Potassium nutrient response in the rice-wheat cropping system in different agro-ecozones of Nepal

Most of the soils of Nepal had a higher potassium (K, expressed as K2O) level inherently. Later in 1976, the Government of Nepal has recommended K fertilizer rate at 30 kg K2O ha-1 in rice-wheat cropping systems. However, those crops began showing K deficiency symptoms in recent decades, which could be due to a large portion of soils with depleted K level or the insufficient input of K fertilizer for crop production. This study explored a limitation of K nutrient in the crops by establishing field trials from 2009–2014 at three agro-ecozones i.e., inner-Terai (2009–2010), high-Hills (2011–2012), and Terai (2012–2014) in Nepal. Seven rates of K fertilizer at 0, 15, 30, 45, 60, 75, and 90 kg K2O ha-1 were replicated four times in a randomized complete block design, where crop yields and yield-attributing parameters of rice-wheat cropping system were recorded. Results revealed that an increase in K rates from 45 to 75 kg K2O ha-1 under inner-Terai and Terai conditions and 45 to 60 kg ha-1 under high-Hills conditions produced significantly higher grain yields compared to the recommended K dose. Economically, the optimum rate of K fertilizer should not exceed 68 kg K2O ha-1 for rice in all agro-ecozones, or 73 kg K2O ha-1 for wheat in inner-Terai and 60 kg K2O ha-1 for wheat in high-Hills and Terai. Our findings suggest to increase potassium application in between 1.5 to 2.5 times of the current K fertilizer rate in rice-wheat cropping system of Nepal that need to be tested further in different locations and crop varieties.

a deficiency of K 83 in the field crops. 84 For instance, the yield of the RWCS is decreasing mainly due to a decline in soil fertility in 85 recent decades [20][21]. Yields of the RWCS are also declining due to the incidence of 86 diseases such as rusts, leaf blight, spot blotch [22] and temperatures in the region during 87 grain filling period [23]. Many reports indicated that there is a decreasing trend of soil 88 fertility, including K concentration in soils of Nepal [24][25].
A recent soil test report from 89 soil testing mobile van program in 9 (out of 75) districts in Nepal revealed that around 33% 90 of total sample tested (n= 1479) had soil K at 10-30 kg K2O ha-1 or even less [26]. Farmers 91 use farmyard manure (FYM), mineralized K from FYM is primarily the K input in the soil. The 92 current FYM application rate among Nepalese farmers is about 2.5-3 t ha-1 annually.
But, 93 the quality of manure is inferior [27] that resulted in a very high FYM recommendation rate 94 (20-30 t ha-1), which is far beyond farmers' achievability [28][29]. So, mineral K sources are 95 necessary to replenish the mined K from the cropping soils. 96 A proper nutrient management plan is necessary to sustain yield in the long run with a 97 revision of possibly outdated K fertilizer rate.
Improved productivity of RWCS in Nepal with 98 a revised rate of K fertilizer is also a major concern to keep up with population growth in 99 Nepal, which is predicted to be 36 million by mid-2050 [30]. Previous K recommendations 100 do not consider the environmental K response factor. Here, we select three agro-ecozones 101 to address this shortcoming. Increasing K levels in the K deficient soils may increase crop 102 productivity in RWCS.
So, we aimed to study the yield response and nutrient response of 103 the additional K fertilizer in rice and wheat in three predominant RWCS agro-ecozones 104 (inner-Terai, Terai, and high-Hills) of Nepal. 105 2 Materials and Methods 106 2.1 Study Area and Climate 107 Nepal is divided into five agro-ecozones (Terai, inner-Terai, mid-Hills, high-Hills, and high-108 Himalayas) elevation ranges from 80 masl in Terai to 8,848 masl in high-Himalayas. 109 Cultivation is done up to the 4,800 masl high-Hills [31].
The climate at all agro-ecozones is 110 wet summer and dry winter subtropical. Average annual rainfall ranges from 1000-1800 mm 111 and more than 80% of annual rainfall distributed from June to September, called monsoon 112 period and the winter (November to February), is generally dry with scanty rainfall [32]. 124 During the research period, Kabre site received 2370 mm annual precipitation with 125 maximum rainfall (790 mm) in August and minimum (0 mm) rainfall from December to 126 February. The annual temperature ranges from 26? C (May and June) to 3? C (January). The 127 average maximum and minimum temperatures were 22? C and 12? C, respectively for the 128 region.
