Runoff-associated nitrogen and phosphorus losses under natural rainfall events in purple soil area: the role of land disturbance and slope length Open Access

Nitrogen (N) and phosphorous (P) are the common nutrient elements in ecosystems (Kleinman et al. 2006; Xing et al. 2016). The runoff-associated N and P losses originating from sloping land can lead to reductions in soil fertility and crop yields, and further cause non-point pollution in downstream aquatic ecosystems (Pote et al. 1996; Parris 2011). Runoff-associated N and P losses are reported to be affected by many factors, such as rainfall amount, rainfall intensity, land disturbance and slope length (Girmay et al. 2009; Xing et al. 2016; Wu et al. 2018b). Beniston et al. (2015) reported that conventional tillage disturbance significantly accelerated soil erosion and carbon and macronutrient losses, and no-till increased the resilience of soils to the high intensity rainfall. Xing et al. (2016) found that runoff rate and runoff-associated TN loss rate decreased with the increasing of slope length and rainfall intensity. Though great efforts have been made to quantify the effects of land disturbance and slope length on erosion and nutrient loss, the dynamics of runoff-associated N and P losses in different forms was still unclear, and the conflict relationships between runoff-associated N and P losses with the rainfall intensity were usually reported. For instance, Song et al. (2017) suggested that the exponential function could describe the correlations between total dissolved nitrates (TDN) concentration and rainfall intensity in the Karst area. However, Wu et al. (2018b) indicated that TDN concentration presented significantly logarithmical relationship with rainfall intensity on the bare slope. Therefore, more data are required to interpret the effects of influence factors on the dynamics of runoff-associated nutrient losses.

Land disturbance significantly affects the runoff-associated N and P losses. Previous studies indicated that artificial disturbances, such as conventional tillage, overgrazing and urbanization expansion, could break up the soil aggregates and thus decrease the resistance and water stability of soil particles (Celik 2005; Navas et al. 2008; Roberts et al. 2009), resulting in acceleration of soil erosion and nutrient losses under the intensive rainfall events (Beniston et al. 2015). Therefore, natural restoration measurement was increasingly adopted to control rainfall erosion and reduce nutrient losses in the recent years. For instance, the Grain-for-Green project in China has achieved great benefits in increase in vegetation coverage and reduction in soil erosion and nutrient losses (Wang et al. 2007; Deng et al. 2012). Martínez-Mena et al. (2020) reported that sustainable land management practices (e.g. reduced tillage, no-tillage and cover crops) significantly improved soil quality and reduced runoff, soil erosion and nutrient losses in comparison to the conventional tillage in Mediterranean rainfed agroecosystems. Despite the land disturbance effects on soil erosion are increasingly intensified, the regularity of runoff-associated N and P losses under different land disturbances remains unknown.

Slope length also significantly affects the runoff and sediment yield along the slope (Boix-Fayos et al. 2006). Xing et al. (2016) reported that the runoff rate decreased and sediment rate increased with the increasing of slope length. This was because the increasing infiltration reduced runoff production (Kleinman et al. 2006; Ghahramani & Ishikawa 2013), while runoff energy increased with the increasing of slope length, and hence led to increase in sediment yield (Boix-Fayos et al. 2006; Giménez & Govers 2008). Additionally, some researches also interpreted the relationships between runoff and sediment yield under different slope lengths using variable mathematical functions (Kothyari et al. 2004; Wang et al. 2010). However, little information was available on the effects of slope length on the changes in runoff-associated N and P losses and their relationships with rainfall intensity and runoff rate (Xing et al. 2016).

Purple soils are widely distributed in the Three Gorges Reservoir Area (TGRA), and provide an important basis for the agricultural production in this area. Purple soils are originated from purplish parent materials within a relatively short duration of weathering, and have high fertility and great potential for crop productions (Zhu et al. 2008), while they are also characterized by relatively shallow soil profile, high dispersibility, poor structure and high erosion vulnerability (Li et al. 2019). Due to the intensive rainfall, steep slope gradients and irrational land uses in the purple soil area, severe soil erosion and intensive nutrient losses from the sloping farmlands occurred in the TGRA (Meng et al. 2001; Jannes et al. 2009), resulting in high risk of non-point pollution and downstream aquatic eutrophication (Wu et al. 2012). Thus, it has great significance to understanding the regularity of land disturbance and slope length on nutrient losses in the purple sloping farmlands of the TGRA.

