Spatial correlations at different spatial scales are themselves highly correlated in isolation by distance processes

Area of Science:

• Spatial statistics
• Population genetics
• Landscape genetics

Background:

• Spatial autocorrelation statistics are well-understood, but correlations between different spatial scales are largely unknown.
• Understanding these correlations is crucial for making accurate inferences in spatial genetics across multiple scales.

Purpose of the Study:

• To investigate correlations among spatial autocorrelation statistics (Moran's I) at different spatial scales.
• To analyze how dispersal levels influence these correlations in isolation by distance processes.

Main Methods:

• Conducted stochastic space-time simulations of isolation by distance processes.
• Analyzed correlations among Moran's I statistics across various mutually exclusive distance classes.
• Varied dispersal amounts for simulated plant and animal populations.

Main Results:

• Stochastic correlations among spatial autocorrelation statistics at different scales were found to be extremely large (>0.90).
• These correlations exhibit a complex relationship with dispersal levels, spatial scale, and spatial lag.
• Signal-to-noise ratios decreased significantly with increasing distance.

Conclusions:

• Existing statistical methods using multiple distance classes must account for these large stochastic correlations.
• Increasing the number of assayed loci is generally more effective for statistical power than increasing sample size.
• Results provide guidance for complex landscape genetics processes and are applicable to various spatial autocorrelation measures.