To ensure consistency across all sites, identical sampling materials and detailed protocols for N₂O and CH₄were prepared and distributed to each collaborating laboratory. Before the start of the sampling cruises, dedicated training sessions on trace gas sampling were provided to all teams, aiming to harmonize procedures and minimize potential biases between sites and laboratories. Duplicate samples for N₂O and CH₄ were collected in 120 mL borosilicate serum vials, sealed with grey chlorobutyl septa and aluminium crimps. To preserve the samples, approximately 200 µL of saturated HgCl₂ solution was added and the vials were stored upside down in the dark until analysis at the AQUANITROMET laboratory (AQUANITROMET service: http://hdl.handle.net/10261/423686, last access: 1 July 2026) of the Instituto de Investigaciones Marinas (IIM-CSIC, Vigo, Spain).

The dissolved concentrations of N₂O and CH₄ were analysed using the static-headspace equilibration technique combined with gas chromatography following de la Paz et al. (2015, 2021). It consists of creating a headspace by automatically introducing a volume (20 mL) of ultrapure N₂ gas into the sample vial using a high-precision automatic burette displacing the same volume out of the vial by piercing the septum with two needles. After reaching equilibrium overnight in a temperature-controlled environment, 18 mL of the headspace is automatically extracted using a high-precision burette and injected into a gas chromatograph (Agilent GC 7890-A), where the signal for each gas is separated using two independent Porapak Q-packed columns. Finally, N₂O is detected with an electron capture detector and CH₄ with a flame ionization detector. For the calibration curve, we used triplicate injections of three combined gas standard mixtures of N₂O and CH₄ in N₂, supplied by NOAA (National Oceanic and Atmospheric Administration) (N₂O : CH₄ = 332 : 1959 ppb) and by AirLiquide France (ratios N₂O : CH₄ = 1000 : 3000 and 3100 : 5000 ppb). The precision estimated from the averaged coefficient of variation of duplicates (n=123) for this study was 1 % for N₂O and 4.5 % for CH₄. Instrumental limits of detection (LOD) and quantification (LOQ) for GC were defined as 3 and 10 times the standard deviation of triplicate injections of the lowest standard (NOAA atmospheric level), divided by the calibration slope, respectively. This yielded LOD ranges of 3–10 ppbv for CH₄ and 0.1–2.9 ppbv for N₂O, and LOQ ranges of 9.8–32.4 ppbv for CH₄ and 0.3–9.7 ppbv for N₂O. Because dissolved gas analysis is fundamentally constrained by detector sensitivity, conservative methodological limits were calculated by substituting the maximum gas-phase LOD and LOQ values into Eq. (1) for seawater at S=35 and T=20 °C. The resulting aqueous LODs were 0.11 nmol L⁻¹ (CH₄) and 0.16 nmol L⁻¹ (N₂O), with corresponding LOQs of 0.35 nmol L⁻¹ (CH₄) and 0.55 nmol L⁻¹ (N₂O). This analytical system was intercalibrated with other laboratories, with very good results, as part of activities during the first large international experiment to compare marine N₂O and CH₄ measurements, organized by the Scientific Committee on Oceanographic Research (Wilson et al., 2018).

The gas concentration using the headspace equilibration was computed following Wilson et al. (2018): 1Cmeas[nmolL-1]=βxPVwp+xPRTVhs/Vwp where β denotes the Bunsen solubility coefficient of N₂O (Weiss and Price, 1980) and CH₄ (Wiesenburg and Guinasso, 1979) expressed in nmol L⁻¹ atm⁻¹; x is the dry gas mole fraction (ppb) determined in the headspace; P is the ambient atmospheric pressure (atm); Vwp and Vhs are the volumes of the water sample and equilibrated headspace, respectively (mL); R is the universal gas constant (0.08205746 L atm K⁻¹ mol⁻¹); and T is the equilibration temperature (K). The CH₄ and N₂O concentrations are also expressed as percent saturation (N₂OSat and CH₄Sat in %), calculated as the ratio between the measured concentration (Cmeas) and the equilibrium concentration in seawater (Cequi) for each gas. The equilibrium concentrations were computed by multiplying Bunsen solubility coefficient (β) of each gas by the monthly average atmospheric molar fraction of N₂O and CH₄ measured at the Lampedusa monitoring station – the only site in the Mediterranean Sea with continuous records of these gases – obtained from the NOAA monitoring network (Lan et al., 2026a, b).

