Optimizing Nitrogen Fertilizer Application for Synergistic Enhancement of Economic and Ecological Benefits in Rice–Crab Co-Culture Systems

Abstract

The rice–crab co-culture (RC) system is a multidimensional integrated farming model with significant potential for balancing ecological and economic benefits in paddy fields. However, improper nitrogen (N) fertilizer application exacerbates greenhouse gas (GHG) emissions, degrades water quality, and disrupts the balance of the RC ecosystem. Therefore, optimizing and improving N management strategies for the RC system is crucial to maximize its ecological and economic benefits. This study conducted a two-year field experiment to assess the impact of optimizing N application on the productivity, sustainability, and economic benefits in RC systems. Comparisons were made to compare rice and crab productions, GHG emissions, and net ecosystem economic benefit (NEEB) between the RC and rice monoculture (RM) systems under different N application rates (0, 150, 210, and 270 kg ha⁻¹) with the aim of identifying the optimal N application rate for the RC system. The results showed that the N application rate of 210 kg ha⁻¹ in the RC system improved the agronomic traits and N use efficiency, leading to a 0.4% increase in rice yield (7603.1 kg ha⁻¹) compared to the maximum rice yield in the RM system at 270 kg ha⁻¹. At this application rate, surface water quality was optimal for crabs, resulting in the highest crab yields (370.1 kg ha⁻¹) and average weights (81.1 g). The lower N application reduced the greenhouse gas intensity (GHGI) of the RC system by 13.7% compared to the RM system. The NEEB at the optimal N application rate of 210 kg ha⁻¹ in the RC system reached 8597.5 CNY ha⁻¹, which was 1265.7% higher than that of the RM system at 270 kg ha⁻¹. In summary, optimizing N application in the RC system conserves N fertilizer resources, increases rice and crab yields, and reduces GHG emissions, thereby synergistically enhancing both economic and ecological benefits. Optimizing the N application rate has greater potential in other innovative RC models, and the productivity, sustainability, and economic efficiency should be further investigated.

1. Introduction

Rice (Oryza sativa L.) is one of the most widely cultivated cereal crops, providing sustenance for approximately 3.3 billion people globally [1]. In contemporary agricultural systems, rice production relies heavily on the use of nitrogen (N) fertilizers [2,3]. However, the excessive application of N fertilizers in traditional rice cultivation systems has resulted in rice paddies exhibiting the highest greenhouse gas (GHG) emissions per unit of yield among all food crops [4,5,6]. Additionally, it has increased the risk of surface water degradation, thereby compromising environmental sustainability and adversely affecting rice production [7,8]. Therefore, appropriate N fertilizer application is crucial for ensuring food security and protecting ecosystems.

Rice–animal co-culture systems represent multidimensional integrated farming models that combine rice cultivation with the farming of aquatic animals such as crabs, shrimp, fish, turtles, and ducks [9]. These systems have been widely adopted by 40% of rice-growing countries worldwide due to their significant economic and environmental benefits [10]. China is the world’s leading producer of rice and has the largest area dedicated to rice–animal co-culture systems [11]. In China, the adoption of animal co-culture systems has reached 2.9 million ha [12], accounting for 9.5% of the rice planting area, providing 21.5 × 10⁶ t of rice and 3.9 × 10⁶ t of aquaculture products [13,14,15]. Among these, the rice–crab co-culture (RC) system is particularly notable, covering an area of 1.6 × 10⁵ ha in China [16,17]. In the RC system, crabs can effectively control weeds and reduce the use of fertilizers and pesticides [18,19], thus helping to mitigate the risk of non-point source pollution and pesticide residues [20,21]. By balancing agricultural productivity with environmental sustainability, the RC system provides substantial economic benefits to farmers while protecting ecosystems [22].

Since rice cultivation and Chinese mitten crab (Eriocheir sinensis) aquaculture are integrated into the RC system, residual feed and crab feces serve as organic N sources that meet the N requirements for rice growth [23,24]. The incorporation of these organic N sources throughout the rice growing season extends the N supply period and enhances N use efficiency, thus reducing reliance on N fertilizer application [25]. Traditional N application strategies in RC systems may waste N fertilizer or even lead to over-application, threatening rice production and sustainability. Within the RC system, rice fields provide a crucial source of natural food for crabs, which can be enriched by N application, thereby enhancing crab yields [26,27]. However, excessive N application may degrade surface water quality by increasing harmful components such as ammonia nitrogen (NH₄⁺-N), potentially jeopardizing the survival of crabs [28]. It is noteworthy that the N application rate of the RC systems is increasing every year [16]. According to the surveys in the typical area for the RC systems in China, the average N application rate was about 215 kg ha⁻¹ in 2013 [29], rising to over 250 kg ha⁻¹ in 2019 [25]. Therefore, it is crucial to determine the optimal N application rate for the RC system to ensure economic benefits for farmers, considering the potential impacts on both rice and crab production.

In addition to the uncertainty regarding the effects on rice and crab production, the N fertilizer application rate significantly influences the impact of RC systems on GHG emissions. Some studies have demonstrated that at an N application rate of 210 kg ha⁻¹, methane (CH₄) emissions in the RC system were reduced by 5.3–63.6% compared to the RM system due to the bioturbations of crabs [22]. In contrast, other studies have shown that the RC system significantly increased CH₄ emissions by 29.2–36.8% at a N application rate of 160 kg ha⁻¹ [30]. While these studies consistently suggest that RC systems reduce nitrous oxide (N₂O) emissions, the inclusion of organic N from feed and crab feces may potentially increase N₂O emissions [31]. N application stimulates CH₄ and N₂O emissions in rice paddies, but the synergistic effects of the RC system and N application rates on GHG emissions remain unclear [32]. This may lead to the optimization of N application in RC systems for yield and economic benefits while overlooking environmental costs, resulting in uncertain net ecosystem economic benefit (NEEB). Therefore, both economic and ecological benefits should be considered when developing an optimized N application strategy for RC systems to maximize NEEB.

