Early warning of red tides using bacterial and eukaryotic communities in nearshore waters

Abstract

Anthropogenic discharge activities have increased nutrient pollution in coastal areas, leading to algal blooms and microbial community changes. Particularly, microbial communities could easily be affected with variation in nutrient pollution, and thus offered a promising strategy to predict early red tides warning via microbial community-levels variation and their keystone taxa hysteretic responses to nutrient pollution. Herein high-throughput sequencing technology from 52 samples were used to explore the variation of microbial communities and find the significant tipping points with aggravating nutrient conditions in Xiaoping Island coastal area. Results indicated that bacterial and microeukaryote communities were generally spatial and seasonal heterogeneity and were influenced by the different nutrient conditions. Procrustes test results showed that the comprehensive index of organics polluting (OPI), total nitrogen (TN), inorganic nitrogen (DIN), and total phosphorus (TP) were significantly correlated with the composition of bacteria and microeukaryotes. A SEGMENTED analysis revealed that the threshold of TN, DIN, and NH4–N for bacterial community were 0.23 ± 0.091 mg/L, 0.21 ± 0.084 mg/L, 0.09 ± 0.057 mg/L, respectively. Tipping points for TN, DIN, and NH4–N agreed with the concentration during Ceratium tripos and Skeletonema costatum blooms. Co-occurrence network results found that Planktomarina, Acinetobacter, and Verrucomicrobiaceae were keystone and OPI-discriminatory taxa. The abundant changes of Planktomarina at station A1 were significantly correlated with the development of C. tripos blooms (r = 0.55, p < 0.05), and also significantly correlated with TN, DIN, and NO3–N (r≥|0.55|, p < 0.05). The abundant changes of Acinetobacter and Verrucomicrobiaceae at station C1 were significantly correlated with the development of C. tripos blooms (r ≥ 0.77, p < 0.05), and also significantly correlated with PO4–P (r ≥ 0.64, p < 0.05). The dynamic abundance of keystone taxa showed that the trend of rapid changes could be monitored 1.5 months before the occurrence of red tide. Therefore, this study provides an assessment method for early warning of red tide occurrence and factors that trigger red tide.