- Email: [email protected]
Modeling the impacts of water and fertilizer management on the ecosystem service of rice rotated cropping systems in China
Agriculture, Ecosystems and Environment 219 (2016) 49–57
Contents lists available at ScienceDirect
Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee
Modeling the impacts of water and fertilizer management on the ecosystem service of rice rotated cropping systems in China Han Chena , Chaoqing Yua,* , Changsheng Lib , Qinchuan Xinc,a , Xiao Huanga , Jie Zhanga , Yali Yuea , Guorui Huanga , Xuecao Lia , Wei Wanga a b c
Center for Earth System Science, Key Lab of Earth System Numerical Simulation, Tsinghua University, China Institute for the Study of Earth, Oceans, and Space, University of NH, USA Department of Geography and Planning, Key Lab of Urbanization and Geo-simulation, Sun Yat-sen University, China
A R T I C L E I N F O
A B S T R A C T
Article history: Received 6 May 2015 Received in revised form 15 November 2015 Accepted 26 November 2015 Available online xxx
Detailed information on the impacts of water use and nutrient application on agro-ecosystem services including crop yields, greenhouse gas (GHG) emissions and nitrogen (N) loss is the key to guide field managements. In this study, we use the DeNitrification–DeComposition (DNDC) model to simulate the biogeochemical processes for rice rotated cropping systems in China. We set varied scenarios of water use in more than 1600 counties, and derived optimal rates of N application for each county in accordance to water use scenarios. Our results suggest that 0.88 0.33 Tg per year (mean standard deviation) of synthetic N could be reduced without reducing rice yields, which accounts for 15.7 5.9% of the N application in China in 2005. Field managements with shallow flooding and optimal N applications could enhance ecosystem services at a national scale, leading to 34.3% reduction of GHG emissions (CH4, N2O, and CO2), 2.8% reduction of overall N loss (NH3 volatilization, denitrification and N leaching) and 1.7% increase of rice yields, as compared to current management conditions. Among provinces with major rice production, Jiangsu, Yunnan, Guizhou, and Hubei could achieve more than 40% reduction of GHG emissions under appropriate water managements, while Zhejiang, Guangdong, and Fujian could reduce more than 30% N loss with optimal N applications. Our modeling efforts suggest that China is likely to benefit from reforming water and fertilization managements for rice rotated cropping systems in terms of sustainable crop yields, GHG emission mitigation and N loss reduction, and the reformation should be prioritized in the above-mentioned provinces. ã 2015 Elsevier B.V. All rights reserved.
Keywords: Water regime Nitrogen fertilization Sustainable management Ecological modeling DNDC
1. Introduction Rice is one of the major commodity crops and feeds more than half of the world’s population (Lobell et al., 2011). Being the largest rice producing country in the world, China contributes 27.3% of global rice production on 18.3% of global paddy fields (FAO, 2013). It is estimated that 20% more rice needs to be produced in China in the coming two decades to meet food demands as population increases (Peng et al., 2009). Meanwhile, rice cultivation has resulted in the degradation of a wide range of agro-ecosystem services in China. For example, China emits approximately 7.41 Tg CH4 per year, one of the primary non-CO2 greenhouse gases (GHG),
* Corresponding author at: Tsinghua University, Mengminwei South Building, Room 916, Beijing 100084, China. E-mail address: [email protected] (C. Yu). http://dx.doi.org/10.1016/j.agee.2015.11.023 0167-8809/ ã 2015 Elsevier B.V. All rights reserved.
to the atmosphere due to rice planting alone (Yan et al., 2009). Increasing rice cultivation activities likely lead to higher methane emissions in China under environmental conditions with rising temperature and increasing atmospheric CO2 concentrations (Van Groenigen et al., 2013). In addition, because rice needs to grow in flooded conditions, rice planting typically involves considerable water consumption and nitrogen (N) loss to the environment (Vlek and Byrnes, 1986), which has been found to aggravate water degradation and nutrient pollution in China (Deng et al., 2006; Zhao et al., 2012). Given environmental issues associated with rice planting, there is a need to explore possible solutions that mitigate conflicts between increased food demands and degraded agroecosystem services for China (Miao et al., 2010). Suitable managements of irrigation and fertilization have shown to be capable of enhancing ecosystem services of rice rotated cropping systems by increasing agricultural yields and minimizing environmental pollutions (Burney et al., 2010; Mueller
50
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
et al., 2012). For water schemes, switching continuous flooding to one or several drainages in paddy fields could promote rice yields slightly and reduce water use (Linquist et al., 2015). Intermittent irrigation could also reduce total GHG emissions from rice planting because the overall radiative forcing of CH4 reduction suppresses that of N2O stimulation (Xiong et al., 2007). For fertilizer management practices, N fertilization have found to strongly influence rice yields and N loss through biotic and abiotic processes such as ammonia volatilization, denitrification and nitrogen leaching (Zhang et al., 1996; Ramankutty and Foley, 1999). To derive optimal rates of N application that associated with maximum grain yields and minimum N loss, approaches have been developed based on (1) soil mineral N tests that identify the differences between predicted and measured soil mineral N (NO3-N and NH4+-N) supply (Chen et al., 2006), and (2) regional evaluation method that use numerous field experiments to quantify yield efficiency, net income, and environmental impacts (Zhu and Chen, 2002). Moreover, as found in field experiments and literature reviews, yield-scaled GHG emissions (i.e., GHG emissions per unit rice yield) tend to have minimized values at the rates of N application that optimize rice yields (Feng et al., 2013). Water managements may influence nitrogen use efficiency through various N pathways like denitrification, leaching and ammonia volatilization thus the identification of optimal N rates to achieve the best yield (Sun et al., 2012; Liu et al., 2013). These efforts clearly indicate that strategies of water and N uses hold promise to improve grain yields while reducing GHG emissions and N loss from rice fields, and the water and N managements should be considered simultaneously. Quantitative evaluations of managements on agro-ecosystem services of rice rotated cropping systems are critical to understand potential environmental benefits from managements, but evaluations over large areas like mainland China are challenging for some reasons. First, conflicting conclusions might be drawn from previous studies when performing metaanalysis, largely owing to incomplete records of field management schemes. For example, the response of CH4 emissions to N inputs varies with water regimes, while activities like drainage frequency and strength are not well documented among sites (Pittelkow et al., 2014). Second, it can be challenging to capture optimal N rates based on in-situ experiments (Cassman et al., 1998). Third, scaling up from individual sites to large geographical regions is difficult due to spatial variation of climate environment, soil conditions, and management practices (Smith et al., 2010). Thus, regional mitigation potential of GHG emissions and N loss with yield constraints through optimal managements remains highly uncertain. Lastly, physiological models that describe underlying mechanisms of biogeochemical cycles in paddy fields hold great promise to address the issues associated with the meta-analysis method or field measurements. However, due to incomplete modeling strategies and limited data access in China, most of modeling studies have not evaluated the integrated impacts of water and N fertilizer managements on rice rotated cropping systems (Li et al., 2005; Cheng et al., 2013). Compounding these concerns, the objectives of this study are to (1) quantify national GHG emissions, rice yields, and N loss to the environment from rice rotated cropping systems under varied management scenarios of water and N uses; and (2) identify optimal management combinations that potentially enhance ecosystem services of rice rotated cropping systems in China. To meet these objectives, we applied a biogeochemical model of DeNitrification–DeComposition (DNDC) with county-scale agricultural database that contains climate, soil, vegetation, and management information in China.
2. Materials and methods 2.1. Model descriptions DNDC is a process-based biogeochemical model that simulates carbon dynamics and trace gas emissions for agro-ecosystems (Li et al., 1992). The DNDC model consists of three primary components that model crop growth, soil environment, and soil microbe processes. The crop growth sub-model tracks phenological development, biomass accumulation and allocation, demand and uptake of water and N, respiration and litter production. The soil environment sub-model simulates soil temperature, soil moisture, soil redox potential (i.e., Eh), soil physical properties, and soil oxygen status. The sub-model of soil microbe processes simulates soil biogeochemistry reactions such as decomposition, nitrification, denitrification and fermentation, in which processes trace gases (such as N2O, CH4, NO, and NH3) are produced. Agricultural management practices, including fertilization, tillage, irrigation, manure amendment and grazing, are considered in DNDC to account for various biogeochemical processes. Though DNDC has considered the impacts of farmland water conservancy facilities together with N contents in flooded water and calculated overflow on N runoff, detailed descriptions of these parameters are challenging to achieve over large regions thus set to constant inside DNDC, which would produce uncertainties in specific regions. Therefore, modeling results of N runoff were not included in this study. DNDC has been widely used to simulate greenhouse gas emission, crop growth, ecosystem carbon dynamics, soil water balance, and vegetation biochemical cycles (Fumoto et al., 2008; Deng et al., 2011; Uzoma et al., 2015). 2.2. Model revision to account for rice transplant Different from other places in the world, rice cultivation in China often involves practices of transplantation. Rice seedlings are typically transplanted from a seedling bed to paddy fields right after the harvests of previous crops. Transplantation saves growing periods to meet requirements of temperature accumulation, thus it is important for double- and triple-cropping systems in China. Current version of DNDC does not include a function of rice transplantation, such that direct modeling of rice growth and yields tends to be biased for rotated cropping systems. Here, we implement a new scheme and modify DNDC to include parameters that define dates of rice transplantation. The schematic descriptions of crop growth sub-models are shown in Fig. S.1 (see supporting information Appendix A) for comparisons between current and revised versions of DNDC. 2.3. National database construction To run the DNDC model for more than 1600 counties with rice cultivation in China, we constructed county-level databases that contain soil, weather and management data, including: (1) daily meteorological data with 2-m air temperature ( C) and precipitation (mm); (2) soil parameters with soil organic matter content (kg C kg1 soil), clay fraction (0–1), pH, and bulk density (g cm3); (3) Crop parameters with maximum grain yield (kg C ha1), effective accumulated temperature ( C), crop water demand (g water g1 dry matter), and annual rice yield increase index (%); (4) crop system information with rotation types and areas; (5) field management practices that involve sowing/harvest dates, N fertilizer amounts (kg N ha1), irrigation ratios within a county (0–1), tillage dates and methods, manure amounts (kg N ha1), crop straw return proportions (0–1), flooding methods, and transplant dates (if rice rotated cropping system). Soil parameters
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
are expressed with upper and lower limits rather than absolute values to indicate the range of soil inputs for each county. Ten rice-rotated systems are considered in model simulations: rice, rice–rice, rice–wheat, rice–rapeseed, rice–rice–vegetable, rice–vegetable, oat–rice, rice–soybean, rice–rice–rapeseed and rice–rice–wheat. The areas of rotation systems in each county were derived by combining the county-scale statistical database of cropsown areas with satellite-based land-cover map for mainland China (Frolking et al., 2002). Meteorological measurements from 620 weather stations were obtained from China Meteorological Data Sharing Service System (http://cdc.cma.gov.cn/home.do). Stationary meteorological datasets from 1996 to 2010 were assigned to the nearest counties. County-level soil parameters were derived from digitized national soil maps [Institute of Soil science, 1986] and other information [National Soil Survey Office of China, 1997]. Soil Mn, Fe, and sulfate contents were set as national average values for paddy soils, where Mn = 30 mg kg1 soil, Fe = 80 mg kg1 soil, and sulfate = 220 mg kg1 soil. Information on agricultural management practices was obtained from Crop Cultivation Bureau of China (http://www.zzys.moa.gov.cn/; in Chinese). The type of N fertilizer in paddy fields was mostly urea and top-dressed into floodwater across China (Peng et al., 2009). The N application rate (kg N ha1) was derived for each province using Huang's method (Huang and Tang, 2010) with statistical data from National Bureau of Statistics of China (http://www.stats.gov. cn/english/). The manure amendment rates were calculated based on province-level human population and animal livestock (Wang et al., 2006). Ministry of Agriculture of China reported that 31.30% of crop residues were left in the field in 2009, of which the percentage value was used for all paddy fields. The conventional tillage and mid-season drainage were applied across China using default settings in DNDC. 2.4. Model calibration and evaluation Simulation of rice growth is essential to the predictions of GHG emission and N loss due to the substrate supply/uptake and gas transport. Given model parameters in DNDC, it is computationally impractical to search the entire value spaces for the best parameter sets for each county in China. Among a variety of parameters, four parameters were found to be the most sensitive ones that influence crop growth, namely effective accumulated temperature ( C), crop water demand (g water g1 dry matter), maximum grain yield (kg C ha1), and annual changing percentage of maximum grain yield (%). Effective accumulated temperature describes temperature requirement for crop growth; crop water demand stands for
51
crop water productivity, maximum grain yield describes potential grain production without stresses from water, nutrient, and temperature; and annual changing percentage of maximum grain yield denotes the changing rate of maximum grain yield due to technology advancement. To find suitable parameter values for each county, we employ the Shuffled Complex Evolution Approach (SCEA) to speed up model calibration processes (Duan et al., 1993). The root-meansquare deviation serves as the cost function in SCEA. County-level rice yields from 2000 to 2008 in China were used as reference data for calibration, whereas yield data from 1995 to 1999 and 2009 were used for model validation. Statistical metrics such as coefficient of determination (R2) and root mean square error (RMSE) were used for model evaluation, and RMSE was calculated as follows: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u n uX 2 u X obs;i X model;i u t i¼1 RMSE ¼ ð1Þ n where X obs;i and X model;i represent observed and modeled values, respectively; n is the total number of samples. 2.5. Scenario analysis Using the calibrated DNDC model, we conducted analysis for varied scenarios of water and N uses. For water use scenarios, we evaluated three widely used irrigation schemes of continuous flooding, mid-season drainage and shallow flooding. In the scheme of continuous flooding, paddy fields are continuously flooded with a 5–10 cm depth of surface water during rice growing period; in the scenario of mid-season drainage, paddy fields experience two drainages and each last for one week at late tillering stage and restore continuous flooding for the rest; and in the scheme of shallow flooding, the water table of rice paddy is constantly fluctuating 5–10 cm above and below the soil surface (Li et al., 2006). For scenarios of N fertilizer uses, we compared results from three application schemes of zero N inputs, optimal rates of N inputs, and actual rates of N inputs. The zero N inputs denote the scheme without using any N fertilizer, the actual rates of N inputs denote the scenarios that N fertilizer are used in field managements in reality, and the optimal rates of N inputs denote the scheme that maximum rice yields can be achieved with minimum N uses. The optimal rates of N inputs are derived for each corresponding scheme of water use and for each county. We varied
Fig. 1. Comparisons of measured and simulated county-level rice yields are shown: (A) model calibration using data in 2000–2008; (B) model validation using data in 1995– 1999; (C) model validation using data in 2009.
