Distribution and origin of sedimentary organic matter in an eutrophic estuary: Pina Sound – NE Brazil

: Eutrophic estuaries receive organic matter (OM) inputs from multiple sources. This study evaluated the distribution and origin of sedimentary OM in an eutrophic estuary (Pina Sound, NE Brazil). Thirteen samples were collected in the sublittoral in addition to major local sources. Biochemical (chlorophyll - Chl), elemental [(C/N) a and C/S ratios] and isotopic ( δ 15 N and δ 13 C) analyses were carried out for characterizing OM and redox conditions. The SIAR mixing model was used to quantify contribution from main sources. At Pina Sound, distribution of OM is associated with mud, refl ecting the hydrodynamics control on deposition of suspended particles. Microphytobenthic production is limited ([Chl a ] < 1000 µg/g organic carbon) in the sublittoral where the Chl degradation products prevail (mean [Pheopigments] = 2643 ± 958 µg/g organic carbon). Anoxic conditions (C/S ratio ≈ 2) are typically observed in sediments of deeper portions of Pina Sound. Such sediments receive high organic loads and are subject to poor water renewal. According to SIAR mixing model, sedimentary OM of Pina Sound is composed of, on average: 50% phytoplankton, 24% sewage and 26% C 3 plants. This refl ects fertilization of Pina Sound with high loads of untreated sewage. Pina Sound has a great potential to retain sewage-derived OM.


INTRODUCTION
Estuaries are ecosystems that retain organic matter (OM) from terrestrial and aquatic sources (Dittmar et al. 2001, Andrews et al. 2008. Additionally, human-impacted estuaries receive nutrients and OM inputs from anthropogenic sources (Mcclelland & Valiela 1998), increasing productivity of aquatic primary producers (Nixon 1995). Such OM inputs are preserved in sedimentary OM (SOM) that refl ects proportional contribution from each source (Lesen 2006, Canuel & Hardison 2016. Thus, SOM is an environmental compartment that records the historical inputs of OM to urban estuaries (Andrews et al. 2008, Barcellos et al. 2017 and it contributes to CO 2 sequestration from atmosphere (Watanabe & Kuwae 2015).
Estuaries exhibit a characteristic pattern of OM mixing from terrestrial and marine sources (Gearing 2013). Contribution of terrestrialderived OM to SOM decreases from upper to lower estuary while marine contribution increases toward the ocean (Gireeshkumar et al. 2013, Sarkar et al. 2016. In contrast, SOM is predominantly derived from aquatic primary producers in eutrophic coastal systems (Zimmerman & Canuel 2001, Carreira et al. 2002, Zhao et al. 2015, Kubo & Kanda 2017. The origin of SOM has been evaluated using biochemical (chlorophyll and pheopigments), elemental (carbon-to-nitrogen -C/N -ratio) and isotopic (stable carbon and nitrogen isotope ratios -δ 13 C and δ 15 N, respectively) proxies (Hardison et al. 2013, Canuel & Hardison 2016. These proxies allow to quantify the relative contribution from multiple OM sources (C 3 higher plants, phytoplankton, benthic algae and domestic sewage) to SOM (Watanabe & Kuwae 2015, Kubo & Kanda 2017. The relative contribution of bulk sewage-derived OM to human-impacted coastal environments has been scarcely reported in the literature (Tucker et al. 1999, Kubo & Kanda 2017. This gap needs to be filled in order to understanding the fate and the impacts of sewage-derived OM on marine ecosystems (Spano et al. 2014, Roth et al. 2016. OM is preserved in sediments according to its composition, sedimentary mineralogical composition and redox conditions in interstitial water (Burdige 2007, Arndt et al. 2013, Barber et al. 2017. Redox conditions regulate the early diagenesis of SOM and diagenetic shifts of its isotopic signature (Freudenthal et al. 2001, Lehmann et al. 2002. Water anoxic conditions favor preservation of OM in sediments (Arndt et al. 2013), and have been commonly reported in eutrophic estuaries (Pinckney et al. 2001). Thus, SOM preserved in anoxic conditions is an appropriate record of long-term OM inputs to eutrophic estuaries.
Pina Sound (2 km 2 ) is an estuary located on the northeastern coast of Brazil (8° S). The sound is delimited by the urban zone of Recife city (218 km 2 ) whose population increased from 1.3 to 1.6 million inhabitants over the past 20 years (IBGE 2019). Pina Sound receives every day inputs from untreated domestic sewage with an estimated outflow between 0.81 and 2.31 m 3 s -1 (IBGE 2011. As a consequence, an eutrophic to hypertrophic and hypoxic to anoxic conditions have been reported at surface and bottom waters of Pina Sound, respectively (Flores Montes et al. 2011, Nascimento et al. 2003, Somerfield et al. 2003. Pina Sound also receives natural OM inputs from local Atlantic forest (~13.4 km 2 ) and mangrove patches (~3.2 km 2 ) (Ferreira & Lacerda 2016). The apportionment of OM inputs from natural and anthropogenic sources is important for understanding their impacts on Pina Sound.
This study investigated the distribution and origin of OM in sediments of Pina Sound, northeastern Brazil. Major potential OM sources to the sound were characterized in terms of elemental and isotopic composition. A stable isotope mixing model was employed for estimating the relative contribution of OM sources to SOM of Pina Sound. Additionally, a non-metric multidimensional scaling was performed for visualizing sample grouping according to distribution and sources of OM. Finally, a factor analysis was also performed for identifying major latent dimensions related to distribution, origin and early diagenesis of SOM in Pina Sound.

