Probiotic bacteria, Paraburkholderia and Delftia promote rice seed germination
The seed germination rate was calculated three days after bacterial inoculation. The germination percentage was higher in the probiotic treated seeds compared to the untreated control. The highest germination percentage (89.23%) was recorded in BTL-M2 (Delftia), followed by BRRh-4 (P. fungorum; 87.69%) and untreated controls (76.92%) (Table 1).
Probiotic bacteria improve shoot and root length of rice seedlings
The root and shoot growth of rice seedlings significantly (p<0.05, One-way ANOVA) varied by the effects of probiotic bacteria at 5 DAI (Figure 1a), and 15 DAI (Figure 1b). The highest average shoot length was observed in the rice seedling obtained from the seeds treated with BRRh-4 at 10 DAI (3.0± 0.95 cm, mean ± SD) and at 15 DAI (10.1±1.08 cm) (Figure 1c). Conversely, the lowest shoot length was recorded in the untreated control plant both at 10 DAI (1.95±0.44 cm) and 15 DAI (6.76±1.19 cm) (Figure 1c). Similarly, the root lengths of the studied rice seedlings also varied significantly when obtained from seeds treated with the probiotic bacterial isolates (Figure S1). The highest average root length was recorded in the plant treated with BRRh-4 at 10 DAI (7.85±1.35 cm), however, at 15 DAI the highest root length was found in the plant treated with BTL-M2 (9.0±1.62 cm) (Figure 1d).
Effects of probiotic bacteria on shoot weight of rice seedlings
The bacterial strains, BTL-M2; Delftia and BRRh-4; P. fungorum inoculation increased the shoot weight of rice seedlings positively in the case of shoot fresh weight at 10 DAI. In this study, the highest average mean weight was recorded in the plants treated with BRRh-4 (7.41±2.8 mg) compared to that of BTL-M2 (4.82±0.7 mg) and the untreated control (3.83±1.19 mg) plants (Figure 2a). However, at 15 DAI, the highest average shoot fresh weight was found in the plants treated with BTL-M2 (33.07±6.18 mg) compared to BRRh-4 treated (31.82±4.9 mg) and untreated control plants (18.27±0.6 mg) (Figure 2a). Similarly, in the case of shoot dry weight, the highest average weight was recorded in the plants treated with BRRh-4 both at DAI 10 (1.16±0.48 mg) and DAI 15 (4.5±0.92 mg), which was statistically dissimilar with the untreated control plants. In this study, the lowest average was recorded in the untreated control plants both at DAI 10 (0.7±0.24 mg) and DAI 15 (2.43±0.74 mg) (Figure 2b). The selected bacterial inoculation also showed a positive effect on the root weight of rice seedlings (Figure 2).
The highest average root fresh weight was recorded in the plants treated with BRRh-4 at DAI 10 (1.21±0.6 mg). However, at DAI 15 the highest average root fresh weight was observed in the plants treated with M2 (33.12±6.52 mg), which was significantly higher than the control plants where the lowest average root fresh weight was recorded at both DAI 10 (6.3±1.8 mg) and DAI 15 (29.64±7.36 mg) (Figure 2c). The mean root dry weight (at 10 DAI) was recorded in BTL-M2 treated plants (1.39±0.29 gm) followed by BRRh-4 (1.21±0.60 gm) and untreated controls (0.26±0.16 gm). Likewise, in the case of root dry weight, the highest average was recorded in the plants treated with BTL-M2 at DAI 15 (4.98±0.97 gm) compared to that of BRRh-4 (4.43±0.43 gm) and untreated control (3.01±0.62 gm) plants (Figure 2d).
