 8).Fluvial SPE-DOM was characterized by lower H/C ratios and higher abundances of molecular formulae identified as PCA (group 1) and HAC (group 2).DOM molecular formulae in group 1 include mobilized dissolved black carbon (DBC) from soils, whereas DOM group 2 includes highly aromatic material with aliphatic side chains (described in more detail in Seidel et al. 2017).Fluvial DOM is often enriched in these aromatic compounds (Medeiros et al. 2015).The relative abundances of the aromatic-rich terrestrial DOM (groups 1 and 2) decreased from the fluvial and mangrove toward the coastal marine section (Fig. 4, Supporting Information Fig. S4e).However, HACs (group 2) were enriched in the mangrove porewater (Fig. 4c) and abundant in the mangrove-fringed estuary (Fig. 4h).Terrestrial DOM is enriched in mangrove porewater and terrestrially influenced coastal sediments (Schmidt et al. 2009;Mori et al. 2019), demonstrating mangrove-porewater as a source for terrestrial DOM in the estuary.Lignin-phenol-based analysis revealed that terrestrial DOM in the Caeté estuary is mainly derived from degraded mangrove plant material, while degraded land plant material from the fluvial section was efficiently removed in the estuarine mixing zone with increasing salinity (Dittmar et al. 2001).Our molecular DOM analysis confirmed these findings, i.e., the addition of mangrovederived DOM to the fluvial DOM in the mangrove section.Generally, along river-to-coastal ocean transition zones, the aromatic fluvial DOM shifts to a more aliphatic DOM composition with increasing salinity (Medeiros et al. 2015).Proton nuclear magnetic resonance spectroscopy analysis revealed that aromatic compounds were almost completely degraded while functionalized aliphatics decreased significantly during photodegradation of porewater DOM from the Caeté estuary (Dittmar et al. 2006).Thus, apart from dilution with seawater, photodegradation is likely responsible for the relatively lower abundance of aromatic DOM in the coastal marine section (Fig. 4, Supporting Information Fig. S3).In addition, the removal of terrestrial DOM in the estuary can occur through the sorption of aromatic hydrophobic DOM and subsequent flocculation of fluvial particles (Sholkovitz et al. 1978).
Sulfide incorporation into DOM through abiotic sulfurization explains the relative enrichment of S-containing DOM in the sulfidic mangrove sediments (Fig. 4, Supporting Information Table S5; Pohlabeln et al. 2017).Sulfate reduction is a key metabolic pathway within the mangrove-fringed area of the Caeté estuary (Cabral et al. 2021;N obrega et al. 2022).Likewise, the incomplete degradation of N-containing organic matter can lead to an accumulation of recalcitrant N-containing DOM in anoxic porewater (Schmidt et al. 2011).The enrichment of N-and S-containing DOM during low tide may indicate a close coupling of microbes involved in N-and S-cycling and DOM composition in the low-oxygen waters of the mangrove section (N obrega et al. 2022).Apart from the relative enrichment of S-and N-containing molecular formulae, we also observed a relatively higher abundance of highly aromatic molecular formulae (group 2 compounds) in porewater DOM (Fig. 4c-e).The relative enrichment of aromatic DOM results from accumulation due to its relative recalcitrance and preferential degradation of aliphatic DOM (Seidel et al. 2014).
Marine, phytoplankton-derived and microbial DOM contains higher proportions of aliphatic, N-, and P-containing compounds than terrestrial DOM (Kujawinski et al. 2004;Sleighter and Hatcher 2008).The changes in N and P content in DOM along the estuarine transect result from a mixture of microbial transformations and a shift in the DOM composition from a typical terrestrial signature (more aromatic, N-and P-poor) to a more coastal marine signature (more aliphatic, Nand P-rich) from marine primary production (Medeiros et al. 2015).Medeiros et al. (2016) introduced a molecular approach for tracing the fate of terrigenous DOM in the ocean (I Terr and t-Peaks), representing relatively recalcitrant terrestrial DOM.I Terr values and the sum of t-Peaks intensities decreased from the fluvial toward the coastal ocean section, demonstrating a shift to a coastal marine (algal and microbial) DOM signature (Fig. 5k, 6b).Still, changes in the composition and concentrations of lignin-derived phenols indicated an exchange of fluvial, terrigenous DOM by mangrove-derived DOM (Dittmar et al. 2001).To track the fate of mangrove-derived porewater DOM, we applied a similar approach using molecular formulae identified as s-Peaks from FT-ICR-MS analysis.Increasing values of I SuP in the mangrove-fringed estuary were evidence for the addition of mangrove-derived DOM to the fluvial DOM along the estuary (Fig. 6a).
Moreover, our samples were separated in the PCoA according to I SuP and I Terr values (Fig. 7).Thus, a differentiation between fluvial and mangrove-derived terrestrial DOM signatures was possible by applying both molecular indices.The I Terr can be considered a proxy for recalcitrant terrigenous DOM because its molecular formulae were present even in deep oceanic samples (Medeiros et al. 2016).Both t-and s-Peaks were detected in our coastal marine section samples (Supplementary Table S9), demonstrating that a part of the terrigenous and mangrove-derived DOM is recalcitrant enough to be transported to the nearshore ocean.We want to emphasize that the relative abundances of t-and s-Peaks in the samples are not quantitative since the detection of molecular formulae via FT-ICR-MS is not quantitative.However, applying FT-ICR-MS-derived molecular indices such as the s-Peaks index can provide an additional approach for tracing porewater DOM further offshore.Our findings confirm that the outwelling of mangrove porewater is a (potentially) recalcitrant DOS source to the tropical coastal marine ocean (since all s-Peaks contain S).Our molecular tracer may, therefore, be useful to increase our understanding of marine DOS sources and transformations.Combined with quantitative measurements of geochemical tracers, it may be a useful proxy for determining the fate of terrestrial and recalcitrant porewater DOM in the ocean.The consistency of results from our new qualitative molecular proxy (I SuP ) with quantitative molecular proxies (lignin-derived phenols (Dittmar et al. 2001) is encouraging in this context.
Establishing correlative linkages between FDOM components and DOM molecular parameters provides deeper insights into molecular DOM transformations.It helps designing future studies and for the molecular interpretation of fluorescence signals without needing resource-demanding molecular analyses (Stubbins et al. 2014).For example, protein-like FDOM component C4 significantly correlated with highly unsaturated and heteroatom-containing molecular formulae (Table 1), which can be associated with autochthonous algal and microbial DOM (e.g., Kujawinski et al. 2004).
Terrestrial FDOM is generally associated with aromatic (Stubbins et al. 2014) and oxygenated DOM (Herzsprung et al. 2012).Highly aromatic molecular formulae positively correlated with FDOM components C1, C2, and C3 (> 35%) (Table 1).The correlation patterns of the three humic-like components with molecular formulae were very similar, indicating an overall common origin and biogeochemical behavior of the FDOM components in the coastal system (Supporting Information Fig. S4).However, there were exceptions to this general observation.The correlation of component C1 with N-containing molecular formulae in the Caeté estuary indicates agricultural and municipal wastewater inputs (Roebuck et al. 2020).Humic-like FDOM components might be helpful tracers for dissolved black carbon (Stubbins et al. 2014).Our data support this since FDOM compounds C1 and C3 correlated with the abundance of highly aromatic and PCA formulae indicative of terrigenous and pyrogenic DOM.
Nearly half of the molecular formulae related to FDOM component C2 were highly unsaturated with relatively high O/C ratios, considered recalcitrant DOM (Lechtenfeld et al. 2014).Although present throughout the whole estuary, component C2 was relatively enriched in microbially processed terrigenous and S-enriched DOM from mangrove porewater (Fig. 7), which is further evidence of a significant porewater DOS and FDOM source in the estuary.The analytical windows of SPE-DOM molecular analysis via FT-ICR-MS and optical FDOM analysis do not necessarily overlap (Stubbins et al. 2014).However, our results illustrate that combining optical and molecular DOM analyses is a powerful approach for deciphering the fate and transformations of fluvial and mangrove-derived DOM along the river to coastal ocean continuum.

