EFFECTS OF SOWING DATES ON WHEAT CROP YIELD: APPLICATION OF AQUACROP MODEL UNDER ALTERNATIVE SOIL MOISTURE REGIMES

Abstract

Kenya faces numerous challenges against the need to increase food production to feed an increasing population, especially with the setbacks posed by the current enigma of climate change and declining crop productivity. Among the key challenges especially for the farmers who depend on rain-fed agriculture is the inability to accurately predict the optimal sowing onsets. The result is a false start or delayed sowing which shorten the growing period leading to decline in production and crop loss. While zero tillage can better mitigate the problem as compared to conventional tillage there is scanty literature in Kenya on the practice. Therefore, this study was carried out to determine effects of sowing dates and tillage practices (conventional tillage (CT) and zero tillage (ZT)) on soil water content and wheat crop yield. The evaluation was based on field experiment set up at Lengetia farm in Laikipia East sub-County and simulation with calibrated AquaCrop model. The objectives were: to evaluate soil moisture variation between the tillage treatments; to evaluate the impact of onset dates, conventional and zero tillage practices on wheat crop yield; to calibrate and validate AquaCrop model and determine the optimal sowing date(s) for rain-fed wheat in Laikipia County based on optimization analysis of simulated grain yields. Rain-fed trials and a water regime trial (control treatment) were planted on a split plot with the two tillage treatments (ZT) and (CT) as the main plots factors, and four sowing dates (SD1, SD2, SD3, WTSD2) as the sub-plots factors randomized and replicated in three blocks. The data collected was subjected to statistical analysis (ANOVA and T-tests). An approach based on AquaCrop sowing dates simulation with 19 years historic climate data and optimization analysis of simulated yield in Microsoft Excel Solver tool allowed determination of optimal sowing dates based on probability of exceeding the estimated target yields. The key findings were; there was significant moisture variation at p<0.05 between the tillage treatments. Under the early onset (SD1), normal (SD2) and late (SD3) the yields were significantly different producing higher yields in zero tillage by 48.9%, 20.6% and 34.1% respectively compared to conventional tillage. AquaCrop model was satisfactorily calibrated and validated in conventional tillage using different data sets. The model performance in simulating canopy cover (CC), biomass (B) and soil water content (SWC) was evaluated using statistical indices, RMSE, d, R2 and EF. The value of R2 of 0.95, 0.80, and 0.51 respectively for CC, B and SWC was obtained for calibration in conventional tillage and R2 of (0.88, 0.87, 0.50) respectively for CC, B and SWC in validation. However, calibration under zero tillage was not satisfactory especially in simulating observed soil water content (R2=0.13). This limited the application of the model to conventional tillage only in sowing date optimization analysis. First and second week of October was found to be the optimal sowing period for wheat in conventional tillage while in zero tillage all the dates within the sowing window were optimal as confirmed from field observation. Therefore, zero tillage is recommended for use by farmers as a strategy to improve soil moisture conservation and thus maximize wheat grain yield.