Measuring spatial aggregation in binary epidemics: correlative analysis and the advantage of fractal-based sampling

Area of Science:

• Plant pathology
• Spatial statistics
• Epidemiology

Background:

• Incomplete sampling of binary epidemics involves analyzing spatial patterns of diseased plants and sampled points.
• The spatial arrangement of diseased plants is dependent on the geometric locations of sampled points.

Purpose of the Study:

• To examine and compare the performance of various sampling schemes for disease incidence.
• To evaluate sampling strategies for detecting spatial patterns and estimating disease incidence in binary epidemics.

Main Methods:

• Simulated binary epidemics using a Neyman-Scott cluster process with varying spatial aggregation.
• Analysis of spatial correlation adapted for binary epidemics, equivalent to a chi-squared test.
• Comparison of regular grid, clustered, random, and fractal sampling designs.

Main Results:

• Fractal sampling designs demonstrated superior efficiency in detecting contagion and provided spatial information over broader distance scales.
• Regular grid sampling consistently yielded the lowest variance in average disease incidence estimates.
• Random sampling patterns exhibited intermediate performance between fractal and grid designs.

Conclusions:

• The choice of sampling scheme significantly impacts the detection of disease contagion and the precision of incidence estimation.
• Fractal designs are optimal for understanding spatial disease dynamics, while grid designs are best for accurate incidence estimation.