Conclusions
The main purpose of our study was to answer the question of how mangrove ecosystems influence the amount and composition of recalcitrant DOM and essential nutrients in a mangrove-fringed estuary.Precipitation and microbial turnover largely controlled dissolved nutrient and trace metal dynamics in the estuary.Concentrations of inorganic nitrogen species exceeded conservative mixing values in the fluvialmangrove transition zone, demonstrating the outwelling of dissolved nitrogen from the mangrove sediments.In contrast, low phosphate concentrations in the fluvial section indicated removal by co-precipitation with Fe(hydr)oxides, whereas high phosphate concentrations in the mangrove section were related to porewater outwelling.In this part of the estuary, significant co-precipitation did not occur, likely because of little Fe input from the mangrove sediments.
Via the application of our novel molecular index (I SuP ) as a tracer for mangrove-derived porewater DOM, we showed that mangroves were a major DOM source to the estuary and that mangrove-derived DOM was recalcitrant on the time scales of estuarine and nearshore mixing.This is consistent with previous findings that mangroves contribute significantly to DOM far offshore on the northern Brazilian shelf (Dittmar et al. 2006).Here, we show that mangrove-derived DOM is strongly associated with organic sulfur.Since sulfur is incorporated in sulfidic sediments worldwide, the I SuP index may serve as a tracer for DOM derived from sulfidic porewaters in anoxic environments in general.The outwelling of recalcitrant DOM from mangrove habitats is relevant in the blue carbon storage context since local carbon burial budgets alone may underestimate coastal carbon sequestration (Dittmar et al. 2006;Cabral et al. 2021).The application of molecular proxies (e.g., I Terr and I SuP ), in combination with trace-metal and optical DOM analyses, is a powerful approach to understand coastal ecosystem biogeochemistry and to differentiate the relative contributions of different DOM sources in highly complex coastal ecosystems.

