Research on ecosystem services has been prospering in recent years. The fast evolution in this field of science results in divergences with regard to definitions and terminologies. Ecosystem services have been defined by de Groot et al. (2010) as direct and indirect contributions to human well-being that originate from ecosystems. Burkhard et al. (2012b) (p. 2) defined ecosystem services as “the contributions of ecosystem structure and function - in combination with other inputs - to human well-being”. Fig. 1 depicts the interrelations and dependencies between the ecosystem and human benefits through ecosystem services within a defined system. The supply of ecosystem services is strongly dependent on ecosystem properties and conditions (Fig. 1). Ecosystem service supply can be specified as potentials (or capacities) and flows (Syrbe et al. 2017, Burkhard et al. 2014, Schröter et al. 2014; see Table 1).

Figure 1.  

Ecosystem service mapping framework (based on Syrbe et al. 2017; aspects presented in italic were assessed in this study).

The beneficiaries of the ecosystem service nutrient regulation are diverse. At first sight, it is obvious that society as such is a beneficiary as it strives for a clean environment (Villamagna et al. 2013). With regard to directives and regulations arising from national as well as European legislation, politics can also be defined as a beneficiary for the assessed ecosystem service, which can eventually also be allocated to society as such. Following the approach of Power (2010), agriculture as such simultaneously provides and consumes ecosystem services. In order to provide high yields amongst others, the agricultural system, also known as the agroecosystem, requires the ecosystem service nutrient regulation (Burkhard et al. 2012a, Power 2010). These different applicable perspectives on the beneficiaries of the ecosystem service nutrient regulation emphasise the importance of the service and, as a consequence, underline the demand for its evaluation.

The following section describes the two study areas: the federal state of Schleswig-Holstein and the Bornhöved Lakes District, both located in northern Germany. We selected these two study areas of different spatial extent in order to allow for some comparisons with regard to the issue of scale. In the first step, we assessed both case study areas with the exact same data and spatial resolution. Thus, in that case, the results of the Bornhöved Lakes District are simply an extract of the results for the federal state of Schleswig-Holstein. In the second step, we used a higher spatial resolution to assess the situation in the Bornhöved Lakes District.

The assessment of the potential supply of the ecosystem service nutrient regulation is based on the ecosystem service matrix by Burkhard et al. (2014). The matrix can be used to distinguish nutrient regulation for different land cover types. CORINE land cover data*7 from the European Union was selected as the underlying dataset. The CORINE dataset is hierarchically organised into three levels differentiating in total between 44 land cover classes (EEA 1995). In this study, the 2012 CORINE dataset in ArcGIS polygon shape format was applied. Fig. 4 presents the spatial distribution of the CORINE land cover classes in the two study areas. For both study areas, Schleswig-Holstein and the Bornhöved Lakes District, non-irrigated arable land and pastures are the dominant land use/land cover (LULC) classes (Fig. 4). Hotspots of continuous urban fabric, discontinuous urban fabric and industry or commercial units are agglomerated around the four urban districts of Lübeck, Kiel, Flensburg and Neumünster and in the area adjacent to Hamburg in the southwest of the study area. More natural LULC classes are distributed rather randomly in both areas.

Figure 4.  

Distribution of CORINE land cover classes in 2012 in Schleswig-Holstein (left) and the Bornhöved Lakes District (right; data from GeoBasis-DE / BKG 2017a).

CORINE land cover classes were chosen as geospatial units to be placed in the rows (y-axis) of the ecosystem services matrix (Burkhard et al. 2014). For each land cover type occurring in the study area, the corresponding potential for nutrient regulation, derived from an expert-based assessment, was specified on a relative scale from zero to five by Burkhard et al. (2012a), where five stands for the highest potential. The expert-based assessment was not executed specifically for this study but the potentials have been adopted from the study by Burkhard et al. (2012a). By means of ArcMap 10.3, the values assigned to the CORINE land cover classes were presented in ecosystem service maps for the study areas.

