The study area is situated in northeastern Germany, approximately 10 km southeast of Rostock (Figure 1). Characterized by conventional arable use covering 94% of the region, the landscape features minimal forested areas, accounting for only 6%. The catchment is located in a glacially formed landscape with a flat topography and only gentle slopes (slope gradients <3%). Mineral soils (mainly sandy loams) are characteristic for the area, with a predominance of Luvisols (48%), Gleysols (43%), and Stagnic Gleysols (9%). The four agricultural fields in the study area (Figure 1) are tile-drained with a drainage depth 1.1 m and a drainage distance from 8 to 22 m.

Water originating from the tile drains flows into a main ditch (Figure 1). The drainage plot (field 4) and the ditch catchment (including fields 1, 2, 3, 4) are 4.2 ha and 179 ha in size, respectively. Tile drain flow is most pronounced during the winter months, extending from November to April. This temporal pattern corresponds to a precipitation surplus attributable to lower temperatures and reduced evapotranspiration rates during this period. Mean annual precipitation was 614 mm and the mean annual temperature was 9.1°C (2001–2022). While precipitation is predominantly in the form of rainfall, certain winters experienced substantial snow cover, notably in 2009 and 2010. Due to the high areal proportion of tile drainage (94%), discharge in the ditch is mainly controlled by tile flow and groundwater flow (Bauwe et al., 2016). In contrast, surface runoff is only sporadically being observed during intense storm events.

The sampling station positioned at the outlet of the drainage plot, linked to the collector drain, is equipped with a Venturi flume (RBC-flume, UGT GmbH, Germany). This apparatus enables continuous and automatic water level measurement at 15 min intervals. The recorded water level data are subsequently converted into flow data at the collector drain through the application of regression equations supplied by the manufacturer. The sampling station at the ditch located at the catchment outlet is equipped with an automatic, ultrasonic, water level measurement device (Teledyne Isco, Inc., Lincoln, NE). We conducted frequent flow measurements, typically on a weekly basis, utilizing an inductive flowmeter (Flo-Mate TM, Marsh-McBirney, Inc., Frederick, MD) to develop rating curves for this station. Rating curves were used to calculate daily flow data using the automatic water level measurements.

Water samples from both the drainage plot and the ditch were systematically collected using ISCO samplers (Teledyne Isco, Inc., Lincoln, NE), typically a minimum of twice a week. After collecting the water samples, they were promptly transported to the laboratory and frozen at −20°C to preserve their integrity until the subsequent analysis for NO₃⁻-N. The analysis of NO₃⁻-N was conducted through ion chromatography, utilizing equipment from Metrohm AG, Herisau, Switzerland. To derive daily, monthly and annual NO₃⁻-N loads for both the drainage plot and the ditch catchment, we used the R software package RiverLoad (Nava et al., 2019). This package provides different methods for load estimation, such as averaging methods, ratio estimators and regression methods. Based on our dataset with continuous flow data and instantaneous concentration data, load estimation using linear interpolation of concentration data was appropriate for this study:

where L is the NO₃⁻-N load, C_j^int (g m⁻³) is the daily NO₃⁻-N concentration linearly interpolated between two measured samples, Q_j (m³ s⁻¹) is the mean daily streamflow and K″ is a conversion factor for the period of load estimation (Moatar and Meybeck, 2005).

It is pertinent to note that the annual values for both hydrological data and NO₃⁻-N data pertain to a hydrologic year spanning from November to October.

Field records from the local farmer, encompassing detailed agricultural practices, served as the foundation for calculating annual N soil surface budgets from 2001 to 2022. The crop rotation during the investigation contained common crops such as winter wheat (Triticum aestivum L.), winter barley (Hordeum vulgare L.), rapeseed (Brassica napus L.), and maize (Zea mays L.) (Table 1). Periodically, sugar beet (Beta vulgaris L.) and legumes such as peas (Pisum sativum L.) or alfalfa (Medicago sativa L.) were also cultivated. This crop rotation is typical for conventional arable farming in northern Germany. Annual N applications varied based on crop selection, as detailed in Table 1. N fertilizers, primarily in mineral form, were applied three to four times a year. Winter wheat and winter barley were typically sown in the latter part of September and harvested in the subsequent summer. Rapeseed was sown about a month earlier, and its harvest coincided with that of winter wheat and winter barley. Maize, sugar beet, and peas were sown in late April and harvested in late summer or early fall of the same year. Table 1 additionally provides information on N in yields for the main crops.

