Effects of Different Biological Amendments on Rice Physiology, Yield, Quality, and Soil Microbial Community of Rice–Crab Co-Culture in Saline–Alkali Soil

Abstract

The yield and quality of rice are influenced by soil conditions, and the soil issues in saline–alkaline land limit agricultural productivity. The saline–alkaline fields in the northern irrigation area of Yinchuan, Ningxia, China, face challenges such as low rice yield, poor quality, low fertilizer utilization efficiency, and soil salinity and alkalinity obstacles. To improve this situation, this study conducted experiments in 2022–2023 in the saline–alkaline rice–crab integrated fields of Tongbei Village, Tonggui Township, Yinchuan. This study employed a single-factor comparative design, applying 150 mL·hm⁻² of brassinolide (A1), 15 kg·hm⁻² of diatomaceous (A2), 30 kg·hm⁻² of Bacillus subtilis agent (A3), and an untreated control (CK) to analyze the effects of different biological amendments on rice growth, photosynthesis, yield, quality, and microbial communities. The results indicated that, compared with CK, the A3 increased the SPAD value and net photosynthetic rate by 2.26% and 28.59%, respectively. Rice yield increased by 12.34%, water use efficiency (WUE) by 10.67%, and the palatability score by 2.82%, while amylose content decreased by 8.00%. The bacterial OTUs (Operational Taxonomic Units) and fungal OTUs increased by 2.18% and 22.39%, respectively. Under the condition of applying 30 kg·hm⁻² of Bacillus subtilis agent (A3), rice showed superior growth, the highest yield (8804.4 kg·hm⁻²), and the highest microbial OTUs. These findings provide theoretical and technical support for utilizing biological remediation agents to achieve desalinization, yield enhancement, quality improvement, and efficiency in saline–alkali rice–crab co–culture paddies.

1. Introduction

China has approximately 99.13 million hectares of saline–alkali land, with Ningxia alone containing 1.532 million hectares of such land [1]. The comprehensive development and utilization of saline–alkali land hold strategic significance for ensuring national food security. Measures for the comprehensive development and utilization of saline–alkali land include hydraulic engineering, agronomic practices, chemical treatments, biological methods, and management strategies. The rice–crab integrated farming system is an important improvement method for moderately to severely saline–alkali lowland areas, which aids in salt leaching, prevents surface crusting, and enhances organic matter accumulation [2]. However, excessive use of chemical fertilizers in rice–crab co-culture systems can damage soil structure, reduce soil aeration and water permeability, and adversely affect the development of rice roots and the growth of aquatic species such as fish and crabs [3,4]. Organic amendments, due to their environmental friendliness and cost-effectiveness, have become a highly advantageous management strategy, including biochar, farmyard manure, compost, crop residues, and biofertilizers [5].

In recent years, research on improving saline–alkali soils for rice cultivation has expanded to include a wide variety of biological amendments, such as microbial inoculants and plant growth regulators [6,7]. These amendments improve soil structure, regulate soil pH, and enhance the availability of essential nutrients, thereby increasing rice tolerance to saline–alkali conditions and improving yields and quality [8]. This provides a new direction for addressing the challenges of utilizing saline–alkali land [9,10]. Among plant growth regulators, brassinosteroids are known as the “sixth class” of plant hormones due to their high physiological activity and unique role in regulating plant growth and development [11]. They are characterized by high efficiency, broad–spectrum activity, and non-toxicity, helping plants mitigate abiotic stress [12]. With a structure similar to steroid hormones in insects and animals, the exogenous application of brassinosteroids to plant tissues induces cell elongation, proliferation, differentiation, and organ bending [13]. Additionally, they influence many other physiological processes, promoting plant growth, coordinating nutrient balance, enhancing yield and quality, improving drought and cold resistance, and strengthening overall stress tolerance [14,15].

Applying brassinosteroids significantly reduces Na⁺ content within plants, facilitates osmotic regulation by increasing the accumulation of organic-compatible solutes, and decreases inorganic ion accumulation [16]. Another plant growth regulator, polysilicon diatomaceous, effectively enhances plant resilience under adverse conditions [17]. Silicon is a common element in rice, and studies have shown that rice xylem sap is rich in silicic acid [18], with silica deposits identified in rice cell walls, which stabilize cell wall structure during cell division [19]. Furthermore, silica deposits in stomata are believed to reduce water loss during transpiration [20]. Silicon, primarily present in natural water as monosilicic acid [21], is readily absorbed by plants. Silicon also activates plant defense mechanisms, alters photosynthesis, and improves water and carbon use efficiencies [22].

Bacillus subtilis is a type of bacteria that colonizes the rhizosphere and promotes plant growth through various mechanisms, such as producing plant hormones, solubilizing essential minerals, nitrogen fixation, and synthesizing antimicrobial agents against plant pathogens [23]. Research has shown that Bacillus subtilis RR4 exhibits chemotaxis toward malic acid, inducing its biosynthesis in rice roots [24]. Since malic acid is an intermediate of the tricarboxylic acid cycle, it plays a role in pH regulation, nutrient assimilation, and stomatal function [25]. Bacillus subtilis colonization also increases plant calcium and magnesium content, enhances dry weight, and protects rice from cadmium stress. Additionally, it promotes phosphate solubilization, indole-3-acetic acid (IAA) production, salicylic acid accumulation in roots, and ethylene regulation via ACC deaminase activity [26,27]. Jamily et al. evaluated the impact of inoculating Bacillus subtilis C–3102 on rice through pot experiments and observed a 10–20% increase in dry matter yield [28]. Furthermore, Zhang et al. isolated a salt–tolerant bacterium, Paenibacillus sp. C1, which exhibited excellent acid production, phosphate solubilization, and exopolysaccharide production under high saline–alkali conditions. This bacterium increased available phosphorus (AP), organic phosphorus (OP), total organic carbon (TOC), polysaccharide (PS) content, and alkaline phosphatase (ALP) activity in saline–alkali soils [29]. Bacillus subtilis solubilizes phosphate by producing organic acids and phytase, which break down insoluble phosphate compounds into soluble forms that plants can absorb. The nitrogen fixed by these bacteria is converted into ammonium or other nitrogenous compounds, which are then taken up by plant roots and used for growth and metabolism [30].

