The central coast of São Paulo state is under the domain of the Wet Tropical Climate, where the rainfall index is characterized by high rainfall with an average above 200 millimeters during the summer months (December-March) (Nunes, 1997). Considering the hydrological system, the Bertioga Channel presents several tributary rivers that came from the Santo Amaro Island (Santo Amaro mountain) and the continent (Serra do Mar—mountain chain), which present short paths due to the narrow plain. Thus, the majority of these tributaries act as riverine-associated inlet systems, thereby being dominated by freshwater runoffs during rainy periods and tidal currents during drier periods (Alfredini & Arasaki, 2009).

Despite the industrial and harbor activities around the neighboring Santos Channel, visibly affected by the economic activities, the banks of the Bertioga Channel are still dominated by a dense mangrove forests (Schmiegelow & Gianesella, 2014). Santo Amaro Island belongs to Guarujá city that presents a fixed population of about 300 thousand inhabitants (IBGE, 2014), but which can duplicate due to the tourist arrival in the summer months. The city has presented critical problems concerning the disposal of wastes over the last decades, and thus the local authorities took action to amplify the Saco do Funil Landfill (Engeoconsult, 2011) (Figure 1).

The Crumaú and Maratuã rivers (located in northern of the Santo Amaro Island) are physically similar, presenting a meandering system conditioned by bidirectional fluxes (tidal and fluvial). The depths range approximately from 1 to 4 meters in most of their extensions (~8 kilometers), and the width are maximum at the mouths (~90 meters). However, the Crumaú River is directly influenced by wastewaters from a small suburban area (neighborhood Morrinhos), agricultural areas (mainly banana crop), and a Landfill (Saco do Funil) (Figure 1). On the other sense, the mangrove areas over the Maratuã riverbanks are in a better preservation state (Sutti et al., 2012).

Both rivers reach the central portion of the Bertioga Channel (Largo do Candinho), where the tide currents meet (tidal-wave convergence). From this region up to Bertioga bar (Figure 1), Miranda et al. (1998) classified the estuary as a partially mixed type, which changes from highly stratified at neap tides to moderately stratified at spring tides. Meanwhile, the increased influence of river discharge and the shallower areas on the upper zone create sub regions where there is dominance of ebb current that drives the stratification (Seiler et al., 2020).

Water samplings were carried out in October/2012 (drier period) and January/2013 (rainy period) in two points of the Crumaú river, i.e. upstream (station 1) and downstream (station 2) and one in the Maratuã downstream (station 3) (Figure 1) to determine E. coli and nutrients. In situ data (flow velocity, water levels temperature, salinity and dissolved oxygen) were measured at these points. The rainfall data were obtained in an official meteorological station located in the Santos city (CIIAGRO, 2013). Along the water system, geographic coordinates and the depths were obtained using a Sonar & GPS M52 (Lowrance®).

In the downstream stations (2 and 3) (Figure 1), the water flow velocities and the water level were measured every 30 minutes (covering the most of the spring tide cycle), respectively by mechanic fluxmeter (General Oceanics® 2030R6) and a metric ruler. In these stations, the estuarine water samples and in situ measures took place in the ebbing tide phase. The fluxmeter was dipped to a depth of 1m over the deepest area (thalweg) of the transverse section. Meanwhile, the estuarine water samples and in situ measures in the station 1 took place during the phase of tide reversion (slack waters), thereby covering periods no movement of water.

At the same time of each surface water collection, dissolved oxygen (DO) and temperature were measured directly in the water column through an oximeter MO-910 (Instrutherm®). Furthermore, DO values were used as a basis to calculate the percentage of dissolved oxygen saturation (%DO) following Grasshoff et al. (1983). Onboard, water aliquots were taken for the salinity determination using a refractometer RTS-101ATC (Instrutherm®). The estuarine water was sampled in triplicate by van Dorn bottle and transferred to autoclaved polyethylene bottles (1L) that, in turn, were maintained inside of thermal boxes (~4˚C) until the laboratory analysis.

Under laboratory environment, the pH was determined using the parameter B474 (micronal®). A volume (400 mL) of water sample was used for the ammoniacal-N (NH₃-N + NH 4 + -N) determination, which followed the volumetric method 4500-D (APHA, 1999). As described in this method, the ammonia (NH₃-N) acts as a base in aqueous solution, acquiring hydrogen ions from H₂O to yield ammonium ( NH 4 + -N) and hydroxide ions (NH_3(aq) + H₂O_(l) ↔ NH 4 + _(aq) + OH⁻_(aq)).

