Assessment of nitrogen application limits in agro-livestock farming areas using quantile regression between nitrogen loadings and groundwater nitrate levels

https://doi.org/10.1016/j.agee.2019.106660Get rights and content

Highlights

  • An empirical relationship between N inputs and groundwater nitrate levels over S. Korea.

  • Results of quantile regression is useful to suggest optimal N application limits.

  • Nitrate contamination is more sensitive at higher quantiles toward low-lying lands.

  • N application limit below 170 ha−1 year-1 is suggested for typical agricultural areas.

Abstract

The excessive application of nitrogen in agro-livestock farming areas has led to serious groundwater contamination around the world; thus, the efficient control of N loads is crucial to manage nitrate contamination of groundwater. In this study, to suggest an optimal N application limit as a key guideline of N management, we examine the impact of anthropogenic N loading on nitrate levels of shallow groundwater, using a large dataset (n = 4,000) collected in 2012–2014 from 100 agro-livestock farming districts in South Korea. Not considering the time lag and legacy problem, quantile regression is performed to overall assess the relationship between nitrate contamination and anthropogenic N input across the full range of conditional distribution of groundwater nitrate concentrations because of heteroscedasticity. As a result, positive gradients (βτ) meaning the rates of changes are found between groundwater nitrate concentrations and land-derived N loads at all quantiles; βτ increases as the quantile is higher and is as large as 36.48 ± 10.73 mg NO3/L per land-derived N input (on a log scale) at the 90% quantile. The quantile map indicates that the high gradient (i.e., the large sensitivity to N loading) is related to agricultural land use, low elevation, and low topographic slopes. In fact, groundwater nitrate concentrations increase as the percentage of agricultural land increases but the percentage of forest, elevation and slope angle decreases, which suggests that land use, elevation, and slope should be considered when attempting to evaluate N application limits. This study suggests the N limits below 170 kg ha-1 year-1 based on the drinking water standard (44.3 mg/L as NO3) at the quantiles higher than 70% which mostly occur in flat agricultural lowland areas. Our method used in this study can be applied elsewhere to establish efficient management practices for groundwater nitrate contamination, considering local environmental factors.

Introduction

Nitrate (N) pollution in freshwater, including groundwater, is an ongoing problem across the globe. Global N inputs (dominantly as dissolved inorganic nitrogen) to fresh water environment have more than doubled over the last 4–5 decades because of anthropogenic pressure with the expanded production of N fertilizers and fossil fuels (Vitousek et al., 1997; Galloway et al., 2004, 2008; Seitzinger, 2009; Seitzinger et al., 2010). Of the total anthropogenic N input, agricultural N input for crops and livestock occupies approximately 85% of the global total N input and is expected to further increase toward 2050, which indicates that agriculture is the main cause of global N pollution (Galloway et al., 2003). Meanwhile, agricultural activities have been intensified (e.g., concentrated animal feeding operations) due to economic competitiveness and production efficiency, which causes unintended environmental consequences such as eutrophication and water pollution (Tilman, 1999; Tilman et al., 2002). Thus, N management practices, especially in intensive agro-farming areas, have been raised as an important issue for water sustainability. Therefore, better understanding of the connections between nutrient loads including N and water quality is required for water quality management (Johnes, 1996; Alexander et al., 2008; Hansen et al., 2017).

In South Korea, N-use efficiency (i.e., the ratio of N output to N input) in the agricultural sector was reported to be lowest (29.47%) among OECD countries in 2002–2004 (OECD, 2008), and eutrophication and water pollution with nutrients (esp. N) have been often observed. According to the Korea Ministry of Environment (KMOE, 2017), the total number of days when the water quality exceeded a minimum reporting level (i.e., cyanobacteria abundance of 1000 cell/mL) increased from 114 days in 2008 to 608 days in 2015 at 12 weirs of major rivers, and phytoplankton blooms increased during this period. In addition, a number of recent papers reported that the overuse of fertilizers and manure in agricultural areas resulted in acidification and nitrate contamination of shallow groundwater in South Korea (Chae et al., 2004; Choi et al., 2007; Kim et al., 2015a, 2019), and suggested the urgent need of reduced application of fertilizers and manure.