N soil surface budget was calculated for all four fields in the catchment accounting for mineral and organic fertilization, N fixation, and harvest as follows:

where N_sur is N soil surface budget, N_min is amount of N in mineral fertilizer, N_org is amount of N in organic fertilizer accounting for ammonia volatilization, N_fix is amount of N fixed by legumes, N_atm is amount of atmospheric N deposition, and N_yld is amount of N in harvested yield. All units are (kg ha⁻¹). Data on the N content of organic fertilizers, harvested crops, N fixation rates, and losses due to ammonia volatilization were sourced from literature (MLUV-MV, 2008, 2019). The atmospheric N deposition was 14 kg ha⁻¹ year⁻¹ for the study catchment.¹ The N content of mineral fertilizers was obtained from manufacturer specifications. Fertilizer application details and crop yields for each field were provided by the farmer. Notably, the N soil surface budgets of the drainage plot align with those of field 4 (Figure 1), while those of the ditch catchment were computed as area-weighted means of the four fields within the catchment.

Streamflow and NO₃⁻-N concentrations at both the collector drain and the ditch exhibited pronounced dynamics throughout the study period (Figure 2). Peak flows and maximum annual NO₃⁻-N concentrations predominantly occurred during the winter, while both streamflow and NO₃⁻-N concentrations approached near-zero levels during the summer months. NO₃⁻-N concentration levels were comparable at the collector drain and in the ditch during winter. Mean flow rates during the winter months (November until April) at the collector drain and in the ditch were 0.4 and 15.5 L s⁻¹, respectively. Conversely, the collector drain often experienced dry conditions in summer (May until October), with an exception in the summer of 2011 when an extraordinary precipitation event (98 mm day⁻¹) led to substantial flow. Mean flow rates in the ditch during summer were relatively low at 5.1 L s⁻¹. Although NO₃⁻-N concentrations were typically below the drinking water limit of 11.3 mg L⁻¹ in summer (mean values of 8.0 and 3.1 mg L⁻¹ at the collector drain and in the ditch, respectively), they frequently exceeded this limit in winter (Figure 2) (mean values of 10.8 and 9.0 mg L⁻¹ at the collector drain and in the ditch, respectively).

The general trend of low NO₃⁻-N concentrations in summer and high NO₃⁻-N concentrations in winter (Figure 2) can be attributed to the activation of different N sources. During the summer, characterized by low groundwater tables, it is reasonable to assume that mainly groundwater, with low NO₃⁻-N concentrations, contributes to the flow. In contrast, during winter when the water table rises, NO₃⁻-rich water from the upper soil layers significantly influences the chemical composition of the discharge. This seasonal NO₃⁻-N pattern in surface waters of agricultural catchments aligns with findings described by various researchers, including Pionke et al. (1999), Dupas et al. (2016), Exner-Kittridge et al. (2016), Ford et al. (2018) and Abbott et al. (2018).

Over our observation period, no discernible trends were identified regarding the magnitude of NO₃⁻-N concentrations. The visual interpretation of Figure 2 and the plotting of NO₃⁻-N concentrations as a function of flow (Figure 3) revealed a positive relationship between flow and NO₃⁻-N concentrations. This positive relationship between streamflow and NO₃⁻-N concentrations was more pronounced in the ditch (Figure 3). This can be attributed primarily to water flow from groundwater with very low NO₃⁻-N concentrations in the ditch during summer, whereas the discharge season at the collector drain is typically confined to the winter, during which NO₃⁻-N concentrations from drainage water are generally higher. The different NO₃⁻-N concentration pattern of the collector drain and the ditch will be visible when plotting the NO₃⁻-N concentrations as histograms (Figure 4). The main difference between both figures is the large number of samples exhibit NO₃⁻-N concentrations from 0 to 5 mg L⁻¹ in the ditch. These samples were taken in the vast majority of cases in summer, when there was no water in the collector drain. Excluding these low concentration samples, both histograms are similar. Only, a small number of samples (104 at the drain collector and 44 in the ditch) contained NO₃⁻-N concentrations beyond 20 mg L⁻¹.

