Estuary Adjacent to a Megalopolis as Potential Disrupter of Carbon and Nutrient Budgets in the Coastal Ocean

The goals were to estimate nutrients and carbon flow rates between Guanabara Bay and the adjacent coastal waters, to characterize the provenance of the exported/imported organic matter. Samples were collected from different depths over 25 h in two seasons at the bay entrance. Measurements included physicochemical parameters, nutrients, chlorophylls, dissolved organic carbon (DOC), particulate organic carbon (POC), particulate nitrogen (PN), carbon (dC) and nitrogen (dN) isotopic composition, sterols in suspended particulate matter (SPM) and bacterioplankton. Most variables showed higher values in ebb tide events. The flow rates calculated on daily basis and estimated on annual basis revealed the exportation to the continental shelf of 1.27 × 10 kmol year dissolved inorganic nitrogen (DIN), 9.52 × 10 kmol year dissolved inorganic phosphorus (DIP), 2.65 × 10 t year DOC, 1.96 × 10 t year POC, and 2.96 × 10 t year PN. The estimates show the bay contributes with 0.01% of the total global carbon influx to the ocean.


Introduction
][4][5] The cycle of OM in coastal systems, as in estuaries and bays, is highly dynamic and complex.7][8][9] Turbulent mixing of fresh-and sea-water can also generate sudden changes in temperature, salinity, turbidity, pH and bioactive element concentrations in time scale of seconds to months. 7perimposed to the aforementioned natural factors, in many urbanized areas raw or only partially treated sewage is released directly into adjacent aquatic systems.The introduction of these nutrient rich effluents into water bodies with restricted water circulation can cause eutrophication. 10,11It is also the cause of less evident problems including high economic losses, 11 changes in the ecological structure 12 and decreased levels of dissolved oxygen (DO). 13,14In addition to alterations directly affecting the receiving body, the nutrient load may be transported to inner and mid shelf areas, in the dissolved and particulate forms, and thereby changing the carbon balance and fixation rates in the marine system. 15It has been considered, for instance, that nutrient and carbon exports originating from areas contaminated by sewage may be directly related to ocean acidification processes, and associated with climatic changes. 1,15,168][19][20] However, these studies usually Vol.][23] Selman et al. 24 provide strong evidences of the growing global impact of eutrophication, identifying 415 eutrophic and hypoxic coastal systems around the globe, including the coastal region of Rio de Janeiro and other areas in the south-southeastern coast of Brazil.
The warm and nutrient poor waters of the Brazil current dominate a large portion of the Brazilian inner and mid shelf.In the mid shelf area off Rio de Janeiro, on the other hand, the upwelling of South Atlantic central waters (SACW) is a strong seasonal phenomenon that creates areas of enhanced primary production.Besides the fertilization by upwelling, the region receives nutrient and OM inputs from the eutrophic Guanabara Bay (GB), whose hydrographic basin houses nearly 12 × 10 6 inhabitants. 25][28] The high eutrophication of GB results in enhanced local rates of primary production, reaching values of 0.17 mol C m -2 day -1 . 29The annual input of 3.2 × 10 10 mol P and 6.2 × 10 10 mol N, mostly originated from untreated sewage discharge, has been estimated by Wagener. 28A fraction of the highly available nutrients and OM derives from industrial activities producing about 150 tons of effluents per day. 27,30n the present study, GB was taken as a model system to investigate the potential impact of highly eutrophic environments on the nutrient and carbon global balance in coastal zone and to estimate its relevance as a source of these materials to the inner continental shelf.
For this purpose, physicochemical and biogeochemical properties of the water column were studied over tidal cycles in the dry and wet seasons providing also information on the origin of particulate organic carbon (POC) through the use of elemental (C and N) and isotopic (d 13 C and d 15 N) composition.Sterols were also considered as additional indicators of the sources of OM in the studied region.

