Stability analysis of reef fish communities based on symbiotic graph model

Abstract

The community stability of coral reefs and fish is the focus of ecological monitoring of coral reefs. Among them, the realization of effective metrics of variations in reef fish communities (i.e., the combined communities of coral reefs and fish) is important for analyzing the stability of communities as well as maintaining the ecological balance of coral reefs. Based on coral reef and fish data collected at St. John's Island from 2004 to 2010, this study proposes a symbiotic graph modeling method to express the biological relationships of reef fish communities, and a Pyramid Match graph kernel method for fusing Attributes (PMA) to quantify community fluctuations to measure interannual variability of communities. The results showed that the community similarity was low in 2006, 2007, and 2008. The total coral cover rate in the study area decreased by 32.04% from 2006 to 2007 and increased by 24% in 2008. The total number of fish fell from 3780 in 2006 to 2596 in 2007 and rose to 6249 in 2008. Among them, the proportion of herbivorous fish decreased to 30.84% in 2007. Furthermore, we have combined the Louvain algorithm with the proposed PMA method to effectively identify the regions that should be prioritized for protection. Experiments were conducted on real datasets with good results, demonstrating the potential of the proposed method to assist in the analysis of community stability and identification of priority conservation areas.