Currently, the effects of biological amendments on rice growth characteristics, photosynthetic efficiency, yield, and microbial communities in rice–crab co-culture systems is unclear. To address this, a two–year (2022–2023) field experiment with three treatments (brassinosteroids, diatomaceous, and Bacillus subtilis) was conducted to investigate their impacts on rice growth, photosynthesis, yield, and quality. The objectives of this study were to (1) examine the effects of different biological amendments on rice yield and quality; (2) analyze how these amendments alter soil bacterial and fungal community structures using high-throughput Illumina MiSeq sequencing technology; and (3) evaluate the influence of soil microbial communities on rice yield and quality. Based on the aforementioned objectives, we hypothesize the following: 1. The application of biological amendments will significantly enhance rice yield and quality compared to soil that has not been treated with biological amendments. 2. The effects of different types of biological amendments on soil microbial communities will vary, with diatomaceous potentially being more effective in promoting microbial diversity. 3. Changes in the soil microbial community will have a significant impact on rice yield and quality, thereby supporting the use of biological amendments in the amelioration of saline–alkali soils.

2. Materials and Methods

2.1. Overview of the Study Area

The experimental site was located at the Tonggui Township Demonstration Base in Yinchuan, Ningxia (38°30′ N, 106°28′ E), characterized by a typical temperate continental arid climate (Figure 1). The average annual temperature, precipitation, and wind speed were 9.3 °C, 195.68 mm, and 2.3 m·s⁻¹, respectively, with an altitude of 1106 m. During the rice growing seasons in 2022 and 2023, the recorded precipitation was 166.78 mm and 76.06 mm, respectively (Figure 1). The soil was sandy loam with a bulk density of 1.43 g·cm⁻³ in the top 40 cm, a field water-holding capacity of 22.5%, and a saturation moisture content of 50.2%. The physicochemical properties of the surface soil (0–20 cm) were analyzed before rice planting, and the results are shown in

Table 1. The physical–chemical properties of the experimental soils.

Year	pH	Organic Matter (g·kg−1)	Total N (g·kg−1)	Olsen Phosphorus (mg·kg−1)	Available Potassium (mg·kg−1)
2022	8.46	12.1	1.2	23.0	220.0
2023	8.17	16.9	1.0	39.0	214.8

.

2.2. Experimental Design

This study used a single-factor comparative design under saline–alkali conditions in a rice–crab co-culture system. Three treatments were applied: brassinosteroid at 150 mL·hm⁻² (A1), diatomaceous at 15 kg·hm⁻² (A2), and Bacillus subtilis at 30 kg·hm⁻² (A3), with an untreated control (CK). Application rates followed the product recommendations. The brassinosteroid contained 0.0075% 1,4-epibrassinolide as the active ingredient; the Bacillus subtilis inoculant had ≥500 million viable bacteria/mL and ≥100 g/L organic matter; and the diatomaceous contained ≥5% enzymatically hydrolyzed alginic acid, ≥5% monosilicic acid, and ≥10% trace elements. To more comprehensively and accurately assess the effects of different treatments on rice and soil, this experiment was conducted as a field trial over two consecutive years, with each treatment replicated three times in a plot of 13,356 square meters (126 m × 106 m).

2.3. Experimental Implementation

The rice variety used was the mid-early-maturing fragrant rice variety “Mengxiang”, with a growing period of approximately 148 days. The crab variety was Liao crab (Eriocheir sinensis). Rice seeds were sown directly into the field using specialized machinery at a row-to-plant spacing of 20 cm × 10 cm and a seeding rate of 337.5 kg·hm⁻². The land remained fallow between the 2022 rice harvest and the 2023 planting season, with no other crops cultivated during this period

In the first year (2022), sowing took place on 1 May, with harvest on 28 September. Basal fertilizer was applied on 21 April, comprising cow manure at 15,000 kg·hm⁻² (organic matter content 30.2%), diammonium phosphate (18-46-0) at 75 kg·hm⁻², compound fertilizer (24-12-12) at 300 kg·hm⁻², and bioorganic fertilizer (3.3-3.3-3.3) at 300 kg·hm⁻². Fertilization was consistent across treatments. Urea at 150 kg·hm⁻² was top-dressed on 15 June and 23 July. Biological amendments were applied according to the experimental design on 28 July. Crab seedlings, pre-soaked in a 20 mg/L potassium permanganate solution for 10–15 min, were introduced into the fields on 10 August. Plastic barriers were set around each treatment area to prevent crab escape. The crab stocking density was 90 kg·hm⁻², and the water was drained and the crabs were collected on 5 September.

In the second year (2023), sowing occurred on 11 May, with harvest on 1 October. Basal fertilizer was applied on 7 May, consisting of compound fertilizer (28-13-10) at 300 kg·hm⁻², compound fertilizer (17-30-8) at 300 kg·hm⁻², and diammonium phosphate (18-46-0) at 150 kg·hm⁻². Urea at 150 kg·hm⁻², compound fertilizer (28-0-0) at 120 kg·hm⁻², and diammonium phosphate (18-46-0) at 90 kg·hm⁻² were top-dressed on 25 June. Biological amendments were applied on 30 July. The crab stocking density was 90 kg·hm⁻², and the water was drained and the crabs were collected on 5 September. The irrigation water was sourced from the Yellow River via the Huinong Canal, with water quality parameters shown in

Table 2. Irrigation water quality.

