MORPH – Lessons learnt from the development and application of an individual-based model

This paper summarises some of the lessons learnt over approximately 25 years from the development and application of the individual-based model, MORPH. The overall purpose of the model is to predict how the survival, condition, and behaviour of populations of animals are influenced by the food resources available and other environmental factors. The model was originally developed for a single bird species at a single site but was designed to be very flexible, hence its name. This flexibility has allowed the model, with its code largely unchanged, to be applied to 30 bird species across 55 sites and three fish species to inform environmental decision-making by predicting the effects on animals of environmental change, including habitat loss, disturbance from human activities, climate change, aquaculture, and shellfishing. The key lessons learnt during this process, associated with maintaining model flexibility and the ability to inform environmental management and policy, are to: (i) think beyond the system for which a model is being developed to prevent the model from becoming too system-specific; (ii) keep as much system-specific detail as possible out of the model code to maintain flexibility; (iii) incorporate individual variation between animals and animal decision-making rules that mimic those in real animals; (iv) keep the scope of the model as simple as possible to minimise model complexity; (v) search for general patterns in parameters that apply across systems to simplify model application to new systems; (vi) minimise the number of parameters with unknown values to reduce the need for model calibration; (vii) identify parameters to which predictions are most sensitive; (viii) test as many model predictions as possible to assess the accuracy of the model; (ix) compare the accuracy of predictions of alternative versions of a model; (x) simplify the process of running and interpreting a model; (xi) search for general patterns in predictions that apply across modelled systems to derive more general insights; and (xii) simplify the process of communicating model predictions. It is hoped that some of these lessons will be familiar to others and be useful for those developing future individual-based models.