Introduction

A large proportion of Korea’s landforms are mountainous, with over 80% of the country’s mountains located in Gangwon (Jung et al., 2004; National Atlas of Korea, 2020). Consequently, approximately 70% of the cultivated lands in Gangwon are situated on slopes greater than 7%, with 15% slope gradients being common. This steep terrain significantly increases the potential for soil erosion (Ryu et al., 2010; Kim et al., 2023a). An estimated 80 Mg ha^-1 of soil is lost annually to erosion in the region (Kang et al., 2021), and the recent expansion of cultivated land in the Highlands has exacerbated this issue (National Atlas of Korea, 2020).

Soil erosion is a major driver of fertilizer inefficiencies, as eroded sediments and runoff wash away fertilizer nutrients, making them less available to crops (Bashagaluke et al., 2018; Vaidya et al., 2023). Given the high cost of fertilizers, it is essential to maximize crop yield with minimal input. Therefore, improving fertilizer use efficiency is crucial for enhancing crop productivity while mitigating the adverse environmental effects of fertilizer residues.

In Gangwon, farmers have adopted heavy fertilizer applications to compensate for nutrient losses and improve crop yield (Lee et al., 2010). Kimchi cabbage (KC), one of the region’s major crops (Lee et al., 2024), requires such large fertilizer doses for optimal productivity. Gangwon is one of the leading producers of Korea’s - 2 million tons of KC produced annually (Na et al., 2016; Kim et al., 2022). Despite high fertilizer inputs, KC productivity fluctuates; influenced by season-specific fertilizer inefficiencies associated with slope-enhanced soil erosion and unpredictable weather conditions (Kim et al., 2022; Kim et al., 2023b). This variability in yield contributes to market price fluctuations, creating economic instability. For example, the 14% decline in KC yield in 2021 compared to 2020 resulted in a - 32% increase in KC prices between August 2020 and August 2021 (Statistics Korea, 2021). It is therefore crucial to assess KC fertilizer use efficiency in relation to slope gradients in Gangwon to optimize agronomic practices and stabilize KC yields.

In recent years, slow release fertilizers (SRFs) have been recommended for the region. These fertilizers are designed to release nutrients gradually, matching plant nutrient demands more effectively over time (Callahan et al., 2005). As a result, SRFs are expected to experience less nutrient losses through erosion compared to fast-release fertilizers. Additionally, SRFs are known to improve macronutrient utilization, conserve energy, and reduce environmental pollution (Priya et al., 2024). In Gangwon’s highland agriculture, SRFs have reduced nutrient losses into the environment and promoted KC production, improving fertilizer efficiency (Brempong et al., 2024). Over the past three years, SRFs have consistently provided similar or higher KC yields compared to fast-release fertilizers (Kim et al., 2023b; Brempong et al., 2024). Elsewhere, SRFs have been shown to reduce nitrogen (N) losses and improve nitrogen use efficiency (Shivay et al., 2001; Giller et al., 2004). For example, Wang et al. (2020a) reported 9.8% increase in N use efficiency with SRF application compared to conventional fast-release fertilizers. As a leafy vegetable, N use is a very important factor affecting KC yield and productivity (Leghari et al., 2016), therefore, our study focused more on N efficiency.

Given these findings, we hypothesized that SRF application would improve N use efficiency and KC yield across all slope gradients, compared to fast-release fertilizer. We also hypothesized that N use efficiency and KC yield would be generally lower on sloped land compared to flat land due to the greater potential for soil erosion. The objective of this study was to evaluate the N content, uptake, and use efficiency, and yield parameters of KC on sloped versus flat land in Gangwon’s highland agriculture.

Materials and Methods

Study site description

Two experiments were conducted during the 2024 growing season at the Highland Agriculture Research Institute (37°40'48.0"N, 128°43'51.0"E) in Gangwon, South Korea. The experiments were carried out in two distinct land conditions: a lysimeter with a 15% slope and flat land (0 - 2% slope). The 15% slope lysimeter simulates the most common slope gradient found in cultivated lands across the Gangwon region. The slope lysimeter was constructed based on the standards outlined by the Universal Soil Loss Equation (USLE) developed by the United States Department of Agriculture (USDA). It has a simulated slope area measuring 2.8 meters in width and 22.4 meters in length, and is filled with clay loam soil. Soil texture in the flatland was loam.