The specific objectives of this study are to (1) quantify the effects of land disturbance and slope length on the changes in runoff-associated N and P concentrations and loss amounts in different forms under the natural rainfall events, and (2) determine the relationships between N and P concentrations in different forms and rainfall intensity as well as runoff rate as affected by land disturbance.

A detailed description of the study site has been given by Guo et al. (2017). Briefly, the study site is located in Xiema Town, Beibei District, Chongqing Municipality City, Southwest China (106°18′-106°25′E, 29°43′-29°48′N) (Figure 1). The elevation varies from 175 to 1,312 m above sea level. The study site belongs to the subtropical warm and humid monsoon area, with average annual temperature of 18.2 °C, average annual sunshine of 1,006.2 hours and a frost-free period of 359 days. The average annual precipitation reaches 1,174 mm, and approximately 75% occurring in the wet season from May to October. The underlying soils in the study area are mainly classified as purple soil equivalent to Regosols in FAO Taxonomy or Entisols in USDA Taxonomy (Lin 2002). The prevailing vegetation is evergreen broadleaved forest, including the major silvicultural tree species of Pinus massoniana (Lamb.), Cunninghamia lanceolata (Lamb.) Hook, Cupressus funebris (Endl.) and Quercus fabric (Hance.) in the study area.

The experimental setup totally consisted of 9 parallel closed plots along the same 15° slope with 20-, 40- and 60-m lengths and fixed to 1 m in width on the southerly aspect in the field (Figure 1). Three plots were set at each plot length and the middle one of them was created with natural restoration, two plots either side set as replications were implemented by artificial disturbances. All plots were surrounded by concrete borders that were 15 cm in width and extended 30 cm below the ground and 20 cm above the ground. A collector and V-shaped outlet made from concrete bricks were fixed at the end of each plot, and a plastic pipe with 10 cm diameter was used to guide the runoff and sediment from the collector into the concrete brick tank (1 m in length, 1 m in width, 1 m in depth) fixed at the end of the collector. The collector and tank were both covered by sheet iron to eliminate the rainfall effects on runoff volume. An automated tipping-bucket rain gauge was fixed next to the plots to record rainfall depth and duration. In this study, the plots with artificial disturbance were managed through weed and plough in the 30 cm of topsoil at 10-day intervals, while the ones with natural restoration were covered dominantly by Conyza canadensis (L.) Cronq and Anredera cordifolia (Tenore) Steenis without any human management.

Before the wet season, soil samples were collected to obtain the background soil properties including soil bulk density, particle size distribution, pH, soil organic matter, total nitrogen and total phosphorus (Table 1). These variables were measured following the recognized protocols (for detailed information please refer to Guo et al. 2017). During the wet season, rainfall monitoring was launched and only the erosive rainfall events generating runoff and sediment were recorded, and a total of 8 erosive rainfall events were recorded from May 13, 2013 to July 18, 2013 (Figure 2). For each erosive rainfall event, the entire volume of runoff and sediment was measured in situ and then 2 plastic bottles (∼500 ml for each one) washed by deionized water were used to collect mixed samples at the different depths. After that, all samples were kept in the dark at 4 °C and delivered to the laboratory immediately within 8 hr.

In laboratory, one bottle of sample was used to accurately determine the concentrations of total nitrogen (TN) and total phosphorus (TP) and the other bottle of sample was preferentially passed through the 0.45-μm microporous filter membrane; the filtered sample was used to analyze the concentrations of total dissolved nitrogen (TDN), ammonium nitrogen (NH₄-N), nitrate nitrogen (NO₃-N), total dissolved phosphorus (TDP) and orthophosphate ion (PO₄-P) for each plot after each rainfall event. TN and TDN were measured by potassium persulfate (K₂S₂O₂) oxidation-ultraviolet spectrophotometry. NH₄-N was determined by indophenol blue colorimetry. NO₃-N was measured by ultraviolet spectrophotometry. TP and TDP were determined by potassium persulfate (K₂S₂O₂) oxidation-molybdenum blue colorimetry. PO₄-P was analyzed by molybdenum blue colorimetry. The measurements for the above variables (i.e. TN, NH₄-N, NO₃-N, TP, TDP, PO₄-P) were performed in terms of the standard Chinese protocols (SEPA 2002).