Years	pH	Total Nitrogen mg·L−1	Total Phosphorus mg·L−1	Ammonium Nitrogen mg·L−1	Mineralization Degreemg·L−1	CO32− g·L−1	HCO3− g·L−1	SO42− g·L−1	Cl− g·L−1	Ca2+ g·L−1	Mg2+ g·L−1	K+ g·L−1	Na+ g·L−1
2022	7.48	0.28	0.044	0.055	397	0.0061	0.13	0.21	0.057	0.056	0.050	0.27	0.26
2023	7.77	9.07	0.283	0.230	394	0.0086	0.32	0.17	0.080	0.060	0.079	0.004	0.085

.

2.4. Observation Items and Methods

2.4.1. Soil Sampling and Analysis

Surface soil (0–20 cm) was sampled at five points, and the soil samples were air-dried in a shaded area before nutrient measurements were conducted. The samples were air-dried, sieved, and then analyzed in the laboratory to determine the following properties [31]. The pH of soil was determined with a glass electrode at a water–soil suspension ratio of 2.5:1, according to a pHS−3C mv/pH detector in Shanghai, China. The total salt content of soil was determined with a glass electrode at a water–soil suspension ratio of 5:1, according to a DDS−307 μs/cm detector in Shanghai, China. Organic matter was analyzed via the potassium dichromate oxidation method, total N was determined by the Kjeldahl method, available phosphorus was assessed using the sodium bicarbonate method, and available potassium was determined by flame photometry.

2.4.2. Measurements of Growth, Photosynthesis, Yield, and Quality

During the growing period, rice plants with uniform growth were randomly selected and marked in each treatment for a total of six plants. Plant height and stem diameter were measured periodically using a measuring tape and vernier calipers. At the grain-filling stage, photosynthetic indices, including net photosynthetic rate, transpiration rate, stomatal conductance, and intercellular CO₂ concentration, were measured on clear sunny days from 10:00 to 12:00 a.m. using a portable photosynthesis system (LI-6400, LI-COR).

Grain yield was determined from a harvest area of 12.0 m² in each plot and adjusted to 14% moisture, and these grain samples were retained for quality determination. Representative samples of six hills of rice from each plot were collected to calculate the number of grains per panicle, seed-setting rate, and grain weight. Leaf chlorophyll content (SPAD value) was measured using a portable chlorophyll meter (SPAD-502, Konica Minolta, Japan). Rice quality traits, including chalkiness degree, chalky grain rate, viscosity, taste value, and amylose content, were evaluated according to GB/T 1354-2018 [32]. Protein content was determined following GB 5009.5-2016 [33].

Water use efficiency (WUE) is defined as the ratio of yield to actual evapotranspiration, calculated using Equation (1). ET_a was determined using the water balance method [34], with the calculation formulas presented in Equations (2) and (3) [35].

$$WUE=Y/{ET}_a$$

(1)

For Equation (1), Y is yield (kg) and ET_a is groundwater recharge (kg·m⁻³).

$${ET}_a=P_{\gamma }+U+I-DP-D-{}_{∆}W$$

(2)

For Equation (2), P_γ is effective precipitation, U is groundwater recharge, I is irrigation quota, D is drainage, DP is surface runoff, and _∆W is the change in soil water content (0–20 cm layer) from the start to the end of the experiment mm, with all parameters expressed in mm.

$$DP=I-{}_{∆}W-E_t$$

(3)

For Equation (3), DP is surface runoff, I is irrigation quota, _∆W is the change in soil water content (0–20 cm layer) from the start to the end of the experiment mm, and E_t is crop evapotranspiration, with all parameters expressed in mm.

2.4.3. Microbial Community Diversity Analysis

Soil samples from the rhizosphere (0–20 cm) were collected from rice fields on 29 August 2022 and 30 August 2023 using a five-point sampling method and were stored at −20 °C. Total soil DNA was extracted using the PowerSoil DNA Isolation Kit (Mobio Laboratories) [36]. The bacterial 16S rDNA (V3 + V4) region was amplified using primers 5′-ACTCCTACGGGAGGCAGCA-3′ and 5′-GGACTACHVGGGTWTCTAAT-3′, while fungal ITS2 regions were amplified with universal primers 5′-TAGAGGAAGTAAAAGTCGTAA-3′ and 5′-TTCAAAGATTCGATGATTCAC-3′. The PCR products were purified and recovered using 1% agarose gel electrophoresis and subsequently sent to Shanghai Majorbio Bio-pharm Technology Co.,Ltd for library preparation and sequencing. Raw sequencing reads were demultiplexed, trimmed, and quality-filtered to remove low-quality sequences and adapter contamination. Denoising was performed to eliminate sequences with errors. Paired-end reads were merged using FLASH v1.2.7 to generate raw tags, which were further filtered using Trimmomatic v0.33 to obtain high-quality tags [37]. Chimeric sequences were identified and removed using UCHIME v4.2, yielding the final effective data. Operational Taxonomic Units (OTUs) were clustered at a 97% sequence similarity threshold using the UCLUST algorithm in QIIME (version 1.8.0). Taxonomic annotations of the OTUs were performed using the Silva database for bacterial sequences. Alpha diversity indices, including Ace, Chao, Shannon, and Simpson, were calculated using Mothur software. Principal coordinate analysis (PCoA) based on Bray–Curtis distances was conducted to examine similarities in microbial community structure among samples. PERMANOVA was used to test the statistical significance of microbial community differences across sample groups. Differential bacterial taxa across groups were identified using LEfSe (Linear discriminant analysis Effect Size) with thresholds of LDA > 4 and p < 0.05 [38]. Functional profiles of microbial communities were predicted using FAPROTAX for bacteria and FUNGuild for fungi [39].