In this study, a two-year field experiment was conducted in Northeast China to measure the rice and crab yields, GHG emissions, and NEEB of the RC and traditional rice monoculture (RM) system under different N application rates (0, 150, 210, and 270 kg ha⁻¹) to assess the impact of optimizing N application on productivity, sustainability, and economic benefits in the RC systems. The objectives of this study were to (1) assess the effects of N application on rice and crab to determine optimal N application strategies for production, (2) compare GHG emissions of the RC and RM systems under different N application rates, and (3) identify optimal N application strategies for maximizing NEEB in the RC system.

2. Materials and Methods

2.1. Study Site

The field experiment was conducted at the experimental farm of the Liaoning Provincial Saline–Alkali Land Utilization and Research Institute (41°04′ N, 122°18′ E) in Panjin, Liaoning Province, China (Figure 1). To increase the reliability of this study, the field experiments were carried out for two years, in 2020 and 2021, respectively. The region experiences a temperate monsoonal climate, with an average temperature of 22.5 °C and total precipitation of 552.3 mm during the rice planting season in 2020, and 21.9 °C and 742.7 mm in 2021 (Figure 2). The soil at the study site is characterized by saline paddy soil with properties of 0–20 cm layer as follows: pH, 7.35; bulk density, 1.28 g cm⁻³; organic matter, 37.60 g kg⁻¹; total nitrogen (TN), 1.51 g kg⁻¹; total phosphorus (TP), 0.63 g kg⁻¹; Olsen P, 42.56 mg kg⁻¹; and total potassium (TK), 34.58 g kg⁻¹.

2.2. Experimental Design

Two farming systems were implemented: the rice–crab co-culture (RC) system and the rice monoculture (RM) system. The rice variety of Yanfeng 47 was used in the experiments, planted at a spacing of 12 cm × 30 cm. Since the N application rate over 270 kg ha⁻¹ reduces the yield of this rice variety, the N application rates used in this study were 0, 150, 210, and 270 kg ha⁻¹ for RC and RM systems, respectively [33]. The N application rates of RC0 (RM0), RC150 (RM150), RC210 (RM210), and RC270 (RM270) as basal fertilizer were 0, 120, 180, and 240 kg ha⁻¹, respectively. The N application rates of tillering fertilizer were 0 kg ha⁻¹ for RC0 (RM0) and 30 kg ha⁻¹ for RC150 (RM150), RC210 (RM210), and RC270 (RM270). Diammonium phosphate (100 kg P₂O₅ ha⁻¹) and potassium sulfate (80 kg K₂O ha⁻¹) were used as basal fertilizers in all treatments. For irrigation water requirements of rice and crabs, and with reference to local water management experience, all treatments were kept flooded with surface water depth maintained at 5 cm to 10 cm throughout the rice growing season [25,30]. The stocking density of crab for all RC treatments was 9000 individuals ha⁻¹. A nylon net (0.4 m in height) was used around the RC treatment plots to prevent crab escape. To prevent crab escape, nylon nets (0.4 m in height) were placed around the RC treatment plots. Crab feed containing 35.0% crude protein was introduced daily at 17:00 from 1 June to 25 September each year. The feed input was 550 kg ha⁻¹ annually in all RC treatments, containing 54.6 g kg⁻¹ of N, 11.7 g kg⁻¹ of phosphorus, and 8.0 g kg⁻¹ of potassium. Each treatment was replicated three times and randomly distributed in 24 experimental plots measuring 15 m × 8 m (Figure 1). Rice and crab are cultivated in a single season each year, starting in May and ending in October, respectively. Details of the agricultural operations are provided in Table S1.

2.3. Sampling and Measurements

At rice maturity over two years, five randomly selected 1 m² areas of rice plants within each plot were harvested, threshed, and manually weighed to determine grain and straw yields. Concurrently, 20 rice plants were randomly selected from each plot to assess the following agronomic traits: number of productive tillers (plant⁻¹), number of filled grains (spike⁻¹), seed setting percentage (%), 1000-grain weight (g), and harvest index (%) [25]. The N concentration of the rice plant samples was determined using the Kjeldahl method to assess N use efficiency indicators, including grain N uptake (GNU, kg ha⁻¹), N recovery efficiency (NRE, %), and N partial factor productivity (NPFP, kg kg⁻¹) [34].

Crabs were harvested from all RC plots annually between 25 September and 1 October. The weight and sex of each crab were recorded to determine the yields, average weights, and economic values of crabs across the different RC treatments.

Surface water characteristics were monitored biweekly following the application of basal fertilizer (Figures S1 and S2). A multi-parameter meter (Professional Plus, YSI, Yellow Springs, OH, USA) was used to measure dissolved oxygen (DO) concentration, temperature, and pH. Additionally, surface water depth was measured using a scale. Surface water samples were collected using a five-point sampling method and analyzed for ammonium nitrogen (NH₄⁺-N) and nitrate nitrogen (NO₃⁻-N) concentrations with a continuous flow analyzer (Auto-Analyzer 3, Norderstedt, Seal, Germany).

Static dark chambers were used weekly to collect GHGs following the application of basal fertilizer, with the frequency of gas collection being increased during periods of fertilization [30]. Before the first collection, the base of the static chamber was inserted into the soil, leaving the channels uncovered to allow free movement of crabs. During gas collection from 9:00 to 11:00, a static chamber equipped with a fan, which operated during sampling, was placed on the base, and water was used to seal around its channels. Gas samples from each chamber were collected at 10 min intervals (0, 10, 20, and 30 min) and analyzed using a Shimadzu GC-2010 Plus gas chromatograph (GC) equipped with an electron capture detector (ECD) and a flame ionization detector (FID).