Similarly, Rampur site received 2290 mm annual precipitation with maximum 129 rainfall (697 mm) in August and minimum rainfall (0 mm) in November and January. The 130 annual temperature ranges from 37? C (April) to 10? C (December and January). The average 131 maximum and minimum temperatures were 31? C and 21? C, respectively. And, Parwanipur 132 site received 1423 mm annual precipitation with maximum rainfall (311 mm) in August and 133 minimum rainfall (0 mm) in November to January, and April. The annual temperature ranges 134 from 37.5? C (April) to 10? C (January). The average maximum and minimum temperatures 135 were 22? C and 12? C, respectively. 136 2.2 Soil 137 A baseline soil sampling from the top 20 cm depth was collected and analyzed for pH, OM, 138 total N, available P, and available K concentrations. The soil of Parwanipur (Terai) was 139 medium fertile silty clay soil with low in organic matter (OM, 1.1%), medium in total N 140 (0.11%), medium in available P (48 kg P2O5 ha-1), medium available K (165 kg K2O ha-1) and 141 slightly acidic pH (6.2).
Soil samples from the top 20 cm were again collected from experimental 147 plots in 2012 (Kabre site), 2010 (Rampur site) and 2013 (Parwanipur site) to observe 148 changes in pH, OM, total N, available P, and available K concentrations before planting of 149 the main crop. 150 2.3 Experimental Design and Treatment 151 At all three sites, the experimental design was layout in Randomized Complete Block Design 152 with four replications.
Treatment included seven rates of K fertilizer as 0, 15, 30, 45, 60, 75 153 and 90 kg K2O ha-1. The N and P fertilizers were applied at the rate of 100 kg N ha-1 and 30 154 kg P2O5 ha-1 in each plot. Half rate of N, full rate of P, and full rate of K was applied as basal 155 at the time of transplanting and the remaining half N was applied at the tillering stage.
156 Urea, triple superphosphate, and muriate of potash were sources of N, P, and K which 157 contains 46% N, 46% P2O5, and 60% K2O, respectively. All these fertilizers were sourced from 158 a local agro-vet shop. Similarly, rice and wheat seeds were sourced from the Botany Division 159 of NARC, which were the recommended varieties for the regions.
WK 1204 variety of wheat 160 and Ram Dhan variety of rice was selected for high-Hill region and Gautam variety of wheat 161 and Hardinath variety of rice was selected for inner-Terai and Terai regions. Seeds of wheat 162 were directly sown in the field, however, 21 days old seedlings of rice were transplanted in 163 the puddled field in a row-row spacing of 20 cm and hill to hill spacing of 20 cm. 164 2.4 Statistical Analysis 165 Yield and yield parameters of rice and wheat were collected during the growing seasons.
166 Length parameters (plant height and panicle length) were measured by using scale and 167 weight parameters (grain yield, straw yield, biomass, and thousand-grain weight) were 168 measured. Tiller number was recorded with manual counting. Relative yield (RY) was 169 calculated by subtracting control yield from treatment yield (delta yield) and divided by 170 170 171 171 control yield and expressed in percentage. The assumption of analysis 180 of variance (ANOVA) was tested and means were separated at a 5% level of significance 181 using Tukey's test using R-studio.
Time (in years) and location were not combined to analyze 182 as fixed variables, and they were treated individually. A correlation test was carried out 183 between different parameters at a 95% level of significance.  Table S1. There were no significant changes in pH, OM, N and P (except in 190 Rampur site) in addition to K at different rates.
Soil K concentration increased with addition 191 of K rates in Kabre site compared to control. Only K rates at 60 kg K2O ha-1 or higher 192 increased soil K concentration compared to control in Rampur site (Supplemental Table S1). 193 3.2 Yield Response 194 3.2.1 Inner Terai (Rampur) 195 Thousand-grain weight, grain yield and straw yield of rice and wheat increased with the 196 addition of K fertilizer in the inner-Terai condition (Table  1). Results suggested that there 197 was more demand for K fertilizer to obtain a significantly higher yield of rice than RDKF, i.e., 198 15 kg K2O ha-1 recommended rate of K fertilizer.
Analysis of variance (ANOVA) results with means for thousand-grain weight, grain yield and straw yield of rice as affected by different rates of potassium 204 fertilizer in rice-wheat cropping system at NMRP, Rampur, Chitwan, Nepal in three consecutive years. Thousand-grain weight, g Grain yield, kg ha-1 Straw yield, kg ha-1 205 †Treatments included 0 (or control), 15, 30, 45, 60, 75, and 90 kg K2O ha-1.
206 ‡Means in a column followed by same lowercase letter are not significantly different. 207 *P < 0.05,* *P < 0.01, ***P < 0.001 and NS = not significant. 208 Thousand-grain weight (TGW) of rice was significantly different than RDKF in the first and 209 third years from 75 and 60 kg K2O ha-1, but no significant difference was observed in second 210 year (Table 1).