52
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
N inputs from zero to maximum values and identified the first point, after which rice yields never increase. This point was then considered as the optimal N rate for the corresponding county. Given three water use scenarios and three N use scenarios, we compared results from nine (3 3) scenarios to assess the impacts of water and fertilizer management interactions on rice yields, GHG emissions, and N losses. To allow direct comparisons, we set a baseline using scenarios of mid-season drainage water use and actual rates of N use for the year of 2010. Based on meteorological datasets, the year of 2005 was identified as a normal year in term of annual average temperature and precipitation. We used climate data in 2005 as baselines and performed a twenty-year simulation for each county, in order to (1) initialize soil background values and relative sizes of each pool in DNDC, and (2) avoid the impacts of inter-annual meteorological variance on the optimal N rates calculation and scenario analysis. The GHG emissions are derived for three key components of CO2, N2O, and CH4 as follows: GHG = [CO2] + 298 [N2O] + 25 [CH4]
(2)
where [CO2], [N2O] and [CH4] are CO2 flux (kg CO2 ha1 yr1), N2O flux (kg N2O ha1 yr1) and CH4 flux (kg CH4 ha1 yr1), and they are converted as [CO2] = C 44/12, [N2O] = N 44/28, [CH4] = C 16/12. C and N are fluxes in carbon (kg C ha1 yr1) and nitrogen (kg N ha1 yr1) units, respectively. 298 and 25 represents the global warming potential in kg CO2 equivalent ha1 yr1 of N2O and CH4 over a 100-yr horizon (Houghton et al., 2001). The yield-scaled GHG emissions are derived as follows: Yield scaled GHG ¼
GHG yield
ð3Þ
To address uncertainties of CH4 and N2O emissions due to varied soil conditions within each county, we applied the approach of Most Sensitive Factor (MSF) (Li et al., 2004). Sensitivity tests
were conducted and soil texture and soil organic matter content (SOC) were identified as two of the most sensitive parameters that influence CH4 and N2O emissions from paddy fields. Then we varied these factors within the range from county-level database, and performed DNDC twice to obtain the range of CH4 and N2O emissions: first with low SOC, low pH, and high clay content, and second with high SOC, high pH, and low clay content to most represent variations of actual county-level fluxes. 3. Results 3.1. Model calibration and validation Fig. 1 shows the results between observed and modeled rice yields for more than 1600 counties with rice cultivation. Overall, our model could explain 84%, 80% and 79% variance of measured rice yield for model calibration periods in 2000–2008 and model validation periods in 1995–1999 and in 2009, respectively. The RMSE were 763.86, 887.11 and 881.10 kg ha1 for calibration periods and two validation periods, respectively. The modeled results might not capture the observations of low yields well for some counties in specific years, which are likely due to disturbance factors such as yield damages caused by insects, typhoon, or floods. Given the large extent of our study area and the long time series of our study period, our model performs well in general. In addition to rice yields, observed and modeled GHG and N loss are shown in Fig. 2 for six field sites across China. The six validation sites are distributed in main rice production regions from Northeast to South (see Fig. S.2 & Table S.1 in supporting information Appendix A). DNDC could simulate the impact of mid-season drainages on N2O stimulation (Fig. 2A). Modeled results in Fig. 2C–E indicate that DNDC is able to trace N leaching and NH3 emission dynamics under nitrogen fertilizer applications. The influence of rice growth stages on CH4 emission rate was
Fig. 2. Observed and modeled time series in paddy fields are shown for (A) N2O emission for Wangcheng county in Hunan province; (B) CH4 emission for Beibei county in Chongqing municipal district; (C) NH3 emission for Wuming county in Guangxi province; (D) N leaching for Qingpu county in Shanghai municipal district; (E) NH3 emission for Kunshan county in Jiangsu province; and (F) CH4 emission for Sanjiang plain in Heilongjiang province.
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
53
Fig. 3. The spatial distribution of estimated ratios of N over-application are shown for water use scenarios of (A) CF denotes continuous flooding, (B) MD denotes mid-season drainage, and (C) SF denotes shallow flooding.
shown in Fig. 2B and F. These results suggest that DNDC could capture the phases and magnitudes of the processes of CH4 and N2O emission, ammonia volatilization and nitrogen leaching for paddy fields under a variety of management practices. Fig. S.3 (see supporting information Appendix A) illustrates how optimal N rates are derived for various rice cropping systems across China. Regional variations of optimal N rates may be mainly caused by the discrepancies of environmental N supplies, soil properties and climate conditions. The maximum yields and minimum yield-scaled GHG emissions occurred at optimal N rates, which agree with previous works (Pittelkow et al., 2014). 3.2. application, N loss, and rice yields According to the yearbook of 2005, the cultivated areas of paddy fields were 27.6 million ha in 2005, average rice productivity was 6279.4 kg ha1, and approximate 5.6 Tg chemical N fertilizer (93.6 kg N ha1) were applied on these lands. Nationwide areaweighed optimal N rates of paddy fields calculated with DNDC are 78.9 4.5 kg N ha1 (mean standard deviation as derived based on the variability of three water management practices; same hereafter unless notified). National yields of rice under optimal N rates are estimated to be 6411 5.6 kg ha1, which increases slightly as compared to current yields. The increased rice yields are largely due to yield enhancement in regions with under-applied mineral N. There are no significant biases in modeled rice yields at the provincial level, as indicated by student's t-test under 95% confidence intervals for all water regimes. This indicates 15.7 5.9% reduction of N inputs can be achieved without rice yield reductions.
Nationwide optimal N rates under mid-season drainage (MD; 80.2 kg N ha1) and shallow flooding (SF; 83.7 kg N ha1) water scenarios were higher than that under the scenario of continuous flooding (CF; 72.9 kg N ha1). Water management practices of MD and SF lead to more N2O emission and relatively lower nitrogen use efficiency (Siebert and Döll, 2010). In the CF scenario, approximately 20% of paddy fields in China are over-applied with more than 40% mineral nitrogen. The area ratios fall to 17% and 14% under MD and SF water regimes, respectively (see Fig. S.4 in supporting information Appendix A). At a national scale, the amounts of overused N fertilizers are estimated to range from 0.60 Tg yr1 (SF) to 1.25 Tg yr1 (CF). The estimated amount of over-applied mineral N was highly uneven across the country (Fig. 3). Given SF water regime scenario, paddy fields with mineral N over-application were mainly distributed in southern China (such as Guangdong and Hunan provinces) and eastern China (such as Fujian and Zhejiang provinces), where more than 31% excessive mineral N were applied than optimal N rates. Paddy fields with under-applied N appeared in Anhui, Yunnan, Guizhou, and Hubei, where more than 20% N fertilizer inputs were needed to fulfill rice growth. As shown in Table 1, N losses from the whole rice rotation systems increase with increased N inputs. Surplus N in the rice season can stimulate further NH3 emission, denitrification and N leaching in other rotated crop seasons. For the whole rice rotated cropping systems, compared to other pathways, ammonia volatilization represents a primary contributor to total N loss and accounts for 25.5–30.9% of total N fertilizer and manure inputs on a national scale. This ratio were higher than that of 9–18% for agricultural lands in 1990 (Xing and Zhu, 2000). Several reasons
Table 1 Pathways of N inputs and N loss under various N fertilizer application rates and water management practices at a national scale. Scenarioa
CF_N_0 CF_N_Opt CF_N_Act MD_N_0 MD_N_Opt MD_N_Act SF_N_0 SF_N_Opt SF_N_Act a
N inputs (kg N/ha/yr)
N loss for rice growth season (kg N/ha/yr)
N loss for rice rotated systems (kg N/ha/yr)
N rate
Manure
NH3 volatilization
Leaching
Denitrification
NH3 volatilization
Leaching
Denitrification
80.47 157.68 174.21 80.47 165.43 174.21 80.47 169.01 174.21
10.48 10.48 10.48 10.48 10.48 10.48 10.48 10.48 10.48
0.66 13.69 15.83 0.75 18.70 17.96 0.80 26.43 22.69
0.74 4.57 6.74 0.58 3.56 4.26 0.58 4.05 5.03
18.95 27.18 32.19 20.70 36.72 41.33 17.24 31.51 36.36
31.22 44.63 47.17 29.85 48.01 47.50 29.68 55.53 51.93
2.62 6.91 9.28 2.59 5.88 6.67 2.72 6.35 7.41
22.90 31.70 37.44 25.07 41.69 46.89 28.95 43.78 49.01
N_0 denotes zero N fertilizer inputs, N_Opt denotes optimal N application rates derived using DNDC, and N_Act denotes actual nitrogen fertilizer application rates.