2.4. Calculation

GNU, NRE, and NPFP were calculated as follows:

$$GNU=N_{grain}\times Y$$

(1)

$$NRE={(TN}_i-{TN}_0)/N_{application}\times 100$$

(2)

$$NPFP=Y/N_{application}$$

(3)

where N_grain indicates grain N content; Y represents rice yield; N_application indicates the N application rate (kg ha⁻¹); TN_i represents the cumulative N uptake (kg ha⁻¹) at N application rate of 150, 210, and 270 kg ha⁻¹; TN₀ indicates the cumulative N uptake (kg ha⁻¹) in N application rate of 0 kg ha⁻¹.

GHG emission fluxes were calculated based on the linear least squares fit of the concentration time series at four points in each plot using the following formula:

$$F=\rho \times H\times dC/dt\times 273/(273+T)$$

(4)

where F represents the gas emission flux (mg m⁻² h⁻¹ for CH₄, μg m⁻² h⁻¹ for N₂O), dC/dt is the rate of change in gas concentration over time as determined by the linear regression equation (μL L⁻¹ h⁻¹ for CH₄, nL L⁻¹ h⁻¹ for N₂O), and H, ρ, and T denote the height from the water surface to the top of the sample chamber (m), the density of CH₄ or N₂O (kg m⁻³) at standard conditions, and temperature (°C), respectively.

Seasonal cumulative CH₄ and N₂O emissions (kg ha⁻¹) were calculated using the following formula:

$$E=\sum _{i=1}^n{(f}_i+f_{i+1})/2\times \left(t_{t+1}-t_i\right)\times 24\times {10}^{-2}$$

(5)

where f represents the flux of CH₄ (mg m⁻² h⁻¹) and N₂O (mg m⁻² h⁻¹), i represents the ith measurement, (t_i+1 − t_i) represents the time interval between two adjacent measurements, and 24 × 10⁻² is used for unit conversion.

The global warming potential (GWP) of CH₄ and N₂O are 28 and 298 times that of CO₂, respectively, over a 100-year time scale [35]. The GWP and greenhouse gas intensity (GHGI) were calculated using the following equations:

$$GWP\left(kg {CO}_2 eq {ha}^{-1}\right)=28\times {CH}_4+298\times N_{2}O$$

(6)

$$GHGI\left(kg {CO}_2 eq {kg}^{-1}\right)=GWP/Y$$

(7)

where CH₄ and N₂O represent seasonal cumulative emissions of CH₄ and N₂O, respectively; Y represents the rice yield.

The net ecosystem economic benefit (NEEB) were calculated to comprehensively evaluate both economic and environmental benefits [36]. The NEEB was calculated as follows:

$$NEEB=NEB-GWP costs$$

(8)

where net economic benefit (NEB) was calculated by subtracting total costs from total incomes. Total costs were calculated by summing expenses for rice seedlings, fertilizers, irrigation, electricity, crab seedlings, feed, pesticides, machinery, land rent, and labor. Expenses were calculated based on local prices for production materials and labor (Table S2). Total incomes were calculated by applying local prices to the sale of grain and crabs. Crab prices varied significantly by sex and individual weight (Table S3). GWP costs refer to the price of total CO₂ equivalent emissions, with the current price set at 174.3 CNY ton⁻¹ of CO₂ equivalent [37].

2.5. Statistical Analysis

The Shapiro–Wilk test was employed to assess the normality of the data distribution. Pearson correlation analysis was conducted to examine the relationships among crab yield, average crab weight, and surface water characteristics using SPSS (Version 29.0, IBM, Armonk, NY, USA). One-way analysis of variance (ANOVA), Duncan’s post-hoc test, and t-test were used to examine the differences between treatments. Statistical significance was evaluated at the 5% and 1% probability levels (p < 0.05 and p < 0.01). A three-way analysis of variance (year, farming system, and N application rate) was performed to analyze differences in rice yield, agronomic traits, N use efficiency indicators, crab yield, average crab weight, seasonal cumulative GHG emissions, GWP, GHGI, NEB, and NEEB among different treatments [31]. The figures were created using Microsoft Excel (Version 2016, Microsoft, Redmond, WA, USA) and Origin Pro (Version 2024, OriginLab, Northampton, MA, USA).

3. Results

3.1. Rice and Crab Production

Increasing the N application rate linearly increased the rice yield in both the RC and RM systems, with significant differences observed between years (p < 0.01; Figure 3a). Rice yield increased linearly from 4622–4950 kg ha⁻¹ to 7484–8403 kg ha⁻¹ in the RC system and from 4303–5111 kg ha⁻¹ to 7121–8031 kg ha⁻¹ in the RM system, as the N application rate increased from 0 to 270 kg ha⁻¹ over two years. At the same N application rate, the rice yield in the RC system was, on average, 2.2–6.1% higher than in the RM system. Notably, the two-year average yield of RC210 was 0.4% higher than that of RM270, despite having an N application rate that was 60 kg ha⁻¹ lower.

The N application rate had significant effects on agronomic traits, including the number of productive tillers and harvest index (p < 0.01). Additionally, the farming system significantly influenced the number of productive tillers, seed setting percentage, and harvest index (p < 0.01;

Table 1. Agronomic traits of different treatments.

