Our Blog

By Felix Jäger

Studying abstract mathematics, you usually don’t know what this should be good for. Of course, you are told that mathematical thinking is needed in countless contexts and that mathematicians can work nearly everywhere. But, as is so often the case, being told something and experiencing its truth are two entirely different things. Let me describe how I started finding my way into socio-environmental modelling and how what began as an internship, led to my first scientific publication.

Having finished my bachelor degree in 2019 - with a specialization in mathematical logic - I was curious to discover what life beyond university could look like for me. Instead of immediately pursuing a master‘s program, I searched the web for internship opportunities. Among others, I applied to UFZ - and they were not only the first who responded, but also accepted me as an intern for two months within POLISES. The communication was very kind and understanding from the first contact, taking away my insecurities about leaving my familiar environment and starting to work in a new city with new colleagues in a new area. Before starting the internship, I was encouraged to participate in the OESA winter school where basics of ecological modelling are taught to students from many disciplines in a 10-days-program. A great experience where I already got to know the positive atmosphere that defines the work at the OESA department.

My two months in Leipzig confirmed my first impression. From the first day, I was treated as a full integral part of the department, I was free to contribute my own ideas and opinions while at the same time all colleagues were ready to answer any questions. And, in this environment and as time proceeded, I discovered: Yes, I can contribute something. Although this has nothing to do with my studies on the level of contents, my way of thinking is valuable while doing this work. Still, I am very grateful to the OESA and POLISES for making this feeling possible.

But what did I actually do?

The project I worked on was concerned with the adoption of nitrogen management practices in California. Excessive agricultural fertilization is a major problem for ecology, climate and society. Only around half of the nitrogen (N) in the fertilizer is taken up by plants, the remaining N pollutes the environment in different forms. Most important in California is the pollution of groundwater in form of nitrate. In order to reduce N losses and thereby improve water quality, farmers may apply certain management practices developed to increase the share of N taken up by the plants. However, the adoption rates of at least some of these practices are low. The two main questions of the project were: (i) What are the main drivers of the farmers' decision to adopt N management practices? And (ii) What is the effect of potential policy measures aiming to increase adoption rates? In order to answer these questions, we used survey data that was provided by colleagues from UC Davis in California. Farmers were asked about a variety of characteristics potentially related to their adoption decision.

An important feature of the project was the method – instead of using the traditional method of a statistical regression model to analyze the data, we wanted to apply so-called Bayesian Belief Networks (BBNs). These are probabilistic models describing systems of variables based on the notion of Conditional Probability. However, in POLISES no one had worked with BBNs before. Therefore, my main task during the internship was to familiarize myself with this method and start constructing first BBNs based on the survey data.

After finishing my internship, I continued to stay in contact with Birgit, my supervisor. Although I started my master studies in mathematics in Bonn, I was offered the opportunity to continue working at the UFZ part time as a research assistant. Remotely I further pursued the nitrogen management project building on the work of my colleague Martin who had already started preparing a paper draft. The main results: Farm income and structural farm characteristics are the most important drivers of N management practice adoption. The effect of policy measures strongly differs according to the practice under consideration. Moreover, innovative farmers respond better to the policy measures we examined than more traditional ones.

Many revisions later, we were finally able to submit our paper in 2021. Recently, in April 2022, it got published.

I still work for the UFZ and I plan to stay in the field of environmental modelling. With this confidence I look forward to make many more exciting experiences, get to know likeable people and, as every scientist, discover the world we live in.