Potato Yield Gaps in North Korea and Strategies to Close the Gaps

Abstract

Potato has become one of the staple crops to improve food security in North Korea since the late 1990s. However, the potato yield has been stagnated around 11–12 t ha⁻¹ for several decades, and a food shortage is still a primary issue in North Korea. Yield gap analyses were carried out using the SUBSTOR-potato model to quantify the potato yield gaps and explore the potential ways to close the yield gaps in two different cropping seasons in North Korea (early- and main-season potatoes). Yield gaps were estimated to be around 80% for both early- and main-season potatoes. Early-season potato yield was substantially determined by water or nitrogen supplies, depending on the year’s weather condition (i.e., with or without spring drought). Irrigation during the vegetative stage could effectively reduce the year-to-year variation in yield as well as the yield gap (+7.0 t ha⁻¹, +66.1%). Meanwhile, additional nitrogen fertilizer in the early-season potatoes was less effective compared to that in the main-season potatoes. For the main-season potatoes, where precipitation was sufficient, the primary limiting factor of yield was nitrogen supply. Since heavy rainfall aggravated nitrogen leaching, additional nitrogen fertilizer is recommended as a top dressing rather than a basal dressing. Additional top dressing at 50 days after planting with the current amount of nitrogen fertilizer was expected to increase the main-season potato yield by 42.0 t ha⁻¹ (+191.4%). This study highlights that the primary limiting factor of potato yield may differ between the cropping seasons. Therefore, our findings suggest that different agronomic strategies should be applied for different cropping seasons to improve potato production in North Korea, where agronomic resources are limited.

1. Introduction

Food security has been a primary issue in North Korea for a long period of time. In 2018–2019, the shortage of food was expected to be 1.3 million tons, and 40 percent of the population was estimated to be under the situation of food insecurity [1]. Since the 1990s, the productivities of main cereal crops (maize and rice) were reduced due to the limited supplies of fertilizers, pesticides, plastic sheeting, and fuels, partly attributing to the international restrictions on trade. Considering a high potential of potato (Solanum tuberosum L.) as a staple crop, which produces far more carbohydrate than most cereal crops, the area planted with potatoes has extended from 61,000 ha in 1990 to 215,000 ha in 2018 [2] to improve food security.

According to the global simulation study using a potato model [3], potato yield in North Korea could be around 30–40 t ha⁻¹ with proper crop management. Additionally, the experimental yield in the northern-highland area is reported to be above 60 t ha⁻¹, indicating that climate condition of North Korea is suitable for potato growth. However, actual potato yield is around 11–12 t ha⁻¹, which is well below the abovementioned potential yield and the potato yield in the neighboring country, South Korea, 25 t ha⁻¹ [2]. Closing such a high yield gap is one of the most promising options for addressing the food security challenge, especially in developing countries [4,5].

Process-based crop models can be used to estimate the potential yield, which is the yield of a crop cultivar grown under non-limiting water and nutrient supply with effective controls of biotic stresses [5]. Since some crop models can integrate the effects of climate, soil, and crop management, different kinds of yields can be simulated in addition to the potential yield, such as water-limited yield, nutrient-limited yield, and water- and nutrient-limited yield. These yields can be used to identify the potential environmental constraints of yield. Several studies used crop models to assess the current state of yield gaps and to establish strategies to increase yield [6,7,8,9]. For example, using the LINTUL-POTATO model, Svubure et al. [9] found a huge potato yield gap of 65–92% in Zimbabwe and suggested that the yield gaps can be reduced by improving water, nutrients, and biocides resource use efficiencies.

Despite the importance of potato production in North Korea, the potato yield gap has never been quantified. This study aimed (i) to quantify potato yield gaps for two different cropping seasons in North Korea (early- and main-season potatoes), (ii) to identify the primary limiting factor (e.g., water or nitrogen) for each cropping season, and (iii) to explore potential ways to close the yield gaps. The SUBSTOR-potato model [10], which is one of the most widely used potato models, was used to simulate the potential yield (Y_p), water-limited yield (Y_w), nitrogen-limited yield (Y_n), and water- and nitrogen-limited yield (Y_wn). In addition, the effects of supplemental irrigation and nitrogen fertilizer were simulated to find effective ways to improve potato productions.