Paraburkholderia and Delftia improve growth and grain yield of rice under nutrient deficient conditions
The application of BTL-M2 and BRRh-4 significantly increased growth and yield of rice (Figure S2). At the zero-dose of recommended N, P, and K fertilizers treatment, the plants with bacteria had a considerably higher results compared to the rice plants treated with no bacteria. Application of both BTL-M2 and BRRh-4 increased plant heights irrespective of the fertilizer treatments. In the case 100% doses of chemical fertilizers, the average plant height obtained was 114.67±1.66 cm, whereas, in the 50% doses of chemical fertilizer combined with the bacterial strains, BTL-M2 and BRRh-4, the average plant heights obtained were 115.33±1.2 cm and 115.52±1.45 cm, respectively, which were statistically almost similar to the 100% doses of chemical fertilizers application (Table 2). The total number of tillers per hill and the effective number of tillers per hill were also higher in rice plants treated with the bacteria. Applying of 100% doses of chemical fertilizers, the total number of tillers per hill was 12.67±0.33. On the other hand, the total number of tillers per hill in the combined application of 50% doses of chemical fertilizers with the bacterial strains, BTL-M2 and BRRh-4 gave statistically similar results (12.00±0.99 and 12.91±0.33, respectively (Figure 3a). Likewise, the number of effective tillers per hill in the combined application of 50% doses of chemical fertilizers with bacterial strains BTL-M2 and BRRh-4 were 9.83±0.88 and 10.27±0.66, respectively which statistically similar with the results obtained by the full doses of chemical fertilizers (10.33±0.66) (Figure 3b) only.
Application of probiotic bacteria significantly effects the filled and unfilled grain of rice. The maximum number of filled grain was recorded in the combination of 100% doses of chemical fertilizers plus BRRh-4 (1155±45.84) treatment followed by the same doses of chemical fertilizers combined with BTLM2 (1146±26.21) and only by the treatment of 100% doses of chemical fertilizers (1146±26.21) (Figure 4a). Interestingly, the number of filled grain in 50% recommended doses of chemical fertilizers in combination with BRRh-4 and BTL-M2 were 1119±13.45 and 1113±27.59, that were statistically similar to the 100% doses of chemical fertilizers (1146±26.21) without treatment of any bacteria. Reasonably, the lowest number of filled grain was obtained from 0% doses of chemical fertilizers (645±32.56) (Figure 4a) with no treatment of bacteria. In the 50% doses of chemical fertilizers, the average number of unfilled grains were significantly lower when combined with treatment of either probiotic bacterial strain BTL-M2 (256±3.48) and BRRh-4 (253±12.35), respectively compared to the rice treated with only chemical fertilizers (382±42.55) (Table 3). The application of the probiotic bacterial strains BRRh-4 or BTL-M2 alone reduced the number of unfilled grains of rice compared to untreated control. The 1,000-grain weight obtained from the 100% recommended full doses of chemical fertilizers was 19.24±0.13 g whereas, the almost similar result (19.53±0.22 g) was obtained from the 0% dose of chemical fertilizers when combined with the treatment of BRRh-4 (Table 4). The total grain yield per pot was higher in the 50% doses of chemical fertilizers treatment when combined with the treatment of BTL-M2 (21.98±0.42 g) or BRRh-4 (21.64±0.54 g) compared to the total grain yield per pot obtained by the treatment of rice with 100% recommended doses of chemical fertilizers (20.27±0.12 g) but all these data were statistically similar (Figure 4b).
Plant growth-promoting rhizobacteria are associated with plant roots and increase plant productivity and resistance to abiotic stresses. In the current study, the application of probiotic bacteria, P. fungorum BRRh-4 and Delftia sp. BTL-M2 remarkably improved the growth and yield of rice. Interestingly, both the probiotic bacteria with 50% recommended doses of the N, P and K fertilizers resulted in almost equivalent growth and yield of rice when treated with 100% doses of these fertilizers. Our results clearly indicate that application of BRRh-4 and BTL-M2 strains could reduce the requirements for 50% N, P, and K fertilizers in rice production. Moreover, the continuous application of synthetic fertilizers (CF) has adversely affected the natural environment.
