By Michael Renton
(Click the image first and see what happens)
Building computer simulations involves deciding which aspects of a real-world system are most important, and then representing them ‘in silico’ by using mathematical equations or computer-coded algorithms and rules. Models cannot include every part of reality, and thus must always simplify reality – in fact this is what makes them useful! A model that includes too much becomes just as difficult to understand and analyse as the real world. For example, if we want to model the effect of climate change on the growth of a tree in South-Western Australia, we don’t represent a butterfly flapping its wings in Brazil in our model! Modelling thus requires decisions about which parts of the world should be represented in the model, and about exactly how to represent them. These decisions will always involves trade-offs between desirable objectives, such as realism, precision, explanatory value, simplicity and generality.
Models can be made for various purposes. Models can be created to help clarify or understand, for purposes of comparison, prediction or management, or in order to educate and communicate ideas, or even to convince people. They can simply help visualise ideas or results, or aim for accurate prediction in a range of conditions, or act as a theoretical framework for experimental investigations. They can save time and money by helping to focus experimental resources, identify knowledge gaps, and synthesise knowledge. Sometimes, if experiments are impossible, too dangerous, too expensive, or would take too long, then modelling is the only option for predicting what will happen under different scenarios. Clearly identifying the purpose of a model is essential for guiding decisions about what to represent in a model and at what level of detail.
In particular, computer simulation modelling may be able to help us better understand, predict and manage the health of forests and woodland ecosystems in the face of changing climates. By modelling important ecological processes happening across a spatially mixed agricultural landscape with bushland fragments, we might be able to predict the effects of different restoration strategies on biodiversity and agricultural production, and thus choose the best strategies. By representing trees as individuals growing in groups including different species, we can predict the way that they deal with reduced rainfall, while taking into account different species’ strategies for water use and individual variability, and explore possible management options, such as burning or thinning. And by modelling the important ecological processes operating across larger scales, such as local extinction and re-colonisation, gene flow, seed dispersal, competition and evolution, we can predict how the range of different species will change with climate change, and whether restoration or assisted migration can help threatened species persist. In all these situations, experimental data is essential but full experimental analysis is impossible, due to the costs and scales involved, and so building and playing with virtual worlds provides an opportunity to help understand and benefit our real one.