54
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
Fig. 4. GHG emissions, yields and yield-scaled GHG emissions of nine scenarios. The error bar indicates the uncertainties of GHG emissions due to spatial heterogeneity of input soil data.
contribute to this overestimate: Firstly, the past two decades has seen the gradually increase of N fertilizer inputs and decrease in N productivity of agricultural soil in China (Zhang et al., 2013), which would induce more NH3 emission in 2005 compared to that in 1990; Secondly, uncertainties exist in Xing et al., 2000’ method to estimate NH3 emission based on the emission factors of different nitrogen fertilizers over regions (Xing and Zhu, 2000). Lastly, extra emission of NH3 during the fallow period is not included in the previous study. For the rice growing season when comparing our results with previous studies, our estimates of NH3 emissions (3.9– 12.5%) are consistent with the observations of 4.4–16.7% in Taihu region (Lin et al., 2007). Meanwhile, our estimation indicates denitrification loss could account for 34.3–45.8% of observed applied N, which agree with that (33%) in Henan province (Zhu et al., 1988) and that (41%) during rice growth period (Cai et al., 1991). N leaching from paddy fields appears to be minor as compared to the other two pathways and is only responsible for less than 5% of total N application in our simulation, which is in accord with previous findings (Ju et al., 2009). At the national scale, reducing N application rates toward the optimal rates could decrease overall N loss to the environment by 6.4 3.7%. Although benefiting from water conservation and GHG emissions mitigation, changing water management from CF to intermittent drainages tends to induce higher N loss (7.6% higher from CF to MD or 15.4% higher from CF to SF). The national average GHG emissions, rice yields and yieldscaled GHG emissions of nine scenarios were calculated and shown in Fig. 4. The annual average GHG flux from paddy fields in China was 8609.4 2659.1 kg CO2 equivalent ha1 and the national total amount of GHG emissions was 237.6 73.4 Tg CO2 equivalent under current N rates. The nationwide flux of yield-scaled GHG emissions from rice fields was 1.37 0.51 kg
CO2-equivalents kg1-grain yield yr1, which fell within previous estimates (3.22 kg CO2-equivalent kg1-grain yield yr1) in continuous flooding rice paddies (Li et al., 2006) and estimates (0.24–0.74 kg CO2-equivalent kg1-grain yield yr1) in rice paddies with midseason drainage and organic manure incorporation (Qin et al., 2010). At a nationwide scale, changing water regimes from CF to MD and SF could lead to 32.3% and 53.6% reduction of GHG emissions, respectively. In comparisons, changing observed N rates to optimal rates could reduce national GHG emissions by 2.2 1.7% (Fig. 4), mostly due to reductions in N2O emission rather than CO2 and CH4 emissions (Table 2). Meanwhile, yield-scaled GHG emissions are reduced from 1.37 0.51 to 1.32 0.52 kg CO2-equivalent kg1grain yield yr1 (or 3% reduction in equivalent). Among the nine scenarios, yield-scaled GHG emissions are minimized under SF water regime with optimal N rates. Compared to the reference scenario (mid-season drainage with actual N rates), changing water management practices from MD to SF contributes to 31.5% reduction of yield-scaled GHG emissions and further 4.2% decreases are achieved through reducing current N inputs to optimal rates. The error bars in Fig. 4 indicated the impacts of soil heterogeneity on yields, GHG emissions and yield-scaled GHG emissions. The variation coefficients of GHG emissions induced by soil heterogeneity are 0.34, 0.01 and 0.29 for CF, MD and SF water regime, respectively, which implies that the uncertainties of estimated GHG emissions from soil inputs minimized under midseason drainage. Similar results are obtained with respect to yieldscaled GHG emissions (0.37, 0.02 and 0.29 for CF, MD and SF water regime). As shown in Fig. 5, among main rice production provinces in China (provinces that together contribute more than 90% of rice
Table 2 Summary of CH4, N2O and CO2 emissions (units: CO2 equivalent ha1 yr1) under varied managements of water and nitrogen fertilizer.
CO2 N2O CH4
CF_N_0
CF_N_Opt
CF_N_Act
MD_N_0
MD_N_Opt
MD_N_Act
SF_N_0
SF_N_Opt
SF_N_Act
1994 445 8113
888 672 10544
919 1107 10145
2075 586 4490
1036 1206 5633
1071 1587 5484
1866 815 2389
880 1431 2917
942 1726 2855
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
55
Fig. 5. Estimated GHG emissions and yield-scaled GHG emissions of different provinces: (a) under shallow flooding with optimal application rates of mineral N; (b) mitigation potential of water and nitrogen managements. The error bars are standard deviations derived from the low, mean and high estimates of GHG emissions due to uncertainties of soil inputs. MD_Obs_N, SF_Obs_N and SF_Opt_N represent mid-season drainage + current application rates of mineral N, shallow flooding+ current application rates of mineral N and shallow flooding + optimal application rates of mineral N. The red line indicates rice cultivated areas of different provinces. Abbreviations of province: AH, Anhui; BJ, Beijing; FJ, Fujian; GD, Guangdong; GS, Gansu; GX, Guangxi; GZ, Guizhou; HEB, Hebei; HEN, Henan; HLJ, Heilongjiang; HN, Hainan; HUB, Hubei; HUN, Hunan; IM, Inner Mongolia; JL, Jilin; JS, Jiangsu; JX, Jiangxi; LN, Liaoning; NX, Ningxia; QH, Qinghai; SAX, Shaanxi; SC, Sichuan; SD, Shandong; SH, Shanghai; SX, Shanxi; TB, Tibet; TJ, Tianjin; XJ, Xinjiang; YN,Yunnan; ZJ, Zhejiang.
cultivated areas in China), the highest GHG emissions (>5700 CO2 equivalent ha1 yr1) occurred in Heilongjiang (high SOC), Zhejiang (multiple rotation cropping systems), Anhui (multiple cropping systems), and Yunnan (high manure inputs). The lowest (<2000 CO2 equivalent ha1 yr1) GHG emissions occurred in Guizhou (high clay fraction), Jiangxi (low manure inputs), and Jiangsu (less multiple cropping systems). When changing water regimes from MD to SF, the highest reduction of GHG emissions occurred in Jiangsu, Yunnan, Guizhou and Hubei (more than 40% reduction) with high SOC, high multiple crop index and low clay fraction. The highest reduction of induced GHG emissions due to optimal N uses could occur in Zhejiang, Guangdong, and Fujian, where serious over-application of mineral N exists.
uncertainties of N2O emission are likely due to climate and soil variance, inconsistent management practices and limited observation datasets among these studies (Akiyama et al., 2005). Reduced N inputs to optimal rates suppressed N2O emission by an average of 26.8%. This change stimulates CH4 emission slightly because CH4 transportation rates through plants are also enhanced with increased rice growth under optimal N application rates (Sass et al., 1990). CO2 emissions are less sensitive to application rates of N fertilizer, but are more sensitive to water management practices and soil properties because rates of both decomposition and crop residue incorporation could be affected by water regimes and soil properties (Li et al., 2005). In sum, nation-average GHG emissions in rice cropping systems in China are minimized under SF water regime with optimal N application rates.
4. Discussion 4.2. Uncertainties in deriving optimal N rates 4.1. The impacts of water and nitrogen management on three GHG emissions In our study, national averaged methane emissions range from 71.7–316.3 CH4–C ha1 yr1 in the scenario of continuous flooding (Table 2). Our estimates fall within observed ranges of CH4 fluxes (9 to 725 kg C ha1 yr1) (Cai et al., 2003) and are close to previous estimates of 90–214 kg C ha1 yr1 (Frolking et al., 2004). CH4 could account for fewer GHG emissions with increased intermittent drainages (82%, 67% and 52% for CF, MD and SF, respectively). National average N2O emission is lower (2.36 kg N ha1 yr1) in CF water regime than in MD and SF scenarios (3.39 and 3.69 kg N ha1 yr1, respectively). Frolking et al. (2004) estimated a higher emission rate of N2O (5.7–58 kg ha1 yr1) over the 100-year simulations for all water regimes. Akiyama et al. (2005) estimated a lower emission flux of 1.82 kg N ha1 yr1 by compiling and analyzing available data from rice paddy fields in Asia. Large
Discrepancies in deriving optimal rates of N application exist when comparing our result of 23.6% reduction for paddy fields in Taihu region to that of 33% reduction using knowledge-based optimum N fertilization (Zhu and Chen, 2002) and that of 15–25% reduction with soil mineral N tests (Hofmeier et al., 2015). Similarly, studies on a rice–wheat rotated cropping system in southwest China have demonstrated 18% reduction with an integrated nutrient management strategy (Fan et al., 2007). The differences in optimal rates of N application are likely due to several factors: first, there are scaling issues among different studies. Here, we uses N fertilizer rates and manure applications at a provincial level, while previous studies obtain application rates of synthetic N fertilizers and manure through field surveys (Fan et al., 2007). Second, water management strategies also involve the frequency and strength of drainages, but related information still lacks in detail (Ju et al., 2009). The missed
56
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
information in our simulation could influence modeling of the processes of denitrification (Cai et al., 1997) and ammonia volatilization (Li et al., 2008). Third, we have not included enhanced N management strategies such as deep placement of fertilizer, slow-release N fertilizers, nitrification, and urease inhibitors (Vitousek et al., 2009; Giltrap et al., 2010), nor we calculated optimal N through matching crop requirements at varied growth stages based on environment N supplies and estimated crop N demands (Qiu et al., 2009). These factors are challenging issues in large-scale modeling efforts and are worthy further explorations. 5. Conclusions Improving food productions while reducing environmental risks have become one of the predominant challenges that contemporary agriculture face nowadays (Godfray et al., 2010). To understand how managements of water and nitrogen fertilizer influence ecosystem services of rice cropping systems in China, we calibrated and validated DNDC models and compared results for nine contrasting scenarios. We estimate 0.88 0.33 Tg of synthetic N per year was overapplied into paddy fields in China. Shallow flooding under optimal rates of N use show promises to enhance agro-ecosystem services, which increases rice yields by 1.7% and reduces N loss and GHG emissions by 2.8% and 34.3%, respectively. The priority to reform water and fertilizer managements should be given to the provinces with heavy over-application of N fertilizer like Zhejiang, Guangdong, and Fujian, and to the provinces with large GHG mitigation potential like Jiangsu, Yunnan, Guizhou, and Hubei. Future researches should be conducted to address uncertainties associated with county-level nitrogen inputs and scaling factors. Acknowledgements We gratefully acknowledge the support of the China 973 project 2013CB956600, CNSF#41371491, CNSF#41401484, Tsinghua20121088052, SEPA201309062 and Tsinghua Fudaoyuan Research Fund. Thanks also go to the anonymous reviewers for their constructive comments. In memory of Prof. Changsheng Li, we especially thank him for his enthusiasm and dedication to the research and education. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.agee.2015.11.023. References Akiyama, H., Yagi, K., Yan, X., 2005. Direct N2O emissions from rice paddy fields: summary of available data. Global Biogeochem. Cycles 19, GB1005. Burney, J.A., Davis, S.J., Lobell, D.B., 2010. Greenhouse gas mitigation by agricultural intensification. Proc. Natl. Acad. Sci. 107, 12052–12057. Cai, G., Cao, Y., Yang, N., Lu, Y., Zhuang, L., Wang, X., Zhu, Z., 1991. Direct estimation of nitrogen gases emitted from flooded soils during denitrification of applied nitrogen. Pedosphere 1, 241–251. Cai, Z., Sawamoto, T., Li, C., Kang, G., Boonjawat, J., Mosier, A., Wassmann, R., Tsuruta, H., 2003. Field validation of the DNDC model for greenhouse gas emissions in East Asian cropping systems. Global Biogeochem. Cycles 17 GB1107. Cai, Z., Xing, G., Yan, X., Xu, H., Tsuruta, H., Yagi, K., Minami, K., 1997. Methane and nitrous oxide emissions from rice paddy fields as affected by nitrogen fertilisers and water management. Plant Soil 196, 7–14. Cassman, K., Peng, S., Olk, D., Ladha, J., Reichardt, W., Dobermann, A., Singh, U., 1998. Opportunities for increased nitrogen-use efficiency from improved resource management in irrigated rice systems. Field Crops Res. 56, 7–39. Chen, X., Zhang, F., Römheld, V., Horlacher, D., Schulz, R., Böning Zikens, M., Wang, P., Claupein, W., 2006. Synchronizing N supply from soil and fertilizer and N demand of winter wheat by an improved Nmin method. Nutr. Cycl. Agroecosyst. 74, 91–98.
