Since all models are wrong the scientist cannot obtain a “correct” one by excessive elaboration. On the contrary following William of Occam he should seek an economical description of natural phenomena. Just as the ability to devise simple but evocative models is the signature of the great scientist so overelaboration and overparameterization is often the mark of mediocrity.
– George Box
Ecology is fundamentally intertwined with our understanding of processes that regulate our environment. However, we find ourselves facing unprecedented human-driven changes in our environment in the forms of urbanization, fragmentation, and climate change. With such monumental changes, we have already observed and can expect to see further differences in how pathogens emerge and spread in human and animal populations. Mathematical models can be a particularly valuable tool that help us understand how epidemics come to be, not only in the context of populations, but across landscapes. You have probably heard the favorite aphorism of statisticians, attributed to George Box: “All models are wrong, but some are useful.” So how does one choose the “least wrong” model for a given host-pathogen system? I do not have a rote or straightforward answer for you, but there will inherently be trade-offs between computational efficiency, requirements for parameterization, and how well the model can be generalized to other systems. These trade-offs are exacerbated when we extend our models to include spatial context.
Some of the very first spatial disease models were used to help understand how rabies spread across Europe in fox (Vulpes vulpes) populations and to evaluate the effects of control strategies for bovine tuberculosis in badger (Meles meles) populations. Metapopulation models, initially applied in the absence of disease to understand the role of conservation corridors in population persistence, also provided some of the first antecedents to current spatial disease models. In our recent review for the Journal of Animal Ecology, my co-authors and I classified the questions, systems, and model-types for over 150 studies. Disease ecology is inevitably moving towards more complex, elaborate, and system specific individual-based models. However, despite advancements in computational and analytical methods, rabies and bovine tuberculosis remain some of the most common pathogens of choice for spatial disease models. The spatial models in our review were also most likely to be in mammals or theoretical hosts rather than other taxa.
One Health or the intersection between human, animal, and environmental health is a hot topic these days, and it is a common refrain to hear for calls of integration between the different realms of science that study each of these areas. We found in our recent review, however, that there has been less integration across disciplines than one might hope to see in terms of spatial disease models. I truly believe that there is further room for integration between the fields of ecoimmunology, movement ecology, genomics and disease ecology. With such integration, our models may end up being more elaborate, but they may also be more “useful” to understanding the pressing issues of human-mediated environmental changes.
Lauren White @LAWhite_Ecology
Department of Ecology, Evolution & Behavior, University of Minnesota
Box, G. E. P. (1976) Science and Statistics. Journal of the American Statistical Association 71: 791–799.
Manlove, K.R., Walker, J.G., Craft, M.E., Huyvaert, K.P., Joseph, M.B., Miller, R.S., … Cross, P.C. (2016) “One Health” or three? Publication silos among the One Health disciplines. PLoS Biology 14: e1002448.
White LA, Forester JD, Craft ME (2017) Dynamic, spatial models of parasite transmission in wildlife: Their structure, applications, and remaining challenges. Journal of Animal Ecology. In Press.