2.4.4. Statistical Analysis and Data Processing

Statistical analyses were conducted using Microsoft Excel 2019 and SPSS 22.0 (SPSS Inc., Chicago, IL, USA). Data visualization was carried out using Origin 2021b. Significant differences among groups were determined using Student’s t-test at the p < 0.05 significance level. Microbial community analysis and related visualizations were generated on the Majorbio Cloud Platform (https://cloud.majorbio.com).

3. Results and Analysis

3.1. Effects of Biological Amendments on Chlorophyll and Photosynthesis

The SPAD values of rice leaves under different treatments showed a trend of first increasing and then decreasing, reaching their peak during the booting stage before declining, as shown in

Table 3. Effect of different treatments on SPAD value.

2022

Treatments

6–16

6–30

7–9

7–13

7–25

8–5

8–15

8–24

9–4

9–14

Mean

A1

32.4 ± 0.5 a

40.5 ± 1.1 a

43.0 ± 0.4 a

44.7 ± 0.9 a

42.3 ± 1.2 a

44.0 ± 1.5 a

42.1 ± 0.5 a

42.9 ± 0.9 a

42.7 ± 1.1 a

17.8 ± 0.3 a

39.2 ± 8.0 a

A2

32.2 ± 2.4 a

40.7 ± 0.9 a

43.1 ± 0.8 a

44.8 ± 1.8 a

42.1 ± 1.1 a

41.8 ± 0.5 ab

42.9 ± 0.6 a

41.7 ± 0.5 a

43.0 ± 1.1 a

17.8 ± 0.6 a

39.0 ± 8.0 a

A3

32.5 ± 3.7 a

40.8 ± 0.5 a

43.1 ± 0.5 a

44.6 ± 0.9 a

42.0 ± 0.5 a

44.0 ± 0.7 a

42.2 ± 0.9 a

41.8 ± 0.4 a

42.8 ± 1.6 a

19.4 ± 1.5 a

39.3 ± 7.6 a

CK

32.2 ± 0.6 a

40.0 ± 0.6 a

42.9 ± 0.6 a

44.5 ± 1.7 a

42.0 ± 0.4 a

40.4 ± 0.6b

42.5 ± 1.0 a

42.2 ± 1.1 a

42.7 ± 0.4 a

17.7 ± 1.0 a

38.7 ± 7.9 a

2023

Treatments

6–13

6–23

7–3

7–13

7–23

8–2

8–12

8–22

9–3

9–13

Mean

A1

28.8 ± 0.7 a

36.3 ± 0.5 a

39.6 ± 1.5 a

44.2 ± 1.9 a

38.3 ± 0.8 a

42.6 ± 1.6 a

40.2 ± 0.7 a

40.0 ± 1.4 a

38.6 ± 0.6 ab

29.3 ± 1.6 a

37.8 ± 5.0 a

A2

29.2 ± 1.0 a

36.0 ± 1.4 a

39.7 ± 0.7 a

43.9 ± 0.2 a

38.2 ± 0.9 a

42.7 ± 0.8 a

40.2 ± 0.3 a

40.1 ± 1.5 a

38.1 ± 0.8 ab

29.1 ± 2.0 a

37.7 ± 5.0 a

A3

29.2 ± 0.6 a

35.4 ± 2.0 a

40.9 ± 1.0 a

43.4 ± 1.1 a

38.5 ± 0.8 a

43.0 ± 1.8 a

40.5 ± 0.3 a

40.5 ± 0.8 a

39.6 ± 2.3 a

30.4 ± 0.7 a

38.1 ± 4.9 a

CK

29.3 ± 0.5 a

33.7 ± 1.2 a

39.5 ± 0.7 a

43.8 ± 1.3 a

39.1 ± 1.7 a

40.4 ± 0.7 a

39.6 ± 0.7 a

39.1 ± 1.7 a

35.4 ± 1.9b

30.1 ± 1.2 a

37.0 ± 4.7 a

Note: Different lowercase letters in the same column indicate significant differences between treatments (p < 0.05).

. The average values over two years revealed the order of SPAD values across treatments as A3 > A1 > A2 > CK. Compared to CK, the SPAD value under the A3 increased by 2.30%, 0.94%, and 0.57% more than under A2 and A1, respectively. Over the two years, the average SPAD value during the growth period under A3 was 1.58% and 3.04% higher than under CK. During the flowering–grain-filling period, the SPAD values of each treatment were significantly higher than those of CK at certain times, but there was no significant difference for most of the period.

Figure 2 illustrates the changes in photosynthetic rate (A), transpiration rate (E), stomatal conductance (Gsw), and intercellular CO₂ concentration (Ci) under different treatments. The two-year average results ranked A values across treatments as A3 > A1 > A2 > CK. Specifically, the A3 increased A by 28.36%, 16.22%, and 13.25% compared to CK, A2, and A1, respectively. In the first year, A, E, Gsw, and Ci under the A3 increased by 35.06%, 10.79%, and 3.38% and decreased by 17.62% compared to CK, respectively. In the second year, these values increased by 22.12%, 9.49%, and 23.08%, with Ci decreasing by 2.09%. The A3 had significant effects on A, Ci, and Gsw (p < 0.05), particularly in the second year.

3.2. Effects of Biological Amendments on Yield, Quality, and Water Use Efficiency

The results from both years indicated that treatments A2 and A3 had significant effects on rice yield (p < 0.05) (

Table 4. Effect of different treatments on yield, variety, and WUE.

