The Minho and Lima estuaries, both situated in the northern part of Portugal (Fig. 1), differ essentially in terms of river discharge, anthropogenic pressures and morphology (Table 1).

Water sampling was undertaken in September 2006 during ebb tide (at spring tide), at nine stations located along a main transect of both estuaries, from the upstream limit of tidal influence to the estuary mouth, covering a full range of salinity of 0-30 corresponding to a distance of 26.5 km in Minho and 15 km in Lima estuary (Fig. 1).

Surface water (0.2 m depth) was collected using 2-L Niskin bottles (General Oceanics) for analysis of salinity (S), temperature (T), pH, dissolved inorganic nitrogen (nitrate NO₃^–, nitrite NO₂^– and ammonium NH₄⁺), dissolved oxygen (DO) and nitrous oxide (N₂O). The hydrological characteristics of the Minho and Lima estuaries and the meteorological conditions observed during the sampling period are presented in Table 2.

Water temperature (T) was measured in situ with a Seabird SBE19/CTD probe with an accuracy of 0.01°C.

Salinity (S) measurements were carried out using a temperature-controlled Guideline Salinometer (Portasal 8410A), and accuracy was 0.03 salinity. Equipment was calibrated with a certified IAPSO Standard Seawater reference.

Meteorological parameters (air temperature, pressure, and wind speed and direction) were determined using a portable meteorological station (Campbell Scientific CR510). Measurements represent the average of physical parameters taken using a sampling time of 5 seconds and a storage time of 1 minute. Wind speed 10-minute average was determined for each sampling station and converted to wind speed values at 10 m height (u₁₀) using a logarithmic correction (Hartman and Hammond 1985).

The Minho and Lima River discharges were calculated as an average of the flow 10 days before sampling, at the hydrometric stations of Foz de Mouro and Ponte da Barca, respectively (SNIRH 2013).

Dissolved oxygen (DO, μmol L^–1) was measured using whole-bottle Winkler’s titration method (Aminot and Chaussepied 1983). A Methrom titrator was used to dispense small amounts of thiosulphate, and starch endpoint was detected visually. Precision of the method was in the range of 0.08% to 0.25%. DO saturation, expressed in percentage (%), was determined as the ratio of the oxygen concentration determined and the equilibrium values of DO calculated with the Weiss (1970) equation. Apparent oxygen utilization (AOU, μmol L^–1) was calculated as the difference between the saturation oxygen concentration and the dissolved oxygen concentration measured in the sample.

Water samples for determination of dissolved inorganic nitrogen were filtered through acetate cellulose filters (pore size 0.45 μm) and stored at –20ºC until analysis. Analyses were carried out using a Traacs autoanalyser following colorimetric techniques outlined by the manufacturer. Estimated precision was ±0.8% for nitrate and nitrite (NO₃^– and NO₂^–) and ±2.0% for ammonium (NH₄⁺), at mid-scale concentrations. Accuracy of nutrient measurements was maintained by using CSK Standards (WAKO, Japan).

pH measurement was carried out immediately after collecting water samples using a Metrohm 704 pH-meter and a combined electrode (Metrohm), standardized against NBS buffers (6.865 and 9.180 pH). Precision of pH measurements was ±0.01.

Water samples for determination of N₂O were collected in triplicate in 20-mL glass headspace vials and poisoned with saturated aqueous mercury chloride (HgCl₂) to stop biological activity. The vials were stored upside down, in the dark, at 4°C in the refrigerator until analysis, performed within 10 days. Dissolved N₂O was determined by a headspace equilibration technique coupled with gas chromatographic analysis (GC-3800, Varian). Briefly, 20 mL of sample was equilibrated with 5 mL of highly purified helium (purity =99.9999%) in a headspace CombiPAL autosampler. Gas chromatographic separation was carried out using a stainless steel column packed with 80/100 (mesh) Porapak. Oven and detector temperature was set at 50°C and 320°C, respectively, and high purity nitrogen (99.9999%) was used as the carrier gas (flow rate 30 mL min^–1). To remove water vapour and carbon dioxide, absorbent columns packed respectively with Mg(ClO₄)₂ and Carbosorb were located in the carrier gas line between the sample loop and the separation column. N₂O peak was detected with a ⁶³Ni electron capture detector (ECD). Calibration of ECD response was performed using standard gas mixtures with 400, 780 and 1980 ppb N₂O in synthetic air (Air Liquide), and method precision was 2.6% (30 replicate measurements using samples containing 10 nmol L^–1 of N₂O). In situ concentration of N₂O (C, nmol L^–1) was calculated from the concentrations measured in the headspace according to the solubility equation of Weiss and Price (1980):

C = β (TS) x′P

where x′ is the measured N₂O dry mole fraction, P is the atmospheric pressure, and β is the solubility coefficient, which is a function of the water temperature (T) and salinity (S). N₂O equilibrium concentrations were calculated assuming an atmospheric N₂O mixing ratio of 320.1 ppb (WMO 2006).

