Partitioned Recirculating Renovation for Traditional Rice–Fish Farming Induced Substantial Alterations in Bacterial Communities Within Paddy Soil

Abstract

Integrated agriculture–aquaculture (IAA), represented by integrated rice–fish farming, offers a sustainable production method that addresses global food issues and ensures food security. Partitioned recirculating renovation based on traditional integrated rice–fish farming is an effective way to facilitate the convenient harvesting of aquatic products and avoid difficulties associated with mechanical operations. To elucidate the impact of partitioned recirculating renovation on the bacterial communities within paddy field ecosystems, we investigated the soil environmental conditions and soil bacterial communities within integrated rice–fish farming, comparing those with and without partitioned recirculating renovations. The findings indicated a significant reduction in the bacterial community richness within paddy soil in the ditch (fish farming area), along with noticeable changes in the relative proportions of the predominant bacterial phyla in both the ditch and the rice cultivation area following the implementation of partitioned recirculating renovation. In both the ditch and the rice cultivation area, partitioned recirculating renovation diminished the edges and nodes in the co-occurrence networks for soil bacterial communities and considerably lowered the robustness index, negatively impacting the stability of bacterial communities in paddy soil. Simultaneously, the partitioned recirculating renovation substantially influenced the bacterial community assembly process, enhancing the relative contributions of stochastic processes such as dispersal limitation, drift, and homogenizing dispersal. In addition, partitioned recirculating renovation significantly altered the soil environmental conditions in both the ditch and the rice cultivation area, with environmental factors being markedly correlated with the soil bacterial community, especially the total nitrogen (TN) and total phosphorus (TP), which emerged as the primary environmental drivers influencing the soil bacterial community. Overall, these results elucidated the ecological impacts of partitioned recirculating renovation on the paddy soil from a microbiomic perspective, providing a microbial basis for optimizing partitioned rice–fish systems.

1. Introduction

In recent years, as the global population keeps rising, the demand for food has been increasing, posing serious challenges to the food security supply and the agricultural ecological environment [1,2]. Rice and aquatic products play crucial roles in ensuring global food security [3,4,5]. The integrated agriculture–aquaculture (IAA) production pattern, represented by integrated rice–fish farming, capitalizes on the symbiotic relationship between rice and aquatic animals, combining rice cultivation with aquaculture to not only stabilize and increase rice yield but also enhance the production of aquatic products [6,7,8,9,10]. Consequently, it enhances agricultural productivity, ensures food security, and provides a stable source of high-quality protein [6,7,8,9,10]. This innovative agricultural production pattern not only strives to meet basic food requirements but also plays a significant role in improving the agricultural ecological environment [9,11]. As of 2023, integrated rice–fish farming in China has expanded beyond 2.86 million ha, making up 9.72% of the nation’s total rice cultivation area [12]. A range of aquatic species are involved in integrated rice–fish farming across China, among which the largemouth bass (Micropterus salmoides), an important freshwater economic fish, has also begun to be widely farmed in paddy fields [13,14].

Although integrated rice–fish farming offers substantial benefits for improving the agricultural ecological environment and ensuring the stable supply of agricultural products, differences in the environmental conditions, management operations, and production cycles required for rice cultivation and fish farming pose obstacles for the development and practical application of integrated rice–fish farming [15,16,17,18]. The integration of rice cultivation and fish farming complicates daily agricultural management, presenting difficulties in mechanized operations and fisheries harvest [15,16,17,18]. Addressing these issues can not only increase the diversity of agricultural landscapes and enhance agricultural profitability but also further facilitate the effective management of paddy fields [10]. In this situation, integrated rice–fish farming systems continue to be optimized. Partitioned recirculating renovation for traditional rice–fish farming helps to reduce the difficulties of harvesting aquatic products and implementing mechanized operations in the rice fields [19,20]. This is achieved by designating distinct areas for fish farming and rice cultivation and establishing a circular water flow between them [19,20]. However, research on such renovations and optimizations of integrated rice–fish farming systems remains primarily focused on the technical aspects, and their profound impact on the agricultural environment, particularly on soil microorganisms, still requires further exploration [19].