During the KC cultivation period, the study area received 322.5 mm of rainfall and had summer temperatures averaging 20°C (Fig. 1). These weather conditions were conducive for KC cultivation in the year.

Fig. 1.

Average monthly precipitation (mm) and average monthly temperatures recorded in the study sites in 2024. Rainfall and temperature data were obtained from the Korean Government’s Weather Station in Daegwallyeong, Pyeonchang-gun in the Gangwon State.

Initial physico-chemical analysis of soil before experiments

Six soil samples were collected from each experimental site using an auger to a depth of 15 cm. The top 1 cm of each core was removed to eliminate plant debris and organic materials. The samples from each site were then composited, air-dried, and sifted through a 2 mm sieve. Standard soil analysis protocols, as prescribed by the National Academy of Agricultural Science (NAAS, 2011), were followed.

Soil texture (particle size distribution) was determined using the hydrometer method. Soil pH was measured at a 1:5 soil-to-distilled water ratio using a pH meter (Orion Versa Star, Thermo Scientific, USA). Soil electrical conductivity (EC) was measured by an EC meter (Orion Star A212, Thermo Scientific, USA) at a soil to distilled water ratio of 1:5, after which the values were multiplied by a dilution factor of 5. Total carbon (TC) concentrations were analyzed with an elemental analyzer (Vario Max, Elementar, Germany). Available phosphorus (P) was quantified calorimetrically using a UV-VIS spectrophotometer (Lambda 365, Perkin Elmer, USA) at a wavelength of 720 nm. Exchangeable cations were extracted with 1 N mono-ammonium acetate and measured using inductively coupled plasma (ICP) spectroscopy (Avio 550 Max, Perkin Elmer, USA). Table 1 presents the soil properties before the start of the experiments. Based on the analysis, the pH, available P₂O₅, K, Ca, and Mg levels in the slope lysimeter were below the recommended thresholds for Kimchi cabbage (KC) production, while the total carbon (TC) and electrical conductivity (EC) were within the recommended ranges (Table 1). In contrast, for the flat land, pH, TC, K, Ca, Mg, and EC were all within the recommended ranges for KC production (Table 1).

Experimental design and treatments

The experiments were conducted in 2024 on a 15% slope lysimeter and flat land. The slope lysimeter and flat land were prepared three weeks before transplanting KC. Quick lime was added to the soil at a rate of 1,130 kg ha^-1. The treatments for the experiment included: Fertilizer type (Polymer coated slow-release fertilizer, fast-release fertilizer and ‘no fertilizer) and slope gradient of the land (15% slope lysimeter and flat land). Polymer coated slow release fertilizer (SRF) was applied at 330:108:108 kg ha^-1 NPK (with micronutrients), conventional fast-release fertilizer (FF) at 330: 42: 99 kg ha^-1 NPK and ‘no fertilizer’ (NOF). Fast-release fertilizer was applied in three splits at basal application (114: 42: 55 kg ha^-1) and at 20 and 40 days after transplant at the rate of 108: 0: 22 kg ha^-1, respectively. SRF treatment was 1.4 times the N recommendation, while FF was 1.4 times the NPK recommendation (Adapted from NIAS, 2022). SRF was applied in one event at the basal application. The treatments were replicated three times on the 15% slope lysimeter and four times on flat land.

Kimchi cabbage planting, harvest and measurementsIn the slope lysimeter, Kimchi cabbage (KC) was transplanted on June 10, 2024, at a spacing of 70 cm × 35 cm, and harvested on August 1, 2024. The lysimeter was mulched with black polyethylene plastic to suppress weed growth. From each slope, ten representative plants were harvested for analysis.

Similarly, on the flat land, KC was transplanted on June 10, 2024, with the same spacing of 70 cm × 35 cm, and harvested on July 31, 2024. The flat land was also mulched with black polyethylene plastic. Ten representative plants were harvested from each plot for data collection. The following data were recorded for both experiments: plant weight, head weight, head height, head width, leaf number, longest leaf length, leaf width, and marketable yield.

Soil loss measurements

Soil loss measurements were made on rainfall events in the15% slope lysimeter. On these dates, runoff water mixed with eroded soil passed through the splitter, which collected water and soil in the ratio of 1:100. Collected soil was dried, quantified and extrapolated to kg ha^-1. Soil loss on the flat land was negligible and therefore, was not measured. In total, slope lysimeters receiving SRF, FF and NF lost 140.31, 158.32 and 224.65 kg ha^-1 soil over the KC growing period.