Study area
Pina Sound is a tropical ecosystem with mean annual temperature and rainfall of 26 °C and 2450 mm, respectively (Schettini et al. 2016a). The sound is formed by the confluence of Pina, Jordão and Tejipió creeks in addition to the southern branch of Capibaribe River (Fig.  1). Annual mean outflow of Capibaribe is 11 m 3 s -1 , ranging from 2 m 3 s -1 in the dry season (September to February) to 30 m 3 s -1 in the wet season (March to August; Schettini et al. 2016a). Concentration of suspended particulate matter has been reported to be in the range 10-60 mg L -1 (Nascimento et al. 2003, Schettini et al. 2016b and mean sedimentation rate is estimated to be 0.45 cm year -1 (Xavier et al. 2017). The sound is a shallow depositional environment with sand bars and mud flats exposed during low tide (Feitosa et al. 1999). In the dry season, algal mats grow on mud flats (Santos et al. 2009) while they are not observed during wet season. Mats are mainly composed of cyanobacteria but may also contain diatoms (Santos et al. 2009, Valença et al. 2016, and are visually identified by their typical blue-green color. In Pina Sound, estuarine phytoplankton production ranges between 2.70 and 256 mg m -3 throughout the year (Feitosa et al. 1999, Nascimento et al. 2003.

Sampling
Surface sediment (top 10 cm) was collected using a stainless steel van Veen grabber from 13 sites at Pina Sound in December 2014 (Fig. 1). In the laboratory, samples were homogenized and frozen at -20 °C until further analysis.
Three potential sources of OM in the sound were sampled: algal mats (AM), suspended particulate OM (SPOM) from an untreated sewage outfall and leaves from higher C 3 plants (HP) -terrestrial plants and mangrove. Marine phytoplankton production in the adjacent shelf was not considered as an important OM source to Pina Sound. Marine primary production is about an order of magnitude lower than estuarine production, with concentrations ranging between 0.05 and 5 mg m -3 throughout the year (Resurreição et al. 1996). Elemental and isotopic signatures of estuarine phytoplankton were reported by Costa (2018). Costa (2018) sampled estuarine SPOM during complete tidal cycles at lower and middle portions of the Capibaribe River estuarine system. The AM samples (n = 5) were taken from mud flats during low tide using a stainless steel spatula. Samples of estuarine SPOM (n = 11) and SPOM from untreated sewage (n = 13) were collected in a plastic bottle (250 mL) and cooled until filtration in the laboratory. Fresh leaves of terrestrial plants (n = 3) and mangrove Avicennia schaueriana (n = 1) were collected using a stainless steel scissor, stored in aluminum containers and processed in the laboratory as soon as possible.

