4.1. Chemical Characteristics of the NAd River Waters
The rivers of the NAd showed high A
T concentrations, with mean values of 3115 ± 536 µmol L
−1 for the Po River, the main freshwater source in this continental shelf, to extremely high values in some smaller karst rivers such as the Timavo, Rižana and Dragonja (
Table 1 and
Table S1). Based on the pH levels (7.50–8.17), the HCO
3− was the predominant form of dissolved inorganic carbon in the river water (2613–5139 µmol L
−1). The Ca
2+ concentration was in the range of 860–2131 µmol L
−1, while the Mg
2+ concentration was in the range of 306–638 µmol L
−1, except for the Dragonja River, which had very high concentrations of these two elements at the mouth of the river due to saltwater intrusion. The mean values of the [A
T − DIC] in the river waters were negative for the Po, Brenta, Piave, Timavo and Dragonja, indicating a frequent excess of the DIC with respect to the A
T, which originated from the CO
2 sovrasaturation in freshwater systems. In the other rivers, the [A
T–−DIC] was slightly positive.
The stable isotopic composition of the dissolved inorganic carbon showed significantly lower values for the Dragonja, Rižana and Timavo than for the other major NAd rivers. The values of the Ca
2+, Mg
2+ and HCO
3− were high compared to the global weighted average concentrations (370, 140 and 870 µmol L
−1, respectively) [
6]. The Ca
2+ and Mg
2+ in the river waters were primarily derived from rock weathering. The Ca
2+ sources consisted of carbonate rocks, such as calcite and dolomite, with a minor contribution from Ca–silicate minerals. Dolomite is also a source of dissolved Mg
2+ in the same ratio as Ca
2+. Calcite and dolomite occur almost exclusively in sedimentary rocks and contribute, on average, 65% of the Ca
2+ concentration in river water [
6].
4.2. Mg2+/Ca2+ Molar Ratios in River Waters
The mean Mg
2+/Ca
2+ molar ratios ranged from 0.12 to 1.07, indicating a relevant variation in the contribution of calcite and dolomite dissolution in the different watersheds (
Figure 2,
Table A4). In particular, they showed a greater contribution of calcite-to-rock weathering for the Po and Isonzo rivers compared to the other major NAd rivers, and a very high contribution for the karst rivers such as the Timavo and Rižana. The catchment area of the Po and its tributaries covers a large part of northern Italy and is surrounded by the Alps in the north and the Apennines in the south. The Apennines are the most affected by rock weathering and sediment transport, where many easily erodible marine sedimentary rocks rich in calcite are found [
39,
40]. On the other hand, the Isonzo catchment develops in the Eastern Alps, where calcite and dolomite rocks are predominant, and in a karst region, where Cretacean carbonates predominate, before flowing through the plain [
41]. The Timavo and Dragonja catchments are smaller and completely embedded in a karst area (
Figure 2).
It should be noted, however, that the relationships between Mg
2+ and Ca
2+ and HCO
3− discharges can be further decoupled by other environmental factors. An overflow of bicarbonates may result from enhanced rock weathering due to acid precipitation, primarily due to the deposition of sulfuric and nitric acids of anthropogenic origin, while outgassing and primary production reduce the HCO
3− concentrations in river waters relative to the Mg
2+ and Ca
2+ concentrations [
6]. The relationships between Ca
2+ and Mg
2+ relative to HCO
3− revealed important differences among the NAd rivers. For the major rivers, from the Po to the Isonzo, the Mg
2+ concentration was linearly related to the HCO
3− concentration with a ratio of 0.18, indicating a non-negligible contribution of dolomite weathering to the HCO
3− transport (
Figure 3a,
Table A1).
The Timavo, Rižana and Dragonja rivers had low Mg
2+ concentrations that were not related to HCO
3−, except for some very high values measured at the mouth of the Dragonja River that were influenced by saltwater intrusion. In contrast, the relationship between Ca
2+ and HCO
3− was significant for all the rivers with a ratio of 0.38 (
Figure 3b,
Table A1). Overall, these ratios indicated simultaneous weathering of dolomite and calcite in the river basins, from the Po to the Isonzo, and predominant calcite weathering in the smaller karst rivers in the easternmost part of the NAd.