A Kolmogorov-Smirnov (K-S) test was used to test the normality of all data and log-transformation (base 10) was conducted when the data did not follow a normal distribution. Kruskal-Wallis H test was adopted to compare the effects of slope length on the concentrations and loss amounts of N and P for each land disturbance. Mann-Whitney U test was used to determine the differences in these variables between artificial disturbed and natural restored plots with the same length. Pearson correlation analysis was employed to detect the correlations among the study variables, after these variables were log-transformed (base 10) to ensure the data followed a normal distribution. The statistical significance was specified at P < 0.05 in this study. All statistical analyses were conducted using SPSS 18.0.

A total of eight erosive rainfall events were recorded during the field monitoring period, including 3 rainstorms on May 24, 28, and July 18, and 5 heavy rainstorms on May 13, June 8, 21, 22, and 30, respectively. The maximum precipitation was recorded on May 13 (71.4 mm), and the minimum on June 18 (30.3 mm). The maximum rainfall intensity was obtained on June 8 (6.66 mm h⁻¹), and the minimum on May 24 (3.33 mm h⁻¹).

For all plots, the maximum TN, TDN, NH₄-N and NO₃-N concentrations were almost observed on May 24, while the minimum values of these variables were mainly found on June 22, 30 and July 18. The maximum TP, TDP and PO₄-P concentrations mostly appeared on May 13 and 24, while the minimum values for these variables were detected on June 30 and July 18 (Table 2).

Slope length caused the TDP concentration to differ significantly in naturally restored plots (Table S1), presenting that TDP concentration was 2.07, 1.92 times that in the 60-m plot than those in the 20- and 40-m plots, respectively (Table 2).

Land disturbance only caused the PO₄-P concentration to significantly differ in the 20 m plots (Table S1), showing that the PO₄-P concentration in the artificially disturbed plot was 1.25 fold than that in the naturally restored plot (Table 2).

For all plots, the maximum N and P loss amounts in different forms were almost found on May 13, and the minimum values for these variables were mainly detected on May 24, June 22 and July 18 (Table 3).

Slope length caused the TDN and PO₄-P loss amounts to significantly differ in the artificially disturbed plots (Table S2). Average TDN loss amount in the 20-m plot was 1.56 and 3.06 times than those in the 40- and 60-m plots, respectively (Table 3). Average PO₄-P loss amount in the 20-m plot was 5.22, 4.34 times those in the 40- and 60-m plots, respectively (Table 3).

Natural restoration significantly reduced the N and P loss amounts in different forms relative to artificial disturbance across all plots (Table 3). The naturally restored plots reduced loss amounts of TN by 62.14–79.05%, TDN by 61.24–81.38%, NH₄-N by 54.38–81.39%, NO₃-N by 69.82–81.51%, TP by 79.28–83.43%, TDP by 65.83–81.68% and PO₄-P by 58.05–86.86% in comparison to the artificially disturbed plots across these plots with different lengths, respectively (Figure 3).

There were significant interrelationships among N and P concentrations in different forms (Table 4, Figure S1). Specifically, for artificially disturbed plots, TN concentration showed significant positive correlations with TDN, NH₄-N and NO₃-N concentrations. TP concentration showed significant positive correlation with NH₄-N and TDP concentrations (Figure S1). In naturally restored plots, TN concentration presented positive correlations with TDN, TP, TDP and PO₄-P concentrations. TP concentrations exhibited significant positive correlations with TN, TDN and TDP concentrations, respectively (Table 4). In the artificially disturbed plots, both rainfall intensity and runoff rate showed negative correlations with NO₃-N concentrations, and total rainfall showed positive correlations with TP and TDP concentrations, while showed no significant relationship with any N and P concentrations in different forms in the naturally restored plots (Tables 4 and S3, Figures S2 and S3).