Grain size analysis
Subsamples of sediment (50 g) were oven dried at 60 °C for at least 96 h. OM was removed with H 2 O 2 (10%, v/v) and grain size was determined according to Suguio (1973). Briefly, samples were wet sieved through a 63 µm sieve with distilled water. The fraction > 63 µm was dry sieved for separating sand (63 to 2000 µm) and gravel (> 2000 µm). These fractions were weighed for determining their contribution to the total sediment. The mud fraction (< 63 µm) was added to a graduated cylinder (1 L) containing sodium pyrophosphate (3.75 mmol L -1 ). Silt (4 to 63 µm) and clay (0.5 to 4 µm) fractions were sampled at specific settling time and depth according to the Stokes' law. These fractions were then weighed for determining their contribution to the mud fraction. Results were plotted on the Pejrup's triangular diagram (Pejrup 1988) which is suitable for classifying sediments according to particle texture and hydrodynamic conditions of the depositional environment.

Chemical analysis
Subsamples of wet sediment were freeze-dried for 24 h in the dark to avoid Chl degradation and ground using mortar and pestle. Pigments were extracted from 0.5 g of sediment (in triplicate) with 10 mL acetone (90%, v/v) for 20 h at -20 °C. Pigments were measured with a spectrophotometer using absorbance readings at 630, 647, 664, 665 and 750 nm. Chl a and pheopigments (Pheop) were estimated according to the Lorenzen's equations (Lorenzen 1967). In addition, Chl b and c (c 1 + c 2 ) were estimated according to equations reported by Jeffrey & Humphrey (1975). Concentration of sedimentary pigments was OC-normalized in order to assess the microphytobenthos contribution (Moreno & Niell 2004). Analytical precision (standard deviation, SD) ranged from 0.4 to 19 µg g -1 dry weight.
For elemental analyses, sediment aliquots (1 g) were weighed in centrifuge tubes and acidified with 10 mL HCl (1 M) for 72 h to ensure complete dissolution of carbonates (Hedges & Stern 1984). After acidification, the aliquots were washed with distilled water and oven dried at 60 °C for 24 h. Carbonate-free sediments were weighed in Sn capsules and analyzed for elemental [total nitrogen (TN), organic carbon (OC) and total sulfur (TS)] and isotopic (δ 15 N and δ 13 C) composition.
Water samples were filtered through Whatman GF/C membrane (Ø = 45 mm) and oven dried at 60 °C for 24 h. One-eighth of each filter was wrapped in tin disk prior to elemental and isotopic analyses. Leaf samples were washed with distilled water for removing salts. Leaf and AM samples were oven dried at 60 °C for at least 24 h and ground using mortar and pestle. Aliquots were weighed in tin capsules. All elemental and isotopic analyses were carried out using an elemental analyzer coupled to an isotope ratio mass spectrometer (EA-IRMS).
The stable N and C isotope ratio values were calculated using delta notation (δ 15 N AIR for TN and δ 13 C VPDB for OC, respectively) according to Eq. 1 (Coplen 2011).
(1) C/N atomic ratio -(C/N) a -and C/S weight ratio were calculated according to Hedges & Stern (1984) and Berner & Raiswell (1984), respectively. Average precision of sample replicates was 0.09% and 0.30‰ for elemental and isotopic analysis, respectively.

Modeling
The SIAR mixing model (version 4.0) was used for quantifying OM contribution from natural and anthropogenic sources (Parnell et al. 2008). Model input data were (C/N) a , δ 13 C and δ 15 N values of sediment and potential local OM sources (including their respective standard deviation), and discrimination factors (DF). This parameter is the magnitude of change in signatures during early diagenesis (Bond & Diamond 2011). DF was assumed to be 0 in OM (C/N) a ratio and δ 13 C, and DF ranging from -5 to +5‰ in OM δ 15 N. This is the range of DF values observed by Lehmann et al. (2002) during laboratory experiments. The model was run through 5 x 10 5 iterations using the 'siarsolo' command (Parnell et al. 2008). Output was mean estimate of the contribution from each source and its 95% credible interval.