Probiotic bacteria application improves diversity of bacteria in root and rhizosphere soils of rice
Structure and composition of bacteriome
To see whether application of probiotic bacteria effect on the population and diversity of bacteriome in rhizosphere soils and the roots, we conducted a metagenomics analysis. The ribosomal (16S rRNA) gene amplicon sequencing yielded a total of 4379,162 reads with an average number of 243,286.78 reads per sample and an average GC content of 50.81% (Data S1). A total of 2,039 unique Operational Taxonomic Units (OTUs) were identified with an average number of 228.28 OTUs per sample (maximum = 608, minimum = 60). The samples of root bacteriome always possessed a higher number of OUTs per sample compared to that of the bacteriome of rhizosphere soils (Figure 5a). The rarefaction curves based on the observed genus or species metrics reached the plateau after on average 4 million reads (Figure 5b) suggesting that the depth of coverage was sufficient to capture the maximum microbial diversity within the samples. We found significant differences (PERMANOVA, p = 0.00035) in alpha-diversity (Observed species and Shannon estimated) between the root (R) and rhizosphere soil (S) metagenomes, showing higher diversity in the microbial ecosystem of root samples (Figure 5c). We compared the distribution of microbial community in chemical fertilizer (CF), CF + BTL-M2, and CF + BRRh-4 treated samples which revealed that the within-sample (alpha) diversity remained significantly higher (PERMANOVA, p = 0.0312) in the samples of combined CF and probiotics treated metagenomes than that of only CF treated group (Figure 5d). The principal coordinate analysis (PCoA) also showed significant microbial disparity (PERMANOVA, p = 0.002) among CF, CF + BTL-M2, and CF + BRRh-4 treated metagenomes (Figure 5e).
The classified sequences were aligned to 12 bacterial phyla in both root and rhizosphere soil samples (Table S1), of which 66.67% (8/12) phyla were shared between the metagenomes, and the root (R) metagenome had a unique association of 33.33% (4/12) phyla (Figure 6a). The root (R) metagenome was mainly composed of Bacteroidetes (42.91%), Firmicutes (29.03%), Proteobacteria (13.51%), Planctomycetes (5.78%), Thermi (4.15%), Actinobacteria (2.01%) and Verrucomicrobia (1.07%) (contributing to ~ 99.0% of the total sequences, Kruskal Wallis test, p = 0.021), and rest of the detected phyla had a relatively lower abundance (< 1.0%) (Figure 6b). On the other hand, the rice rhizosphere was mostly comprised of Bacteroidetes (67.25%), Firmicutes (17.32%), and Proteobacteria (14.19%) (Figure 6b, Table S1). Moreover, we detected 42 bacterial orders in both root and soil metagenomes (41and 21 orders in R and S metagenomes, respectively), of which 50% of the order was shared between the conditions (Figure 7a, Table S2). Bacteroidales (67.25%) was found as the single most predominant bacterial order detected in soil metagenome, however, other abundant orders in this metagenome were Clostridiales (15.47%), Vibrionales (6.27%), Rhodospirillales (3.53%), Bacillales (1.54%), Burkholderiales (1.37%) and Actinomycetales (1.12%) (Figure 7b, Table S2). The predominant bacterial orders in rice root metagenome were Bacteroidales (39.60%), Bacillales (19.19%), Clostridiales (9.44%), Planctomycetales (5.21%), Deinococcales (4.15%), Flavobacteriales (3.16), Sphingomonadales (2.70%), Vibrionales (2.56%), Rhizobiales (2.51%), Burkholderiales (1.80%), Actinomycetales (1.68%) and Rhodospirillales (1.34%), and rest of the orders had a relatively lower abundance (<1.0%) (Figure 7b, Table S2). Similar levels of bacterial diversity in rice root and rhizosphere soil have been reported. However, this is the first report on the modulation of bacterial diversity in the rice roots and rhizosphere soils due to the application of BRRh-4 and BTL-M2 with varying doses of N, P, and K fertilizers.
Microbial community of rice roots differs from rhizosphere soils
In contrast to rice rhizosphere soils communities, bacterial communities in root showed significantly high diversity (p = 0.019, Kruskal Wallis test). At the genus level, we obtained a total of 185 bacterial genera from both the root and rhizosphere soil metagenomes, of which 171 and 69 genera were detected in root and rhizosphere soils. The unique bacterial genera found in both rice root and rhizosphere soil metagenomes were 62.70% (116/185) and 7.57% (14/185), respectively. However, 29.7% (55/185) genera were shared between these two metagenomes (Figure 8a, Data S1). Even though, Prevotella was found as the predominant genera in both metagenomes, relative abundance of this genus remained two-fold higher in rhizosphere soil metagenome (52.02%) than the rice roots (25.04%) (Table S3). The other predominant bacterial genera detected in rhizosphere soil metagenome were Bacteroides (12.38%), Faecalibacterium (9.50%), Vibrio (5.94%), Roseomonas (3.40%), Delftia (3.02%), Ruminococcus (1.76%), Bacillus (1.20%), Dialister (1.16%), Butyrivibrio (107%). And rest of the genera had a relatively lower abundance (<1.0%) (Figure 8b, Data S1). Conversely, Bacillus (11.07%), Planctomyces (4.06%), Faecalibacterium (3.91%), Deinococcus (2.97%), Bacteroides (2.61%), Chryseobacterium (2.30%), Exiguobacterium (1.91%), Vibrio (1.78%), and Novosphingobium (1.60%) were the most abundant bacterial genera in rice roots. The rest of the genera had a lower relative abundance (<1.0%) (Table S3). Surprisingly, the root metagenome had many-fold higher unclassified bacterial genera (24.86%) than the soil metagenome (3.01%) (Figure 8b, Data S1).