Due to the lack of standardised QC tools or reference materials in aqueous phase for N₂O and CH₄, such as those available for the carbonate system (CRMs produced by A. Dickson at Scripps USA), our QC procedure relied on duplicates seawater samples and reported results as the coefficient of variation (CV, %). The presence of outliers was established based on the CV value between duplicates; hence, samples were discarded with a CV higher than 4.5 % (n=3) for CH₄ and 1 % for N₂O (n=1). In addition, some glass flasks were broken during transport to Vigo (n=14), reducing the total number of duplicates samples.

The air-sea fluxes (F) were calculated using the following equation: 2F=kw(Cmeas-Cequi) where k is the gas transfer velocity (cm h⁻¹) and (Cmeas–Cequi) is the air-sea concentration gradient of N₂O and CH₄, being Cmeas and Cequil the measured and equilibrium gas concentrations defined in the previous section. By convention, and following the sign of the gas concentration gradient, negative values of F correspond to a transfer of N₂O or CH₄ from the atmosphere to the ocean (i.e., a sink for the atmosphere) and positive values of F correspond to a transfer of those gases from the ocean to the atmosphere (i.e., a source for the atmosphere). There is a long-term debate over which is the best expression for gas transfer rate k, but here we adopted as the expression given by Wanninkhof (2014), since it is widely used in the last RECCAP2 analysis for GHG emissions in the coastal ocean (Resplandy et al., 2024). The gas transfer velocity by Wanninkhof (2014), gave the expression kw=0.251〈U2〉 (Sc/660)^−0.5, where 〈U2〉 is the average of the square wind speed and Sc is the Schmidt number (dimensionless). For our study, due to lack of local sources for wind data on land or buoy for some of the stations, we used the monthly averaged of the square of 6 hourly wind data provided by the Cross-Calibrated Multi-Platform Ocean Surface Wind Vector 3.0 (CCMP) wind data product that combines satellite and buoy data with model predictions (Mears et al., 2022). This wind speed product is the optimum choice using the k proposed by Wanninkhof (2014), which is computed using this same data product, and has been demonstrated to be an optimum wind source for coastal regions (Otero et al., 2013). The Sc, dependent on the temperature, salinity, and specific gas molecule, is calculated according to the coefficients for N₂O and CH₄ and the equation reported by Wanninkhof (2014). Then, the Schmidt number for the in situ salinity at the Mediterranean station was calculated by interpolating Sc for fresh water and seawater.

To determine the annual average air-sea fluxes of N₂O and CH₄ at each station, calculations were performed over a 12-month period. Missing monthly values of N₂O, CH₄, temperature, and salinity during this timeframe were reconstructed through temporal linear interpolation based on the closest available cruise measurements.

To determine whether the differences in N₂O and CH₄ concentrations and other variables across stations were statistically significant, we used Tukey's test with a significance level of p<0.05. Additionally, we compute the Spearman correlation coefficients between surface water CH₄ and N₂O concentrations and potential environmental factors, such as water temperature, salinity, dissolved oxygen, inorganic nutrients, and Chl a. All analyses were done using the R software.

Overall, the seasonal variability of sea surface temperature (SST) was very similar in every region, ranging from 14.5 °C in BAL-PB (March 2024) to 31.2 °C in LEV-BEY (August 2023) (Fig. 2). ALB-M'DIQ exhibits unique characteristics, particularly the unexpectedly low temperatures observed during the summer. These anomalies are likely linked to the eddy-driven vertical motion in the Alboran sub-basin. This connection is supported by several studies that have documented mesoscale eddies with intensified vertical motion and subsurface anticyclonic eddies along the African coast (Tintoré et al., 1991; Viúdez et al., 1998). Hence, except for ALB-M'diq, the maximum (August) and minimum surface temperatures (from December to March) showed an overall synchronised cycle between the stations, with lower seasonal SST amplitudes in the southern coast of the Alboran sub-basin (9.5 °C ALB-REM) to higher seasonal SST amplitudes in the stations of the Balearic Sea (13.3 °C in BAL-PB).

The sea surface salinity ranged from 36.6 in BAL-CS in June 2024 to 39.5 in LEV-ETS (Fig. 2b). Compared to the other regions, the seasonal cycle of salinity was clearly observed in the surface waters of Lebanon, where both stations LEV-B1 and LEV-BEY showed higher salinity values from July to November, owing to intense evaporation and lower values from January to April owing to precipitation/continental runoff (Abboud-Abi Saab and Hassoun, 2017).