It is a challenge, however, to optimize the application amount of fertilizers and manure for human resources (e.g., food and energy production) while minimizing nitrate pollution (Tilman et al., 2002; Ju et al., 2009; Hansen et al., 2011, 2017). To solve this optimization problem, many researchers and environmental agencies have made an effort: the USDA Natural Resources Conservation Service provides guidelines for addressing nutrient imbalance concerns based on a nutrient management standard (code 590) (NRCM, 2011). In California, N fertilizer guidelines for major crops are available online (https://apps1.cdfa.ca.gov/go/FREPguide) suggesting an environmentally safe and sound N fertilizer application rate. The Nitrate Directive (91/676/EEC) (European Commission, 1991) imposed the maximum application amount of livestock manure for nitrate vulnerable zones (i.e., 170 kg manure-N ha−1  year-1). A few EU countries including Denmark, Germany, and the Netherlands revised the N application rate on the basis of scientific data (Oenema, 2004; Schröder and Neeteson, 2008; D’Haene et al., 2014).

However, most fertilizer response studies have been conducted in a single small watershed and tried to provide a site-specific N fertilizer application limit, while there were only a few regional- or global-scale studies about the relationship between anthropogenic N inputs and nitrogen fluxes (McIsaac et al., 2001; Seitzinger et al., 2010; Sinha et al., 2017). The N fertilizer application limits from small-scale studies may not reflect the response difference due to the spatial change of environmental factors such as soil variability (Kitchen et al., 1995; Rashid and Voroney, 2005). In particular, in many countries including South Korea, which are characterized by diverse geomorphological features, a large spatial variation makes it difficult to apply a single N fertilizer rate to a whole region. Consequently, the application of an N fertilizer rate to a whole region can result in over- or under-fertilization. Thus, a comprehensive guideline is needed to suggest sustainable N application rates, especially in areas with varying environmental conditions (e.g., elevation) like South Korea.

Based on a large dataset collected from a national-scale groundwater quality survey in South Korea, Kim et al. (2019) overall evaluated the status of nitrate contamination and associated hydrochemical changes (esp., acidification) of shallow groundwater. The present study using the same dataset aims (1) to find relationships between N loadings and groundwater nitrate levels in 100 agro-livestock farming districts (section 3),(2) to investigate the relative importance of environmental factors, i.e., land use, elevation, slope angle, soil depth (section 4.1.1.) and manure management (section 4.1.2) in nitrate contamination of groundwater, and (3) to suggest an optimal N limit (section 4.2). We use a regional-scale database for N loading and environmental factors that covers South Korea (Korea Ministry of Environment (KMOE, 2014a). In particular, quantile regression (QR) is adopted to provide a complete view of the relationship between median groundwater nitrate concentrations and land-derived N inputs (on a log scale) for 100 agro-livestock districts by estimating rates of changes, called gradients (βτ) (Koenker and Bassett, 1978; Yu et al., 2003). QR is useful for identifying heterogeneity across the quantiles of the response variable (i.e., groundwater nitrate concentrations in this study) when the response variable is affected by more than one factor or has a large variance, which is common in hydrochemical data. QR has been widely used in econometric (Koenker and Hallock, 2001), ecological (Sankaran et al., 2005), meteorological (Elsner et al., 2008; Hirschi et al., 2011), and hydrological studies (Wasko and Sharma, 2014), while there are only a few research studies related to hydrochemical dataset using QR, including Monteith et al. (2014) which used QR to assess the acidification of surface water.

Section snippets

Study area

For this study following Kim et al. (2019), 100 representative agro-livestock farming districts were evenly chosen in five provinces: Gyeonggi, Gangwon, Chungcheong, Gyeongsang, and Jeolla (n = 20 for each province in Fig. 1b). Each district consisted of two to 33 animal feeding operations (AFOs), and as a result, a total of 613 AFOs were included in the study area (Table 1). According to the KMOE (2014a), 80% of the 613 AFOs collected manure and used it directly on the farm as an

Results

A total of 3,928 groundwater samples are used for data interpretation in this study, with excluding 72 samples whose charge balance errors are larger than ± 10%.