Y	S	N (kg ha−1)	Number of Productive Tillers (plant−1)	Filled Grains (spike−1)	Seed Setting (%)	1000-Grain Weight (g)	Harvest Index (%)
2020	RC	0	9.3 ± 0.3 e	93.3 ± 2.9 ab	93.3 ± 0.7 a	28.0 ± 0.2 a	45.9 ± 1.1 a
		150	12.9 ± 0.3 bc	97.0 ± 6.5 ab	93.3 ± 1.5 a	28.4 ± 0.3 a	43.9 ± 0.4 ab
		210	14.2 ± 0.2 ab	91.0 ± 1.3 ab	93.0 ± 1.0 a	28.4 ± 0.2 a	44.3 ± 2.7 ab
		270	14.5 ± 0.4 a	87.8 ± 5.3 b	94.3 ± 0.9 a	28.4 ± 0.2 a	41.5 ± 1.1 bc
	RM	0	7.7 ± 0.3 f	106.0 ± 9.5 a	92.7 ± 2.3 a	27.8 ± 0.6 a	39.8 ± 2.4 c
		150	10.9 ± 0.2 d	100.0 ± 5.3 ab	94.7 ± 0.9 a	28.3 ± 0.1 a	44.2 ± 0.9 ab
		210	12.1 ± 1.5 cd	94.8 ± 3.6 ab	93.0 ± 1.2 a	28.3 ± 0.4 a	41.4 ± 1.3 bc
		270	12.9 ± 0.4 bc	88.1 ± 5.3 b	94.7 ± 0.9 a	27.7 ± 0.5 a	39.1 ± 0.9 c
2021	RC	0	8.8 ± 0.2 c	113.1 ± 1.7 a	91.8 ± 0.1 a	26.6 ± 0.6 a	47.9 ± 0.8 a
		150	12.7 ± 0.2 b	109.5 ± 9.0 a	88.6 ± 0.8 b	26.4 ± 0.2 a	45.9 ± 0.9 b
		210	14.6 ± 0.4 a	110.9 ± 7.3 a	88.7 ± 0.9 b	26.5 ± 0.7 a	49.4 ± 0.6 a
		270	15.6 ± 0.8 a	110.7 ± 4.3 a	88.7 ± 0.6 b	26.2 ± 1.0 a	44.8 ± 0.8 bc
	RM	0	9.4 ± 0.3 c	109.5 ± 6.2 a	87.2 ± 0.4 c	26.6 ± 0.6 a	44.8 ± 0.9 bc
		150	12.8 ± 0.4 b	108.5 ± 0.1 a	85.8 ± 0.2 d	26.4 ± 0.7 a	43.3 ± 0.6 cd
		210	14.5 ± 0.6 a	109.3 ± 1.6 a	85.8 ± 0.3 d	26.3 ± 0.9 a	45.0 ± 0.7 bc
		270	15.3 ± 0.4 a	107.6 ± 6.2 a	85.0 ± 0.6 d	26.2 ± 0.1 a	42.5 ± 0.6 d
F value	Y		23.2 **	31.0 **	222.7 **	24.3 **	75.4 **
	S		12.1 **	0.2	17.6 **	0.1	70.6 **
	N		106.9 **	1.1	1.3	0.1	14.6 **
	Y × S		14.9 **	1.7	24.6 **	0.1	0.3
	Y × N		1.1	0.8	5.8 **	0.1	5.7 **
	S × N		0.2	0.2	0.9	0.0	4.4 *
	Y × S × N		0.1	0.3	0.1	0.0	3.4 *

Y: year; S: farming system; N: nitrogen application rate; RC: rice–crab co-culture; RM: rice monoculture. The values mentioned in the herein are indicated as mean ± S.D. Diverse lowercase letters within each column stand for differences of statistical significance (p < 0.05). * p < 0.05, ** p < 0.01.

). In the RC system, the number of productive tillers, seed setting percentage, and harvest index changed by −1.1–17.2%, −0.3–4.1%, and 6.9–7.0%, respectively, compared to the RM system at the same N application rate over two years.

GNU, NRE, and NPFP were significantly influenced by the farming system and N application rate, with notable variations observed between years (

Table 2. N use efficiency indicators of rice in different treatments.

Y	S	N (kg ha−1)	GNU (kg ha−1)	NRE (%)	NPFP (kg)
2020	RC	0	44.7 ± 2.3 b	-	-
		150	62.7 ± 1.5 a	25.8 ± 1.4 a	44.7 ± 1.0 a
		210	65.4 ± 4.1 a	23.3 ± 1.1 a	34.2 ± 2.1 b
		270	67.2 ± 3.4 a	22.6 ± 1.4 a	27.7 ± 1.4 c
	RM	0	40.2 ± 4.6 b	-	-
		150	60.0 ± 2.6 a	24.0 ± 2.1 a	42.8 ± 1.9 a
		210	63.1 ± 3.7 a	22.2 ± 1.9 a	31.8 ± 1.9 b
		270	66.6 ± 2.8 a	22.2 ± 1.2 a	26.4 ± 1.1 c
2021	RC	0	42.6 ± 0.5 d	-	-
		150	65.0 ± 2.6 c	31.0 ± 0.6 a	45.5 ± 2.3 a
		210	79.2 ± 2.6 ab	28.9 ± 1.0 b	38.2 ± 1.5 b
		270	83.4 ± 3.8 a	28.2 ± 1.1 b	31.1 ± 1.2 c
	RM	0	43.7 ± 1.8 d	-	-
		150	61.6 ± 3.5 c	27.1 ± 1.1 b	44.5 ± 1.2 a
		210	73.6 ± 3.0 b	24.1 ± 0.6 c	36.4 ± 1.5 b
		270	77.1 ± 3.4 b	24.2 ± 0.5 c	29.7 ± 1.3 c
F value	Y		42.4 **	58.2 **	21.6 **
	S		8.0 **	27.1 **	6.5 *
	N		163.4 **	11.0 **	197.4 **
	Y × S		0.2	9.6 **	0.1
	Y × N		9.5 **	0.1	1.8
	S × N		0.2	0.2	0.1
	Y × S × N		1.2	0.3	0.0

GNU: grain nitrogen uptake; NRE: nitrogen recovery efficiency; NPFP: nitrogen partial factor productivity. Diverse lowercase letters within each column stand for differences of statistical significance (p < 0.05). * p < 0.05, ** p < 0.01.