2. Materials and Methods

2.1. Study Area

The assessment was conducted for early- and main-season potatoes across North Korea, which lies between latitude 37°–43° N and longitude 124°–131° E. The early-season potatoes are grown in lowland areas (Figure 1), where potatoes are planted from March to April and harvested from June to July (Figure 2a). The main-season potatoes are grown in highland areas (Figure 1), where potatoes are planted from late April to June and harvested from August to September (Figure 2b). For the areas of the early-season potatoes, monthly maximum temperature ranges from −0.2 to 28.9 °C, and monthly minimum temperature ranges from −10.1 to 20.8 °C (Figure 2a). Annual precipitation is 1100 mm, and more than half of the annual rainfall is received in July and August (Figure 2a). For the areas of the main-season potatoes, temperatures and precipitation are much lower than those for the early-season potato areas. Monthly maximum temperature ranges from −7.7 to 24.8 °C, and monthly minimum temperature ranges from −21.5 to 14.8 °C (Figure 2b). Annual precipitation is 800 mm, and around half of the annual rainfall is received in July and August (Figure 2b).

2.2. Yield Data

Potato yield data of North Korea are available from the Food and Agriculture Organization Corporate Statistical Database (FAOSTAT) or Global Information and Early Warning System (GIEWS) [11]. FAOSTAT provides overall potato yield data on a nation level, while GIEWS provides potato yield data for different cropping seasons such as the early- and main-seasons. Potato yield of the main-season is twice that of the early-season, and limiting factors may differ between the cropping seasons. Hence, the yield data from 2009 to 2018 were retrieved from GIEWS rather than FAOSTAT. GIEWS yield data are based on the North Korean government’s official yield data. The official data are further verified by the FAO/WFP Crop and Food Supply Assessment Mission using the field visit data and satellite-based imagery data. The early-season potato yield data are not reported for 2009 and 2018. Thus, the yield data from 2010 to 2017 were used for the early-season potatoes.

2.3. Crop Growth Model

The SUBSTOR-potato model was used for the yield gap analysis, which is part of the Decision Support Systems for Agrotechnology Transfer, DSSAT v4.7.5 [12,13]. The SUBSTOR-potato model simulates phenological developments, dry matter production, and partitioning on a daily basis using daily weather data. The phenological development (e.g., emergence and tuber initiation) is simulated using the functions of temperature and daylength [14]. Dry matter production is calculated based on the intercepted radiation and radiation use efficiency. The computed dry matter production is further modified by the response factors of CO₂, temperature, water, and nitrogen. The potential growth rates of organs (i.e., tuber, leaf, stem, and root) are computed based on the daily weather and phenological stage. The actual growths of each organ are calculated by partitioning the produced dry matter first to tubers and then to the other organs. The SUBSTOR-potato model uses five genetic coefficients of potential leaf expansion rate (G2, cm² m⁻² d⁻¹), potential tuber growth rate (G3, g m⁻² d⁻¹), determinacy for tuber growth (PD, dimensionless), sensitivity of tuber initiation to photoperiod (P2, dimensionless), and critical temperature for tuber initiation (TC, °C). The model has been extensively validated across contrasting growing conditions [14]. The model is also tested with the high-temperature experiments from the adjacent country, South Korea [15]. For the current study region, the model evaluation was carried out using the annual potato yields reported by GIEWS (Figure 3). The yield data from 2011 and 2012 were excluded from the evaluation because the shortage of seed potatoes due to the severe winters caused substantial yield losses [16], which cannot be accurately considered in the simulations.

