Translating the agricultural N surplus hazard into groundwater pollution risk: Implications for effectiveness of mitigation measures in nitrate vulnerable zones
Introduction
In the Mediterranean region, water is limited in quantity and quality. Climate change may aggravate this problem due to the combined effect of reduced recharge of aquifers, increase in crop water requirements and rising sea levels (Da Cunha et al., 2007; Kilsby et al., 2007; Kovats et al., 2014; Stigter et al., 2014). Hence, the conservation of aquifers as water reservoirs constitutes a major environmental challenge (Arauzo and Bastida, 2015). In this context, nitrogen (N) pollution in water has been a major issue, as it produces negative impacts on human health and biodiversity (Sutton et al., 2011; Erisman et al., 2013).
Diffuse sources of N associated with agricultural activity have been considered the major cause of elevated N in groundwater in the European Union (EU) (Sutton et al., 2011). This is particularly true in Nitrate Vulnerable Zones (NVZs) which are designated as areas of land draining into ground and surface waters vulnerable to pollution from nitrogen compounds from agricultural sources. In the case of groundwater this means nitrate levels exceeding (or likely to exceed) 50 mg L− 1 (maximum acceptable value, MAV). In the NVZs the continuation of agricultural activities for food production purposes, depends upon the capacity to bring the groundwater quality back to good status.
Mitigation programmes and Directives have been implemented across the EU to reduce nitrogen loads to water bodies. The European Community (EC) Nitrates Directive (ND) (Council Directive 1991/676/EEC) aims to reduce water pollution through nitrates from agricultural sources. The ND requires the Member States to establish Codes of Good Agricultural Practices (CGAP) to be implemented by the farmers on a voluntary basis, and National Action Programs (AP) with mandatory measures for the identified vulnerable zones (Cameira et al., 2019). However, reports by EU Member States regarding the status of their water bodies show that the effects of these measures remain insufficient at the European scale (Bouraoui et al., 2011). Thus, it is essential to develop approaches to evaluate the effectiveness of the ND related measures/practices upon the water bodies. Such approaches must also allow the identification of areas with critical sources of diffuse pollution.
Nutrient balances, in particular the gross N balance (GNB) or N surplus, have been used extensively as a proxy for agriculture environmental pressure because of their simplicity at local, regional, and national scales (Bouraoui and Grizzetti, 2011; van Grinsven et al., 2012; Lassaletta et al., 2012; Cameira et al., 2019; Serra et al., 2019). Hansen et al. (2012) found a correlation between N surpluses and nitrate concentrations in groundwater for Denmark. However, large N surpluses do not always coincide with high nutrient losses to the environment (Grizzetti et al., 2008; Baily et al., 2011) particularly in arid climates were recharge rates are low and also when the unsaturated zones are thick (Hagedorn et al., 2018). Thus, the N balance can only indicate the potential hazard. As a next step, this hazard needs to be converted into pressure on the environment and translated into a groundwater pollution risk, by considering the main factors involved in the transport from the bottom of the root zone to the groundwater surface.
The use of experimental methods to quantify actual N losses to the water bodies is limited because routine application of such labour-intensive methods is mostly not viable. Furthermore, measurements are often only made after management decisions have already been taken (i.e., too late). Also, experimental data used for the calculation of N fluxes, e.g. N concentrations in soils, are often not generalizable, due to inter-annual variability in weather patterns, management practices, fertilizer application rates, etc. Alternatively, with more or less complex physically based N transport models, it is possible to quantify N losses for various environmental conditions and management practices (e.g. Cameira et al., 2014; Molina-Herrera et al., 2016; Kasper et al., 2019). Some process-based models can be used in a Geographical Information System (GIS) to predict the temporal and spatial distribution of nitrates in groundwater. It is the case of DAISY-MIKE SHE (Refsgaard et al., 1999), NLEAP-GIS (Li et al., 2020), RZWQM2 (Ahuja et al., 2000), and SWAT (Arnold et al., 2012) models, among others. However, such models require detailed input data, contain many weakly constrained parameters and are often difficult to operate. Furthermore, some models, particularly those to be applied at large river basins are usually set up without including the current management practices (Malagó et al., 2017) which influences considerably N fluxes and storage. Instead, simplified non-process models have been developed for indicative N loss assessment and to identify critical source areas, requiring fewer and more accessible input data (Buczko and Kuchenbuch, 2010). A group of these simplified non-process models is based on assigning ratings to various physical attributes and are called index methods. However, most of them concentrate on groundwater vulnerability assessment rather than on the pollution hazard and groundwater pollution risk. Vulnerability merely indicates whether the characteristics of the subsurface prevent or favour the transport of pollutants into groundwater, without taking into account the actual pollutant loading, which represents the hazard. Thus, these models can indicate high vulnerability but no pollution hazard given the absence of a pollution load. It is the case of SINTACS (Civita and De Maio, 1997) and DRASTIC (Aller et al., 1985; Leone at al., 2009; Kazakis and Voudris, 2015; Meng et al., 2020), among others. Initially this type of models dealt only with groundwater intrinsic vulnerability, which is brought upon by the natural, hydrogeological factors of an aquifer (Vrba and Zoporozec, 1994; van Beynen et al., 2012). Revised versions of the models incorporated few aspects of specific vulnerability (natural parameters and human activities) namely information on land cover (Arauzo, 2017; Salman et al., 2019; Vogelbacher et al., 2019) and performed better than the purely intrinsic methods (Stigter et al., 2006). Nevertheless, there is some doubt that the concept of aquifer vulnerability is a valuable tool for groundwater quality protection, due to discrepancies between nitrate pollution maps and vulnerability maps (Stigter et al., 2006; Foster et al., 2007, 2013; Rizeei et al., 2018). Land cover/use alone does not incorporate information on agricultural practices (e.g. fertilization and irrigation). On the other hand, in addition to groundwater vulnerability, the pollution risk also depends on the existence of a significant pollutant loading, which represents the hazard (Uricchio et al., 2004; Kazakis and Voudris, 2015; Pisciotta et al., 2015). According to Kazakis and Voudouris (2015), the groundwater pollution or pollution risk assessment is achieved by overlaying hazard and vulnerability, which is the concept adopted in the present work.
