2011 Annual Report
1a.Objectives (from AD-416)
Infectious diseases such as foot and mouth disease (FMD) and the Ug99 strain of wheat stem rust (Ug99) present significant threats to national security because of the potential for the rapid spread of these diseases on a continental scale resulting in severe social and economic disruption. In most scenarios, these dislocations would certainly involve extended embargo of agricultural products, restriction of interstate commerce and costs associated managing the outbreak itself (e.g., diagnosis, testing, removal and quarantine).
The potential for high socio-economic impact makes it essentially impossible to conduct designed experiments to validate epidemiological models for the spread of these diseases in real-world settings. In the absence of data from controlled experiments to test specific hypotheses and predictions, it is common to rely on endorsements by trusted subject matter experts or by fitting to extant observational data. In contrast, plant diseases such as wheat leaf rust, sudden oak death and late blight of potato and tomato offer the opportunity to conduct extensive validation of models and thereby to gain better insight into the spatiotemporal dynamics of infectious diseases.
Since infectious diseases of plants and animals can spread over similar scales of space and time, it makes sense to use plant diseases as proxies for highly contagious foreign animal diseases such as FMD in order to address generic, high-level questions relating to commonalities among pathosystems and methodologies for constructing quantitative models. By first addressing some more general, high-level questions in the context of particular well-characterized model plant systems, we propose to identify and address generic statistical and methodological issues, and then to test solutions to these problems in the context of specific models. This knowledge can then be used to guide the development, implementation and testing of models for specific animal diseases, despite the differences in the details of the models and biology of the organisms involved.
1b.Approach (from AD-416)
The primary focus of this work will be on several generic aspects of epidemiological models: (a) identification and examination of existing models and datasets as case studies; (b) analysis of models to examine dynamics of disease spread, parameter estimation, and contact network structure in models of interest; (c) analysis of statistics of extreme events and first passage times in models of interest; (d) development of a flexible toolkit for implementing spatio-temporal models for national-scale outbreaks of agriculturally-related infectious diseases; (e) applications to fungal diseases of plants capable of rapid dispersal over long distances. Related issues arise in essentially all models for FAD epidemics on a national scale.
The USDA and DHS, in collaboration with researchers at Cornell University, are currently working together closely to address these problems. Within USDA and Cornell, this performance site is uniquely qualified to provide expertise in statistical physics, dynamical systems, applied mathematics, computational science, and close cooperation with subject matter experts in plant diseases. In addition, there are well-established relationships across a range of technical and scientific disciplines relevant to the project. The performer has already established a working relationship with DHS S&T and USDA-APHIS, as well as other DHS S&T performers (e.g., RAPPID).
Since award of this IAA on 21 January 2010, the Ithaca group has established a highly productive working relationship with several RAPIDD Working Groups. In addition, the group is now collaborating with researchers at Cornell (Ithaca, NY) and Cambridge (Cambridge, UK).
The Ithaca group has been developing methods that exploit land use and related geographic and remote sensing data into spatial models of disease outbreaks. This is a unique and time critical opportunity for the DHS S&T FAD Modeling Program. Existing funding will be used to support two students, two full time senior scientific staff members and two postdoctoral researchers. These positions will all be supported for the remainder of this IAA (through Jan. 2014).
Substantial progress has been made in several areas. First, two different kinds of statistical methods have been developed to characterize the fragmented structure of agricultural land use from the Cropland Data Layer supplied by the USDA National Agricultural Statistics Service. This dataset covers the entire US at very high resolution (30 meters). These methods simultaneously measure the typical sizes, area to perimeter ratio and spacing between patches of the same land use class. This information is critical for developing models for the spread of crop pathogens on national and regional levels. In addition, collaborators are currently using it to guide the collection of field data on diseases of wheat and maize in upstate New York. Second, mathematical methods have been developed to predict the probability of outbreaks of infectious diseases in collections of subpopulations connected by a transportation network. These methods depend on a few basic epidemiological parameters, the sizes of the subpopulations and the structure of the network itself. Unlike previous methods for this class of problems that rely heavily on computer simulations, these analytical methods yield simple formulas for the probability of epidemics in cases where the sizes of the subpopulations are large, even for very large networks. These methods are generally useful for modeling the spread of diseases where transportation plays an important role in long-distance transmission between otherwise isolated subpopulations such as movement of infected horticultural and garden nursery stock through consumer-oriented distribution chains and the spread of foot and mouth disease by transport of infected cattle and swine between various production facilities.