Modelling the effect of directional spatial ecological processes at different scales

Area of Science:

• Ecology
• Spatial Statistics
• Biogeography

Background:

• Ecological models increasingly incorporate spatial relationships for species distribution.
• Eigenfunction-based methods like Moran's eigenvector maps (MEM) and principal coordinates of neighbour matrices (PCNM) are used for spatial structuring.
• Existing methods are limited to non-directional spatial processes.

Purpose of the Study:

• Introduce the asymmetric eigenvector map (AEM) framework for modeling directional spatial processes.
• Demonstrate AEM's applicability across various sampling schemes, data types, and spatial scales.
• Compare AEM's performance against MEM and PCNM for spatial distribution modeling.

Main Methods:

• Developed the asymmetric eigenvector map (AEM) framework.
• Applied AEM to three case studies: crustacean distribution in a river, bacterial production in a lake, and crustacean distribution on an oceanic shelf.
• Compared AEM with Moran's eigenvector maps (MEM) and principal coordinates of neighbour matrices (PCNM) without environmental variables.

Main Results:

• AEM demonstrated strong predictive power in all applications.
• AEM explained 59.8% of Atya distribution, 51.4% of bacterial production variation, and 38.4% of copepodite distribution.
• AEM consistently outperformed MEM and PCNM in the analyzed spatial modeling scenarios.

Conclusions:

• Asymmetric eigenvector map (AEM) is a powerful and appropriate tool for spatial modeling of species distributions driven by directional forces.
• AEM enhances understanding of ecological processes influenced by directional factors such as currents, wind, and historical events.
• The AEM framework offers a significant advancement in spatial ecology for analyzing complex species distributions.