Simple and easy estimation of network properties based on linear correlation analysis

• Main Author: Yanhong Qi
• Format: Article
• Language: English
• Published: International Academy of Ecology and Environmental Sciences 2015-12-01
• Series: Network Biology
• Subjects: network; direction interactions; Pearson linear correlation; partial linear correlation; estimation
• Online Access: http://www.iaees.org/publications/journals/nb/articles/2015-5(4)/simple-and-easy-estimation-of-network-properties.pdf

An ecological network can be constructed by calculating the sampling data of taxon by sample type. A statistically significant Pearson linear correlation means an indirect or direct linear interaction between two taxa, and a statistically significant partial (net, or pure) correlation based on Pearson linear correlation means a candidate direct linear interaction between two taxa. In many cases, statistically significant partial correlations are not available, or we only need to estimate some of network properties. Based on sampling data of arthropods in different countries and periods, in present study I proved that the number of candidate direct linear interactions (y) increases with the number of indirect + direct linear interactions (x) calculated by Pearson linear correlation (y=-0.2757+0.5343x, r2=0.859, p<0.00001), and the former is approximately half of the later. The proportion of candidate direct interactions in possible maximum interactions (y percent) is approximately two-thirds of mean Pearson linear correlation (x) (y=1.9060+64.6084x, r2=0.339, p=0.023). These conclusions are expected to provide simple and easy quantities to estimate some of network properties.