Introduction

Nitrogen (N) is an essential plant nutrient that influences the yield and quality of arable and horticultural plants (Jalali 2005). However, intensive N fertilizer application to agricultural soils can cause N nutrients to leach below the crop’s root zone and threaten groundwater quality in agricultural catchments (Hallberg 1989; Jalali 2005), resulting in large-scale and adverse impacts on economic, ecosystem, and human health (Kolpin et al. 1994; Panno et al. 2001; Jalali 2005). For example, high groundwater N concentrations in drinking water may lead to methemoglobinemia, gastric cancer, and non-Hodgkin’s lymphoma in humans (Kolpin et al. 1994; Panno et al. 2001; Jalali 2005). Numerous studies have reported that increasing groundwater N concentrations in agricultural catchments are strongly associated with environmental factors such as climate, soil, and topography, and also with anthropogenic agricultural practices such as irrigation and fertilization (Hallberg 1989; Kolpin et al. 1994; Rivers et al. 1996; Jalali 2005; Andrade and Stigter 2009). However, the mechanisms by which agricultural land use types influence groundwater N concentrations, especially at the catchment scale, remain unclear.

In general, various agricultural land use types require different agricultural practices and fertilization (Jalali 2005). When N fertilizers are applied to agricultural soils in amounts that exceed crop uptake, excess N nutrients can potentially leach into groundwater (Groffman 2000; Jalali 2005; Wang et al. 2011). As a result, serious groundwater N pollution is common throughout the world where intensive N fertilizer application is employed, such as in Iran, USA, China, and Europe (Kolpin et al. 1994; Rivers et al. 1996; Jalali 2005). Jalali (2005) found an average nitrate-N (NO3-N) concentration of 49 mg L−1 in groundwater in Hamadan, Iran, and related this finding to excessive fertilizer application. Similarly, Rivers et al. (1996) reported increasing ammonium-N (NH4-N) concentrations in shallow and deep groundwater in Nottingham, England, caused by agricultural activities, i.e., fertilizer application and release of soil organic N due to tilling. Given the importance of N fertilizer application on groundwater N concentrations, this issue has received significant international attention (Kolpin et al. 1994; Rivers et al. 1996; Jalali 2005).

Agricultural land use type has a direct impact on groundwater N concentrations (Tang 2005; Young and Briggs 2005). In particular, the agricultural land use type influences the ecohydrological characteristics of the agricultural land and N use efficiency in agroecosystems, both of which are connected to N leaching processes into groundwater systems (Du et al. 2011; Wang et al. 2014b; Young and Briggs 2005). For example, rice agriculture involves a periodically flooded ecosystem, and the flooded conditions in paddy fields can increase the N leaching and limit nitrification in soil (Feng et al. 2004; Krupa et al. 2011), different from nonflooded agricultural ecosystems (e.g., upland agriculture and pasture), which likely influences the N species and increases N concentrations in groundwater (Tang 2005). However, the impact of agricultural land use type on groundwater N concentrations usually appears to be “specific in situ” to season, soil texture, and groundwater depth (Jin et al. 2004; Tang 2005; Du et al. 2011). Wang et al. (2011) found that agroforestry system and mono-peanut cropping system were prone to leaching more N during the rainy season than during dry season in Jiangxi province, China. Honisch et al. (2002) reported that soil texture contributed a lot to the N leaching into shallow groundwater system in the agricultural catchment. Bouman et al. (2002) found that the shallow groundwater system may be sensitive to groundwater N pollution in the agricultural catchment. These results suggested that, due to the “specific in situ” effects, the influence of agricultural land use type on groundwater N pollution may vary, and its impacts still need to be examined in greater detail (Chen et al. 2005; Du et al. 2011).

China is the largest global producer and consumer of N fertilizer, with an annual N consumption in 2011 estimated at 38 million tons (FAO 2011). This excessive N application has caused increasing concerns for groundwater quality in Chinese agricultural regions. Zhao et al. (2007) investigated groundwater NO3-N concentrations in 1139 wells under different agricultural land use types and found that the average groundwater NO3-N concentration reached 11.9 mg N L−1 in seven provinces surrounding the Bo-Hai Sea in northern China, with approximately 34.1 % of the samples exceeding the WHO standard for drinking water (NO3-N > 10 mg N L−1) (WHO 2011). These elevated groundwater N concentrations were associated with intensive fertilizer application in the region. Moreover, Du et al. (2011) reported an average groundwater NO3-N concentration of 13.8 mg N L−1 under vegetable field land use in Beijing, 2.8 times greater than in groundwater under row crops. However, these studies focused on northern China, where the groundwater is generally deep. Indeed, Wang et al. (2008) measured groundwater depth in 39 wells in the North China Plain and found that the average groundwater depth reached as high as 21.6 m. In contrast, in subtropical central China, precipitation is abundant, and the groundwater is generally shallow, potentially making the groundwater more sensitive to fertilizer application and agricultural land use type (Ouyang et al. 2013). In fact, Jin et al. (2004) reported that the NO3-N concentration in 40.5 % of wells in subtropical eastern China exceeded the WHO standard for drinking water, suggesting the existence of groundwater N pollution in this region. Because the local farmers obtain their domestic drinking water directly from this shallow groundwater without any treatment prior to consumption, the potential health hazards due to groundwater N pollution are considerable in subtropical central China (Jin et al. 2004; Du et al. 2011).