Study area
Guanabara Bay (Figure 1), one of the largest bays in the Brazilian coast, is circumscribed by the metropolitan area of Rio de Janeiro (22 o 40'-23 o 00' S, 43 o 00'-43 o 20' W) serving as estuary for more than 45 rivers and streams, of which only six are responsible for 85% of total annual freshwater discharge.
The surface area of the bay is approximately 384 km 2 and the average volume of water is around 1.87 × 10 9 m 3 .[33] The influx of fresh water to GB is not sufficient to modify the flow pattern of the water in the bay because the ratio of fresh water runoff to tidal prism for a period of 24 h is less than 5%. 33However, riverine inputs are important for determining some patterns and water mass exchanges.The effects of riverine contribution on the receiving system depend on the use and occupation of the drainage basin.In systems adjacent to highly industrialized areas, such as GB, in addition to nutrient inputs, environmental conditions are significantly altered due to releases of trace metals, hydrocarbons, other potentially toxic substances and OM. 33,34he tidal influence is very important in the bay especially due to the intensity of the current velocities, which range from 80 to 150 cm s -1 at the bay mouth decreasing to 30-50 cm s -1 in the inner areas.The average tidal amplitude is of 0.7 m, varying from 1.1 m in spring tide to 0.3 m in neap tide. 35Salinity varies in the range of 29.5 ± 4.8 and reported average surface water temperature is 24.2 ± 2.6 o C. 33

Sampling
A sampling station (Figure 1) was strategically positioned as to allow the observation of tidal influence and exchange of materials between the bay and the open coastal area.Two sampling campaigns (the dry season sampling (C1) was in June 2011, and the wet season campaign (C2) occurred in November 2011) were performed, and the total monthly precipitation for C1 was 32.7 mm and for C2 it was 93.3 mm.Both samplings occurred in spring tide due to the higher amplitude variation (3.4 m for both) between low and high tides.Water samples were taken at each 2 h from three different depths: surface (S, < 1 m), mid-depth (M, 5 m) and close to the bottom (F, 25 m).Sampling was conducted aboard the parcel ship and the local depth varied from 28 to 30 m.
Salinity, temperature and depth were monitored through a conductive, temperature and depth (CTD) system (SBE, model 19 plus V2 SEACAT; frequency 4 Hz) to obtain a full water profile characterization.The data were processed on MatLab (R2008a) using only the data collected when the instrument was submerged and the average calculation for each 0.5 m of depth.The pressure inversions data were removed and also all the unrealistic values were removed based on the thermohaline index for the local water mass.The temperature and salinity data were filtered in order to remove peaks that exceeded values 3 times the standard deviation around the mean.An acoustic doppler current profiler (ADPC) WorkHorse 600 KHz was installed to obtain current data.Current components were corrected for magnetic decline and a small rotation of 22 o W in relation to the geographic North.DO was measured with Alfakit AT 150 oximeter (range 0.01 to 20.00 mg L -1 ) and the pH was measured with Thermo Scientific Orion 3-Star (range -2 to 19.999), both in situ.
Water samples for organic compounds (sterols), elemental and isotopic compositions (C and N) of the suspended particulate matter were collected at each depth by using an amber glass bottle that was inserted closed into the water column.Bottles containing samples were kept under ice until filtration in the lab.For nutrients and chlorophyll determinations 4L Go-Flo bottles were used for sampling and samples were stored in 1.5 L polyethylene bottles under refrigeration.Samples for bacterioplankton counting (Bact) were stabilized with p-formaldehyde immediately after sampling and thereafter kept in liquid N 2 .Samples for ammonium (NH 4 + ) determination were stored in 50 mL glass tubes with screw cap after fixation in situ as described in Grasshoff et al. 36 For separation of the particulate matter samples were filtered in the lab.Those intended for sterol determination were filtered in Macherey-Nagel 142 mm diameter, 0.7 µm porosity glass filters.Samples for POC, particulate nitrogen (PN), d 13 C and d 15 N were filtered in Macherey-Nagel 47 mm, 0.7 µm glass filters.For chlorophyll determination, filtration through cellulose acetate filters was applied before filter immersion in 90% acetone-water solution following Parsons. 37The filtrate was stored at -20 o C in polyethylene bottles for dissolved inorganic nutrient analyses.