Cheng, K., Ogle, S.M., Parton, W.J., Pan, G., 2013. Simulating greenhouse gas mitigation potentials for Chinese Croplands using the DAYCENT ecosystem model. Global Change Biol. 20, 948–962. Deng, J., Zhou, Z., Zhu, B., Zheng, X., Li, C., Wang, X., Jian, Z., 2011. Modeling nitrogen loading in a small watershed in southwest China using a DNDC model with hydrological enhancements. Biogeosciences 8, 2999–3009. Deng, X., Shan, L., Zhang, H., Turner, N.C., 2006. Improving agricultural water use efficiency in arid and semiarid areas of China. Agric. Water Manage. 80, 23–40. Duan, Q., Gupta, V.K., Sorooshian, S., 1993. Shuffled complex evolution approach for effective and efficient global minimization. J. Optim. Theory Appl. 76, 501–521. Fan, M., Lu, S., Jiang, R., Liu, X., Zeng, X., Goulding, K.W., Zhang, F., 2007. Nitrogen input, 15N balance and mineral N dynamics in a rice–wheat rotation in southwest China. Nutr. Cycl. Agroecosyst. 79, 255–265. FAO. 2013. FAO statistical databases. Available at http://faostat3.fao.org/home/E. (verified 19 Nov. 2014). Feng, J., Chen, C., Zhang, Y., Song, Z., Deng, A., Zheng, C., Zhang, W., 2013. Impacts of cropping practices on yield-scaled greenhouse gas emissions from rice fields in China: a meta-analysis. Agric. Ecosyst. Environ. 164, 220–228. Frolking, S., Li, C., Braswell, R., Fuglestvedt, J., 2004. Short- and long-term greenhouse gas and radiative forcing impacts of changing water management in Asian rice paddies. Global Change Biol. 10, 1180–1196. Frolking, S., Qiu, J., Boles, S., Xiao, X., Liu, J., Zhuang, Y., Li, C., Qin, X., 2002. Combining remote sensing and ground census data to develop new maps of the distribution of rice agriculture in China. Global Biogeochem. Cycles 16 1091. Fumoto, T., Kobayashi, K., Li, C., Yagi, K., Hasegawa, T., 2008. Revising a process-based biogeochemistry model (DNDC) to simulate methane emission from rice paddy fields under various residue management and fertilizer regimes. Global Change Biol. 14, 382–402. Giltrap, D.L., Li, C., Saggar, S., 2010. DNDC: a process-based model of greenhouse gas fluxes from agricultural soils. Agric. Ecosyst. Environ. 136, 292–300. Godfray, H.C.J., Beddington, J.R., Crute, I.R., Haddad, L., Lawrence, D., Muir, J.F., Pretty, J., Robinson, S., Thomas, S.M., Toulmin, C., 2010. Food security: the challenge of feeding 9 billion people. Science 327, 812–818. Hofmeier, M., Roelcke, M., Han, Y., Lan, T., Bergmann, H., Böhm, D., Cai, Z., Nieder, R., 2015. Nitrogen management in a rice–wheat system in the Taihu Region: recommendations based on field experiments and surveys. Agric. Ecosyst. Environ. 209, 60–73. Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C., 2001. Radiative forcing of climate change. Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK and New York, USA 881 pp.. Huang, Y., Tang, Y., 2010. An estimate of greenhouse gas (N2O and CO2) mitigation potential under various scenarios of nitrogen use efficiency in Chinese croplands. Global Change Biol. 16, 2958–2970. Ju, X., Xing, G., Chen, X., Zhang, S., Zhang, L., Liu, X., Cui, Z., Yin, B., Christie, P., Zhu, Z., 2009. Reducing environmental risk by improving N management in intensive Chinese agricultural systems. Proc. Natl. Acad. Sci. 106, 3041–3046. Li, C., Frolking, S., Frolking, T.A., 1992. A model of nitrous oxide evolution from soil driven by rainfall events: 1. Model structure and sensitivity. J. Geophys. Res. (1984–2012) 97, 9759–9776. Li, C., Frolking, S., Xiao, X., Moore III, B., Boles, S., Qiu, J., Huang, Y., Salas, W., Sass, R., 2005. Modeling impacts of farming management alternatives on CO2, CH4, and N2O emissions: a case study for water management of rice agriculture of China. Global Biogeochem. Cycles 19, GB3010. Li, C., Mosier, A., Wassmann, R., Cai, Z., Zheng, X., Huang, Y., Tsuruta, H., Boonjawat, J., Lantin, R., 2004. Modeling greenhouse gas emissions from rice-based production systems: sensitivity and upscaling. Global Biogeochem. Cycles 18, GB1043. Li, C., Salas, W., DeAngelo, B., Rose, S., 2006. Assessing alternatives for mitigating net greenhouse gas emissions and increasing yields from rice production in China over the next twenty years. J. Environ. Qual. 35, 1554–1565. Li, H., Liang, X., Chen, Y., Tian, G., Zhang, Z., 2008. Ammonia volatilization from urea in rice fields with zero-drainage water management. Agric. Water Manage. 95, 887–894. Lin, D., Fan, X., Hu, F., Zhao, H., Luo, J., 2007. Ammonia volatilization and nitrogen utilization efficiency in response to urea application in rice fields of the Taihu Lake region, China. Pedosphere 17, 639–645. Linquist, B.A., Anders, M.M., Adviento, B., Maria Arlene, A., Chaney, R.L., Nalley, L.L., da Rosa, E.F.F., van Kessel, C., 2015. Reducing greenhouse gas emissions, water use, and grain arsenic levels in rice systems. Global Change Biol. 21, 407–417. Liu, L., Chen, T., Wang, Z., Zhang, H., Yang, J., Zhang, J., 2013. Combination of sitespecific nitrogen management and alternate wetting and drying irrigation increases grain yield and nitrogen and water use efficiency in super rice. Field Crops Res. 154, 226–235. Lobell, D.B., Schlenker, W., Costa Roberts, J., 2011. Climate trends and global crop production since 1980. Science 333, 616–620. Miao, Y., Stewart, B.A., Zhang, F., 2010. Long-term experiments for sustainable nutrient management in China. A review. Agron. Sustain. Dev. 31, 397–414. Mueller, N.D., Gerber, J.S., Johnston, M., Ray, D.K., Ramankutty, N., Foley, J.A., 2012. Closing yield gaps through nutrient and water management. Nature 490, 254– 257. Peng, S., Tang, Q., Zou, Y., 2009. Current status and challenges of rice production in China. Plant Prod. Sci. 12, 3–8.
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57 Pittelkow, C.M., Adviento, B., Maria, A., van Kessel, C., Hill, J.E., Linquist, B.A., 2014. Optimizing rice yields while minimizing yield-scaled global warming potential. Global Change Biol. 20, 1382–1393. Qin, Y., Liu, S., Guo, Y., Liu, Q., Zou, J., 2010. Methane and nitrous oxide emissions from organic and conventional rice cropping systems in Southeast China. Biol. Fertil. Soils 46, 825–834. Qiu, J., Li, C., Wang, L., Tang, H., Li, H., Van Ranst, E., 2009. Modeling impacts of carbon sequestration on net greenhouse gas emissions from agricultural soils in China. Global Biogeochem. Cycles 23 GB1007. Ramankutty, N., Foley, J.A., 1999. Estimating historical changes in global land cover: croplands from 1700 to 1992. Global Biogeochem. Cycles 13, 997–1027. Sass, R., Fisher, F., Harcombe, P., Turner, F., 1990. Methane production and emission in a Texas rice field. Global Biogeochem. Cycles 4, 47–68. Siebert, S., Döll, P., 2010. Quantifying blue and green virtual water contents in global crop production as well as potential production losses without irrigation. J. Hydrol. 384, 198–217. Smith, W., Grant, B., Desjardins, R., Worth, D., Li, C., Boles, S., Huffman, E., 2010. A tool to link agricultural activity data with the DNDC model to estimate GHG emission factors in Canada. Agric. Ecosyst. Environ. 136, 301–309. Sun, Y., Ma, J., Sun, Y., Xu, H., Yang, Z., Liu, S., Jia, X., Zheng, H., 2012. The effects of different water and nitrogen managements on yield and nitrogen use efficiency in hybrid rice of China. Field Crops Res. 127, 85–98. Uzoma, K.C., Smith, W., Grant, B., Desjardins, R.L., Gao, X., Hanis, K., Tenuta, M., Goglio, P., Li, C., 2015. Assessing the effects of agricultural management on nitrous oxide emissions using flux measurements and the DNDC model. Agric. Ecosyst. Environ. 206, 71–83. Van Groenigen, K.J., Van Kessel, C., Hungate, B.A., 2013. Increased greenhouse-gas intensity of rice production under future atmospheric conditions. Nat. Clim. Change 3, 288–291.
57
Vitousek, P.M., Naylor, R., Crews, T., David, M., Drinkwater, L., Holland, E., Johnes, P., Katzenberger, J., Martinelli, L., Matson, P., 2009. Nutrient imbalances in agricultural development. Science 324, 1519. Vlek, P.L., Byrnes, B.H., 1986. The efficacy and loss of fertilizer N in lowland rice. Fertil. Res. 9, 131–147. Wang, F., Ma, W., Dou, Z., Ma, L., Liu, X., Xu, J., Zhang, F., 2006. The estimation of the production amount of animal manure and its environmental effect in China. China Environ. Sci. 26, 614–617. Xing, G., Zhu, Z., 2000. An assessment of N loss from agricultural fields to the environment in China. Nutr. Cycl. Agroecosyst. 57, 67–73. Xiong, Z., Xing, G., Zhu, Z., 2007. Nitrous oxide and methane emissions as affected by water, soil and nitrogen. Pedosphere 17, 146–155. Yan, X., Akiyama, H., Yagi, K., Akimoto, H., 2009. Global estimations of the inventory and mitigation potential of methane emissions from rice cultivation conducted using the 2006 Intergovernmental Panel on Climate Change Guidelines. Global Biogeochem. Cycles 23 GB2002. Zhang, F., Chen, X., Vitousek, P., 2013. Chinese agriculture: an experiment for the world. Nature 497, 33–35. Zhang, W.L., Tian, Z.X., Zhang, N., Li, X.Q., 1996. Nitrate pollution of groundwater in northern China. Agric. Ecosyst. Environ. 59, 223–231. Zhao, X., Zhou, Y., Min, J., Wang, S., Shi, W., Xing, G., 2012. Nitrogen runoff dominates water nitrogen pollution from rice–wheat rotation in the Taihu Lake region of China. Agric. Ecosyst. Environ. 156, 1–11. Zhu, Z., Cai, G., Simpson, J., Zhang, S., Chen, D., Jackson, A., Freney, J., 1988. Processes of nitrogen loss from fertilizers applied to flooded rice fields on a calcareous soil in north-central China. Fertil. Res. 18, 101–115. Zhu, Z., Chen, D., 2002. Nitrogen fertilizer use in China—contributions to food production, impacts on the environment and best management strategies. Nutr. Cycl. Agroecosyst. 63, 117–127.