Statistical analysis
Pearson product-moment correlation was performed for investigating relationships between sedimentary elemental contents and isotopic ratios. Linear regression between TN and OC was performed for confirming prevalence of organic nitrogen (ON) in the sedimentary N pool. The critical level of significance for all statistical tests was set at α = 0.05.
Data were log transformed [log(50 + x)] and normalized previously multivariate analysis (Hair et al. 2006). Potential relationships among variables (mud content, OC, sedimentary pigments (Chl a + Pheop), (C/N) a , δ 13 C, C/S, Pheop/Chl a and δ 15 N) were summarized using a factor analysis (FA) (Hair et al. 2006). Factors with eigenvalues above 1 were extracted using the principal component (PC) method followed by varimax rotation (Kaiser 1970, Hair et al. 2006. Two non-metric multidimensional scaling (MDS) plots were constructed using the Euclidian distance similarity matrix (Clarke & Warwick 2001). MDS plots evaluated sample grouping according to sand content and OM contribution from HP.

Granulometric composition, sedimentary organic matter distribution and redox conditions
Pina Sound sediments exhibited sand and mud contents ranging from 7 to 80% and from 16 to 91%, respectively (Table I). Samples were separated according to their sand content into two groups: sandy sediments with higher (≥ 45%) sand content collected mostly from central portions of Pina Sound, and sediments with lower sand (< 45%) content collected mainly from bank portions (Fig. 1). Samples had a predominance of clay in the mud fraction and they were plotted along hydrodynamic section II of Pejrup's triangular diagram (Fig. 2).
Mean TS and C/S ratio were 0.94% (range: 0.11 to 1.36%) and 3.14 (range: 1.66 to 9.58), respectively (Table I). C/S ratio was lower than 2.5 in sediments from upper sound (sites [1][2][3][4][5] and at the confluence with the Capibaribe River main stem (site 13; Fig. 1). Site 10 showed the lowest TS content and the highest C/S ratio (Table I) because it was collected from a sand bar exposed to the atmosphere during low tides.

Elemental and isotopic signatures of OM sources
HP, sewage-derived SPOM, AM and estuarine phytoplankton were considered as potential OM sources. Local HPs have the highest (C/N) a ratio and the lowest δ 13 C and δ 15 N values (Table II). The lowest (C/N) a value and the highest δ 13 C and δ 15 N values were reported in AM samples collected from the intertidal zone of Pina Sound (Table II). Sewage-derived SPOM exhibited intermediate (C/N) a , δ 13 C and δ 15 N values (Table II). Elemental and isotopic signatures of phytoplankton collected in the Capibaribe River estuary are reported by Costa (2018). According to the author, phytoplankton had mean (C/N) a , δ 13 C and δ 15 N values around of 6.67 ± 0.33, -25.83 ± 1.37‰ and -1.57 ± 1.18‰, respectively, in the middle estuary, and 6.22 ± 0.39, -21.14 ± 1.95‰ and +3.36 ± 3.53‰, respectively, in the lower estuary.  (Table I).