Plant height was affected by K rate 75 kg K2O ha-1 and significantly higher 211 than RDKF, but panicle length and tillers number did not differ significantly with RDKF 212 (Supplemental Table 2). All these yields attributing traits were positively correlated with 213 yield, but a strong positive correlation (r2 = 0.87, p = 0.000) was reported with plant height 214 (Supplemental Table 5).
215 Wheat grain and straw yield significantly differ from RDKF with an additional level of K 216 ( Table 2). The significant yield of rice recorded from 75 to 45 kg K2O ha-1 and wheat from 90 217 to 60 kg K2O ha-1 in the first and second year, respectively. And subsequent decline of 38% 218 in grain yield and 20% in straw yield was observed over years (Table 2). 219 Table 2. ANOVA results with means for thousand-grain weight, grain yield and straw yield of wheat as affected by different rates of potassium fertilizer in rice-wheat 220 cropping system at NMRP, Rampur, Chitwan, Nepal in three consecutive years. Thousand-grain weight, g Grain yield, kg ha-1 Straw yield, kg ha-1 221 †Treatments included 0 (or control), 15, 30, 45, 60, 75, and 90 kg K2O ha-1. 222 ‡Means in a column followed by same lowercase letter are not significantly different. 223 ***P < 0.001 and NS = not significant. 224 224 225 225 226 226 227 Thousand-grain yield was significantly different at 60 kg K2O ha-1 than RDKF. Plant height 228 was significantly different than RDKF with the highest plant height reported from 90 kg K2O 229 ha-1 in the first and 75 kg K2O ha-1 in the second years (Table 2).
There was no significant 230 difference in panicle length and tiller number with RDKF and no significant difference in 231 tillers number with control on the second and third year (Supplemental Table S4). Highest 232 positive correlation was observed between wheat grain yield and plant height (r2 = 0.815, p 233 = 0.000) and panicle length (r2 = 0.743, p = 0.000) (Supplementary Table S5). 234 3.2.2 High Hills (Kabre) 235 Additional K rate had no significant response over RDKF in straw yield and grain yield of 236 wheat in both year and rice in the first year (Table 3). Rice grain and straw yield were 237 significantly different in the second year of potassium application with 30 kg K2O ha-1. Over 238 the years, grain yield of rice increased by 10%, but straw yield declined by 22%. 239 Table 3. ANOVA results with means for thousand-grain weight, grain yield and straw yield of rice and wheat as affected by different rates of potassium fertilizer in rice-240 wheat cropping system at HRS, Kabre, Dolakha, Nepal in two consecutive years. Rice Wheat (2011) Thousand-grain weight, g Grain yield, kg ha-1 Straw yield, kg ha-1 Thousand-grain weight Grain yield, Straw yield, 241 †Treatments included 0 (or control), 15, 30, 45, 60, 75, and 90 kg K2O ha-1. 242 ‡Means in a column followed by same lowercase letter are not significantly different. 243 ** P < 0.01, ***P < 0.001 and NS = not significant 244 Thousand grain weight of the rice not significantly different in the first and second year 245 (Table 3).
Plant height of rice significantly differed at 45 kg K2O ha-1 application in the second 246 year, but tiller number, panicle length of rice and plant height and tiller numbers of wheat 247 were not significantly different with RDKF (Supplemental Table S6). Panicle length of wheat 248 was significantly different at 60 kg K2O ha-1 than RDKF.
But, a significant response of additional K fertilizer was reported from 60 Kg K2O 254 ha-1 in wheat. An incremental yield of rice was achieved up to 60 kg K2O ha-1 but declined 255 afterward. Over the years, yield increased up to 30% and incremental highest yield over 256 control treatment was reported up to 80%.
The addition of K did not increase the thousand-257 grain of rice in both years. A 33% increase in wheat grain yield was reported from 60 kg K2O 258 ha-1 compared to the RDKF. Wheat straw yield was similar (3.7 t ha-1) at 60 kg K2O ha-1 in 259 both years. 260 Table 4. ANOVA results with means for thousand-grain weight, grain yield and straw yield of rice and wheat as affected by different rates of potassium fertilizer in rice-261 wheat cropping system at RARS, Parwanipur, Bara, Nepal. Rice Wheat Thousand-grain Thousand-grain weight, g Grain yield, kg ha-1 Straw yield, kg ha-1 weight, g Grain yield,  Significance NS NS ** *** *** *** *** * *** *** *** *** 262 †Treatments included 0 (or control), 15, 30, 45, 60, 75, and 90 kg K2O ha-1. 263 ‡Means in a column followed by same lowercase letter are not significantly different. 264 *P < 0.05,* *P < 0.01, ***P < 0.001 and NS = not significant. 265 Thousand grain of wheat was obtained significantly from 30 kg K2O ha-1 in the first year and 266 90 Kg K2O ha-1 in the second year (Table 4). In both crops except tillers number other yield 267 attributes traits (plant height and panicle length) significantly differ from control but similar 268 with RDKF.