Rice paddies are among the most important agricultural ecosystems widely distributed across the world, containing the richest and densest known microbial populations to date [21]. In paddy soil, microbes play crucial roles in material cycling and energy flow, maintaining and contributing to these processes [22]. Environmental microbes can facilitate the biotransformation of crucial biogenic elements, enhance soil condition and fertility, and ultimately promote agricultural yields [23,24,25,26]. Employing molecular biology methods to characterize the abundance and community composition of microbes in the soil is an important approach to evaluating shifts in agricultural ecosystems and tracking ecological health [27,28,29]. Diverse production methods or management strategies may result in various environmental changes in agricultural ecosystems, thereby affecting the composition and metabolic potential of the microbiome in paddy fields [30,31,32,33,34]. Previous research has indicated that partitioned recirculating renovation in integrated rice–fish farming can alter the composition and stability of microbial communities in paddy water [20]. However, there is no existing literature studying its impact on paddy soil microbes. As a coupled agro-aquatic ecosystem, rice–fish farming demonstrates multilevel complex interactions involving primary producers (rice), secondary consumers (fish/crustaceans), decomposer networks (bacteria/fungi), and environmental modulators [34,35]. Clarifying the impact of partitioned recirculating renovations on microbial communities in paddy soil is crucial for understanding ecosystem dynamics in integrated rice–fish farming and advancing sustainable agricultural practices.

Therefore, our study utilized amplicon sequencing to analyze the variations in the soil properties and bacterial communities in integrated rice–largemouth bass farming systems with/without partitioned recirculating renovation. The impact of partitioned recirculating renovation on the structure, diversity, assembly mechanisms, and intrinsic interactions of the soil bacterial communities was explored in this study. The hypothesis was that partitioned recirculating renovation would significantly affect the structure and composition of bacterial communities within paddy soil while altering the bacterial assembly processes. The objective was to assess how partitioned recirculating renovation impacted agricultural ecosystems through microbiomics, with the aim of fostering the advancement of scientifically sound, efficient, and sustainable farming methods. This study builds on our previous work but focuses specifically on the bacterial community changes within paddy soil induced by partitioned recirculating renovation in rice–fish systems, a novel approach that has not been fully explored in prior research [20].

2. Materials and Methods

2.1. Experimental Site and Duration

This experiment took place at the research facility of the Freshwater Fisheries Research Center of the Chinese Academy of Fishery Sciences in 2021. This research spanned three months, starting on August 1 and concluding on 31 October 2021.

2.2. Experimental Design

Six standardized paddy fields were selected for integrated rice–largemouth bass farming, each covering approximately 4 mu (≈2664 m²). Three paddy fields underwent partitioned recirculating renovation (PRR), while the other three employed traditional rice–fish farming (TRFF) as a control. The traditional rice–fish farming included creating a moat-like ditch encircling the field, delving 1.2 m beneath the surface level of the field (Figure 1). The ditch served as the fish farming area; however, due to there being no barriers between the ditch and rice cultivation areas, once the paddy field was filled with water, both the fish and water could move and flow freely throughout the entire rice paddy (Figure 1). Partitioned recirculating renovation involved intensive fish farming within the ditch that is isolated from the rice cultivation area by waterproof walls (Figure 1). Water pumps facilitated the connection between the ditch (fish farming area) and the rice cultivation area (Figure 1). Ditch water was channeled into the rice cultivation area and recirculated back. Daily, thirty percent of the ditch water underwent recirculation, with pumps inactive during nighttime. The physical separation of rice and fish simplifies the implementation of mechanized operations, reducing the complexity of managing both rice and fish simultaneously. Additionally, centralized fish cultivation allows for more efficient harvesting. In this study, the rice was sown on June 15, with the variety being Nanjing 5055. Before sowing, the fields were harrowed, leveled, irrigated, and base fertilizer was applied. Tillering fertilizer was applied 40 days after sowing, followed by the application of panicle fertilizer 60 days after sowing. Largemouth bass (M. salmoides; initial weight: 40.00 ± 2.83 g) were stocked July 30 at 600 fish/paddy field. Throughout the experiment, the farmed fishes were fed twice daily (08:00, 17:00) using commercial feed provided by Changzhou Haida Biological Feed Co., Ltd. (Changzhou, China), which guaranteed minimum levels of 47.0% crude protein, 5.0% crude fat, and 18.0% crude ash. The total daily feed amount was 4% of their total body mass.