Discussion

Because Fig. 5 indicates higher sensitivity of groundwater to land-derived N loading at higher quantiles, external factors that affect sensitivity (i.e., heteroscedastic data) are discussed in Section 4.1 and an optimal N limit is suggested for each quantile group in Section 4.2.

Conclusion

Groundwater in South Korea is highly contaminated by nitrate due to pervasive livestock agricultural activities. This study investigates the relationship between N loading and nitrate levels in groundwater using quantile regression to deal with the heteroscedastic data. The positive gradients βτ at all quantiles (0.1–0.9) gradually increase as the quantile is higher, which indicates that the groundwater is more sensitive to N loading at high quantiles. A quantile map shows that the high

Acknowledgments

This study was performed with support from a research project (11-1480523-002666-01) sponsored by the Korean Ministry of Environment (MOE) and Korea National Institute of Environmental Research (NIER). Partial support was also provided by Korea Environment Industry & Technology Institute (KEITI) through Subsurface Environmental Management (SEM) Project, funded by Korea Ministry of Environment (MOE) (2018002440002), Basic Research Project (GP2018-002; 19-3415) of the Korea Institute of

References (73)

  • D.T. Monteith et al.

    Trends in the hydrochemistry of acid-sensitive surface waters in the UK 1988-2008

    Ecol. Indic.

    (2014)
  • C. Reimann et al.

    Background and threshold: critical comparison of methods of determination

    Sci. Total Environ.

    (2005)
  • J.J. Schröder et al.

    Nutrient management regulations in the Netherlands

    Geoderma

    (2008)
  • R.B. Alexander et al.

    Differences in phosphorus and nitrogen delivery to the Gulf of Mexico from the Misssissippi River Basin

    Environ. Sci. Technol.

    (2008)
  • J.K. Böhlke et al.

    Combined use of groundwater dating, chemical, and isotopic analyses to resolve the history and fate of nitrate contamination in two agricultural watersheds, Atlantic coastal plain, Maryland

    Water Resour. Res.

    (1995)
  • B.S. Cade et al.

    A gentle introduction to quantile regression for ecologists article

    Front. Ecol. Environ.

    (2003)
  • B.S. Cade et al.

    Estimating effects of limiting factors with regression quantiles

    Ecology

    (1999)
  • D. Chen et al.

    Influence of lag effect, soil release, and climate change on watershed anthropogenic nitrogen inputs and riverine export dynamics

    Environ. Sci. Technol.

    (2014)
  • C.M. Clark et al.

    Loss of plant species after chronic low-level nitrogen deposition to prairie grasslands

    Nature

    (2008)
  • R.A. Duce et al.

    Impacts of atmospheric anthropogenic nitrogen on the open ocean

    Science

    (2008)
  • J.B. Elsner et al.

    The increasing intensity of the strongest tropical cyclones

    Nature

    (2008)
  • M.E. Exner et al.

    Nebraska’s groundwater legacy: nitrate contamination beneath irrigated cropland

    Water Resour. Res.

    (2014)
  • European Commission

    Directive 91/676/EEC

    Off. J. Eur. Communities L

    (1991)
  • J.N. Galloway et al.

    Nitrogen fixation : anthropogenic enhancement-environmental response reactive N

    Global Biogeochem. Cycles

    (1995)
  • J.N. Galloway et al.

    The nitrogen cascade

    BioScience

    (2003)
  • J.N. Galloway et al.

    Nitrogen cycles: past, present, and future

    Biogeochemistry

    (2004)
  • J.N. Galloway et al.

    Transformation of the nitrogen cycle: recent trends, questions, and potential solutions

    Science

    (2008)
  • T. Gleeson et al.

    The global volume and distribution of modern groundwater

    Nature Geosci.

    (2016)
  • B. Hansen et al.

