Spatial Methods in the Study of Vector-Borne

Jerome E. Freier

USDA, APHIS, Veterinary Services Centers for Epidemiology and Animal Health Fort Collins, Colorado Topics

• Vector and vector-borne considerations

• Spatial epidemiology introduction

• Spatial analysis methods for population/disease studies

• Remote sensing and habitat modeling

• Training in spatial epidemiology 1. Vector and Vector-Borne Disease Considerations Tetrad Model for Vector-borne Diseases • Developing a model of any vector-borne disease is complex due to: – Number of necessary components – their variability – their spatio-temporal relationships. Vector-borne diseases of importance to animals Domestic

• Disease • vector – Anaplasmosis – (biological) and biting (mechanical) – Bluetongue ( serotypes) – Culicoides (biting midges) – Eastern equine encephalomyelitis – Mosquitoes – Equine infectious anemia – Biting flies (mechanical) – – Ticks – Q fever – Ticks and other – St. Louis encephalitis – Mosquitoes – – Tabanid flies and ticks – Western equine encephalomyelitis – Mosquitoes – West Nile – Mosquitoes Vector-borne diseases of importance to animals Exotic

• Disease • Arthropod vector – – Culicoides (biting midges) – African swine fever – Soft ticks – Akabane disease – Culicoides (biting midges) – – Ticks – Bluetongue (exotic serotypes) – Culicoides (biting midges) – Bovine ephemeral fever – Culicoides and mosquitoes – East Coast fever – Ticks – Epizootic lymphangitis – Flies (mechanical) – Equine encephalosis – Culicoides (biting midges) – Getah – Mosquitoes – Heartwater – Ticks – Japanese encephalitis – Mosquitoes Vector-borne diseases of importance to animals Exotic

• Disease • Arthropod vector – Louping-ill – Ticks – Lumpy skin disease – Mosquitoes and biting flies (mechanical) – Nairobi sheep disease – Ticks – – Mosquitoes (biological) – Biting flies (mechanical) – Screwworm myiasis – Calliphorid flies – Sheep and goat pox – Mosquitoes and biting flies (mechanical) – Tropical myiasis – Calliphorid flies – African animal trypanosomiasis – Glossinid flies (biological) – Venezuelan equine encephalitis Tabanid flies (mechanical) – Vesicular stomatitis – Mosquitoes – Sand flies, black flies, and – Wesselsbron disease Culicoides – Mosquitoes Factors to Consider in the Emergence of VBD • Microbial adaptation and change • Expanded animal range • Climate and weather • Changing ecosystems • Economic development and land use • Human demographics and behavior • Technology and industry • International travel and commerce • Breakdown of animal health protection measures • Poverty and social inequality • War and famine • Lack of political will • Intent to harm Conditions Leading to the Resurgence and Emergence of VBD – Institute of Medicine, 2007

• Population growth and unplanned urbanization

• Poverty, social inequalities, and the emergence of the throw-away society

• Globalization and tracking of humans, , vectors, and genes

• Erosion of public health infrastructure, including human resource capacity in and vector

• Lack of new targets and approaches to control vectors and vector- borne diseases

• Loss of for real and perceived environmental issues, development of resistance in vectors, and economic disincentives to new and formulation development

• Lack of robust models and information systems to predict, prevent, and control vector-born diseases. Knowledge Gaps for Vector-Borne Disease

• Need to fill knowledge gaps to – target specific surveillance and control efforts – minimize surveillance costs over large areas – forecast risk and anticipate range expansion for VBD – develop containment and exclusion strategies for VBD

• Knowledge gaps in basic biological information – Quantitative data on disease agent cycles in all hosts – Measurement of disease agent potential by known and potential vectors. – Occurrence, distribution, and abundance of competent disease agent vectors. – Mechanisms of host – Mechanisms of pathogenesis – Spatio-temporal distributions of vectors and environmental conditions in determining conditions or circumstances for disease emergence. 2. Spatial Epidemiology Spatial Epidemiology

Definition: The discipline in epidemiology concerned with describing and understanding spatial variation in disease risk, particularly at the small area level. Examples of Activities

