Introduction

Methane (CH4) and nitrous oxide (N2O) are prominent greenhouse gases (GHGs) that exhibit substantial global warming potentials (GWPs) of 28 and 265 times, respectively, compared with the GWPs of CO2 on a mass basis over a 100-year timespan1. Together, CH4 and N2O contribute approximately 20% of the overall radiative forcing of the atmosphere stemming from anthropogenic activities2. Agricultural wetlands, including aquaculture ponds, rice paddies, and rice–fish fields, are the most important emission sources of CH4 and N2O3,4. In these agricultural lands, the periodic patterns of waterlogging and drainage alteration during farming can create a conducive environmental condition for CH4 and N2O production. The abundant inputs of organic and inorganic fertilizers or fish baits in agricultural wetlands serve as rich substrates, which in turn significantly enhance the production of CH4 and N2O, respectively5,6. Notably, while agricultural wetlands are substantial contributors to atmospheric CH4 and N2O emissions, agricultural land-use changes can effectively mitigate net CO2-equivalent emissions by strategically enhancing carbon sinks and reducing GHGs fluxes through practices like altered tillage, water management, and fertilizer application7,8.

Currently, the Chinese government has initiated a strategic program aimed at converting inland rural aquaculture ponds into cultivated farmland to ensure high grain yields, resulting in the transformation of nearly 90,000 hectares of former rural aquaculture ponds into reclaimed paddy fields, including 12,000 hectares of former shrimp ponds9. Nevertheless, the economic returns for rice cultivation are generally lower compared with pond aquaculture, and many of these reclaimed lands have therefore been abandoned after brief cultivation periods. To address this issue, the reclaimed rice paddies have been further adapted into rice–shrimp fields, aiming to achieve the government objective of “steady success in grain and fish production”. This shift from SP to RM and RS, which has become one of the most prevalent forms of agricultural reclamation since 2018, exerts profound and intricate influences on nutrient cycling in agricultural wetlands owing to substantial alterations in hydrological conditions and management strategies10,11,12.

Agricultural production in China contributes approximately 40% and 60% of the national overall CH4 and N2O fluxes in China, respectively13,14. Rice fields in particular represent significant contributors to atmospheric CH4 and N2O emissions, accounting for roughly 25% of China’s total agricultural emissions15. Extensive field measurements have been conducted to investigate the emission patterns of CH4 and N2O from rice paddies, yielding comprehensive estimates of 6 to 10 Tg/yr for CH4 emissions and 32 to 51 Gg/yr for N2O emissions from Chinese rice fields9. Rice paddies, as a typical type of agricultural land, have been extensively documented as a prominent emissions source of atmosphere CH4 and N2O, which is primarily attributed to water management practices, along with the application of both organic and inorganic fertilizers16,17.

As the world’s largest aquaculture producer, accounting for roughly 60% of the global total, the field records pertaining to GHG emissions from rural aquacultural ponds remain comparatively limited in comparison with those from rice paddies18,19. Likewise, substantial proportions of the feed carbon and nitrogen were transformed into CO2, CH4, and N2O by aquatic organisms and microbes or sequestered within pond farming systems20,21. These extensive inputs of carbon (C) and nitrogen (N) have the potential to drive aquaculture ponds to become significant anthropogenic sources of CH4 and N2O emissions22,23. To date, some reports have evaluated GHG fluxes from these ponds, particularly at the Min River estuary in southeast China, where research has been conducted to gain a deeper understanding of these emissions21,24,25.

The rice–shrimp coculture approach has been widely adopted in several Asian countries, including China, Vietnam, and Indonesia due to its economic advantages26,27,28. Nevertheless, the GHG emissions stemming from this type of farming practice have the potential to exert adverse environmental impacts, and the associated greenhouse effects over time have drawn significant attention on a global scale29. Our previous research suggests that the transition from traditional paddy fields to rice–Macrobrachium rosenbergii coculture systems can lead to substantially elevated N2O emissions, with an increase of up to 41.7% compared with normal fertilizer application practices in rice monoculture30. Previous reports have highlighted that alterations in land use can introduce substantial uncertainties when assessing the contribution of agricultural wetlands to atmospheric GHG budgets9,31,32. Consequently, comprehensive field studies that encompass spatiotemporal dimensions are critical for gaining a deeper understanding of the variability in GHG emissions from rice–shrimp coculture ecosystems and for improving its large-scale assessments.

