The project MODELCOL is integrated in the framework of the urgent need for solutions for compensating human deterioration of the environment. New environmental policies point out natural ecosystems as key elements for providing ecological services such as water purification, soil detoxification, climate regulation and advocate for the creation of new ecosystems (as surrogates for the degraded natural systems) for increasing their positive effect on the environment. The design of such new ecosystems needs to be optimized for providing the best ecological services. The project MODECOL aims at developing an hybrid model, sufficiently realistic, that will simulate a prairial ecosystem (modelled through IBM) correlated through feedbacks to environmental conditions (modelled through PDEs). This approach aims to provide a virtual laboratory for testing ecological hypothesis on complex ecological systems.
The first step deals with the development of an IBM model for simulating a prairie which can be considered as interactions of plant populations. An original aspect of this project is to take into account plant vegetative reproduction e.g. the ability of each perennial grass to colonize space through numerous reiterations of new shoots connected through aerial or subterranean connexions. This network form carries resource and information within the individual.
The second step of the project is to match this IBM to PDE models which will simulate the space and time variations of the environmental conditions (for instance nitrate or pesticides concentration). This point refers to considerations about i. the coupling of models with different time and space scales, ii. the behavioural determinism of an individual plant and its sensitivity at those scales. The validation of this hybrid model will be conducted thanks to in situ surveys in the study site of Pleine-Fougères referenced in the international LTER network.
The third step deals with the development of an aggregated model as a decision support system used by stake holders and end users. Based firstly on dynamic temporal models, a PDE model will be developed using spatial data from the IBM. A cross validation with experimental data will be conducted.
The fourth step of the project aims at using this model to a concrete application in the context of the recent agri-environmental directives of the EU, Common Agricultural Policy (CAP). A key measure concerns the creation of herbal strips along rivers to purificate water coming from agricultural fields before it reaches the river. Our goal is to predict the optimal design of these herbal strips for this ecological service. We will beneficiate from the collaboration of an advisory committee comprising representatives of stake holders, farmers, agronomists and researchers, insuring that our results match the requirements of agricultural practices and constraints.
These tasks will beneficiate from: i the development of numerical ecological tools for matching data from simulations and experiments. These tools will detect spatial plant patterns, test the correlation between plant patterns and environmental conditions at different scales; ii the recent advances in computational procedures to realize simulations at large scale and low cost. This project will be the first green project to use the platform BOINC.
This project falls closely into the scopes of the SYSCOM program. It includes many original aspects, is built on a strong collaboration between ecologists and modellers, and proposes a concrete application in collaboration with stake holders and end users for a better adaptation of our tool. In addition, this project will develop an hybrid model which is sufficiently flexible to be applied, after minor modifications, for other agronomical or environmental purposes linked with herbaceous ecosystems.