Another water fractions were filtered in cellulose acetate membranes (AP40—Millipore®) with porosity 0.45 μm to attend the determination of nitrate, nitrite, silicate, and phosphate. These nutrients were quantitated by the colorimetric method using spectrophotometric (E-225-D—CELM®) following APHA (2005) method. The Dissolved Inorganic Nitrogen (DIN) was obtained by the sum of ammoniacal-N ( NH 4 + + NH₃), nitrite ( NO 2 − ), and nitrate ( NO 3 − ). Due to the low concentrations of NO 2 − , the NO 2 − + NO 3 − concentrations were expressed as NO 3 − .

The Escherichia coli concentration was determined by the membrane filtration technique (APHA, 2005), which consisted of the filtration of a known volume of sample (after successive dilutions) through sterile membranes (porosity 0.45 μm). After this step, the membranes were put in Petri dishes containing a selective and differential chromogenic crop (Agar Biochrome Coliform—Biolog®). The results were expressed by units of bacterial colonies to each 100 mL of samples (CFU/100mL). The results obtained in triplicate were expressed in column graphics as average (±standard deviation) using the GraphPad Prism® for Windows (version 5.03). The same software was used to generate the Pearson correlation (significance degree at p ≤ 0.05), which included all parameters except the flow velocities. Each sampling period was represented by one Pearson correlation with 9 samples and 10 parameters.

In January (2013), the rainfall volume was more than twice the October (2012) rainfall one, reaching values similar to the historical average under data computed since 1960 (CII AGRO, 2013). Also, it is important to highlight the expressive rainfall that occurred before the sampling days of January/2013 (Figure 2). Considering it, the sampling days of October (2012) represented a period moderately dry of springtime, and January (2013) ones, a common wet period of summertime. In general, the flow velocities in the ebb tide were higher than the observed in flood tide in the downstream stations (2 and 3). In these stations, the highest river flow was indicated by the higher ebbing currents during the rainy period (Figure 3).

The surface water temperature reached the lowest values in the sampling days of the rainy period. The salinity values ranged from 5.0 (station 1—rainy period) to 23.0 (station 3—drier period). During the rainy period, the surface water of the three stations presented pH, dissolved oxygen (DO) and saturation (%DO) values lower than the observed in the drier period, showing the influence of freshwater input by continental drainage in this system. Spatially, these values increased toward the lower estuary (from station 1 to stations 2 and 3).

Overall, considering the variation of salinity as a function of the tidal range and freshwater input, the nutrients presented a great range of variation. The nutrient concentration ranged from 0.65 μmol·L⁻¹ (nitrate in the station 2) to 114.05 μmol·L⁻¹ (ammoniacal-N in the station 1) in the drier sampling, and from 5.82 μmol·L⁻¹ (ammoniacal-N in the station 2) to 112.27 μmol·L⁻¹ (ammoniacal-N in the station 2) in the rainy period (Figure 4). In comparison to the drier period, the rainy period revealed a nitrate and silicate increase and an ammoniacal-N decrease in the downstream stations (2 and 3), whereas in the Crumaú upstream (station 1) was observed a phosphate decrease and a nitrate increase.

In the drier period, the E. coli concentration ranged from 0.00 ( 3 colony forming units (CFU/100mL) in the station 1 (Crumaú upstream). Meanwhile, in the rainy period, the lowest value (3.0 × 10 ¹ CFU/100mL) was found in the station 3, and the highest value (1.47 × 10 ⁴ CFU/100mL) in the station 1 ( Figure 4).

The Pearson correlation matrix showed important correlations between nutrients and E. coli contributing to infer the sewage influence on the eutrophication and sanitary risks in this system. The correlations established between the physicochemical variables (T, S, DO, and pH) and the nutrient and E. coli concentrations (Figure 5), in turn, can be useful to understand the influence of

these abiotic members on the microbial metabolism in this system. In general, E. coli presented a significant negative correlation with S, pH, and DO (−0.97, −0.96 and −0.71 respectively) in dry sampling, maintaining this behavior for S and pH (−0.91, −093 respectively) in the rainy period. In general, it is possible to verify stronger (p < 0.05) correlations between E. coli and all nutrients in the rainy period (i.e. NH 4 + , NO 2 − , NO 3 − , PO 4 3 − and Si of 0.99, −0.99, −0.81, 0.93 and 0.96, respectively). On the other hand, in the drier period, just the phosphate and silicate presented strong correlations (r > 0.96) with E. coli.

Regarding the nitrogen metabolism, it is important to mention that the ammoniacal-N was the most positively correlated to E. coli (0.99 in both periods), whereas the oxidized forms (nitrite and nitrate) presented strong negative correlations (−0.99 and −0.81, respectively) with E. coli, just in the rainy period. These results suggest a higher influence of fresh sewage on the whole aquatic system during the rainy period and a more intense N-oxidation in inner areas of the Crumaú River during the drier period.

