2019 Elsevier B.V. Uncertainty is a pervasive problem in the development of ecological models precisely because uncertainty is a pervasive problem in ecology. As we have seen, uncertainty pervades the process of model evaluation. The broad array of qualitative and quantitative aspects of model structure and behavior that we examined in Chapter 6 as part of model evaluation all are focused implicitly on identifying and confronting our uncertainties regarding usefulness of our model. Sensitivity analysis, in particular, helps us determine the level of confidence that we should have in model projections in view of the uncertainty with which we have estimated those parameters that most affect model behavior. This parametric uncertainty often results from a lack of data with which to estimate model parameters. However, as we have seen, even if data from the real system are available, they are not an infallible standard for judging performance of the model. Thus, evaluation of uncertainty associated with field or laboratory data played a prominent role in both of the methodological approaches to model evaluation that we discussed in the previous chapter. Also from our discussion in Chapter 10, it is clear that another source of uncertainty associated with model evaluation arises from the semantics associated with discussions of the topic. This linguistic uncertainty arguably plays an under-appreciated role in the model development process.