In summary, both flow patterns and the dynamic behavior of NO₃⁻-N concentrations at the collector drain and in the ditch exhibit a high temporal parallelism (Figure 2), emphasizing the significance of the artificial drainage system in discharge generation. Furthermore, it identifies the drainage system as a notable source of NO₃-losses at larger spatial scales such as ditches, which confirms an earlier study conducted in the same study area analyzing a 3-years period of time (Tiemeyer et al., 2006).

NO₃⁻-N loads (Equation 1) were clearly driven by discharge. This holds true both for a daily and for an annual evaluation (Figure 5). The relationship between flow and NO₃⁻-N losses was even stronger compared to NO₃⁻-N concentrations. Annual NO₃⁻-N loads varied greatly. Highest annual NO₃⁻-N loads were observed at the collector drain (50 kg ha⁻¹) and at the ditch (52 kg ha⁻¹) in 2007 followed by very high loads in 2011, while NO₃⁻-N loads were lowest in 2019. In that year, no NO₃⁻-N was transported via the collector drain, since the collector drain did not carry any water and annual NO₃⁻-N load was 1 kg ha⁻¹ at the ditch. These 2 years reflect extreme weather conditions. The high NO₃⁻-N loads in 2007 and 2011 were mainly caused by very wet conditions in summer, while the low NO₃⁻-N loads in 2019 were a result of a prolonged drought period in 2018. Mean annual loads at the collector drain (19.7 kg ha⁻¹) and the ditch (19.9 kg ha⁻¹) were similar.

Due to the strong relationship between flow and NO₃⁻-N loads it is not surprising that NO₃⁻-N loads follow a pronounced seasonal pattern (Figure 6). High discharge rates in winter, in particular in January, February and March, caused high NO₃⁻-N loads during that time. Consequently, highest monthly NO₃⁻-N loads were recorded in January, both at the collector drain (4.7 kg ha⁻¹) and at the ditch (4.4 kg ha⁻¹), while NO₃⁻-N loads were lowest in October at the collector drain (0.2 kg ha⁻¹) and in September at the ditch (0.3 kg ha⁻¹).

A significant amount of NO₃⁻-N was leached during a short period of time and the majority of NO₃⁻-N losses were restricted to winter. For example, the extraordinary high rainfall event (98 mm) on July 23, 2011 led to 10% of the annual NO₃⁻-N load being discharged within the next 7 days. To underpin the importance of seasonality of NO₃⁻-N leaching, it should be noted that 65 and 61% of the annual NO₃⁻-N load is being between January and March at the collector drain and the ditch, respectively (Figure 6). These NO₃⁻-N export mechanisms can be illustrated when the daily cumulative NO₃⁻-N load starting with the highest load is plotted against time (Figure 7).

When analyzing Figure 7, it is striking that the majority of the NO₃⁻-N load is being discharged during a very short period of time. Twenty-seven days (out of 4,259 discharge days) with the highest NO₃⁻-N loads were enough to contribute 10% of the total load at the collector drain. Even more extreme, 22 days out of 6,679 discharge days contributed to 10% of NO₃⁻-N load at the ditch. 50% of the total NO₃⁻-N load was discharged, when 10 and 5% of the considered time period has passed at the collector drain and the ditch, respectively. At the other end of the spectrum, it can be stated that only 10% of the total NO₃⁻-N load is being discharged in 50 and 75% of the total time at the collector drain and the ditch, respectively. The graph of the collector drain in Figure 7 is flattened compared to the graph of the ditch. This is mainly because there is no flow and therefore no NO₃⁻-N leaching at the collector drain in summer with very low NO₃⁻-N loads. Generally, the range of the NO₃⁻-N loads at the collector drain is smaller compared to the ditch due to missing low summer values.

From 2001 to 2022, The N input at the drainage plot ranged from 40 kg ha⁻¹ year⁻¹ to 275 kg ha⁻¹ year⁻¹, while the N input varied from 109 kg ha⁻¹ year⁻¹ to 228 kg ha⁻¹ year⁻¹ in the ditch catchment. N output ranged from 26 to 232 kg ha⁻¹ year⁻¹ and from 99 to 163 kg ha⁻¹ year⁻¹ for the drainage plot and the ditch catchment, respectively. These values align with typical conventional farming practices in northeastern Germany. N soil surface budgets (Equation 2) in our study area mostly exhibited positive values due to higher input than export through harvest (Figure 8 and Table 1). Notably, the ditch catchment exhibited less variability in N input, output, and soil surface budgets, benefiting from the inclusion of four fields with different cultivation practices instead of one single field (drainage plot). When the cumulative N soil surface budgets and NO₃⁻-N losses via the collector drain and the ditch over the course of the study period are subtracted, a surplus of 295 kg N (drainage plot) and 268 kg N (ditch catchment) is evident. This suggests a long-term accumulation of N in the soils despite the fact that, based on the prevailing soil types and impact of groundwater, ca. 10–30 kg N ha⁻¹ year⁻¹ may have been lost through denitrification (Kreins et al., 2010). NO₃⁻-rich soil water is potentially available for leaching and becomes active when the groundwater table rises. This probably explains the observed seasonal patterns of NO₃⁻-N concentrations and the positive relationships between flow and NO₃⁻-N concentrations in our study catchment. The significance of groundwater head in nitrate export has been highlighted in previous studies (Musolff et al., 2016).