Analytical procedures
Nutrients (nitrate (NO 3 -), nitrite (NO 2 -), ammonium (NH 4 + ), inorganic dissolved phosphorus (P inorg )) were determined following the methods described by Grasshoff et al. 36 Chlorophylls (Chl-a, Chl-b and Chl-c) were determined using the method of Parsons 37 and equations given by Jeffrey and Humphrey. 38A DMS 100 UV-Vis spectrophotometer (Intralab, PerkinElmer) was used and all variables were determined in duplicate.The analytical variability was < 2%.Quality control included calibration curves with r > 0.99.Blanks were treated as samples for zero calibration in the spectrophotometer.Limits of detection (LOD) for nutrient determinations were the lowest concentration detected with precision 0.01 µmol L -1 for NO 3 and NO 2 -; 0.1 µmol L -1 for NH 4 + ; and 0.02 µmol L -1 for P inorg .Bact was analyzed by flow cytometry. 39,40issolved organic carbon (DOC) was determined in a Shimadzu total organic carbon (TOC)-VCPN analyzer (samples determined in duplicate).The LOD was equal to 3 × standard deviation (SD) of 10 blanks / slope of the calibration curve = 0.35 mg L -1 and limit of quantification (LOQ) = 3.33 × LOD was 1.15 mg L -1 .
The filters used for particulate carbon and nitrogen were sub sampled and only the ones used for POC and d 13 C analysis were treated with HCl fumes for over 18 h to remove carbonates and dried at 60 o C overnight.Filter blanks were treated in the same way as the samples.The standard reference material (SRM) 1944 of The National Institute of Standards and Technology (NIST) (New York/ New Jersey Waterway Sediment; TOC: 4.4 ± 0.3% d.w.) was used for accuracy testing and for n = 4 the average TOC concentration obtained was of 4.65 ± 0.29% d.w.The accepted variation coefficient was 0.06%; when higher values were obtained for duplicates sample analysis was repeated.A Thermo Scientific Flash 2000 elemental analyzer was used for POC and PN determination.Instrumental calibration was performed with standard aspartic acid (C = 36.09%d.w., N = 10.52% d.w.) and only calibration curves showing r = 0.999 or higher were accepted.LOD and LOQ were calculated using the smallest detectable amount of the standard and n = 7.As the minimum mass was used in these measurements LOQ was considered equal to LOD: 0.003 mg C and 0.01 mg N. At each 10 samples batch a standard was analyzed to check the calibration.Vol. 27, No. 10, 2016 d 13 C and d 15 N of the particulate organic matter was determined in a Thermo Scientific Flash EA 1112 Delta V Plus isotope-ratio mass spectrometer (IRMS).For reference gases calibration and determination of LOD and LOQ the standard reference material U. S. Geological Survey (USGS) 40 (L-glutamic acid) purchased from the International Atomic Energy Agency (IAEA) (-26.389± 0.042‰ Vienna Pee Dee Belemnite (VPDB) for d 13 C and -4.5 ± 0.1‰ air N 2 for d 15 N) was used.Reported values are the mean of triplicate determinations and those considered as valid measurements were generating pulses > 500 mV for carbon and > 800 mV for nitrogen.
Extraction of sterols was based on the U. S. Environmental Protection Agency (EPA) 3540C method. 41ilters containing the suspended particulate matter (SPM) were freeze dried and weighed before addition of the surrogate standard (5α-androstan-3β-ol) and Soxhlet extracted with 250 mL of dichloromethane over 24 h.Following extract concentration in a rotary evaporator, solvent was exchanged to hexane before fractionation in a glass column (7 g alumina, 2% deactivated silica gel and 1 g Na 2 SO 4 ) to isolate aliphatic (F1) and aromatic hydrocarbons (F2) not considered in the present work.Only the fraction of interest, containing the sterols (F3), was then cleaned up using dichloromethane:methanol (9:1).To F3, evaporated to dryness under N 2 flow, 250 µL of acetonitrile and 100 µL of bis(trimethylsilyl)trifluoroacetamide (BSTFA) were added for derivatization under 80 o C for 1 h.Lastly, the extracts were dried and the final volume adjusted to 1 mL with dichloromethane after addition of cholestane (2500 ng) as internal standard.

Statistical data evaluation
The software Statistica 11 was used for data evaluation through Pearson correlation test, Kruskal-Wallis test, and principal component analysis.Further information is given in Results and Discussion.All discussions referring to differences or similarities are based on the results of the statistical analysis using p < 0.05 for decision.