Years	Treatment	Yield kg·hm−2	Panicle Length cm	Panicle Weight g	100-Grain Weight g	Seed-Setting Rate %	WUE kg/m3	Crab Yield
2022	A1	9154.6 ± 200.1 a	21.5 ± 1.4 a	2.9 ± 0.4 a	3.1 ± 0.2 a	91.2 ± 4.0 a	0.64 ± 0.01 a	361.5 ± 18.0 a
	A2	8937.8 ± 305.7 ab	20.5 ± 0.5 a	2.8 ± 0.2 a	3.0 ± 0.2 a	92.8 ± 1.4 a	0.63 ± 0.02 a	345.0 ± 23.6 a
	A3	9087.9 ± 275.5 a	20.4 ± 0.7 a	2.9 ± 0.4 a	3.1 ± 0.0 a	93.1 ± 1.2 a	0.64 ± 0.02 a	355.5 ± 21.3 a
	CK	8320.8 ± 605.1 b	20.4 ± 0.8 a	2.7 ± 0.2 a	3.0 ± 0.1 a	92.7 ± 2.3 a	0.60 ± 0.04 a	309.5 ± 7.4 b
2023	A1	7976.2 ± 369.2 ab	20.8 ± 1.8 a	3.1 ± 0.1 ab	3.0 ± 0.2 ab	92.3 ± 1.2 bc	0.64 ± 0.03 ab	343.0 ± 15.1 a
	A2	8276.4 ± 459.2 a	20.4 ± 2.9 a	3.1 ± 0.2 ab	2.9 ± 0.1 b	94.8 ± 1.5 ab	0.65 ± 0.04 a	352.5 ± 18.2 a
	A3	8520.9 ± 371.4 a	20.3 ± 1.7 a	3.3 ± 0.1 a	3.2 ± 0.1 a	94.9 ± 1.8 a	0.67 ± 0.03 a	370.5 ± 13.7 a
	CK	7353.7 ± 280.5 b	20.1 ± 1.2 a	2.8 ± 0.1 b	2.9 ± 0.0 ab	90.1 ± 0.6 c	0.58 ± 0.02 b	311.5 ± 8.5 b

Note: The effective rainfall in 2022 is 92.2 mm, and the drainage is 150 mm; the effective rainfall in 2023 is 20.5 mm, and the drainage is 105 mm. Different lowercase letters in the same column indicate significant differences between treatments (p < 0.05).

). The average yield across two years ranked as A3 > A2 > A1 > CK, with A3 increasing yield by 12.34%, 2.79%, and 2.29% compared to A1, A2, and CK, respectively (Table S1). In both years, the yield of A3 treatment increased by 10.43% and 15.87% over CK, while the yield of A1 treatment increased by 11.34% and 8.47% over CK, respectively. In the first year, water use efficiency among treatments followed the order A3 = A1 > A2 > CK, while in the second year, the order was A3 > A1 > A2 > CK. The second-year water use efficiency of all treatments was significantly higher than CK, with A3 showing the highest increase of 6.69% and 14.65% over CK. The crab yield for each treatment showed a significant increase compared to the control (CK), but the differences between the treatments were minimal.

In the second year, treatment A3 significantly influenced the number of grains per panicle, while its effects on panicle length, panicle weight, panicle number, and 100-grain weight were not statistically significant (

Table 5. Effect of different treatments on rice quality.

Years	Treatments	Chalkiness Degree %	Chalkiness Rate %	Protein Content %	Viscosity mm	Taste Score	Amylose Content %
2022	A1	6.5 ± 0.6 a	19.8 ± 2.3 a	5.5 ± 0.3 b	135.8 ± 4.1 a	69.0 ± 1.0 ab	14.3 ± 0.6 b
	A2	5.2 ± 0.3 bc	13.6 ± 0.9 b	5.9 ± 0.4 b	113.5 ± 11.5 b	67.0 ± 2.0 b	16.9 ± 0.4 a
	A3	5.7 ± 0.6 ab	14.2 ± 0.7 b	7.6 ± 0.4 a	111.0 ± 9.6 b	75.0 ± 2.6 a	14.1 ± 0.5 b
	CK	4.5 ± 0.4 b	10.7 ± 1.1 c	5.3 ± 0.3 b	134.3 ± 3.3 a	71.0 ± 6.1 ab	16.0 ± 0.8 a
2023	A1	7.1 ± 0.4 a	14.6 ± 0.6 a	6.9 ± 0.2 b	127.5 ± 4.8 ab	69.0 ± 2.0 b	14.4 ± 0.3 ab
	A2	6.8 ± 0.4 a	13.2 ± 0.3 b	7.5 ± 0.2 a	120.5 ± 3.4 ab	68.0 ± 1.7 b	14.8 ± 0.3 a
	A3	6.6 ± 0.5 a	13.8 ± 0.5 ab	7.4 ± 0.4 ab	129.7 ± 1.4 a	73.0 ± 1.0 a	14.0 ± 0.3 b
	CK	5.5 ± 0.3 b	13.4 ± 0.4 b	7.4 ± 0.2 ab	116.6 ± 11.4 b	73.0 ± 1.0 a	14.6 ± 0.3 a

Note: Different lowercase letters in the same column indicate significant differences between treatments (p < 0.05).

). The chalkiness degree and chalky grain rate of A1 rice were significantly higher than those of CK (p < 0.05), while its eating quality score and amylose content were significantly lower in some years compared to CK. The amylose content of A2 rice was higher than that of CK, though not significantly, while its eating quality score was significantly lower than that of CK. The protein content and eating quality score of A3 showed no significant differences from CK and were slightly higher in some cases.

3.3. Effects of Biological Amendments on Microbial Community

The high-throughput sequencing results are shown in

Table 6. Effect of different treatments on OUTs and alpha diversity index.