There is an ongoing and expanding trend of converting earthen shrimp ponds into rice paddies, which are further converted into rice–shrimp coculture fields in southeast China. CH4 and N2O emissions are intricately linked to agricultural practices and vary under different agricultural production ecosystems20,33. Previous studies have indicated that the spatial and temporal variation of GHG fluxes may bring additional uncertainties when evaluating the contribution of land use change to the total GHG emissions16,34. More importantly, most existing research on GHG fluxes in aquaculture ponds, rice paddies, and rice–shrimp fields only focused on an individual study site, neglecting simultaneous cross-comparisons among these areas using a unified method26,35. Therefore, there is an urgent and evident need to obtain a greater understanding of the costs associated with GHG emissions during the ongoing shift of shrimp ponds into rice paddies and integrated rice–shrimp fields.

To investigate the effects of the transition of shrimp ponds (SP) to rice monoculture (RM) and rice–shrimp coculture (RS) fields on GHG emissions, this two-year study measured the year-round fluxes of CH4 and N2O from reclaimed RM and adjacent RS fields converted from SP three years ago, as well as the neighboring SP, in the Taihu basin in Huzhou, China. The specific objectives of this study were to (1) obtain insights into the heterogeneity of CH4 and N2O emissions from SP, reclaimed RM, and RS fields; (2) elucidate the primary environmental factors regulating the emissions of CH4 and N2O from those three ecosystems; and (3) compare the emission factors, GWP, and environmental implications following the reclamation of SP to RM and RS fields.

Materials and methods

Site description

The experimental site was conducted in a single-crop district within the Huzhou district (30°43’41"N, 120°10’10"E), a representative region of southeast China, where SP were converted into paddy fields by removing the pond ridge between 2017 and 2018 (Fig. 1). Subsequently, in 2018–2019, some reclaimed RM area was further transformed into RS fields. Field experiments were conducted simultaneously in conventional SP, reclaimed RM, and RS fields, all of which were located at the Huzhou Comprehensive Test Station of the National Shrimp and Crab Industry, China. These experiments were performed over the rice or shrimp farming cycle from April 2020 to April 2022. The experimental region is characterized by a typical north subtropical monsoonal climate, featuring an annual mean temperature of 16.2 °C and a total precipitation of 1348 mm across the experimental cycle. Detailed physicochemical properties of the soil and sediment prior to the initiation of experiments are provided in the supplementary files (Table S1).

Fig. 1
Fig. 1
Full size image

Location of the study area and sampling sites for the SP, RM and RS fields in the Taihu Lake watershed, Southeast China. This map was created from the standard base map (Map Review No: GS(2019)1686) of the National Administration of Surveying, Mapping and Geographic Information (http://bzdt.ch.mnr.gov.cn/) using ArcGIS Pro 3.5 (https://www.esri.com/).

Experimental design and field management

Field experiments in SP

Adjacent to the investigated reclaimed RM and RS farming systems, a local conventional shrimp aquaculture system was established as a reference, comprising three independent replicated ponds. The dimensions of these ponds were as follows: pond 1 (length × width × depth: 110 m × 62 m × 1.8 m), pond 2 (length × width × depth: 122 m × 65 m × 1.8 m), and pond 3 (length ×width× depth: 115 m×64 m×1.8 m).

Field experiments in RM

A parallel field experiment on local RM was conducted from April through October in 2020 and 2021. Three randomly selected rice monoculture plots (plot 1: length 132 m×width 68 m, plot 2: length 142 m×width 69 m, plot 3: length 138 m×width 68 m) were set up as experimental replicates.

Field experiments in RS

A parallel field experiment utilizing a newly constructed RS system was also conducted to evaluate the GHG fluxes compared with the SP and RM systems in the experimental site. Three randomly reclaimed paddy fields (plot 1: length 145 m × width 71 m, plot 2: length 149 m × width 73 m, plot 3: length 144 m × width 65 m) were selected to conduct the integrated RS experiments from April to October in both 2020 and 2021.

A comprehensive overview of the key management practices employed for SP, RM, and RS coculture fields is presented in the supplementary files (S2.2 and Table S2).

Sampling and measurement of environmental variables

Measurement of meteorological factors

During each sampling event, the air temperature and wind speed in each test plot were measured using a portable weather meter (Kestrel-3500, USA).

Water sampling and measurement

During the farming period, the water depth was maintained at approximately 1.5 ± 0.1 m in the aquaculture ponds, 0.3 ± 0.06 m in the rice paddies, and 0.5 ± 0.08 m in the rice-crayfish co-culture fields. In each plot, five samples were collected once every two weeks from a depth of 20 centimeters beneath the water surface during the farming periods. The five samples were then mixed to create a composite sample for each plot, which was preserved in a plastic bottle maintained at 4 °C. The composite samples were transported to the laboratory for analysis within 24 h.