Significantly the highest tillers number of rice (260 m-2) and wheat (312 m-2) was 269 obtained from 90 Kg K2O ha-1 (Supplemental Table S9). Strong positive significant 270 correlation exists between yield and yield attributing traits (plant height and panicle length) 271 of rice (r2 = 0.82, p = 0.000) and wheat (r2 = 0.76, p = 0.000) but a weak non-significant 272 correlation (r2 = 0.32, p = 0.87) with thousand grain weight (Supplemental Tables S10 and 273 S11).  (Table 5). Physical maximum rate of the K for maximum production was 278 required highest for the wheat in inner-Terai which is 4.8 times the recommended rate and 279 the least for the rice in Terai which is 3.4 times the recommended rate. Similarly, the K 280 requirement for the wheat should be increased by 4.7 times in inner-Terai and 3.9 times in 281 high-Hills and Terai. 282 282 283 Table  5. Physical maximum rate and economic optimum rate of potassium fertilizer recommended 284 for the rice-wheat cropping system in different agro-ecozones of Nepal.
The addition of K rate was expected to increase the yield of rice 290 maximally by 1.8, 0.9, and 1.3 t ha-1 in inner-Terai, Terai, and high-Hills, respectively. 291 Similarly, the residual fertility yield of wheat was 0.6 t ha-1 in inner-Terai and around 1 t ha-292 1 in Terai and high-Hills. The addition of K rate was expected to increase the yield of wheat 293 maximally by 2.6, 1.8, 1.4 t ha-1 in inner-Terai, Terai, and high-Hills, respectively.
The 294 response of K was high in inner-Terai due to the low residual potassium level. 295 295 296 [ Figure 2] 297 Figure 2. Aggregated data analysis to compare the different levels of potassium application on expected yield 298 (represents by bar graph) and relative yield (represents by green line graph) over residual fertility yield 299 (represent by a straight blue line) in rice (Figure 2A, 2C, 2E) and wheat ( Figure 2B, 2D, 2F) in inner-Terai (2A, 300 2B), Terai (2C, 2D), and high-Hills (2E, 2F). 301 4 Discussion 302 Increased yields of rice and wheat at inner Terai and Terai regions of Nepal with K rate at 303 not less than 60 kg K2O ha-1 evident that these regions have low availability of nutrients due 304 to sandy soil, high rainfall intensity, and frequent leaching loss of K [34]. Other findings also 305 reported low availability of nutrients in those regions [5,[35][36].
The soils of these regions 306 are generally characterized as fertile soil, which is made up of recent alluvial deposits mostly 307 fine sand and silt with light to medium texture, and even Terai region of Nepal is considered 308 as the basket of grain [37][38]. Crop cultivation in this region started after clearing the forest 309 in 1927 [39].
A robust agricultural production was witnessed with an alluvial deposit, 310 medium-textured, forest soils. But in later decades, low organic input, nutrient removal due 311 to crop harvesting, heavy tillage work, continuous erosion, and poor crop and land 312 management resulted in low nutrient reserve [40][41]. Hence, K, an active part of the 313 nutrient cycle, is also impacted by low nutrient reserve.
314 In high-Hills, increased crop yields were recorded with an additional K rate in the second 315 year from 30 kg K2O ha-1 not exceeding 70 kg K2O ha-1. Soil in the region is evident with low 316 in OM content, high erosion rate, and sandy loam texture with acidic soil pH [42][43]. 317 Additionally, mid-and-high-Hills are characterized by erosion rate of 32-38 t ha-1 yr-1 [44].
