4.1. Seasonal and Interannual Variations of Discharge, Nutrient Concentrations and Stoichiometry
Mid-term (over some consecutive years) and frequent (at least monthly) monitoring of rivers that drain large watersheds allows to provide reliable datasets to analyse the temporal trends in discharge and nutrient transport [
34]. This monitoring time span and frequency requirements are due to three main reasons: (1) Climatic microcycles may last for multiple years and are characterized by prolonged dry or wet phases; (2) climate, through hydrology, affects nutrient cycles; and (3) the effects of ongoing climate changes interact with those of implementation of policy directives that target nutrient reduction (e.g., land use, sewage water treatment plants, and the use of fertilizers) [
36,
50,
51,
52]. Due to reasons (1) and (2), a one- or two-year sampling program may fall in unusually low or high discharge periods that are not representative of a climatic microcycle and therefore lead to an incorrect conclusion regarding discharge (and nutrient) trends. Due to reason (3), long-term temporal series (10–30 years) are needed to disentangle the effects of climate from those of socio-economic ongoing changes.
The results from this study suggest that the discharge and nutrient concentrations at the Nemunas River closing section displayed a strong and regular seasonality (see
Seasonal index in
Table 2). The discharge was uncoupled from precipitation, suggesting air temperature, plant cover, and activity and evapotranspiration as factors that regulate the link between precipitation and river flow [
53]. The Nemunas River discharge generally peaked in spring, coinciding with snowmelt, and reached a minimum in summer, coinciding with the highest levels of precipitation and air temperatures (
Figure 2). Similar results were described for the Daugava River, which has a comparable annual discharge and lies in the same geographical area [
54]. The discharge is affected by a combination of precipitation patterns/intensities and evapotranspiration/accumulation in groundwater, as well as by steep changes in temperature that either drive snow/ice melt or water freezing [
55]. The air temperature therefore co-regulates river discharge, and the discharge affects the pattern of nutrient delivery from the watershed to the river.
During 2012–2016, the annual discharge was below the historical records, a result that aligns with that predicted by Jakimavičius and Kriaučiūnienė [
41] via climate change models. The different variabilities between precipitation and discharge can be explained by processes that occur within the Nemunas watershed. In particular, the large soil surface, which is covered by active vegetation during summer, may efficiently buffer the discharge from the effects of variable precipitation patterns by varying the evapotranspiration rates, resulting in constant flows [
56]. Due to the absence of active vegetation in cultivated land, this does not occur during winter.
All of the dissolved inorganic nutrients underwent pronounced variations during the 5-year study period, with alternating peaks and minima that were opposite to those recorded for their particulate forms. The seasonal trends were stable for NH
4+ and NO
3−, while they were more variable for DIP and DSi. Within the study period, only NO
3− increased significantly, a trend that is opposite to the predicted trend and desired political actions [
50]. According to recent inventories, agriculture and the associated conversion of pastures into croplands remain the main contributors of N loads transported by the Nemunas River [
50]. A major fraction of this N (and P) is generated from the section of the Nemunas watershed that belongs to Belarus, while a minor fraction has a natural origin [
50,
57]. Agricultural activities affect hydrology and nutrient cycling within croplands by altering filtration, groundwater recharge, base flow, and run-off from catchments [
58]. At the closing section of the Nemunas River, NO
3− dynamics were comparable to those recorded in other Baltic Sea rivers, with peaks during winter/spring and minima during summer/autumn [
54,
59]. During colder months, NO
3− loss from the watershed is expected to be higher, as biogeochemical processes that are able to transform NO
3− into organic or molecular N via uptake or denitrification are limited, while the retention time is reduced by precipitation and snow melting. In summer, high N uptake by crops and other vegetation, denitrification processes in soils and N uptake by phytoplankton all drive the lower downstream flux. The relationship between the riverine NO
3− concentration and discharge varied among years, likely due to different leaching from the land. We expect that during the driest years, the water table migrates vertically, allowing air to penetrate deeper within soils and nitrification processes to occur. Nitrate may accumulate within dry soils during summer and then be suddenly and massively transported when it starts raining, resulting in proportionally higher transport per unit discharge compared to wetter years (e.g., [
51]). In fact, during wetter years, when longer periods of soil saturation and anoxia and higher rates of denitrification are expected, proportionally lower nitrate export occurs [
60].
