Potassium management in rice–maize systems in South Asia

Abstract

AbstractPotassium (K) availability in soils is largely governed by their mineralogical composition. The extent of weathering of primary K‐bearing minerals, the chemical pathways through which weathering takes place, as well as the dynamic equilibrium between various K fractions in soils are factors which determine different soil types of varying K‐supplying capacity. The marked variability of K availability in soils in South Asia needs to be taken into account when formulating K‐management strategies in intensive cereal‐based systems in response to K application. Evidence from long‐term fertilizer experiments in rice–rice (R‐R) or rice–wheat (R‐W) systems strongly indicates significant yield responses to K application and negative K balances where K application is either omitted or applied suboptimally. However, K‐fertilizer recommendations in South Asia are generalized over large areas while farmers neglect K application to crops and remove crop residues from fields. These practices may strongly affect yield and soil K‐fertility status in the emerging rice–maize (R‐M) systems in different locations of South Asia. The dry‐matter yield of the R‐M system is usually much higher than that of the R‐R or R‐W system causing high withdrawal of nutrients from the soil. The current review assesses various K forms and K availability in diverse soil types of South Asia supporting rice‐based systems. Aspects considered include: long‐term crop yield and its response to added nutrients, K balance for intensive rice‐based systems, and the role of crop residues in supplying K to crops. Emerging data from either completed or on‐going experiments on the R‐M systems in India and Bangladesh have revealed very high system productivity and variable responses and agronomic K‐use efficiency of maize and rice. Potassium responses of maize are extremely high and variable for soils in Bangladesh. Finally, a plant‐based strategy for field‐specific nutrient management is presented and the need for models and decision support systems for developing efficient K management of the R‐M system is also discussed.