Characterization of the water column and physicochemical properties
The average values, standard deviations and range for all physicochemical and geochemical parameters are given in Table 1.
The temperature (T) and salinity (Sa) diagram revealed distinct water column stratification especially in C2 (Figure 2).
In C1, temperatures were around 22 o C at all depths and the salinity varied slightly in the range of 32 to 34.5, with lower values in the surface.These values indicate that coastal waters, which are characterized by Sa < 33, dominated most samples collected in C1 at all depths.In C2, there was a linear correlation between T and S, marked by predominance of cold (T < 18 o C) and saline waters (Sa > 34.5) at the bottom and in mid-depth.These properties indicate the influence of SACW identified in the bottom by temperatures below 20 o C and salinities in the range of 34.6 to 36.2. 42The plot of salinity against temperature for C1 and C2 can be seen in Supplementary Information Figure S1.The current intensity and direction is shown in Figure 3.
The pH values were higher in the surface due to interaction between respiration, production and wateratmosphere exchange processes.The consumption of CO 2 by photosynthesis causes an increase in pH, while the production of CO 2 during respiration leads to a decrease in pH.In general pH was more elevated in C2 (p < 0.05), especially at the bottom, due to the intrusion of SACW. 43O saturation varied from 131% at the surface to 89% at the bottom in C1 and from 75 to 39% in C2.The significant positive linear correlations between DO and Chl-a (r = 0.635; p < 0.01), of DO and bacteria counting (r = 0.550; p < 0.01), as well as of Chl-a and bacteria counting (r = 0.423; p < 0.01) only in C2 may be taken as an indication of respiration activity (bacteria and other grazing) exceeding primary production in the entire water column including the photic zone. 44This may occur subsequent to the incidence of an algae bloom.Algae blooms have been frequently observed in the outer bay region as well as in the adjacent coastal region. 44DO and pH are correlated in the linear data fits (C1: DO = -17.98+ 3.15 pH, r = 0.603, p < 0.05; C2: DO = -27.21+ 3.78 pH, r = 0.807, p < 0.05) demonstrating dominance of biochemical process 45 as driver of DO and pH values from surface to bottom waters.Similar slopes but very different intercepts (C1 = -17.9and C2 = -27.2) also support that in C2 respiration exceeded primary production thereby generating a greater decrease in DO than in pH.In the internal areas of the bay DO usually reaches saturation > 100% in surface waters, while in many occasions the bottom is suboxic.The photic zone in the bay varies from 0.5 m in the inner areas to 2.5 m in the central channel mostly due to the autochthonous production of particulate matter. 29,46Kjerfve et al. 33 reported DO data obtained by the state environmental agency from 1980-1993 for a site in the central channel, which are similar (surface: 124%; bottom: 73%) to those found in C1.The lower DO levels in C2 compared to the 13 years average reveal occasions of eutrophic conditions aggravation in the outer region of the bay specially during the wet period.
The intensification of eutrophic conditions in areas of the bay with higher circulation is confirmed by the nutrient concentrations which were up to three times higher than those reported previously for the same area. 33,47This should be expected due to the combined effect of a submarine outfall installation in 2003 downstream of the sampling station and the population growth in recent years. 25NH 4 + was the most abundant nitrogen species, which showed strong seasonal variations with higher concentrations in C1 (p < 0.05).In both samplings concentrations of ammonium were more elevated at mid-depth during ebb tide proving that OM degradation in the water column just below the euphotic zone is very intense as observed by Wagener 28 and that the inner areas of the bay are important source of NH 4 + derived from sewage release.Nitrite showed low seasonal variation being more elevated in surface waters in C2, while for C1 higher nitrate concentration occurred during flood tide (Figure 4) and higher concentration of NH 4 + occurred in the ebb tide (Figure 4).Higher nitrate concentrations (3.7 µmol L -1 ) were reported in the past by Rebello et al. 29 for surface waters and by Kjerfve et al. 33 for bottom waters (average for 13 years: 4.5 ± 3.4 µmol L -1 ).Such change is an additional indicator of a tendency towards further decay of the system with increasing NH 4 + .P inorg predominated in C2 especially at the bottom and variations in both samplings occurred without tide dependence.Higher concentrations in C2 bottom waters reveal the input from the SACW but a contribution from the sediments cannot be discarded in both C1 and C2. 33,48he ratio dissolved inorganic nitrogen (DIN) / dissolved inorganic phosphorus (DIP) was calculated considering all DIN species and the results are given in Table 2.By using the Redfield ratio N / P (16:1) it is possible to estimate that P is the limiting nutrient in C1 and also that the remobilization of P from sediments is substantial as the sharp DIN / DIP decrease from surface to bottom waters, especially in C1, and the trend to higher concentrations in the bottom suggest.In C2 N / P values were often below 16 indicating a strong relative P consumption. 49PM was higher in C2 (p < 0.05) (see Table 1) possibly due to the intense continental runoff during the rainy season.Kalas et al. 50found in the sampling region values for SPM in the range reported here.
The intense primary production in the bay is represented by the high Chl-a concentrations during the samplings, with concentrations reaching more than 80 µg L -1 (Table 1).According to the Organisation for Economic Cooperation and Development (OECD), 51 Chl-a concentrations above 25 µg L -1 are typical of eutrophic systems.The presence of higher values in C1 may be ascribed to the lower SPM found in the dry season since, as observed by Rebello et al., 29 primary production in the bay is limited by the water transparency.Chl-c is also a significant component of the photosynthesized pigments and such observation concurs with the reported abundance of dinoflagellates in the bay. 52,53eterotrophic bacteria are important for the structure and dynamics of food chains, as well as for the biogeochemical cycles, being responsible for the inter-conversion of DOC and POC in the carbon cycle. 54Bacterioplankton  abundance was similar in C1 and C2 as indicated by the Kruskal-Wallis test (Table 1) although variations with tide were very intense as indicative of the heterogeneity of the system.Maximum abundance was present at mid-depth, below the photic zone, as expected due to the fast recycling of organic matter typically occurring in the bay down in the water column. 55In C2 bacterioplankton abundance was significantly correlated to NH 4 + (r = 0.764, p = 0.001) and to POC (r = 0.500, p = 0.001).A low correlation is observed with respect to d 13 C (r = 0.398, p = 0.01, s = 1.83).These associations during periods of water stratification when vertical transport is hindered as in C2 prove the intensity of bacterial activity at mid-depth resulting in residual organic material enriched in 13 C and liberation of NH 4 + .DOC was on average higher at the surface.The very elevated concentrations sporadically present (up to 400 mg C L -1 ) demonstrate the time variability of the system due to the influence of diffuse sources of sewage around the bay and to the sewage outfall located downstream of the sampling site (Figure 5).
