Drought patterns: their spatiotemporal variability and impacts on maize production in Limpopo province, South Africa

Abstract

Due to global climate change, droughts are likely to become more frequent and more severe in many regions such as in South Africa. In Limpopo, observed high climate variability and projected future climate change will likely increase future maize production risks. This paper evaluates drought patterns in Limpopo at two representative sites. We studied how drought patterns are projected to change under future climatic conditions as an important step in identifying adaptation measures (e.g., breeding maize ideotypes resilient to future conditions). Thirty-year time horizons were analyzed, considering three emission scenarios and five global climate models. We applied the WOFOST crop model to simulate maize crop growth and yield formation over South Africa's summer season. We considered three different crop emergence dates. Drought indices indicated that mainly in the scenario SSP5-8.5 (2051-2080), Univen and Syferkuil will experience worsened drought conditions (DC) in the future. Maize yield tends to decline and future changes in the emergence date seem to impact yield significantly. A possible alternative is to delay sowing date to November or December to reduce the potential yield losses. The grain filling period tends to decrease in the future, and a decrease in the duration of the growth cycle is very likely. Combinations of changed sowing time with more drought tolerant maize cultivars having a longer post-anthesis phase will likely reduce the potential negative impact of climate change on maize.

Keywords: Climate change; Crop modeling; Droughts; Maize; South Africa.

Fig. 1

Experimental sites Syferkuil and Univen in Limpopo, SA. Total precipitation (mm) and mean temperatures (°C) monthly climatology (period: 1984–2014). Missing values are shown in gray

Fig. 2

RMSE and MBE for Univen (red) and Syferkuil (blue), calculated based on historical simulations and observed data (1984–2014). The boxplots indicate the errors of the different models for each index and each month (from October to March) considered in this analysis. PRCPTOT represents the total precipitation (mm), DD is the number of dry days (days), LDP is the longest dry period (days), LWP is the longest wet period (days), RX5D is the maximum consecutive 5-day precipitation (mm), and SPI is the standardized precipitation index (-)

Fig. 3

Ensemble 30-year average of the index: a PRCPTOT, b LWP, and c SPI in the Limpopo province for historical simulation (baseline) and future scenario SSP5-8.5 (2021–2050, 2051–2080). Sites are represented with a black circle (Syferkuil), and a black triangle (Univen)

Fig. 4

Number of years with drought conditions (DC) in Univen and Syferkuil. Colors indicate the difference between future scenarios and historical simulation. Shades of red show drier conditions in the future compared to historical simulation, and blue shades show wetter conditions in the future

Fig. 5

Historical and future water-limited maize yield (t.ha − 1), yield gap (Yp-Ywl) (ton/ha), grain filling period (days), and growth duration (days), in Univen and Syferkuil, according to different emergence dates (15 October (DOY 288), 15 November (319), and 15 December (349) days). The dotted grey lines indicate the median values, and the shade of grey indicates the maximum and minimum values of the historical simulation