Soil/sediment sampling and measurement

On the same day as water sampling, soil or sediment samples were collected from a depth of up to 20 cm using a five-point sampling method within each plot. These five samples were combined to create a composite sample for each plot for phychemical property and functional gene analysis.

Measurement of CH4 and N2O fluxes

At the same day as water sampling, CH4 and N2O fluxes were measured using the static chamber method in the three agroecosystems30.

A brief description of measurement of water and soil/sediment samples as well as CH4 and N2O fluxes highlighted in the supplementary files (S2.3, Table S3 and Table S4).

Data analysis

The statistical analyses were conducted utilizing the Statistical Package for the Social Sciences software, specifically version 19.0 (SPSS v19.0). The normality of the research data was evaluated based on the Shapiro–Wilk test. Depending on the outcome of the normality test, further comparisons were conducted using either Tukey’s parametric test or Dunn’s non-parametric test. Specifically, Tukey’s test was employed when the data adhered to the normality assumption, whereas Dunn’s test was utilized when normality could not be presumed. Differences among the three treatments were considered significant at P < 0.05.

To investigate the hypothetical pathways of the effects of biotic and abiotic factors on CH4 or N2O emissions, structural equation modeling (SEM) was conducted in SPSS Amos 22.0. Models were considered to have an acceptable fit when the following criteria were met: the chi-square value normalized by degrees of freedom (χ2/df) remained below 3, the goodness-of-fit index (GFI) exceeded 0.9, and the root mean square error approximation (RMSEA) was less than 0.0836.

Results

Water/sediment physicochemical and biological properties

During the experimental periods, no significant differences were observed for W.T, pH, TP, and NO3−-N of the overlying water among the three ecosystems (P > 0.05) (Table 1). Yet, obvious differences were found for other water parameters (DO, Eh, NH4+–N, TN, and TOC) (P < 0.05). SP had the highest concentrations of TN, NH4+–N, and TOC, followed by RM and RS fields. After the field experiments, the sediment NH4+–N, NO3−–N, TN, TP, and TOC contents in SP were obviously higher than those in both RM and RS fields (P < 0.05). Conversely, the production patterns in RS fields led to improvements in the physicochemical properties of the reclaimed fields compared with RM. Specifically, soil collected from RS fields exhibited higher values of soil porosity, CEC, NH4+–N, NO3−–N, TN, TP, and TOC.

Table 1 Physicochemical characteristics of meteorology, overlying water and sediment from SP, RM and RS fields. (n = 4)
Full size table

There were no significant differences in the copy number of the AOB amoA gene among these ecosystems (P > 0.05) (Fig. 2). However, the abundance of the AOA amoA gene of the pond sediment was significantly higher than that of the RM and RS fields (P < 0.05). Similarly, the pond sediment exhibited significantly higher copy numbers of denitrification genes, including narG, nirK, nirS, and nosZ, relative to RM and RS fields (P < 0.05). The copy numbers of mcrA in RS coculture fields (18.26 ± 2.16)×108 were the highest, followed by RM (16.85 ± 1.85)×108 and SP (13.25 ± 1.38)×108 during the farming periods. Although no significant difference was found in the copy number of mcrA among the three systems, there were significant differences in the pmoA gene in the order of SP > RS > RM (P < 0.05).

Fig. 2
Fig. 2
Full size image

Boxplots of functional gene copies in the SP (a, b), RM (c, d) and RS (e, f) in the farming and non-farming periods, respectively.

N2O and CH4 emission fluxes

The seasonal dynamics of N2O fluxes were not obviously dependent on the air temperature in the SP, RM, and RS during the experimental periods (Fig. 3). The N2O flux from the RM (7.39 ± 0.73 kg/ha) was significantly larger than that from the SP (4.21 ± 0.5 kg/ha) and RS (4.21 ± 0.53 kg/ha) (P < 0.01) (Fig. S1). Over the two-year study period, the average ratios of N2O fluxes during the farming periods were 53.5%, 64.1%, and 48.0% of the annual fluxes from the SP, RM, and RS, respectively (Fig. S1).