Distribution of SOM, redox conditions and diagenesis of labile OM
Distribution of SOM in Pina Sound was evaluated by bulk parameters (TN and OC contents) and pigment (Chl a, b and c) contents. Bulk TN and OC parameters were significantly related, indicating that TN is a good estimate for organic nitrogen. This allows the use of (C/N) a to infer SOM sources (Hedges et al. 1986). A strong relationship between TN and OC has been commonly reported for organic-rich estuarine sediments, and indicates sorption of SOM onto clay minerals (Andrews et al. 1998, Resmi et al. 2016, Sarkar et al. 2016. Sedimentary Chl is a proxy for both planktonic and benthic primary production (Burford et al. 1994). Low concentrations ( Redox conditions at the sediment-water interface were evaluated using TS and C/S ratio as proxies (Berner & Raiswell 1983. Boundary TS and C/S values for oxic marine sediments are 0.6% and 2.8, respectively (Goldhaber 2005). Under anoxic conditions, sedimentary TS tends to increase while C/S tends to decrease (Berner & Raiswell 1983, Goldhaber 2005. Bottom water anoxic conditions are commonly reported for eutrophic estuaries (Pinckney et al. 2001, Bricker et al. 2008), including Pina Sound (Nascimento et al. 2003, Somerfield et al. 2003. This is related to the balance between sewage discharges and estuarine hydrodynamic conditions (Cardoso-Mohedano et al. 2016), which regulate the dispersal and dilution of sewage. At Pina Sound, sewage discharges (0.81-2.31 m 3 s -1 ) can be similar to river discharge during dry season (≤ 2 m 3 s -1 ) (Schettini et al. 2016a). At Pina Sound, sulfur proxies indicated predominantly anoxic conditions at sites 1-5 and 13 (TS > 1.1% and C/S < 2.4). This is probably related to OM input from streams that drain into the sound (see Fig. 1) and receive high loads of untreated sewage. At sites 6-12, sulfur proxies (TS < 1% and C/S ratio > 3) indicated predominantly oxic conditions at the sediment-water interface. At these sites, mesotides and shallow depths (ca. ~ 2.7 m) facilitate wastewater dilution, leading to oxic conditions in sediment. According to Valença & Santos (2013), there is a high density of macrobenthic fauna (up to 40,000 individuals m -2 ) in surface sediments of the lower sound, which is additional evidence for predominantly local oxic conditions. Diagenesis of labile OM was evaluated using Pheop/Chl a ratio (Rasiq et al. 2016). Such ratio indicated a prevalence of chlorophyll degradation products in sediments of Pina Sound. This might be primarily related to analysis of ancient sediments deposited during the last 20 years (Xavier et al. 2017), which record past planktonic primary production of Pina Sound. Additionally, low light conditions are prevalent in surface sediments from sublittoral zone of Pina Sound (Nascimento et al. 2003). Thus, microphytobenthos does not have an important contribution to surface SOM from sublittoral. Microphytobenthos have been found to play an important role in primary production of bottom sediments where the Pheop/Chl a ratio is close to 1 (Hardison et al. 2013, Valença & Santos 2013, Gontharet et al. 2015.

Signatures of OM sources
At Pina Sound, potential OM sources exhibited distinct elemental and isotopic signatures ( Fig.  3 and Table II). Local HPs and sewage-derived SPOM showed (C/N) a , δ 13 C and δ 15 N values similar to signatures reported in the literature (Dover et al. 1992, Kuramoto & Minagawa 2001, Waldron et al. 2001, Gearing 2013, Francisquini et al. 2014).
The δ 13 C values of AM were similar to the mean δ 13 C (-17‰) reported for marine benthic algae (France 1995, Bouillon et al. 2011, Noh et al. 2019. In contrast, δ 15 N of AM was higher than that reported for marine N-fixing cyanobacteria (δ 15 N ca. 0‰) (Yamamuro et al. 1995). This suggests assimilation of dissolved inorganic nitrogen derived from nitrogen-rich wastewaters discharged into Pina Sound (Rejmánková et al. 2004).
At Capibaribe River estuarine system, estuarine phytoplankton exhibits striking differences in isotopic signatures along the estuarine gradient (Costa 2018). 13 C and 15 N-depleted signatures of phytoplankton in middle estuary suggest fixation of dissolved inorganic carbon (DIC) and assimilation of NH 4 + from depleted pools, respectively (Waser et al. 1998, Montoya 2007, Bouillon et al. 2011. Conversely, 13 C and 15 N-enriched signatures of phytoplankton in lower estuary suggest fixation of marine DIC and NO 3 -, respectively (Montoya 2007, Bouillon et al. 2011). Thus, both depleted and enriched signatures should be included in the mixing model in order to obtain accurate OM contributions from estuarine phytoplankton.