	Years	Sample Name	OTUs	Simpson	Ace	Coverage
Bacteria	2022	A1	3822.00 ± 389.75 a	0.002 ± 0.0001 b	4863.64 ± 249.71 a	0.974 ± 0.0098
		A2	3266.33 ± 102.32 c	0.003 ± 0.0003 a	4405.56 ± 50.98 b	0.966 ± 0.0071
		A3	3420.33 ± 36.61 ab	0.003 ± 0.0002 a	4652.83 ± 49.03 ab	0.96 ± 0.0004
		CK	3497.33 ± 149.53 ab	0.002 ± 0.0002 b	4570.96 ± 52.77 b	0.969 ± 0.0081
	2023	A1	4129.67 ± 44.96 a	0.002 ± 0.0000 b	5068.10 ± 15.53 a	0.977 ± 0.0003
		A2	4120.00 ± 176.03 a	0.003 ± 0.0007 a	5099.33 ± 290.32 a	0.976 ± 0.0020
		A3	4388.67 ± 243.75 a	0.003 ± 0.0002 b	5399.45 ± 308.12 a	0.975 ± 0.0016
		CK	4118.33 ± 77.59 a	0.002 ± 0.0001 b	4965.81 ± 88.95 a	0.978 ± 0.0004
Fungus	2022	A1	341.33 ± 7.23 a	0.021 ± 0.0025 b	963.02 ± 70.66 ab	0.998 ± 0.0012
		A2	258.33 ± 43.52 b	0.054 ± 0.0161 ab	731.98 ± 152.58 c	0.998 ± 0.0006
		A3	352.33 ± 17.9 a	0.074 ± 0.0385 a	1062.19 ± 85.01 a	0.997 ± 0.0004
		CK	278.33 ± 41.19 b	0.041 ± 0.0162 ab	798.07 ± 135.72 bc	0.998 ± 0.0008
	2023	A1	219.00 ± 10.00 c	0.525 ± 0.2693 a	233.91 ± 7.31 b	1.000 ± 0.0001
		A2	316.33 ± 64.84 b	0.152 ± 0.0620 b	339.89 ± 72.92 a	0.999 ± 0.0002
		A3	401.33 ± 30.07 a	0.168 ± 0.0805 b	420.74 ± 35.27 a	0.999 ± 0.0002
		CK	337.00 ± 22.52 ab	0.068 ± 0.0164 b	352.02 ± 16.58 a	1.000 ± 0.0001

Note: The data in the table are the means ± standard deviations; different lowercase letters in the same column indicate significant differences between treatments (p < 0.05).

. In 2022, a total of 643,432 reads were obtained from 12 bacterial samples (three replicates for each of the three treatments and the control), with an average sequence length of 416 bp. In the same year, 654,622 reads were obtained from 12 fungal samples, with an average sequence length of 235 bp. In 2023, a total of 923,006 reads were obtained from 12 bacterial samples, with an average sequence length of 417 bp, and 1,111,173 reads were obtained from 12 fungal samples, with an average sequence length of 242 bp.

The average OTU values over the two years showed that the bacterial OTUs ranked as A1 > A3 > CK > A2, while the fungal OTUs ranked as A3 > CK > A2 > A1. The bacterial and fungal OTUs in A3 increased by 2.54% and 22.48%, respectively, compared to CK. The average values of the richness indices over the two years showed that the bacterial and fungal richness, from high to low, ranked as A3 > A1 > CK > A2. The bacterial diversity in A3 increased by 5.41% and 4.13% (Simpson, Ace) compared to CK, and the fungal diversity in A3 increased by 28.94% and 28.86% (Simpson, Ace) compared to CK. Only in the first year were the fungal OTUs and Ace index in A3 significantly higher than those in CK, while no significant differences were observed in other cases.

Figure 3A,B shows the top 10 bacterial community compositions. The bacterial community compositions of CK and A1 were similar, while those of A2 and A3 were also similar. The results from both years revealed that, compared to CK, the relative abundances of Chloroflexi, Actinobacteriota, Acidobacteriota, and Gemmatimonadota in A2 and A3 were significantly reduced, while the relative abundances of Proteobacteria, Bacteroidota, and Desulfobacterota were increased (Table S1).

The results of the first year at the genus level showed a total of 737 taxa identified across all soil samples, with 23 genera having a relative abundance of >1%. CK exhibited higher proportions of Marmoricolas and Bacillus, A2 had the highest proportion of Romboutsia, and A3 had higher numbers of Geothermobacter compared to other treatments. In the second year, 842 bacterial genera were identified, with Thiobacillus being the most abundant genus, followed by Marmoricolas, Pseudarthrobacter, Phycicoccus, Anaeromyxobacter, Nocardioides, Sphingomonas, Ellin6067, Bacillus, and Gaiella. Compared to other treatments, A1 had a higher proportion of Bacillus, while A3 had higher numbers of Thiobacillus and Phycicoccus.

The top 10 fungal communities are shown in Figure 3C,D. In the first year, the relative abundance of Ascomycota ranged from 36.32% to 43.38%, while in the second year, it exceeded 60%, especially in CK (83.17%). The relative abundance of Rozellomycota decreased from 15.46–25.12% in the first year to less than 2% in the second year. In both years, the relative abundance of Basidiomycota in A3 was significantly higher than that of CK (p < 0.05).

At the genus level, there were clear differences in fungal community composition across the four treatments (Table S1). unclassified_p_Rozellomycota (9.065–23.361%) was a common dominant genus in all four treatments, while Aspergillus (11.455%) was an additional dominant genus identified in CK. The dominant fungal genera in A2 included Mortierella (5.394%), Talaromyces (6.363%), and Sporormiella (17.842%). In A3, the dominant genera also included Sporormiella (14.683%) and Vishniacozyma (10.345%). In the second year, the differences in fungal community composition at the genus level further intensified, and there was no common dominant genus. Sporormiella (67.18% and 35.65%) was dominant in A1 and A3, respectively, while Alternaria (16.60%) and Mortierella (14.75%) were dominant in A1. The dominant genera in A3 were more diverse, including Coprinellus (28.34%), Schizothecium (19.76%), Sporormiella (11.44%), and Mortierella (5.75%).

