In our ‘Science to solutions’ series we talk to Cambridge scientists working with industry, business, NGO and other partners to help solve real-world problems using cutting-edge research.
Rachel Russell is a 3rd year PhD student in Professor Nik Cunniffe’s Quantitative Plant Health group in the Department of Plant Sciences. Her research focuses on optimising the control of invasive plant disease using approaches from the world of robotics and machine learning (ML).
Before her PhD she was an electronics engineer working on electronics to accelerate machine learning tasks at the UK-based semiconductor design company, Arm. She has returned to academia to explore how ML technology can be applied in agriculture and conservation.
Last year Rachel was funded by Cambridge University's Impact Acceleration Account to take some time out from her core PhD work and take part, with Professor Cunniffe, in a European Food Safety Authority (EFSA) working group, contributing to policy-facing work. EFSA is the body responsible for providing scientific advice to the European Union regarding food and, more importantly in this context, plant health.
What real-world problem are you tackling through this partnership?
The European Commission is planning to update their regulations regarding a plant pathogen called Xylella fastidiosa. Xylella is a vector-borne bacterium which affects a wide range of host species, including olive, almond and grape. The largest European outbreak is in Apulia, Italy, where over one-third of their 60 million olive trees have been lost.
Current legislation requires the plant health authorities in each member state to remove (or in some cases, exhaustively test) all the Xylella host plants within a 50 metre radius of every infected plant. This means that the cost of control and the impact of this legislation on the affected landscapes can be significant..
The mathematical models and analyses that informed the existing legislation were developed largely in the context of the outbreak in Apulia. However, the legislation applies to Xylella outbreaks across a wider range of plant hosts and agricultural systems. In some of these settings, disease impacts and patterns of spread appear to differ from those observed in Apulia, which means plant health authorities may face challenges when applying management measures that were originally developed in a different epidemiological context.
To address this, we have built a more flexible model to explore the effectiveness of different control options across a much wider range of hosts. Working with the EFSA team and Xylella experts, we have compared options for updating the legislation to work out which variations are likely to be most effective.
What do you find the most rewarding about this work?
I already knew that I wanted to get more involved in practical modelling work so when Nik mentioned this as an opportunity, I was keen to take part. I was also looking forward to working as part of a bigger team again. A PhD gives you a lot of freedom to explore different approaches and decide what is most interesting but compared to working in industry it can be more solitary, and it is often harder to determine what is important.
What’s the best part of teaming up with EFSA?
EFSA had a clear problem to solve and access to data and wider expertise which we would have struggled to access on our own.
They have longstanding relationships with experts across the field of Xylella research, including specialists on the insect vectors as well as inspectors working in the field and researchers specialising in the development of the disease in the plant. They also have established processes for quantitatively combining that expertise in a justifiable and reasoned way.
From our side, we brought the modelling tools needed to bring the data and expertise together to answer their questions.
What has this partnership taught you that you couldn't learn in a lab?
As part of the work, we were able to take part in field visits hosted by plant health authorities in some of the different regions affected by the disease. I was cynical about this before we did it – I thought it might be a waste of money to send us to look at an olive field.
However, it did change how I thought about the modelling in some very real ways. For example, in Apulia, a lot of the olive trees aren’t in the olive fields at all – they’re everywhere! If you are trying to eradicate Xylella and looking for places where the disease might be hiding, this makes a big difference.
It also gave me a much better understanding of the practical constraints facing inspectors. For example, we visited Occitanie in France where there is an outbreak affecting the natural vegetation. Including inaccessible regions where inspectors cannot reach or remove hosts in a mathematical model feels far more justifiable when you are walking down a long path with impenetrable walls of thorn trees on either side.
What’s the next step for this project?
The report from this work is now being finalised and should go to public consultation later this year as part of an updated EFSA Scientific Opinion. This will form a key part of the evidence base for future updates to the regulations.
Image: Rachel Russell. Credit: Roberto Mecca / University of Cambridge.