Unlike N2O, the seasonal pattern of CH4 fluxes was overwhelmingly dependent on the air temperature in the SP, RM, and RS during the study periods (Fig. 3). The annual CH4 flux was the largest in the RM (490 ± 46 kg/ha), followed by the RS (189 ± 18 kg/ha) and SP (103 ± 10 kg/ha). Grouping the data revealed a seasonal pattern, with lower emission fluxes in autumn and winter, whereas higher fluxes were detected in spring and summer for the three culture modes. Notably, the majority of CH4 emissions occurred during the rice or shrimp growing periods (Apirl-October), during which the emissions were significantly higher compared with non-farming periods (November-next March)(P < 0.01). The mean ratios of CH4 fluxes during the farming period reached 75.99%, 72.81%, and 74.99% of the annual fluxes from the SP, RM, and RS, respectively (Fig. S1).

Methods S2.2 and Table S2 detailed the timing of fertilizer application and water replacement, along with their corresponding values. In general, peak N2O fluxes mainly occurred after the application of “rich water” fertilizer in the SP and RS fields, and following the application of organic basal fertilizer or inorganic topdressing fertilizer in the RM and RS fields, after which the N2O fluxes decreased rapidly until the next application of fertilizer (Fig. 3). Replacing surface water of a depth of 20 cm also caused the short obvious pulse of N2O flux in SP. In the rice paddy, drainage for drying the field slightly stimulated N2O emissions. By contrast, the aquaculture practice of applying quicklime in both SP and RS fields restrained N2O emissions to a great extent. Similarly, applying quicklime reduced CH4 emissions, and in contrast to N2O, replacing the surface water quickly increased the CH4 fluxes in both the SP and RS fields, and the CH4 fluxes decreased rapidly until the next event occurred. In addition, drainage for drying the field and drainage for rice harvest clearly enhanced the peaks of N2O fluxes. Conversely, irrigation during the rice-growing period led to decreased N2O emissions. Thereafter, with the rising air temperature and application of feed input among the three systems, the CH4 fluxes remained relatively high during the shrimp- or rice-growing periods. Following the harvest of rice or shrimp, CH4 fluxes gradually declined due to the decrease in air temperature and subsequently remained at a low level (Fig. 3).

Fig. 3
Fig. 3
Full size image

Annual variation of CH4 and N2O fluxes with water temperature in the SP, RM and RS fields. Vertical bars are the standard errors of the means (n = 3). The gray sections represent the farming periods (Apirl-October).

Relationships between environmental parameters and N2O/CH4 fluxes

Over the annual waterlogged farming cycle, the CH4 fluxes were positively associated with the air/water temperature, water depth, Eh, and TOC, as well as the sediment Eh and TOC (P < 0.05), while the CH4 fluxes were negatively affected by the water DO (P < 0.05), water NO3−–N, and sediment NO3−–N (Fig. 4). Over the annual non-farming period, CH4 fluxes were only significantly and positively correlated with the air temperature and sediment Eh. The CH4 fluxes from SP showed a more robust response to water depth, water NH4+–N, sediment Eh, and sediment TOC, while the CH4 fluxes from the RS were more dependent on the air temperature and sediment TOC.

The environmental factors that significantly impacted N2O fluxes differed from those that significantly affected CH4 fluxes. N2O fluxes in SP and RS fields were positively correlated with NH4+–N, NO3−–N, TN, and TOC (P < 0.05) over the annual waterlogged farming cycle. Similarly, positive correlations between N2O fluxes and sediment Eh, NH4+–N, NO3––N, and TN were observed in all three agroecosystems. In addition, the positive response of N2O fluxes to sediment N was greater than their response to water N (Fig. 4). During the non-farming period in RM and RS, the N2O fluxes were significantly correlated with the air temperature (P < 0.05).

To identify the biotic factors controlling N2O/CH4 production in sediment in the three different wetlands, correlation analyses were conducted among functional gene abundances and N2O/CH4 fluxes (Fig. 4). Based on Pearson correlations, the changes in CH4 emissions in the farming period were strongly driven by mcrA gene abundance and strongly restrained by pmoA gene abundance among the SP, RM, and RS systems. N2O fluxes were positively associated with AOA amoA, AOB amoA, narG, nirK, and nirS gene abundance only during the farming periods (P < 0.05), while they were negatively affected by nosZ gene abundance in both the farming and non-farming periods (P < 0.05).

The ratios of CH4/N2O-producing genes to CH4/N2O-oxidizing genes, that is, mcrA/pmoA and (nirK + nirS)/nosZ, respectively, were used to represent the CH4/N2O production potential. In the current study, N2O fluxes were significantly correlated with the ratios of (nirK + nirS)/nosZ (P < 0.05) during the farming periods, while there was no significance during the non-farming periods (P > 0.05). In SP, CH4 fluxes were significantly related to mcrA/pmoA in both the farming and non-farming periods (P < 0.05), yet the correlation relationship in RM and RS was only significant during the farming periods (P < 0.05).

