Understanding the relative impacts of climate change on crop production using data from the Rothamsted long-term experiments

Crop yields are affected by many variables. In the context of climate change, higher temperatures tend to reduce yield. The Rothamsted Long-Term Experiments on winter wheat (Triticum aestivum) (in Broadbalk), spring barley (Hordeum vulgare) (Hoosfield), and herbage (Park Grass) are some of the world’s oldest continuous agricultural experiments. This Thesis investigates inter-annual variability in yield in the response to climate change, and principally, variations in weather. A multivariate approach to quantify climate change was developed in which 10 different clusters of similar annual weather characteristics from 1892 to 2016 were identified. Most years in the 21st century had their own distinct cluster of a generally warmer climate, which occurred infrequently in the 20th century. FYM treatments of wheat and barley from these warm and dry years had a total biomass of 3.05 and 1.18 t ha-1 lower compared to years from a typical 20th century climate. Between-year variations in temperature and rainfall were associated with variations in the yield of wheat, barley and forage. Warmer temperatures in the early-summer were shown to have a negative effect on the yield of cereal crops. By modelling variations in a Nitrogen response curve, annual yields of wheat and spring barley to Nitrogen were also influenced by variations in rainfall and temperature, where warmer temperatures reduced asymptotic yield of the response to Nitrogen. Simulated wheat yields were estimated to increase by 9.12 to 9.87% from 1892 to 2016 due to rises in atmospheric CO2 when all other variables were fixed, but this effect was largely negated by the actual rise in temperature over this period. The Rothamsted Long-Term Experiment data provided a unique insight into the association between weather and yield and potential mitigations to increase food production. The statistical approaches developed within this Thesis may be applied to other long-term crop-weather datasets