10-kg 319 nitrogen ha-1, 7-kg P ha-1 and 15 kg hectare ha-1. Around 3000 kg K ha-1 has been removed 320 from the soil with the productivity of rice and wheat 7 and 5 t ha-1, respectively [46]. In 321 contrast, rice-growing farmers in the mid-hill regions have benefited from the accumulation 322 of eroded sediments (for example, mica), a source of exchangeable K [47][48][49]. 323 Comparing the K fertilizer consumption under RWCS in the IGP regions, Nepal uses less 324 amount of potassium fertilizer in rice (0.8 kg ha-1) and wheat (1.7 kg ha-1) compared to China 325 in rice (33.2 kg ha-1) and wheat (26.6 kg ha-1). Other countries, such as India uses 7.8 kg ha-326 1 in rice and 3.2 kg ha-1 in wheat, whereas Bangladesh uses 8.3 kg ha-1 in rice and 6.6 kg ha-327 1 in wheat [2]. A long term rice-wheat fertility experiment conducted in Bhairahawa, Nepal 328 showed that annual K balance was negative and K showed a large response in rice [5].
The 329 increase in yield of rice and wheat was evident in all three agro-eco zones of Nepal. 330 Nutrient response functions showed the physical and economic limits of K rate were more 331 or less equal at all locations. The K requirement of the rice should be increased by 4.4, 3.8, 332 3.3 times RDKF in high-Hills, inner-Terai, and Terai, respectively.
In the current study, the 333 economic optima of the K ranged from 50-67 kg K2O ha-1 in rice and 58-71 kg K2O ha-1 in 334 wheat in different locations. The availability of K in soil is proportional to the additional K 335 fertilizer [50], but a higher rate of K addition results in fixation rather than availability [48]. 336 Potassium being a 'luxury consumption' nutrient, sometimes additional K may result in loss 337 [51].
Also, K is a mobile element additional K results in leaching under flooded conditions in 338 case of rice cultivation [49]. So, the current study was unable to attain a more economic 339 yield in 90 kg K2O ha-1. Hence it is not worthy of applying more than 70 kg K2O ha-1 starting 340 from 50 kg K2O ha-1 depending upon crop type and agro-eco zone. 341 At all locations, an additional K rate increased relative yield between 40-50 and 60-80% in 342 rice and wheat. The rice yield was estimated to increase up to 3 to 4.5 t ha-1 and wheat yield 343 was estimated to increase 2.5-3 t ha-1 from 60 kg K2O ha-1. This different K response in rice 344 and wheat is witnessed due to the K limitation in the soil and the availability of K after K 345 fertilizer application.
As the soils of the studied region were high in mica content. In soils 346 containing high mica, even 1-2 % of total K is enough. However, continual crop removal of 347 K and restricted K application in soil for a long time resulted in the weathering of mica to 348 biotite or vermiculite, which is an avenue for K limitation [52]. Similarly, the exchange of K 349 between available and fixed pools is possible.
More than 50% of total K availability in RWCS 350 is obtained from fixed K pools that may further deplete the K reserve from the soil [36]. So, 351 it is imperative to apply K fertilizer to maintain K levels and soil fertility in the cropland. 352 The current study reported that there were no differences in soil total N after a year of 353 addition of K fertilizer, suggesting no adverse effects on total N due to treatment 354 application.
These days, Nepalese farmers are aware of the importance of chemical 355 fertilizers in crop production and the application of chemical fertilizers is gradually 356 increasing. Most of the farmers use only N related fertilizers, which may reach the potential 357 yields. However, it can increase the cost of cultivation in short term and reduction in soil 358 quality and productivity in long term.
A continuous and increased application N fertilizer is 359 not enough to replenish lost plant nutrients and maintain soil productivity [25,53]. So, 360 balanced fertilization is necessary to increase the productivity of RCWS. 361 To summarize, the current study recommends revising the RDKF (15 kg K2O ha-1) to 60-70 362 kg K2O ha-1 in different agro-eco zones.
The use of a single variety of rice and wheat, which 363 was suitable to the soils of research sites, could be a potential pitfall of this study. The K 364 rate might differ for hybrid cultivars of rice and wheat and different soil types. So, a series 365 of experiments are suggested with different varieties of crop in different soil types in 366 different crop rotations. 367 5 Conclusion 368 Improvements in crop yield and soil nutrient response with addition of K fertilizer in RWCS 369 at three agro-ecozones of Nepal suggests that the current recommendation rate of K 370 fertilizer should be increased about 3-4 times for rice and 4-4.5 times for wheat to achieve 371 optimum economic production. These recommendations are made for the improved 372 varieties with the yield potential 2-4 t ha-1 of rice and wheat, which may vary for other high 373 yielding varieties.
Fertilizer recommendation is a dynamic and continuous process that 374 largely depends on soil type, crop response, inherent fertility, grain to fertilizer price ratio, 375 and environment. Recommendations should be revised regularly considering these factors 376 over time. Therefore, K, an integral crop nutrient component contributing to soil fertility 377 and optimum crop production, should be applied in the recommended