Hauptinhalt

Topinformationen

32nd Systems Science Colloquium

Winter semester 2025/26

The colloquium takes place on Wednesdays from 16:15 to max. 17:45, in room 66/E01 (Barbarastraße 11).

Program

17 October 2025

Dr Marianna Cerasuolo, School of Mathematical and Physical Sciences, University of Sussex (UK)

Mathematical Insights into Soil Carbon and Sustainable Agriculture

Climate change affects various aspects of agricultural production, including fluctuations in temperature and precipitation that impact nutrient cycling, soil moisture, pest occurrences, and plant diseases. In response, crop diversification and a range of management practices are being explored as methods to address these challenges. The amount of carbon in soil is often used as a proxy for soil health, and increasing the capture and storage of carbon dioxide has the potential to positively impact climate change mitigation. However, soil carbon dynamics are slow, and data-informed mathematical models are necessary to support research in agriculture, such as predicting the impact of innovative cropping strategies.

In this seminar, I will present results from the research undertaken within the Diverfarming project (www.diverfarming.eu), which aimed to promote sustainable agriculture in Europe through crop diversification and reduced-input farming. A central aim was to apply mathematical and statistical methods to evaluate the environmental and economic consequences of diversified cropping systems. Our analyses showed that while such systems may yield less in the short term, they often enhance resilience, improve soil function, and lead to comparable or improved profitability over time. Using the experiment-based mathematical model ECOSSE-M, developed within the project, we examined the effects of diversified cropping systems across six European countries, considering soil–water–atmosphere– plant interactions at both farm and landscape scales. In particular, I will focus on the Po Valley in Northern Italy, where mathematical, statistical, and machine learning approaches were employed to investigate the spatial dynamics of soil carbon in intensive agricultural systems. This allowed us to quantify the effects of different management practices and identify key drivers of carbon storage.

Our results provide science-based evidence on the role of crop diversification in improving soil health and mitigating climate change, and demonstrate how sustainable agricultural practices can enhance resilience, reduce environmental impacts, and support long- term ecosystem health.

22 October 2025

Dr. Christoph Schünemann, Leibniz-Institut für ökologische Raumentwicklung (IÖR) Dresden (Germany)

Integrating Social Complexity in Policy Advice to better understand Transformation Processes

Traditional policy modelling often relies on rational economic assumptions, overlooking real-world decision-making complexities influenced by psychological, social, and systemic factors. Existing psychological or socioeconomic decision-making and behavioural change theories provide insights into how cognitive biases, social identity, values and norms or bounded rationality affect behaviour. Despite their relevance, these theories are underrepresented in policy assessment and modelling, often due to their conceptual fragmentation and the lack of psychological expertise in the policy development process. To address this gap, the use of systemic behavioural change modelling is proposed, combining psychological decision theories with complexity science, i.e. system modelling approaches. A case study explores this approach by developing a systemic decision-making model for homeowners’ decisions to adopt energy retrofit of their houses. The model builds on the Rubicon Model and maps decision-making phases—pre-decisional to post-actional—while integrating societal influences like media and social norms in several feedback loops.

05 November 2025

Dr. Judith Klein, Fraunhofer-Institut für Molekularbiologie und Angewandte Oekologie IME (Germany)

From Molecules to Models: Computational Approaches for Ecological Risk Assessment

Assessing the environmental risks of chemicals requires an understanding of both exposure and effect processes. This presentation provides an overview of key modelling approaches used for environmental risk assessment—from exposure models such as FOCUS PELMO and GeoPELMO, which estimate chemical concentrations in groundwater, to effect models like the GUTS model that describe survival of organisms under dynamic exposure conditions. The integration of mechanistic modelling, bioinformatics, and geospatial data enables more realistic, science-based evaluations to support regulatory decision-making and sustainable environmental management.

12 November 22025

Prof. Dr. Andreas Beyer, Universität zu Köln (Germany)

Analysis of molecular high-throughput data to understand cellular systems

Cells are very complex systems that are composed of billions of highly diverse biomolecules. These biomolecules execute enormously complex cellular functions in a highly coordinated way. Clearly, understand how a cell works requires a systems approach. We have contributed computational methods that extract mechanistic biological insight from high-throughput profiling data of DNA, RNA and protein molecules (genomics, transcriptomics and proteomics data). I will present exemplary present methods that integrate such data using network-based approaches, machine learning, and relatively simple, yet powerful statistical approaches. Using these examples I will show how insight about the functioning of molecular systems and their malfunctioning during aging and disease can be inferred from large molecular datasets.

19 November 2025

Dr. Geeske Scholz, Bremen University (Germany)

Formalizing social science theory for computational models

Behaviour and behavioural change are crucial to many of our current challenges, such as climate change mitigation and adaptation. The importance of social dynamics and situated behaviour (which is triggered by and follows situational cues) is something I studied empirically and also addressed with agent-based models. Agent-based models reflect great potential to better understand the macro effects of behavioural dynamics. However, the representation of human behaviour in models is often oversimplified and not backed up by research, with significant consequences for model validity. Using the broad body of theories of human behaviour remains a significant challenge for modellers, and the learnings of formalization and modelling for social science research still need to be explored. In this talk, I discuss the potential of combining social science research with social simulation models, focusing on the formalization of theory. I will discuss current work on selecting, formalizing, and incorporating theory from the social sciences in computational models.

26 November 2025

Dr. Maria Gerullis, University of Göttingen (Germany)

Economics of Social-Ecological Diversity: Governance Pathways for Crop Genetic Resource Systems

03 December 2025

Dr. Karin Mora, Institute for Earth System Science and Remote Sensing, Leipzig University (Germany)

Phenoplexity: Phenology of the Future

Plant phenology studies short-term dynamics, such as daily and seasonal changes, measured on the ground as well as with satellite sensors. However, what does an observation of an individual plant, recorded via a smartphone using apps such as Flora Incognita, tell us about the wider plant community or ecosystem, and how well they cope with the impact of climate change? To answer such questions, we have to adapt machine learning approaches typically developed using synthetic data to work with these complicated, noisy, real-world datasets: from analysing time series under driving forces to linking observations from different sources—such as country-scale citizen science data, satellite-derived vegetation indices, and eddy-covariance flux tower data—across species and spatial and temporal scales. I will give an introduction to these multi-scale plant phenology measurements before discussing specific methodological challenges in time series analysis, such as extracting dynamical characteristics of the seasonal cycle in noisy data using dimension reduction approaches, and the use of neural networks for modelling vegetation responses. I will show how these approaches can reveal surprising patterns of synchrony and asynchrony across species and on a country-wide scale. Ultimately, this work aims to predict ecosystem responses to climate change and improve our understanding of feedback mechanisms and plant resilience at global scales, as well as whether biodiversity can have mitigating effects. I will discuss how, on the one hand, big data can help overcome knowledge gaps and yield spatio-temporal patterns, and on the other hand, how it can pose significant computational limitations that themselves constrain the types of system-level questions we can currently ask.

10 December 2025

(online) Dr. Maria Claudia Lopez Perez, Michigan State University

tba.

17 December 2025

Prof. Dr. Johannes Halbe, Osnabrück University (Germany)

tba.

07 January 2026

Prof. Dr. Stefanie Lutz, Utrecht Univerity (The Netherlands)

tba.

14 January 2026

Dr. Elke Kellner, Hochschule Luzern (Switzerland)

tba.