In Schleswig-Holstein, the anthropogenic transformation of the nutrient cycle primarily leads to nutrient oversupply, as excess nutrients are introduced into the environment by means of agricultural practices (Trepel 2016, Kreins and Henseler 2015, Taube et al. 2015, Larsson and Granstedt 2010). At these locations, there is a significant demand for the ecosystem service nutrient regulation. In order to obtain data on the nutrient surplus, we calculated nutrient budgets. Data from the Landwirtschaftszählung 2010 (Statistikamt Nord 2010) *8 served as the basis for the budget calculations. We included the following parameters in the budgets: Organic fertiliser (from livestock) and mineral fertiliser with corresponding losses, compost, biological fixation, nitrogen deposition (UBA 2015), yield, digestate from biogas plants and sewage sludge. An overview on inputs and outputs and on the corresponding methods used for calculating the different parameters is presented in Table 2.

Due to the data privacy law, data for some municipalities was not available on the scale of municipalities within the Landwirtschaftszählung 2010 (Statistikamt Nord 2010). In order to avoid such data gaps, the available data on the scale of municipalities was compared to the data on the scale of counties. The calculated differences between the two datasets were allocated to the affected municipalities. The relative extent of the agricultural area of the corresponding municipalities was used as a factor for the allocations.

With regard to the issue of scale, we used different spatial references for the nutrient budgets in the two case study areas. For the federal state of Schleswig-Holstein, we allocated the nutrient budgets solely to the corresponding municipalities. Thus, the resolution of the result corresponds to the resolution of the input data, the Landwirtschaftszählung (Statistikamt Nord 2010). In order to allow for comparability, we used the same approach in the Bornhöved Lakes District. However, in addition to that, aiming to assess the smaller case study area in more detail, we allocated the nutrient budgets to the respective areas within each municipality.

In order to assess the potential ecosystem service supply and demand, two more important datasets were collected: groundwater nitrate concentration and the nitrate leaching potential. After nitrogen is introduced into the environment, the potential for the ecosystem to regulate nitrogen inputs influences the quantity of nitrogen available for nitrate leaching. However, the actual nitrate leaching is also dependent on other factors. The Landesamt für Landwirtschaft, Umwelt und ländliche Räume (LLUR 2011a) assessed the nitrate leaching potential in Schleswig-Holstein. According to the LLUR (2011a), the following factors are relevant for computation: soil type and texture, soil horizon classification, humus content, state of decomposition, state of consolidation, substance volume, leak water rate, mean capillary rise, precipitation and the climatic water balance. The data were classified into categories from one to five, whereby five represents the highest nitrate leaching potential. Data on the measured nitrate concentration in the groundwater bodies in Schleswig-Holstein were downloaded from the online portal “Wasserkörper- und Nährstoffinformationssystem Schleswig-Holstein” (MELUR 2016) for the year 2013. In order to obtain continuous spatial data, the measured values were interpolated using the IDW interpolation method (software: ArcMap 10.3).

The measured groundwater nitrate concentration was defined as the focal indicator for the nitrogen status in the study area. The influence of the ecosystem service nutrient regulation, the nitrate leaching potential and the actual nitrogen budget on the nitrogen status in the study area was tested. In order to analyse this relationship statistically, the correlations between the parameters were calculated for Schleswig-Holstein and the Bornhöved Lakes District. In order to assess the issue of scale and spatial resolution, in total five different datasets (Table 3) were statistically analysed.

A rank-order correlation analysis was executed, whereby the parameters groundwater nitrate concentration, nitrogen budget, nitrate leaching potential and nutrient regulation potential were incorporated. According to literature, the following correlations were expected: nitrogen surpluses correlate positively to the groundwater nitrate concentration (Wick et al. 2012, Power 2010, Rumohr et al. 1996, European Council 1991, Tivy 1987). In addition to that and in consistency with the assumptions and definitions given in this study, a negative correlation was expected between the nutrient regulation potential and the groundwater nitrate concentration (Gaines and Gaines 1994). As explained earlier, the LLUR assessed the nitrate leaching potential incorporating several site-specific characteristics. Thus, we expect that a positive correlation exists between the nitrate leaching potential and the groundwater nitrate concentration. As a consequence, a negative correlation presumably will arise between the nitrate leaching potential and the nutrient regulation potential. The correlation between the nitrogen budget and the nutrient regulation potential allowed us to draw conclusions on the coincidence of ecosystem service potential supply and demand.