Many studies have shown that excessive N nutrients from fertilizer application can leach into shallow groundwater, with the amount of leached N nutrients differing according to various agricultural land use types. The NH4-N, NO3-N, and TN concentrations in shallow groundwater were investigated in a red soil agricultural catchment in subtropical central China from 2010 to 2012. This study aimed to (i) investigate the temporal variations of N concentrations in shallow groundwater under various agricultural land use types, (ii) evaluate the status of N pollution in shallow groundwater under different agricultural land use types, and (iii) determine the influence of agricultural land use type on the N concentrations in shallow groundwater.

Materials and methods

Geographic location and climate

The study site is located in the Jinjing catchment and part of the Changsha Research Station for Agricultural and Environmental Monitoring (27° 55′ –28° 40′ N, 112° 56′–113° 30′ E, elevation of 46–452 m) of the Chinese Academy of Sciences (CAS) in Hunan Province, China (Fig. 1a). The area has a typical subtropical monsoon climate with an annual mean air temperature of 17.5 °C and a mean annual rainfall of 1340 mm. In the studied region, most of the rainfall occurs from April to July due to the moist summer monsoon, while evaporation usually exceeds rainfall from July to October.

Fig. 1
figure 1

Land use map (a) and groundwater observation locations (b) in the Jinjing catchment

Full size image

Geology and hydrology characteristics

The characteristic landform pattern consists of low hills and flood plains; the former are generally found in the western and northern parts of the catchment, whereas the latter are located along streams and in southwest regions. A soil survey was conducted in the catchment during 2010–2011, and 1439 soil samples were collected. The survey suggested that the soils in the catchments developed from Quaternary red earth (Paleosol develops from Quaternary sediments under relatively hotter climate conditions) and highly weathered granite and were classified as Ultisols and Anthrosols (Soil Survey Staff 2010). The depth of the soils overlying the impermeable sandstone and Quaternary red earth ranged from 1.8 to 2.0 m. Chemical analysis of the soil samples in the catchment revealed a soil pH of 3.9–4.0, a soil organic carbon (SOC) content of 11.4–22.1 g kg−1, and a total soil nitrogen (TSN) of 1.3–1.9 g N kg−1 (Wang et al. 2014a).

The catchment hydrology was previously described by Wang et al. (2014a). Briefly, the headwaters typically originate from the upper hills in the western and northern parts of the catchment and flow through ditches/streams and paddy fields to the catchment outlet. The hydrology is strongly affected by anthropogenic activities in the catchment (Fig. 1a). Because rice agriculture requires a large amount of irrigation and surface drainage, there are many artificial irrigation-drainage channels, pools, and reservoirs in the catchment; the paddy fields are intensively irrigated from these nearby artificial water bodies during the growing period. This phenomenon is common in many rice agriculture regions (Wang et al. 2014b). The soils in the catchment are characterized by high permeability. The saturated soil hydraulic conductivity (Ks) of surface forest and arable soil ranges from 19.2 to 56.8 cm day−1, and from 15.1 to 60.0 cm day−1 for surface paddy soil (Zou 2012). Because the soils are permeable and the soil depths are thin in the catchment, the groundwater is generally shallow, at depths of 0.6–1.8 m; moreover, the water table in some paddy fields adjacent to the surface water bodies can be as shallow as 0.2 m beneath the land surface.

Land use types

In the catchment, forest is the dominant land use type and covers 65.1 % of the total catchment area. The forest is primarily composed of secondary Masson pine (Pinus massoniana Lamb.) woodland (approximately 30 years old). Other plants also grow in the forest, such as sawtooth oak tree (Quercus acutissima CARR.), azalea (Rhododendron simsii Planch.), and cape jasmine (Gardenia jasminoides Ellis). In recent decades, tea fields have been established on hilly terraced slopes following the clearing of secondary Masson pine woodland. To maintain the yield and quality of the tea, a large quantity of N fertilizer, including organic and chemical fertilizers, is commonly applied during tea cultivation (Fu et al. 2012; Wang et al. 2014b). Most of the vegetable fields are cultivated on a scattered scale by individual families for household purposes, and the digested or decomposed organic wastes from septic tanks and animal yards are commonly applied as fertilizer. The upland crops include corn, rape, and soybean, but these crops make up only a small part of the local agricultural production. Rice agriculture is the dominant agricultural land use type, comprising 27.1 % of the catchment area (Wang et al. 2014a), and paddy fields are typically distributed in the valleys and flood plains. In the catchment, there are two types of rice agriculture systems: double-rice and single-rice systems. The double-rice system refers to two harvests in a single year, with early rice crops transplanted in mid-April and harvested at the end of June and late rice crops transplanted in mid-July and harvested in mid-October. The single-rice system consists of a single harvest per year, and the rice is generally cultivated from June to September. However, because of the increasing cost of labor and decreasing economic benefit of rice agriculture, many paddy fields have been abandoned over the last decade, especially in the valleys, which are difficult to irrigate. Hence, abandoned paddy fields have become more prevalent in the studied catchment, comprising 5.1 % of the catchment area.