2.3. Samples Collection

Soil samples from the paddy fields were gathered on September 15, at the midpoint of the cultivation period, using equal-distance sampling and diagonal sampling methods, from the ditch and rice cultivation area in each paddy field, respectively. Soil samples were taken from the top 5 cm using a Kajak column sediment sampler (KC Denmark A/S, Silkeborg, Denmark), with impurities such as gravel and stones removed. Five soil samples were collected from the ditch and rice cultivation area in each respective paddy field, respectively. There was a total of 15 soil samples collected for the ditch and rice cultivation area in each group, respectively. After thorough mixing, 15 soil samples collected in the ditch or rice cultivation area for each group were separated into five equal segments, each segment representing an independent replicate for the ditch or rice cultivation area in each group, respectively. Soil samples were processed in two ways: a portion was dried at 60 °C and stored at −20 °C for analysis of total nitrogen (TN), total phosphorus (TP), ammonia nitrogen, nitrate, and nitrite contents, and another was placed directly into sample bags for bacterial community analysis.

2.4. Soil Properities Measurements

For the paddy soil samples, TN was determined by modified Kjeldahl digestion while TP was measured through alkali fusion–Mo-Sb anti-spectrophotometry [36,37]. Ammonia, nitrate, and nitrite were measured through extraction with potassium chloride solution followed by spectrophotometry [38]. All methods adhere to Chinese national standards: HJ 717-2014, HJ 632-2011, and HJ 634-2012 [36,37,38]. During chemical analyses, quality control was primarily achieved through the following two aspects: at least one blank test was performed for each batch of samples, with all blank test results confirmed to be below the method detection limits; the correlation coefficient (R²) of all target analyte calibration curves consistently exceeded 0.999 [36,37,38].

2.5. Soil Bacterial DNA Extraction and Sequencing

Bacterial DNA was extracted from paddy soil using the E.Z.N.A.^® Soil DNA Kit, followed by PCR amplification of the 16S rRNA V3-V4 region using primers 341F/806R. A total of 10 ng of DNA was used to construct a 20 µL PCR reaction mixture. All samples were amplified in triplicate. After extraction and purification, the resulting amplicons were quantified and pooled equimolarly, followed by sequencing on the Illumina paired-end (PE250) platform according to standard protocols.

Raw sequencing reads underwent quality filtering with FASTP (version 0.18.0, https://github.com/OpenGene/fastp (accessed on December 2021)) [39]. Clean paired-end reads were subsequently merged into raw tags using FLASH (version 1.2.11, https://ccb.jhu.edu/software/FLASH/index.shtml (accessed on December 2021)), utilizing a minimum overlap of 10 bp and allowing for a mismatch error rate of 2% [40]. Operational taxonomic units (OTUs) were then clustered at a 97% similarity threshold using UPARSE (version 7.1, http://drive5.com/uparse/ (accessed on December 2021)), and chimeric sequences were identified and removed using UCHIME (https://www.drive5.com/usearch/ (accessed on December 2021)) [41,42,43]. Following this, taxonomic assignment for each sequence was conducted with the RDP classifier (version 2.2, http://rdp.cme.msu.edu/) against the Silva 16S rRNA database (version 138) with a matching threshold of 80% [44,45]. Lastly, any OTU abundance data falling below 0.01% were excluded during normalization.