). GNU increased significantly by 50.3–95.8% for RC270 compared to RC0 and 65.7–76.4% for RM270 compared to RM0 over two years. However, increased N application significantly reduced NRE and NPFP in both the RC and RM systems. On average, GNU, NRE, and NPFP were 4.8–5.0%, 4.7–16.9%, and 4.0–5.7% higher in the RC system compared to the RM system at the same N application rate. Notably, the number of productive tillers, filled grains, seed setting percentage, 1000-grain weight, harvest index, GNU, NRE, and NPFP of RC210 were, on average, 2.8%, 3.2%, 1.3%, 1.8%, 4.8%, 0.5%, 12.3%, and 29.1% higher than those of RM270 over two years, despite having an N application rate was 60 kg ha⁻¹ lower. Overall, the N application enhanced rice yield, with the RC system exhibiting higher yields compared to the RM system. Agronomic traits and N use efficiency were improved in the RC system. Despite a lower N application rate, rice yield in RC210 was higher than in RM270.

N application significantly affected crab yield and average weight, both of which exhibited an increasing and then decreasing trend (p < 0.01; Figure 3b,c). The crab yield and average weight of RC0 were 100.1–195.8 kg ha⁻¹ and 69.5–72.2 g, respectively. Compared to RC0, the crab yield in RC150 changed by −1.3–68.5%, reaching 168.7–193.3 kg ha⁻¹, while the average weight altered by −0.5–5.3%, reaching 71.8–73.2 g. The highest crab yield and average weight were achieved in RC210. As the N application rate increased to 210 kg ha⁻¹, the crab yield increased by 93.4–260.9% compared to RC0, reaching 361.4–378.7 kg ha⁻¹ over two years. Similarly, the average weight of crabs in RC210 increased by 11.7–17.4%, reaching 80.6–81.6 g. The crab yield and average weight in RC270 were 341.3–360.0 kg ha⁻¹ and 80.4–79.6 g, reduced by 4.9–5.5% and 0.3–2.5%, respectively, compared to RC210 over two years. Therefore, the N application rate of 210 kg ha⁻¹ was the most suitable for crab production in the RC system.

3.2. Effect of Surface Water Characteristics on Crab Production

Crab yield and average weight demonstrated a significant positive correlation with the average concentrations of DO, NH₄⁺-N, and NO₃⁻-N in surface water during the crab growing season (Figure 4). N application significantly increased the average concentrations of DO, NH₄⁺-N, and NO₃⁻-N in surface water. However, the RC system significantly decreased DO concentration in surface water compared to the RM system (Figure 5a–c). In 2020 and 2021, an N application rate of 210 kg ha⁻¹ resulted in the highest concentration of DO in surface water within the RC system, ranging from 6.6 to 6.9 mg L⁻¹. NH₄⁺-N and NO₃⁻-N concentrations in surface water peaked at a N application rate of 270 kg ha⁻¹ in the RC system, reaching 1.2–1.9 mg L⁻¹ and 1.3–1.4 mg L⁻¹, respectively. A N application rate of 210 kg ha⁻¹ resulted in the most favorable DO concentration in surface water for crabs while avoiding excessive NH₄⁺-N and NO₃⁻-N levels.

3.3. GHG Emissions, GWP, and GHGI

CH₄ emission fluxes for all treatments remained low immediately after the application of the basal fertilizer, gradually increasing after 40 days and peaking around 70 days (Figure 6a). As the N application rate increased from 0 to 270 kg ha⁻¹, the peaks of CH₄ emission flux rose linearly from 7.2–15.5 mg m⁻² h⁻¹ to 21.3–26.4 mg m⁻²h⁻¹ in the RC system and from 8.0–17.0 mg m⁻² h⁻¹ to 24.5–29.3 mg m⁻² h⁻¹ in the RM system. CH₄ emission fluxes declined gradually and nearly ceased by the time of crab harvest. After the application of the basal fertilizer, N₂O emission fluxes increased rapidly, peaking within 1–2 weeks (Figure 6b). As the N application rate increased from 0 to 270 kg ha⁻¹, the maximum N₂O emission flux increased linearly from 19.0–101.1 µg m⁻² h⁻¹ to 111.2–285.2 µg m⁻² h⁻¹ in the RC system and from 19.8–88.7 µg m⁻² h⁻¹ to 111.6–330.0 µg m⁻² h⁻¹ in the RM system. N₂O emission fluxes declined rapidly, increased again within one week after the application of the tillering fertilizer, and then decreased rapidly to almost zero.

The cumulative emissions of CH₄ and N₂O were significantly influenced by the N application rate and varied notably between years (p < 0.01;

Table 3. Seasonal cumulative CH₄ and N₂O emissions, GWP, and GHGI of different treatments.

