Browsing by Subject "Biogeochemie"

Now showing 1 - 3 of 3

Equifinality, sloppiness and emergent minimal structures of biogeochemical models
(2019) Marschmann, Gianna; Streck, Thilo
Process-based biogeochemical models consider increasingly the control of microorganisms on biogeochemical processes. These models are used for a number of important purposes, from small-scale (mm-cm) controls on pollutant turnover to impacts of global climate change. A major challenge is to validate mechanistic descriptions of microbial processes and predicted emergent system responses against experimental observations. The validity of model assumptions for microbial activity in soil is often difficult to assess due to the scarcity of experimental data. Therefore, most complex biogeochemical models suffer from equifinality, i.e. many different model realizations lead to the same system behavior. In order to minimize parameter equifinality and prediction uncertainty in biogeochemical modeling, a key question is to determine what can and cannot be inferred from available data. My thesis aimed at solving the problem of equifinality in biogeochemical modeling. Thereby, I opted to test a novel mathematical framework (the Manifold Boundary Approximation Method) that allows to systematically tailor the complexity of biogeochemical models to the information content of available data.
Modeling microbial regulation of pesticide turnover in soils
(2022) Chavez Rodriguez, Luciana; Streck, Thilo
Pesticides are widely used for pest control in agriculture. Besides their intended use, their long-term fate in real systems is not well understood. They may persist in soils, thereby altering ecosystem functioning and ultimately affecting human health. Pesticide fate is assessed through dissipation experiments in the laboratory or the field. While field experiments provide a close representation of real systems, they are often costly and can be influenced by many unknown or uncontrollable variables. Laboratory experiments, on the other hand, are cheaper and have good control over the governing variables, but due to simplification, extrapolation of the results to real systems can be limited. Mechanistic models are a powerful tool to connect lab and field data and help us to improve our process understanding. Therefore, I used mechanistic, process-based models to assess key microbial regulations of pesticide degradation. I tested my model hypotheses with two pesticide classes: i) chlorophenoxy herbicides (MCPA (2-methyl-4-chlorophenoxyacetic acid) and 2,4-D (2,4-Dichlorophenoxyacetic acid)), and ii) triazines (atrazine (AT)), in an ideal scenario, where bacterial degraders and pesticides are co-localized. This thesis explores some potential controls of pesticide degradation in soils: i) regulated gene expression, ii) mass-transfer process across the bacterial cell membranes, iii) bioenergetic constraints, and iv) environmental factors (soil temperature and moisture). The models presented in this thesis show that including microbial regulations improves predictions of pesticide degradation, compared to conventional models based on Monod kinetics. The gene-centric models achieved a better representation of microbial dynamics and enable us to explore the relationship between functional genes and process rates, and the models that used transition state theory to account for bioenergetic constraints improved the description of degradation at low concentrations. However, the lack of informative data for the validation of model processes hampered model development. Therefore, in the fourth part of this thesis, I used atrazine with its rather complex degradation pathway to apply a prospective optimal design method to find the optimal experimental designs to enable us identifying the degradation pathway present in a given environment. The optimal designs found suggest to prioritize determining metabolites and biomass of specific degraders, which are not typically measured in environmental fate studies. These data will lead to more robust model formulations for risk assessment and decision-making. With this thesis, I revealed important regulations of pesticide degradation in soils that help to improve process understanding and model predictions. I provided simple model formulations, for example the Hill function for gene expression and transition state theory for bioenergetic growth constraints, which can easily be integrated into biogeochemical models. My thesis covers initial but essential steps towards a predictive pesticide degradation model usable for risk assessment and decision-making. I also discuss implication for further research, in particular how mechanistic process-based modeling could be combined with new technologies like omics and machine learning.
Models for the representation of ecological systems? The validity of experimental model systems and of dynamical simulation models as to the interaction with ecological systems
(2001) Haag, Daniel; Kaupenjohann, Martin
Models guide the investigation of ecological phenomena and the managemant of man-environment interactions. Based on six papers, this thesis critically examines characteristic features, limitations and the scientific and societal role of experimental model systems (as well-tried instruments of knowledge production) and of dynamical simulation models (as representatives of relatively recent computer models). Experimental model systems are described as materially and conceptually closed systems with a limited number of parameters. They consist of a material component which is encoded into a formal (numerical) system through the measurement of defined parameters. The transfer of statements derived from model systems to natural systems is critically discussed. Dynamical systems - the paradigm for the representation of ecosystems - permit the simultaneous handling of a large number of parameters. Dynamical systems are conceptually closed systems and are based on the notion of an abstract state (focussing on 'being'). I contrast this view with an image of ecosystems as conceptually open systems ('becoming') which emphasizes the evolutionary openness of ecological systems, the internal production of novelty, and the emergence of system level properties. Taking the nitrogen cycle and its human alterations as an example, model concepts and limitations to the derivation of cause-effect-relationships in ecological systems are illustrated. Acknowledging the limited predictive capacity of simulation models and the intrinsic perspectivity of the identification of 'relevant' phenomena and parameters and drawing on new forms of knowledge production (as described by science studies), a modified role for model building and for simulation models - particularly with respect to science for policy - is sketched.