Figure 4 presents the predicted microbial functional capacities. Compared to CK, the number of annotated functional microorganisms in A2 and A3 increased by 15.54% and 18.41%, respectively. Fermentation was the primary C metabolism mode across all treatments, followed by nitrate reduction, nitrate respiration, nitrogen respiration, and aromatic compound degradation. The abundance of annotations for these five metabolic pathways followed the order of A3 > A2 > A1 > CK. In terms of nutritional types, chemoheterotrophic and aerobic chemoheterotrophic microorganisms dominated the rhizosphere soil of the rice–crab co-culture system. Phototrophy, photoautotrophy, cyanobacteria, and oxygenic photoautotrophy microorganisms were present in all treatments, with the annotation abundance following the order of CK > A2 > A3 > A1.

After two years of application, the absolute abundance of bacterial groups related to the respiration of sulfur compounds, dark oxidation of sulfur compounds, dark sulfide oxidation, sulfite respiration, and sulfur respiration significantly increased, following the order of A2 > A3 > A1 > CK. In contrast, the absolute abundance of groups related to ligninolysis, chitinolysis, cellulolysis, and xylanolysis showed a continuous upward trend, with the annotation abundance following the order of A3 > A2 > A1 > CK. Sulfur cycling is often intricately linked with carbon, nitrogen, and metal cycling, primarily involving the interactions between sulfur-metabolizing and sulfur-reducing microorganisms in elemental cycles. The absolute abundance of groups related to manganese oxidation, iron respiration, black iron oxidation, and manganese respiration in the metal cycle showed a continuous upward trend, with annotation abundance following the order of A3 > A1 > A2 > CK.

To explore the signature microbes between different groups from phylum to genus, LEfSe analysis was conducted. For the bacterial communities, the LEfSe (LDA = 4) analysis identified 18 and 15 soil microbial biomarkers over the two years (Figure 5A,B). At the genus level, the biomarkers in the first year included the genus norank_f__norank_o__Vicinamibacterales for the CK and norank_f__norank_o__SBR1031 for the A2, while in the second year, norank_f__norank_o__Vicinamibacterales in the CK and Thiobacillus in the A2 were identified as biomarkers. For fungal communities, the LEfSe (LDA > 4) analysis identified 27 and 31 microbial biomarkers over the two years (Figure 5C,D). At the genus level, in the first year, Aspergillus was significantly higher in the CK, unclassified_p__Rozellomycota and unclassified_f__Coniochaetaceae were significantly higher in the A1, and Sporormiella, Talaromyces, and Pseudeurotium were significantly higher in the A2. In the second year, Alternaria, Cladosporium, and Vishniacozyma were significantly higher in the CK, while Sporormiella, unclassified_p__Ascomycota, and Pseudeurotium were significantly higher in the A1, A2, and A3, respectively.

4. Discussion

4.1. Response of Microbial Inoculants to Rice Physiology, Yield, Quality, and Soil Microorganisms

Compared to the control (CK), the application of Bacillus subtilis significantly improved the photosynthetic parameters, SPAD values, and yield of rice, with no significant decrease in quality. In terms of microorganisms, the application of B. subtilis significantly increased the relative abundance of Bacteroidetes, with a two-year average increase of 98.73%, particularly a 136.35% increase in the first year. Bacteroidetes are often plant pathogens, but actinobacteria can produce antibiotics to suppress plant pathogens and decompose soil organic matter [40], with Bacteroidetes primarily involved in the late-stage degradation of hemicellulose [41], suggesting that the increased abundance of Bacteroidetes contributes to carbon source utilization in rice soil.

The two-year experimental results showed that the application of Bacillus subtilis significantly increased the relative abundance of non-dominant genera (p < 0.05), especially those involved in sulfur (S) and iron (Fe) cycling. Thiobacillus and Desulfatiglans are important sulfate-reducing bacteria [42], while Rhodoferax and Geothermobacter may play functional roles in key processes such as denitrification, polyphosphate accumulation, and iron reduction [43,44]. Notably, in the A3, the relative abundance of Bacillus did not differ much in the first year, but decreased by 53.03% in the second year compared to the CK. This decrease was due to ecological niche overlap between the microbial inoculant and indigenous microorganisms, which made the colonization of the introduced Bacillus subtilis more difficult [45].

Furthermore, probiotics rapidly proliferate in the soil, suppressing harmful microorganisms by occupying ecological niches, generating large amounts of carbohydrates and extracellular polysaccharides, improving soil aggregate structure, and, under certain conditions, participating in humus formation [46]. These actions increase soil organic matter content and reduce soil salinity [47]. The enhanced cycling of soil chemical elements also further improved leaf photosynthetic efficiency, inhibited chlorophyll degradation, and promoted rice photosynthesis, ultimately increasing yield [48]. The yield of the A3 increased by 10.43% and 15.87% compared to CK in the first and second years, respectively, showing a gradual increase. This trend was related to the differences in soil bacterial and fungal abundance and the time it took for Bacillus subtilis to colonize and alter the soil microecology.

In terms of quality, Bacillus subtilis improved the cooking quality of rice while reducing the appearance quality. Cooking quality parameters (amylose content, protein content, viscosity) are core factors in rice quality, with protein and starch being the most significant determinants of rice cooking quality [49]. Lower protein and amylose content, along with higher viscosity, typically improve the taste score of rice, making the taste score slightly higher than that of the control. Chalkiness and the chalky grain rate influence the overall appearance quality of rice. Since the same amount of nitrogen fertilizer was applied in all treatments, the A3 showed varying degrees of increased chalkiness and chalky grain rate, indicating that rhizosphere microorganisms effectively participate in nutrient cycling, particularly influencing nitrogen in the carbon-to-nitrogen ratio [50]. Nitrogen promotes faster nutrient synthesis in rice plants, leading to fluctuations in grain-filling speed, which is unfavorable for the accumulation of assimilates and grain filling, thereby increasing chalkiness [51]. Research has shown that the application of Bacillus subtilis stimulates the supply of nitrogen in the soil rhizosphere (through microbial nitrogen fixation and the utilization of nitrogen sources in organic matter), slows down the nitrification process, and enhances the denitrification process in the soil [52,53].