N₂O saturation, expressed in percentage (%), was determined as the ratio between the measured dissolved N₂O concentration and the equilibrium concentration. The N₂O water-air flux (F_N2O) was estimated according to the following equation:

F_N2O = k_N2O ΔN₂O

where ΔN₂O, the excess of N₂O, is the difference between the measured concentration and the equilibrium concentration with the atmosphere in the estuarine water at the local temperature and salinity; and k_N2O (cm h^–1) is the N₂O transfer velocity, which is expressed as a function of the wind speed and the Schmidt number (Sc). Since no direct measurements of k_N2O were made in the Minho and Lima estuaries, both the k-wind relationships of Carini et al. (1996) (hereinafter referred to as C96) and Raymond and Cole (2001) (hereinafter referred to as RC01) were, respectively, used to compute k:

k_C96 = 0.045 + 2.0277u₁₀

k_RC01 = 1.91 e^0.35u10

The gas transfer velocities and air-sea fluxes were estimated using in situ wind speeds normalized to 10 m height (u₁₀). The k coefficients were corrected for in situ temperature using the following relationship:

k_N2O /k₆₀₀ = (Sc_N2O /600)^–0.5

where Sc_N2O is the Schmidt number for N₂O calculated according to the equation of Wanninkhof (1992):

Sc = 2301.1 – 151.1 T + 4.7364 T² + 0.057431 T³

where T is the temperature (ºC).

An unpaired t-test was used to identify statistical differences in levels of N₂O and other environmental variables between and along estuaries.

Pearson’s correlation analyses were performed to evaluate the existence of relationships between ΔN₂O and the variables NH₄⁺, AOU, NO₂^– and pH assumed to be connected with production pathways of this biogas.

The multivariate techniques principal component analysis (PCA) and cluster analysis were applied to environmental data in order to identify and compare inter-relationships between these variables in both estuaries. Data were log(x+1) transformed and handled using correlation-based PCA on the basis of standard Euclidean distance between samples to define their dissimilarity. PRIMER (version 6) was employed for the multivariate analysis.

Figures were created in Golden Software Grapher program (version 9.6.1001).

Concentrations of N₂O and the studied environmental parameters plotted against salinity along both Minho and Lima estuaries are shown in Figure 2.

Distribution of N₂O exhibits a declining tendency towards the mouths of both estuaries. In general, concentrations were higher in Lima than in Minho (Fig. 2A). In Lima values ranged from 10.1 to 20.0 nmol L^–1 and were below the conservative mixing line. The maximum value was reached in the upper estuary, suggesting that the Lima River was the main source of N₂O to the estuarine system. From salinity 0 to 4 a sharp decrease was observed, suggesting a greater N₂O loss in the zone, presumably through water-air gas exchange. Downstream of salinity 4, N₂O decreased more slightly and the mixing with the N₂O-poorer seawater is well perceptible. In the Minho estuary, concentrations varied from 8.6 to 14.4 nmol L^–1 and by contrast with those observed in Lima were above the conservative mixing line, pointing to the existence of N₂O sources within the estuary. A major concentration increase from 10.5 to 14.4 nmol L^–1 was detected between 0 and 2.5 salinity, suggesting the existence of internal N₂O sources in this zone of the estuary, presumably from manufacturing industries located nearby. N₂O saturation values ranged from 101% to 166% along the Minho estuary and from 113% to 227% along the Lima estuary (see Table 4), indicating that both estuaries are potential N₂O sources to the atmosphere.

Surface waters of both estuaries were well oxygenated during our study period. Concentrations of DO were higher than 280 µmol L^–1 in the upper Lima estuary at salinity 0.2, decreasing to 240 µmol L^–1 at salinity 2.8 (Fig. 2C). Seawards of this salinity, an increasing tendency was observed and a concentration of 255 µmol L^–1 was also reached in the vicinity of the estuary mouth (salinity 30.2). In the Minho estuary, a decrease in DO was also detected in the upper estuary and the concentration dropped from 243 µmol L^–1 at 0.1 salinity to 199 µmol L^–1 at 2.3 salinity. Afterwards, a sharp increase in DO was measured along the estuary and a maximum value of 264 µmol L^–1 was reached at the estuary mouth (salinity 29.7). DO-enriched seawater probably accounted for the similar increasing trend seaward in both estuaries. Saturation values were higher than 70% in the Minho estuary and 90% in the Lima estuary.

In the Minho estuary pH increased from 7.3 in the more river-influenced zone to 8.1 at the estuary mouth (Fig. 2D). In the Lima estuary pH showed a larger range of values, increasing from 6.7 in the most river-influenced site to 8.0 at the estuary mouth.

