Date of Award

Summer 7-15-2016

Degree Type

Thesis

Degree Name

Master of Science in Integrative Biology (MSIB)

Department

Biology

Major Professor

Dr. Antonio Golubski

First Committee Member

Dr. William Ensign

Second Committee Member

Dr. Erik Westlund

Third Committee Member

Dr. Stuart Borrett

Abstract

The complexity of ecological systems makes it difficult to predict how one species will react to the disturbance of another. Complex systems of species’ interactions can be described as ecological networks. One way in which ecological networks can give information concerning one species’ response to the perturbation of another is through the quantification of species’ proximity to one another in the network. In this study, we evaluate communicability, a topological metric that accounts for all of the direct and indirect interactions between species in a food web without additional information concerning the strengths of species interactions. Communicability is then compared to shortest path distance, a metric only containing information about the shortest path between two species. We found that communicability outperformed shortest path distance in 89% of the significant model treatments (91% were significant) when analyzed using polynomial regression and in 75% of the significant model treatments when analyzed using the linear regression of natural logarithm transformed metric data (58% were significant). Yet, when comparing the effects of varying structural model properties, we found conflicting results between polynomial and linear analysis. Consequently, we were able to conclude that because communicability accounts for the totality of effects based on link structure between two species, it is a better predictor of how a species will respond to a perturbation. However, because of conflicting results in some of our statistical analyses, it is unclear what roles structural network properties play in communicability’s predictive abilities.

Available for download on Saturday, July 24, 2021

Share

COinS