Yield reduction under climate warming varies among wheat cultivars in South Africa

Understanding climate change impacts on staple crops like wheat is crucial for food security, particularly in dryland farming systems of Southern Africa. South Africa, historically a significant wheat producer, faces increasing vulnerability to extreme heat due to recent weather patterns and projected climate shifts. Existing research on heat impacts on wheat yields often lacks empirical, long-term, multi-location data, especially from African contexts. This study addresses this gap by analyzing a large dataset of South African dryland wheat production (18,881 observations across 71 cultivars, 17 locations, and 17 years) to quantify heat impacts and explore cultivar-level differences. South Africa's advanced agricultural sector and breeding programs provide a unique opportunity to investigate the potential for adaptation through cultivar improvement. Historically, South Africa has been the second largest wheat producer in Sub-Saharan Africa after Ethiopia. However, since 1998 wheat acreage has declined due to market deregulation. The country has fluctuated between importing and exporting wheat, playing a critical role in regional wheat prices and food security. Shocks to South African wheat production due to heat stress can affect food security within the country and across Southern Africa. The increasing demand for wheat coupled with stagnant production underscores the need to understand and mitigate the impacts of extreme weather and climate change on wheat yields. Given the limited availability of empirical data for agriculture in Southern Africa, this study's large dataset and analysis provide a foundation for improved understanding of climate change impacts and development of adaptation strategies.

Previous studies investigating weather and heat impacts on wheat yields have primarily been conducted outside of Africa, focusing on small experimental plots, controlled experiments, or biophysical crop models. Empirical studies of wheat in open-air field trials over extended periods and multiple locations in Southern Africa are scarce. This study fills this gap by using daily weather data, a large multi-temporal dryland wheat dataset from across South Africa, and extensive regression models to assess the impact of extreme heat on wheat yields. It also investigates the differential impacts of warming temperatures on various cultivars, providing insights into potential adaptation strategies for producers.

This study uses a comprehensive dataset of 18,881 observations from open-air field trials of dryland wheat production in South Africa. Data spans 17 locations, 71 cultivars, and 17 years (1998-2014), paired with daily weather information (maximum and minimum temperatures, total precipitation) from nearby weather stations. Daily temperature exposure within 5°C intervals was calculated using sinusoidal interpolation, summing intervals across growing seasons. The primary analysis employs a regression model that includes location, cultivar, and year fixed effects; a quadratic polynomial for cumulative precipitation; and eight temperature exposure bins (including a bin for temperatures >30°C). Standard errors are clustered by province-year to account for spatial and temporal correlations. A multilevel model with varying slopes for the >30°C temperature bin was used to assess cultivar-specific heat effects. Uniform warming scenarios (+1°C, +2°C, +3°C) were simulated by increasing daily temperatures and recalculating exposure bins. The Delta Method was used for point estimation of warming impacts. Several robustness checks were conducted, including alternative model specifications, different temperature bin thresholds and sizes, inclusion of low-precipitation indicators, and analysis of different maturity periods. Moran’s I was calculated to assess spatial autocorrelation in yield data and regression errors. The study also examined the interaction of heat effects with cultivar release year and breeder.

The analysis reveals a strong negative relationship between high temperatures and wheat yields. An additional 24 hours of exposure to temperatures above 30°C is associated with a 12.53% yield reduction (t(30) = 40.26, p = 0.000). The marginal effects of temperature show both beneficial and detrimental effects across temperature ranges, with yield reductions above 30°C outweighing benefits from moderate warmth. Uniform warming scenarios indicate substantial yield reductions: 8.5% under +1°C (Delta Method = −3.21, p = 0.001), 18.4% under +2°C (Delta Method = −3.68, p = 0.000), and 28.5% under +3°C (Delta Method = −4.16, p = 0.000). The average effect is a 9% yield reduction per degree Celsius of warming. Interactions between low precipitation and heat exposure were assessed, but the interaction effect was not statistically significant, suggesting the heat effects are robust to variations in soil moisture. The study found substantial heterogeneity in heat effects across cultivars, with more recently released cultivars showing higher average yields but also larger negative heat effects. A hypothetical shift from the cultivar with the highest heat effect to the lowest would reduce yield impacts by 5.5, 11.5, and 17.6 percentage points under +1°C, +2°C, and +3°C scenarios, respectively. This highlights a potential avenue for adaptation through targeted breeding efforts. The five cultivars with the lowest heat effects above 30°C were PAN3118, PAN3144, SST124, SST399, and Tugela-Dn; while the highest were PAN3368, SST367, SST356, Limpopo, and PAN3355.

The findings highlight the significant threat of extreme heat to wheat yields in South Africa and underscore the urgency of adaptation strategies. The substantial yield reductions projected under uniform warming scenarios emphasize the need for proactive measures. The observed heterogeneity in heat effects across cultivars suggests a promising avenue for adaptation through selective breeding and cultivar selection. The potential to mitigate warming impacts by shifting from heat-sensitive to heat-tolerant cultivars demonstrates the value of leveraging genetic diversity. While recently released cultivars exhibit higher yields, their increased sensitivity to heat suggests a trade-off that future breeding programs can potentially optimize. The results also support the need to further investigate the interaction of heat and other climatic factors, especially drought stress, to develop more comprehensive strategies for enhancing the resilience of wheat production in Southern Africa. The study’s large dataset and extensive analysis provide valuable empirical evidence to inform both regional and global efforts to address climate change impacts on agriculture.

This study provides critical empirical evidence on the significant impacts of extreme heat on dryland wheat yields in South Africa. The findings emphasize the urgent need for adaptation strategies, particularly through selective breeding to integrate heat tolerance into high-yielding cultivars. The large cultivar-level differences in heat response highlight the potential for substantial yield improvements. Future research should focus on exploring genotype-environment interactions further, investigating the combined effects of heat and drought stress, and evaluating the potential for other adaptation strategies such as optimized planting dates and irrigation.

The study focuses on dryland wheat production, omitting irrigated wheat, which could influence results. The analysis relies on weather station data, which may not perfectly represent the microclimate conditions experienced by the wheat in the field plots. The study focuses on South Africa; generalizing findings to other regions in Southern Africa requires caution and further research. The model does not explicitly include CO2 effects, potentially affecting long-term climate impact estimates.