Abstract
Understanding functions and co-occurrence patterns of microbial communities in various ecosystems enriches the knowledge on ecosystem characteristics and microbial ecology. However, such analyses have rarely been reported. Herein, functions and inter-taxa correlations of microbial communities in a set of natural environments (farmland (SA), forest soil (SB) and Caspian Sea sediments (CSS)) and engineered ecosystems (wastewater treatment plants (FW, WA and WB) and anaerobic digesters (AD)) were studied based on FAPROTAX and network analyses, respectively, by a collection of 115 samples from seven published 16S rRNA gene datasets generated by high-throughput sequencing. The results show that chemoheterotrophy related populations were the most abundant in almost all the communities. Their relative abundances (RAs) in the AD systems were the highest (43.7%±4.2%), followed by those of the soil environments (40.2%±1.9% in SA and 36.4%±2.0% in SB). For each ecosystem, the indicative community and overall community showed differentiations in several function categories. For example, the SA and SB indicative communities showed higher RAs in aerobic chemoheterotrophy, the CSS indicative community showed higher RAs in sulfate respiration, the AD indicative community showed higher RAs in fermentation, and the WB indicative community included higher RAs of predatory/exoparasitic bacteria. Three molecular ecological networks of the communities from the AD, WB and SB datasets were constructed, respectively. The WB network showed the highest proportion of negative correlations (70.4%), possibly attributed to the environmental pressure which aggravated microbial competition. The positively correlated taxa showed lower phylogenetic distances than the negatively correlated taxa on average in each network.