Seasonal variability in N₂O concentrations was observed at all stations, with concentrations increasing during winter and reaching minimum values in summer (Fig. 2, Table 2). The highest N₂O concentration of 9.95 nmol kg⁻¹ was measured in BAL-PB in January 2024, whereas the minimum N₂O concentration of 6.3 nmol kg⁻¹ corresponded to LEB-B1 in August of 2023. Monthly N₂O concentrations showed a clear intra-annual cycle, with amplitudes ranging from 1.3 nmol kg⁻¹ in ALB-M'DIQ to 3.1 nmol kg⁻¹ in BAL-PB Bay, which were directly correlated with SST amplitude. Despite some inter‐station differences, the timing of the peaks and troughs was broadly consistent across the Mediterranean coast.

The CH₄ concentrations in surface waters ranged from 2.9 nmol kg⁻¹ measured in March 2024 in BAL-CA to 38.2 nmol kg⁻¹ measured in September 2023 in LEV-ETS (Table 2). The distribution of CH₄ concentrations were very skewed, and the minimum CH₄ was comparable between the stations (60 % of values were lower than 5 nmol kg⁻¹), and CH₄ peaks were evenly distributed and consistently high along the coast of the Levantine sub-basin. Only BAL-CS showed a moderate seasonal trend, with a gradual increase in CH₄ concentrations in June, peaking in July, and gradually decreasing through September 2023. Apart from this, no clear seasonal pattern was observed for CH₄ concentrations.

The saturation status of N₂O in surface water for most of the period was oversaturated (Fig. 2, Table 2), except in May 2023 at the stations of the Balearic sub-basin (BAL-CS, BAL-CA) and in ALB-M'DIQ in October 2023 and May 2024 with a moderate seasonal trend of lower N₂OSat in winter and spring, and higher oversaturation in summer months, with the N₂OSat ranging from 96±0 % in BAL-CS in May 2023 and 127±1 % in LEV-ETS in September 2023. The surface water at all the stations was supersaturated in CH₄ (Fig. 2, Table 2), with minimum values during the winter and early spring (110±4 % in BAL-CA in March 2024) and higher values in summer (up to 400 %), with very high CH₄Sat peaks in the eastern Basin of the Levantine sub-basin that reached the 1973±4 %.

Sea Surface Temperature distribution was relatively homogeneous within each subbasin (Fig. 3) (Tukey test; p>0.05). However, a significant longitudinal gradient was observed across the Mediterranean Sea. The mean SST increased from 17.7±1.8 °C at the westernmost station (ALB-M'diq, Alboran Sea) to 25.6±4.6 °C at the easternmost station (LEV-ETS, Levantine Sea). Consequently, SST values were significantly higher in the Levantine Sea (B1, BEY, and ETS) compared to those in the Alboran and Balearic sub-basins (M'Diq, REM, and CA; p<0.05), while no significant differences were found between the latter two subbasins. A similar zonal trend was found for salinity, with mean values increasing from 37.18±0.12 in Alboran coastal waters to 39.11±0.46 in the Levantine sub-basin. This reflects a significant salinification in the Eastern basin (ETS, B1, and BEY) compared to the western Moroccan coast (M'diq, REM) and the Balearic sub-basin (CS, CA, PB), with a maximum inter-basin gradient of 1.9.

The regional distribution of CH₄ across the Mediterranean Sea exhibited a marked longitudinal contrast (Fig. 3). Although temporal variability was systematically higher in the eastern stations, significantly elevated mean CH₄ concentrations were only observed at LEV-ETS (13.6±9.8 nmol kg⁻¹; Tukey's test, p<0.05). In the remaining stations, average concentrations ranged from 3.8±0.8 nmol kg⁻¹ at BAL-CA to 8.1±5.2 nmol kg⁻¹ at LEV-BEY. Despite the occurrence of high episodic values at LEV-B1 and LEV-BEY, their median concentrations did not differ significantly from those in the Alboran and Balearic sub-basins (Tukey's test, p>0.05). Surface waters throughout the Mediterranean Sea were persistently supersaturated in CH₄, with mean saturation levels ranging from 168±46 % (BAL-CA) to 659±500 % (LEV-ETS), following the spatial pattern of absolute concentrations. The CH₄Sat oversaturation peaks along with their associated variability observed at the LEV-ETS station (up to 659±500 %) suggest the influence of localised seafloor sources, such as cold seeps or shallow gas venting, which are well documented in the Levantine sub-basin (Coleman and Ballard, 2001). Alternatively, nearshore/river filaments or submarine groundwater discharge (SGD), which can be brackish to saline and may not show as a salinity minimum, can significantly raise CH₄, SST, and salinity remain mostly unchanged. In this region, the presence of coastal groundwater systems and SGD is supported by hydrogeochemical/isotopic evidence for the Mersin–Erdemli and Lamas areas (Kuyumcu, 2023). Both scenarios, seafloor sources and SGD, can likely contribute to the episodic enrichment of surface waters in this region, however further research is needed to confirm them.