First, the statistical analysis was executed for average values from each municipality. Second, the same analysis was executed using an original values excerpt via a point grid (approximately 1 kilometer spacing). A spatially more explicit analysis was executed for the Bornhöved Lakes District case study area. Here, the nitrogen budget was allocated to the corresponding CORINE land cover classes, thereby generating a spatially more explicit demand for nutrient regulation. In addition to the point grid, an additional finer point grid was created with approximately 250 m spacing (Fig. 5). The rank-order correlation analysis was repeated for these datasets.

Figure 5.  

Assessment units for Schleswig-Holstein and the Bornhöved Lakes District: Large and fine point grid (data on administrative boundaries from GeoBasis-DE / BKG 2017b).

Fig. 6 gives a schematic overview on the workflow to compute potential supply and demand for the ecosystem service nutrient regulation and on the statistical analyses.

Figure 6.  

Overview of the methodological procedure for the assessments of the ecosystem service nutrient regulation and the nutrient status on the two scales Schleswig-Holstein and Bornhöved Lakes District (CORINE = Coordination of Information on the Environment*24).

The potential supply of the nutrient-regulating ecosystem service was evaluated and mapped for Schleswig-Holstein and the Bornhöved Lakes District using the ecosystem service matrix (Burkhard et al. 2014) based on CORINE land cover types (Fig. 4). The qualitative values on nutrient regulation for the matrix assessment have been adopted from Burkhard et al. (2014) (p. 15). In Table 4, the relevant values are shown. Fig. 7 shows the spatial distribution of the ecosystem service nutrient regulation potential supply.

Figure 7.  

Nutrient regulation ecosystem service potential supply in 2012 in Schleswig-Holstein (left) and the Bornhöved Lakes District (right) (application of ecosystem service potential matrix by Burkhard et al. 2014; CORINE Land Cover data from GeoBasis-DE / BKG 2017a).

The largest part of the study area is characterised by low nutrient-regulating potentials (Fig. 7). High regulatory potentials are distributed rather patchily over Schleswig-Holstein. However, areas with high potentials are agglomerated in the south and southeast of the study area (Fig. 7). Consistent with the distribution of the nutrient regulation potential in Schleswig-Holstein, the largest part is also featured with rather low nutrient regulation potentials, with similar potentials also being evident in the Bornhöved Lakes District. Medium and high potentials can be found mainly in the central part of the area. There are also areas with no relevant nutrient regulation potential. These are distributed randomly throughout both study areas.

The largest oversupply of nitrogen can be found in municipalities located in the northwest of the study area (Fig. 8). An agglomeration of municipalities with relatively low nitrogen budgets (lower than 40 kilograms of nitrogen per hectare per year) is located in the southeast of the study area. With regard to the regional landscapes, municipalities in the Geest show a tendency for larger nitrogen surpluses. There, the nitrogen budget increases from south to north. In the Bornhöved Lakes District, the majority of municipalities have a nitrogen budget lower than 40 kilograms per hectare per year. Municipalities with higher nitrogen budgets can be found in the west of the study area. The municipality with the highest nitrogen surplus is Ruhwinkel. Compared to the adjacent municipalities, Ruhwinkel is featured by relatively high nitrogen inputs through the application of digestate and organic fertilisers from livestock.

Figure 8.  

Total nitrogen budget in 2010 in Schleswig-Holstein Holstein (left) and the Bornhöved Lakes District (right) on the municipality scale (data on agricultural census 2010 from Statistikamt Nord 2010*25; data on nitrogen deposition 2009 from UBA 2015; data on administrative boundaries from GeoBasis-DE / BKG 2017b).

