A comparison of network, neighborhood, and node levels of analyses in two models of nitrogen cycling in the Cape Fear River Estuary

• Published in: Ecological modelling 2014-12, Vol.293, p.210-220
• Main Authors: Hines, David E., Borrett, Stuart R.
• Format: Article
• Language: English
• Subjects: Animal and plant ecology; Animal, plant and microbial ecology; Ascendency; Biological and medical sciences; Brackish; Brackish water ecosystems; Centrality; Cycles; Ecological buffer capacity; Ecological monitoring; Ecological network analysis; Ecosystems; Estuarine environments; Fresh water ecosystems; Freshwater; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Hierarchy; Mathematical models; Methods and techniques (sampling, tagging, trapping, modelling...); Network analysis; Network environ analysis; Networks; Rivers; Synecology
• ISSN: 0304-3800; 1872-7026
• DOI: 10.1016/j.ecolmodel.2013.11.013

Ecological network analysis is a set of algorithms that provide a holistic approach to the study of ecosystems. These analyses operate on at least three different hierarchical levels: network, neighborhood, and node. Network level analyses capture whole-system interactions and provide a broad view of the system; neighborhood level analyses provide relational information for specific parts or sub-networks; node level analyses offer descriptive characteristics of individual nodes. This work investigated the insights gained from each of these levels of analysis in an ecological network analysis case study. We compared two nitrogen cycling network models constructed at sites with different salinities, one oligohaline and one polyhaline, in the Cape Fear River Estuary, NC, USA as a case study to demonstrate the differences between levels of analysis. We evaluated the nitrogen cycling models at both the network and node levels, and compared these results to existing results of a neighborhood level analysis. We further compared the ecological implications of the nitrogen network comparison produced by each hierarchical level to test the null hypotheses that there would be no difference between the conclusions resulting from these levels of analysis. We found that while network level analyses showed little difference between the two nitrogen models, differences with potential ecological importance for the availability of nutrients to phytoplankton could be seen using node level analyses. The results of the existing neighborhood level analyses exhibited characteristics with similarities to the results of both the network and node level analyses. We show that higher hierarchical levels, which integrate the information contained at the lower levels, can mask potentially important signals when describing network attributes. Therefore, we conclude that ecosystem networks should be analyzed at multiple hierarchical levels to provide a complete description of system function.