Nitrogen content, uptake and use efficiency

KC samples were analysed in the laboratory for nitrogen content (%) of the dry matter. Values obtained were used to calculate nutrient uptake as follows:

Results for KC nutrient content and uptake for this experiment are presented in Table 3.

Nitrogen Use Efficiency (NUE)

Nitrogen use efficiency of the treatments were calculated as follows:

Statistical analysis of dataAll collected data were subjected to analysis of variance (ANOVA) using IBM SPSS Statistics 20. ANOVA was first run for fertilizer types (SRF, FF, NF) separately in each slope gradient (15% slope lysimeter or flat land). Secondly, both fertilizer type and slope gradient were treated as factors and run as Factorial ANOVA to determine their main effects and interactions. Statistically significant treatment means were separated with Fischer’s Least Significant Difference (lsd) at 5% probability. T-test analysis was done for statistics on NUE in either slope lysimeter or flatland, as only two treatment groups (SRF and FF) were analyzed. Levene’s test for equality of variances were applied.

Final soil sampling and chemical analysis

Six soil samples were collected from each plot after harvest with an auger at 15 cm depth. Samples from plots in either flatland or slope lysimeters were composited and analyzed for pH, total C, Av. P₂O₅, Exchangeable cations (K, Ca, Mg) and EC, following the procedures outlined under ‘Initial physico-chemical analysis of soil before experiments’ section above. Soil properties after harvest are presented in Table 5.

Results and Discussion

Kimchi cabbage production in slope lysimeter and flat land

In either slope lysimeter or flatland conditions, the application of SRF resulted in comparable values for plant fresh weight, head fresh weight, head height, head width, leaf length, leaf width, and marketable yield to those observed in plots treated with FF (Table 2). Across all parameters, SRF and FF treatments consistently outperformed the NF control (Table 2). Overall, SRF and FF demonstrated similar performance across all Kimchi cabbage (KC) production parameters, regardless of land gradient. However, plant fresh weight and head height were higher in the flatland compared to the 15% slope lysimeter (Table 2). The comparable performance of SRF and FF in both land gradients suggests that nutrient availability was similar between the two, despite the higher phosphorus (P) and potassium (K) doses supplied via SRF. Since both treatments provided equal amounts of nitrogen (N), our findings reinforce the conclusion that N availability is a primary determinant of yield in leafy vegetables such as Kimchi cabbage, as also reported by Leghari et al. (2016). A significant interaction was observed between fertilizer type and land gradient for plant weight (Table 2), indicating that fertilizers generally enhanced fresh plant weight more effectively in flatland conditions than on sloped land. The interaction effect can likely be attributed to greater water uptake by KC plants and better baseline soil chemical properties (Table 1) in flatland conditions. While SRF and FF did not show significant differences in their main effects on fresh plant weight (Table 2), enhanced soil water infiltration and reduced runoff in flatlands likely improved water availability to plants, unlike the sloped terrain (Parihar et al., 2023). The lack of statistical differences between KC yield in the slope lysimeter and flatland suggests that yield potential was not fully limited by the slope but perhaps other factors such as water availability.

Nitrogen content and uptake

The effects of SRF and FF on N content and uptake were similar in both the slope lysimeter and flatland conditions, and both were significantly higher than the NF treatment (Table 3). Given that a slightly higher N dose (1.4 times the recommended rate) was applied in both SRF and FF treatments, it is likely that any N losses associated with the different nutrient release mechanisms of the two fertilizer types were offset. This aligns with Baldwin (2006), who noted that the quantity of diffusible nutrients plays a critical role in plant nutrient uptake. In our study, total soil losses during the Kimchi cabbage (KC) growing period in the 15% slope lysimeter were 140.31, 158.32, and 224.65 kg ha^-1 under SRF, FF, and NF treatments, respectively. In contrast, soil losses in the flatland were negligible (data not shown). Although soil erosion typically results in offsite nutrient loss (Meena et al., 2017), these losses did not significantly affect N content or uptake by KC plants. There were no significant differences in N content and uptake between the slope lysimeter and flatland, suggesting that the observed soil losses in the sloped plots were not substantial enough to reduce soil N availability. This is further supported by the fact that the recorded losses were far below the 89 Mg ha^-1 reported by Kim et al. (2023a) under KC production in the same region.