Contents lists available at ScienceDirect
Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee
Modeling the impacts of water and fertilizer management on the ecosystem service of rice rotated cropping systems in China Han Chena , Chaoqing Yua,* , Changsheng Lib , Qinchuan Xinc,a , Xiao Huanga , Jie Zhanga , Yali Yuea , Guorui Huanga , Xuecao Lia , Wei Wanga a b c
Center for Earth System Science, Key Lab of Earth System Numerical Simulation, Tsinghua University, China Institute for the Study of Earth, Oceans, and Space, University of NH, USA Department of Geography and Planning, Key Lab of Urbanization and Geo-simulation, Sun Yat-sen University, China
A R T I C L E I N F O
A B S T R A C T
Article history: Received 6 May 2015 Received in revised form 15 November 2015 Accepted 26 November 2015 Available online xxx
Detailed information on the impacts of water use and nutrient application on agro-ecosystem services including crop yields, greenhouse gas (GHG) emissions and nitrogen (N) loss is the key to guide field managements. In this study, we use the DeNitrification–DeComposition (DNDC) model to simulate the biogeochemical processes for rice rotated cropping systems in China. We set varied scenarios of water use in more than 1600 counties, and derived optimal rates of N application for each county in accordance to water use scenarios. Our results suggest that 0.88 0.33 Tg per year (mean standard deviation) of synthetic N could be reduced without reducing rice yields, which accounts for 15.7 5.9% of the N application in China in 2005. Field managements with shallow flooding and optimal N applications could enhance ecosystem services at a national scale, leading to 34.3% reduction of GHG emissions (CH4, N2O, and CO2), 2.8% reduction of overall N loss (NH3 volatilization, denitrification and N leaching) and 1.7% increase of rice yields, as compared to current management conditions. Among provinces with major rice production, Jiangsu, Yunnan, Guizhou, and Hubei could achieve more than 40% reduction of GHG emissions under appropriate water managements, while Zhejiang, Guangdong, and Fujian could reduce more than 30% N loss with optimal N applications. Our modeling efforts suggest that China is likely to benefit from reforming water and fertilization managements for rice rotated cropping systems in terms of sustainable crop yields, GHG emission mitigation and N loss reduction, and the reformation should be prioritized in the above-mentioned provinces. ã 2015 Elsevier B.V. All rights reserved.
Keywords: Water regime Nitrogen fertilization Sustainable management Ecological modeling DNDC
1. Introduction Rice is one of the major commodity crops and feeds more than half of the world’s population (Lobell et al., 2011). Being the largest rice producing country in the world, China contributes 27.3% of global rice production on 18.3% of global paddy fields (FAO, 2013). It is estimated that 20% more rice needs to be produced in China in the coming two decades to meet food demands as population increases (Peng et al., 2009). Meanwhile, rice cultivation has resulted in the degradation of a wide range of agro-ecosystem services in China. For example, China emits approximately 7.41 Tg CH4 per year, one of the primary non-CO2 greenhouse gases (GHG),
* Corresponding author at: Tsinghua University, Mengminwei South Building, Room 916, Beijing 100084, China. E-mail address: [email protected] (C. Yu). http://dx.doi.org/10.1016/j.agee.2015.11.023 0167-8809/ ã 2015 Elsevier B.V. All rights reserved.
to the atmosphere due to rice planting alone (Yan et al., 2009). Increasing rice cultivation activities likely lead to higher methane emissions in China under environmental conditions with rising temperature and increasing atmospheric CO2 concentrations (Van Groenigen et al., 2013). In addition, because rice needs to grow in flooded conditions, rice planting typically involves considerable water consumption and nitrogen (N) loss to the environment (Vlek and Byrnes, 1986), which has been found to aggravate water degradation and nutrient pollution in China (Deng et al., 2006; Zhao et al., 2012). Given environmental issues associated with rice planting, there is a need to explore possible solutions that mitigate conflicts between increased food demands and degraded agroecosystem services for China (Miao et al., 2010). Suitable managements of irrigation and fertilization have shown to be capable of enhancing ecosystem services of rice rotated cropping systems by increasing agricultural yields and minimizing environmental pollutions (Burney et al., 2010; Mueller
50
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
et al., 2012). For water schemes, switching continuous flooding to one or several drainages in paddy fields could promote rice yields slightly and reduce water use (Linquist et al., 2015). Intermittent irrigation could also reduce total GHG emissions from rice planting because the overall radiative forcing of CH4 reduction suppresses that of N2O stimulation (Xiong et al., 2007). For fertilizer management practices, N fertilization have found to strongly influence rice yields and N loss through biotic and abiotic processes such as ammonia volatilization, denitrification and nitrogen leaching (Zhang et al., 1996; Ramankutty and Foley, 1999). To derive optimal rates of N application that associated with maximum grain yields and minimum N loss, approaches have been developed based on (1) soil mineral N tests that identify the differences between predicted and measured soil mineral N (NO3-N and NH4+-N) supply (Chen et al., 2006), and (2) regional evaluation method that use numerous field experiments to quantify yield efficiency, net income, and environmental impacts (Zhu and Chen, 2002). Moreover, as found in field experiments and literature reviews, yield-scaled GHG emissions (i.e., GHG emissions per unit rice yield) tend to have minimized values at the rates of N application that optimize rice yields (Feng et al., 2013). Water managements may influence nitrogen use efficiency through various N pathways like denitrification, leaching and ammonia volatilization thus the identification of optimal N rates to achieve the best yield (Sun et al., 2012; Liu et al., 2013). These efforts clearly indicate that strategies of water and N uses hold promise to improve grain yields while reducing GHG emissions and N loss from rice fields, and the water and N managements should be considered simultaneously. Quantitative evaluations of managements on agro-ecosystem services of rice rotated cropping systems are critical to understand potential environmental benefits from managements, but evaluations over large areas like mainland China are challenging for some reasons. First, conflicting conclusions might be drawn from previous studies when performing metaanalysis, largely owing to incomplete records of field management schemes. For example, the response of CH4 emissions to N inputs varies with water regimes, while activities like drainage frequency and strength are not well documented among sites (Pittelkow et al., 2014). Second, it can be challenging to capture optimal N rates based on in-situ experiments (Cassman et al., 1998). Third, scaling up from individual sites to large geographical regions is difficult due to spatial variation of climate environment, soil conditions, and management practices (Smith et al., 2010). Thus, regional mitigation potential of GHG emissions and N loss with yield constraints through optimal managements remains highly uncertain. Lastly, physiological models that describe underlying mechanisms of biogeochemical cycles in paddy fields hold great promise to address the issues associated with the meta-analysis method or field measurements. However, due to incomplete modeling strategies and limited data access in China, most of modeling studies have not evaluated the integrated impacts of water and N fertilizer managements on rice rotated cropping systems (Li et al., 2005; Cheng et al., 2013). Compounding these concerns, the objectives of this study are to (1) quantify national GHG emissions, rice yields, and N loss to the environment from rice rotated cropping systems under varied management scenarios of water and N uses; and (2) identify optimal management combinations that potentially enhance ecosystem services of rice rotated cropping systems in China. To meet these objectives, we applied a biogeochemical model of DeNitrification–DeComposition (DNDC) with county-scale agricultural database that contains climate, soil, vegetation, and management information in China.
2. Materials and methods 2.1. Model descriptions DNDC is a process-based biogeochemical model that simulates carbon dynamics and trace gas emissions for agro-ecosystems (Li et al., 1992). The DNDC model consists of three primary components that model crop growth, soil environment, and soil microbe processes. The crop growth sub-model tracks phenological development, biomass accumulation and allocation, demand and uptake of water and N, respiration and litter production. The soil environment sub-model simulates soil temperature, soil moisture, soil redox potential (i.e., Eh), soil physical properties, and soil oxygen status. The sub-model of soil microbe processes simulates soil biogeochemistry reactions such as decomposition, nitrification, denitrification and fermentation, in which processes trace gases (such as N2O, CH4, NO, and NH3) are produced. Agricultural management practices, including fertilization, tillage, irrigation, manure amendment and grazing, are considered in DNDC to account for various biogeochemical processes. Though DNDC has considered the impacts of farmland water conservancy facilities together with N contents in flooded water and calculated overflow on N runoff, detailed descriptions of these parameters are challenging to achieve over large regions thus set to constant inside DNDC, which would produce uncertainties in specific regions. Therefore, modeling results of N runoff were not included in this study. DNDC has been widely used to simulate greenhouse gas emission, crop growth, ecosystem carbon dynamics, soil water balance, and vegetation biochemical cycles (Fumoto et al., 2008; Deng et al., 2011; Uzoma et al., 2015). 2.2. Model revision to account for rice transplant Different from other places in the world, rice cultivation in China often involves practices of transplantation. Rice seedlings are typically transplanted from a seedling bed to paddy fields right after the harvests of previous crops. Transplantation saves growing periods to meet requirements of temperature accumulation, thus it is important for double- and triple-cropping systems in China. Current version of DNDC does not include a function of rice transplantation, such that direct modeling of rice growth and yields tends to be biased for rotated cropping systems. Here, we implement a new scheme and modify DNDC to include parameters that define dates of rice transplantation. The schematic descriptions of crop growth sub-models are shown in Fig. S.1 (see supporting information Appendix A) for comparisons between current and revised versions of DNDC. 2.3. National database construction To run the DNDC model for more than 1600 counties with rice cultivation in China, we constructed county-level databases that contain soil, weather and management data, including: (1) daily meteorological data with 2-m air temperature ( C) and precipitation (mm); (2) soil parameters with soil organic matter content (kg C kg1 soil), clay fraction (0–1), pH, and bulk density (g cm3); (3) Crop parameters with maximum grain yield (kg C ha1), effective accumulated temperature ( C), crop water demand (g water g1 dry matter), and annual rice yield increase index (%); (4) crop system information with rotation types and areas; (5) field management practices that involve sowing/harvest dates, N fertilizer amounts (kg N ha1), irrigation ratios within a county (0–1), tillage dates and methods, manure amounts (kg N ha1), crop straw return proportions (0–1), flooding methods, and transplant dates (if rice rotated cropping system). Soil parameters
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
are expressed with upper and lower limits rather than absolute values to indicate the range of soil inputs for each county. Ten rice-rotated systems are considered in model simulations: rice, rice–rice, rice–wheat, rice–rapeseed, rice–rice–vegetable, rice–vegetable, oat–rice, rice–soybean, rice–rice–rapeseed and rice–rice–wheat. The areas of rotation systems in each county were derived by combining the county-scale statistical database of cropsown areas with satellite-based land-cover map for mainland China (Frolking et al., 2002). Meteorological measurements from 620 weather stations were obtained from China Meteorological Data Sharing Service System (http://cdc.cma.gov.cn/home.do). Stationary meteorological datasets from 1996 to 2010 were assigned to the nearest counties. County-level soil parameters were derived from digitized national soil maps [Institute of Soil science, 1986] and other information [National Soil Survey Office of China, 1997]. Soil Mn, Fe, and sulfate contents were set as national average values for paddy soils, where Mn = 30 mg kg1 soil, Fe = 80 mg kg1 soil, and sulfate = 220 mg kg1 soil. Information on agricultural management practices was obtained from Crop Cultivation Bureau of China (http://www.zzys.moa.gov.cn/; in Chinese). The type of N fertilizer in paddy fields was mostly urea and top-dressed into floodwater across China (Peng et al., 2009). The N application rate (kg N ha1) was derived for each province using Huang's method (Huang and Tang, 2010) with statistical data from National Bureau of Statistics of China (http://www.stats.gov. cn/english/). The manure amendment rates were calculated based on province-level human population and animal livestock (Wang et al., 2006). Ministry of Agriculture of China reported that 31.30% of crop residues were left in the field in 2009, of which the percentage value was used for all paddy fields. The conventional tillage and mid-season drainage were applied across China using default settings in DNDC. 2.4. Model calibration and evaluation Simulation of rice growth is essential to the predictions of GHG emission and N loss due to the substrate supply/uptake and gas transport. Given model parameters in DNDC, it is computationally impractical to search the entire value spaces for the best parameter sets for each county in China. Among a variety of parameters, four parameters were found to be the most sensitive ones that influence crop growth, namely effective accumulated temperature ( C), crop water demand (g water g1 dry matter), maximum grain yield (kg C ha1), and annual changing percentage of maximum grain yield (%). Effective accumulated temperature describes temperature requirement for crop growth; crop water demand stands for
51
crop water productivity, maximum grain yield describes potential grain production without stresses from water, nutrient, and temperature; and annual changing percentage of maximum grain yield denotes the changing rate of maximum grain yield due to technology advancement. To find suitable parameter values for each county, we employ the Shuffled Complex Evolution Approach (SCEA) to speed up model calibration processes (Duan et al., 1993). The root-meansquare deviation serves as the cost function in SCEA. County-level rice yields from 2000 to 2008 in China were used as reference data for calibration, whereas yield data from 1995 to 1999 and 2009 were used for model validation. Statistical metrics such as coefficient of determination (R2) and root mean square error (RMSE) were used for model evaluation, and RMSE was calculated as follows: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u n uX 2 u X obs;i X model;i u t i¼1 RMSE ¼ ð1Þ n where X obs;i and X model;i represent observed and modeled values, respectively; n is the total number of samples. 2.5. Scenario analysis Using the calibrated DNDC model, we conducted analysis for varied scenarios of water and N uses. For water use scenarios, we evaluated three widely used irrigation schemes of continuous flooding, mid-season drainage and shallow flooding. In the scheme of continuous flooding, paddy fields are continuously flooded with a 5–10 cm depth of surface water during rice growing period; in the scenario of mid-season drainage, paddy fields experience two drainages and each last for one week at late tillering stage and restore continuous flooding for the rest; and in the scheme of shallow flooding, the water table of rice paddy is constantly fluctuating 5–10 cm above and below the soil surface (Li et al., 2006). For scenarios of N fertilizer uses, we compared results from three application schemes of zero N inputs, optimal rates of N inputs, and actual rates of N inputs. The zero N inputs denote the scheme without using any N fertilizer, the actual rates of N inputs denote the scenarios that N fertilizer are used in field managements in reality, and the optimal rates of N inputs denote the scheme that maximum rice yields can be achieved with minimum N uses. The optimal rates of N inputs are derived for each corresponding scheme of water use and for each county. We varied
Fig. 1. Comparisons of measured and simulated county-level rice yields are shown: (A) model calibration using data in 2000–2008; (B) model validation using data in 1995– 1999; (C) model validation using data in 2009.