Guenther et al. 44 reports DOC values for the outer bay area similar to those found in the present work except for the extremely high concentrations (overall DOC average and standard deviation after removing the extremes: 2.19 ± 0.48 mg L -1 , n = 34 out of 39 samples in C1; 2.39 ± 0.68 mg L -1 , n = 31 out of 39 samples in C2).Also, similar values are reported for the Pearl River in China. 56n polluted European estuaries, as the Sheldt estuary in Belgium and the Thames estuary in England, average concentrations are in the range of 5.8 to 6.8 mg C L -1 while in less altered estuaries, such as those of Gironde in France and of Douro in Portugal, DOC concentrations are from 2.5 to 3.1 mg C L -1 . 57he data for Guanabara Bay fits well in the latter range in spite of being highly polluted.This can be understood on the basis of the fast degradation of organic matter under high temperatures and microbial activity, and the intensity of water exchange.According to Kjerfve et al. 33 the time for renewal of 50% of the water volume (ca. 10 9 m 3 ) in Guanabara Bay is only 11.4 days and this would explain the better water conditions in the most water-circulated areas, such as those near the bay mouth where the sampling point is located.
POC was more elevated at the surface during daytime in both samplings but in C1 the variations with depth were more prominent (Figure 5).The overall average POC concentration was of 1.00 ± 0.50 mg L -1 in C1 and 1.01 ± 0.39 mg L -1 in C2.In both samplings, DOC correlates significantly with POC if extreme DOC values are removed.The linear correlation parameters in C1 are: s = 0.41, r = 0.680, p < 0.05; in C2: s = 0.79, r = 0.715, p < 0.05.
POC and PN are very strongly correlated even if the entire data collection is considered (r = 0.997; p < 0.05).A linear slope of 5.1 indicates predominance of autochthone organic matter in both campaigns. 50The C / N ratio of 4.4 (range 3.4 to 6.2; Table 2) indicates marine phytoplankton (C / N between 4 and 10) 51 and bacterioplankton (C / N ca.5) predominance. 58,59rbon and nitrogen stable isotopes The plot of d 13 C vs. d 15 N (Figure 6) highlights the different properties of the system in the two samplings  2).In C1 the average for d 13 C was -21.35 ± 1.75‰ with only two values appearing below -23‰.These isotopic shifts are typically reported for autochthone OM (-19 to -25‰), according to Meyers. 60The presence of sewage-derived organic carbon, whose d 13 C is approximately -24‰, 34 cannot be discarded; however, as observed by Carreira and Wagener, 61 under the high regional temperatures the turnover rate of sewage OM and release of nutrients occur largely before reaching the marine environment.Kalas et al. 50reported d 13 C of -21.30 to -15.10‰ for particulate organic matter in GB, similarly to the range found here in C1.In C2, the average for d 13 C shifted to -18.41 ± 1.11‰ and OM enriched in 13 C (d 13 C = -16.40‰)was present.Such enrichment may result from fast fixation rates and utilization of enriched dissolved inorganic carbon (DIC) produced during intense phytoplankton activity.Cifuentes et al. 62 reported for periods of elevated primary production and during algae blooms carbon isotopic shift in the range of -18 to -12‰.The presence of POC enriched in 13 C and other indications of phytoplankton predominance in C2 suggest utilization of dissolved carbon enriched in 13 C and seem to reinforce the possible occurrence of an algae bloom just prior to the sampling.The significant correlation between the POC, made up mainly of autochthonous OM, and d 13 C (r = 0.632; p < 0.05, s = 1.80)only in C2 may also be considered an indication of 13 C enrichment and primary fixation rates association.
C1 and C2 average values for d 15 N were 4.92 ± 1.13 and 8.79 ± 1.20‰, reaching a maximum of 7.11 and 11.08‰, respectively.In the marine environment, the predominant nitrogen incorporation mechanism includes nitrate reduction that generates nitrogen isotopic shifts between -2 and 4‰. 63According to Fogel and Cifuentes, 64 d 15 N between 7 and 11‰ occurs principally when nitrate is the limiting nutrient, since in periods of elevated primary production phytoplankton consumes the available nitrate leading to enrichment in the heavier nitrogen isotope in the dissolved form. 65In coastal areas the contribution of dissolved nitrogen derived from anthropogenic activities (d 15 N of 7 to 30‰) has strong influence on the d 15 N of the SPM.Also, sediment trap studies have shown that microbial degradation of phytoplankton leads to increase in 15 N in the residual OM. 66,67 The 15 N enriched OM in C2, as compared to C1, may result from the algae bloom occurring before the sampling day.The lower nitrate concentration and OM d 15 N close to 9-10‰ in C2 surface waters may derive, in addition, from nitrate consumption by dinoflagellates, the presence of which is confirmed by elevated Chl-c concentrations.Low values of d 15 N in the range of -3 to 1‰, typical of N 2 fixation, 64 were not observed here.This is expected due to the abundance of dissolved nitrogen species in both samplings.
A factor analysis (FA) with varimax rotation was carried out with and without data normalization.As described in Massone et al., 68 the standard normalization is intrinsic to principal components analysis (PCA) as long as it is based on correlation matrix.The variables were grouped in four factors totalizing of 73% of variance.Salinity and temperature were very distinct in the two samplings and therefore were not used in the FA to avoid obscuring the differentiation of chemical properties.In the first factor (F1, 32% of the variance) are all variables that discriminate the two samplings: P inorg , DO, SPM, d 13 C, d 15 N, NH 4 + and pH.Samples with higher concentration of P inorg , SPM, d 13 C, d 15 N and pH were positively correlated in F1, while those negatively correlated in F1 showed lower values of DO and NH 4 + .In the second factor (F2, 21% of variance) PN and POC are strongly correlated, which separates the samples according to the sampling depth.A third factor (F3, 12% of variance) includes DOC and POC and distinguishes the few samples appearing with unusually high concentration.The three chlorophylls are strongly correlated in F4 (8% of variance) and were not efficient in separating groups of samples, possibly due to the influence of depth and tidal variations.Figure 7 shows the graphical representation for the cases.