The variability of the concentrations of DIP and DSi was comparatively smaller, suggesting limited control by processes within the watershed (on both elements) or a constant background input from sewage treatment plants (P). Dissolved Si is also very important in the Curonian lagoon since it is necessary for diatom growth, which is a dominant group in spring phytoplankton [
61]. The DSi concentrations were highest during cold periods due to weathering of bedrock and groundwater transport. They decreased in late spring, likely due to Si uptake by both cultivated plants and riverine communities of blooming diatoms [
56]. The same trends were found in other Baltic rivers [
54,
62]. The nutrient concentration ranges in the Nemunas River were similar to those reported in the Daugava River [
54].
The ecological stoichiometry of the three nutrients is central to the analysis of watersheds and nutrient export to coastal areas [
63,
64]. Within the BONUS program, the COCOA project (Nutrient Cocktails in Coastal zones of the Baltic Sea, 2014–2017) targeted the origin, transport, and fate of the three nutrients in different environments of the Baltic Sea. This approach is important since the relative availability of nutrients, much more than their single abundances, regulates algal blooms or drives the efficiency of decomposition processes [
65]. Frequent cyanobacterial blooms in the shallow lagoons along the coast of the Baltic Sea suggest favourable nutrient ratio for their dominance [
66,
67]. Our results show that the seasonality of nutrient transport resulted in a variable ecological stoichiometry of N, Si, and P. In general, we observed a large excess of N in autumn, winter, and spring (DIN:DIP > 16), but a marked N deficiency (and to a minor extent of Si) from May to August during the five analysed years. With respect to downstream water bodies (Curonian Lagoon and Baltic Sea), this strong N and Si summer limitation may favour the succession of phytoplankton communities from diatom-dominated (spring) to cyanobacteria-dominated (summer) [
13,
24]. During warm months, cyanobacteria have a competitive advantage as they do not require silica for their exoskeleton and can fix relatively inert dinitrogen (N
2) when N is limiting. Any P excess may therefore favour their development [
31]. Recent findings suggest that cyanobacteria dominance can offset the attenuation of N load by the Curonian lagoon via denitrification, anammox, uptake, and internal storage, while enhancing its export to the Baltic Sea [
68]. Moreover, cyanobacteria blooms have large economic impacts in the Curonian Lagoon due to high rates of respiration in the water that can favour night-time hypoxia (<62.5 µM·O
2, [
27]) and mortality of benthic organisms and fish [
27]. The present study suggests that future political actions should target further P reductions in the Nemunas watershed to tackle the unbalanced nutrient stoichiometry [
30].
4.2. Nutrient Trends: Past, Present and Ongoing Changes
Compared to previously published data, the loads of dissolved inorganic as well as the total forms of N and P measured from 2012–2016 have changed, while to our knowledge there is no previously published information on Si (
Table 5). From 1997–2002 to 2012–2016, the NO
3− loads decreased moderately, whereas the load reductions of NH
4+ and DIP were much more noticeable from the 1986–1991 and 1980–1993 datasets, respectively (
Table 4). Since riverine discharge remained relatively similar, we address the substantial decrease of NH
4+ and DIP loads to the modernization of water treatment and construction of new sewage treatment plants after the entrance of Lithuania into the European Union. Interestingly, the socio-economic changes in society such as the increasing use of phosphorus-free detergents will be responsible for further decrease of P inputs to natural environments from households [
18]. The trend in the DSi loads remains rather unclear. In the present study, the estimated nutrient loads were similar to those reported for rivers entering the Gulf of Riga [
54]. Humborg et al. [
69] and Conley et al. [
70] showed that the DSi concentrations and loads tended to decrease during the past few decades in other large tributaries of the Baltic Sea due to river damming, other anthropogenic activities in catchments, or changes in the composition of detergents. These changes in loads might correspond to an increasing Si limitation over N, which may result in a shift from diatom to green algae blooms during spring.