2.4. Model Inputs

Daily weather data of 27 locations across North Korea, including precipitation and maximum and minimum temperature, were collected from the Korea Meteorological Administration. Solar radiation was estimated using the function of diurnal temperature range, daily average temperature, precipitation introduced by Wu et al. [17]. The coefficients for the solar radiation function were estimated using the historical weather data (2009–2018) from the weather stations (Daegwallyeong, Suwon, Wonju, Incheon, and Chuncheon) in the northern part of South Korea (r² = 0.71; n = 18,259). Default soil (medium sandy loam; IB00000008) was used for simulation since detailed soil data are not available for specific locations in North Korea.

For each location, the planting date was estimated using 10-year (2009–2018) average daily minimum temperature. Since potatoes are generally planted as early as the soil can be worked during the frost-free season, the planting date was set to a day after the last date of the daily minimum temperature below zero. The 27 locations were categorized into the early- and main-season potato areas using the estimated planting dates. If the estimated planting date was within 100 day-of-year (DOY), the location was categorized into the early-season potato area, and the harvesting date was set to 100 days after planting. If the estimated planting date was after 100 DOY, the location was categorized into the main-season potato area, and the harvesting date was set to 130 days after planting. This classification well-represented the actual distribution of early- and main-season potato areas in North Korea (Figure 1).

Planting density was set to 7.14 plants m⁻² [18]. Fertilizer rates were set to the total annual supply of fertilizers divided by the total planted area, which are obtained from GIEWS. For the early-season potatoes, cv. Superior was used, which is a leading cultivar for the early potatoes in South Korea. For the main-season potatoes, cv. Desiree was used, which is one of the cultivars imported to North Korea and a popular maincrop potato in Europe. The genetic coefficients for cv. Superior were collected from Kim and Lee [15]: 1000 for G2, 20 for G3, 0.6 for PD, 0.3 for P2, and 20 for TC. For cv. Desiree, the default values included in DSSAT v4.7.5 were used: 2000 for G2, 25 for G3, 0.9 for PD, 0.6 for P2, and 16 for TC.

2.5. Yield Simulations

The model was run for every year and location to obtain the potential yield (Y_p), water-limited yield (Y_w), nitrogen-limited yield (Y_n), and water- and nitrogen-limited yield (Y_wn). Since supplemental irrigation at the critical growth stage is considered as an effective solution to reduce yield gap [8], simulations with supplemental irrigation during two different growth stages were conducted: water- and nitrogen-limited yields with full irrigation during the vegetative stage (Y_vi; from planting to tuber initiation) and with full irrigation during the reproductive stage (Y_ri; from tuber initiation to harvest). Under the low input systems as in North Korea, crop yield can be enhanced by additional fertilizers and improved fertilizer management, such as split nitrogen fertilization at the early tuberization stage [19]. Therefore, three different N fertilizer applications were tested: water- and nitrogen-limited yields under the current rate of N with basal-top dressing ratio of 1:1 (Y₅₀₊₅₀), under the doubled rate of N with all basal dressing (Y₂₀₀), and under the doubled rate of N with basal-top dressing ratio of 1:1 (Y_100+100). Basal and top dressing were applied at 0 and 50 days after planting (around tuber initiation), respectively.

GIEWS reports potato yields in a nation-level. Hence, nation averages of simulated yields were calculated using the province-level crop area data for the early- and main-season potatoes, which can be obtained from GIEWS. In detail, for each year, simulated yields were first averaged across the locations within each province; and then, nation-level yields were calculated as area-weighted averages. Afterward, eight- and ten-year averages of simulated yields (Y_p, Y_w, Y_n, Y_wn, Y_vi, Y_ri, Y₅₀₊₅₀, Y₂₀₀, and Y_100+100) and reported yields (Y_a) were calculated for the early- and main-season potatoes, respectively.

3. Results

3.1. Yield Simulations

Mean temperature (average of the maximum and minimum temperatures) averaged over the growing season was within the optimum range of potato yield, 14–22 °C [20] and references cited therein], i.e., 16.2 and 16.4 °C for the early- and main-season potatoes, respectively. Total precipitation over the growing season was 259 and 585 mm for the early- and main-season potatoes, respectively. The early-season potatoes achieved a limited rainfall, particularly during the vegetative stage (from planting to tuber initiation, which accounts for 40% of growing season length), with an average of 73 mm across the years and a minimum of 35 mm in 2014. Relatively sufficient rainfall was achieved during the reproductive stage, 186 mm from tuber initiation to harvest. For the main-season potatoes, rainfall was sufficient during the vegetative (141 mm) and reproductive stages (444 mm).