Application of probiotic bacteria remarkably changes the rice associated bacterial community
The most evident differences between CF and CF + probiotics (CF + BTL-M2 and CF + BRRh-4) treatments were detected in the fraction of reads assigned to a genus. The microbial community in the treated rice metagenomes varied significantly (p = 0.002, Kruskal Wallis test) showing the potential efficacy of combined CF and probiotics therapy over sole application of CF. Of the detected bacterial genera (n = 185), 133 and 161 genera were detected in CF, and CF + probiotic bacteria treated rice groups, respectively (Figure S3a). Therefore, treatment of rice by a combination of CF and probiotic bacteria had the inclusion of 52 (28.11%) bacterial genera (Figure S3a). Of them, Planctomyces (2.98%), Chryseobacterium (1.68%), Exiguobacterium (1.40%), Novosphingobium (1.14%) and Ancylobacter (0.74%) were the most abundant genera (<0.5%) (Data S1). Although, Prevotella remained as the most abundant bacterial genus (47.19%) in both CF and CF + probiotics metagenomes, the relative abundance of this genus was found higher in CF treated samples (49.35%) than the CF + probiotics (34.26%) sample groups. The relative abundance of the rest of the bacterial genera also varied significantly (p = 0.002, Kruskal Wallis test) in both CF and CF + probiotics treated samples with predominantly identified unclassified bacterial genera in CF + probiotics samples (18.86%) (Figure 9, Data S1). The composition of microbial community and their associated relative abundances also varied significantly (p = 0.044, Kruskal Wallis test) within CF + probiotics treated samples keeping higher number of bacterial genera (153) in CF + BRRh-4 treated samples than CF + BTL-M2 treated samples (n=120) (Figure S3b). However, out of 175 genera detected in both metagenomes, 56% (98/175) genera were found to be shared (Figure S3b, Data S1). The CF + BRRh-4 treated samples were therefore enriched with higher percentage (31.42%) of unique bacterial genera compared to CF + BTL-M2 treated metagenomes (12.57%) (Figure S3b, Data S1).
Considering the divergence of microbiome composition across all of the sample groups (CF, CF + BTL-M2 and CF + BRRh-4) (Figure S4a), we found that 20.54% (38/185) bacterial genera were shared among all of the three metagenomes (Figure S4b, Data S1). These shared bacteriome were mostly represented by Prevotella (23.23 to 49.35%), Bacteroides (1.11 to 12.77%), Faecalibacterium (4.13 to 8.83%), Vibrio (1.66% to 7.12%), Roseomonas (0.87% to 3.95%) (Figure 9, Data S1). Moreover, in this metagenomic study, we also found dysbiosis effect of fertilizer dosage on the composition and diversity of the associated microbiomes in different treatment groups. For instance, we detected 25 bacterial genera in CF1 and CF2 with zero dose chemical fertilizer whereas 56 and 69 bacterial genera were detected in CF3 – CF4 and CF5 – CF6 after application of half and full doses of chemical fertilizer, respectively. Similarly, we detected 26 bacterial genera in I1CF1 and I1CF2 with Delftia + zero dose chemical fertilizer while 45 and 65 bacterial genera were detected in I1CF3 – I1CF4 and I1CF5 – I1CF6 after application of Delftia + half dose and Delftia + full dose of chemical fertilizer, respectively. Correspondingly, 29, 51 and 50 genera were detected in I2CF1 – I2CF2, I2CF3 – I2CF4 and I2CF5 – I2CF6 after application of Paraburkholderia + zero dose, Paraburkholderia + half dose, Paraburkholderia + full dose of chemical fertilizer (Figure 9, Data S1).