Fig. 4
Fig. 4
Full size image

Person correlation coefficients between CH4/N2O fluxes, and various environmental variables in the SP, RM and RS fields.

Relative importance of environmental factors in N2O and CH4 emissions

This study further determined the relative contributions of different factor types to N2O (a) and CH4 (b) emission fluxes due to land use change (Fig. 5). SEM of the CH4 emissions revealed that land use change positively affected the water DO, water depth, water TOC, and sediment TOC. The sediment TOC had a positive impact on the abundances of mcrA, pmoA, and CH4 emissions. The water/sediment Eh had a positive influence on CH4 emissions. The mcrA gene was positively correlated with CH4 emissions, while the pmoA gene was negatively correlated with CH4 emissions. The water DO had a significantly negative effect on water Eh and CH4 emissions.

SEM of the N2O emissions revealed that land use change positively affected the water DO, sediment TN, sediment TOC, and sediment Eh. The sediment Eh exhibited a positive correlation with nosZ and nirK. The sediment TOC contents were positively correlated with nirG, AOB amoA, and AOA amoA. Water NH4+–N was positively correlated with the AOA amoA. Conversely, water Eh demonstrated a negative correlation with N2O emissions. N2O emissions were driven by denitrification (nirS, nirK, and nrfA) and nitrification genes (AOA amoA and AOB amoA), as well as the sediment TN.

Fig. 5
Fig. 5
Full size image

The effects of different land-use practices on CH4 and N2O emissions. Panels (a) and (b) represent the influence on CH4 and N2O emissions, respectively. Solid lines indicate significant effects, while dashed lines represent non-significant relationships. The width of the arrow is proportional to the strength of path coefficients. Numbers adjacent to the arrows are path coefficients, indicative of the effect size of the relationship. Blue arrows signify positive relationships, whereas red arrows indicate negative relationships. Significance levels are as follows: *p-value < 0.05, and **p-value < 0.01. The low chi-square normalized by degrees of freedom (χ2/df < 3), high goodness-of-fit index (GFI > 0.9), and low root mean square error approximation (RMSEA < 0.08) ) indicate the model with a good fit.

Emission factors and GWP

As shown in Table 2, the mean scaled GWP (SGWP) and net ecosystem economic profit (NEEP)-scaled GWP (NEEP-scaled SGWP) of the RM were obviously higher than those of the RS and SP (P < 0.05). Combining the annual emissions from the three wetlands, this study estimated that the emission factors for CH4 (EF−C) and emission factors for N2O (EF−N) were in the order of RM > RS > SP, and the difference among them was significant (P < 0.05).

Table 2 Annual average total CH4 and N2O emissions, SGWP, NEEP-sacled SGWP, and the emission factors (EF) of CH4 and N2O from the SP, RM and RS fields.
Full size table

Discussion

Effects of land use change on water/sediment properties

The conversion of aquaculture ponds to RM and RS fields substantially alters the physicochemical characteristics of water and sediment, with a marked influence on the availability of C and N substrates that results in a series of changes in nutrient cycling processes38. Inefficient feed utilization by aquatic organisms has led to widespread eutrophication within intensive aquaculture systems and adjacent river nets30. Only 20.4% of feed-N and 22.8% of feed-P were assimilated into shrimp biomass, leaving considerable amounts of C, N, and P resident in the pond aquaculture system30. Furthermore, the field survey revealed that shrimp monoculture systems exhibited the highest C, N, and P contents in both sediment and water, followed by RM and RS systems (Table 1), despite the periodic removal of pond sludge every two to three years. The present investigation demonstrated that shrimp aquaculture significantly increased the sediment TOC, NH4+–N, NO3−–N, and NO2−–N contents due to the decomposition of substantial residual feeds and excrement25.

In rice–aquaculture systems, N and P from unconsumed feeds and excrement can be assimilated and utilized by rice plants39. In addition, shrimp excretions, which include fecal matter and excess N excreted as ammonia and urea, can be directly utilized through rice roots35,40. This utilization of C, N, and P by rice plants potentially mitigates the accumulation of these nutrients in the sediment environment. However, the field experimental data obtained in the present study revealed that the C, N, and P contents in the field water and soil of integrated RS systems were significantly elevated compared with those in RM systems (Table 1). This phenomenon may be attributed to the substantial quantities of feed administered in shrimp monoculture systems. Although the precise quantification of the N and P assimilated by rice plants in RS system remains challenging, the present field experiment suggests that rice plants do indeed absorb considerable amounts of N and P within RS systems, aligning with findings reported by Burford et al. (2020)41.