3.2. Model Performance

The SUBSTOR-potato model was able to simulate the potato yield in North Korea (R² = 0.72) (Figure 3a). The model slightly overestimated potato yield in most cases, which is not surprising since it does not simulate the effects of some other reducing factors such as P and K fertilizers, and pests. However, early-season potato yields were under-estimated in the years with severe spring droughts (i.e., rainfall below 60 mm during the vegetative stage in 2014 and 2017). This indicates that minimum irrigation would have been applied during the severe spring droughts, which was not considered in the simulated yields (Y_wn). Thus, by applying the minimum irrigation of 25 mm during the severe spring droughts in 2014 and 2017, R² increased to 0.83 (Figure 3b).

3.3. Yield Gaps

For the early-season potatoes, potential yield (Y_p) was 52.7 t ha⁻¹ with a coefficient of variation (CV) of 8.3% (Figure 4a). Around half of the yield loss could be caused by water shortage, i.e., water-limited yield (Y_w) was 25.9 t ha⁻¹ with a CV of 40.8%. The high CV indicates that severe drought in some years (e.g., 2014), particularly during the vegetative stage (Figure 2a), hugely affected the early-season potato yield and yield variability. Nitrogen-limited yield (Y_n) was 19.7 t ha⁻¹ with a relatively smaller CV of 14.7%. Water- and nitrogen-limited yield (Y_wn, 10.7 t ha⁻¹) was lower than Y_w and Y_n, while the CV of the Y_wn (53.9%) was higher than that of Y_p, Y_w, and Y_n. These indicate that the early-season potato yield is limited by water or nitrogen depending on the year’s weather condition. Reported yields (Y_a, 9.6 t ha⁻¹ with a CV of 22.9%) was lower than Y_wn, which indicates additional limiting factors exist in reality, such as P and K fertilizers, quality and quantity of seed potatoes, and pests. Total yield gap and exploitable yield gap (i.e., yield gap from 80% of Y_p) for the early-season potatoes were 43.1 (81.7%) and 32.5 t ha⁻¹ (77.2%), respectively.

For the main-season potatoes, Y_p was 93.4 t ha⁻¹ with a coefficient of variation (CV) of 3.4% (Figure 4b). Higher Y_p compared to the early-season potato mainly attributes to the longer growing season length, 100 days for the early-season potatoes and 130 days for the main-season potatoes. Unlike the early-season potatoes, the yield gap between Y_p and Y_w was very low, 8.2 t ha⁻¹ (8.8%). The CV for Y_w (9.0%) was much lower than that for the early-season potatoes (40.8%). These results indicate that water shortage may not be a major concern for the main-season potatoes. Meanwhile, huge yield gaps were found between Y_p and Y_n, 72.7 t ha⁻¹ (77.9%), which indicates that nitrogen supply primarily limits the main season potato yield. Interestingly, Y_wn (21.9 t ha⁻¹ with a CV of 21.1%) was slightly higher than Y_n (20.7 t ha⁻¹ with a CV of 16.2%), indicating that the irrigation may enhance the loss of soil nitrogen by leaching. Y_a (17.8 t ha⁻¹ with a CV of 25.0%) was lower than the simulated yields, which indicates some issues may exist in fertilizing P and K, preparing seed potatoes, and controlling pests. The total yield gap and exploitable yield gap for the main-season potatoes were 75.6 t ha⁻¹ (81.0%) and 57.0 t ha⁻¹ (76.2%), respectively.