2.6. Statistical Processing

Observed species, Shannon, Chao 1, and Pielou_J indices were computed to assess bacterial community diversity. The variance in the composition of soil bacterial community across different groups was analyzed using Principal Coordinates Analysis (PCoA) grounded in weighted Bray–Curtis distances, in conjunction with Permutational Multivariate Analysis of Variance (PERMANOVA). Variations in environmental variables, alongside the diversity, composition, and functional attributes of soil bacterial communities, were statistically examined through Tukey’s Honestly Significant Difference (Tukey HSD) test. To discern the potential biological interactions within soil bacterial communities, bacterial co-occurrence network patterns based on Spearman’s rank correlation analysis were evaluated by identifying correlations deemed significant statistically, with an absolute correlation coefficient exceeding 0.6 and a p-Value below 0.05 [46]. Further, the robustness and vulnerability metrics for the bacterial co-occurrence networks were calculated to gain insights into bacterial community stability [47]. Furthermore, null model analysis and a neutral community model were employed to determine the relative contributions of deterministic and stochastic processes to bacterial community assembly in paddy soil [48,49]. The ecological processes shaping the soil bacterial community were identified by the β-nearest taxon index (βNTI) and the Raup-Crick metric (RC) in null model analysis (Figure 2) [50]. The βNTI is a standardized measurement used to assess the mean phylogenetic distance of the nearest taxa between samples/communities, and the RC is based on the null model analysis of Bray–Curtis taxonomic β-diversity index [51,52,53]. Distance-based redundancy analysis (db-RDA) was used to evaluate the relationship between soil bacterial community and environmental variables, and the relative contributions of environmental factors to soil bacterial community changes were analyzed by Random Forest analysis.

3. Results

3.1. Differences in Soil Bacterial Community Diversity

The diversities of soil bacterial communities within both the ditch and the rice cultivation area are shown in Figure 3. Compared to the TRFF group, the implementation of partitioned recirculating renovation considerably reduced the Chao1 index of the bacterial communities within paddy soil in the ditch (Figure 3a, p < 0.05). However, no notable differences in the indices of Observed species, Shannon, and Pielou_J were observed for soil bacterial communities between the TRFF and PRR groups, whether in the ditch or in the rice cultivation area (Figure 3a,b, p > 0.05). The PCoA results also revealed distinct separation of soil bacterial communities between the TRFF and PRR groups across both the ditch (PERMANOVA test, R² = 0.15, p < 0.05) and the rice cultivation area (PERMANOVA test, R² = 0.15, p < 0.05), indicating that partitioned recirculating renovation significantly influenced the beta diversity of soil bacterial communities in these environments (Figure 3c, p < 0.05).

3.2. Differences in Soil Bacterial Community Composition

In the ditch, the top ten dominant bacterial phyla in terms of relative abundance within paddy soil were Proteobacteria, Acidobacteriota, Chloroflexi, Actinobacteriota, Desulfobacterota, Bacteroidota, Verrucomicrobiota, Nitrospirota, Myxococcota, and MBNT15 (Figure 4a). The predominant bacterial families by relative abundance included Defluviicoccaceae, Anaerolineaceae, Nitrosomonadaceae, Sutterellaceae, Pedosphaeraceae, Vicinamibacteraceae, Steroidobacteraceae, Rhodocyclaceae, Caldilineaceae, and Sulfurimonadaceae (Figure 4a). In comparison with the TRFF group, the relative proportion of Chloroflexi in the ditch substantially increased in the PRR group, whereas the phylum Verrucomicrobiota and the family Pedosphaeraceae significantly decreased (Figure 4b,c, p < 0.05).