Sterols
Table 3 shows the results obtained for phytol and 15 sterols identified in the SPM.The average concentration of total sterols for C1 was 2,728 ± 2,058 ng L -1 , which was slightly higher than the average value measured in C2, of 2,381 ± 1,006 ng L -1 .Only 4 out of the 15 identified compounds made up the majority of the total sterols found in the present study.Cholest-5-en-3β-ol (cholesterol or 27∆ 5 ) was one of the most abundant sterols, representing, on average, 20.8 ± 24.2 and 15.5 ± 18.8% of the total sterols in C1 and C2, respectively.There was no clear evidence of 27∆ 5 concentration gradient with depth: the highest average value was at the mid-depth in C1 (800.5 ± 613.4 ng L -1 ) and in bottom water in C2 (408.3 ± 244.9 ng L -1 ).27∆ 5 has been usually observed in natural waters at high concentrations due to its widespread occurrence in planktonic organisms and other animals. 69It is also found in raw sewage, 70 and thus in contaminated coastal waters, such as in GB, 26 the input of domestic effluents can be a relevant additional source of this sterol.

Sterols as source assignments of OM
Sterols and phytol were used as a complimentary tool, in addition to bulk properties and isotopic composition of POC (as previously discussed), for source assignments of OM in the studied site.As these compounds have multiple and, in some cases, non-specific sources 80 the sources assignment of OM using sterols was attained after a specific FA.The factor analysis applied to the entire sterol data set after normalization was not efficient in discriminating sampling campaigns and samples.Normalized data from C1 and C2 were then analyzed separately.
It is interesting to note in Figure 8 that in spite of 27∆ 5 being positioned distant from the other most abundant phytosterols, it appears always correlated with 27∆ 5,22 and 28∆ 5 .Such grouping is not easy to explain; it might well reveal the occurrence of grazing, as 27∆ 5 is more abundant in zooplankton and/or fecal pellets. 79Anyway, there are few samples (scores represented in Figures 8c and 8d) associated with this group.The grouping of 29∆ 5 with the phytosterols in both samplings suggest an autochthonous origin for this sterol in the present study.
Whereas the FA aided to evaluating the origin of some sterols, the most relevant outcome of the analysis was obtained from the scores (Figures 8a and 8b).Most of the surface and mid-depth samples are positioned in the same quadrant as the main phytosterols (excluding 27∆ 5 ).Although observed for some C1 samples (Figure 8c), this feature is strikingly evident for C2 surface samples (Figure 8d) and is in agreement with a post-bloom scenario in C2 as indicated by the physicochemical and isotopic data.
Finally, another important outcome of the FA scores is the position of the bottom samples, which are in most of the cases in the quadrant where coprostanol dominates.This suggests the presence of sewage particles.The ratio 5β / (5β + 5α) sterol proposed by Grimalt et al. 81 was calculated in order to check out the sewage contamination using coprostanol.The threshold value of 0.6 proposed by Carreira et al. 26 was considered for indicating sewage contamination.Figure 9 shows that in almost all bottom samples the ratio is > 0.6.The higher incidence of sewagederived OM in bottom samples can be explained by the nearby discharge of the sewage outfall and the presence of settling aggregated particles; the sample highlighted with a red circle in Figure 9 showed the highest coprostanol concentration and the highest 5β / (5β + 5α) ratio.