4.3. Isotopic Signature and Weathering Intensity of the Riverine DIC
The stable isotopic composition of the DIC in freshwater (δ
13C
DIC) is generally between −25 and 0‰, depending on the main processes affecting the balance of the bicarbonates in the river basins. The weathering of carbonate rock produces δ
13C
DIC values around 0‰, while the respiration of the soil and aquatic organic matter produces negative values (down to −30‰) due to the preferential release of isotopically light carbon (
12C) during the remineralization of organic matter. The primary production that preferentially assimilates
12CO
2, as well as the equilibration of the freshwater DIC with air CO
2 (δ
13C
DIC ≈ −8‰), generally results in progressively fewer negative δ
13C
DIC values in the rivers downstream. The δ
13C
DIC values in freshwater may also change seasonally due to the annual cycles of the biogeochemical processes in the river ecosystems [
24,
42].
The δ
13C
DIC data in the NAd rivers fell within a range of values from −12.4 to −5.7‰ (
Figure 4,
Table A1), and three different groups of data were identified.
The first group had the highest AT, DIC and HCO3− and the most negative δ13CDIC values, including the Dragonja, Rižana and Timavo. In the small rivers flowing in the karst catchments, the δ13CDIC values from the weathering of carbonate rocks were significantly shifted toward the negative values by the contribution of the remineralization of organic matter, which also produced the highest DIC and HCO3− concentrations in freshwater.
The second group had a highly variable AT, DIC and HCO3− and medium to low values of the δ13CDIC. This group included rivers such as the Brenta, Piave and Isonzo, which were probably influenced by a combination of processes along their river networks, such as the weathering of carbonate rocks, primary production and CO2 exchange with the atmosphere, which increased the variability of these parameters. Minimal information was available on the δ13CDIC values in the Po River, but the data showed that these were intermediate values compared to the other NAd rivers, at least at the mouth of this large basin.
The third group presented a low AT, DIC and HCO3− and the least negative δ13CDIC values from the Adige and Tagliamento Rivers, which might have been due to the predominance of carbonate weathering in all the available sampling sites.
Despite the large variability in the river ecosystem characteristics of the NAd catchment, significant overall relationships were found between the A
T, DIC, HCO
3− and δ
13C
DIC (
Figure 4a–c). These relationships were typical of sub-tropical temperate rivers, where the large DIC budgets in freshwater result from a variety of factors and not just from the decomposition of fresh organic carbon, as found in tropical rivers [
43].
The NAd rivers basins were also characterized by a high runoff intensity compared to their extent, due to high annual precipitation and snowmelt in the surrounding mountain ranges [
44]. Roy et al. [
45] also found that interrelated factors such as the lithology, water residence time, mechanical erosion, etc. have a higher influence together than individually. Most studies on weathering in alpine regions concluded that enhanced mechanical erosion in these environments also increases chemical weathering [
46,
47,
48]. The stress on the mineral and an increased rock surface area create conditions under which minerals are more easily dissolved.
The global theoretical models of the CO
2 consumption in carbonate watersheds showed a value of the A
T near 3000 µmol L
−1 determined from a best-fit line [
49]. Although this value was reasonable as an average for all the NAd watersheds, many drainage basins have waters with much higher HCO
3− concentrations. The NAd river values ranged from 2.0 to 4.5 mmol HCO
3− per liter of runoff (
Figure 5,
Table A3).
In particular, the Tagliamento, Piave, Brenta and Livenza rivers showed a high weathering intensity of the HCO
3− and DIC. This finding was consistent with the fewer negative δ
13C
DIC values found in some of these rivers, suggesting a stronger contribution of carbonate rock weathering to the total transport of the HCO
3−. The Po River had a similar ratio for the HCO
3− weathering intensity to the specific runoff as the Danube (∽3 mmol L
−1), although the Danube had a catchment area approx. ten times larger and a median annual flow approx. four times greater [
30]. This result confirmed that the largest European rivers have high HCO
3− and DIC discharges compared to rivers in other continents (
Figure 5,
Table A3). The DIC weathering intensity is also high in Asian tropical rivers, which have the highest values (25.85 mmol C km
−2 s
−1) among tropical rivers worldwide (8.69 mmol C km
−2 s
−1) [
50]. The DIC concentrations in Asian tropical rivers are lower than in the NAd rivers, possibly due to the abundant vegetation that reduces weathering and greater aquatic photosynthesis [
50,
51]. However, the contribution of European freshwater discharges should be better evaluated by including the data for medium and small catchments [
43,
52] and for submarine groundwaters [
53,
54].