Sampling and chemical analysis

To observe the groundwater N concentrations in the catchment, a total of 161 groundwater observation wells were installed in 2010 under the different land use types described above: forest, abandoned paddy fields, vegetable fields, uplands, tea fields, single-rice fields, and double-rice fields (Fig. 1b). The wells were constructed using polyvinyl chloride (PVC) pipes with inside diameters of 5 cm. The PVC pipes were installed vertically to depths of 1.5 or 1.8 m, such that the bottom was at the level of the nearly impermeable parent material. The tops of the pipes were above ground and were covered by a screw cap to prevent unintended pollution from the outside; holes were drilled into the bottoms of the pipes to allow the percolation of groundwater into the pipe. To prevent contamination by soil particles and sediment during sampling, the bottom ends of the pipes were covered by a high-intensity mesh tape.

To construct the observation wells, first, a hole with a diameter of 5 cm was manually bored using a soil sampler. The soils generated during the boring were kept in a plastic tank, after the removal of little stones and semi-weathered residual parent material. Second, some water was added to the plastic tank to produce a muddy mixture of soil and water, and the mud was then used to fill back into the hole. Third, the pipe was directly inserted into the holes, and the soil around the pipe walls was manually compacted to prevent bypass flow along the pipe walls. Fourth, the inside of the pipe was purged and washed with clear water, and the pipe was filled with 10 cm thick clear sand. Eventually, an intact groundwater observation well was established. To evaluate the performance of the groundwater observation wells, the groundwater was sampled after a rainstorm in both the rainy and dry seasons, and no obvious sediment in the extracted groundwater samples was found. The N concentrations from filtered (through a 0.45-μm membrane) and unfiltered groundwater samples were also compared, and no significant differences were observed. These results suggested that the wells would function effectively for groundwater observation.

During the groundwater sampling, the screw cap was first carefully removed, the depth of shallow groundwater was measured with a tape, and the groundwater was sampled using a handheld vacuum pump. The groundwater samples were collected in 1-month intervals from April 2010 to November 2012; however, some pipes failed in the dry season when the soil was very dry. The collected water samples were immediately transported to the laboratory at the experimental station and stored below 4 °C until analysis. Three N parameters were determined: NH4-N, NO3-N, and TN. Briefly, the concentration of TN (i.e., the total organic and inorganic N) was determined using a fully automated injection system (Tecator FIAstar 5000 Analyzer, Foss Tecator, Sweden) after digestion with a K2S2O8-NaOH solution. The NO3-N and NH4-N concentrations (i.e., inorganic N) were determined directly using the fully automated injection system.

Soil texture and nitrogen fertilizer application rate

The soil at each groundwater observation well was collected during the construction of the wells, and the soil particle size distribution was determined by the hydrometer method (Liu 1996) according to the US texture classification system, i.e., sand content (2.0–0.05 mm), silt content (0.05–0.005 mm), and clay content (less than 0.002 mm). The soil texture was determined based on the soil particle size distribution using the online Soil Texture Calculator Program on the USDA website (accessed using http://www.nrcs.usda.gov).

To understand how N fertilizer application rates influence N concentrations in shallow groundwater, a questionnaire was given to the landowner of each field where a groundwater observation well was constructed to estimate the N fertilizer types and application rates used. The average value of N fertilizer application rates was calculated for each category of land use type, and these average values reflected the mean levels of N fertilizer application rates under each land use type in the studied catchment.

Statistics and graphical representation

The pollution status in the catchment was evaluated using the median values of groundwater N concentration during the observation period according to the ordinary Kriging interpolation in ArcGIS 10.0 software (ESRI, Redlands, CA, USA) and based on the Chinese Environmental Quality Standards for Groundwater of GB/T 14848-93 (General Administration of Quality Supervision of China 1993).

Factorial correspondence analysis (FCA), a data analysis technique for nominal categorical data, was applied for the observed N concentrations in shallow groundwater to determine their relationship to land use type, soil texture, and groundwater depth in the catchment. An important feature of FCA is the multivariate treatment of the data through simultaneous consideration of multiple categorical variables. This multivariate nature enables FCA to reveal relationships among variables that would not be detected in a series of pair-wise comparisons of variables.

