Seasonal variation of geochemical properties and bacterial abundance
Water temperature and pH ranged from 20.3 - 31°C and 6.2 - 8.6, respectively. The concentrations of DOC and TN varied from 0.7 - 8.9 mg L−1 and 0.2 - 3.9 mg L−1, respectively. The value of EC and TDS ranged from 38 - 115.1 mg L−1 and 19 - 58.4 mg L−1, respectively. The concentration of Chl a ranged from 0.5 - 83.4 µg L−1. Bacterial abundance in two seasons ranged from 0.07 - 2.99 x 106 cells mL-1 (Table S1). As expected, the maximum value of bacterial abundance was measured in the spring season. Comparison based on the Kruskal-Wallis test indicated a significant (p < 0.05) difference in temperature, pH, Chl a, DOC, and TN in spring and autumn (Fig. S2).
An interesting correlation between water temperature, Chl a and water nutrients was reported in our study. Temperature showed significant positive correlation with DOC (r = 0.55, P < 0.05) and Chl a (r = 0.67, P < 0.01). Chl a showed significant positive correlation with DOC (r = 0.72, P < 0.001) and TN (r = 0.50, P < 0.05), while it was significantly negatively correlated with TDS (r = -0.47, P < 0.05). EC and TDS correlated strongly with each other i.e., (r = 0.99, P < 0.001) (Fig. S3).
Variations in diversity and community composition
After removing chimeric sequences, 753,880 high quality reads were obtained, with 30,395 - 54,782 sequences (mean = 41882 ± 6457) in each sample.
Three measures of alpha diversity indices representing diversity (Shannon diversity index), evenness (Pielou's evenness), and richness (Chao1 richness) were used. In study sites, the diversity, evenness, and richness ranged from 5.3 - 8.3, 0.50 - 0.73, and 1539 - 4810, respectively (Fig. 2). Values of diversity and evenness were reported to be significantly (p < 0.001) higher in autumn, meanwhile, the richness was significantly (p < 0.05) higher in spring.
Clear separation of samples based on the different seasons was observed in the ordination space of NMDS (Fig. 3a). Dissimilarity tests also confirmed the pattern, showing significantly distinct bacterial composition in spring and autumn bacterial communities (MRPP, ANOSIM, and perMANOVA, all P = 0.001, Table S2). Thus, pronounced seasonal variation of BCC was observed in Ramsar sites of the Central Himalayas. Furthermore, a comparison of bacterial community dissimilarity (Bray-Curtis distance) in two seasons uncovered that bacterial β-diversity in autumn was significantly higher than that in the spring (p < 0.001) (Fig. 3b).
Distribution of taxa in autumn and spring
Out of 16,599 OTUs obtained in this study, a maximum number of unique OTUs were found in the autumn, i.e., 58% (9630). 28% (4652) of the total OTUs were unique in spring. Meanwhile, the proportion of shared OTUs in autumn and spring was 14% (2317) (Fig 4).
Sequences belonging to Gammaproteobacteria (25%) and Actinobacteria (7%) dominated the shared OTUs in the autumn and spring seasons (Fig. 5a). In autumn, Actinobacteria (11%) dominated the unique OTUs (Fig. 5b). However, sequences belonging to Planctomycetes (22%) and Bacteroidetes (10%) dominated the unique OTUs in spring (Fig. 5c). SIMPER analysis revealed that 12 OTUs explained for 33% of bacterial community dissimilarity in two seasons (contribution cutoff > 1%) (Fig. S4). Of them, 5 OTUs belonged to Gammaproteobacteria, 3 OTUs belonged to Actinobacteria, 2 OTUs belonged to Verrucomicrobia, and each in Alphaproteobacteria and Planctomycetes. Among all indicator OTUs, the most remarkable was OTU0, classified into Acinetobacter johnsonii. This OTU showed an average abundance of 34% in spring while 3% in autumn, contributing to 18% of the total bacterial community dissimilarity.
Water physicochemical properties influencing bacterial biodiversity
Bacterial abundance showed a significant positive correlation with DOC (r = 0.67, P < 0.01) (Fig. S4). Shannon diversity index displayed a significant positive correlation with pH (r = 0.52, P < 0.05), but negative with Chl a (r = -0.47, P < 0.05), and temperature (r = -0.78, P < 0.001). Evenness showed significant negative correlation with temperature (r = -0.70, P < 0.001), DOC (r = -0.47, P < 0.05) and Chl a (r = -0.58, P < 0.05), respectively. Richness presented significant positive correlation with Chl a (r = 0.48, P < 0.05), and DOC (r = 0.48, P < 0.05), respectively (Fig. S5).
According to the forward selection of environmental variables (sequential test in DISTLM_forward program, 999 permutations), temperature and TDS were the significant environmental factors explaining the BCC variation. In total, all these factors explained 26% of BCC variation (Table 1). Furthermore, PLS-PM illustrated the direct and indirect effects of different environmental factors in bacterial community composition and richness. Goodness of fit (GoF) statistics values for BCC and richness were 0.54 and 0.52, respectively. GoF values signified the ease of our hypothetical path model. Water physicochemical properties negatively affected nutrients, productivity, and bacterial abundance (i.e., a total effect of -0.32, -0.41, and -0.25, respectively). Meanwhile, nutrients positively affected productivity (a total effect of 0.81) and bacterial abundance (0.65). Wetlands productivity showed a positive effect on bacterial abundance (0.23). For bacterial community composition, the total effects of physicochemical properties, nutrients, productivity, and bacterial abundance were 0.13, 0.42, 0.66, and 0.22, respectively. For richness, the total effects of physicochemical properties, nutrients, productivity, and bacterial abundance were 0.34, 0.60, -0.10, and -0.20, respectively (Fig. 6).
Potential functions of microbial communities in two seasons
A comprehensive assignment of microbial taxa to function was performed to identify bacterial potential ecological and pathogenic roles in two seasons. The result of predicted functions indicated that the majority of putative functions were enriched for aerobic chemoheterotrophy (16%), chemoheterotrophy (14%), animal parasites or symbionts (9%), aromatic compound degradation (8%), and human pathogens (8%) (Fig. 7). We also noticed that the abundance of bacteria belonging to all the aforementioned putative functions was higher in spring. Acinetobacter, Enhydrobacter, Sphingomonas, Pseudomonas, Sphingobium, Aeromicrobium, and Flavobacterium were the most abundant genera associated with aerobic chemoheterotrophy and chemoheterotrophy (Fig. S6 and S7). Acinetobacter, Candidatus Xiphinematobacter, and Clostridium were the predominant genera associated with animal parasites or symbionts (Fig. S8). Similarly, the dominant genera associated with aromatic compound degradation were Acinetobacter and Rhodococcus (Fig. S9). Bacterial members under the genus Acinetobacter, Clostridium, and Stenotrophomonas showed a higher abundance among the potential human pathogens (Fig. S10).
Though the average abundance was comparatively lesser than the five functions mentioned above, the abundance of bacterial genera associated with functions like oxygenic photoautotrophy, photoautotrophy, and phototrophy was higher in autumn. The genus Synechococcus showed a higher abundance in oxygenic photoautotrophy (Fig. S11). Synechococcus and Rhodoplanes showed a higher abundance in photoautotrophy (Fig. S12). Similarly, bacterial members of the genus Synechococcus and Rhodobacter were more abundant in phototrophy (Fig. S13).