NO₃^– was the dominant species of inorganic nitrogen in both estuaries, reaching a similar maximum concentration in the river input (47.6 µmol L^–1 in Minho and 44.4 µmol L^–1 in Lima) (Fig. 2E). Values decreased seawards and in general followed the theoretical conservative mixing line. Along the Minho estuary both NO₂^– and NH₄⁺ exhibited an irregular behaviour, though the system seemed to function as an NO₂^– source and an NH₄⁺ sink (Fig. 2F, G). Between salinity 0 and ~5-7 the decline in NH₄⁺ (~4.0 to 0.5 µM L^–1) was simultaneous with an increase in NO₂^– (~0.6 to 1.0 µM L^–1) and N₂O (~10 to 13-14 nmol L^–1), suggesting the occurrence of nitrification.

Figure 3 displays relationships between ΔN₂O and AOU, NH₄⁺, NO₂^–, NO₃^– and pH in the Minho estuary. In the mentioned salinity zone (0 and ~5-7) a significant positive correlation was found between ΔN₂O and AOU (R²=0.75). This indicates the occurrence of nitrification as a source of N₂O and the respective slope provides an estimate of the biological N₂O yield per mole O₂ consumed (Yoshinari 1976). Further, the simultaneous (negative) correlations between ΔN₂O and the primary substrate for nitrification, NH₄⁺ (R²=0.40), and pH (R²=0.80), and the positive correlation between ΔN₂O and the byproduct of nitrification, NO₂^– (R²=0.41), are consistent with the predominance of nitrification as a mechanism of N₂O production in the upper part of the Minho estuary. No correlation was found with NO₃^–_, whose high concentrations were mostly riverine derived.

In the Lima estuary no relationships were found, suggesting the occurrence of nitrification (Fig. 4).

The application of PCA to the studied environmental variables in the Minho and Lima estuaries allowed us to identify two main composite variables, PC1 and PC2 (eigenvalues >1.0), which explain 82% of the variance (Table 3) and represent a good description of the environmental structure across the estuarine sampled sites.

PC1 explained 53% of variance and had the highest positive loading for NO₃^–, T and N₂O and a negative loading for S and pH (Fig. 5). This component represents the separation of major river-influenced stations from major marine-influenced ones in both estuaries. PC2 explained 27% of the variance and correlated positively with DO and negatively with NO₂^– and NH₄⁺. This component appears to represent relevant parameters to N dynamics, particularly in Minho estuary. In fact, projection of stations along PC2 reveals a clear separation of sites from the upper Lima estuary (L1-L5) (Fig. 5), mostly associated with higher values of N₂O, and sites from the upper Minho (M1-M5) more associated with higher NO₂^– and NH₄⁺, apparently from river origin, as the N load from the Minho River is considerable at this point (Table 1). It was also observed that the stations from both middle/lower estuaries did not differ in terms of studied environmental variables.

N₂O water-air fluxes are shown in Figure 6. Positive values prevailed at all stations but decreased, in general, between the upper and lower zone of the estuaries. This tendency was more pronounced in the Lima estuary, where higher N₂O fluxes in the river-influenced area were about twice (~12.0 μmol m^–2 d^–1; C96) (Fig. 6B) those observed in the upper part of the Minho estuary (~6.0 μmol m^–2 d^–1; C96) (Fig. 6A). The higher fluxes in the Lima estuary were mainly associated with the higher levels of N₂O observed in the upper estuary area (St.1 to St.3; Fig. 2), indicating that the low salinity zone (0-5) is an important source of N₂O to the atmosphere.

Using the two different parameterizations (C96, RC01) to calculate the gas transfer coefficients, the averaged N₂O water-air fluxes from Minho estuary to the atmosphere ranged between 4.0±3.3 µmol m^–2 d^–1 (RC01) and 4.1±2.8 µmol m^–2 d^–1 (C96), corresponding to a mean N₂O concentration of 11.3±1.3 nmol L^–1 (132±22% saturation) and a mean wind speed of 3.9±0.1 m s^–1. Slightly higher N₂O water-air fluxes were found in the Lima estuary, with averaged values ranging between 4.7±1.9 µmol m^–2 d^–1 (RC01) and 5.0±2.0 µmol m^–2 d^–1 (C96), corresponding to a higher mean N₂O concentration of 13.7±1.6 nmol L^–1 (153±26% saturation) and a lower mean wind speed level (2.4±0.1 m s^–1).

N₂O fluxes from the Minho and Lima estuaries were regressed versus the first two PCs’ ordination of station scores to test their ability to predict the fluxes. We found out that only PC1 showed to be correlated with N₂O flux, with a strong positive correlation (R²=0.61) (Fig. 7). N₂O flux also increased along a gradient of increasing NO₃^–, T and N₂O. These results suggest that future global changes in these parameters will result in an increase of N₂O flux in these estuarine systems.