FCA was performed using SPSS18.0 software (SPSS Inc., Chicago, IL, USA), following the method of Andrade and Stigter (2009). Briefly, two quantitative variables (N concentration and groundwater depth) and three qualitative variables (soil texture, season, and land use type) were selected (Table 1). Therefore, the analysis accounted for climate, soil, vegetation, groundwater depth, N fertilizer, and groundwater N pollution status in the catchment. It is widely reported in the literatures that, based on climate and soil water regime in the region, each hydrological year can be separated into three seasons: soil moistening (from February to June), soil exhausting (from July to October), and soil restoring (from November to January) (Li et al. 2004; Wang et al. 2011). Because the soil N that leached into the shallow groundwater system was largely controlled by soil moisture, the classification of seasons in Table 1 refers to this season classification method. Prior to FCA, the data were preprocessed by transforming the two quantitative variables into ordinal variables and subdividing each of them into three classes based on an equal sample distribution criterion. The two qualitative variables were subdivided into classes according to the practical situation at each of the groundwater observation wells. Land use types were recorded during the construction of the groundwater observation wells, and soil textures were obtained using the USDA Soil Texture Calculator Program. All classes and frequencies of the quantitative and qualitative variables are shown in Table 1. The original data matrix was then codified in a binary fashion (1 if the sample falls within a variable class, 0 if not) and transformed through calculating the so-called marginal frequencies of each sample and modality (Stigter et al. 2008; Andrade and Stigter 2009). Finally, the original data matrix was run in the SPSS program, in which the first two factors were retained based on eigen value stabilization in the scree plot and the researchers’ knowledge of the case study (Andrade and Stigter 2009).

Table 1 Classes of quantitative variables and qualitative parameters in factorial correspondence analysis (FCA)
Full size table

Results

Annual nitrogen fertilizer application rates and nitrogen concentrations

The median values of N fertilizer application rates for each land use type ranged from 0 to 450 kg N ha−1 a−1 in the catchment. N fertilization rates were the lowest in the forest and abandoned paddy fields and the greatest in the tea fields (Table 2). The fertilizer was applied in different forms according to land use type: urea and compound fertilizers were applied in the single-rice and double-rice fields, and organic fertilizers were applied in the vegetable fields.

Table 2 Forms and average application rates of N fertilizers under different land use types within the study catchment
Full size table

Overall, the 25 % quantile, 50 % quantile (median), and 75 % quantile values of the TN concentrations in shallow groundwater were 0.73, 1.38, and 2.70 mg N L−1 for all the land use types during the observation period. The highest values were observed in the tea field: 1.87, 3.87, and 10.8 mg N L−1 for the 25 % quantile, median, and 75 % quantile, respectively (Table 3). However, the 100 % quantile (maximum) values of the TN concentrations were found beneath the single-rice and double-rice fields, at 80.9 and 98.2 mg N L−1, respectively. The median and 75 % quantile values of NH4-N were 0.15 and 0.91 mg N L−1 under all land use types, and the highest values were found beneath the double-rice fields, at 0.23 and 1.21 mg N L−1, respectively. The maximum NH4-N concentration observed was 96.6 mg N L−1. The 25 % quantile, median, and 75 % quantile values of NO3-N concentrations across all land use types were estimated to be 0.10, 0.39, and 1.04 mg N L−1. The highest 25 % quantile value of NO3-N concentration was found beneath the upland field, at 0.40 mg N L−1, and the median and highest 75 % quantile values were found beneath the tea field at 1.51 and 8.66 mg N L−1, respectively. With regard to the chemical composition of the groundwater N, NH4-N and NO3-N were determined to be the main N species in groundwater, accounting for 32.2 and 44.9 % of the TN concentration, respectively. However, the N chemical composition was different under the various agricultural land use types; the abandoned paddy fields, single-rice fields, and double-rice fields were characterized by a higher proportion of NH4-N, accounting for 48.3, 41.2, and 45.3 % of the TN concentration, respectively.

Table 3 Descriptive statistics of ammonium-N (NH4-N), nitrate-N (NO3-N), and total N (TN) concentrations in groundwater under different land use types during the observation period
Full size table

Groundwater quality evaluation

The NH4-N, NO3-N, and TN concentrations in shallow groundwater were classified to illustrate the pollution status in the Jinjing catchment. According to the Chinese Standards for Groundwater, NH4-N pollution exists if the NH4-N concentration in the groundwater exceeds 0.2 mg N L−1. As indicated in Fig. 2, based on the median values of NH4-N concentrations, 50.5 % of the catchment area was categorized as containing severe NH4-N pollution, especially in the valleys and flood plains along streams where the paddy fields were distributed. According to the threshold value of NO3-N concentrations for groundwater pollution defined by the Chinese Standards for Groundwater (20.0 mg N L−1), almost none of the catchment area was polluted with NO3-N, with median values of groundwater NO3-N ranging from 0 to 2.0 mg N L−1. Hence, the spatial distribution of NO3-N was not shown in Fig. 2. The Chinese Standards for Groundwater do not set criteria for TN concentrations (Fig. 2), and the TN concentrations were classified according to the NO3-N concentration classification. A large portion of the catchment area had the median values of groundwater TN concentrations ranging from 2.0 to 4.0 mg N L−1. In general, higher median values of groundwater TN occurred in the valleys and flood plains along streams, similar to the spatial distribution of NH4-N concentrations.