Flow rates
Flow rates were calculated based on water mass transport data obtained from the current intensity and where: T = transport, M = mean concentration, s-m = integration of surface-mid-depth data, m-f = integration of mid-depth-bottom data, and i = sampling.The flow rates calculated on a daily basis are given in Table 4 and indicate net export of materials to the inner shelf in both samplings, except for DOC in C2.Exports of DIN, Chl-b and Chl-c were of the same order in both campaigns; those of DIP, Chl-a and sterols predominated in C2 while SPM, DOC, POC and PN were more important in C1.DOC importation in C2 may derive from the inflow during high tide of organic carbon associated with the suggested algae bloom event prior to the sampling day and also from influence of the sewage outfall located closer to the bay mouth with respect to the sampling station.The widespread delivery of untreated sewage into the bay causes variability in chemical properties of the system and difficulties of source identification.
As the two samplings were performed in distinct conditions typically found during the dry and wet period,  obtained data on daily basis can be used for estimating the magnitude of the exchanges on an annual basis as necessary for comparison with literature data.The export rates to the inner continental shelf so far estimated are: 1.27 × 10 4 kmol year -1 DIN; 9.52 × 10 2 kmol year -1 DIP; 2.65 × 10 10 g year -1 DOC; 1.96 × 10 10 g year -1 POC; and 2.96 × 10 10 g year -1 PN.
Souza et al. 83 report the following integrated flow rates for 24 rivers in the northeast and southeast coast of Brazil: average DIN ca.3.0 × 10 2 kmol day -1 and DIP = 4.4 and 37 kmol day -1 in the months of September and December 2000, respectively.These flow rates are approximate since they were calculated based on published data (National Water Agency).Figueiredo et al. 84 evaluated the outputs of C and N from the Paraiba do Sul River (hydrographic basin: 22,000 km 2 , average drainage: 900 m 3 s -1 ) to the Atlantic Ocean and found 48 × 10 9 g year -1 particulate carbon (PC), 83 × 10 9 g year -1 DOC and 8 × 10 8 g year -1 PN.The data of the present work reveal that on a regional scale GB contributes with about 10% of the dissolved N and P that enters into the continental shelf through river inputs of the Northeast and Southeast Brazil.If compared to the Paraiba do Sul River estimates, GB contributes with an export of DOC of the same order of magnitude and with 24 times higher PN output.
The international program Land-Ocean Interactions in the Coastal Zone (LOICZ) produced numerous studies targeting nutrient flow rates to the oceans in coastal areas.Table 5 shows the results for nutrient flow rates from several systems in the Baltic Sea. 85pra 86 estimated flow rates in the range of 10 2 to 10 6 kmol year -1 DIN and 10 2 to 10 5 kmol year -1 DIP for 30 ecosystems in the Southeast Asia region.Wepener, 87 by using the LOICZ model, found the following average flow rates for the Nhlabane, Thukela and Mvote Estuaries, respectively: 0.076, 15, and 0.634 kmol day -1 for DIP and 1.7, 399, and 0.015 kmol day -1 for DIN.Arndt et al., 88 using mechanistic model, estimated for the Scheldt Estuary, Belgium, a DIN flow rate of the order of 10 5 kmol year -1 .The GB exports of DIN and DIP are of the order of one thousandth of the outputs into the Baltic Sea, and are in the center of the flow rate range reported for the South Asia Sea.He et al. 6 estimated a mangrove-derived POC flux of 5.3 × 10 5 to 1.0 × 10 6 kg year -1 POC from the Shark River into the Gulf of Mexico.
The estimated global nutrient loads from rivers are of 0.32-0.64Tmol N year -1 for DIN and 13-27 Gmol P year -1 for DIP.The contribution of GB on a global scale may be estimated using these flow rates and turns out to be of 1.2 × 10 -2 % for organic carbon, 2.6 × 10 -3 % DIN and 4.7 × 10 -3 % DIP.