In contrast, N₂O concentrations showed much lower spatial variability, with station averages ranging from 7.2±0.7 nmol kg⁻¹ (LEV-ETS) to 8.5±0.3 nmol kg⁻¹ (ALB-M'DIQ) (Table 2, Fig. 3). The N₂O distribution showed an inverse relationship with SST, and significant differences (Tukey, p<0.05) were observed only between the Eastern and Western extremes of the Mediterranean Sea. Notably, N₂OSat levels followed an opposite longitudinal trend to absolute concentrations, with higher values at the Eastern Basin (ranging from 100±4 % at BAL-CS to 116±11 % in the Levantine stations). Despite the widespread N₂O oversaturation, these values remain relatively low compared to those in other coastal seas (Charpentier et al., 2007; Sommer et al., 2025). This finding suggests that both external inputs and in situ biological production, such as nitrification and denitrification processes in the water column or benthic compartment, are comparatively weak in coastal waters of the studied stations, consistent with the ultra-oligotrophic nature of the Mediterranean Sea.

Nutrient concentrations reflected a predominantly oligotrophic regime, although episodic enrichment was observed near the coast (Table 2). Maximum values of nitrate (up to 7.78 µM at LEV-BEY) and phosphate (up to 6.23 µM at LEV-B1) suggest localized continental influence. Specifically, the elevated silicate levels in the Eastern Basin (mean > 2 µM) coupled with high nutrient peaks point towards terrestrial runoff and potential anthropogenic inputs from major coastal urban centers as the primary drivers of coastal nutrient variability. Nutrient stoichiometry indicated a shift from the typical phosphorus-limited regime of the open Mediterranean towards nitrogen-limited conditions in several coastal sites. Specifically, low N : P ratios (<10) in stations such as LEV-B1 and BAL-CA, coupled with high silicate concentrations (up to 5.61 µmol L⁻¹), suggest a strong influence of terrestrial runoff and urban effluents. This is consistent with the description of the study sites (Sect. 2.1), where the proximity of anthropogenic activities is identified as a potential driver of local biogeochemical variability. The highest phytoplankton biomass was recorded at LEV-BEY (mean 0.45 ± 0.22 µg L⁻¹), where a more balanced N : P ratio (≈ 18.7) appears to support enhanced primary production compared to the more nitrogen-limited eastern sites.

The N₂O concentrations showed a strong negative correlation with temperature for the entire database collected from all study sites; showing that both the temporal and spatial scales of variability were highly controlled by temperature (Fig. 4). This negative correlation points to a thermodynamic control of the variability through solubility, which increases N₂O in colder water. However, there was a positive correlation between N₂OSat and temperature (N₂OSat = 0.82 ⋅ T + 89.96; r=0.58, n=109, p≪0.01), with higher values in summer in the Eastern Basin than in winter in the Western Basin, corresponding to an enhancement of N₂O microbial production with increasing temperature. In contrast, CH₄ concentrations showed the opposite behaviour to N₂O, with greater variability and a positive correlation with temperature, indicating mechanisms other than solubility controlling the variability and important sources of CH₄ in the Eastern Basin.

