Investigating Disease Spread between Two Assessment Dates with Permutation Tests on a Lattice

Area of Science:

• Plant pathology
• Spatial statistics
• Epidemiology

Background:

• Analyzing plant disease spread over time is crucial for understanding disease dynamics.
• Spatiotemporal analysis of disease status in regularly spaced plantings requires robust statistical methods.
• Existing methods may lack flexibility in handling missing data or complex spatial structures.

Purpose of the Study:

• To develop and evaluate a permutation method for analyzing spatiotemporal disease maps in plant populations.
• To assess the independence of newly diseased plant locations from previously diseased plants.
• To provide a flexible tool for exploring disease spread processes with minimal prior assumptions.

Main Methods:

• Utilizes a permutation approach with Monte Carlo tests to analyze binary spatiotemporal disease data.
• Accounts for spatial structures at each time point to differentiate between inherent spatial patterns and disease dependence.
• Simulates independent patterns by random shifting or reallocation of diseased plant positions for comparison.
• Employs distance-based statistics to compare simulated and observed disease patterns.

Main Results:

• The proposed method accurately identifies simulated independent and dependent bivariate point patterns, demonstrating robustness.
• The permutation test effectively separates nonrandomness due to spatial structure from nonrandomness caused by disease dependence.
• Real-world case studies illustrate the method's ability to provide insights into contrasting disease spread processes.

Conclusions:

• The developed permutation method offers a valuable, flexible tool for analyzing plant disease spread in spatiotemporal maps.
• It can supplement biological investigations and serve as an exploratory step before mechanistic model development.
• The method handles missing data and complex spatial dependencies, enhancing its applicability in ecological studies.