Y	S	N (kg ha−1)	CH4 Emissions (kg ha−1)	N2O Emissions (kg ha−1)	GWP (kg CO2-eq ha−1)	GHGI (kg CO2-eq kg−1)
2020	RC	0	181.1 ± 16.0 c	0.09 ± 0.02 c	5098 ± 453 c	1.10 ± 0.04 d
		150	263.0 ± 31.0 b	0.48 ± 0.07 b	7495 ± 854 b	1.12 ± 0.10 cd
		210	292.0 ± 3.5 ab	0.65 ± 0.06 a	8353 ± 93 ab	1.17 ± 0.08 bcd
		270	330.6 ± 31.1 a	0.72 ± 0.02 a	9449 ± 875 a	1.26 ± 0.06 abcd
	RM	0	199.8 ± 30.4 c	0.09 ± 0.04 c	5621 ± 847 c	1.31 ± 0.13 abc
		150	274.3 ± 29.4 b	0.49 ± 0.08 b	7814 ± 823 b	1.22 ± 0.08 abcd
		210	313.0 ± 15.4 ab	0.64 ± 0.06 a	8934 ± 438 ab	1.34 ± 0.03 ab
		270	340.6 ± 23.0 a	0.69 ± 0.10 a	9724 ± 620 a	1.37 ± 0.10 a
2021	RC	0	117.8 ± 1.7 d	0.36 ± 0.04 c	3397 ± 60 d	0.69 ± 0.02 b
		150	207.5 ± 21.1 c	0.75 ± 0.07 b	6011 ± 579 c	0.89 ± 0.12 a
		210	248.1 ± 12.7 ab	0.75 ± 0.06 b	7148 ± 377 ab	0.89 ± 0.04 a
		270	265.2 ± 7.5 a	0.95 ± 0.08 a	7678 ± 194 a	0.91 ± 0.04 a
	RM	0	109.3 ± 21.4 d	0.33 ± 0.05 c	3152 ± 616 d	0.62 ± 0.12 b
		150	228.8 ± 14.2 bc	0.69 ± 0.08 b	6594 ± 420 bc	0.99 ± 0.09 a
		210	267.5 ± 26.6 a	0.82 ± 0.03 ab	7711 ± 738 a	1.01 ± 0.13 a
		270	282.5 ± 7.1 a	0.93 ± 0.12 a	8159 ± 230 a	1.02 ± 0.06 a
F value	Y		63.7 **	70.9 **	59.9 **	135.6 **
	S		3.5	0.1	3.5	11.9 **
	N		83.6 **	106.6 **	90.4 **	9.1 **
	Y × S		0.0	0.0	0.0	1.7
	Y × N		0.9	0.9	0.9	5.1 **
	S × N		0.1	0.1	0.2	0.3
	Y × S × N		0.3	0.4	0.3	1.1

CH₄: methane; N₂O: nitrous oxide; GWP: global warming potential; GHGI: GHG intensity. Diverse lowercase letters within each column stand for differences of statistical significance (p < 0.05). ** p < 0.01.

). The N application rate linearly increased the cumulative emissions of CH₄ and N₂O in both the RC and RM systems. In 2020 and 2021, the cumulative CH₄ emissions of the RC system increased linearly by 82.5–125.0% from 117.8–181.1 kg ha⁻¹ to 265.2–330.6 kg ha⁻¹, as the N application increased from 0 to 270 kg ha⁻¹. CH₄ emissions of the RM system increased linearly by 70.5–158.3% from 109.3–199.8 kg ha⁻¹ to 282.5–340.6 kg ha⁻¹. The increase in N₂O emissions was more pronounced. The cumulative N₂O emissions of the RC system increased linearly by 160.1–648.9% from 0.09–0.36 kg ha⁻¹ to 0.72–0.95 kg ha⁻¹ as N application increased from 0 to 270 kg ha⁻¹. N₂O emissions of the RM system increased linearly by 175.6–635.0% from 0.09–0.33 kg ha⁻¹ to 0.69–0.93 kg ha⁻¹. Under the same N application rate, cumulative CH₄ emissions in the RC system exhibited average reductions of 3.7–5.8% compared to the RM system over two years. Meanwhile, cumulative N₂O emissions of the RC system increased by 1.5–2.1%. RC210 reduced cumulative CH₄ and N₂O emissions by an average of 13.3% and 12.4%, respectively, compared to the RM270 over two years.

GWP and GHGI were significantly influenced by the N application rate and year, with GHGI showing significantly reduced in the RC system (Table 3). As the N application rate increased from 0 to 270 kg ha⁻¹, the GWP of the RC system increased linearly from 3397–5098 kg CO₂-eq ha⁻¹ to 7678–9449 kg CO₂-eq ha⁻¹, while that of the RM from 3152–5621 kg CO₂-eq ha⁻¹ to 8159–9724 kg CO₂-eq ha⁻¹. Similarly, GHGI increased linearly from 0.69–1.10 kg CO₂-eq kg⁻¹ to 0.91–1.26 kg CO₂-eq kg⁻¹ for the RC system and from 0.62–1.31 kg CO₂-eq kg⁻¹ to 1.02–1.37 kg CO₂-eq kg⁻¹ for the RM system. At the same N application rate, GWP and GHGI in the RC system exhibited reductions of 3.6–5.7% and 5.5–11.1%, respectively, compared to the RM system over two years. Compared to RM270, RC210 reduced GWP and GHGI by an average of 13.3% and 13.7% over two years. Overall, N application significantly increased the emission fluxes and cumulative emissions of CH₄ and N₂O. The RC system significantly reduced GHGI compared to the RM system, and RC210 has a greater GHG reduction potential than RM270.

