By: Carlos Pérez Carmona, Graciela M. Rusch and David N. Barton
This example illustrates the use of a Bayesian Network for modelling This example illustrates a way to model ecological functions that underpin the provision of ecosystem services in a silvopastoral systems in Senegal based on functional traits of 23 common tree species in the region as indicators. We link the ecological nodes with nodes that represent farmers. preferences which gives the opportunity to explore the consequences of farmers. choices on ecological functions and the overlap between science-based and local ecological knowledge. We show how BBN software can give rapid access to multiple types of knowledge and assist in reasoning about conditions under which different agro-silvopastoral practices will be more likely. The plant functional group models are based on tree traits that are indicators of the effect of trees on pasture productivity; on soil fertility and on fodder quality for goats, sheep and cattle. The socio-economic model is based on the farmers. perception on useful trees, the specific uses they make of the trees, and the kinds of agro-silvopastoral practices the farmers are interested in. The sub-models on trees and pasture productivity and soil are built following the assumption that the effect of the trees can be positive, negative or neutral, depending on two interacting features: the characteristics of the tree functional group and the environmental characteristics.
Examples of questions that can be explored:
Functional group - environment interactions on pasture productivity. Grassland productivity is affected differently by the different Grassland productivity functional groups, and in interaction with site conditions. Click for instance on Productivity group 2 in the Grass productivity node in the set of Tree functional groups nodes. The trees in this group were mostly located in non-salty soils and promoted pasture productivity, i.e., in about 90 % of the cases, trees in this group increased pasture productivity.
The effects of trees associated with soil chemical properties. Trees considerably increased soil C, but these effects were independent of functional group and of the environment. Click on Carbon group 1 in the Soil carbon node in the set of Tree functional groups nodes, and then on Carbon group 2, and you will notice no difference in the probability distribution of the soil carbon response.
Trade-offs and synergies in farmers' choices. Farmers make various uses of the trees and choices based on uses may synergetic or result in trade-offs. Click for example on Pharmacopaeia in the Uses of trees node in the Farmers' uses and practices set of nodes and on Fodder for goats node in the Ecosystem services set of nodes and you will observe that plants used for medicinal purposes are also highly consumed by goats. Also species used as food are highly consumed by goats.
Carlos Pérez Carmona, David N. Barton, Katim Touré and Graciela M. Rusch. (2013) Towards integrating plant functionality traits and ecosystem services in a model of silvopastoral systems in Senegal using Bayesian Belief Networks.Graciela M. Rusch, Carlos Pérez Carmona, Mayécor Diouf, Cristina Armas, Dioumacor Fall, Francisco I. Pugnaire and David N. Barton. (2013) Trait-based models for single AFS functions.
Useful references for those interested in Bayesian networks include:
Kjærulff, U. B. and Madsen, A. L. (2013) Bayesian Networks and Influence Diagrams: A Guide to Construction and Analysis. Springer, Second Edition.
Graciela M. Rusch firstname.lastname@example.org David N. Barton email@example.com
Norwegian Institute for Nature Research (NINA) www.nina.no