Origin of SPOM and SOM
At Pina Sound, SPOM and SOM are mixtures of OM from HPs, estuarine phytoplankton and sewage-derived SPOM. Apparently, AM is not an important OM source to Pina Sound (Fig. 3). Low contribution of AM is likely a consequence of their seasonal growth (September to February) on restricted areas (intertidal mud flats).
Sedimentary δ 15 N and δ 13 C were not significantly related. A positive correlation between these proxies tend to be observed in estuaries dominated by OM inputs from terrestrial C 3 plants and marine algae (Middelburg & Nieuwenhuize 1998, Gearing 2013. Conversely, the lack of correlation between δ 15 N and δ 13 C has been related to either high OM inputs from a third source (Wada 2009) or diagenetic shifts in δ 15 N of organic nitrogen (Kurian et al. 2013). At Pina Sound, inputs of untreated domestic sewage (an anthropogenic source) are likely the major reason.
Estuarine phytoplankton was the major OM source to both SPOM and SOM, reflecting the eutrophic to hypertrophic condition commonly observed in surface waters of Pina Sound (Flores Montes et al. 2011). High phytoplankton contribution suggests sewage fertilization effect on planktonic primary production. The contribution of sewage-derived OM at Pina Sound was higher than that reported for Tokyo Bay (10%) (Kubo & Kanda 2017). Differences in sewage-derived SOM between Pina Sound and Tokyo Bay likely reflect local treatment of sewage and hydrodynamic conditions of these marineinfluenced systems.
The mixing model revealed that on average HP-derived OM in Pina Sound comprises 26% of local SOM (Table III). Contribution of that source to Pina Sound SOM is lower than those reported for other estuaries and coastal zones (Table III). This is likely related to the small area occupied by Atlantic forest (13.4 km 2 ) and mangrove patches (3.2 km 2 ) in the highly urbanized Recife city (218 km 2 ) (Ferreira & Lacerda, 2016). Additionally, MDS indicated a high contribution of HP to sandy sediments (see Fig. 5). This suggests the accumulation of large (> 63 µm) HP-derived detritus in the sand fraction (Megens et al. 2002).

Major factors controlling SOM
FA was used to infer major latent factors which may be related to distribution and origin of SOM in Pina Sound. PC1 indicated that SOM distribution was regulated by sedimentary mud content (Fig. 5a), reflecting hydrodynamics control on deposition of fine-grained (silt and clay) suspended particles. This has been commonly observed in estuarine and coastal ecosystems (Keil et al. 1994, Fagherazzi et al. 2014. Estuaries are retention zones for finegrained suspended particles (Schettini et al. 2013). This has been previously observed along the longitudinal axis of Pina Sound , resulting in high local sedimentation rate (0.45 cm year -1 ) of fine-grained suspended particles (Xavier et al. 2017). At Pina Sound, calm hydrodynamic conditions are prevalent according to indicated by the Pejrup's triangular diagram (Fig. 2) (Pejrup 1988). FA also indicated that anoxic conditions were inversely related to SOM content (Fig. 5a), reflecting the control of dissolved oxygen concentrations on preservation of sedimentary OC (Goldhaber 2005).
PC2 is likely related to OM degradation/ preservation that is directly influenced by sediment redox conditions (Lehmann et al. 2002). Low C/S ratio indicates anoxic sediments that facilitate OM preservation. In contrast, high Pheop/Chl a along with high δ 15 N reflect OM degradation. A positive shift in δ 15 N of SOM may be related to selective removal of labile organic compounds (e.g. Chl a) or microbial fractionation during degradation of N compounds (Freudenthal et al. 2001). On average, Chl a is 15 N-depleted by 5‰ when compared to the total biomass of marine primary producers (Sachs et al. 1999). Thus, selective degradation of Chl a would result in 15 N enrichment of the non-degraded biomass of primary producers. Intense N isotope fractionation occurs during microbial assimilation of aminoacids, altering the δ 15 N value of bulk SOM (Macko & Estep 1984,  PC3 probably reflects the conservative mixing of protein-rich and 13 C-depleted OM from estuarine phytoplankton (mean C/N = 6.5 and mean δ 13 C = -23.49‰), and 13 C-enriched OM from sewage SPOM (mean C/N = 9.2 and δ 13 C ≈ -20.73‰).

CONCLUSIONS
Distribution of SOM is regulated by hydrodynamic conditions in Pina Sound. Sediments from subtidal zone have limited microphytobenthic production with predominance of chlorophyll degradation products. Pina Sound exhibits nonuniform redox potential at the sediment-water interface, with anoxic conditions prevalent in the upper sound and at the confluence with the Capibaribe River main stem. SOM of Pina Sound is predominantly composed of OM from estuarine phytoplankton and sewage followed by a lower contribution from higher plants. This reflects fertilization of Pina Sound by high loads of untreated domestic sewage. Additionally, Pina Sound has a great potential to retain sewagederived OM and adsorbed contaminants.