Details zur Publikation
|DOI / URL|
|Volltext||Publikationsdokument einer UFZ-Veröffentlichung|
|Titel (primär)||Linking pattern and processes in natural forests: innovative modelling methods across scales|
|Journal / Serie||PhD Dissertation|
|POF III (gesamt)||T53;|
|UFZ Bestand||Leipzig, Bibliothek, Reportsammlung, 00525299, 18-0218 F/E|
|Abstract||The dynamics of ecosystems are often analysed by using the concept of processes. It is essential to explore relations across scales and understand the linkage of pattern and processes (e.g. using ecosystem modelling), because the scale at which an ecological process acts may differ from the scale at which the process generates patterns. In this thesis natural forest are the focus of investigation as they provide important services like conservation of biodiversity and carbon storage. In the first chapter we give an introduction on the different parts of a scientific inference framework in which this thesis is embedded. We emphasize the importance of innovative modelling methods and present in the following chapters three modelling studies in the context of forests.
In the first study we present an inverse modelling framework based on stochastic search methods. This framework allows the fast calibration and uncertainty assessment of vegetation models. The framework is tested using a dynamic forest model in a virtual, single species and multiple species experiment. The developed methods were able to provide reliable parameter and uncertainty estimates using 10 times less computational resources than comparable methods based on Bayesian techniques.
In the second study we developed a neutral model of tropical rain forest including size structure and neighbourhood competition. The model was able to explain the main dynamics of six spatial and non-spatial biodiversity patterns using a single parameter set. This includes the species abundance distribution, the species-area relationship and the individual tree size distribution. A global sensitivity analysis revealed a highly correlative structure between mortality, dispersal and migration, which points to possible model enhancements.
In the third study we focused on the quantification of global deforestation, as this is the main driver for biodiversity loss and species extinction. We present results regarding the structure and state of forest fragments on a global scale and in three different forest biomes using an innovative cluster detection algorithm, unique data compression methods and high-resolution (30 m) forest cover maps. The analysis showed that the forests worldwide contain 409 million forest fragments and 36% of forested area lies within
100m of forest edges. Additionally, the fragment size and shape index frequency distributions have a similar shape across the forest biomes, which can be described through power-laws.
This thesis demonstrates that complex questions regarding the understanding of natural forests can be answered using innovative modelling methods in conjunction with recent advances in available observation data. The developed methods are formulated in such a general way that they can serve as a basis to test future hypotheses about ecological communities and vegetation dynamics.
|Lehmann, S. (2018):
Linking pattern and processes in natural forests: innovative modelling methods across scales
Dissertation, Universität Osnabrück
PhD Dissertation 1/2018
Helmholtz-Zentrum für Umweltforschung - UFZ, Leipzig, 99 pp.