Identifying Drivers for Maize Response to Fertilizer in Ghana
Loading...
Date
2022-01
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
CONTEXT: Maize is the main cereal crop produced in Ghana, but its yield is severely affected by several biotic and abiotic factors. Increasing the overall productivity of maize is essential to ensure food security and lift farmers out of poverty.
OBJECTIVE: Therefore, this study was done to quantify the effect of these factors on maize yield response to fertilizer using 978 data points from on-farm and on-station trials so that intervention practices can be identified.
METHODS: The Quantitative Evaluation of the Fertility of Tropical Soils (QUEFTS) model and five regression algorithms to build the maize yield model – Ordinary Least Squares Regression (OLSR), Multivariate Linear Regression (MLR), Stepwise Multiple Linear Regression (SMLR), and Random Forest Regression (RFR) – were used to analyze the data.
RESULTS AND CONCLUSIONS: The results show that the QUEFTS model cannot significantly explain yield variability at on-station and on-farm levels (adj. R²=3% and adj. R²=22%, respectively). However, the MLR, SMLR, and RFR models explained more than 60% of the variability in maize yield. SMLR and RFR show that the type and rate of fertilizer applied, temperature, variety, and root zone depth are significant factors in explaining maize yield variability. They reveal that soil physical properties explain more of the yield variability (adj. R²=32%) on-station than environmental parameters (adj. R²=1%), with soil chemical properties explaining the highest percentage (adj. R²=36%). On-farm, environmental covariates (adj. R²=33%) explain more of the variability in yield response than physical (adj. R²=25%) and chemical (adj. R²=19%) soil variables. Detailed analytics pinpointed that high temperature and high precipitation, combined with shallow rooting depth (<50 cm), were key factors reducing the efficiency and effectiveness of on-farm fertilizer application. Calculation of nutrient use efficiency shows an average partial factor productivity of 29-80 kg grain and an average agronomic efficiency of 13-46 kg grain per kg of applied N.
SIGNIFICANCE: While fertilizer recommendations are generally based on soil chemical properties, the results of this study indicate that methods to arrive at such recommendations should take soil physical properties and climatic variables into consideration as well.
Description
Keywords
Use efficiency