Our results showed complex changes in runoff-associated N and P concentrations in different forms as affected by the land disturbances and rainfall characteristics. The maximum N concentration in different forms occurred under the long duration and soft rainfall events across most plots, while the maximum P concentration in different forms showed the contrasting results in the artificially disturbed plots, and similar results in the naturally restored plots. This was possibly attributed to the distinct loss mechanisms of N and P elements. Previous studies revealed that dissolved nitrogen was the main mechanism for N loss, while particulate phosphorus was the main way for P loss (Daverede et al. 2003; Udawatta et al. 2006). Our results were similar to the findings given by Wu et al. (2018b), who found that higher TDN concentration appeared under the lower rainfall intensity. This was likely because high infiltration and low runoff flow adequately leached soil N under long duration and soft rainfall events, resulting in increase in exportation of soil N to the downstream by the runoff flow (Kleinman et al. 2006). In contrast, the artificially disturbed plots provided rich erosive materials, and the intensive rainfall accelerated the soil aggregates breaking up and increased the exposure of soil P, leading to the increase in runoff-associated P concentration in the long duration and intensive rainfall events (Daverede et al. 2003; Liu et al. 2016; Wu et al. 2018a). While the vegetation coverage pronouncedly reduced the runoff production and sediment yield, prolonged the runoff retention and elevated the leaching export of P in the naturally restored plots under the long duration and soft rainfall events (Sharpley 1981; Kleinman et al. 2006; Liu et al. 2014). The minimum N and P concentrations were found under the latter intensive rainfall events, probably owing to the preliminary dilution of N and P elements that led to the lower runoff-associated N and P concentrations.

Unlike the changes in the concentrations of N and P elements, the maximum N and P loss amounts were all recorded in the long duration and intensive rainfall events across the two land disturbances, due largely to the substantial runoff production. Guo et al. (2017) previously found that the runoff rate presented a positive relationship with rainfall intensity under these two land disturbances, and nutrient loss amounts were greatly dependent on the runoff production (Xing et al. 2016). Naturally restored plots significantly reduced N and P loss amounts in different forms compared to the artificially disturbed plots. This result was consistent with the results of Liu et al. (2014), who found the losses of N and P were negatively related with vegetation cover in the tidal soil, meadow cinnamon soil, and calcareous cinnamon soil, respectively, in the North China Plain. This was probably attributed to artificial disturbance damaging the topsoil structure and resulting in easy soil erosion and nutrient losses under the natural rainfall (Ramos et al. 2019). In contrast, vegetation cover could protect the soil from raindrop splashing, effectively increase the infiltration and reduce the runoff yield, leading to decrease in nutrient losses by runoff (Liu et al. 2016).

Slope length did not cause mostly N and P concentrations and loss amounts to differ in the different forms, except the TDP concentrations in the naturally restored plots and PO₄-P loss amounts in the artificially disturbed plots (Tables S1 and S2). This result was in partial agreement with the findings of Xing et al. (2016) who found that slope length did not cause runoff-associated TN concentration to significantly differ while causing the TN loss amounts to significantly differ in the Kastanozem in the HeLinGe'Er, Hohhot, Inner Mongolia, China. Although most N and P loss amounts in different forms were not significantly different among slope lengths, these variables were in the sequence 20-m > 40-m > 60-m (Table S2). The runoff rate and sediment yield under different slope lengths were the same with the changes in N and P loss amounts in different forms (Guo et al. 2017). However, there were no uniform changed trends in N and P concentrations in different forms under different slope lengths. Further, the variations in runoff rate and sediment yield fluctuated more than in N and P concentrations in different forms under the different slope lengths (Table S1, Guo et al. 2017). This result was in good line with the findings of Xing et al. (2016), who found the slope length significantly affected the runoff rate rather than the TN concentration. These results suggested that runoff amounts played a dominant role in affecting nutrient loss amounts in the erosive rainfall events. Generally, many previous studies have focused on the effects of slope length on the changes in runoff and sediment yield, but not on the runoff-associated nutrient loss, probably due to the complexity of the influencing processes and the diversity of the influencing factors, such as the initial soil nutrient contents, the antecedent soil moisture content and hydraulic residence time, which collectively affected N and P loss amounts under the different slope lengths (Kleinman et al. 2006; Ghahramani & Ishikawa 2013; Liu et al. 2014, 2016). However, given that the runoff-associated N and P are the important sources of the non-point source pollution downstream (Pote et al. 1996; Parris 2011), more data are needed to quantify the slope length effects on the runoff-associated N and P losses.