    Trend reversal of nitrate in Danish groundwater − a reflection of agricultural practices and nitrogen surpluses since 1950

    Environ. Sci. Technol.

    (2011)
  • B. Hansen et al.

    Groundwater nitrate response to sustainable nitrogen management

    Sci. Rep.

    (2017)
  • T. Harter et al.

    Addressing nitrate in California’s drinking water

    Calif. State Water Resour. Control Board.

    (2012)
  • M. Hirschi et al.

    Observational evidence for soil-moisture impact on hot extremes in southeastern Europe

    Nat. Geosci.

    (2011)
  • T. Ju et al.

    Reducing environmental risk by improving N management in intensive Chinese agricultural systems

    Proc. Natl. Acad. Sci.

    (2009)
  • H.R. Kim et al.

    Spatial-temporal variations of nitrate levels in groundwater of Jeju Island, Korea: evaluation of long-term (1993–2015) monitoring data

    Econ. Environ. Geol. (Korea)

    (2018)
  • N.R. Kitchen et al.

    Comparison of variable rate to single rate nitrogen fertilizer application: corn production and residual soil NO3-N

    Site-Specific Management for Agricultural Systems (1995)

    (1995)
  • R. Koenker

    quantreg: Quantile Regression [www document]

    (2018)
  • Cited by (18)

    • Development of an integrated hydrochemical index for delineating livestock manure-derived groundwater plumes in agro-livestock farming areas

      2022, Ecological Indicators
      Citation Excerpt :

      In the agricultural fields, which were developed in an upland at elevations of 44 ∼ 50 m asl, crops such as peppers, sweet potatoes and beans were cultivated with the application of fertilizers (100–150 kg ha−1 year−1 as of 2013 from the communication with local farmers) including silicate fertilizers. It should be noted that the application of fertilizers must have occurred all over the study area before 1995 (Supplementary Fig. S1) and probably more intensively given that the overuse of fertilizers resulted in the acidification and nitrate contamination of shallow groundwater in South Korea (Kim et al., 2019b). The use of chemical fertilizers has been regulated since 2004, decreasing the use of chemical fertilizers down to 46% between 1994 and 2014, whereas the use of livestock manures increased by 30% during the same period in South Korea (OECD, 2019).

    • Investigating distribution of nitrate concentration using ensemble nonparametric quantile regression

      2021, Science of the Total Environment
      Citation Excerpt :

      Choi et al. (2007) investigated the impact of land-use type on NO3− concentration; however, an extremely small study region with 12 wells was considered. Kim et al. (2019b) used a quantile linear regression method for NO3− concentration to suggest the optimal N application limit to manage NO3− groundwater pollution in South Korea, but they only considered agro-livestock farming districts. Additionally, while we are interested in less polluted areas, they focused on more polluted regions.

    • Spatiotemporal evolution of groundwater nitrate nitrogen levels and potential human health risks in the Songnen Plain, Northeast China

      2021, Ecotoxicology and Environmental Safety
      Citation Excerpt :

      Nitrate groundwater concentrations are higher in the pastoral areas of the southeastern plain along the Atlantic coast (North Carolina) (Messier et al., 2014; Gabriel et al., 2018). Nitrate concentrations in groundwater are higher in densely populated areas and residents face serious health risks (Kim et al., 2019). In some parts of Europe, the observed perennial value of groundwater nitrate in the Paris Basin was 22.4 mg/L, and nitrate concentrations in areas with developed agricultural activities have also increased correspondingly (Tavakoly et al., 2019).

    • Application performance and nutrient stoichiometric variation of ecological ditch systems in treating non-point source pollutants from paddy fields

      2020, Agriculture, Ecosystems and Environment
      Citation Excerpt :

      With rapid agricultural development and the effective control of point source pollution such as industrial wastewater and urban domestic sewage, agricultural non-point source (NPS) pollution is becoming increasingly prominent (Sun et al., 2012), with deteriorating water quality in many regions such as Europe (Fischer et al., 2017), USA (Badruzzaman et al., 2012), and South Korea (Kim et al., 2019).

    View all citing articles on Scopus
    View full text