Carbon footprint of crop production in Heilongjiang land reclamation area, China

: In the context of global warming, agriculture, as the second-largest source of greenhouse gas emissions after industry, had attracted widespread attention from all walks of life to reduce agricultural emissions. The carbon footprint of the planting production system of the Heilongjiang Land Reclamation Area (HLRA), an important commodity grain base in China, was evaluated and analyzed in this paper. On this basis, this paper sought feasible strategies to reduce carbon emissions from two aspects: agronomic practices and cropping structure adjustment, which were particularly crucial to promote the low-carbon and sustainable development of agriculture in HLRA. Therefore, using the accounting methods in IPCC and Low Carbon Development and Guidelines for the Preparation of Provincial Greenhouse Gas Inventories compiled by the Chinese government, relevant data were collected from 2000 to 2017 in HLRA and accounted for the carbon emissions of the planting production system in four aspects: carbon emissions from agricultural inputs, N 2 O emissions from managed soils, CH 4 emissions from rice cultivation and straw burning emissions. Then carbon uptake consisted of seeds and straws. Finally, with farmers' incomes were set as the objective function and carbon emissions per unit of gross production value was set as the constraint, this paper simulated and optimized the cropping structure in HLRA. The results showed that there was a “stable-growing-declining” trend in the total carbon emissions and carbon uptake of the planting production system in HLRA, with total carbon emissions of 2.84×10 10 kg and total carbon uptake of 7.49×10 10 kg in 2017. In the past 18 years, carbon emissions per unit area and carbon emissions per unit of gross production had both shown a decreasing trend. To achieve further efficiency gains and emission reductions in the planting production system, it was recommended that the local governments strengthen the comprehensive use of straw resources, optimize irrigation and fertilization techniques, and adjust the cropping structure, i.e., increase the planting area of maize and soybeans and reduce the planting area of rice, and increase subsidies to protect the economic returns of planters.


Introduction
In 2018, the Intergovernmental Panel on Climate Change (IPCC) special report on the impacts of global warming of 1.5°C revealed that if global warming continued to increase at the current rate, intensity and frequency of some climate and weather extremes might be higher [1] . Reducing greenhouse gas (GHG) emissions is regarded as an effective approach to mitigate the worsening global climate change [2][3][4] . Agricultural production releases a large quantity of GHG, which accounts for 52% and 84% of global anthropogenic methane and nitrous oxide emissions [5,6] . So it is urgent to mitigate GHG emissions in agriculture and promote sustainable development [7] .
In China, facing huge population pressure, agricultural outputs must remain increasing to meet food demand, which leads to GHG biochar amendment application [20] might be effective strategies. Thirdly, Yang et al. [27] showed that the carbon footprint of diversified crop rotation systems (sweet potato → cotton → sweet potato → winter wheat → summer maize) is lower than the conventional intensive crop production system (winter wheat → summer maize) in North China Plain. What's more, maize → tomato rotation in India [26] and maize → soybean → oats rotation in the USA [25] were also better than the traditional rotation. Besides, irrigated crops produced more grain with a lower carbon footprint [24] so an appropriate irrigation system was chosen for the better environmental benefit [23] . Finally, compared with the separated crop production system or separated livestock production system, the crop-livestock integrated system would not only recycle agricultural waste, such as livestock manure and crop residue but also cut down the integrated system GHG emissions [28,29] .
Most previous studies showed that carbon footprint was an effective indicator to develop cleaner agricultural technologies at the operational level. Meanwhile, improving crop planting structure would be also an important way to reduce GHG emissions [30] . At present, researches related to crop planting structure optimization through carbon footprint remain scarce. So this study attempts to use this idea by a case study.
In this research, the carbon footprint was used to estimate carbon emission and carbon uptake of crop production from 2000 to 2017 in the Heilongjiang Land Reclamation Area, China. Besides, in order to mitigate GHG emissions and promote agricultural economic development, a multi-objective optimization model was built to make the adjustments to the crops planting structure in Heilongjiang Land Reclamation Area. Through commutating with farmers, they were willing to plant more soybean due to market requirements, which was also considered in this model.

Study area
As China's principal grain production area, Heilongjiang Land Reclamation Area (HLRA) is located in the south of Lesser Hinggan Mountains in northeastern China, including the Songnen Plain and Sanjiang Plain. The Heilongjiang Land Reclamation Administration is located in Harbin. The total land area is 5.54 million hm 2 . The annual average temperature is 1.7°C-4.8°C. The annual average rainfall is 430-600 mm, and the frostless season is 110-140 d. The accumulated temperature of ≥10°C is 2300°C-2700°C.
where, Frac ATD is a fraction of chemical fertilizer and organic fertilizer that volatilizes as NO X and NH 3 , 11.2% [45] ; EF 13 is the emission factor for atmosphere deposition of N on soils, 0.01 kg N/kg [5] . N 2 O emissions from N leaching and runoff are calculated by Equation (10) where, Frac leach is a fraction of N leaching and runoff, 12.6% [45] ; EF 14 is emission factor for N leaching and runoff, 0.0075 kg N/kg [5] .
3) CH 4 emissions from rice cultivation CH 4 emissions from rice cultivation are calculated by Equation (11) where, S B is straw burning rate, %. According to official survey data by HLRA, the ratio of straw burning was 40% from 1999 to 2008, and it was 28% from 2009 to 2017. EF 16 is the emission factor for CO 2 released by straw burning, 1.390 4 kg CO 2 -eq/kg [50] . Carbon emissions from CH 4 released by straw burning are calculated by Equation (14). 4 straw-burning-CH g, S/G, where, EF 17 is the emission factor for CH 4 released by straw burning, 0.002 19 kg CH 4 /kg [50] .

