Prof. Dr. Carsten Dormann
Carsten Dormann, Prof. Dr.
Department of Biometry and Environmental System Analysis
phone: ++49 761 203-3749
Zur Vorlesung "Statistik" für BSc Umweltnaturwissenschaften, Waldwirtschaft und Umwelt, Geographie jetzt das Lehrbuch:
Ein früheres Machwerk für Umsteiger auf R:
Dormann, C.F., I. Kühn. 2008. Statistische Analyse biologischer Daten (mit dem freien Programmpaket R). 2. Auflage.
In my research, I seek to understand the mechanistic functioning of ecological communities, in order to be able to predict effects of Global Environmental Change (GEC, comprising Climate and Land-Use Change, as well as loss of biodiversity). The key concept here is "biotic ecosystem services". Under this framework, I want to understand how biological services (such as pollination, carbon sequestration, soil nutrient cycling and biological control) are affected by GEC.
My research methods fall into three categories: manipulative field experiments, statistical ecology and mechanistic modelling. They go hand-in-hand:
I employ statistical methods to explore potential causes of ecological pattern, thereby generating hypotheses on the processes and mechanisms underlying these pattern. My main field or research is here species distribution modelling, i.e. the analysis of large-scale data sets on the occurrence pattern of plant and animal species, as well as regional biodiversity pattern.
For plant communities, I then try to manipulate processes I suspect to be important. These have mainly been resource competition and herbivory, but also mycorrhization. Experiments take place in the field as well as in the lab, in greenhouses and botanic gardens. My favourite system is low-growing vegetation: arctic tundra (where I did my PhD), salt marshes or temperate grasslands.
Finally, I try to predict the future: Based on the insights from exploratory statistical analyses and mechanistic understanding of different systems, I employ mechanistic modelling to investigate the effects of land-use and climate change.
Details and projects: The data-theory interface and the role of statistical ecology
This interface between theory and data is a very challenging one, since theoretical ecology is very advanced in this field, but appropriate experimental data are very hard to come by. In this field I try to bring as much theoretical-mechanistic understanding into field work and experiments. I believe that theory requires experimental support to become usefull, while at the same time theory actually provides the hypotheses to be tested. So, in principle, ecological theory and experiments should lead a harmonious relationship. Due to different research traditions, this is not always the case, and I try to do my bit to help bridge the existing gaps.
Plant community processes
I could show that competition is of high importance for plant community structure for different systems (e.g. salt marshes, High Arctic tundra, temperate grassland, laborartory bryophyte communities). When I tried to transfer this understanding into community theory, the theoretical predictions and the experimental data did not match (Dormann & Roxburgh 2005). Plant biodiversity will affect local biogeochemical processes, so it is of relevance to properly understand the mechanisms that produce biodiversity in the first place.
How plant diversity mediates climate change effects on litter decomposition and nutrient cycling in grasslands and forest understorey (funded as part of the DFG biodiversity exploratories); together with Dr. Michael Scherer-Lorenzen, ETH Zürich
Species Distribution Modelling
Throughout the world, large datasets on species occurrences are available. Their analysis can be used to understand the drivers of species occurrences and species richness, and possibly to project their ranges under GEC. However, the analysis of these data is burdened with statistical problems (such as spatial autocorrelation) and ecological assumptions (such as lack of adaptation to new environmental conditions). Investigating the consequences of violating these assumptions and finding solutions to some of the problems is my aim. In particular the integration of important ecological processes (biotic interactions, microevolution, dispersal/migration, changing limiting factors) are the future steps in my research.
Species Distribution Modelling and the effects of Global Environmental Change (various small, DFG-funded workshops); main collaborators are Boris Schröder, Uni Potsdam, Ingolf Kühn, UFZ Halle, Björn Reineking, Uni Bayreuth.
Diversity pattern of plants, birds and arthropods: quantifying the importance of climate and soil vs. land-use intensity and landscape structure (e.g. within the EU project GREENVEINS); main collaborators are Dr. Oliver Schweiger, UFZ Halle
In many studies, visitation recordings of pollinators to plants have been analysed to investigate the stability of pollination networks to Global Environmental Change or loss of pollinators. However, several assumptions behind the sophisticated analysis of pollination networks are untested (e.g that the observed network is representative), others are untenable (e.g. that temporal changes in pollination network structure are negligible). Within a larger research project (Biotic Ecosystem Services), I work on simulations to investigate violations of some assumptions. For that purpose I also maintain an R-package devoted to the visualisation and analysis of bipartite networks (see package bipartite on CRAN: www.r-project.org).
Bipartite: Providing the tools to analyse pollination networks (with Bernd Gruber, Uni Canberra, Nico Blüthgen, Uni Darmstadt, and Jochen Fründ, Uni Göttingen)
Landscape-scale pattern of pollination and biocontrol services
From various descriptive studies we have gained an intuitive insight into how landscape structure and composition affects the diversity and abundance of pollinators and pest-suppressing biocontrol agents. At the landscape scale, manipulative experiments are nigh impossible, so an alternative approach, spatially explicit mechanistic models, comes into play. By formulating our knowledge as behavioural rules or as mechanistic processes, we predict species abundances in time and space, which can then be validated with field data. Once a model is able to a priori predict correctly, we can use land-use and climate change scenarios to explore how pollination and biocontrol services will be affected.
I lead the UFZ' Helmholtz-University Group "Biotic Ecosystem Services", together with the Agroecology at the University Göttingen under Prof. Teja Tscharntke, which combines field work and modelling in the way outlined above.
"Modelling wild bee pollination at the landscape scale", a PhD study by Jeroen Everaars, as part of the ALARM project (www.alarmproject.net), in collaboration with PD Dr. Joseph Settele, UFZ Halle and Prof. Ingolf Steffan-Dewenter, Uni Bayreuth