Matt Castle is a postdoctoral research associate in the Department of Plant Sciences. His work focuses on using mathematical modelling to predict the spread of plant diseases and present possible scenarios for tackling outbreaks.
How did you first get involved in modelling plant diseases?
I initially studied Maths at Pembroke, and later decided to go into mathematical biology. I was interested in how you can use maths to understand how an ecosystem might be stable or unstable. I came back to Cambridge to do a PhD and ended up working on plant diseases; the maths is actually quite similar. Then I became a post-doc here and never left. I like the breadth of what we’re able to do here: it’s not just the maths I’m interested in, although as a mathematician I can say that it’s really fun, but also the application. We can look at the results of our models and give them to ministers or politicians, and we can see our work having an impact.
I really love that in the morning I can be working with coding and just sitting at a computer, and then in the afternoon I’m down in London meeting government officials from DEFRA, discussing how to use the information from that model on a national scale. It’s pretty cool to be able to go from something that’s very abstract to something that’s tangible within one day, so I feel very lucky to be able to do that. With most research, the length of time it takes to see the work having an impact is enormous, and I think it’s quite nice to be able to see the whole spectrum of the work. I don’t know what that says about me!
Can you tell us a bit about your research?
I’m part of the Epidemiology and Modelling Group, which is led by Professor Chris Gilligan. The group looks at plant disease and epidemiology – the spread of infectious diseases, at a national or international scale. We construct models of plant disease epidemics, recreating what people have observed so far and then constructing surveillance plans to determine where we should go and look in order to identify the next possible area of infection. We also look into control scenarios, to help government or international agencies to decide where they should survey and how to best manage whatever it is that’s going on, for example whether they should prioritise any applications of pesticides or any felling of infected trees.
We work in real time. We tend to do a lot of risk-based work with the British government, and we work with DEFRA on a range of things. Two or three years ago, for instance, there was an issue concerning ash dieback, caused by a fungus called Chalara fraxinea which was found in some mature ash trees in East Anglia and Kent. We were contacted a few days after this was brought to the government’s notice. They’d sent out a massive body of people to survey the entire country, and they wanted us to find out where the disease had come from, where it was going to go next, and whether we could eradicate it. We constructed models of the spread of the disease, and spent a lot of time working through the night trying to get information to the government ministers within the next few days. It didn’t feel very academic. We managed to get our work into a government COBRA meeting in the end!
I had to fly to Lithuania the next morning, because the disease had spread into Europe as well. Denmark was losing 90% of its ash trees. We were concerned, because ash in the UK is a native species – it’s one of the most popular trees, one of our main hedgerow trees. There are a few other species which depend on ash trees, so there’s a biodiversity impact as well as an amenity value. But the primary reason, from a national point of view, is that we don’t want to be unable to control our own ecosystem.
When a tree gets infested, the fungus goes into the leaves, and the leaves and then branches slowly die back until there are no leaves left and the tree can’t support itself – which doesn’t look very nice. What’s worse is that it weakens the tree, so it gets affected by other fungi, and the base of the tree starts to rot; so the tree looks a little bit ill, a little bit sickly, then suddenly it falls over. As a result of this, we’ve been working with local councils, the Environment Agency and Network Rail to work out which stretches of road or railway might be affected in future years by falling ash trees.
So our work early on was to find out where the disease came from, how it got here, and where it might go in the future. Then, later on, it was to predict the rate of spread and find out if we can do anything to stop it. Our work was coupled with a genomics project, looking at the genome of ash trees, with the aim of being able to breed more resistant ash trees in the future. The disease is now established in the UK and we’re still looking to mitigate the effects and make sure that ash stays part of the British landscape.
What’s the process for creating these models?
Firstly we try to figure out the epidemiology. We look at the life cycle of the tree, any historical epidemics, and the location of the host trees within the UK. Then we construct a computer simulation of the distribution of host trees. We start out with all the trees on the map being healthy, then adjust the map to reflect where the infected areas are. We can then plot how far the spores or bacteria are likely to disperse, so we can judge the probability of other trees nearby becoming infected. In the model, we then infect some of them, let the infection spread, and we end up with a series of maps showing how the disease spreads. We do that hundreds of thousands of times, so we get the probability – if the trees in one area are definitely infected, what’s the potential risk of those in the surrounding area becoming infected?
Then you can start to look at the potential impact. That’s one of the things we spend most of the time doing – talking to various agencies who want to know how best to spent their budget. For instance, you could spend your entire budget at the beginning of the outbreak and do a massive surveillance programme, or you could spend a little bit upfront, wait and see what happens, and then spend more money later on, once you’ve seen where the disease is going.