Fig. 2
figure 2

Spatial distribution of the 50 % quantile value (median value) of groundwater ammonium-N (NH4-N), nitrate-N (NO3-N), and total N (TN) concentrations using ordinary Kriging interpolation in the Jinjing catchment (n = 161)

Full size image

The NH4-N, NO3-N, and TN concentrations in shallow groundwater were categorized by agricultural land use type (Fig. 3). A total of 36.3 % of the groundwater samples exhibited greater than 0.2 mg N L−1, suggesting the occurrence of NH4-N pollution according to the definition of the Chinese Standards for Groundwater. Specifically, in the single-rice and double-rice fields, 43.0 and 48.6 %, respectively, of the groundwater samples exceeded the pollution criterion of 0.2 mg N L−1 in the Chinese Standards for Groundwater. In contrast, only 3.9 % of the groundwater samples from the single-rice and double-rice fields exceeded the maximum safe NO3-N concentration for drinking water recommended by the WHO, and almost none of the samples exceeded the pollution criterion set by the Chinese Standards for Groundwater. The greatest frequency of groundwater NO3-N pollution (> 20 mg N L−1, Chinese Standards for Groundwater) was observed in the tea fields (24.4 %), followed by the vegetable fields (5.2 %). The distribution of TN concentrations in groundwater was similar to that of NO3-N, and higher groundwater TN concentrations were found beneath the tea fields and vegetable fields. Considering all land use types, the TN concentrations in 5.7 % of the groundwater samples exceeded 20 mg N L−1.

Fig. 3
figure 3

Frequency distribution for groundwater ammonium-N (NH4-N), nitrate-N (NO3-N), and total N (TN) concentrations in the studied catchment (n = 161) based on the Chinese Environmental Quality Standards for Groundwater (GB/T 14848-93) (General Administration of Quality Supervision of China 1993)

Full size image

Temporal variations in nitrogen concentration dynamics

The 25 % quantile, median, and 75 % quantile values of the N concentrations in shallow groundwater under the different agricultural land use types were calculated monthly to reveal temporal variations (Fig. 4). The average N concentrations in shallow groundwater fluctuated strongly according to the season and displayed different variations under the various land use types. In general, the N dynamics had more variations during the soil exhausting season than during the soil moistening and soil restoring season in a single hydrological year beneath the abandoned paddy, single-rice, and double-rice fields, while the NO3-N and TN dynamics had more variations during the soil moistening season than during the soil exhausting and soil restoring season beneath the upland and tea fields. A comparison of the different land use types showed that the variation of groundwater N dynamics was smaller and the groundwater N concentrations were relatively stable beneath the forests and vegetable fields; in contrast, the N dynamics were greater beneath the abandoned paddy, upland, tea, single-rice, and double-rice fields. Because NO3-N was the dominant N species to the TN concentrations under beneath the upland and tea fields, their concentrations varied consistently with the season. However, beneath the single-rice and double-rice fields, the influence of NO3-N concentration on the TN dynamics was relatively small because NH4-N was the dominant N species to the TN concentrations during certain periods. For instance, during the period from July 2010 to January 2011, the NH4-N concentration accounted for 46.8 % of TN beneath the double-rice fields; accordingly, the TN dynamics were consistent with the NH4-N dynamics but not with the NO3-N dynamics.

Fig. 4
figure 4

Dynamics of the 25 % quantile, 50 % quantile (median), and 75 % quantile values of groundwater ammonium-N (NH4-N), nitrate-N (NO3-N), and total N (TN) concentrations under the different agricultural land use types during observation period of 2010–2012. The lower boundary of the gray shadow indicates the 25 % quantile value, and the upper boundary indicates the 75 % quantile value. MS indicates soil moistening season during from February to June, ES indicates soil exhausting season during from July to October, and RS indicates soil restoring season during from November to January. It is noted that the concentration ranges on the y-axis vary for the different N species under the different land use types

Full size image

Factorial correspondence analysis

The variable classes in the FCA are projected on the primary plane in Fig. 5, and the variable classes of NH4-N, NO3-N, and TN concentrations in shallow groundwater are connected by a line in increasing class order. These steps are typically performed to facilitate interpretation (Stigter et al. 2008). In the primary plane, the first two factorial axes were retained, explaining 78.8 % of the total data variance. The first factor (x-axis) represents the highest to the lowest classes of NH4-N concentration, whereas the second factor (y-axis) represents the highest to the lowest classes of NO3-N and TN concentration. Therefore, the first factor can be interpreted as reflecting the degree of NH4-N pollution and its correlation with other variables, with the second factor reflecting NO3-N and TN pollution.