First, it is important to mention that station 1 (Crumaú upstream) is located near to the point of reversal current where the local circulation is less intense. Highest values of salinity and weaker ebbing currents observed during the October sampling (2012) indicated a low river flow and hence a high influence of tidal creek dynamic (e. g. a higher efficiency of vertical transport during the tidal cycles) on the Crumaú and Maratuã mouths (stations 2 and 3). Also, the long period under rainfall absence before the sampling days in this period probably allowed a high local nutrient regeneration in the Crumaú upstream (station 1), where the water movement is effective just under high river flows.

On the other hand, the sampling period of January (2013) presented a high river discharge driven by the increase of precipitation, which was indicated by the low salinity variation observed between stations 1 and 2 (5 and 9, respectively) and the intense ebbing current in the Crumaú downstream (station 2). This hydrodynamic condition, in turn, reflected a period of water renewal in the Crumaú upstream and hence, favoring a transport of regenerated nutrients toward the lower zone. Besides, it is important to reinforce that the rainfall intensification normally increases the nutrient leaching from watersheds to the aquatic systems.

The strong significant negative correlation between salinity and silicate reflects the silicate input from terrestrial sources (e.g. leaching of sediments and sedimentary rocks) and its dilution by seawater toward the lower estuary. This longitudinal distribution of silicate was documented in several estuaries worldwide (Bell, 1994; Braga et al., 2000; Conley & Malone, 1992). The present study evidenced distinct silicate inputs to the aquatic system considering the evaluated sampling periods. On the one hand, the high rainfall index reported before the January (2013) sampling days (rainy period) probably unleashed diffuse terrestrial inputs from banks located between the stations 1 (Crumaú upstream) and 2 (Crumaú downstream), thereby explaining the silicate increasing observed in the downstream stations (2 and 3). In this case, this increment can be attributed to erosion processes that usually are intensified in deforestation areas during events of high rainfalls.

In contrast, the saline intrusion was evidenced as an important hydrodynamic factor to lead the dissolved nutrients from sediment pore water to the water column during the dry period. This vertical dynamic of nutrients was observed in several estuaries (Ovalle et al., 1990; Paudel et al., 2015; Li et al., 2017). In the phosphate case, the input from sediment can occur with the salinity increase during the saline water intrusion, since the sulfate from sea salt tends to promote the complexation between sulfide and iron, disfavoring the phosphate sequestration from the water column and maintaining it in the dissolved form (Hartzell & Jordan, 2012). In the Pearl River Estuary (Subtropical eutrophic estuary in China), Li et al. (2017) reported that, during a low river flow, a strong vertical mixing increased the levels of dissolved phosphate (1.44 ± 0.57 μmol·L⁻¹) in the surface waters, whereas a river discharge input created a stratification and enhanced the phosphorous transport (associated to the suspended particulate matter) toward the lower zone.

In the Crumaú river, Ferreira (2002) verified at surface sediments, a gradient of sewage contamination indicated by phosphorus concentration, which ranged from 31 (upstream) to 16 mg·kg⁻¹ (downstream). This suggests that the surface sediment in practically all the Crumaú river can represent important phosphate source to the water column, mainly when the saline intrusion (salt-wedge) reaches great distance toward the upper zone. As previously exposed, this dynamic was observed in the dry period under low river flow and high salinities. Meanwhile, the high river flow during the rainy period probably increased the transport of anthropogenic phosphorus from upstream to downstream, thereby explaining the phosphate increment in surface waters of the stations (2 and 3) and the strong correlation performed between this nutrient and E. coli.

The nitrogen transformation in subtropical and tropical estuaries also is highly driven by the magnitude of river discharges (Bianchi, 2007). Souza et al. (2006) reported considerable nitrate (12.1 μmol·L⁻¹) and ammonium (2.8 μmol·L⁻¹) values in the rainwater composition in the Ilha Grande region (about 300 km from Santos estuary), thereby demonstrating that the wet deposition also can be an important nitrogen input to regional estuaries. In turn, the organic matter decomposition in mangrove floodplains is an important input of DIN to the estuarine waters (Dittmar & Lara, 2006), which normally is intensified during the rainfall periods. In the present study, these nitrogen inputs probably were important contributions of nitrate to surface waters, as pointed out by the high concentrations observed in the samples collected in the rainy period. Besides, this period presented a considerable nitrate increment from the station (1) to stations (2) and (3), which probably responded to the leaching increase in agricultural areas over the eastern banks of the Crumaú river and the N-oxidation associated to sewage input in Crumau upstream.

