Validating models of complex, stochastic, biological systems

The highly variable nature of many biological entities, particularly those occurring in natural ecosystems, makes the validation of models and simulations extremely important. Valuable tools such as simulations should not be discounted simply because of a lack of cookbook validation procedures. This paper considers the process of validation as it relates to complex biological systems, and examines some methods to compare the inherently multivariate values required to characterise biological entities. A model of Pinus radiata root system morphology is used as a case study. It is concluded that, while time consuming, development of system-specific objective indices may often be the most useful way to compare simulation output and field data when complex biological entities are modelled.