For the Bornhöved Lakes District, the nitrogen budget was mapped in a spatially more explicit manner (Fig. 9). In the first step, the nitrogen budget was mapped excluding nitrogen deposition. Furthermore, the remaining nitrogen budget was not assigned to the corresponding municipalities, but assigned to the CORINE LULC classes “land principally occupied by agriculture” (243), “non-irrigated arable land” (211) and “pastures” (231) in the corresponding municipalities. Thus, areas with CORINE classes, other than these, have a nitrogen budget equal to zero. High nitrogen budgets (61 - 80 kg N/ha*year) can be found in the west of the Bornhöved Lakes District. Medium nitrogen budgets are located in the south; the remaining area shows low nitrogen budgets. In the next step, atmospheric nitrogen deposition was mapped separately. The greatest share of the area is characterised by nitrogen deposition rates between 16 and 18 kilogram nitrogen per hectare per year. Higher and lower nitrogen deposition values are distributed rather randomly throughout the study area. In the third and last step, both values were combined in order to obtain the total nitrogen budget. The largest share of the area shows medium nitrogen budgets (41 – 60 kg N/ha*year). Very low and low budgets are distributed randomly throughout the Bornhöved Lakes District and can be found along the eastern border of the study area, respectively. High nitrogen budgets are located in the southwest and southeast. The relatively high nitrogen inputs on the agricultural grounds that belong to the municipality of Ruhwinkel can still be identified.

Figure 9.  

Nitrogen budgets in the Bornhöved Lakes District: (a) nitrogen budget 2010 (excl. nitrogen deposition) for each municipality allocated to arable land, (b) atmospheric nitrogen deposition in 2009, (c) total nitrogen budget 2009/2010 (data on agricultural census 2010 from Statistikamt Nord 2010*26; data on nitrogen deposition 2009 from UBA 2015).

Consistent with Taube et al. (2015) and the Nitrate Directive (European Council 1991), the groundwater nitrate concentration was selected as the focal indicator for the nitrogen status of the environment. It can be assumed that responses to changing parameters (i.e. nitrogen budget) can be detected in the measured nitrate concentration after a specific time lag (Rumohr et al. 1996). In order to take that time lag into account and remain in accordance with recent publications on the nutrient status of Schleswig-Holstein (e.g. Taube et al. 2015), data from 2013 was selected.

In 2013, the groundwater nitrate concentration was distributed rather heterogeneously (Fig. 10). Lowest concentrations (<10 mg/l) were found in the northeast (from Flensburg to Eckernförde) of Schleswig-Holstein and along the central part of Schleswig-Holstein’s North Sea coastline. Hotspots of nitrate concentrations were distributed rather patchily throughout the federal state, mostly in the hinterland. The largest area with relatively high nitrate concentrations (>50 mg/l) was located in the east of Schleswig-Holstein. In the remaining part, medium concentrations (10-50 mg/l) were mostly dominant. The Bornhöved Lakes District is located at the western boundary of the largest hotspot. Thus, high nitrate concentrations (>50 mg/l) were found in approximately half of the case study area. From southeast to northwest, the nitrate concentrations decreased. However, only a very small part of the Bornhöved Lakes District showed low nitrate concentrations (<10 mg/l).

Figure 10.  

Interpolated groundwater nitrate concentration in Schleswig-Holstein (left) and the Bornhöved Lakes District (right) in 2013 (data from Wasserkörper- und Nährstoffinformationssystem Schleswig-Holstein from the MELUR 2016).

Low nitrate leaching potentials can be found along both coastlines of Schleswig-Holstein (Fig. 11). Lowest potentials agglomerate at the central part of the Baltic Sea coastline. High potentials are located in the hinterland, increasing from south to north. Thus, the area with high potentials coincides with the spatial extent of the Geest. In accordance with this pattern, in the Bornhöved Lakes District, the potential decreases from south to north (Fig. 11).

Figure 11.  

Nitrate leaching potentials in Schleswig-Holstein (left) and the Bornhöved Lakes District (right) (data from LLUR 2011b).