The recently completed H2020 Twinning project NitroPortugal coordinated by the University of Lisbon (grant number 692,331) built on the European Nitrogen Assessment report (Sutton et al., 2011), highlighted the need to review current scientific understanding of nitrogen sources and paths specifically for Portugal and wider Mediterranean systems. Recently, Cameira et al. (2019) showed that measures implemented by Portuguese farmers, within the scope of the EC Nitrates Directive and the national Action Plans, have partially succeeded in reducing N surplus at farm level in a designated Vulnerable Zone. One remaining question is how this N surplus reduction impacted on groundwater. Thus, the specific objectives of the study were to: (i) develop an index based methodology for the evaluation of groundwater pollution risk from the agriculture activity (GRI); (ii) assess the global risk of groundwater pollution by nitrates from agricultural activity in a nitrate vulnerable zone and its evolution after the implementation of the EC nitrates directive; (iii) obtain a better understanding of the spatio-temporal distribution of nitrate concentration in the groundwater over a period of 13 years; (iv) discuss the relation between groundwater pollution risk and nitrate pollution status; (v) identify critical areas and predict the global risk under targeted mitigation scenarios.
Section snippets
Physical features
Currently Portugal has designated nine Nitrate Vulnerable Zones on the mainland. The Tagus Nitrate Vulnerable Zone (TVZ) is the largest one, representing 60 % of the total designated NVZ area of Portugal. The TVZ is located in the Portuguese part of the river Tagus catchment, which is a transboundary river flowing from Spain to Portugal. The climate across the TVZ is Mediterranean (Csa according to the Köppen system) with hot dry summers and mild wet winters. Daily mean and maximum air
N surplus hazard
Fig. 3 shows the spatial distribution of the N hazard index (INH), classified from “very low” (1) to “very high” (5). In the early study period, the hazard was very high (> 100 kg ha−1 yr−1) across most of the TVZ, reducing substantially over time, especially in the northern part where the irrigated crops are the main agricultural activity. In these areas the hazard is classified as moderate (25 < N surplus < 50 kg ha−1 yr−1) in 2016. This decreasing trend is partly associated with the NVZ
Conclusions
An index-based methodology is presented to evaluate the impact of the EU Nitrates Directive (ND) and the Nacional Action Plans (AP) measures upon groundwater pollution with agricultural nitrates. Its originality lies in the fact that it incorporates two information layers defined with detailed municipality data: the N surplus, used to classify the N loading hazard and the water surplus were irrigation has an important role in the calculation of the aquifer recharges in Mediterranean zones. Its
Declaration of Competing Interest
The authors report no declarations of interest.
Acknowledgments
The authors acknowledge NitroPortugal, H2020-TWINN-2015, EU coordination and support action n. 692331 for funding. João Rolim was funded by FCT through the researcher contract DL 57/2016/CP1382/CT0021. LEAF (UID/AGR/04129/2019), CEF(UID/AGR/00239/2013) and CMAFcIO (UID/MAT/04561/2013) are research units funded by Fundação para a Ciência e a Tecnologia I.P. (FCT), Portugal.
References (93)
Vulnerability of groundwater resources to nitrate pollution: a simple and effective procedure for delimiting Nitrate Vulnerable Zones
Sci. Total Environ.
(2017)- et al.
Drainage and N-leaching in alluvial soils under agricultural land uses: implications for the implementation of the EU Nitrates Directive
Agric. Ecosyst. Environ.
(2013) - et al.
Spatio-temporal dynamics of nitrogen in river-alluvial aquifer systems affected by diffuse pollution from agricultural sources: implications for the implementation of the Nitrates Directive
J. Hydrol. (Amst)
(2011) - et al.
Spatial and temporal variations in groundwater nitrate at an intensive dairy farm in south-east Ireland: insights from stable isotope data
Agric. Ecosyst. Environ.
(2011) - et al.
Regression model for aquifer vulnerability assessment of nitrate pollution in the Osona región (NE Spain)
J. Hydrol. (Amst)
(2013) - et al.
Sustainability and environmental assessment of fertigation in an intensive olive grove under Mediterranean conditions
Agr. Water Manage.
(2014) - et al.
Spatial distribution and uncertainties of nitrogen budgets for agriculture in the Tagus river basin in Portugal–Implications for effectiveness of mitigation measures
Land Use Policy
(2019) - et al.
Climate change projections for the Mediterranean region
Global Planet Change
(2008) - et al.
Assessing aquifer vulnerability from lumped parameter modeling of modern water proportions in groundwater mixtures: application to California’s south Coast Range
Sci. Total Environ.
(2018) - et al.
7 Rank-based analyses of linear models
Handbook of statistics
(1997)