52
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
N inputs from zero to maximum values and identified the first point, after which rice yields never increase. This point was then considered as the optimal N rate for the corresponding county. Given three water use scenarios and three N use scenarios, we compared results from nine (3 3) scenarios to assess the impacts of water and fertilizer management interactions on rice yields, GHG emissions, and N losses. To allow direct comparisons, we set a baseline using scenarios of mid-season drainage water use and actual rates of N use for the year of 2010. Based on meteorological datasets, the year of 2005 was identified as a normal year in term of annual average temperature and precipitation. We used climate data in 2005 as baselines and performed a twenty-year simulation for each county, in order to (1) initialize soil background values and relative sizes of each pool in DNDC, and (2) avoid the impacts of inter-annual meteorological variance on the optimal N rates calculation and scenario analysis. The GHG emissions are derived for three key components of CO2, N2O, and CH4 as follows: GHG = [CO2] + 298 [N2O] + 25 [CH4]
(2)
where [CO2], [N2O] and [CH4] are CO2 flux (kg CO2 ha1 yr1), N2O flux (kg N2O ha1 yr1) and CH4 flux (kg CH4 ha1 yr1), and they are converted as [CO2] = C 44/12, [N2O] = N 44/28, [CH4] = C 16/12. C and N are fluxes in carbon (kg C ha1 yr1) and nitrogen (kg N ha1 yr1) units, respectively. 298 and 25 represents the global warming potential in kg CO2 equivalent ha1 yr1 of N2O and CH4 over a 100-yr horizon (Houghton et al., 2001). The yield-scaled GHG emissions are derived as follows: Yield scaled GHG ¼
GHG yield
ð3Þ
To address uncertainties of CH4 and N2O emissions due to varied soil conditions within each county, we applied the approach of Most Sensitive Factor (MSF) (Li et al., 2004). Sensitivity tests
were conducted and soil texture and soil organic matter content (SOC) were identified as two of the most sensitive parameters that influence CH4 and N2O emissions from paddy fields. Then we varied these factors within the range from county-level database, and performed DNDC twice to obtain the range of CH4 and N2O emissions: first with low SOC, low pH, and high clay content, and second with high SOC, high pH, and low clay content to most represent variations of actual county-level fluxes. 3. Results 3.1. Model calibration and validation Fig. 1 shows the results between observed and modeled rice yields for more than 1600 counties with rice cultivation. Overall, our model could explain 84%, 80% and 79% variance of measured rice yield for model calibration periods in 2000–2008 and model validation periods in 1995–1999 and in 2009, respectively. The RMSE were 763.86, 887.11 and 881.10 kg ha1 for calibration periods and two validation periods, respectively. The modeled results might not capture the observations of low yields well for some counties in specific years, which are likely due to disturbance factors such as yield damages caused by insects, typhoon, or floods. Given the large extent of our study area and the long time series of our study period, our model performs well in general. In addition to rice yields, observed and modeled GHG and N loss are shown in Fig. 2 for six field sites across China. The six validation sites are distributed in main rice production regions from Northeast to South (see Fig. S.2 & Table S.1 in supporting information Appendix A). DNDC could simulate the impact of mid-season drainages on N2O stimulation (Fig. 2A). Modeled results in Fig. 2C–E indicate that DNDC is able to trace N leaching and NH3 emission dynamics under nitrogen fertilizer applications. The influence of rice growth stages on CH4 emission rate was
Fig. 2. Observed and modeled time series in paddy fields are shown for (A) N2O emission for Wangcheng county in Hunan province; (B) CH4 emission for Beibei county in Chongqing municipal district; (C) NH3 emission for Wuming county in Guangxi province; (D) N leaching for Qingpu county in Shanghai municipal district; (E) NH3 emission for Kunshan county in Jiangsu province; and (F) CH4 emission for Sanjiang plain in Heilongjiang province.
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
53
Fig. 3. The spatial distribution of estimated ratios of N over-application are shown for water use scenarios of (A) CF denotes continuous flooding, (B) MD denotes mid-season drainage, and (C) SF denotes shallow flooding.
shown in Fig. 2B and F. These results suggest that DNDC could capture the phases and magnitudes of the processes of CH4 and N2O emission, ammonia volatilization and nitrogen leaching for paddy fields under a variety of management practices. Fig. S.3 (see supporting information Appendix A) illustrates how optimal N rates are derived for various rice cropping systems across China. Regional variations of optimal N rates may be mainly caused by the discrepancies of environmental N supplies, soil properties and climate conditions. The maximum yields and minimum yield-scaled GHG emissions occurred at optimal N rates, which agree with previous works (Pittelkow et al., 2014). 3.2. application, N loss, and rice yields According to the yearbook of 2005, the cultivated areas of paddy fields were 27.6 million ha in 2005, average rice productivity was 6279.4 kg ha1, and approximate 5.6 Tg chemical N fertilizer (93.6 kg N ha1) were applied on these lands. Nationwide areaweighed optimal N rates of paddy fields calculated with DNDC are 78.9 4.5 kg N ha1 (mean standard deviation as derived based on the variability of three water management practices; same hereafter unless notified). National yields of rice under optimal N rates are estimated to be 6411 5.6 kg ha1, which increases slightly as compared to current yields. The increased rice yields are largely due to yield enhancement in regions with under-applied mineral N. There are no significant biases in modeled rice yields at the provincial level, as indicated by student's t-test under 95% confidence intervals for all water regimes. This indicates 15.7 5.9% reduction of N inputs can be achieved without rice yield reductions.
Nationwide optimal N rates under mid-season drainage (MD; 80.2 kg N ha1) and shallow flooding (SF; 83.7 kg N ha1) water scenarios were higher than that under the scenario of continuous flooding (CF; 72.9 kg N ha1). Water management practices of MD and SF lead to more N2O emission and relatively lower nitrogen use efficiency (Siebert and Döll, 2010). In the CF scenario, approximately 20% of paddy fields in China are over-applied with more than 40% mineral nitrogen. The area ratios fall to 17% and 14% under MD and SF water regimes, respectively (see Fig. S.4 in supporting information Appendix A). At a national scale, the amounts of overused N fertilizers are estimated to range from 0.60 Tg yr1 (SF) to 1.25 Tg yr1 (CF). The estimated amount of over-applied mineral N was highly uneven across the country (Fig. 3). Given SF water regime scenario, paddy fields with mineral N over-application were mainly distributed in southern China (such as Guangdong and Hunan provinces) and eastern China (such as Fujian and Zhejiang provinces), where more than 31% excessive mineral N were applied than optimal N rates. Paddy fields with under-applied N appeared in Anhui, Yunnan, Guizhou, and Hubei, where more than 20% N fertilizer inputs were needed to fulfill rice growth. As shown in Table 1, N losses from the whole rice rotation systems increase with increased N inputs. Surplus N in the rice season can stimulate further NH3 emission, denitrification and N leaching in other rotated crop seasons. For the whole rice rotated cropping systems, compared to other pathways, ammonia volatilization represents a primary contributor to total N loss and accounts for 25.5–30.9% of total N fertilizer and manure inputs on a national scale. This ratio were higher than that of 9–18% for agricultural lands in 1990 (Xing and Zhu, 2000). Several reasons
Table 1 Pathways of N inputs and N loss under various N fertilizer application rates and water management practices at a national scale. Scenarioa
CF_N_0 CF_N_Opt CF_N_Act MD_N_0 MD_N_Opt MD_N_Act SF_N_0 SF_N_Opt SF_N_Act a
N inputs (kg N/ha/yr)
N loss for rice growth season (kg N/ha/yr)
N loss for rice rotated systems (kg N/ha/yr)
N rate
Manure
NH3 volatilization
Leaching
Denitrification
NH3 volatilization
Leaching
Denitrification
80.47 157.68 174.21 80.47 165.43 174.21 80.47 169.01 174.21
10.48 10.48 10.48 10.48 10.48 10.48 10.48 10.48 10.48
0.66 13.69 15.83 0.75 18.70 17.96 0.80 26.43 22.69
0.74 4.57 6.74 0.58 3.56 4.26 0.58 4.05 5.03
18.95 27.18 32.19 20.70 36.72 41.33 17.24 31.51 36.36
31.22 44.63 47.17 29.85 48.01 47.50 29.68 55.53 51.93
2.62 6.91 9.28 2.59 5.88 6.67 2.72 6.35 7.41
22.90 31.70 37.44 25.07 41.69 46.89 28.95 43.78 49.01
N_0 denotes zero N fertilizer inputs, N_Opt denotes optimal N application rates derived using DNDC, and N_Act denotes actual nitrogen fertilizer application rates.
