Can modelling help plant breeders and farmers create and choose varieties to cope with climate change? Andrew Blake reports.
Developing and trialling crop varieties is essential to ensure food security in a changing climate.
That is the spur for an AHDB-sponsored part-time PhD study* by Anisa Aubin of Sussex University. She is also technical solution manager at Agrimetrics.
My research contributes to tackling the problem of ongoing food security in line with changing climate, by evaluating the response of wheat varieties to different climate conditions in the UK, says Ms Aubin.
This builds on a similar approach analysing maize in America.
My research uses a learned model (the one that best fits the data) to predict the wheat yield of a given variety in each location/region given the alternative weather scenarios.
Ms Aubin is training the model using historical varietal yield data alongside historical rainfall, sunshine hours, and temperature. AHDB provided access to trials data records for the Recommended List of wheat in the UK, with each record detailing when and where each variety was planted, its harvest date and yield.
I currently have access to record (see fig A) spanning the years 2002-2017, and it is expected that this will soon be extended to 2022.
I also have access to climate data from the Met Office, and Im building the relationship between the yield and rainfall, temperature, and sunshine hours.
Because the AHDB uses consistent inputs and management practices across all the trial sites Im assuming that the variations in yield will be mainly due to weather, soil, or other external factors, says Ms Aubin.
We see the greatest variation in 2008, 2010, and 2015, with the best yields in 2008 and the worst in 2012. Both summers were in the top five wettest summers, but 2012 had lower temperatures and fewer sunshine hours.
Fig A
Trials data from AHDB, highlighting minimum and maximum yields for each year with the volume of yields centred around the median.
A project named FACYNation, run by the University of Sussex and the Met Office using maize data for the US, was the starting point for Ms Aubins work.
That initial model considered the impact of only rainfall and temperature on yield.
While Ive investigated using additional variables, sunshine hours and humidity, they did not result in the improved prediction accuracy that Id hoped for, so Ive reverted to using just those two variables.
It is early days for this research and thus far, her wheat model has not been as successful as was that for maize in the US, she admits.
That could be for several reasons. The first is that wheat is in the ground for much longer than maize, so there are more factors that affect it. For maize just summer temperature and rainfall were used as predictors, which was sufficient.
Secondly, our trials dont cover a wide enough range of weather and environmental conditions to be able to fully constrain the relationship between wheat yield and the climate.
Thirdly, the maize model drew on much larger data sets than are available via the AHDB, she notes. (See Figs B & C)
Fig B
(Caption: UK average monthly temperature data for the period corresponding to the AHDB trials data. The summer months are warmer (lighter colouring), with July in 2006, 2013 and 2014 being the warmest approaching 18degC.)
Fig C
(Caption: AHDB RL wheat trials data (t/ha) where the darker the green shows better the yield in that county/year compared to the average over all years for each corresponding county.)
The AHDB trial sites data can be noisy due to localised effects.
Regional/county level results for the UK may provide a better representation of the system, says Ms Aubin.
[Panel]
[Best use of data required for resilient wheats]
Maintaining UK food security requires a sufficient and stable supply of safe and nutritious food, says Dr Edward Pope, Science Fellow at the Met Office.
This will be made more challenging by a growing population, a changing climate, and the need to improve sustainability.
Wheat plays a vital part in the UKs diet, environment, and economy, and is likely to remain an important component of maintaining food security for the UK.
Environmental and economic conditions, as well as cultural preferences, influence the location and types of wheat grown here. At present, a combination of weather, soil and orography mean that the most productive wheat-growing areas are East Anglia and the East Midlands.
However, the climate is changing, and has already increased the likelihood of extremes, such as the widespread, record-breaking temperatures seen this summer in the UK.
Because of climate change, achieving stable domestic food production may require changes in where we grow our wheat, and will require new varieties that are resilient to the full range of natural climate variability and the uncertain impacts of climate change.
Breeding new wheat varieties is a challenging and lengthy process that usually spans 8-10 years.
In a changing climate, crop breeders will need to design new varieties that are suited to the expected future conditions, while still being able to accommodate the impacts of natural climate variability.
To address this challenge, its essential to make best use of the available data, alongside scientific and technical understanding. As such, there is a strong incentive for farmers, agronomists, crop breeders and climate modellers to work together to develop resilient wheat varieties.
Anisas work is directly contributing to this aim by combining data science approaches with expert knowledge to extract information about crop-climate relationships, which should ultimately help build a more resilient, sustainable UK food system.
[box]
*AHDB project 21130071 - A model for wheat cultivars and optimisation for climate scenarios Sim Farm 2030 (PhD)
Oct 1, 2020-April 30, 2025
Funding: Total 84,100 (AHDB 74,100 + Quant Foundry 10,000 [tbc])
Others involved: Seb Oliver and Julie Weeds (Sussex University); Jake Bishop (Reading University); Edward Pope (Met Office).