The functional genes involved in nitrification, such as AOA amoA and AOB amoA, and those involved in denitrification, including narG, nirK, nirS, and nosZ, are crucial for catalyzing the reduction of nitrate and nitrite to N2O and NO, respectively42. These genes play pivotal roles in regulating the rates of nitrification and denitrification43. The elevated number of gene copies, particularly of nosZ, in SP can be attributed to the increased availability of sediment C and N substrates, indicating that denitrification occurs robustly and efficiently, potentially resulting in N primarily being lost as N2. Furthermore, the higher gene abundances observed during the farming periods compared with the non-farming periods suggest that agricultural practices may not be the most dominant factors influencing sediment properties (such as the soil water content, temperature, and nutrient state). These findings align with the majority of previous studies44. As for CH4, the significantly higher abundance of the pmoA gene in the surface sediment of SP compared with RM and RS fields suggests that the CH4 produced may be promptly oxidized to CO2, ultimately resulting in lower CH4 emissions9,45. However, gene abundance at the DNA level does not fully capture microbial activities46. Therefore, further experimental investigations in future studies are essential to clarify the role of microorganisms in the reduction of sediment organic matter in response to land use changes.

Effects of land use change on N2O and CH4 emissions

In this present study, annual CH4 and N2O fluxes from inland shrimp ponds were comparable to those reported in other managed aquaculture earthen ponds using floating chambers20. Previous researches have extensively focused on investigating CH4 and N2O emissions from various types of aquacultural ponds, including fish ponds16, shrimp ponds25, and crab ponds6. The conversion of aquacultural ponds to rice paddies led to significant increases in annual CH4 and N2O emissions by 375.8% and 75.2%, respectively. However, CH4 and N2O emissions from the reclaimed rice paddies are relatively lower than those reported in most previous studies on rice paddies (Table S5). Several factors may explain the higher CH4 and N2O emissions observed following the reclamation of SP to RS. First, although the soil/sediment microbial substrate (TOC) for methanogens and denitrifiers was relatively higher in SP than in reclaimed RM (Table 1), the emissions of CH4 and N2O from SP were not higher than those from the RM. This can be attributed to the fact that CH4 and N2O emissions are predominantly inversely correlated with the water DO concentration (Fig. 4), resulting in lower emissions from aquaculture ponds with higher DO concentrations (> 5 mg/L) than from rice paddies. Second, the fate or release rate of CH4 and N2O from freshwater ponds is largely determined by the transport efficiency from the sediment–water interface to the atmosphere, with the fluxes exhibiting a negative relationship with water depth47. Third, the alternating patterns of dry and wet conditions coupled with fertilizer application in RM can substantially enhance the generation of CH4 and N2O from paddy fields3,13. As shown in Fig. 2, the abundances of narG, nirK, nirS, and nosZ were all higher in RM than in RS in the farming periods. However, the rate of N₂O production (catalyzed by nirK/nirS) exceeded the rate of N₂O consumption (catalyzed by nosZ), leading to substantial N₂O accumulation and emissions. Meanwhile, the abundance of the methanogenic mcrA was higher in RM than in RS, while the abundance of the methanotrophic pmoA was lower in RS. This combination led to an increase in CH4 emissions.

Meanwhile, the further conversion of RM paddies to RS fields obviously reduced the annual N2O emissions from 7.39 ± 0.73 to 4.32 ± 0.42 kg/ha, which was only slightly greater than the annual N2O emissions of SP (4.22 ± 0.53 kg/ha). Similarly, this further conversion reduced the CH4 emissions from 490.97 ± 45.5 kg/ha in RM to 189 ± 18 kg/ha in RS fields, although this amount remained higher than the 103 ± 10 kg/ha emitted by SP. Several potential factors may contribute to the reduction in CH4 and N2O emissions following the conversion of RM to RS fields. First, mechanical aeration and shrimp disturbance promote the diffusion of O2 from the air into the overlaying water/sediment, leading to N2O and CH4 being further oxidized to form N2 and CO2 through complete denitrification under the relatively high O2 conditions. Second, to maintain the normal growth of M. rosenbergii, the average water depth of 50 cm in the RS fields was greater than the depth of 30 cm in RM. This deeper water inhibited the transport efficiency of CH4 and N2O into the atmosphere. Third, within the RS system, shrimp, as a type of zoobenthos, exhibit vigorous movements and consume biodetritus on the sediment surface, thereby reducing the substrates available for CH4 and N2O generation26. Moreover, the burrowing behavior of giant river prawns in rice-planted areas has the potential to disrupt the soil’s anaerobic environment. This activity facilitates the penetration of O2 into the water, which can suppress CH4 production while simultaneously enhancing CH4 oxidation, especially in rice paddies with shallow water21,30.