Fig. 5
figure 5

Primary factorial plane of factorial correspondence analysis (FCA) based on the variables of ammonium-N (NH4-N), nitrate-N (NO3-N), total N (TN), soil texture, groundwater depth, and agricultural land use type. The classes of quantitative variables and qualitative parameters are listed in Table 1

Full size image

Concerning land use type, all the nonflooded land use types were found to be distributed on the left side of the primary plane, whereas the flooded land use types were on the right side. The land use types with low applied N fertilizer rates, for example, forest (0 kg N ha−1 a−1), abandoned paddy fields (0 kg N ha−1 a−1), and single-rice fields (187 kg N ha−1 a−1), were located on the upper part of primary plane, and the land use types with high applied N fertilizer rates were located on the lower part of the primary plane. This situation suggested that the land use type and resulting N fertilizer application rate had distinctive influences on the N concentration in shallow groundwater. The highest NH4-N class (AN3) was located on the positive end of the first factor and was close to the single-rice and double-rice field land use types, indicating that the single-rice and double-rice fields have great potential to impose NH4-N pollution in shallow groundwater. The highest NO3-N and TN classes (NN3 and TN3) were found to be present on the negative end of the second factor and correlated with the maximum applied N fertilizer rates in the tea fields. Hence, as expected, large N fertilizer application rates would lead to increasing NO3-N and TN pollution in shallow groundwater. Regarding groundwater tables, the location of all the groundwater depth classes (G1, 2, 3) near the origin of the coordinate suggests that the groundwater table has the smallest effect on the N concentration in shallow groundwater. With regard to soil texture, sandy clay loam (S2) and silt (S3) were close to the high NH4-N, NO3-N, and TN classes, indicating that sandy clay loam and silt loam soils have the potential to leach more N into the shallow groundwater system. Interestingly, the soil exhausting season (ES) was located on the positive end of the first factor and was close to the high NH4-N class, single-rice class, and double-rice class, indicating an increase in NH4-N concentrations during the soil exhausting season, especially under flooded land use types.

Discussion

Groundwater quality

Groundwater was observed for 33 months in this study. However, because the groundwater observation wells did not function well under extremely dry conditions, the groundwater N concentrations were only sampled under relatively wet soil conditions over the 33 months of the study. Intensive N fertilizer application under various agricultural land use types generally causes severe groundwater pollution due to the excess N nutrients that leach from the soil (Jalali 2005). The groundwater NH4-N concentrations in the studied catchment partly exceeded the pollution criteria of the Chinese Standards for Groundwater and showed a moderate NH4-N pollution level (Figs. 2 and 3). Numerous studies have reported severe groundwater NO3-N and TN pollution in agricultural regions (Kolpin et al. 1994; Rivers et al. 1996; Jalali 2005). However, the groundwater NO3-N and TN concentrations in the studied catchment tended not to meet pollution criteria. These differing results indicate that groundwater N pollution varies greatly for different agricultural catchments and the “specific in situ” relationship between agriculture land use types and groundwater quality need to be examined in greater detail.

Nitrogen species and temporal variations of groundwater N concentrations

The pollution status as reflected by NH4-N concentrations indicated that rice agriculture is the main source of NH4-N in shallow groundwater (Fig. 2). The high NH4-N concentrations beneath the paddy fields can be attributed to two mechanisms: (1) the application of urea, which can be quickly hydrolyzed into NH4-N during rice cultivation and (2) the anaerobic and reducing conditions that dominate the denitrification process and limit the nitrification process in paddy soil (Feng et al. 2004; Krupa et al. 2011). These two mechanisms make the N nutrients in paddy soil and surface ponding water prone to leaching into the groundwater in the form of NH4-N. The tea and abandoned paddy fields were the other sources of NH4-N in shallow groundwater (Table 3). Given the high N fertilizer application rate of 450 kg N ha−1 a−1 (Table 2), the high NH4-N concentrations beneath the tea fields are not surprising. Regardless, the higher groundwater NH4-N concentrations beneath the abandoned paddy fields likely resulted from the residual NH4-N from previous rice cropping and the anaerobic and reducing conditions in the fields because no fertilizer was applied to these fields during the study period. Although the paddy fields had abandoned, the periodical flooding still occurred in the fields, as the pothole feature of paddy fields permits the retention of a certain amount of rainfall during rainstorms. Meanwhile, these abandoned paddy fields commonly distributed in valleys, where groundwater was generally shallow, our observations showed that the median value of groundwater depth beneath the abandoned paddy fields was approximately 0.26 m. Therefore, the waterlogging and resulting anaerobic conditions beneath the abandoned paddy fields promote the leaching of N into the groundwater system in NH4-N form (Feng et al. 2004; Wang et al. 2014a). In the catchment, these paddy fields had been abandoned for more than 3 years, demonstrating that once the groundwater has been polluted by NH4-N, the removal of this N species is a relatively slow process.

