KNOWLEDGE
CENTRE

Technology [photo] 

Technology

Predicting maize yield potential from available soil water at planting

Louis Ehlers – Manager: Agronomic Services

Introduction

The profitability of maize production in the semi-arid regions of South Africa depends on the efficient use of limited rainfall. Maize profitability can vary widely due to the highly variable nature of rainfall, both in timing and quantity. The most critical growth period for maize is during silking through grain-fill when grain yield is highly dependent on sufficient rainfall or stored soil moisture. Because of these factors, and the high cost of inputs, farmers want a tool that can be used to predict the probability of producing a certain yield before they invest in inputs such as seed, fertilizer and herbicides.

Spatial and temporal variation in crop yield

Crop yield vary spatially due to variation in soil characteristics and temporally due to the interactions of environmental and management factors. The variability caused by different soils is typically high due to the combined effects of physical, chemical and biological processes that operate at different intensities and scales. Good quality data is therefore required to identify the major sources of spatial and temporal variation and to quantify its effect on crop yield. Soil moisture is regarded as one of the most limiting factors to crop production under rainfed conditions.

Response of maize yield to available soil water at planting

Many researchers have shown a relationship between available soil water at planting and subsequent yield for several crops (Table 1). A similar relationship is shown for maize in Bothaville (Figure 1). According to these results, there is always a positive response in crop yield to an increase in available soil water at planting under rainfed conditions. However, these responses are extremely variable (indicated by the coefficient of determination). A logical explanation for this could be that the influence of available soil water at planting becomes less important as the amount of rainfall during the critical growth stages increases. Therefore, any knowledge of the amount of available water before planting, without a reliable forecast of growing season rainfall (amount and distribution), is not sufficient information to adequately predict grain yield.

Table 1: Crop yield response to available soil water at planting

Figure 1: Maize yield response to plant available water at planting in Bothaville

Determination of available soil water

Direct measurement of soil water content is often considered time consuming and costly. However, this method provides a measure of soil water content that is accurate, repeatable and reliable. It is also the standard method against which all other methods are compared and calibrated. This service is now available through OmniPrecise® where soil water content is determined at field scale following the OmniZone™ approach. Soil samples are taken at 30 cm intervals up to a maximum depth of 210 cm at representative sampling points. Volumetric soil water content and texture is determined as non-routine analyses from Omnia’s Chemtech laboratories. Water content at field capacity and permanent wilting point is predicted using pedotransfer functions of the soil’s particle size distribution. Plant available water is calculated as the difference between the actual water content and permanent wilting point. Profile water shortage is calculated as the difference between field capacity and actual water content, expressed as a percentage of field capacity. Figure 2 shows an example of a laboratory analysis report for soil water content.

Figure 2: Example of a laboratory analysis report for soil water content

Prediction of crop yield

A multivariate regression model is used to simulate crop yield for different rainfall scenarios. This model makes use of meteorological variables (rainfall, temperature, solar radiation and humidity), soil properties (soil texture, soil water content and effective depth) and several crop variables (growth period, rooting characteristics and fraction cover) to simulate their impact on the soil water balance and crop growth. As mentioned before, for this tool to be truly useful, access to long-term site-specific rainfall data from a location near (or representative of) a given farm is required. Daily water uptake (as influenced by soil matric stress) is simulated by computing the water supply of a rooted soil profile. As the soil dries, the water supply will decrease until the requirement of the crop cannot be satisfied, which will cause a reduction in water uptake and yield. Therefore, not only the total amount of rainfall, but also the rainfall distribution and the rate at which the soil can supply it to the plant roots is therefore taken into account during the simulation.

Conclusion

Spatial and temporal variation in crop yield confirm the high risk nature of dryland maize production in the central regions of South Africa. The yield response of dryland maize to available soil water at planting varies with rainfall, especially during the critical yield formation growth stage. However, the predictable nature of these responses allow for highly accurate (greater than 75%) estimation of yield potential using long-term rainfall records and the amount of soil water at planting. In specific production areas, every effort should therefore be employed to increase rainfall storage efficiency during non-crop periods.