3.4. NEB and NEEB

NEB and NEEB were significantly influenced by the N application and year, with both being significantly higher in the RC system (Figure 7). N application significantly affected the NEB and NEEB in the RC system, both of which exhibited an increasing and then decreasing trend. In 2020 and 2021, as the N application rate increased from 0 to 210 kg ha⁻¹, the NEB of the RC system increased from −4616.7 and −7014.0 CNY ha⁻¹ to 9095.8 and 9725.4 CNY ha⁻¹, reaching the maximum. However, the NEB in the RC system decreased by 14.4% and 10.0% as the N application rate increased from 210 to 270 kg ha⁻¹. Similarly, as the N application rate increased from 0 to 210 kg ha⁻¹, NEEB of the RC system increased from −5505.3 and −7606.1 CNY ha⁻¹ to 7639.9 and 9555.0 CNY ha⁻¹, then decreased by 19.7% and 12.2% as the N application rate increased to 270 kg ha⁻¹. As the N application rate increased from 0 to 270 kg ha⁻¹, the NEB in the RM system increased linearly from −4220.6 and −1877.6 CNY ha⁻¹ to 867.6 and 3508.3 CNY ha⁻¹, and NEEB in the RM system increased linearly from −5200.3 and −2427.0 CNY ha⁻¹ to −827.2 and 2086.3 CNY ha⁻¹ (Figure 7a,b). The maximum NEB and NEEB in the RC system (RC210) were, on average, 354.7% and 1265.7% higher than those in the RM system (RM270), respectively, over two years. Therefore, considering both economic and environmental benefits, the optimal N application rates for the RC and RM systems were 210 kg ha⁻¹ and 270 kg ha⁻¹, respectively, with both NEB and NEEB being significantly higher in the RC system under the optimal N application rate.

4. Discussion

4.1. Optimizing N Application Potential for Improving Rice and Crab Production

Due to the sensitivity of rice and crab yields to N application rates, it is critical to analyze the optimization of N application to guide appropriate N management practices in the RC system [22]. This study demonstrated that the N application rate of 270 kg ha⁻¹ maximized rice yield in the RM system (Figure 3a), consistent with results from the developer of the rice variety Yanfeng 47 [33]. Additionally, the rice yields in the RC system also reached a maximum of 270 kg ha⁻¹, suggesting that rice yields in the RC system responded to the N application in the same way as in the RM system. In the RC system, unutilized feed and crab feces provided supplementary N for rice growth, extending the period of N availability [27,38] and thereby improving GNU, NRE, and NPFP by 4.9%, 10.8%, and 4.8%, respectively, compared to the RM system (Table 2). Furthermore, crabs limit the growth of weeds and ineffective rice tillers by feeding on them [27,39], thereby reducing N fertilizer waste and improving agronomic traits by 1.9–8.1% (Table 1). The increase in rice yield was attributed to enhanced agronomic traits and N use efficiency [40,41], which were enhanced in the RC system (Table 1), leading to higher rice yields in the RC system compared to the RM system (Figure 3a). These improvements are consistent with previous studies [23,25]. Based on earlier research, this study further demonstrates that the agronomic traits and N use efficiency of RC210 were superior to those of RM270 over two years despite the N application rate being 60 kg ha⁻¹ lower (Table 1), resulting in a higher rice yield at 7603.1 kg ha⁻¹ (Figure 3a). Therefore, a judicious reduction in the N application rate in the RC system can yield higher rice production than the maximum yield of the RM system due to improvements in agronomic traits and N use efficiency.

Crabs play a decisive role in enhancing the economic efficiency of the RC system [42]. The application of N enriches the natural food available to crabs in RC systems, especially macrophytes, phytoplankton, zooplankton, and zoobenthos [28,43]. Increased N application rates boost the nutrient supply for crabs, leading to higher crab yields and average weights (Figure 3b,c). Consequently, crab yields and average weights were significantly higher at N application rates of 210 and 270 kg ha⁻¹ compared to 0 and 150 kg ha⁻¹ (Figure 3b,c), indicating that N application is a crucial factor in crab production. Although few studies have investigated the effects of N application on crab production in RC systems, the observed positive impact aligns with the benefits of N application on aquatic production in pond aquaculture systems [44,45].

Unlike the response of rice yield to N application, the increase in N application rate did not linearly increase crab yield and average weight in our research (Figure 3b,c). Since crabs in the RC system inhabit surface water, their respiration consumes oxygen [46], leading to a significant decrease in DO concentration (Figure 5a), which threatens their survival. N is essential for enhancing the photosynthesis of aquatic plants and increasing DO concentrations in surface water [47,48,49]. Consequently, increased N application raises the NH₄⁺-N and NO₃⁻-N concentrations in surface water (Figure 5b,c), facilitating algae propagation and increasing DO concentrations (Figure 5a). While these effects led to a linear increase in DO concentrations with increasing N application rates from 0 to 270 kg ha⁻¹ in the RM system, the N application rate of 210 kg ha⁻¹ resulted in the highest DO concentration in surface water in the RC system (Figure 5a). The probable reason is that excessive N application at 270 kg ha⁻¹ resulted in high concentrations of NH₄⁺-N and NO₃⁻-N in the surface water (Figure 5b,c), which were toxic to the crabs and led to their mortality. The decomposing crab carcasses consumed oxygen in the surface water, thereby decreasing DO concentration. Additionally, larger crabs have a greater need for oxygen, which resulted in the RC210 having the largest average weight. Therefore, an N application rate at 210 kg ha⁻¹ is a viable strategy to ensure optimal crab production. Meanwhile, the surface water quality at this N application rate was most suitable for crabs, resulting in maximum crab yield and average weight. The N application rate of 210 kg ha⁻¹ is the optimal N application strategy for the RC systems.