Although the tea and vegetable fields exhibited undeniable NO3-N pollution in shallow groundwater (Fig. 3 and Table 3), these fields accounted for only a small area of the entire catchment. Thus, their influence on the groundwater NO3-N pollution became negligible at the catchment scale (Fig. 2). In recent decades, the booming tea industry has caused the establishment of numerous tea gardens in the catchments (Fu et al. 2012; Wang et al. 2014b), and our results suggest that such change of land use may threaten groundwater quality. Du et al. (2011) reported that the median value of groundwater NO3-N concentrations reached 13.8 mg N L−1 beneath the vegetable fields in Beijing, higher than the value of 0.89 mg N L−1 in our catchment (Table 3). These different values may result from different N application rates; the N fertilizer application rate reached 1741 kg N ha−1 a−1 in Beijing, much higher than the 305 kg N ha−1 a−1 rate in our catchment (Table 2). Although the N fertilizer application rate was lower in the vegetable fields than in the double-rice fields, the groundwater NO3-N concentrations were higher beneath the vegetable fields; this result can be explained by the aerobic soil environment in the vegetable soil, which encourages nitrification (Ouyang et al. 2013), and by the higher permeability in the vegetable soil compared to paddy soil that has a less permeable plow pan (Sander and Horst 2007).

Almost none of the 75 % quantile values of groundwater TN concentrations exceeded 10 mg N L−1 under the land use types, except for the tea fields (Fig. 3), due to the high fertilization rates. Similar results were reported by Lu and Yin (2008), who found that high fertilizer application rates can increase TN concentrations in shallow groundwater under agricultural land use types in Tianjin in northern China. Despite the observation that NH4-N only accounted for 32.2 % of the groundwater TN concentration at the catchment scale (Table 3), the spatial distribution of TN concentrations in the catchment was similar to that of NH4-N; this is because the NH4-N was the dominant N species beneath the rice agriculture fields in the valleys and flood plains where higher TN concentrations were observed (Fig. 2). This situation demonstrates the impact of NH4-N and rice culture on the spatial distribution of TN concentrations.

The different groundwater N temporal dynamics show the obvious influence of land use type on the seasonal variation of N concentrations in shallow groundwater (Fig. 4). The forest exhibited less variation with the seasons during the observation period, which likely resulted from the lack of fertilization. The N dynamics were also seasonally stable for the vegetable fields. The vegetable growth cycle is generally short, and fertilizer application and irrigation are commonly frequent, such a land management helps the vegetable soil maintain stable NO3-N concentrations and reduce seasonal variation of N concentrations in shallow groundwater (Du et al. 2011). Although fertilizer application occurred in different seasons in the tea and upland fields (soil moistening season for uplands and soil restoring season for tea fields), the groundwater N concentrations (particularly NO3-N and TN) varied consistently and presented higher N concentrations during the soil moistening season (Fig. 4). These suggest that abundant precipitation and wet soil region had a significant contribution to the leaching of N into the groundwater in the tea and upland fields. The NH4-N was the dominant N species and the NH4-N dynamics varied consistently beneath the single-rice and double-rice fields. Because the fertilizer timing was different in the single-rice and double-rice fields, this consistent variation of NH4-N indicates that not fertilization but season controls the seasonal variation of NH4-N under the two land use types.

Influence of agricultural land use type on groundwater nitrogen concentrations

Agriculture has been identified to be a major source of anthropogenic N for groundwater systems in rural areas (Jalali 2005; Du et al. 2011). The FCA results suggest that all the N species assessed had a strong connection to the agricultural land use type, and such connections likely resulted from the interaction of natural factors and anthropogenic activities associated with agricultural land use types (Hallberg 1989; Kolpin et al. 1994; Rivers et al. 1996; Jalali 2005; Andrade and Stigter 2009). For instance, the flooded, anaerobic, and reducing conditions in paddy fields can inhibit the N nitrification process and thus increase the potential for NH4-N to leach into the groundwater system (Tang 2005). This is why the shallow groundwater NH4-N concentrations had such a strong association with the flooded land use types but not with the nonflooded agricultural land use types (Fig. 5). According to the FCA results, high NH4-N concentrations were present during the soil exhausting season beneath the single-rice and double-rice fields. This finding was consistent with the results shown in Fig. 4, in which the NH4-N concentrations exhibited obvious peaks in the soil exhausting season. This phenomenon can be attributed to the accumulation of groundwater NH4-N that leached from the upper soil during the previous wet soil season and also to the large amount of irrigation during the soil exhausting season in the single-rice and double-rice fields. Indeed, the soil water flow was the main driving force of solutes moving into the groundwater system (Jalali 2005; Du et al. 2011; Wang et al. 2011). Therefore, NH4-N reduction practices in rice agriculture catchments should be instituted to protect shallow groundwater quality during the soil moistening season and the soil exhausting season in subtropical central China.

