Eucarpia

Eucarpia was an ancient city in Phrygia, near the source of the river Meander whose name has given us the word for an indirect route. Eucarpia took its name from the fertile district in which it lies (“eucarpia” means “well-fruited” in Greek), which also gave its name to a goddess of fertility.

The name has been taken by the European Association for Research on Plant Breeding, which had its annual conference in Edinburgh last week. Rosemary was a plenary speaker, so once again I found myself attending a conference far from my usual subject.

It shared some similarities (and even some delegates) with the conference at COBORU in Poland the week before, but there was a significant contrast. The COBORU meeting was on variety testing, so their inputs come from commercial plant breeders and their role is to test them and recommend some for adoption by farmers. At Eucarpia, although they are well aware that farms are the ultimate destination of their product, the majority of delegates were from the plant breeders. Inevitably there is some tension. Variety testers are required to apply a DUS test (this means “Distinctness, Uniformity and Stability”); the D requires a new variety to have traits signiificantly different from existing approved varieties. One comment from the audience suggested that this put a damper on plant breeding research, since a new variety might be genetically quite different from existing varieties but have similar traits to one of them, and it will not be approved.

So the main jobs that statistics has in this community are teasing out the interaction of genetic and environmental effects, remote recognition of traits in growing crops (possibly using AI), and (a dream of plant breeders) prediction of phenotype from genotype, which could cut a year from the time a new variety takes to be ready to send to the variety testers. (At present you have to grow your seeds and see how they perform, but sequencing a genome is now extremely fast.)

The use of AI here is mostly the more simple “Is it a cat or a dog?” variety, and I have to say its performance is not impressive.

The only thing I heard that really touched on mathematics was a speaker who was working to optimise several traits at once, and so had a multivariate problem to solve; the problem turned out to be ill-conditioned, but the speaker got round this by transforming using orthogonal polynomials. This is not my field either, but I was a bit surprised; the coefficients of the function being optimised depend on the data and so are uncertain, and if the problem is ill-conditioned there will be large uncertainties in the solution, and I fail to see how transformations could fix this.

There was also an interesting talk on testing maize varieties in several African countries. They cut the time taken to develop a variety by getting farmers to do some testing. Various farmers are each given seeds of three test varieties. Instead of measuring the yield, they ask the farmers to rank the three varieties; farmers are able to take other factors which matter to them into consideration. In this particular case, the farmers’ rankings are combined in a numerical scale (making certain assumptions). I would have thought there were advantages to keeping the data discrete a bit longer. There is an interesting problem here. Given a collection of 3-element subsets of a set, and an ordering of each of the subsets, how do you find the ranking on the whole set which best reflects these? Also, what is the optimal block design for doing this?

But there were interesting talks in other areas. One was on the genetics of winter oilseed rape. It seems that a cold spell in winter is required in order for the plants to flower in spring. Climate change means that this is an obvious problem in future. The speaker, a geneticist, had identified the gene responsible, as well as several other genes which now have a role in the process, and a simple cross had produced a line which had high yield as well as tolerance to warm winter weather.

I was a bit startled by one claim the speaker in this talk made. I think he said that the Russian invasion of Ukraine had cost the British oilseed rape industry five hundred million pounds as prices doubled overnight. I believe prices increased, and expected that growers would have been very happy, although I can see that users would have taken a hit. But I remembered my former colleague Ian Percival at Queen Mary, who suffered terribly from hay fever and dreaded the rape flowering season; he used to ask whether anyone had counted the cost to the country in lost productivity caused by the large-scale cultivation of rape.

One of the poster presenters was studying a salt-tolerant cereal which grows in the Colorado delta of the Gulf of California in Mexico. Salination of soils is a big problem in many parts of the world, and this is a simple and practical way that the effect fould be ameliorated in the short term.

I felt a couple of times that the elephant in the room is genetic engineering. At present, the plant breeders do not use it, since the results could not be commercially used and so would not earn money. But a couple of speakers addressed the following problem in ingenious ways. We have a number of varieties, and have identified that some of them have good blocks of genes (for what we require) in various locations: derive a breeding program which is likely to result in a new variety combining all the good blocks. I imagine that a genetic engineer let loose on this would simply cut and paste.

On the free afternoon, we had a guided tour of the Edinburgh botanic garden, with interesting information about its history, about plant hunters in China, and so on. Then in the evening, the Conference dinner in the remarkable Mansfield Traquair Center, a former Catholic Apostolic church (this sect, which no longer exists, had no connection at all with the Roman Catholics).

The meeting was organised by a team from the Roslin Institute; I had initially hoped it would be on their premises, and I could finally get to see Rosslyn Chapel, but it wasn’t to be. The meeting was in the Royal College of Physicians building in Queen Street in Edinburgh New Town. In sharp contrast to the COBORU meeting which had about 2o people, this was a big meeting, with over 260 delegates, and the building was only just able to cope with the numbers, so there were occasionally long queues; but hats off to the organisers who kept it running smoothly. I was moved to see the respect in which Rosemary is held in that community (she began her career in agricultural statistics and still calls an experimental unit a “plot”). So very great thanks to Daniel Tolhurst and his team, who despite all the problems associated with such a big meeting, were able to smooth things for Rosemary.