Fig. 4 .
Fig. 4. Boxplots of DOM molecular parameters in porewater, low, and high tide samples in the intertidal creek (a-e) as well as in the fluvial, mangrovefringed, and coastal ocean sections of the sampling transect (f-j).Displayed are intensity-weighted molecular averages of H/C (a and f) and O/C ratios (b and g), as well as relative abundances of highly aromatic molecular formulae (group 2 compounds, HAC, c and h), N-or S-containing molecular formulae (d and i or e and j, respectively).Asterisks indicate if the molecular parameters differed significantly from the other samples (p < 0.05, Wilcoxon Rank-Sum test).

Fig. 6 .
Fig. 6.Observed trends for I SuP and I Terr values (a) along the river-to-coastal ocean transect, and the relationship of I SuP vs.I Terr values (b) with colorcoded dissolved Ba concentrations and station numbers, that is, 1-5 coastal marine, 6-18 mangrove-fringed, and 18-23 fluvial section (see Fig. 1).

Fig. 7 .
Fig. 7. PCoA based on a Bray-Curtis dissimilarity matrix of the relative abundances of FT-ICR-MS derived molecular formulae in the mangrove tidal cycle and porewater samples, as well as transect samples from fluvial, mangrove, and coastal ocean sections.FT-ICR-MS-derived compound classes and molecular properties (blue vectors), environmental parameters, and PARAFAC components (dark gray vectors) were fitted post hoc to the eigenvalues of the PCoA (significant with p < 0.05).The percentages give the explained variability for each axis.DOM molecular compound groups were: PCA, polycyclic aromatic; HAC, highly aromatic; HUC, highly unsaturated; UA, unsaturated aliphatic compounds; N, molecular formulae containing N; S, containing S; P, containing P. Correlations with PARAFAC component C1 and dissolved phosphate concentrations (PO 4 3À

Fig. 8 .
Fig. 8. Conceptual representation of the biogeochemical parameters along the Caeté River estuary.FDOM components C1 and C3 were present in fluvial and mangrove-fringed sections, component C2 was mainly associated with porewater DOM, and component C4 was enriched in the marine section.NO x À concentrations exceeded conservative mixing values within the upper mangrove-fringed section but decreased strongly at mid-salinity ranges, while phosphate concentrations exceeded conservative mixing values within the lower mangrove-fringed section.PCA, polycyclic aromatic compounds; HAC, highly aromatic compounds; HUC, highly unsaturated compounds; UA, unsaturated aliphatic compounds; S and N, S-or N-containing DOM compounds.# FT-ICR-MS derived parameters.This graphic was produced using the Integration and Application Network (IAN, ian.umces.edu/media-library).