Dependence of CH4 and N2O emissions on environmental parameters

CH4 and N2O emissions exhibit strong correlations with water and soil/sediment parameters among various wetland systems. However, the dependence of CH4 and N2O fluxes on these physical and chemical parameters differs among the three fields (Fig. 4). In general, CH4 fluxes exhibit positive correlations with the air/water temperature and soil/sediment TOC contents. Conversely, they show a negative relationship with the water/sediment Eh and water DO48,49, which is consistent with the findings of the current study across the three wetland systems. In particular, a stronger dependence of CH4 fluxes on the soil/sediment Eh and TOC contents was observed in SP compared with RM and RS systems (Fig. 5). This was likely because the persistent anaerobic conditions resulting from water depth exceeding 1.5 m significantly enhanced the availability of TOC in sediment substrates and the metabolic activity of methanogens, which in turn promoted CH4 generation50,51. In addition, compared with RM and RS, the negative correlation between Eh and DO in the sediment microenvironment of SP played a greater role in driving CH4 release through simultaneously influencing its production and the transfer of CH4 from the surface of the sediment to the water–sediment interface, ultimately facilitating its release into the atmosphere52,53.

Notable differences were found in the relationship between N2O fluxes and water quality parameters among RM, RS, and SP systems. Specifically, in SP, the N2O fluxes during the growing period exhibited significant correlations with the water Eh, NH4+–N, NO3−–N, and TN. However, no such significance was observed in RS coculture and RM wetlands. This discrepancy is primarily attributed to the higher concentrations of these compounds in SP systems compared with RS and RM systems. According to a recent review by Murray et al. (2015)54, high NO3−–N and NO2−–N concentrations can significantly enhance N2O production, whereas abundant NH4+–N can in turn stimulate N2O generation via nitrification in either the sediment or water column (Pärn et al., 2018)55.

Overall, high soil/sediment TOC, NH4+–N, NO2−–N, and TN concentrations can supply adequate C and N for the growth and metabolism of methanogens as well as nitrifying and denitrifying bacteria, thereby enhancing the production of CH4 and N2O56. Likewise, this study also identified a significant positive linear correlation between CH4 and N2O emissions and the availability of sediment C and N substrates, particularly during the cultivation period. Thus, there exists an urgent requirement for effective management strategies to reduce the emissions of CH4 and N2O from reclaimed RM and RS farming systems, thereby mitigating their release into the atmosphere. Addressing this issue will not only benefit the environment but also provide a scientific basis for future assessments of the potential global warming contributions associated with the expansion of RS coculture systems.

Effects of land use change on the EF and GWP

The annual net GWPs stemming from CH4 and N2O emissions have undergone a marked increase subsequent to the transformation of SP into RM and RS farming systems. Furthermore, the annual NEEP for these three wetland types has been approximated by deducting the aggregate costs from the economic revenues generated3. On average, the NEEP of SP was about 29 and 2.7 times greater than that of RM and RS systems, respectively (Table 2). Compared with SP, the NEEP-scaled SGWP of the RM and RS fields showed notable increases by 230% and 12,446%, respectively. Therefore, converting inland SP to RM results in high climatic impacts but low ecosystem economic benefits. Interestingly, the further conversion of RM to RS fields can drastically reduce the GWP while increasing the economic benefits. This, in turn, substantially mitigates the environmental and profitability issues stemming from reclaimed rice fields.