The Spearman correlation matrix accounts for overall variability at both temporal and spatial scales, as shown in Fig. 5 for N₂O and CH₄ concentrations. The correlation analysis showed that N₂O was strongly correlated with temperature (-0.95, p<0.01) and significantly negatively correlated with salinity (-0.51, p<0.01), Chl a (-0.26, p<0.05), and silicate (-0.35, p<0.01) and was highly positively correlated with O₂ (r=0.81, p<0.01). The CH₄ showed an opposite behaviour, with a significant negative correlation with oxygen (-0.42, p<0.01) and positively correlation with salinity (r=0.59, p<0.01), temperature (0.49, p<0.01), Chl a (0.4, p<0.01), and silicate (0.4, p<0.01). Temperature was clearly the main controlling factor, especially for N₂O, and the overall correlation with salinity was stronger than when the analysis was performed at every station. This could be due to the higher salinity in the Levantine Sea and the zonal distribution pattern of salinity in the basin itself. Furthermore, the analysis of the Spearman correlation matrix for each station revealed additional local mechanisms acting on the seasonal scale that could be relevant, such as a significant positive correlation in O₂ among Western Mediterranean stations (BAL-CA, BAL-CS, BAL-PB, and BAL-REM) that was not significant in the eastern Levantine sub-basin. In addition, the strong negative correlation between N₂O and temperature was ubiquitous across stations but was only correlated with salinity in LEV-B1 (negative correlation) and ALB-M'DIQ (positive correlation). Finally, Chl a was only significant for temporal variability at the LEV-BEY and BAL-CS stations. Although oceanic CH₄ often exhibits a strong depth dependence (Weber et al., 2019), station depth was included in our initial correlation matrix but yielded no significant relationship (p>0.05). This lack of correlation indicates that CH₄ variability found among stations in our study in our study area is primarily driven by other sedimentary sources rather than depth-dependent benthic remineralization.

Our study is consistent with previous studies in the Eastern (Bange et al., 1996) and Western Mediterranean Sea (de la Paz et al., 2015; Flecha et al., 2023, 2025), where temperature was revealed to be the main controlling factor of surface N₂O and CH₄ variability.

Surface waters were consistently supersaturated with CH₄ relative to the atmosphere at all the coastal Mediterranean stations (Table 2). Consequently, the region acted as a net CH₄ source, with air-sea flux exhibiting high temporal variability (Table 3). Fluxes ranged from 0.97 to 11.41 µmol m⁻² d⁻¹ in the Balearic sub-basin, 1.74 to 15.4 µmol m⁻² d⁻¹ in the Alboran sub-basin, and reached a maximum of 35.2 µmol m⁻² d⁻¹ in the Levantine Sea (Fig. 6, Table 2). While no uniform seasonal trend was observed basin-wide, a notable regional decoupling occurred: maximum CH₄ outgassing in the Balearic sub-basin (July–November 2023) coincided with the lowest flux periods in the Alboran and Levantine sub-basins.

In contrast, N₂OSat and subsequent air-sea exchange followed a more distinct seasonal trend. Most stations remained moderately supersaturated, acting as a minor N₂O source, with the exception of brief undersaturation events (>98 %) in the Balearic sub-basin in May and November 2023. N₂O fluxes peaked in the Balearic sub-basin at 2.37 µmol m⁻² d⁻¹ (January 2024), driven by a combination of supersaturation and intensified winter wind speeds in the Western Mediterranean. Overall, the N₂O flux variability mirrored seasonal concentration changes but was strongly modulated by wind-driven gas transfer.

The annual flux average (Table 3) confirmed that all stations behaved as net annual sources of both CH₄ and N₂O. The highest mean CH₄ fluxes were recorded at the Erdemli site (LEV-ETS: 13.68 ± 9.57 µmol m⁻² d⁻¹). At this site, the substantial discrepancy between the mean and median (8.39 µmol m⁻² d⁻¹) highlights a heavily right-skewed distribution caused by episodically high-concentration events unique to the Levantine Sea. Conversely, annual N₂O fluxes were remarkably homogeneous across subbasins, ranging from 0.42 ± 0.26 µmol m⁻² d⁻¹ (ALB-M'DIQ) to 1.11 ± 0.64 µmol m⁻² d⁻¹ (BAL-PB), showing a symmetric distribution with closely aligned mean and median values.

The magnitude of the air-sea fluxes calculated in this study is consistent with the sparse literature on the Mediterranean Sea (Table 4). Our CH₄ fluxes in the Alboran sub-basin (1.74 to 15.4 µmol m⁻² d⁻¹) aligned with the ranges reported by de la Paz et al. (2015) in the Strait of Gibraltar and the recent observations by Flecha et al. (2023, 2025), who highlighted the coastal Alboran and Balearic seas as persistent sources of CH₄ emissions. The annual mean fluxes of N₂O (0.42 to 1.11 µmol m⁻² d⁻¹) are in excellent agreement with the values reported by Bange et al. (1996) for the Aegean Sea (0.1 to 1.2 µmol m⁻² d⁻¹), confirming that both the Eastern and Western basins act as moderate but steady sources of N₂O to the atmosphere. However, the episodic high-flux events recorded at the LEV-ETS station represent some of the highest CH₄ emission rates documented for the offshore Levantine sub-basin, emphasising the role of localised coastal hotspots in the regional greenhouse gas budget.