NO₃⁻-N loads were completely independent from the N management on the arable fields and highest loads were observed during the wettest years 2007 and 2011 (Figure 8). In contrast to findings in other studies (Larsson and Jarvis, 1999; Ford et al., 2018; Liu et al., 2022), our results confirm earlier studies of our working group, in which no direct relationships between the N soil surface budget and N export (Bauwe et al., 2020) was identified. Contrarily, NO₃-export in our study catchment and also in other catchments of the Baltic Sea Region (Bechmann et al., 2014) appear to be entirely independent of N fertilizer application rates and is solely driven by hydro-meteorological factors.

The data analysis in this study had two significant findings. Firstly, the previously described seasonal fluctuations in NO₃⁻-N concentrations and discharge had profound implications for NO₃-export. Considering the entire time frame, the majority of NO₃⁻-N has been lost via the collector drain or the ditch during a very short period (Figure 7). Secondly, no direct relationships between N soil surface budget and NO₃-export were discerned at either the drainage plot or the ditch (Figure 8). Although the N soil surface budget generally exhibited a decreasing trend, this reduction did not correlate with changes in NO₃⁻-N concentrations (Figure 2).

These findings have implications for a catchment management aiming at mitigating NO₃⁻-N losses. Efforts should be intensified to further reduce the N surplus. While there is an overarching trend in Germany and other European countries toward gradually decreasing agricultural N surplus (Batool et al., 2022), it is important to recognize that the positive impact on riverine NO₃-export may take many years to materialize. The predominant reason for this prolonged delay, known as the “N legacy effect,” refers to the enduring accumulation of N in the root zone (Van Meter et al., 2016). The long-term N accumulation in the upper soil layers might also explain, why there are no functional seasonal relationships between agricultural activities such as N fertilization and NO₃⁻-N concentrations in our study area. In contrast, there are instances, in which NO₃⁻-transport in the river was affected by seasonal changes of the N soil system budgets (Pinardi et al., 2022). Under climate change conditions, both NO₃⁻-N export mechanisms and biogeochemical cycles may change, which makes predictions regarding NO₃⁻-N losses difficult. On the one hand, expected wetter winters in Central Europe (Frei et al., 2006) will probably lead to increased NO₃⁻-N loads because of elevated discharge in winter. On the other hand, rising temperatures will presumably result in higher denitrification rates (de Klein et al., 2017), which will might have a positive effect on NO₃⁻-N concentrations and loads in surface waters. For example, a downward trend in N export during the last three decades was observed for the Po River in Italy, and it was associated with an upward trend in water temperature along with enhanced denitrification rates (Soana et al., 2024).

To expedite the transition to a sustainable N management in agricultural catchments, technical measures should be implemented. Given that artificial drainage systems exert significant control over the hydrology and hydrochemistry of larger scales (Tiemeyer et al., 2006), reduction measures should primarily target drainage plot outlets. Effective strategies may include the establishment of constructed wetlands (Kovacic et al., 2000; Crumpton et al., 2020; Mitchell et al., 2023) or the implementation of reactive barriers (Christianson et al., 2012, 2021; Mitchell et al., 2023). Both approaches also have the potential to mitigate high flows, thereby positively influencing NO₃⁻-N concentrations during discharge events. However, in the scientific literature it is also argued that conservation practices (less fertilization, crop rotation etc.) possibly lead to faster success than controlled drainage, bioreactors, or saturated buffers (Liu et al., 2022).

Because of remaining uncertainties, probably the best approach for a sustainable N management in agricultural catchments is the implementation and strengthening of conservation practices together with technical solutions in hot spot areas.