In the rice cultivation area, the predominant bacterial phyla (top 10 ranked by relative abundance) in descending order of relative abundance included Proteobacteria, Acidobacteriota, Chloroflexi, Actinobacteriota, Desulfobacterota, Bacteroidota, Nitrospirota, Myxococcota, Verrucomicrobiota, and MBNT15. The ten most abundant families were Defluviicoccaceae, Nitrosomonadaceae, Vicinamibacteraceae, Anaerolineaceae, Pedosphaeraceae, Xanthobacteraceae, Rhodocyclaceae, Sutterellaceae, Gemmatimonadaceae, and Steroidobacteraceae (Figure 5a). In the PRR group, there was a notable reduction in the proportional abundance of the phylum Acidobacteriota, alongside a marked rise in the phylum Desulfobacterota within paddy soil in the rice cultivation area when compared to the TRFF group (Figure 5b,c, p < 0.05). At the family level, no obvious differences in the predominant soil bacteria (top 10 ranked by relative abundance) were observed between the TRFF and PRR groups in the rice cultivation area (Figure 5c, p < 0.05).

3.3. Differences in Soil Bacterial Co-Occurrence Networks

Co-occurrence networks of the soil bacterial communities are illustrated in Figure 5. In the ditch, the bacterial co-occurrence network for the PRR group included 131 nodes and 930 edges, while the TRFF group comprised 144 nodes and 950 edges (Figure 6a). The clustering coefficient was 0.60 for the PRR group and slightly lower at 0.55 for the TRFF group (Figure 6a). Within the rice cultivation area, the PRR group possessed 111 nodes and 712 edges, while the TRFF group had 125 nodes and 741 edges (Figure 6b). Both groups displayed identical clustering coefficients of 0.57 (Figure 6b). Moreover, in both the ditch and the rice cultivation area, the soil bacterial community’s robustness index was significantly lower in the PRR group than in the TRFF group (Figure 6c, p < 0.05), and the vulnerability index in the PRR group exceeded that of the TRFF group (Figure 6d).

3.4. Differences in Soil Bacterial Assembly Processes

Implementing partitioned recirculating renovation in integrated rice–fish farming has changed the bacterial assembly processes within paddy soil (Figure 7). Overall, in both the ditch and the rice cultivation area, the fits of the NCM model (R²) for the soil bacterial communities were all above 0.5, indicating that the assembly processes of these bacterial communities were primarily dominant by stochastic processes (Figure 7a). Additionally, the m values of the bacterial communities in the rice cultivation area for PRR and TRFF groups were 1.012 and 1.129, respectively, while in the ditch, the m values were 0.549 and 1.133, respectively (Figure 7a). This might suggest that partitioned recirculating renovation effectively limited the bacterial migration ability between communities.

Null model analysis also revealed that in both the PRR and TRFF groups, the bacterial assembly processes were primarily dominated by stochastic processes (Figure 7b,c). Regardless of whether it was in the ditch or rice cultivation area, the proportion of deterministic processes in the bacterial community assembly in the PRR group was obviously reduced compared to that in the TRFF group (Figure 7c). In addition, for our study, the ecological processes involved in shaping the soil bacterial communities included dispersal limitation, drift, homogenizing dispersal, homogeneous selection, and heterogeneous selection, of which dispersal limitation, drift, and homogenizing dispersal were stochastic processes, while homogeneous selection and heterogeneous selection were deterministic processes (Figure 7d). More specifically, in the ditch (fish farming area), the relative contributions of the ecological processes to the soil bacterial community assembly in the PRR group were 50% drift, 30% heterogeneous selection, and 20% dispersal limitation (Figure 7d). For the TRFF group, these values were 40% drift, 40% heterogeneous selection, 10% dispersal limitation, and 10% homogeneous selection (Figure 7d). In the rice cultivation area, the relative contributions for the PRR group were 60% drift, 30% homogenizing dispersal, and 10% heterogeneous selection, while the TRFF group exhibited relative distributions of 60% drift, 20% heterogeneous selection, and 20% homogeneous selection (Figure 7d).