Conclusions
The present results show that GB comprises one of the most relevant systems in the southeastern Brazilian coast as  DIN: dissolved inorganic nitrogen; DIP: dissolved inorganic phosphorus.
far as the export of organic carbon and nutrient species is concerned.The comparison of the obtained flow rates with regional and global data demonstrates the magnitude of potential impact on the biogeochemistry of the inner shelf.The major source of nutrient and productivity enrichment is sewage input derived from densely populated municipalities in the hydrographic basin.The use of sterols and stable isotopes as markers of organic matter origin successfully indicated the system variability due to interaction with ocean waters, seasonality and tidal oscillations.The study shows that anthropogenic inputs to the bay are rapidly mineralized and used to produce autochthonous material, which is then exported to the inner shelf.Such POC, differently from the natural terrestrial organic carbon, is of high nutritional value and thus may contribute significantly to the ecology of the pelagic and benthic secondary producers in the shelf.Moreover, undesirable effects, such as red tides occurring in the open coastal area off Guanabara Bay and appearance of large bacteria community may result in indirect effects upon the health of the coastal oceans.
The discharge of raw or insufficiently treated sewage directly into estuaries is not unique of Rio de Janeiro but often occurs in developing as well as in developed countries.Therefore, the global impact of these sources on the carbon and nutrient cycles must be better quantified and understood.

Figure 1 .
Figure 1.Map of Guanabara Bay showing the sampling station.

Figure 3 .
Figure 3. Current intensity and direction in C1 and C2 samplings.

Figure 5 .
Figure 5. DOC and POC variation during 25 h tide in C1 and C2.

Figure 8 .
Figure 8. Factor analysis including sterols in surface (S, in red), mid-depth (M, in blue) and bottom (F, in black) samples; (a) and (c) stand for C1 and (b) and (d) stand for C2 sampling.

Figure 9 .
Figure 9. 5β / (5β + 5α) sterol ratio as a function of coprostanol concentration.The broken line highlights the threshold for indication of sewage contamination.

Table 2 .
Ranges and medians for DIN / DIP ratio, C / N ratio, d 13 C VPDB (‰) and d 15 N air (‰) at the surface, mid-water and bottom in C1 and C2

Table 3 .
Means, SD and range for the 15 determined sterols in C1 and C2 samples 82rection measurements.Physical data were integrated for the surface-mid-depth and mid-depth-bottom sectors and then multiplied by the mean concentration of each chemical variable in the same sectors, as follows:82

Table 5 .
Flow rates for DIN and DIP of some ecosystems of the Baltic Sea