4.2. Effects of the Optimizing N Application on GHG Emissions in RC System

During the early stages of the planting season, the application of N resulted in elevated concentrations of N in surface water and soil (Figures S1e,f and S2e,f), along with high levels of DO (Figures S1e and S2e), which enhanced nitrification and denitrification [50,51], leading to a surge in N₂O emissions (Figure 6b). In contrast, N₂O emission fluxes after the tillering stage were suppressed due to the intensification of anaerobic conditions in the soil as a result of persistent flooding, as well as the lack of N [52]. Anaerobic conditions favored CH₄ production [53,54], and the increase in carbon substrates from root secretions and decaying leaves of growing rice [55], along with the formation of advanced aerenchyma [56], led to a gradual increase in CH₄ emissions after the tillering stage (Figure 6a). CH₄ emission fluxes gradually decreased in the later stages of rice growth as temperature decreased and the ventilation tissues aging and breaking [57]. Similar results were observed in other studies [30,58]. Our research further demonstrated that RC systems can reduce GHG emissions by optimizing N application. This is likely due to the precise application of N fertilizer, which enhances N use efficiency of rice while reducing the carbon substrates available for CH₄ production [59]. The availability of substrates for nitrification and denitrification was decreased [60]. Consequently, optimized N application reduced both CH₄ and N₂O emissions in rice paddies [32,61]. Furthermore, our findings indicate that crabs reduce CH₄ emissions from RC systems through physical disturbance, consistent with the results of Khoshnevisan et al. [22]. However, the reduction potential is limited at each N application rate [30]. The limitation may be due to the low population density of crabs and their lack of burrow construction in paddy fields, resulting in minimal disturbance [24,31]. Additionally, the RC system significantly reduced DO in surface water compared to the RM system (Figure 5a), diminishing its potential to reduce CH₄ emissions. Another notable finding was that the RC system did not significantly reduce N₂O emissions; rather, it slightly increased them (Table 3), contrary to previous studies [22,30]. This discrepancy may be due to the timing of GHG monitoring, which in this study, began with the application of basal fertilizer, whereas previous studies typically started at transplanting, thus overlooking N₂O emissions during the initial period. The presence of N in unutilized feed and crab feces increased the N substrate [39], which in turn slightly elevated N₂O emissions during this period. Although the RC system did not significantly affect GHG emissions (Table 3), the N application significantly stimulated GHG emissions. Reducing N application to 210 kg ha⁻¹ in the RC system could significantly reduce GHG emissions, resulting in lower GWP and GHGI. Thus, in RC systems, optimized N application at 210 kg ha⁻¹ enhances productivity while more effectively reducing GHG emissions, thereby further decreasing GWP and GHGI.

4.3. Synergize Economic and Ecological Benefits with Optimizing N Application

Taking into account both economic and ecological benefits, the maximum NEEB in RC and RM systems was achieved at N application rates of 210 kg ha⁻¹ and 270 kg ha⁻¹, respectively (Figure 7a). Despite the reduced N application rate, the NEEB of RC210 exhibited an average increase of 1265.7% over a two-year period compared to RM270, attributable to enhanced productivity and reduced GHG emissions. The optimized N application rate resulted in a more significant NEEB enhancement in the RC system compared to previous studies [30]. Although the differences in rice yields at the optimal N application rate were minimal, the crab yield led to a notable 354.7% increase in NEB in RC210 compared to RM270. To ensure food security, the government typically maintains low rice prices through monopoly purchases. In contrast, aquatic products such as crabs are sold in the free market and command higher economic value, aligning with the findings of previous studies [9,62]. Concurrently, the GWP cost of the RC system is reduced due to the lower GHG emissions resulting from decreased N application, thereby maximizing NEEB. Since farmers are primarily interested in NEB, which serves as the intrinsic motivation for the transition from the traditional RM system to the RC system, 210 kg ha⁻¹ represents the optimal N fertilizer application rate that balances economic and environmental benefits, maximizing benefits for farmers’ incomes [30,62]. Based on the technology and experience of existing RC systems, it can be reasonably concluded that the RC system is well-suited for adoption by small farmers [16,22]. This adoption increases farmers’ incomes, thereby supporting China’s poverty alleviation goals. Therefore, it is recommended that farmers be encouraged to adopt the RC system and utilize NEB to guide the optimization of N application, thereby achieving a synergistic increase in economic and ecological benefits.

This study elucidated that optimizing the N application rate improves the productivity, sustainability, and economic efficiency of the RC systems. However, N application may affect the metabolism of nitrogen in both rice and crabs, thus altering the quality and taste; therefore, further research is needed [63,64]. It is worth noting that the RC system used in this study is the most widely accepted model. Innovative RC models, such as the addition of azolla and the construction of aeration devices, could also have the potential to increase crab yield and reduce CH₄ emissions by enhancing DO concentrations within the paddy field [65,66]. Compared to other optimization strategies, adjusting the N application rate is the most feasible and easy-to-implement improvement for farmers, as the main constraints to the development of RC systems are the aging labor force and the technical capacity of available technology [22]. Combining optimized N application with other innovative RC models may yield even greater potential, and its productivity, sustainability, and economic efficiency should be further investigated.

5. Conclusions

This study compared productivity, sustainability, and economic benefits between the RC and RM systems at different N application rates. Agronomic traits, N use efficiency, and rice yield were improved in the RC system compared to the RM system, with the rice yield of RC210 still 0.4% higher than RM270 despite a lower N application rate. Additionally, RC210 maximizes crab yields (370.1 kg ha⁻¹) and average weights (81.1 g) by optimizing surface water quality, particularly by maximizing the DO concentration to 6.6–6.9 mg L⁻¹. At the same N application rate, the GHGI was reduced by 5.5–11.1% in the RC system compared to the RM system and further reduced by 13.7% in RC210 compared to RM270. The maximum NEEB in the RC system reached 8597.5 CNY ha⁻¹ at an N application rate of 210 kg ha⁻¹, which was 1265.7% higher than that of the RM system at 270 kg ha⁻¹. In conclusion, the optimized N application rate of 210 kg ha⁻¹ in the RC system was lower than the 270 kg ha⁻¹ required in the RM system, while simultaneously enhancing both economic and ecological benefits.