Multiscale codependence analysis: an integrated approach to analyze relationships across scales

Area of Science:

• Ecology
• Ecological modeling
• Spatial analysis

Background:

• Spatial and temporal organization are key in landscape ecology and metapopulation dynamics.
• Understanding ecological processes requires identifying relevant spatial or temporal scales.
• Computational approaches are needed to integrate scale information into ecological models.

Purpose of the Study:

• To introduce a new computational method, multiscale codependence analysis (MCA).
• To test the statistical significance of correlations between variables at specific spatial or temporal scales.
• To validate MCA's performance and assess its practical utility.

Main Methods:

• Developed multiscale codependence analysis (MCA) for scale-specific correlation testing.
• Validated MCA using Monte Carlo simulations to evaluate type I and type II error rates and statistical power.
• Applied MCA to model river habitat for juvenile Atlantic salmon.

Main Results:

• MCA was found to be valid regarding type I error rate and possess sufficient statistical power.
• The method's assumptions are met across diverse ecological scenarios.
• Scale-dependent relationships were identified for Atlantic salmon parr abundance: substrate composition (0.4-4.1 km) and channel depth (200-300 m).

Conclusions:

• MCA is a statistically sound and powerful tool for identifying ecological process scales.
• Proper assessment of spatial structuring enhances ecological understanding and model representativeness.
• MCA provides valuable insights into scale-dependent ecological relationships, exemplified by Atlantic salmon habitat modeling.