Nitrogen use efficiency

In the slope lysimeter, NUE under SRF was comparable to that under FF (Table 4). This outcome was expected, as both SRF and FF supplied nitrogen at 1.4 times the recommended rate, resulting in similar N content and uptake by KC plants in the lysimeter (Table 3). Similarly, in the flatland, NUE under SRF did not differ significantly from that under FF, again reflecting the comparable N supply and uptake across these treatments. However, overall NUE was significantly lower in the slope lysimeter than in the flatland (Table 4). Although N uptake was similar between the two terrains, slightly lower KC yields in the slope lysimeter (based on absolute values) contributed to reduced NUE in that setting. This suggests that factors other than N—such as soil moisture, starting soil conditions or erosion—may have influenced crop growth on the slope. Considering that the initial soil in the flatland was relatively more fertile than that in the slope lysimeter, with higher levels of available P₂O₅, K, Ca, and Mg (Table 1), it can be inferred that several other growth-promoting factors were also more favorable to KC yield and NUE in the flatland. These nutrients play critical roles in nitrogen metabolism and nitrate absorption, ultimately enhancing NUE (Aulakh and Malhi, 2005). (The higher NUE observed in the flatland may be attributed to a more stable soil environment that favors nutrient retention and uptake (Quinton et al., 2010). These findings highlight the need to account for both nutrient input and environmental conditions (e.g., slope, erosion risk) when developing fertilizer management strategies aimed at optimizing crop productivity and sustainability.

Soil chemical properties after KC harvest

No significant differences were observed among the SRF, FF, and NF treatments across all soil chemical properties in the slope lysimeter after harvest (Table 5). On the flatland, a similar trend was observed for most soil properties, except for pH and EC (Table 5). However, SRF and FF influenced soil chemical properties differently depending on the land gradient. Compared to the baseline values (Table 1), SRF and FF treatments reduced soil pH to below 5.6 in the slope lysimeter, while pH was either maintained or increased in the flatland (Table 5). This decline in pH on sloped land may be due to fertilizer-induced acidification and increased erosion-related loss of base cations—common in sloped terrains (De et al., 2008; Wang et al., 2020b). This is supported by the observed post-harvest K losses under SRF and FF in the slope lysimeter (Table 5). The flatland showed more stable soil pH conditions, likely due to better buffering capacity and reduced erosion. Despite higher P application through SRF, post-harvest P levels in the slope lysimeter were comparable to the low baseline values (Table 1) and remained lower than in the flatland (Table 5). This may result from increased P fixation under acidic conditions and surface runoff in the slope lysimeter, (Ch’ng et al., 2014) or luxury P consumption in the flatland (USDA, 2023).

Soil nutrient levels under NF remained stable or increased after harvest in both terrains (Table 5), possibly due to lower plant nutrient uptake and residual fertility. SRF minimized P losses compared to initial levels, whereas FF caused greater P depletion, likely due to its lower P input. Elevated soil EC under both treatments and land gradients may result from nutrient accumulation through mineralization (Mohanavelu et al., 2021).

Conclusion

This study evaluated the fertilizer efficiency of SRF and FF in enhancing Kimchi cabbage production and N use across different land slope gradients in the Highlands of Gangwon. Both SRF and FF significantly improved KC growth and yield on both the 15% slope lysimeter and flatland, outperforming the no-fertilizer control. Despite variations in P and K inputs between SRF and FF, their similar nitrogen supply led to comparable outcomes in N content and uptake, fresh plant weight, head weight, height and width, leaf characteristics, and marketable yield. It confirmed nitrogen’s dominant role in KC productivity. Flatland consistently supported higher fresh plant weight, head height and width, and NUE, likely due to more favorable soil moisture retention and nutrient dynamics compared to the slope. While some soil and nutrient losses occurred on the slope, N availability and uptake were not significantly affected in the short term, suggesting that fertilization can mitigate immediate productivity losses. However, the reduced NUE under slope conditions suggests long-term risks of nutrient inefficiency and proves the need for integrated soil, nutrient and water conservation strategies. Flatland systems were also more effective in buffering soil pH and minimizing some nutrient losses, making them more resilient. Phosphorus management on sloped lands is especially critical, as runoff and increased acidity exacerbate P losses.

However, the potential for EC build-up in slope-SRF systems likely due to nutrient accumulation in lower landscape positions warrants long-term monitoring. Future research should focus on the cumulative effects of repeated fertilizer application under varying topographies, with emphasis on soil health, water availability, erosion control, and sustainable nutrient cycling.