3.4. Strategies to Close Yield Gaps

The simulation results based on two different irrigation strategies are presented in Figure 5. For the early-season potatoes, full irrigation during the vegetative stage increased Y_wn by 7.0 t ha⁻¹ (66.1%) and reduced CV from 53.9% to 16.1%, while full irrigation during the reproductive stage was less effective (2.8 t ha⁻¹, 26.2%). Meanwhile, for the main-season potatoes, full irrigation during the vegetative stage decreased Y_wn by 1.5 t ha⁻¹ (−6.9%), which may attribute the enhanced nitrogen leaching. Full irrigation during the reproductive stage slightly increased Y_wn by 0.8 t ha⁻¹ (3.6%), but the effect was much lower than that in the early-season potatoes.

The simulation results based on different N fertilization strategies are presented in Figure 6. For the early-season potatoes, doubling the N fertilizer rate increased Y_wn by 5.4 t ha⁻¹ (50.5%), which was further enhanced by splitting N fertilization into the basal-top dressing ratio of 1:1 (8.8 t ha⁻¹, 82.5%). However, under the current N fertilizer rate, split fertilization reduced Y_wn by −1.1 t ha⁻¹ (−10.0%), which may cause nitrogen deficit during the vegetative stage. For the main-season potatoes, the effects of additional N fertilizer and split fertilization were much higher compared to those in the early-season potatoes. Doubling the N fertilizer rate increased Y_wn by 15.0 t ha⁻¹ (68.6%), and splitting fertilization with the doubled N fertilizer rate dramatically increased Y_wn by 42.0 t ha⁻¹ (191.4%). Under the current N fertilizer rate, split fertilization slightly increased Y_wn by 1.5 t ha⁻¹ (7.0%).

4. Discussion

This study is the first study evaluating the yield gaps of the early- and main-season potatoes in North Korea with the consideration of the strategies to reduce the yield gaps. The main outcome is that total yield gaps are around 80% of Y_p for both early- and main-season potatoes. In detail, (i) the early-season potato yield is limited by water or nitrogen supplies depending on the year’s weather condition (i.e., droughts during the vegetative stage), and (ii) it can be effectively increased by irrigation during the vegetative stage. Whereas (iii) nitrogen supply primarily limits the main-season potato yield, and (iv) it can be dramatically increased by additional top dressing of N fertilizer.

4.1. Early-Season Potatoes

Potato is vulnerable to drought due to its shallow root system [21]. The drought-induced losses of tuber yield and dry-matter accumulation primarily attribute to the lowered intercepted radiation caused by the reduced leaf expansion and suppression of branching (i.e., reduced shoot growth) [22]. In South Korea, which is located right below North Korea, limited precipitation during spring is one of the primary constraints for rainfed potato yields. For example, early drought from emergence to tuberization reduced shoot growth, delayed canopy development, and decreased tuber yield [23]. Similarly, in this study, spring drought was one of the primary constraints for the early-season potatoes in North Korea (Figure 2a and Figure 4a). Particularly, in the year with severe early drought and the earliest tuber initiation (i.e., 2014), Y_w and corresponding maximum leaf area index during the growing season were 12.4 t ha⁻¹ (which is less than the half of mean Y_w over the years) and 0.6 (mean value over the years is 3.2), respectively. Early drought is reported to have more pronounced impacts on early cultivars than late cultivars. Poor shoot establishment or leaf shedding by early drought cannot be recuperated for early cultivars since they allocate more dry matter to the tubers and less to the foliage during the early growth stages compared to late cultivars [24,25]. Thus, using late cultivars may alleviate the drought stress by providing more time for shoot growth. However, in this case, tuber yield may not be increased by using late cultivars since the tuber bulking period would be shortened due to the limited growing season length of around 100 days.

Full irrigation throughout the growing season is simulated to increase Y_wn of the early-season potatoes by 85% (Figure 4a). However, full irrigation throughout the growing season may not be feasible since water resources and irrigation infrastructures are limited in North Korea. According to the simulation results, irrigation should be concentrated during the vegetative stage rather than the reproductive stage in terms of yield and its stability (Figure 5a). Additionally, in terms of water resources, irrigation during the vegetative stage of the early-season potatoes may not significantly affect the available amount of water for the main-season/staple crops such as maize and rice, which account for 60% of irrigated areas [26]. Since precipitation is limited during spring, developing and applying effective irrigation technologies, such as drip and subsurface irrigation, or agronomic managements to reduce water loss, such as mulching, will be needed to increase water use efficiency. In addition, the renovation of irrigation infrastructures such as water reservoirs and water pipelines should be prioritized in the lowland areas for the early-season potatoes.