The average isotopic signature of the NAd rivers was estimated to be −10.0 ± 1.7‰, while the Timavo, Rižana and Dragonja had −11.6 ± 0.4‰, using the equation in
Section 3.3. These data are close to a δ
13C
DIC signature of −10‰, which is now considered to be representative of the DIC riverine inputs in ocean carbon cycle modeling [
38]. Tropical rivers, such as the Amazon, Congo and Niger, often have strongly negative δ
13C
DIC values due to the large contribution of fresh organic carbon decomposition to the freshwater HCO
3− fluxes. They are also characterized by a rather low HCO
3− weathering intensity, consistently with the global pattern of continental carbonate rocks and with their high discharges of organic carbon [
5,
24,
43].
The variation in the δ
13C
DIC signature can be compared to the ratio [H
2CO
3]/[HCO
3−] in freshwater, as already proposed in the literature [
49]. The plot of the δ
13C
DIC versus the ratio [H
2CO
3]/[HCO
3−] shows that two groups of rivers can be distinguished (
Figure 6).
Samples from the Timavo Rižana and Dragonja rivers, which show larger variations in the [H2CO3]/[HCO3−] ratio, ranging between 0 and 0.6 and low values of the δ13CDIC (−11 to −12.5‰), indicating no exchange with atmospheric CO2, but the presence of biogenic CO2.
The second group, with a very low [H2CO3]/[HCO3−] ratio (<0.1), exhibits greater variations of the δ13CDIC (−10 to −5‰), indicating different biogeochemical processes in the rivers such as an exchange with atmospheric CO2, biogenic CO2 and carbonate weathering.
4.4. Impact of the Riverine AT and DIC in the NAd
In the 2010s, the annual freshwater discharge of the Po River (27.2–75.1 km
3 yr
−1) was approx. an order of magnitude higher than the other NAd rivers (0.6–10.7 km
3 yr
−1). The year 2014 showed particularly high runoff, while 2017 was the driest year of this decade (
Figure 7a,
Table A2). For the other rivers, the freshwater discharge was in the order of the Adige > Brenta > Livenza > Piave. The transport of the A
T and carbonate species followed the freshwater discharge, maintaining the same distinction between the Po and the other NAd rivers (
Figure 7b–d,
Table A2).
The annual transports of the AT, DIC and HCO3− by the Adige, Brenta and Livenza rivers were very similar, despite the differences in the freshwater discharges among these rivers. The Piave River almost always had the lowest freshwater and carbonate loads. The interannual variability of the freshwater and carbonate discharges was pronounced, reaching approx. 65% of these amounts when the extremely wet (2014) and dry (2017) years were compared.
On average, the transport of the A
T, DIC and HCO
3− through the Po River reached the values of 144 ± 43, 149 ± 41 and 140 ± 41 Gmol yr
−1, respectively in the 2010s (
Figure 8,
Table A2). The other NAd rivers were in the order of the Brenta, Livenza, Adige and Piave. Despite the limited availability of the data, the Tagliamento and Isonzo rivers were considered to be of minor importance at a basin scale, although they had a significant impact on the carbonate marine system at the sub-regional scale [
24,
55]. Historical data for the A
T river concentrations in the 1970s were used by Copin Montegut [
15] to estimate the A
T discharge in the Mediterranean basin and reused by Cossarini et al. [
18] for the Adriatic Sea. These estimates were made only for the alkalinity in a few rivers using old data. Our current estimate is 5–15% higher than that calculated on the basis of the Copin Montegut [
15] data, primarily due to a higher A
T concentration in the riverine waters. The NAd rivers contributed to 64% of the total A
T discharged in the Adriatic basin, according to the estimate of Cossarini et al. [
18] and to 12% of the total A
T discharged by all Mediterranean rivers, according to the estimate of Copin Montegut [
15], or to 23%, according to the estimate of Cossarini et al. [
18].