The potential supply of nutrient regulation seems to be distributed randomly throughout the study areas Schleswig-Holstein and Bornhöved Lakes District. Small- and medium-sized patches of high nutrient regulation potentials are scattered throughout a landscape dominated by low nutrient regulation potentials. The biogeochemical properties of different nutrients vary and, as a result, the corresponding processes in the environment also change. For reasons of simplification, we analysed the ecosystem service nutrient regulation by using the nutrient nitrogen as an example. Here, one must note that the potential supply values for nutrient regulation which have been adopted from the expert-based assessment by Burkhard et al. (2014), do not differentiate between different nutrients. The potential supply is thus not specifically harmonised for nitrogen but depicts a more general trend. Furthermore, the spatial distribution modelling was solely dependent on the primary vegetation and land use pattern, because the potential supply of nutrient regulation was mapped qualitatively based on the matrix values provided by Burkhard et al. (2014) and CORINE land cover classes. As nutrient regulation is dependent on further parameters such as soil type, texture and moisture, it is questionable whether the allocation of nutrient regulation potentials, based solely on land cover classes, is sufficient in order to characterise local conditions and nutrient regulation is dependent. However, this approach is an efficient way to gain an overview of the investigated study areas. The quantification of the demand for nutrient regulation on the municipality scale supports the assumption by Taube et al. (2015) on the regional differentiation of the nitrogen budget in Schleswig-Holstein. A tendency for higher nitrogen budgets in the west of the study area was identified, which partly confirms the nitrogen budgets published by Taube et al. (2015) (p. 19). In addition, the relative distribution of the nitrogen budget, to a large extent, conforms to the nitrogen budget calculated on the scale of landscape forms, published by Taube et al. (2015) (p. 20). In the Bornhöved Lakes District on the municipality scale, high nitrogen budgets were found in the west. The greatest share of the area had relatively low nitrogen budgets (<40 kg N/ha*year). However, the spatially more explicit nitrogen budgets revealed a more heterogeneous picture. Low nitrogen budgets (<40 kg N/ha*year) could only be found along the eastern border of the study area and in small fragments throughout the Bornhöved Lakes District. High nitrogen budgets (>60 kg N/ha*year) could still be found in the west but also in the southeast of the study area. The remaining area showed medium nitrogen budgets (41-60 kg N/ha*year).

With respect to nitrogen budgets, the following aspects need to be kept in mind: The computation of the nitrogen budget is highly dependent on input data. Most parameters were based on the data of the Landwirtschaftszählung 2010 (Statistikamt Nord 2010*27). Due to data privacy, the agricultural census data on the scale of municipalities was not complete, as some data had been censored. The data had been corrected on the municipality scale in order to obtain more realistic results. Still, the correction factor for the municipalities is solely based on the extent of the agricultural area in the corresponding municipality. This extrapolation may downgrade the obtained results. The parameters relevant for the nitrogen budget could be calculated by employing different calculation techniques and methodologies. Thus, the nitrogen budgets are influenced by the selected approach. For instance in this study, nitrogen inputs from organic fertilisers (livestock) were corrected by stable, storage and application losses, as the DüV (2007) allows the input values to be corrected by these parameters, even thoughthese emissions will also end up in a waterbody, eventually. The same is true for the deduction of nitrogen losses through the application of mineral fertilisers.

The methodology employed for the calculation of mineral fertiliser inputs was adopted from Taube et al. (2015) (p. 14) which is based upon Bach et al. (2014) (p. 18) and consists of a calculation which estimates mineral fertiliser quantities in relation to the quantities of organic fertiliser, digestate, biological fixation and yield for the area of Schleswig-Holstein in 2010. The approach by Taube et al. (2015) is based on data from 416 model farms (adapted from 14,700 real farms – InVeKoS*28 data, https://www.zi-daten.de/). Taube et al. (2015) claim to provide a more realistic nitrogen budget. However, their input data and model were not available for this research. Thus, this methodology could not be adopted for comparison.

One must also keep in mind that, for the calculation of the nitrogen input through biological nitrogen fixation, the nitrogen fixed by legume cultivation was neglected because the relevant data was not available on the municipality scale but only on the county scale and only added up to 0.2 kilograms of nitrogen per hectare in 2010. Another point of potential criticism may arise from the fact that the analysis of this study did not consider seasonal variability. Rumohr et al. (1996) report, amongst others, on the seasonal variability of groundwater nitrate concentrations in the Bornhöved Lakes District. The variability can primarily be attributed to the diverging nitrogen assimilation through plants throughout the year (Rumohr et al. 1996). None of this information could be checked by this study.