E. coli presented significant strong positive correlations with ammoniacal-N and phosphate, pointing out the sewage sources around the Crumaú upstream as important inputs of these nutrients during the rainy period. The phosphate also is originated from minerals (e.g. apatite mineral), but its source into the terrestrial crust is considerably lower than the silicate one (e.g. aluminosilicate minerals) (Troeh & Thompson, 2007). However, the nitrogen remineralization from the organic matter decomposition can represent an expressive ammoniacal-N input to the water column of polluted estuaries due to the high presence of labile organic matter (e.g. sewage) (Braga et al., 2000; Azevedo & Braga, 2006; Berbel et al., 2015).

The relative high nitrate concentration in the station 3 (Maratuã downstream) observed in the rainy period can be related to the influence of the Crumau River plume during the flood tide, since the Maratuã upstream is not impacted by any significant human activity. The high E. coli presence in the station (3) observed in the rainy period reinforces this hypothesis. The fecal coliforms do not survive for a long time in marine environments, especially due to the action of salinity, high solar radiation, high water temperatures, and ecological factors (such as predation and competition) (Davies-Colley et al., 2008; Hughes, 2003; Jovanovic et al., 2017). Thus, the high river flow characterized in the rainy period probably minimized these inhibitory effects on the development of this bacterial group, thereby allowing a higher traveled distance alive toward the lower estuary. On the other hand, the phytoplankton assimilation (evidenced by the highest %OD values) can have contributed to the lower DIN concentration observed in the downstream stations (2 and 3) during the drier period. According to Ferguson et al. (2004), the DIN uptake by algal blooms and sediments accounted for most of the uptake DIN loading in the Brunswick estuary during low river flows.

According to Piveli and Kato (2006), the high ammoniacal-N percentual within the DIN total can be an indication of recent sewage inputs in urban aquatic environments due to the intense decomposition of organic nitrogen (ammonification process). In typical domestic wastewater, ammonia nitrogen represents about 55% - 60%, organic nitrogen about 40% - 45%, and nitrates plus nitrites together about 0 - 5% of the total nitrogen (Huang & Shang, 2006). These theoretical aspects can explain the high ammoniacal-N concentration found in the Crumaú upstream (station 1), as well as the strong positive correlation that the ammoniacal-N obtained with the E. coli. Also, it is important to report that the leachate from the Saco do Funil landfill can be a DIN input even more expressive than the releases of raw sewage to the Crumaú upstream. The bacterial decomposition makes intensified inside landfills due to the drainage process across the older leachate, from upper to deeper layers, leading to a decrease of organic matter concentration, and consequently contributing to an accumulation of the nitrogen reduced forms (Kalyuzhnyi & Gladchenko, 2004).

Multidisciplinary research has been carried out for years on the environmental status of the Santos estuary, which is one of the most important polluted estuaries in Brazil. The comparatively clean Cananéia estuary nearby (southern coast of São Paulo) is not impacted by any significant human activity and thus has been considered a suitable reference site (Azevedo & Braga, 2006). In comparison to average values obtained by the authors (op. cit.) in the Cananéia estuary, the present study revealed DIN and phosphate values around nine times higher. Meanwhile, similar average values of phosphate (5.5 μmol·L⁻¹) and DIN (150 μmol·L⁻¹) were observed in a tributary river of the Guanabara Bay (about 450 km from Santos estuary) (Brandini et al., 2016).

The highest values of salinity observed in the downstream stations (2 and 3) responded to the greater marine influence on these estuarine sites, explaining the strong correlations performed between salinity and pH values. Generally, river waters present lower pH than marine waters, and a linear increase in the values occur with the salinity increase. Within a tidal cycle, the lower zones (mouths) of tributary rivers usually present a higher water renewal than in upper zones. In the mixture zone, the carbonate-bicarbonate-CO₂ system influenced by the buffer capacity of the seawater acts on the neutralization of H⁺ ions, and consequently, promotes the pH increase. Besides, the pH also can increase with the primary productivity, since the phytoplankton community absorbs CO₂ from water for the photosynthesis process. Under ideal sunlight conditions, Zhang et al. (2019) reported that the estuarine eutrophication resulted in high biomass of phytoplankton and elevated the pH in surface waters. This biological process can have contributed to the pH and DO values observed in the downstream stations (2 and 3), mainly during the dry period due to higher oxygen saturation (% DO).