The mean DIC flux of the Po was slightly lower than the flux of the Rhône (162 G mol C yr
−1) [
56], but almost four times lower than the flux of the Danube (629 G mol C yr
−1) [
5]. The total DIC flux into the NAd represented 13–15% of the total DIC input by rivers in the Mediterranean Sea, estimated by Sempéré et al. [
56].
The considered rivers did not represent the total discharge of carbonates in the NAd, which also consisted of small watercourses and freshwater discharges through coastal lagoons and submarine springs along the east coast. Nevertheless, a comparison can be made between the marine budgets of the A
T, DIC and HCO
3− in the NAd by calculating the turnover rates of these components. In the 2010s, the mean freshwater discharge of the NAd rivers was 65.7 km
3 yr
−1, which represented 24.7% yr
−1 of the seawater volume of the region shown in
Figure 1 (i.e., 266 km
3). These rivers also discharged an average of 205.2 Gmol yr
−1 of A
T, 212.8 Gmol yr
−1 of DIC and 202.6 Gmol yr
−1 of HCO
3−. Assuming the typical concentrations in seawater of A
T = 2745.4 ± 34,5 µmol L
−1, DIC = 2428.4 ± 44.7 µmol L
−1 C and HCO
3− = 2181.9 ± 73.6 µmol L
−1 (see
Section 3.3), the total budgets of these parameters in the NAd can be estimated to be 729.2, 645.0 and 579.5 Gmol, respectively. These data indicate that river discharges have an important influence on the carbonate system in the NAd, accounting for 28, 33 and 35% yr
−1 of the budget of these parameters in the sea, respectively (
Table 2).
The Northern Adriatic region under consideration receives the largest runoff in the entire Mediterranean compared to its seawater volume, since it is a shallow continental shelf surrounded by an extremely large catchment area. As a result, the mean annual outflows of the total nitrogen and total phosphorus can be as much as seven times higher than the budgets of these biogenic elements in the NAd [
28,
29]. The present study showed, for the first time, the importance of continental inputs of the A
T, DIC and HCO
3− in the context of river discharge, as they account for approx. one-third of the marine budgets of these parameters at the annual scale. However, human activities can alter the terrestrial weathering and land surface hydrology, and thus, the HCO
3− fluxes in the rivers [
14,
57]. Pollution, acid rainfall and the use of high N fertilizers can increase the acidification of the soils and promote carbonate dissolution [
5].
The potential effect of river discharges on ocean acidification in the NAd marine ecosystem can also be inferred from the composite parameter [A
T − DIC], which is a property conservative with the mixing and almost linearly correlated to the changes in the pH and the saturation state of the calcium carbonate [
58]. The river waters in the NAd often have higher concentrations of the A
T and DIC than the seawater (
Table 1), but their values of the [A
T − DIC] (−628 to 46 µmol L
−1) are often lower than the values usually found in seawater (321.6 ± 131.9 µmol L
−1). Negative [A
T − DIC] values characterized the Po, Brenta, Piave, Timavo and Dragonja, while the Adige, Livenza, Tagliamento Isonzo and Rižana had slightly positive values. Negative values are common to most of the rivers worldwide as they originate through the CO
2 sovrasaturation in freshwater systems [
58]. This feature means that, in these cases, the river discharges can potentially decrease the pH in the NAd coastal waters [
7,
14,
59]. However, it should also be kept in mind that large phytoplankton blooms occur in the river plumes due to the discharge of land-borne nutrients, often reaching eutrophic and hypertrophic conditions in the NAd [
28,
35]. Primary production can reduce the DIC concentrations, thereby increasing the pH in seawater.
River discharges are also characterized by a multiscale spatio-temporal variability due to the effects of climatic fluctuations and a continuous evolution of anthropogenic pressure in their catchments. For these reasons, more comprehensive monitoring of the carbonate system in freshwater ecosystems should be conducted to better model the future trends and variations of ocean acidification in the coastal areas [
1,
2].