The groundwater nitrate concentrations, which served as indicators for the nutrient status, were found to be distributed heterogeneously. The distribution partly followed a trend expected from the distribution of the nitrate leaching potential and the computed nutrient budget. In the north and west of the federal state, the occurrence of high groundwater nitrate concentration was roughly located in areas with high nutrient budgets and high nitrate leaching potentials. These conditions were particularly found in the area of the Geest. Conforming to the poor sandy soils of the Geest area, the nitrate leaching potentials were relatively high. The same co-occurrence was found for low concentrations and potentials.

In the eastern part of Schleswig-Holstein, the groundwater nitrate concentration cannot be easily explained by the distribution of the nitrogen budget and nitrate leaching potential. Even though, the area is characterised by low to medium nitrate-leaching potentials as well as relatively low nutrient budgets, a large area is featured by relatively high groundwater nitrate concentrations. As the Bornhöved Lakes District is located at the western boundary of the hotspot, high groundwater nitrate concentrations (>50 mg/l) were found in approximately half of the case study area, declining from southeast to northwest. This trend follows the distribution of the nitrate leaching potentials. In order to find out more about the interrelations of the datasets, including the potential supply of the ecosystem service nutrient regulation, a statistical evaluation was performed in the form of a rank-order correlation analysis. The parameters, nitrogen budget, nutrient regulation potential, nitrate leaching potential and groundwater nitrate concentration, were correlated with each other for all different datasets (Fig. 12).

The correlation analysis revealed a positive correlation between the nitrate leaching potential and the groundwater nitrate concentration. This correlation was expected. For the Bornhöved Lakes District, this correlation was stronger than for the datasets covering the whole federal state. The correlation between nutrient regulation potential and the groundwater nitrate concentration only fulfilled the (expected) negative correlation for the Bornhöved Lakes District. As a consequence, the same was true for the correlation between the nitrate leaching potential and the nutrient regulation potential. It is noticeable that the spatially more explicit datasets indicated a negative correlation between the nitrogen budget and the nutrient regulation potential. This correlation implies that there is a spatial mismatch between the ecosystem service potential supply and demand in the study area. The more aggregated nitrogen budget datasets suppress this condition. However, the statistical analysis could not prove the expected positive influence of the nitrogen budget on the groundwater nitrate concentration. This circumstance provokes the question whether thegroundwater nitrate concentration dataset is suitable as a nitrogen status indicator as the opposite effect could be expected from existing knowledge (Wick et al. 2012).

Previous publications (Taube et al. 2015, Rumohr et al. 1996) and Directives have specified groundwater nitrate concentration as an indicator for the nitrogen status of an area. On this account, the same indicator was selected for this research. Official data on groundwater nitrate concentrations at the measuring stations was provided by the MELUR (2016) in the online portal Wasserkörper- und Nährstoffinformationssystem Schleswig-Holstein. Thus, the same data on nitrate concentrations in Schleswig-Holstein’s groundwater bodies as in previous studies were used in this study. The data available from the measuring stations was interpolated (employed interpolation method: IDW) for the whole study area. The fact that this indicator is not based on a spatially continuous original dataset but only on approximately 230 values (measuring stations) for the whole area of Schleswig-Holstein has to be considered a limitation. This limitation became even more evident for the Bornhöved Lakes District, because only one original data point is located within the study area (Fig. 10). The surrounding area of the Bornhöved Lakes District with high groundwater nitrate concentrations is also featured by only a few measuring stations. This circumstance decreased the certainty and the accuracy of the extent of the groundwater nitrate concentration hotspots and may lead to incorrect correlations with regard to the nutrient status. Rumohr et al. (1996) measured groundwater nitrate concentrations in the Bornhöved Lakes District in the first half of the 1990s at eight measuring stations. They discovered that the groundwater nitrate concentrations on agricultural grounds were higher compared to other ecosystems. It can be assumed that the results of the statistical analysis with such accurate data on the current groundwater nitrate concentrations would correspond better with the expected values. Besides, the diverse measurement and quantification units of the different parameters also decrease the explanatory power of the statistical analysis. In particular, the comparison between physical entities (i.e. kilogram nitrogen per hectare per year) and the specification of nutrient regulation potential and nitrate leaching potential on a relative scale from zero and one to five, can lead to decreasing quality of the outcomes.

