US-UK COLLAB: LONG-DISTANCE DISPERSAL AND DISEASE SPREAD UNDER INCREASED ECOLOGICAL COMPLEXITY

<b>Project Methods</b><br> 1. Locations of multiple sources of epidemic outbreak can be imputed from population dynamic models. We will build upon a framework that we previously developed by using a Bayesian approach for estimating the locations of infection sources when they are unknown. We will start with a prior on the number of sources and then conditioned on the number of sources. The estimates of the locations of the sources would be obtained based on the posterior distribution. This approach will be applied to the following diseases: For cucurbit downy mildew, we will conduct inoculated field experiments to validate the model at relatively short distances (hundreds of meters). For hop powdery mildew we will conduct retrospective analyses with extant census data of during 2014 to 2017 and apply the modeling framework described above for identification of unknown disease sources at the mesoscale. For foot-and-mouth disease significant effort will be devoted to data from the 2001 FMD epidemic in the UK owing to the high quality of that data set. A second available data set is from an FMD outbreak in Japan in 2010. We will determine if the original and simulated sources of different can be correctly imputed at different time points in the simulated epidemics, test effects of stochastic effects on imputations, and determine if multiple sources can be correctly imputed. For wheat stripe rust, two replicated experiments will be conducted, each repeated in two consecutive years. In the first experiment, treatments will involve artificially inoculating replicated plots with either one, two, three, or four outbreaks of equal size. In the second experiment, all plots will be artificially inoculated with three foci, with foci being either small or large, in different combinations.2. Ecologic spillover effects influence the spread of LDD epidemics in a predictable manner, and strongly influence observed relationships between disease spread and host diversity. For sudden oak death, we will use five main host species of differing susceptibility and transmission values to test spillover effects. We will simulate spread in two different series with the PoPS model, which has been designed to incorporate plant community structure. The first series will test the effect of species richness and the second set will examine plant diversity by altering species proportion. For foot-and-mouth disease, there are substantial differences of pathogen fitness components on different host species. There are a multitude of ways in which effects of spillover can be evaluated via modeling. The modeling studies may provide clues to the combinations of susceptibility and transmission by which different types of epidemic responses are observed. For West Nile virus, natural transmission cycles involve mosquito vectors and avian hosts. We will consider three scenarios based on contact networks, where birds are nodes, and connections among nodes represent the possibility of disease transmission from one infected bird to another susceptible bird by mosquito vectors. For cucurbit downy mildew, we will interplant cucumber (moderately susceptible to A1 mating type lineage ) and squash (moderately susceptible to A1 mating type lineage ) at two plant densities and create two primary disease sources of differing size on either the leeward or windward side of a plot.3. Robust predictions of pathogen transmission require understanding the effects and sources of uncertainty. Predicting pathogen transmission requires that forecasts be robust and uncertainty understood. Uncertainty in forecasts is primarily driven by initial disease conditions, data drivers, parameter estimates, and model selection (e.g., choice of dispersal kernel distribution). To account for the multiplicity and complexity of all factors affecting the progress and evolution of epidemics, we will draw a parameter value from parameter distributions for our estimated model parameters for each PoPs simulation. Each simulation will produce one realization of the stochastic process. We will run many iterations of the model to produce an average projection and a cone of uncertainty around the average projection. We will quantify the effect of different sources of uncertainty on model predictions to highlight areas that would provide the greatest impact on predictive ability, and how these sources of variation combine. Data for all sources of variation will be available for SOD, wheat stripe rust and cucurbit downy mildew from previous experiments of the PIs, field experiments to be conducted within this proposal, or from the literature.4. A unifying framework of biological processes emerges across LDD diseases incorporating diverse hosts, pathogens, and environments evolving with timeWe will develop a generalized modeling platform that can accommodate a wide breadth of pathogen biology and ecological questions. Our team's previous research has shown several diverse disease systems share the same biological framework concerning their specific long-distance dispersal (LDD) characteristics. We will build a unifying framework that will combine data from field experiments, citizen science, laboratory studies, and other sources with ensemble modeling of biological processes. The power of a multi-model unifying biological framework is that it connects pathogen biology to a suite of models that allow testing hypotheses about LDD across landscape structures. Our team has a strong history of using both network and raster-based models to understand and forecast epidemic outbreaks. Once developed, we will parameterize and validate multilayer raster-based and network models against the many field data sets available for the six diseases to be studied in this proposal. Once the models have been shown to handle the biological diversity represented by these six diseases, we will move on to more general questions in disease ecology. We will test effects of and interactions among pathogen fitness components, spillover effects, sources of uncertainty, spatial patterns of epidemic outbreaks, weather patterns, host density, and host landscape structure.5. Optimizing epidemic mitigations. Several senior personnel have long-time experience with one of the specific diseases proposed for study, and a strong desire to contribute to more complex analyses of societal impacts and to reach out to underrepresented groups. The overall goal of these analyses is to optimize epidemic mitigations in terms of policies, societal impacts, and epidemic spread. Though a thorough evaluation would require a separate proposal, there are several projects for which the senior personnel have decades of experience, substantial data and a significant start, thus making the proposed contributions feasible for foot-and-mouth disease, sudden oak death, hop powdery mildew, cucurbit downy mildew, and West Nile virus.