Modelling of processes lets one understand the functions of interacting components, helps to identify parts of processes, and can predict outcomes of changes in the system. Unfortunately, what was a major area of financial modelling is now largely discredited, much to the cost of the rest of us; other areas such as insurance are becoming so constrained by rules and regulation as to be useless. Biological modelling, in contrast is advancing rapidly, whether with respect to subcellular events, whole organism development, or disease epidemiology. This week, Professor Xueron Mao has organized a meeting (previous blog post) at the University of Strathclyde in Glasgow, Scotland, on “Stochastic Modelling in Ecosystems” (link to meeting programme).
In the week marking the 20th anniversary of the 1992 United Nations Conference on Environment and Development (“Rio Summit”), I wondered about the impact of ecosystem modelling on the major policies being discussed at the Rio+20 summit in June 2012. Had I missed a whole area of literature in the last 10 years? Weekly, the news media tell us about the results from the latest models of climate change, while I read papers every month about crop and photosynthesis models, not to say stunning work on many individual plant species, including special issues and collected papers from Annals of Botany. Like many UN-related organizations and meetings, the Sustainable Development conference has a large amount of underpinning ‘grey literature’ – commissioned reports and research with strong scientific content (albeit, often not complete or definitive, and hence not suitable for publication in refereed journals). However, a search of the UN website does not show any attempts at ecosystem modelling (or, indeed, ‘modeling’): the 14 discussions of the topic of modelling are all about economics and finance. The Google Scholar search shows few major papers in the last decade with keywords of modelling/modeling and ecosystems.
Maybe the challenge of modelling a whole ecosystem is too difficult: a model needs to define inputs, outputs and flux through a system. The ecosystem involves cycles and networks involving hundreds of species and millions of interactions from sub-cellular level upwards. In my own talk opening the meeting, I concluded that the outputs can be classified in three areas. Firstly, chemical energy, largely in the form of the fixed carbon that is used as food, feed, fibres and fuel outside the direct ecosystem. Secondly, a small but important fraction of the flux is removed from the system, particularly to the long-term carbon stores in limestone and fossil fuels. The final group of outputs can be considered as ‘ecosystem services’ including purified (or indeed polluted) water that is changed from the input state in both purity and flow rate, or oxygen reduced from carbon dioxide. The slides from my talk are on Slideshare.com under pathh, and maybe I will make a shortened commentary for YouTube at some point.
In the meeting, we were treated to a range of talks ranging from models of carbon cycles, through population and vegetation dynamics, through to disease epidemiology models. It is always exciting when different research communities come together, so it was very valuable to hear from and talk to the mathematicians at the meeting, even if there is some differences in our languages!
It is always invidious to pick out particular talks from a full programme, and the full listing is given here. Since this blog is plant-related, I will note the impressive talks from Mathew Williams (University of Edinburgh) discussing how gigatons of carbon move around the terrestrial (and indeed atmospheric) carbon cycles using global measurements in an experiment named FLUXNET, which, along with space-based measurements could examine large-scale forest biomass changes over timescales of only three years. My collaborator Jongrae Kim (University of Glasgow http://www.robustlab.org/) gave the next talk, discussing some formal approaches to modularization of complex networks in his talk on robustness analysis of community structures, of great relevance to making very large networks amenable to analysis. Francesco Accantino presented a model of abundance and changes of three Acacia species in humid savannas adding stochasticity to a matrix model, which linked nicely to Pierre Couteron (IRD, Montpellier) working at other sites in sub-Saharan Africa. Pierre modelled the distribution patterns of patchy vegetation, showing effects of rainfall and slope in both stable systems and the changes in the last 50 years. Remote sensing is giving much more data than ecologists have ever had, and interestingly Pierre is able to use freely available Google Earth for many of his analyses. After valuable talks related to aspects of epidemiology in several systems, the closing paper by Carlo de Michele (Politecnico di Milano, Italy) built on earlier talks about water as a main determinant of vegetation type – the topic of ecohydrology as the study of hydrology that underpins ecology. Like several other talks, modelling of water could give a bistable system with two solutions of bare soil (low rainfall) or of vegetative ground cover (high rainfall), taking into account the effects of rainfall stochasicity on soil water linked to vegetation systems. The surprise was that not only did the results describe behaviour of desert compared to topical forest ecosystems, but also annual changes in savannas with dry, bare periods followed by vegetation-covered wet seasons.
“Stochastic modelling in ecosystems” has some way to go before it becomes “Stochastic modelling of ecosystems”. Genetics, measurement methods and parameterization of properties are coming from the biologists are beginning to meet the modelling community with their increased understanding of robustness, oscillation and network reduction as well as computational approaches. I am looking forward to decisions at Rio+30 being underpinned by recommendations based on rigorous and robust models showing how we can exploit ecosystems without destroying the earth.