Still, by evaluating the results of the statistical analysis, it was possible to rank the relative validity of the results according to the spatial resolution and scale/study area (Fig. 13). For that matter, the validity of the outcomes was defined as the degree of conformity of the results with the expectations gained from the literature review. The results on the smallest scale with the finest grid and the highest spatial differentiation were found to deliver the most valid results. Whereas the computation of average values on the municipality scale led to imprecise or even incorrect data. The fact that municipalities are administrative with no geological or natural units most likely emphasises the inexpediency of the averaged values.

Figure 13.  

Relative dependency between validity of results, scale and spatial resolution.

The results obtained by this research contribute to the recently ongoing debate on the topic of nutrient management. On the one hand, the research supports the suggested regional differentiation (Taube et al. 2015). On the other hand, the methodological uncertainties detected confirm the necessity for strict regulations with regard to the application as well as declaration of all kinds of fertilisers on farmland. In order to determine the value of this study for ecosystem service research, one needs to highlight the general picture rather than the explicit results obtained from this research. Contrary to the theoretical evaluation of the ecosystem service concept, the integration of ecosystem services in applied sectors such as environmental management features challenges (de Groot et al. 2010). Egoh et al. (2007) state, amongst others, that the integration of ecosystem services into different research areas lags behind. Another gap exists between the theory and practice with regard to the quantification of ecosystem services (Crossman et al. 2013). Hou et al. (2013) go even further by pointing out uncertainties evolving from ecosystem services research on the landscape scale. Amongst others, uncertainties originate from respondents’ preferences (in expert-based approaches), initial data uncertainty, technical problems, methodological uncertainties and complexities of natural systems (Hou et al. 2013).

Burkhard et al. (2012a) executed research with regard to mapping of ecosystem service supply, demand and budgets. They state that the “[…] supply of multiple goods and services by nature should match the demands of society, if self-sustaining human-environmental systems and a sustainable utilization of natural capital are to be achieved” (Burkhard et al. 2012a, p. 1). According to them, suitable indicators are required and diverse datasets need be integrated for assessments of both supply of and demand for ecosystem services on different temporal and spatial scales. Methods for ecosystem service quantifications are diverse, from mapping, modelling and field measurements to expert opinion (see ESMERALDA*29 project). According to Seppelt et al. (2011), the methodologies should be weighted differently according to the reliability of the data and results.

As specified in Chapter 2, ecosystem services are defined as the benefit human beings experience from the environment. The definition implies the existence of a demand by human society (Burkhard et al. 2012a). This study deals, to a great extent, with the identification of the demand for the ecosystem service nutrient regulation. Through the calculation of nutrient budgets in the study areas, nutrient surpluses were identified. As eutrophication has been recognised as a threat to society in research as well as in decision-making, the localisation of hotspots emphasises the demand for the ecosystem service nutrient regulation in the study areas. The practical implementation through this study confirms the constraints identified in current research (Schulp et al. 2014, Hou et al. 2013). Constraints arise from the usage of different methodologies and datasets for the computation of supply and demand of/for the ecosystem service as well as from the comparison of a rather abstract concept to specific data on nitrogen surplus. The results from the statistical analysis support this notion. In order to combat this issue, more research is necessary assessing the practical implementation of ecosystem services and, as specified by Hou et al. (2013), on ecosystem service assessments on the landscape scale in general.

In the meantime, it would be interesting to undertake the calculation of mineral fertiliser according to Taube et al. (2015) on the municipality scale. Most likely, higher and more realistic mineral fertiliser quantities will emerge. More accurate information on fertiliser usage will increase the validity of the results and thereby enhance the usability of resultsfor decision-making and sustainable land management. For future works, it is also of interest to assess the nutrient situation detached from administrative units, e.g. subdividing the study areas into catchment areas rather than municipalities. The outcome of the correlation analysis supports the usefulness of the transformation from administrative boundaries to natural boundaries. Besides, it would be interesting to assess the potential supply of nutrient regulation in more depth. A quantification of the potential nutrient regulation, incorporating side-specific characteristics such as data on soil types and textures, is an interesting approach, worthy of pursuing.