3.5. Correlations Between Soil Bacterial Community and Soil Environmental Conditions

The implementation of partitioned recirculating renovation led to significant differences in soil nutrient profiles across different groups (Figure 8a,b). In the ditch, the soil TN and TP contents in the PRR group were remarkably greater compared to the TRFF group; however, concentrations of ammonia, nitrate, and nitrite showed no marked differences between the two groups (Figure 8a, p < 0.05). In the rice cultivation area, the soil TN, TP, ammonia, nitrate, and nitrite levels in the PRR group were considerably lower relative to the TRFF group (Figure 8b, p < 0.05).

The db-RDA analysis further highlighted that the soil TN, TP, ammonia nitrogen, and nitrate levels were notably correlated with the soil bacterial community (Figure 8c, p < 0.05), whereas nitrite content did not exhibit a significant correlation (Figure 8c, p > 0.05). Furthermore, Random Forest analysis pinpointed TN and TP as primary environmental determinants influencing the soil bacterial community, demonstrating a more significant impact on these bacteria than other environmental variables (Figure 8d).

4. Discussion

Previous research has found that various agricultural production patterns and agricultural engineering measures can cause changes in soil microbial communities, and these changes are almost always driven by alterations in environmental factors [54,55,56]. Similarly, the current study found that, in contrast to TRFF group, partitioned recirculating renovation significantly altered the diversity, composition, co-occurrence networks, and assembly mechanisms of bacterial communities in rice paddy. Simultaneously, variations in the soil bacterial communities showed substantial correlations with soil TN, TP, ammonia, and nitrate levels. The ecological impacts of integrated rice–fish farming on environmental microbes in rice paddy are also primarily due to the alterations in environmental factors brought about by the introduction of exogenous feed and the excretion of cultured animals [34,57,58]. In the current study, the partitioned recirculating renovation transformed the ditch into an independent, isolated intensive fish farming area, which directly led to an obvious rise in the soil TN and TP levels within the ditch, while substantially reducing the soil TN and TP contents in the rice cultivation area. Nutrients in the soil, especially the TN and TP, significantly influence the growth, abundance, and activity of bacteria, playing a critical role in shaping bacterial communities [59,60]. Consequently, the notably altered soil TN and TP levels in both the ditch and the rice cultivation area caused by partitioned recirculating renovation could be the most important environmental variable driving changes in the soil bacteria.

In the PRR group, in contrast to traditional integrated rice–fish farming, the Chao1 index of the soil bacterial communities in the ditch was dramatically reduced. This might indicate the adverse effects of the partitioned recirculating renovation on the bacterial community stability within the ditch [61,62,63]. Further, the partitioned recirculating renovation notably affected the predominant bacterial phyla in the ditch and the rice cultivation area. It diminished the relative abundance of Verrucomicrobiota and elevated that of Chloroflexi within paddy soil in the ditch; it reduced the relative abundance of Acidobacteriota and enhanced that of Desulfobacterota in the soil of the rice cultivation area. Chloroflexi primarily participates in the carbon cycle and organic matter degradation in soil, while Desulfobacterota is involved in important biogeochemical processes such as methane production, sulfur oxidation/reduction, and iron/ferrous metabolism [64,65,66]. Phyla Verrucomicrobiota and Acidobacteriota play key roles in degrading complex organic matter and regulating C/N/S biogeochemical cycles in soil [67,68,69]. In addition, Chloroflexi, Verrucomicrobiota, and Acidobacteriota are all typical oligotrophic bacteria characterized by slow growth rates [68,70,71,72]. Therefore, the impact of partitioned recirculating renovation on the relative abundance of dominant bacteria in paddy soil in this study might illustrate variations in the ecological functions and growth strategies of soil bacterial communities.

