Optimizing deficit irrigation and fertilizer application for off-season maize production in Northern Benin

Abstract

Context

Soil water and fertility management have been the main challenges of crop production in West Africa, and their impacts are exacerbated by climate variability. While research has been conducted to optimize fertility and water applications for rainfed crops production in this region, little is known about the management of these resources for off-season cereal crops production.

Objective

This study assessed the optimal combination of irrigation and fertilizer levels for off-season maize production in Benin, using the DSSAT CERES-Maize crop model.

Methods

Two years’ experiments (2018 and 2019) of 4 levels of deficit nutrient (DN) and two years’ experiments (2019 and 2020) of 4 levels of deficit irrigation (DI) were conducted and data were collected on maize growth and yield. DSSAT model was calibrated using crop data from DN experiment in 2018 (DN2018) and DI experiment in 2019 (DI2019), and validated using the DN2019 and the DI2020 experimental data. Then, a long-term scenarios analysis (40-years, 1980–2019) was performed to optimize (i) DI levels, (ii) DN rates; and (iii) combined DI levels and DN rates.

Results

The model predicted the grain yield (GY) and total aboveground biomass (TB), with a relative root mean square error and a coefficient of efficiency of 18.3 % and 0.38 for the GY and 11.7 % and 0.50 for the TB during the validation, respectively. However, the model did not account for the effects of DI or DN on the phenological dates, which led to similar predicted values for the anthesis and maturity dates among DI and DN treatments during calibration and validation. Moreover, the model was sensitive to periods with high values of temperature (>45°C) recorded during the DI period, inducing a reduction of the grain filling rate in DI treatments. DI treatments were more sensitive to a change in DUL, SLL, SAT, RGFIL and RUE than the DN treatments; while the DN treatments were more sensitive to the CTCNP2. Reducing maize water requirements by 40 % at the vegetative stage resulted in similar predicted grain yield as in the full irrigation treatment; while reducing the water requirements by 60 % resulted in similar predicted water use efficiency (WUE) as in the full irrigation treatment. Furthermore, the inter-annual variability of grain yield was lower under the optimal DI combined with no fertilizer but higher under high DI combined with higher fertilizer rates. Finally, a combination of 40–60 % of deficit irrigation at the vegetative stage and one-third to half of the recommended fertilizer rates depending on resources availability was the optimum combination of DI and DN rates for off-season maize production.

Conclusions

The projected grain yield and WUE under optimal DI and DN levels were likely underestimated due to shortcomings in the model structure to deal with effects of water and nutrient stresses on phenological dates. For reliable assessments of the effects of water and nutrient stresses on grain yield and WUE, there is need to update parameterization and code of the CERES crop models in DSSAT to have a sufficiently strong effect of water and nutrient stress on phenological dates, and the contribution of phenology to LAI and yields predictions.