On the other hand, the organic matter decomposition process in aquatic environments normally produces CO₂ and consumes the DO. This process likely contributed to the lowest pH and %DO (less than 50%) values observed in the Crumaú upstream (station 1), since this local is under low hydrodynamic and is closer to the sewage sources. According to Breitburg (2002), hypoxic conditions in coastal waters are favored under %DO values less than 50%. Diaz & Rosenberg (2008), in turn, reported that DO values lower than 2.5 mg·L⁻¹ is considered one of the most threats to coastal waters worldwide. Moreover, the eutrophic estuarine zones may provide bacterial biomass higher than phytoplankton biomass (Abreu et al., 1992), since the respiration rates in bacteria may exceed phytoplankton production (Giorgio et al., 1997). In both periods, this scenario of heterotrophic metabolism was strongly evidenced in the station 1 (Crumaú upstream).

The concentration ranges of ammoniacal-N observed in the station 1 were similar to reported in bioreactor environments by Anthonisen et al. (1976). These authors, concerning the two steps of the nitrification process in bioreactors, observed that free ammonia inhibited the Ammonia-Oxidizing Bacteria (AOB) at concentrations as high as 10 mg·L⁻¹ (~110 μmol·L⁻¹) and the Nitrite Oxidizing Bacteria (NOB) at concentrations as low as 0.1 mg·L⁻¹ (~1.1 μmol·L⁻¹). Besides, under temperatures higher than 20˚C and alkaline conditions, a considerable amount of ammonium ( NH 4 + ) cations is converted to ammonia gas (NH₃) in the aqueous phase (Huang & Shang, 2006). Despite the pH values next to 7, the high temperatures (26˚C - 28˚C) associated with the high ammoniacal-N concentration (~110 μmol·L⁻¹) likely disfavored the nitrification process in the Crumaú upstream (station 1).

For a long time, Denitrification (Dettmann, 2001; Burgin & Hamilton, 2007) and Ananmox (Trimmer et al., 2003) have been pointed out as the main processes in the transformation of dissolved inorganic nitrogen (DIN) into gaseous products in eutrophic estuaries worldwide. In subtropical estuaries, Fernandes et al. (2012) showed that denitrification is more important than anammox in sediments, whereas Zhu et al. (2018) showed that the water column can present a high denitrifying activity mediated by suspended particulate matter. However, recent reports on the contribution of dissimilatory nitrate reduction to ammonium (DNRA) to nitrogen removal in these systems indicated a similar or higher importance (Koop-Jakobsen & Giblin, 2010; Dong et al., 2011; Giblin et al., 2013). Moreover, the NO 3 − affinity manifested of nitrate ammonifier bacterias is higher than that by denitrifying bacteria at temperatures above 10˚C (King & Nedwell, 1984; Oglivie et al., 1997), and the heterotrophic DNRA is stimulated in environments with high availability of organic carbon (Tiedje, 1988; Yin et al., 2002). These theories suggest that the Crumaú upstream zone, due to the high temperatures and the constant labile organic matter ‘inputs (e. g. sewage), is an environment propitious to this microbiological pathway on the nitrogen transformation.

According to Einsle et al. (1999), DNRA is a facultative, two-step anaerobic process involving nitrate ( NO 3 − ) reduction to nitrite ( NO 2 − ) followed by the 6-electron reduction of nitrite to ammonium ( NH 4 + ). Regarding the last step, the nitrite reduction to ammonium can be catalyzed by the cytoplasmic NADH-dependent nitrite reductase NirB or its to two-subunit variant NirBD and/or the periplasmic pentaheme cytochrome c nitrite reductase NrfA (Harborne et al., 1992). In this case, it is important to know that Escherichia coli was shown to harbor and express genes for both enzymes (Cole, 1996; Wang & Gunsalus, 2000). In addition, Bonin (1996) observed in strains (isolated from coastal marine sediment) of Escherichia coli specie that the accumulation of acetate and formate occurred with NO 2 − reduction to NH 4 + .

The present study has not an analytical structure to point out which was the most important microbiological pathway (Anammox, Denitrification, or DNRA) on the nitrogen transformation at each sampling station. However, we raise the hypothesis that the DNRA obtains importance when E. coli that is enough representative in a group more extended of bacteria in upper zones of subtropical eutrophic estuaries, in this case, submitted to high sewage discharges under periods of high river drainage. The rainy period established a strong negative correlation between the oxidized forms of nitrogen (nitrite and nitrate) and E. coli colonies, thereby evidencing a considerable nitrate reduction associated to this bacterial group. This evidence is reinforced in the Crumaú upstream (station 1) during the rainy period since the heterotrophic metabolism of this area was associated with high values of ammoniacal-N/Nitrate ratio and colony-forming units of E. coli.