Our results also showed that runoff-associated TN, TDN, NH₄-N, and NH₃-N concentrations exhibited significant interrelationships in the artificially disturbed plots, while TN and TDN concentrations were both significantly correlated with TP, TDP, PO₄-P concentrations in the natural restoration plots. Runoff-associated N and P concentrations in different forms mostly showed no significant relationships with rainfall intensity, runoff rate, total rainfall and runoff volume in both land disturbances, except NO₃-N, TP and TDP concentrations in the artificially disturbed plots, pointing to the decoupling of relationships between nutrient concentration and driven factors (i.e. rainfall intensity and runoff rate) in this study. The reason was probably the combination of effects of many influencing factors (e.g. coverage, shoots and roots) contributing to the weak linkage between nutrient concentration and rainfall intensity, runoff rate, total rainfall as well as runoff volume. In the previous study, Guo et al. (2017) found that significant relationships between rainfall intensity and runoff rate, and between rainfall intensity and sediment yield were found in the artificially disturbed plots rather than in the naturally restored plots. Significant relationships between NO₃-N concentration and rainfall intensity as well as runoff rate in the artificially disturbed plots might result from potentially consistent changes in NO₃-N concentration and runoff flow. Song et al. (2017) reported that rainfall intensity showed a negative relationship with TDN concentration in bare cropland and uncultivated land covered with grass in the karst area. A similar observation was given by Wu et al. (2018b). This was likely due to the dilution of N concentration under the heightened rainfall intensity and runoff rate. In the artificially disturbed plots, TP and TDP concentrations were significantly correlated with total rainfall, while not with rainfall intensity, indicating that the P loss was dominantly affected by the interaction of rainfall duration and rainfall intensity. In this study, there were significant relationships among TN, TDN, NH₄-N, TP, TDP, PO₄-P concentrations in the natural restoration plots, while there were not in the artificial disturbance plots. This indicated that human disturbance altered the coupling relationships between N and P concentrations in different forms during the erosion processes.

Our results showed that natural restoration pronouncedly reduced the runoff-associated N and P loss amounts in different forms in comparison to artificial disturbance. This was consistent with other studies in the different regions (Liu et al. 2014; Qian et al. 2014), and indicated that human disturbance aggravated the soil erosion and nutrient loss, leading to the increase in the risk of soil fertility decrease and eutrophication in the downstream water bodies. In addition, the results also suggested that N and P loss amounts were determined primarily by the runoff volume rather than the concentrations of N and P in runoff, suggesting that the key strategy for controlling nutrient loss should be reducing the runoff generation and erosion in the study area. Overall, our results will provide useful information for optimizing the measures of sustainable land management and soil and water conservation in the purple soil area.

Field monitoring was conducted to determine the effects of land disturbance and slope length on the concentrations and loss amounts of runoff-associated N and P elements in different forms during the wet season in the purple soil area. The highest N concentration was always found in the long duration and soft rainfall events. The highest P concentration in the artificially disturbed plots was observed in the long duration and intensive rainfall events, and was also observed in the long duration and soft rainfall events in the naturally restored plots. The highest loss amounts for N and P in different forms were mainly observed under the high rainfall intensity. Land disturbances caused PO₄-P concentration to differ in the 20-m plot, and the loss amounts of N and P in different forms across all plots. Slope length caused TDP concentration to differ in the natural restored plot, and TDN and PO₄-P loss amounts in the artificially disturbed plots. Loss amounts of N and P in different forms in the naturally restored plots were dramatically lower than in the artificially disturbed plots. N and P concentrations in different forms presented significant interrelationships in both the artificially disturbed and naturally restored plots, while only NO₃-N concentration showed significant correlations with rainfall intensity and runoff rate, and TP and TDP concentrations showed significant correlations with total rainfall in the artificially disturbed plots. Our results showed that nutrient loss amount is mainly determined by runoff volume rather than nutrient concentration in runoff, implying that the key strategy for controlling nutrient loss should be reducing the runoff generation and erosion in the study area. The results also highlight the dominant role of natural restoration in reducing erosion and nutrient loss. This study will help to understand the mechanisms of nutrient losses induced by natural rainfall and provide useful information for optimizing the measures for soil and water conservation in the study area.

This study was funded by the National Natural Science Foundation of China (U20A20326, 41771312), the Chongqing Science and Technology Commission (cstc2018jscx-mszdX0055), and the State Cultivation Base of Eco-agriculture for Southwest Mountainous Land, Southwest University.

The authors declare that they have no conflict of interest.

Data cannot be made publicly available; readers should contact the corresponding author for details.