As the FCA results indicate that the groundwater N concentrations had strong connections with the N fertilizer application rates according to agricultural land use type (Fig. 5), the leaching of NO3-N and TN into shallow groundwater was largely controlled by N rate employed under different land use types. Previous studies have reported that a marked increase in N fertilizer rates applied to agricultural soils causes N to leach into the groundwater system via soil vertical drainage (Hallberg 1989; Jalali 2005; Du et al. 2011), and our results were in good agreement with these findings. Although the N application rate is an intrinsic feature of agriculture land use type, various regions worldwide differ with regard to recommended and optimized fertilizer rates as a result of efforts to minimize the amount of N that leaches into the groundwater system (Jalali 2005). In our catchment, the strong connection between N fertilizer application rate and the groundwater concentrations of both NO3-N and TN indicates that the application exceeds the needs of the crops for growth (Du et al. 2011; Jalali 2005; Tang 2005). Therefore, N fertilization rates should be optimized to prevent NO3-N and TN pollution under different agricultural land use types.

The results of FCA suggest that groundwater depth is not an important factor for groundwater N concentrations under the studied agricultural land use types (Fig. 5). The first possible explanation for this observation is that precipitation is abundant and groundwater is generally shallow across the entire catchment; these properties facilitate the diffusion of N into groundwater through soil water flow. The second possible explanation is that the soil in the catchment, which was developed from Quaternary red earth and highly weathered granite, is rich in macropores that accelerate the transport of soil N toward shallow groundwater through preferential flow (Sander and Horst 2007; Wang et al. 2011). The first explanation is supported by the findings of Ouyang et al. (2013), who reported that abundant precipitation can make groundwater more sensitive to fertilizer application under different agriculture land use types. The second explanation is corroborated by the FCA results, which showed that the soil texture had a strong influence on the N concentrations in shallow groundwater (Fig. 5), whereas another possible explanation for these results is the little variation in the observed shallow groundwater table (less than 100 cm, see Table 1). As the shallow groundwater depth was divided into 3 classes during the FCA analysis, such little variations in the classes of groundwater depth may unlikely to be able to explain the difference in groundwater N concentrations.

Many studies have demonstrated the importance of controlling N fertilizer application rates to alleviate groundwater N pollution in agricultural catchments (Jalali 2005; Young and Briggs 2005). Our results further confirm the strong association of agricultural land use type and N fertilizer application rates with groundwater N concentrations and highlight the importance of appropriate landscape planning and optimized N fertilizer application to protect groundwater quality under different land uses. However, further studies are required to identify the mechanisms that cause rice agriculture systems to promote the accumulation of NH4-N in shallow groundwater and to assess the environmental and agronomic effectiveness of various agricultural land use types; in particular, efforts to control N fertilizer application rates and provide reasonable planning of landscapes and land use types must be evaluated (Palma et al. 2007; Wang et al. 2014b). In general, to reduce N pollution in shallow groundwater in the agricultural catchments, N leaching must be reduced through an optimization of N fertilization or by appropriate landscape planning.

Conclusions

In the studied red soil agricultural catchment in subtropical central China, the median NH4-N, NO3-N, and TN concentrations in shallow groundwater were determined to be 0.15, 0.39, and 1.38 mg N L−1, respectively. These values indicate moderate NH4-N pollution in the shallow groundwater system based on the Chinese Environmental Quality Standards for Groundwater (GB/T 14848-93). Although the tea and vegetable fields were found to be prone to increasing NO3-N and TN concentrations, there was no large area of NO3-N and TN pollution in the shallow groundwater system at the catchment scale. For all of the land use types evaluated, the NH4-N concentration showed a strong association with the flooded agricultural land use (i.e., single-rice and double-rice fields) but not with the N fertilizer application rate, indicating that intensive rice agriculture is the main reason for NH4-N pollution in shallow groundwater. The NH4-N concentrations were especially high in shallow groundwater under the flooded agricultural land use types in the soil exhausting season from July to October. In contrast, the NO3-N and TN concentrations were significantly associated with the N fertilizer application rate. This situation suggests that excessive fertilizer rates may cause NO3-N and TN pollution in shallow groundwater in subtropical central China.

The study also highlights the importance of agricultural land use type and N fertilizer application rates on the quality of groundwater in subtropical central China. The underlying mechanisms of N leaching into the groundwater and the effectiveness of the environment of different agricultural land use types require further study. However, the results of this study indicate that the environmental risks of groundwater N pollution may be reduced by evaluating fertilizer application rates and employing appropriate agricultural land use types to decrease the migration of N into shallow groundwater.