[Paper Review] A new dominance concept and its application to diversity-stability analysis

Zhanshan Ma, Aaron M. Ellison|arXiv (Cornell University)|Mar 26, 2017

Gut microbiota and health76 references20 citations

TL;DR

This paper introduces a novel dominance concept based on Lloyd's mean crowding index, defining three new metrics that link species and community-level dominance—enabling improved modeling of diversity-stability relationships. The method is applied to human vaginal microbiome data, revealing new insights into microbial community stability and disease etiology by replacing population abundance in dominance network reconstructions and serving as proxies for diversity in stability models.

ABSTRACT

We introduce a new dominance concept consisting of three new dominance metrics based on Lloyd's (1967) mean crowding index. The new metrics link communities and species, whereas existing ones are applicable only to communities. Our community-level metric is a function of Simpson's diversity index. For species, our metric quantifies the difference between community dominance and the dominance of a virtual community whose mean population size (per species) equals the population size of the focal species. The new metrics have at least two immediate applications: (i) acting as proxies for diversity in diversity-stability modeling (ii) replacing population abundance in reconstructing species dominance networks. The first application is demonstrated here using data from a longitudinal study of the human vaginal microbiome, and provides new insights relevant for microbial community stability and disease etiology.

Motivation & Objective

• To develop a new dominance framework that integrates species and community-level dominance, addressing limitations of existing community-only metrics.
• To enable more accurate modeling of diversity-stability relationships in ecological systems, particularly in microbial communities.
• To provide metrics that can serve as proxies for diversity in diversity-stability modeling, improving predictive power.
• To replace population abundance in reconstructing species dominance networks, offering a more stable and interpretable representation of community structure.

Proposed method

• The community-level dominance metric is derived as a function of Simpson's diversity index, quantifying overall community dominance.
• For species, the dominance metric measures the difference between actual community dominance and that of a virtual community with mean population size equal to the focal species.
• The metrics are based on Lloyd's (1967) mean crowding index, adapted to quantify dominance at both species and community levels.
• The method enables the reconstruction of species dominance networks using the new metrics in place of raw population abundances.
• The approach is validated using longitudinal data from the human vaginal microbiome, demonstrating utility in real-world ecological systems.
• The new metrics are applied in diversity-stability modeling, replacing traditional abundance-based measures to assess stability.

Experimental results

Research questions

• RQ1 How can dominance be quantified at both species and community levels in a way that captures their interdependence?
• RQ2 Can the new dominance metrics improve the accuracy of diversity-stability modeling in ecological systems?
• RQ3 How do these metrics perform when replacing population abundance in reconstructing species dominance networks?
• RQ4 What insights do the new metrics provide about microbial community stability in the human vaginal microbiome?
• RQ5 Can the new dominance concept enhance understanding of disease etiology linked to microbial community instability?

Key findings

• The new dominance metrics successfully link species and community-level dominance, offering a more comprehensive view of community structure than previous community-only metrics.
• Application to the human vaginal microbiome revealed new patterns in microbial community stability, suggesting implications for disease etiology.
• The metrics serve as effective proxies for diversity in diversity-stability modeling, improving model interpretability and performance.
• Replacing population abundance with the new dominance metrics in network reconstruction produced more stable and biologically meaningful dominance networks.
• The method demonstrated practical utility in longitudinal ecological data, supporting its use in complex microbial community analysis.