In Lithuania, the updated HELCOM compilation of trend analyses in nutrient loads projected a decrease of 1142 t of TN and 63.4 t of TP during the 1995–2010 period [
30]. Our results show that until 2014, the annual loads of TN and TP followed this trend even faster than expected. However, the trend changed later, when TN started to increase. To meet the targets for sustainable total N and P loads to the Baltic Sea, Lithuania agreed to reduce annual TN and TP loads to 11,750 t and 880 t, respectively, by 2021 [
71]. These targets appear to be realistic for P, but they remain challenging for TN, as the mean load calculated in the present study exceeds the projected threshold by ~30%. Šileika et al. [
29] suggest that converting 20% of arable land to pasture, together with sewage treatment plant improvements, would reduce the yearly TN export to 12,000 t, which is close to the target set for Lithuania. Improvement of agricultural practices would not necessary result in lower contribution of diffuse sources to N loads as the transfer of N from arable land to surface and ground water largely depends on storage and immobilisation processes within soils [
36,
52]. During the Soviet Union period, the low cost and the intensive application of N fertilizers in agriculture likely caused N accumulation in soils. Leaching of previously accumulated fertilizers can contribute nutrient concentrations in rivers for decades, as was demonstrated in southern Europe watersheds [
52]. Such elevated background input may mask the expected positive effects of more sustainable agricultural practices (e.g., nutrient reduction). According to the Lithuanian Department of Statistics, over the last decade, the pasture area has remained relatively constant, while cattle number decreased by 7% from 2004 to 2016 and croplands have increased by 42% in the same period (
Figure 6). The changes in agricultural practices coincided with new subsidies after the entrance of Lithuania in the European Union in 2004; as market price and subsidies for crops increased, farmers maintained or expanded this agricultural practice. Furthermore, land cover changes are accompanied by increased application of fertilizers, which can lead to higher leaching of N in dry years, when crop yields are low [
10]. Modelled future intensification scenarios of agricultural practices for selected catchments in the Southeast Baltic region (Poland and Baltic States) reveal an increase in N loading by nearly 30% [
72]. Climate change may also result in increased crop production, especially in Belarus, further stimulating both the conversion of forests or pastures into arable land as well as the use of fertilizers. Under such a scenario, the achievements of nutrient reduction plans along the Nemunas watershed become challenging for
downstream countries as Lithuania due to major anthropogenic N and P loadings generated
upstream, in Belarus. Nevertheless, trends in net anthropogenic N and P inputs can be confounded by ongoing regional changes of lifestyles [
14,
73]. Therefore, future studies addressing the genesis of nutrient loads in the Nemunas watershed should integrate the combined effects of climate, agriculture, and socio-economic changes.
A major fraction of nutrient loads from large catchments to the Baltic Sea passes through large estuarine systems, such as the Curonian, Oder, and Vistula lagoons, where nutrients can be partly retained [
21,
71]. It was demonstrated that 14–88% of the TN loads and 27–89% of the TP loads were retained during their transport through the large Baltic lagoons [
24,
74]. Within lagoons, retention can be sometimes offset by cyanobacterial blooms, which result in large export of algal cells, may induce bottom anoxia, and may favour P release from anoxic sediment and its transport to the Baltic Sea [
24,
68,
75]. The latter point demonstrates that the TN and TP loads can differ substantially depending on whether they are measured at the end of the catchment (the river closing section) or at the interface between the estuary and the Baltic Sea (the lagoon mouth). However, in the Baltic Sea Action Plan (BSAP) adopted by HELCOM, the maximal allowable nutrient loads to the Baltic Sea are measured at the river’s gauging section, excluding the processes within estuaries. This generally emphasizes the need to review BSAP and adapt regional BSAP-related goals [
76]. In fact, the N and P reduction targets for different Baltic Sea countries could be achieved more rapidly if retention processes in transitional waters are included in models and calculations. Continuous monitoring of major rivers, such as the Nemunas, and of the processes that occur in downstream environments, such as estuarine systems, is of key importance as recent studies have demonstrated frequent switches between the role of transitional aquatic environments as nutrient sinks and sources [
24,
27,
75].