Determinants of irrigated rice yield in the Senegal River valley

Determinants of irrigated rice yield were studied in 45 rice fields in the Senegal River valley during 2 years in both wet and dry growing seasons. Grain yields were highly variable, ranging from 0 to 9.5 t ha-1, with an average yield over years and seasons of 5 t ha-1, indicating an average yield gap of about 4-5 t ha-1 between actual farmers’ yields and potential yields. Soil P and K status of farmers’ fields were mostly above critical limits and pH and electrical conductivity values were in the optimal range for irrigated rice. Average yield levels for three out of the four seasons were not significantly different. However, cold temperatures at the onset of the dry season induced slower development rate and lower above-ground biomass despite higher plant and tiller density at panicle initiation (PI). Analysis of growth cycle and spikelet sterility levels in the sample fields using the RIDEV simulation model showed that yield losses due to cold or heat stress around flowering could be neglected. Yield variability was, therefore, assumed to be mostly due to differences in crop management (other than choice of sowing and harvest dates) at the farmer level. Plant density in the dry season was significantly higher than in the wet season in both transplanted and direct-seeded fields, with farmers compensating for retarded growth at the onset of the dry season due to cold temperatures. Low transplanting density and the use of relatively old seedlings seriously affected plant and tiller densities determined at PI. Principal component analyses conducted with three variables, i.e. plant density, tiller density and rice--above-ground dry weight determined at PI and yield components at maturity allowed to determine the effect of crop management. Sub-optimal weed and N fertilizer management were major factors driving yield variability in farmers’ fields. Increasing the productivity of irrigated rice will require collective action at both field and scheme level.