Additional N top dressing is projected to increase the early-season potato yield (Figure 6a). However, supplemental fertilization may not be a realistic solution for improving overall food security since the efficiency of additional N fertilizer is much lower in the early-season potatoes compared to the main-season potatoes, and fertilizers are limited in North Korea.

4.2. Main-Season Potatoes

For the main potato crops in Europe, 150 kg ha⁻¹ of mineral N is required for high yield [27]. In South Korea, 137 kg ha⁻¹ of N fertilizer is recommended for the summer potato [28], a cropping system similar to the main-season potatoes in North Korea. Hence, the current amount of N fertilizer applied in North Korea, which was estimated to be around 90 kg ha⁻¹ (the total annual supply of fertilizers divided by the total planted area, which are obtained from GIEWS), may not be sufficient for the main-season potatoes. Thus, doubling the N fertilizer rate (183.6 kg ha⁻¹) was expected to increase the main-season potato yield by 68.6% (Figure 6b).

Split application of N fertilizer is a common practice to improve crops’ N use efficiency by synchronizing N supply and N demand from crops [29]. Splitting N fertilizer into basal and top dressing could increase tuber yield compared with all N fertilizer applied as basal dressing, particularly in sandy soils with heavy rainfall, where N loss by leaching is pronounced [19,30]. In Harbin, which is located right above North Korea, potato yield under the basal dressing of 100 kg N ha⁻¹ with the top dressing of 50 kg N ha⁻¹ was higher than that under the basal dressing of 150 kg N ha⁻¹ [31]. Similar results were obtained in this study, that the effect of additional N fertilizer of 91.8 kg ha⁻¹ was expected to be much higher when applied as a top dressing, particularly in the main-season potatoes where N leaching is aggravated by heavy rainfall (Figure 2b and Figure 6b). Meanwhile, the effect of split application of N fertilizer with the current fertilizer rate (45.9 + 45.9 kg N ha⁻¹) was not significant (Figure 6b), indicating that the split application of N fertilizer is not recommendable under the low input system as in North Korea since it may limit the vegetative growth.

4.3. Other Reducing Factors

Yield gaps between Y_a and Y_wn (Figure 4) indicate that other reducing factors exist for the potato production in North Korea. The effects of P and K on potato yield are not considered in this study since the SUSBTOR-potato model does not simulate the effects. However, potato yield largely depends on the K supply. Increasing the K fertilizer rate increases tuber yield [32] by enhancing the translocation of carbohydrates into the tubers, which may also increase the leaves’ assimilation rate [33]. Additionally, an adequate supply of P during tuber initiation increases the number of tubers and tuber yield [34]. Since much lower amounts of P and K fertilizers are available in North Korea compared to N fertilizers [1], the effects of P and K should be taken into consideration in the further studies and improvements of crop growth models in simulating the effects of P and K will be needed. Limited quantity and quality of seed potatoes may also reduce potato yield in North Korea, particularly in the years after cold winters when over-winter losses of seed potatoes are enormous [16]. In addition, insufficient pesticide supply due to the international restrictions on trade may also cause significant yield losses due to inadequate pest management. Particularly for the early-season potatoes, heavy rainfall during the harvest season (July; Figure 2a) may also cause decay of tubers and yield loss.

5. Conclusions

This study is the first study that quantified the yield gaps of the early- and main-season potatoes in North Korea and explored effective ways to increase each yield. Yield gaps were 80% for both early- and main-season potatoes, indicating sufficient room for yield increase. Considering the limited agronomic resources in North Korea, irrigation during the vegetative stage and additional N top dressing are recommended for the early- and main-season potatoes, respectively. The results of this study may help in establishing a practical assistance plan for North Korean agriculture.