54
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
Fig. 4. GHG emissions, yields and yield-scaled GHG emissions of nine scenarios. The error bar indicates the uncertainties of GHG emissions due to spatial heterogeneity of input soil data.
contribute to this overestimate: Firstly, the past two decades has seen the gradually increase of N fertilizer inputs and decrease in N productivity of agricultural soil in China (Zhang et al., 2013), which would induce more NH3 emission in 2005 compared to that in 1990; Secondly, uncertainties exist in Xing et al., 2000’ method to estimate NH3 emission based on the emission factors of different nitrogen fertilizers over regions (Xing and Zhu, 2000). Lastly, extra emission of NH3 during the fallow period is not included in the previous study. For the rice growing season when comparing our results with previous studies, our estimates of NH3 emissions (3.9– 12.5%) are consistent with the observations of 4.4–16.7% in Taihu region (Lin et al., 2007). Meanwhile, our estimation indicates denitrification loss could account for 34.3–45.8% of observed applied N, which agree with that (33%) in Henan province (Zhu et al., 1988) and that (41%) during rice growth period (Cai et al., 1991). N leaching from paddy fields appears to be minor as compared to the other two pathways and is only responsible for less than 5% of total N application in our simulation, which is in accord with previous findings (Ju et al., 2009). At the national scale, reducing N application rates toward the optimal rates could decrease overall N loss to the environment by 6.4 3.7%. Although benefiting from water conservation and GHG emissions mitigation, changing water management from CF to intermittent drainages tends to induce higher N loss (7.6% higher from CF to MD or 15.4% higher from CF to SF). The national average GHG emissions, rice yields and yieldscaled GHG emissions of nine scenarios were calculated and shown in Fig. 4. The annual average GHG flux from paddy fields in China was 8609.4 2659.1 kg CO2 equivalent ha1 and the national total amount of GHG emissions was 237.6 73.4 Tg CO2 equivalent under current N rates. The nationwide flux of yield-scaled GHG emissions from rice fields was 1.37 0.51 kg
CO2-equivalents kg1-grain yield yr1, which fell within previous estimates (3.22 kg CO2-equivalent kg1-grain yield yr1) in continuous flooding rice paddies (Li et al., 2006) and estimates (0.24–0.74 kg CO2-equivalent kg1-grain yield yr1) in rice paddies with midseason drainage and organic manure incorporation (Qin et al., 2010). At a nationwide scale, changing water regimes from CF to MD and SF could lead to 32.3% and 53.6% reduction of GHG emissions, respectively. In comparisons, changing observed N rates to optimal rates could reduce national GHG emissions by 2.2 1.7% (Fig. 4), mostly due to reductions in N2O emission rather than CO2 and CH4 emissions (Table 2). Meanwhile, yield-scaled GHG emissions are reduced from 1.37 0.51 to 1.32 0.52 kg CO2-equivalent kg1grain yield yr1 (or 3% reduction in equivalent). Among the nine scenarios, yield-scaled GHG emissions are minimized under SF water regime with optimal N rates. Compared to the reference scenario (mid-season drainage with actual N rates), changing water management practices from MD to SF contributes to 31.5% reduction of yield-scaled GHG emissions and further 4.2% decreases are achieved through reducing current N inputs to optimal rates. The error bars in Fig. 4 indicated the impacts of soil heterogeneity on yields, GHG emissions and yield-scaled GHG emissions. The variation coefficients of GHG emissions induced by soil heterogeneity are 0.34, 0.01 and 0.29 for CF, MD and SF water regime, respectively, which implies that the uncertainties of estimated GHG emissions from soil inputs minimized under midseason drainage. Similar results are obtained with respect to yieldscaled GHG emissions (0.37, 0.02 and 0.29 for CF, MD and SF water regime). As shown in Fig. 5, among main rice production provinces in China (provinces that together contribute more than 90% of rice
Table 2 Summary of CH4, N2O and CO2 emissions (units: CO2 equivalent ha1 yr1) under varied managements of water and nitrogen fertilizer.
CO2 N2O CH4
CF_N_0
CF_N_Opt
CF_N_Act
MD_N_0
MD_N_Opt
MD_N_Act
SF_N_0
SF_N_Opt
SF_N_Act
1994 445 8113
888 672 10544
919 1107 10145
2075 586 4490
1036 1206 5633
1071 1587 5484
1866 815 2389
880 1431 2917
942 1726 2855
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
55
Fig. 5. Estimated GHG emissions and yield-scaled GHG emissions of different provinces: (a) under shallow flooding with optimal application rates of mineral N; (b) mitigation potential of water and nitrogen managements. The error bars are standard deviations derived from the low, mean and high estimates of GHG emissions due to uncertainties of soil inputs. MD_Obs_N, SF_Obs_N and SF_Opt_N represent mid-season drainage + current application rates of mineral N, shallow flooding+ current application rates of mineral N and shallow flooding + optimal application rates of mineral N. The red line indicates rice cultivated areas of different provinces. Abbreviations of province: AH, Anhui; BJ, Beijing; FJ, Fujian; GD, Guangdong; GS, Gansu; GX, Guangxi; GZ, Guizhou; HEB, Hebei; HEN, Henan; HLJ, Heilongjiang; HN, Hainan; HUB, Hubei; HUN, Hunan; IM, Inner Mongolia; JL, Jilin; JS, Jiangsu; JX, Jiangxi; LN, Liaoning; NX, Ningxia; QH, Qinghai; SAX, Shaanxi; SC, Sichuan; SD, Shandong; SH, Shanghai; SX, Shanxi; TB, Tibet; TJ, Tianjin; XJ, Xinjiang; YN,Yunnan; ZJ, Zhejiang.
cultivated areas in China), the highest GHG emissions (>5700 CO2 equivalent ha1 yr1) occurred in Heilongjiang (high SOC), Zhejiang (multiple rotation cropping systems), Anhui (multiple cropping systems), and Yunnan (high manure inputs). The lowest (<2000 CO2 equivalent ha1 yr1) GHG emissions occurred in Guizhou (high clay fraction), Jiangxi (low manure inputs), and Jiangsu (less multiple cropping systems). When changing water regimes from MD to SF, the highest reduction of GHG emissions occurred in Jiangsu, Yunnan, Guizhou and Hubei (more than 40% reduction) with high SOC, high multiple crop index and low clay fraction. The highest reduction of induced GHG emissions due to optimal N uses could occur in Zhejiang, Guangdong, and Fujian, where serious over-application of mineral N exists.
uncertainties of N2O emission are likely due to climate and soil variance, inconsistent management practices and limited observation datasets among these studies (Akiyama et al., 2005). Reduced N inputs to optimal rates suppressed N2O emission by an average of 26.8%. This change stimulates CH4 emission slightly because CH4 transportation rates through plants are also enhanced with increased rice growth under optimal N application rates (Sass et al., 1990). CO2 emissions are less sensitive to application rates of N fertilizer, but are more sensitive to water management practices and soil properties because rates of both decomposition and crop residue incorporation could be affected by water regimes and soil properties (Li et al., 2005). In sum, nation-average GHG emissions in rice cropping systems in China are minimized under SF water regime with optimal N application rates.
4. Discussion 4.2. Uncertainties in deriving optimal N rates 4.1. The impacts of water and nitrogen management on three GHG emissions In our study, national averaged methane emissions range from 71.7–316.3 CH4–C ha1 yr1 in the scenario of continuous flooding (Table 2). Our estimates fall within observed ranges of CH4 fluxes (9 to 725 kg C ha1 yr1) (Cai et al., 2003) and are close to previous estimates of 90–214 kg C ha1 yr1 (Frolking et al., 2004). CH4 could account for fewer GHG emissions with increased intermittent drainages (82%, 67% and 52% for CF, MD and SF, respectively). National average N2O emission is lower (2.36 kg N ha1 yr1) in CF water regime than in MD and SF scenarios (3.39 and 3.69 kg N ha1 yr1, respectively). Frolking et al. (2004) estimated a higher emission rate of N2O (5.7–58 kg ha1 yr1) over the 100-year simulations for all water regimes. Akiyama et al. (2005) estimated a lower emission flux of 1.82 kg N ha1 yr1 by compiling and analyzing available data from rice paddy fields in Asia. Large
Discrepancies in deriving optimal rates of N application exist when comparing our result of 23.6% reduction for paddy fields in Taihu region to that of 33% reduction using knowledge-based optimum N fertilization (Zhu and Chen, 2002) and that of 15–25% reduction with soil mineral N tests (Hofmeier et al., 2015). Similarly, studies on a rice–wheat rotated cropping system in southwest China have demonstrated 18% reduction with an integrated nutrient management strategy (Fan et al., 2007). The differences in optimal rates of N application are likely due to several factors: first, there are scaling issues among different studies. Here, we uses N fertilizer rates and manure applications at a provincial level, while previous studies obtain application rates of synthetic N fertilizers and manure through field surveys (Fan et al., 2007). Second, water management strategies also involve the frequency and strength of drainages, but related information still lacks in detail (Ju et al., 2009). The missed
56
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57
information in our simulation could influence modeling of the processes of denitrification (Cai et al., 1997) and ammonia volatilization (Li et al., 2008). Third, we have not included enhanced N management strategies such as deep placement of fertilizer, slow-release N fertilizers, nitrification, and urease inhibitors (Vitousek et al., 2009; Giltrap et al., 2010), nor we calculated optimal N through matching crop requirements at varied growth stages based on environment N supplies and estimated crop N demands (Qiu et al., 2009). These factors are challenging issues in large-scale modeling efforts and are worthy further explorations. 5. Conclusions Improving food productions while reducing environmental risks have become one of the predominant challenges that contemporary agriculture face nowadays (Godfray et al., 2010). To understand how managements of water and nitrogen fertilizer influence ecosystem services of rice cropping systems in China, we calibrated and validated DNDC models and compared results for nine contrasting scenarios. We estimate 0.88 0.33 Tg of synthetic N per year was overapplied into paddy fields in China. Shallow flooding under optimal rates of N use show promises to enhance agro-ecosystem services, which increases rice yields by 1.7% and reduces N loss and GHG emissions by 2.8% and 34.3%, respectively. The priority to reform water and fertilizer managements should be given to the provinces with heavy over-application of N fertilizer like Zhejiang, Guangdong, and Fujian, and to the provinces with large GHG mitigation potential like Jiangsu, Yunnan, Guizhou, and Hubei. Future researches should be conducted to address uncertainties associated with county-level nitrogen inputs and scaling factors. Acknowledgements We gratefully acknowledge the support of the China 973 project 2013CB956600, CNSF#41371491, CNSF#41401484, Tsinghua20121088052, SEPA201309062 and Tsinghua Fudaoyuan Research Fund. Thanks also go to the anonymous reviewers for their constructive comments. In memory of Prof. Changsheng Li, we especially thank him for his enthusiasm and dedication to the research and education. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.agee.2015.11.023. References Akiyama, H., Yagi, K., Yan, X., 2005. Direct N2O emissions from rice paddy fields: summary of available data. Global Biogeochem. Cycles 19, GB1005. Burney, J.A., Davis, S.J., Lobell, D.B., 2010. Greenhouse gas mitigation by agricultural intensification. Proc. Natl. Acad. Sci. 107, 12052–12057. Cai, G., Cao, Y., Yang, N., Lu, Y., Zhuang, L., Wang, X., Zhu, Z., 1991. Direct estimation of nitrogen gases emitted from flooded soils during denitrification of applied nitrogen. Pedosphere 1, 241–251. Cai, Z., Sawamoto, T., Li, C., Kang, G., Boonjawat, J., Mosier, A., Wassmann, R., Tsuruta, H., 2003. Field validation of the DNDC model for greenhouse gas emissions in East Asian cropping systems. Global Biogeochem. Cycles 17 GB1107. Cai, Z., Xing, G., Yan, X., Xu, H., Tsuruta, H., Yagi, K., Minami, K., 1997. Methane and nitrous oxide emissions from rice paddy fields as affected by nitrogen fertilisers and water management. Plant Soil 196, 7–14. Cassman, K., Peng, S., Olk, D., Ladha, J., Reichardt, W., Dobermann, A., Singh, U., 1998. Opportunities for increased nitrogen-use efficiency from improved resource management in irrigated rice systems. Field Crops Res. 56, 7–39. Chen, X., Zhang, F., Römheld, V., Horlacher, D., Schulz, R., Böning Zikens, M., Wang, P., Claupein, W., 2006. Synchronizing N supply from soil and fertilizer and N demand of winter wheat by an improved Nmin method. Nutr. Cycl. Agroecosyst. 74, 91–98.