Co-occurrence networks can reveal complex dynamics among environmental microbes [73]. After the implementation of partitioned recirculating renovation, notable changes occurred in the co-occurrence networks for soil bacterial communities within both the ditch and the rice cultivation area. In both areas, the bacterial co-occurrence networks in PRR group were characterized by a reduced number of edges and nodes relative to those observed in the TRFF group. A greater number of edges and nodes in the bacterial co-occurrence network generally suggests a more intricate structure within the bacterial community [29,34]. A higher level of complexity and connectivity for the bacterial community structure indicates the increase in shared ecological traits and functional redundancy among bacteria, supporting the bacterial communities in withstanding external environmental challenges and sustaining stability [74,75]. Meanwhile, in both the ditch and rice cultivation area, the robustness index for the soil bacterial community in PRR group was remarkably reduced compared to that in the TRFF group, whereas its vulnerability was increased. These results collectively revealed that the stability of the soil bacterial communities in both the ditch and rice cultivation area significantly decreased after the implementation of partitioned recirculating renovation [76,77]. These findings were also well corroborated by alterations in the Chao1 index discussed above.

In the present study, stochastic processes dominated soil bacterial community assembly, similar to previous research findings. Li et al. [78] discovered that the microbial community assembly in the surface soil of rice paddy was mainly governed by stochastic processes. Li et al. [79] studied the periphyton microbes in a subtropical paddy field and found that their community assembly processes were also dominated by stochastic mechanisms. In contrast to traditional integrated rice–fish farming, partitioned recirculating renovation amplified stochastic contributions to the soil bacterial assembly in both the ditch and the rice cultivation areas. Specifically, in the ditch, the stochastic processes that were mainly promoted included dispersal limitation and drift, whereas in the rice cultivation area, the enhanced stochastic process was homogenizing dispersal. Prior research has shown that the assembly of the bacterial community is closely related to environmental variables, and variations in environmental conditions can effectively influence different ecological processes during bacterial assembly [80,81,82]. Hence, in the current study, partitioned recirculating renovation dramatically altered the soil environmental conditions, thereby affecting the soil bacterial assembly. Additionally, the continuous water cycling created by the partitioned recirculating renovation might also increase the ecological processes related to dispersal and drift, thereby enhancing the contribution of stochastic processes in soil bacterial assembly [29,83].

5. Conclusions

To facilitate fish harvesting and mechanized operations, partitioned recirculating renovation built upon traditional integrated rice–fish farming by establishing separate areas for fish farming (ditch) and rice cultivation, which were interconnected through recirculating water. This study builds on our previous work but focuses specifically on the bacterial community changes within paddy soil induced by partitioned recirculating renovation in rice–fish systems, a novel approach that has not been fully explored in prior research. Overall, after implementing the partitioned recirculating renovation, there was a significant reduction in the soil bacterial community richness in the ditch, and obvious alterations in the relative abundances of dominant bacterial phyla in both ditch and rice cultivation area. In both the ditch and rice cultivation area, partitioned recirculating renovation reduced the edges and nodes within the soil bacterial co-occurrence and considerably lowered the robustness index, negatively impacting the stability of bacterial communities in paddy soil. Simultaneously, partitioned recirculating renovation substantially influenced the assembly process of the soil bacterial community, enhancing the relative contributions of stochastic processes such as dispersal limitation, drift, and homogenizing dispersal. The partitioned recirculating renovation significantly altered the soil environmental conditions in both the ditch and the rice cultivation area. These environmental factors were markedly associated with the bacterial community in paddy fields, with TN and TP identified as principal environmental drivers. Our findings elucidated the ecological impacts of partitioned recirculating renovation on paddy soil from a microbiomic perspective, providing reference data for further optimizing and improving this agricultural production pattern.

However, considering that this study did not obtain data on water quality and rice growth, future research incorporating these factors would provide a more comprehensive understanding of the ecological impacts of partitioned recirculating renovation. Furthermore, this study was limited to a one-year production cycle, and currently, there is no research indicating that the decline in the stability of paddy soil bacterial communities and changes in their assembly processes can significantly affect rice yield in the short term. Therefore, it is very necessary to conduct multi-year continuous monitoring to understand the dynamic changes in paddy soil and rice growth characteristics over a longer production cycle. Long-term and multi-season studies are essential to generalize these microbial findings to broader agroecological outcomes.