There was another example last summer, which hit the news less widely than ash dieback, involving the sweet chestnut trees in Kent. There’s an insect called the Oriental Chestnut Gall Wasp, a tiny thing a couple of millimetres long, which lays its eggs in the buds of sweet chestnut trees. It doesn’t kill the tree but it creates swelling on the buds, which are called galls, and the larvae develop in there and then hatch. It doesn’t kill the trees, but the chestnuts don’t develop. The outbreak in Kent was actually found by people called gall-watchers, or gall-fanciers, who go around looking at the galls and patterns on trees. They reported it, and we found this massive infestation in the area but we had no idea how it arrived – none whatsoever. Then they found another outbreak a hundred miles away, completely isolated, in the centre of a city – and we had no idea how it could’ve got there.
I spent some time initially trying to find out where all the sweet chestnut trees are, and it turns out they’re primarily in Kent. I talked to the Forestry Commission and found out that this was due the lack of clean drinking water in London in the 1700s. There was a huge population in London and they needed a huge amount of beer, because that was the safest thing to drink. To grow enough beer, you need enough hops, and in Kent they were able to grow lots of hops. But you need a wire for the hops to grow up, and to support the wire you need nice straight wooden poles. Sweet chestnut, when you coppice it, grows very straight trunk offshoots. So they used to coppice sweet chestnuts in Kent to make the poles to support the wires to grow the hops, to provide enough beer for people in London.
What does the rest of your work tend to involve?
Because of the work we did for DEFRA and the UK government over the past five or six years, we’ve become effectively the go-to modelling company for the government. They’ve recently set up for plant diseases something called the Quantitative Modelling Standing Capacity framework. They’ve had this in place for animal diseases for about seven years, since the Foot-and-Mouth outbreak. During that outbreak they had several different groups provide modelling reports, as well as the government’s own modelling work. Now we’ve got both animal and plant modelling on a par, so there’s now a similar framework for the plant side of things, and the group at Cambridge is involved in that.
Off the back of our work with DEFRA, I was asked to undertake a review of the UK plant health modelling capability, looking at who else was modelling plant diseases in the UK and at the national strategy for modelling diseases – everything surrounding the national plant health framework. We’ve also spent some time looking at how to study human health and how to study animal health. When you’re modelling for diseases, it’s essentially the same thing. The basic idea is you’ve got a healthy thing, you’ve got an infected thing, and you’re trying to work out how the infection might spread; so the maths is the same, the theory is the same, although the biology is very different. We’ve recently been doing a bit of work with some of the animal health groups, looking at how to integrate these ideas together. When we had the Foot-and-Mouth outbreak, the data that came out of it – it sounds really weird to say it was fantastic, because obviously it was devastating and horrific – but the data that came out of it really helped us to understand how a disease like that spreads on a national scale, and it was very useful for modelling purposes. We’re not concerned with the impact of a disease on an individual, we’re looking at it on a national level, how the disease spreads and how everything links together across the country. So it was a good thing to come out of Foot-and-Mouth, in a way. The application for animals and humans is obviously a bit different from plants, because you wouldn’t just go out and cull an entire city the same way you’d cut down an infected area of trees, but the maths behind the modelling is the same.
We also do some work internationally. One of the things we’re working with is the cassava plant, a tuber-like plant which is a staple food crop in parts of sub-Saharan central Africa. As it’s primarily a subsistence crop, there isn’t a huge amount of investment in creating new varieties of the plant. A few years ago a virus swept through central Africa, the Cassava mosaic virus, which affects the plants by causing mottling on the leaves and making the tubers a bit smaller. This was problematic for the people who were reliant on the crop, but there wasn’t much investment in cassava because, bluntly, we don’t eat it in the west. More recently there’s been a second virus, Cassava brown streak, which actually infects the tubers themselves: they go rotten and brown inside. With the first disease the tubers got smaller, which was bad enough; with this one, they were completely inedible. A lot of struggling families depend on this crop as their last-resort food source, so we need to be able to manage it sensible. The Gates Foundation got involved and decided to try and understand how the disease spreads, to think about what control scenarios might work. Part of this issue is that a clean seed initiative depends on planting healthy plants and then preventing them getting infected, which is very difficult. The virus is spread by white flies which feed on cassava plants, so they can spread the infection from infected plants to healthy plants very easily. So we’re trying to model scenarios of other options, which might provide more manageable solutions in the future.