Wednesday, August 5, 2009 - 3:20 PM

COS 75-6: Eigen analysis of connectivity and throughflow in biological networks

Jonathan L. Bowers1, Meridith Bartley1, Stuart R. Borrett2, and Albert J. Meier1. (1) Western Kentucky University, (2) Univeristy of North Carolina Wilmington

Background/Question/Methods

Food webs and matrices are important tools assisting in the understanding of feeding relationships and ecology.  One way of presenting the direct relationships between predators and prey is with an adjacency matrix, a binary matrix which utilizes with direct links shown as one’s and no direct link between nodes as zero’s.  Species alterations were performed on a variety of published food webs ranging from pine forests in the United States to tussock grasslands in New Zealand.  This produced a set of food webs varying in number of distinguishable taxa present, functional diversity; and from various climates and habitats.  By diversifying habitats and taxa, results are then not specific to a given system.  Predators and prey were chosen from observed food webs by using those which best fit linkage density for the given system and altered into universal predators and universal prey.  Identification of standardized eigenvectors reporting the spatial distribution of energy throughflow potential were obtained through the use of MATLAB programming language.  In addition, further efforts have created a function capable of modifying matrices and extracting their eigenvectors to report this throughflow potential.   

Results/Conclusions

When a universal prey species was created, a greater number of nodes experienced indirect paths and a greater number of total paths were observed. Creation of a universal predator also increased paths but this effect was more localized to top predators than was observed with the creation of a universal prey item. Both introductions yielded a subweb in the biological network which allows for efficient energy cycling in an ecosystem: the strongly connected component.  The dominant eigenvalue of this network reports the rate of path increase while the corresponding dominant eigenvector in the strongly connected component is the node experiencing the most throughflow potential.  The role and ecological interpretation of eigen analysis in the broader Network Environ Analysis may contribute more to the role of system structure and connectivity to the stability of ecosystems as well as keystone nodes in a network.