• Disease mapping

• Geographical correlation studies

• Assessment of spatial and environmental risk factors

• Cluster detection and evaluation

• Predictive spatial modeling

• Spatial data applications development, testing, and consulting 3. Spatial analysis methods for population, vector, and infectious disease studies Analysis of Population Distributions

Describe the how a population is distributed in space • Space occupied • Centers of the population • Population directions • Population density Describing Population Distributions Central Tendency Measures for distributions

Mean center

# # # # # # # Weighted mean center # # # # #

# # # Median center

# # # # # # # # Central feature # # # Standard distance # Standard deviational ellipse Ohio

Case site Mean Center

Population of Test-Positive Animals

Mean Center (unweighted)

Defined: The average x-coordinate and average y-coordinate for all N features in a study area Study Area Mean Center Tracking Disease Progression in Animal Populations

Week 7

Week 6 Week 5

Week 4 Week 3

Week 2

Week 1 N

Study Area Central Feature Population of Test-Positive Animals

Central Feature (unweighted)

Defined: The feature having the shortest total distance to all other N features in a study area Study Area Standard Distance Defined: A statistic that measures the extent to which the distances between the mean center and features vary from the average distance

Points vary more than the standard distance from the mean

Points vary less than the standard distance from the mean

A measure of compactness of points Standard Distance = 291,883 meters around the mean Standard Deviational Ellipse Defined: A measure of the standard deviation of features from the mean center separately for the x-coordinates and the y-coordinates

Shown for one standard deviation First Law of Geography

Everything is related to everything else, but near things are more related than distant things.

- W. R. Tobler, 1970 Principle of Spatial Causality

Something at a given location directly influences the characteristics of nearby locations Estimating Spatial Autocorrelation by Calculating Moran’s I

•Moran’s I is an index that assesses feature similarity in location and attribute value and evaluates whether a pattern is clustered, dispersed, or random.

•The Moran’s I spatial statistics tool calculates Moran’s I Index value, a Z score, and a p-value estimating the significance of the index.

•Moran’s I Index values typically range from +1.0 (clustering) to -1.0 (dispersion), but more extreme values are possible. Cluster Analysis with the Nearest Neighbor Distance Method The The Average Nearest Neighbor Distance This method measures the distance between each feature (or feature centroid) and its nearest neighbor feature (or centroid). All nearest neighbor distances are averaged. If the distance is less than the average for a hypothetical random distribution, the distribution of the features are considered clustered. If the average distance is greater than a hypothetical random distribution, then the features are considered dispersed. High/Low Cluster Analysis with the Getis-Ord General G Statistic

Dairies – Investigations

Considers number of cows on each farm

Is spatial clustering occurring relative to an attribute? Multi-Distance Spatial Analysis Using Ripley’s K Function Multi-Distance Spatial Cluster Analysis • Using Ripley’s K Function determines whether a feature class is clustered at multiple different distances. • Summarizes spatial dependence over a range of distances.

• Results are presented both as a table and a graphic.

Spatial dependency declines at 125 meters and ends at 240 meters Cluster/Outlier Analysis using Anselin’s Local Moran’s I Local Moran’s I • This method identifies clusters of features with values similar in magnitude, in addition to identifying outliers.

• This tool calculates a Local Moran’s I value, a Z score, a p- value, and a code representing the cluster type for each feature.

• The Z score and p-value represent statistical significance of the computed index value. Cluster/Outlier Analysis using Anselin’s Local Moran’s I

Local Moran’s I Z-score Hot Spot Analysis of Point Data using the Getis-Ord G* Statistic

Gi Z-score Interpolation by Mathematical Methods

• Taking known values at points and predicting the value at other points • Output is a surface of predicted values • Assumes that the points are spatially correlated • Number and distribution of points affect output • Search radius, and thus sample size, can be selected • Can set barriers in the interpolated surface, e.g., mountain range Interpolation by Mathematical Methods Inverse Distance Weighted (IDW)

Small Area

Sample Points Large Area

Sample Points Interpolated Surface Interpolation by Statistical Methods Semivariogram Modeling Interpolation by Statistical Methods Kriging Indicator Kriging Kriging for Binary Data American Dog Tick Collection Reports

American Dog Tick Records Rocky Mountain Wood Tick Records Indicator Kriging Kriging for Binary Data American Dog Tick Error Distribution