1. Louis Gross, Ecology and Evolutionary Biology and Mathematics, University of Tennessee Thoughts on Raccoon Rabies Models

2. Overview Objectives - what do we wish to gain from models? Lots of models exist so why are more needed? Are multiple models more enlightening than confusing? Linking to data - how much “reality” do we need to be effective? Models for managers - are rules of thumb all we need?

3. Everyone here plus: Erica Asano (UTK) Keith Langston (UTK) Tim Clayton (UTK) Collaborators

4. Objectives - what do we wish to gain from models? Generality - so that suggestions for actions in one situation may be applied, suitably modified, to other situations without requiring a completely new model Applicability - containing sufficient reality and capacity for calibration with observations that stakeholders feel comfortable using it Extensibility - capacity for elaboration /modification as additional data become available, without great additional effort Testability - projections of system responses (and potentially components of systems) that are amenable to evaluation and determination as to whether model requires modification

5. Objectives - specific to raccoon rabies Account for dynamics of localized (individual and population) response to vaccination schedules as affected by seasonality Account for spatial aspects of disease spread including impacts of spatio-temporally-explicit vaccination patterns Provide methods to evaluate alternative management scenarios (e.g. vaccination, culling) as specified by stakeholders

6. Objectives - specific to raccoon rabies As potential comparators to simpler approaches, develop methods for determining the “best” management using criteria established by stakeholders, including economic ones Provide the capability to utilize spatial data readily available to management agencies Provide methods to evaluate trade-offs such as between expenditures for barriers and surveillance programs

7. Objectives - specific to raccoon rabies Provide methods to account for genetics of transmission and determination of interspecies transmission impacts Provide methods to account for impacts of population structure (age) on spatial spread Provide methods to carry out risk assessments for management actions at various spatial scales, considering impacts of multiple species and spatial and temporal variation in habitat conditions (including climate change impacts)

8. Lots of models exist so why are more needed? Lots of very good people have constructed models using most every kind of approach available ( differential equations, difference equations, matrix models, metapopulation models, individual-based models, cellular automata, GIS-based, etc.) The models appear to have had relatively little impact on practice, so something is lacking (perhaps just communication?) Perhaps the models are either too abstract for potential application (PDEs), are too complex to be practically utilized (Cook County model) Models that walk that fine line between generality and applicability seem to be rare, though certainly lots of papers have been written (one objective, as is getting $ to have fun playing with models)

9. Are multiple models more enlightening than confusing? Models are often constructed in an hierarchical manner - start with simple assumptions and elaborate to more complex ones - this is the nature of the “making up of stories” that is part of science. Determining which models to keep and which to replace with more elaborate ones depends upon the objectives. Given the multiple objectives associated with rabies analyses, multiple models would appear appropriate - no one model will do it all. The choice of which to keep depends upon the patterns we wish to mimic and explore. Mathematically and computationally, it is often appropriate to build up from the simple to the more elaborate, but this process is opaque to stakeholders who just may wish the best guidance available based upon the best available science. Determine what stakeholders need/expect and present model(s) that addresses these issues (may be more than one)

10. Linking to data - how much “reality” do we need to be effective? Determine which data stakeholders care about and include it. Thus if spatial maps that identify areas to carry out an ORVP are needed by managers, include spatial maps at the level of spatial resolution appropriate Use the hierarchy of models to determine whether or not added detail produces significantly different results, where the measure of “significance” corresponds to what can be measured in the field. Models should be tuned to the scales of interest - ignore data at finer or coarser grain than these scales (e.g. details of coon reproductive physiology and sunspot cycles) Remember that the purpose of modeling is insight, not perfect replication of reality

11. Models for managers - are rules of thumb all we need? Managers are diverse and multiple levels of decision- making are to be expected. For some of these (perhaps many of these) explanatory methods that are based upon good science, though not having the details of the best available science, are more likely to have impact (e.g. actually be applied) Determining how appropriate a rule of thumb is for a particular situation requires a comparison to either experimental data (difficult to obtain) or some comparator evaluative set and likely this latter is best obtainable from more elaborate models. Be sure to determine how big the stakeholders thumbs are before developing a rule of thumb from a more complex model.