Soil bacterial community composition in rice–fish integrated farming systems with different planting years

Soil properties in different rice farming systems

Five treatments were designed in the three selected rice fields, including (1) rice monoculture field (RM); (2) planting area in the 1st year of rice–fish field (OP); (3) aquaculture area in the 1st year of rice–fish field (OA); (4) planting area in the 5th year of rice–fish field (FP); (5) aquaculture area in the 5th year of rice–fish field (FA). The soil properties of the five treatments were shown in Table 1. The highest soil available nitrogen (AN) content was observed in FP and was significantly higher than that in RM, OP and OA. The highest soil available phosphorus (AP) content was observed in RM and was significantly higher than that in the other 4 treatments. The highest soil available potassium (AK) content was measured in the 1st year of rice–fish field (OP and OA), followed by RM and the 5th year of rice–fish field (FP and FA), and significant differences were observed among different rice fields. The highest soil organic matter (OM) content appeared in the 5th year of rice–fish field (FP and FA), and was only significantly higher than that in OA. In addition, the soil pH in the 1st year of rice–fish field (OP and OA) was significantly lower than that in RM and the 5th year of rice–fish field (FP and FA). In summary, significant differences of soil properties were observed among the different rice farming systems.

Soil bacterial community composition

A total of 1,346,468 sequences were obtained by 16S rRNA MiSeq sequencing analysis after basal quality control (reads containing ambiguous bases were discarded; only overlapping sequences longer than 10 bp were assembled; Operational taxonomic units (OTUs) were clustered with 97% similarity). These sequences were classified as 46 phyla, 800 genera and 5335 OTUs. As shown in Fig. 1, the dominant bacterial phyla across different treatments were Proteobacteria (26.06–29.41%) and Chloroflexi (20.07–27.99%), followed by Actinobacteria (7.22–20.87%), Acidobacteria (11.36–14.46%) and Nitrospirae (3.11–8.50%). Since the implementation of rice–fish farming regime, the soil bacterial community composition has greatly changed. For example, Actinobacteria abundance decreased from 20.87% in RM to 7.22% in FA, while Nitrospirae abundance greatly increased from 3.11% in RM to 8.50% in FA. Between different areas in a same rice–fish field (i.e. OP vs OA or FP vs FA), the bacterial community composition were similar. The PCoA analysis on OTU level also showed that different areas within the same rice–fish field had high similarity in bacterial community composition. In contrast, the bacterial community composition differed distinctly among different rice farming systems (Fig. 2). Bacterial alpha diversity indices, as evaluated by Shannon, Simpson, ACE and Chao1, were shown in Table 2. Student’s t-test was adopted to evaluate the difference among treatments. The results showed that the alpha indices of FP were significantly lower than other treatments, except for Simpson index.

Based on the Kruskal–Wallis test, the statistical differences among treatments were evaluated in the abundances of the top 15 phyla. The results showed that 5 phyla, including Actinobacteria, Nitrospirae, Bacteroidetes, Unclassified_k_norank and SBR1093 were observed significant differences among treatments, and the most significant phylum was Nitrospirae (Fig. 3). In order to trace the source of the significant differences, the Wilcoxon tests were conducted between every two rice cultivation patterns separately (Fig. 4). The results indicated that the significant differences were mainly derived from the comparison between RM and F_group (FP & FA), as well as the comparison between the O_group (OP & OA) and F_group. In the comparison between the RM and O_group, only the phylum Gemmatimonadetes was observed to have a significant difference. Furthermore, we also compared the differences of the top 15 phyla between planting area (P_group) and aquaculture area (A_group) within rice–fish fields, and the results showed no phyla observed with significant differences in the abundances.

Cluster analysis on genus level

The community heatmap of the top 30 genera is shown in Fig. 5. The genera Nitrospira, Anaerolineaceae and Acidobacteria showed higher abundances than the other genera. The community composition on genus level also differed markedly across the different experimental groups. The clustering tree indicates that the different areas in a same rice–fish field (i.e. OP vs OA or FP vs FA) showed high similarity on genera composition and clustered together first. Among the different rice farming system, the genera composition was clear distinct with each other. Moreover, the statistical difference among the 5 experimental groups of the top 30 genera was checked with Kruskal–Wallis test. The results showed that 11 genera were observed significant differences among treatments (Fig. 6, only significance phyla presented). Some genera, such as Nitrospira, norank_f_Nitrosomonadaceae, norank_c_Ardenticatenia and norank_o_NB1-j were enriched in the 5 years of rice–fish field (FP and FA), while some genera, such as Pseudarthrobacter, Sphingomonas and Nocardioides were enriched in RM. This results indicated that the soil bacterial community composition on genus level has changed greatly since the implementation of rice–fish farming regime, which is consist with previous analysis on phylum level. In addition, we used LEfSe analysis to show the differences in the taxa from the phylum to the genus level among the 5 experimental groups (Figure S1). A total of 150 taxa were observed to have significant differences in abundances, of which 60 taxa were enriched in RM, 27 taxa were enriched in OA, 24 taxa were enriched in FA, 22 taxa were enriched in OP and 17 taxa were enriched in FP.

Correlation between bacterial community composition and soil properties

Redundancy analysis (RDA) at the OTU level was performed to establish the linkages of soil properties with bacterial community composition (Fig. 7). The results showed that the soil properties together explained 32.99% of the total variations in bacterial community composition. The bacterial community in F_group (FP and FA) was positively correlated with soil factors, including AN and OM content, EC and pH value. In contrast, the bacterial community in O_group (OP and OA) was only positively correlated with soil AK content. In addition, the Mantel test was employed to confirm the significance between soil factors and bacterial community composition. The results (Table 3) indicated that the soil community composition was significantly (P < 0.05) correlated with the selected soil factors, except for soil AP content. Soil AK content was the most influential factor that correlated with bacterial community composition.

Rice yield, quality and economic benefit

Rice yield, several quality indicators and the net economic benefit for the different rice cultivation regimes were also evaluated. As shown in Table 4, rice yield was decreased in rice–fish integrated farming systems, especially in the 1st year of rice–fish field. However, the net economic benefit in rice–fish field of the 5th year was increased due to the high economic value of aquatic animals. In the 1st year of rice–fish farming regime, the aquatic animal was not captured for sale as it had not yet reached the marketable size. Therefore, the net benefit of the 1st year of rice–fish field was lower than that in the rice monoculture. For the quality indicators, the protein content and milled rice ratio of rice–fish field were higher than rice monoculture, while the amylose content was opposite. More details for the quality and benefit analysis of rice–fish integrated farming system could be found in previous publication⁵.

Source: Ecology - nature.com