Cheng, K., Ogle, S.M., Parton, W.J., Pan, G., 2013. Simulating greenhouse gas mitigation potentials for Chinese Croplands using the DAYCENT ecosystem model. Global Change Biol. 20, 948–962. Deng, J., Zhou, Z., Zhu, B., Zheng, X., Li, C., Wang, X., Jian, Z., 2011. Modeling nitrogen loading in a small watershed in southwest China using a DNDC model with hydrological enhancements. Biogeosciences 8, 2999–3009. Deng, X., Shan, L., Zhang, H., Turner, N.C., 2006. Improving agricultural water use efficiency in arid and semiarid areas of China. Agric. Water Manage. 80, 23–40. Duan, Q., Gupta, V.K., Sorooshian, S., 1993. Shuffled complex evolution approach for effective and efficient global minimization. J. Optim. Theory Appl. 76, 501–521. Fan, M., Lu, S., Jiang, R., Liu, X., Zeng, X., Goulding, K.W., Zhang, F., 2007. Nitrogen input, 15N balance and mineral N dynamics in a rice–wheat rotation in southwest China. Nutr. Cycl. Agroecosyst. 79, 255–265. FAO. 2013. FAO statistical databases. Available at http://faostat3.fao.org/home/E. (verified 19 Nov. 2014). Feng, J., Chen, C., Zhang, Y., Song, Z., Deng, A., Zheng, C., Zhang, W., 2013. Impacts of cropping practices on yield-scaled greenhouse gas emissions from rice fields in China: a meta-analysis. Agric. Ecosyst. Environ. 164, 220–228. Frolking, S., Li, C., Braswell, R., Fuglestvedt, J., 2004. Short- and long-term greenhouse gas and radiative forcing impacts of changing water management in Asian rice paddies. Global Change Biol. 10, 1180–1196. Frolking, S., Qiu, J., Boles, S., Xiao, X., Liu, J., Zhuang, Y., Li, C., Qin, X., 2002. Combining remote sensing and ground census data to develop new maps of the distribution of rice agriculture in China. Global Biogeochem. Cycles 16 1091. Fumoto, T., Kobayashi, K., Li, C., Yagi, K., Hasegawa, T., 2008. Revising a process-based biogeochemistry model (DNDC) to simulate methane emission from rice paddy fields under various residue management and fertilizer regimes. Global Change Biol. 14, 382–402. Giltrap, D.L., Li, C., Saggar, S., 2010. DNDC: a process-based model of greenhouse gas fluxes from agricultural soils. Agric. Ecosyst. Environ. 136, 292–300. Godfray, H.C.J., Beddington, J.R., Crute, I.R., Haddad, L., Lawrence, D., Muir, J.F., Pretty, J., Robinson, S., Thomas, S.M., Toulmin, C., 2010. Food security: the challenge of feeding 9 billion people. Science 327, 812–818. Hofmeier, M., Roelcke, M., Han, Y., Lan, T., Bergmann, H., Böhm, D., Cai, Z., Nieder, R., 2015. Nitrogen management in a rice–wheat system in the Taihu Region: recommendations based on field experiments and surveys. Agric. Ecosyst. Environ. 209, 60–73. Houghton, J.T., Ding, Y., Griggs, D.J., Noguer, M., van der Linden, P.J., Dai, X., Maskell, K., Johnson, C., 2001. Radiative forcing of climate change. Climate Change 2001: The Scientific Basis. Contribution of Working Group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK and New York, USA 881 pp.. Huang, Y., Tang, Y., 2010. An estimate of greenhouse gas (N2O and CO2) mitigation potential under various scenarios of nitrogen use efficiency in Chinese croplands. Global Change Biol. 16, 2958–2970. Ju, X., Xing, G., Chen, X., Zhang, S., Zhang, L., Liu, X., Cui, Z., Yin, B., Christie, P., Zhu, Z., 2009. Reducing environmental risk by improving N management in intensive Chinese agricultural systems. Proc. Natl. Acad. Sci. 106, 3041–3046. Li, C., Frolking, S., Frolking, T.A., 1992. A model of nitrous oxide evolution from soil driven by rainfall events: 1. Model structure and sensitivity. J. Geophys. Res. (1984–2012) 97, 9759–9776. Li, C., Frolking, S., Xiao, X., Moore III, B., Boles, S., Qiu, J., Huang, Y., Salas, W., Sass, R., 2005. Modeling impacts of farming management alternatives on CO2, CH4, and N2O emissions: a case study for water management of rice agriculture of China. Global Biogeochem. Cycles 19, GB3010. Li, C., Mosier, A., Wassmann, R., Cai, Z., Zheng, X., Huang, Y., Tsuruta, H., Boonjawat, J., Lantin, R., 2004. Modeling greenhouse gas emissions from rice-based production systems: sensitivity and upscaling. Global Biogeochem. Cycles 18, GB1043. Li, C., Salas, W., DeAngelo, B., Rose, S., 2006. Assessing alternatives for mitigating net greenhouse gas emissions and increasing yields from rice production in China over the next twenty years. J. Environ. Qual. 35, 1554–1565. Li, H., Liang, X., Chen, Y., Tian, G., Zhang, Z., 2008. Ammonia volatilization from urea in rice fields with zero-drainage water management. Agric. Water Manage. 95, 887–894. Lin, D., Fan, X., Hu, F., Zhao, H., Luo, J., 2007. Ammonia volatilization and nitrogen utilization efficiency in response to urea application in rice fields of the Taihu Lake region, China. Pedosphere 17, 639–645. Linquist, B.A., Anders, M.M., Adviento, B., Maria Arlene, A., Chaney, R.L., Nalley, L.L., da Rosa, E.F.F., van Kessel, C., 2015. Reducing greenhouse gas emissions, water use, and grain arsenic levels in rice systems. Global Change Biol. 21, 407–417. Liu, L., Chen, T., Wang, Z., Zhang, H., Yang, J., Zhang, J., 2013. Combination of sitespecific nitrogen management and alternate wetting and drying irrigation increases grain yield and nitrogen and water use efficiency in super rice. Field Crops Res. 154, 226–235. Lobell, D.B., Schlenker, W., Costa Roberts, J., 2011. Climate trends and global crop production since 1980. Science 333, 616–620. Miao, Y., Stewart, B.A., Zhang, F., 2010. Long-term experiments for sustainable nutrient management in China. A review. Agron. Sustain. Dev. 31, 397–414. Mueller, N.D., Gerber, J.S., Johnston, M., Ray, D.K., Ramankutty, N., Foley, J.A., 2012. Closing yield gaps through nutrient and water management. Nature 490, 254– 257. Peng, S., Tang, Q., Zou, Y., 2009. Current status and challenges of rice production in China. Plant Prod. Sci. 12, 3–8.
H. Chen et al. / Agriculture, Ecosystems and Environment 219 (2016) 49–57 Pittelkow, C.M., Adviento, B., Maria, A., van Kessel, C., Hill, J.E., Linquist, B.A., 2014. Optimizing rice yields while minimizing yield-scaled global warming potential. Global Change Biol. 20, 1382–1393. Qin, Y., Liu, S., Guo, Y., Liu, Q., Zou, J., 2010. Methane and nitrous oxide emissions from organic and conventional rice cropping systems in Southeast China. Biol. Fertil. Soils 46, 825–834. Qiu, J., Li, C., Wang, L., Tang, H., Li, H., Van Ranst, E., 2009. Modeling impacts of carbon sequestration on net greenhouse gas emissions from agricultural soils in China. Global Biogeochem. Cycles 23 GB1007. Ramankutty, N., Foley, J.A., 1999. Estimating historical changes in global land cover: croplands from 1700 to 1992. Global Biogeochem. Cycles 13, 997–1027. Sass, R., Fisher, F., Harcombe, P., Turner, F., 1990. Methane production and emission in a Texas rice field. Global Biogeochem. Cycles 4, 47–68. Siebert, S., Döll, P., 2010. Quantifying blue and green virtual water contents in global crop production as well as potential production losses without irrigation. J. Hydrol. 384, 198–217. Smith, W., Grant, B., Desjardins, R., Worth, D., Li, C., Boles, S., Huffman, E., 2010. A tool to link agricultural activity data with the DNDC model to estimate GHG emission factors in Canada. Agric. Ecosyst. Environ. 136, 301–309. Sun, Y., Ma, J., Sun, Y., Xu, H., Yang, Z., Liu, S., Jia, X., Zheng, H., 2012. The effects of different water and nitrogen managements on yield and nitrogen use efficiency in hybrid rice of China. Field Crops Res. 127, 85–98. Uzoma, K.C., Smith, W., Grant, B., Desjardins, R.L., Gao, X., Hanis, K., Tenuta, M., Goglio, P., Li, C., 2015. Assessing the effects of agricultural management on nitrous oxide emissions using flux measurements and the DNDC model. Agric. Ecosyst. Environ. 206, 71–83. Van Groenigen, K.J., Van Kessel, C., Hungate, B.A., 2013. Increased greenhouse-gas intensity of rice production under future atmospheric conditions. Nat. Clim. Change 3, 288–291.
57
Vitousek, P.M., Naylor, R., Crews, T., David, M., Drinkwater, L., Holland, E., Johnes, P., Katzenberger, J., Martinelli, L., Matson, P., 2009. Nutrient imbalances in agricultural development. Science 324, 1519. Vlek, P.L., Byrnes, B.H., 1986. The efficacy and loss of fertilizer N in lowland rice. Fertil. Res. 9, 131–147. Wang, F., Ma, W., Dou, Z., Ma, L., Liu, X., Xu, J., Zhang, F., 2006. The estimation of the production amount of animal manure and its environmental effect in China. China Environ. Sci. 26, 614–617. Xing, G., Zhu, Z., 2000. An assessment of N loss from agricultural fields to the environment in China. Nutr. Cycl. Agroecosyst. 57, 67–73. Xiong, Z., Xing, G., Zhu, Z., 2007. Nitrous oxide and methane emissions as affected by water, soil and nitrogen. Pedosphere 17, 146–155. Yan, X., Akiyama, H., Yagi, K., Akimoto, H., 2009. Global estimations of the inventory and mitigation potential of methane emissions from rice cultivation conducted using the 2006 Intergovernmental Panel on Climate Change Guidelines. Global Biogeochem. Cycles 23 GB2002. Zhang, F., Chen, X., Vitousek, P., 2013. Chinese agriculture: an experiment for the world. Nature 497, 33–35. Zhang, W.L., Tian, Z.X., Zhang, N., Li, X.Q., 1996. Nitrate pollution of groundwater in northern China. Agric. Ecosyst. Environ. 59, 223–231. Zhao, X., Zhou, Y., Min, J., Wang, S., Shi, W., Xing, G., 2012. Nitrogen runoff dominates water nitrogen pollution from rice–wheat rotation in the Taihu Lake region of China. Agric. Ecosyst. Environ. 156, 1–11. Zhu, Z., Cai, G., Simpson, J., Zhang, S., Chen, D., Jackson, A., Freney, J., 1988. Processes of nitrogen loss from fertilizers applied to flooded rice fields on a calcareous soil in north-central China. Fertil. Res. 18, 101–115. Zhu, Z., Chen, D., 2002. Nitrogen fertilizer use in China—contributions to food production, impacts on the environment and best management strategies. Nutr. Cycl. Agroecosyst. 63, 117–127.