Management choices in cereal production: Lessons from the Broadbalk Long-term trial

The Broadbalk Wheat Experiment provides information essential for our understanding of the long-term viability of agricultural systems. It demonstrates that it is possible to grow wheat on this soil type with either organic manure or inorganic fertilisers for >180 years. Started by Lawes and Gilbert on a flinty silty clay loam at Rothamsted in 1843, treatment strips are not replicated or randomized, but are large, which allowed later sub-division. Treatments varied in the first eight years, but from 1852 - 1967 remained largely unchanged. They included organic manures, combinations of fertiliser nitrogen (N), phosphorus (P), potassium (K), magnesium (Mg) and sodium (Na) and a control receiving neither fertilisers nor manures. Results from the early years of the experiment showed farmers how to increase their yields and helped lay the foundations for modern agriculture. Later, using data from Broadbalk, Fisher revolutionised statistics with new concepts on the analysis of variance and experimental design. Modifications have been necessary to ensure the experiment’s continuation. Division into five sections in 1926, enabled regular fallowing to control weeds and in 1957 herbicides were introduced. To maintain soil pH, differential chalk applications began in 1954. The relevance of Broadbalk to current agriculture has been critically reviewed several times and in 1968, major changes were made. The five existing sections were divided to create 10 sections; some reverted to continuous wheat while some included other arable crops in rotation with wheat. Wheat varieties changed from long-strawed to short-strawed cultivars with a greater grain yielding potential. Larger N rates and fungicides have been used to exploit the potential of newer varieties. Following the change to short-strawed cultivars average grain yields of continuous wheat given sufficient NPK doubled. Yields of the 1st wheat after a 2-yr disease break have been much larger than those of continuous wheat. With cv. Crusoe, yields of the 1st wheat have exceeded 12 t ha-1 and were, on average, >3 t ha-1 larger than those of continuous wheat. In autumn 2017 the rotation changed to beans, wheat, wheat, oats, wheat. Mean maximum yields of cv. Zyatt were c. 7.9, 9.2 and 9.9 t ha-1 for continuous wheat, 1st wheat after oats and 1st wheat after beans, respectively. Losses of N and P in drainage and the use of archived crop and soil samples to quantify atmospheric inputs, to track changes in the mineral composition of grain and to help construct models describing the turnover of soil organic matter are discussed. We highlight the value of the electronic database which contains a large amount of crop and soil data, together with analyses and background information, making it a valuable resource for researchers.