In conclusion, the application of Bacillus subtilis to the soil promotes the growth and development of microorganisms by enhancing their ability to acquire nutritional resources (including nitrogen fixation, iron, and phosphate) [54], significantly altering the soil microbial community. The relative abundance of non-dominant genera (p < 0.05) was significantly increased, particularly for bacteria involved in sulfur and iron cycling. Soil microorganisms, during the processes of weathering rock minerals and decomposing soil organic matter, convert organic matter into inorganic forms that are available for rice, and by closely interacting with the roots, enhance the activity of functional microorganisms [55]. This process facilitates the cycling of carbon, nitrogen, sulfur, manganese, and iron, providing sustained nutrients to the rice, thereby improving its photosynthetic efficiency and yield. Since the microbial influence on the soil–root–rhizosphere–leaf system is gradual, the transient introduction of microbial inoculants causes changes in microbial diversity that can persist for a long time [56]. Higher microbial diversity in agricultural soils can enhance functional redundancy, improving soil health and resilience to environmental stress, which may support stable crop yields [57].

4.2. Differences in the Response of Biological Amendments to Rice Physiology, Yield, Quality, and Soil Microorganisms

Brassinosteroids (BRs) have been shown to significantly enhance the yield of rice, a finding consistent with the results of our study [58]. However, in terms of rice quality, our conclusions differ from Liu’s research [59]. Specifically, we observed that the application of BRs led to an increase in rice chalkiness and a decrease in the eating quality, which affected the overall quality of the rice. This discrepancy could be attributed to the concentration of BRs applied and the variety of rice used. The increase in yield is closely associated with the enhancement of photosynthesis, as brassinosteroids reduce the leaf angle by promoting cell division and elongation at the leaf joints. This alteration directly affects the structure of the leaves and improves light distribution across the leaf surface. As a result, the net CO₂ assimilation rate is enhanced, and the carboxylation rate of Rubisco is increased, as reported by previous studies [60]. In parallel, the development of roots and stems enhances the plant’s demand for soil nutrients. Endogenous hormones also improve the rice plant’s resistance and increase the efficient utilization of root nutrients [61]. Microorganisms in the rhizosphere, which have a chemotrophic nature, colonize and propagate according to soil conditions [62]. This leads to the aggregation and utilization of soil’s effective and readily available nutrients by functional microorganisms. The results showed that BRs increased plant height, stem thickness, photosynthetic parameters, and SPAD values, although these effects were not statistically significant, possibly due to differences in BR concentration and rice variety. The increase in panicle weight, dry matter, and yield was mainly due to BRs enhancing lodging resistance, which ultimately contributed to the increased yield [63]. However, the chalkiness and percentage of chalky grains increased by an average of 44.44% and 85.05%, respectively. This resulted in faster water absorption during cooking, more soluble solids, and a higher iodine value, leading to poorer eating quality and, thus, a decrease in taste value [64].

Two years of experiments showed that the application of silicon increased rice yield, with a significant effect only in the second year. However, the quality of the rice was lower than that of the control group (CK). This decrease was mainly due to the competition between phosphorus and silicon for adsorption sites on clay particles, which caused the release of phosphorus. Due to the large surface area of mineral silicon, phosphorus was more easily utilized by the plant through physicochemical mechanisms of absorption [65]. Moreover, the ability of silicon to combine with and adsorb other minerals on clay particles prevents more Na or Cl ions from binding with essential nutrients or the soil particle surface. This significantly reduces the effectiveness of soil salinity and the absorption of nutrients by plant roots [65]. Changes in nutrients and salinity lead to alterations in the soil microbial community, with a structure similar to that observed after applying microbial inoculants. The relative abundance of Firmicutes, a phylum involved in carbohydrate fermentation, nitrogen compound utilization, and the biotransformation of bile acids and other steroids, increased significantly [66]. Additionally, many bacteria in Firmicutes gain energy by hydrolyzing polysaccharides such as cellulose, xylan, arabinogalactan, and pectin [67]. Furthermore, sulfate-reducing prokaryotes (Desulfobacca, Desulfovibrio, Syntrophobacter, Desulforhopalus, Desulfarculus, Desulfobulbus, Thiobacillus) were also dominant in the microbial community. These organisms combine sulfate or sulfur respiration with nitrogen fixation, thus enhancing the utilization of carbon and nitrogen in the rhizosphere community [68]. The increase in functional microorganisms accelerates nutrient utilization in the soil, although the number of microbes involved in nitrogen metabolism, autotrophy, and manganese/iron metal metabolism was lower than in A3. Consequently, the rice yield under the diatomaceous treatment was higher than the CK but lower than A3, with rice quality falling between the two treatments.

5. Conclusions

In this study, under the rice–crab co-culture conditions, the application of brassinosteroids, diatomite, and Bacillus subtilis all demonstrated certain yield-increasing potential and showed promising prospects for rice production. Among these, the improvement effect of Bacillus subtilis at 30 kg·hm⁻² was the strongest. Compared to the control, this treatment significantly increased rice yield and promoted the diversity of soil bacteria and fungi. Therefore, the application of Bacillus subtilis is an effective and soil-friendly biological measure to mitigate the productivity barriers commonly found in rice–crab integrated systems on low-lying saline–alkali lands. Future research should further explore the long-term effects of microbial fertilizers on soil health and rice quality and evaluate their applicability in large-scale agricultural practices across different regions.