The microbial conversion of anthropogenic C and N inputs into CH4 and N2O within agricultural farmlands has been recognized as a pivotal process contributing to global CH4 and N2O emissions1. The EF− C and EF− N of rice-based cropping systems have been extensively documented15,21, whereas the direct EFs of anthropogenic C and N from pond aquaculture and rice–aquaculture farmlands are poorly known, primarily owning to a lack of in-situ field measurements18,57,58. In pond aquaculture systems, CH4 and N2O are emitted as byproducts during the conversion of ammonia and urea to nitrate through the process of nitrification and subsequently from nitrate to N2 gas via denitrification6,16. In the present study, CH4 and N2O emissions from inland shrimp ponds amounted to 103 ± 10 kg CH4/ha and 4.22 ± 0.53 kg N2O/ha, respectively. These emissions were equivalent to approximately (5.65 ± 0.46)% and (0.47 ± 0.031)% of the total C and TN inputs in the feeds, respectively, expressed as EF− N. The estimated EF− C for shrimp ponds was lower than those reported for fish and crab ponds by Fang et al. (2022)59, which were 10.19% and 7.99%, respectively (Table 2). Although the estimated EF− N was slightly below the range of 0.46–1.60 previously obtained in fish ponds, crab ponds, rice–crayfish systems, and rice–fish systems (Table 2), the direct field measurements of the N2O flux exceeded the IPCC’s default emission of 1.69 N2O–N kg per unit of aquaculture production2,57. The conversion of SP to RM and further to RS obviously increased the CH4 and N2O emission, especially CH4 (Table 2). The main possible reason may be due to rice root exudates, which enrich the sediment substrates for heterotrophs that consume oxygen. This, in turn, results in the depletion of DO and augmentation of anaerobic conditions, thereby enhancing the potential for CH4 production, especially in the absence of impeller aerators in SP and RS systems.

The EF− N induced by fertilizer in RM system was estimated to be 2.06% (Table 2), exceeding the IPCC’s default value of 1.00% for N2O in agricultural soils37. During the two-year experimental period, the feed EF− C and EF− N from RS coculture were estimated to be 7.81% and 0.90%, respectively, marking significant decreases compared with the findings in RM systems. Furthermore, the estimated EF− N from the rice–giant river prawn coculture was lower than those estimated in rice–crayfish coculture by Fang et al. (2023)3. The relatively lower EF− C and EF− N values observed in RS coculture systems in the present study indicate that these integrated agricultural approaches can markedly enhance the utilization rates of C and N from artificial feeds while simultaneously reducing the CH4 and N2O emissions compared with reclaimed RM.

Implications and future outlook

To guarantee food security and stabilize cultivated land resources, the Chinese government has implemented a series of policies regarding the reclamation of aqua cultural ponds into rice paddies. The findings indicate that such a land use change would significantly exacerbate greenhouse gas emissions and decrease farm incomes, particularly when considering both the GWP and the NEEP. However, further conversion to rice–fish coculture systems could offer advantages in mitigating greenhouse gas emissions and increasing agricultural profits compared with rice monoculture. Given the extensive distribution of reclaimed fields in southeast China, RS coculture, as an integrated rice and aquaculture system, could serve as an ideal compromise to solve the problems caused by land use change, although CH4 and N2O emissions still need to be reduced to mitigate the negative environmental impacts. The results will also be useful for other Asian countries that possess paddy fields suitable for rice–aquaculture coculture strategies, such as India, Vietnam, and Malaysia.

By experimenting with diverse feed formulations and implementing innovative management strategies, such as adjusting feeding frequencies and integrating beneficial bacteria, it is possible to enhance the utilization efficiency of C and N in feed, boost the productivity of coculture, and reduce the GHG discharge. Additional agricultural practices, including improved aeration, water level adjustment, and the reduced application of quicklime, can further contribute to decreasing CH4 and N2O emissions from RS coculture systems. In addition, the measurement conducted in this study was limited to a particular location, climatic conditions, water exchange rate, and stocking density, focusing on a specific set of aquaculture species. Consequently, generalizing the average fluxes obtained in the present study to national estimates of CH4 and N2O emissions from inland aquaculture ponds and reclaimed rice–aquaculture fields is inappropriate. Considering the importance of various factors, including the regional climate, land use type, fish/shrimp species, feeding rate and amount, and specific management practices, there is a pressing need to obtain multiple sets of direct field measurements.

Conclusion

This study provides early insights into the variations in CH4 and N2O emissions caused by the reclamation of SP to RM, followed by a shift to RS coculture across a two-year measurement period. The results revealed that the conversion of SP to RM significantly increased CH4 and N2O emissions. Nevertheless, the subsequent transition of RM into RS coculture fields drastically reduced the CH4 and N2O emissions. Moreover, the present study explored the main environmental factors influencing CH4 and N2O emissions, as well as the GWP and environmental implications, from the three systems. This study proposes the conversion of SP to RS coculture production as a viable solution to partially address the challenge of balancing the growing demand for grain against the low profit and high GWP of RM. In addition, this work suggests strategies to further mitigate GHG emissions in rice–aquaculture systems, including adjusting water levels, modifying drainage practices, utilizing aerators, and reducing the frequency of quicklime application.