Accounting method for carbon uptake
Carbon uptake is calculated by Equation (15).

Evaluation indicators for carbon footprint
In order to analyze carbon emissions and benefits, the four evaluation indicators are selected in this study as follows.
1) Carbon emissions per unit of area reflect the carbon emission level per area of the crop production system, which is calculated by Equation (16).
where, CE A is carbon emissions per unit of area, kg CO 2 -eq/hm; A is the total crop planting area, hm 2 .
2) Carbon emissions per unit of gross production value are carbon productive forces, which are calculated by Equation (17).
where, CE V is carbon emissions per unit of gross production value, 10 -4 kg CO 2 -eq/RMB; V is the gross production value of the crop production system, 10 000 RMB.
3) Net carbon sequestrations per unit of area reflect the net carbon sequestration level per area of the crop production system, which is calculated by Equations (18) Accounting method for planting structure simulating optimization As a major carbon emission source, agriculture is the socioeconomic development foundation.
Taking into consideration agriculture's sustainable development, it is imperative to mitigate the total amount of carbon emissions in agriculture. During crop production, cleaner technologies, which improve agricultural productivity and the efficiency of agricultural machinery operations [59] , can achieve the goals of GHG emissions reduction. At the same time, previous researches [30,60] pointed out that adjustment of crop planting structure is also an effective way to reduce GHG emissions. So this study attempts to use this idea in HLRA. As the staple crops in HLRA, rice, soybean and maize are chosen as the research objects. This study conducted a multi-objective optimization model based on the adjustment of three staple crop planting structures, in which maximizing farmers' income is set as the objective function. Besides, reducing the carbon emissions per unit of gross production value could be a way to alleviate the contradiction between agricultural economic development and GHG emissions reduction. So the carbon emissions per unit of gross production value are taken as a constraint. The model is as follows: 1) Objective function In order to protect farmers' interests, the maximization of farmers' incomes is the objective function in this study, which is expressed as Equation (20). 2) Constraint for three staple cereal crop planting area According to statistics, the planting area of three staple cereal crops in HLRA was 2.742-2.780 million hm 2 from 2014 to 2017, which fluctuated slightly from year to year. Therefore, the variation range of three staple cereal crop planting areas in HLRA is taken as a constraint, which is expressed as Equation (21). 1 2 3 2779705 s s s + + ≤ 3) Constraint for carbon emissions per unit of gross production value The adjusted carbon emissions per unit of gross production value of three staple cereal crops in HLRA are not greater than that in 2017. And this study conducted a simulation analysis of the impact on the planting structure of three staple cereal crops if carbon emissions per unit of gross production value declined by a certain proportion, which is expressed as Equation (22).  (22) where, cp i is carbon emissions per area of crop i, kg/hm 2 . According to the carbon emission calculation method and field investigation of three staple crops production process, the carbon emissions per unit area of rice is 7999.40 kg/hm 2 , soybean is 2385.98 kg/hm 2 and maize is 2866.39 kg/hm 2 ; cev is carbon emissions per unit of gross production value of three staple crops in 2017, 0.2999 kg/RMB, which is calculated based on the statistic of planting area in 2017; α is the simulating decline proportion of carbon emissions per unit of gross production value, %. 4) Constraint for soybean yield As the major soybean importer, the expansion of the domestic soybean planting area would be conducive to meeting market demand in China. Therefore, this study conducted a simulation analysis of the impact on the three staple cereal crop planting structures if the total soybean production increases by a certain proportion, which is expressed as Equation (23).
where, β is the growth ratio of simulated soybean production, %; Y 2 is the total soybean production in 2017, 2 118 782 t. In this study, two scenarios were simulated, both with the same objective function of maximizing farmers' incomes. In the first scenario, the constraint β was set to 0 for the total soybean production growth, and the change in the planting area of the three crops was simulated when α varied from 0 to 5%, which indicated the decrease rate in carbon emissions per unit of gross production value. In the second scenario, α was set to 2%, and the change in the planting area of the three crops was simulated when β varied from 0 to 50%.  The annual carbon emissions of agricultural inputs were presented in Figure 3. The carbon emissions of agricultural inputs tended to be stable from 2000 to 2003. But the carbon emissions of agricultural inputs increased rapidly from 2003 to 2012, with an average annual growth rate of 8.87%, reaching a maximum value of 6.91 billion kg in 2012. After that, the carbon emissions of agricultural inputs inclined to be stable. Moreover, the carbon emissions of nitrogen fertilizer, agricultural machinery, diesel and electricity accounted for more than 74.04% of the total, which was the main component. Carbon emissions of agricultural machinery exceeded nitrogen fertilizer for the first time in 2015, which accounted for the largest part of agricultural inputs and was about a quarter of the total.   Figure 5 demonstrated the annual carbon uptake of the crops production system in HLRA from 2000 to 2017. In general, carbon uptake tended to be stable from 2000 to 2003. But the carbon uptake increased rapidly from 2003 to 2012, with an average annual growth rate of 12.92%, reaching a maximum value of 90.3 billion kg in 2012. After that, carbon emissions decreased with an average annual decline rate of 3.27%. In 2017, the total carbon uptake in HLRA was 7.49×10 10 kg. Besides, the carbon uptake was composed of grain carbon uptake and straw carbon uptake, which accounted for 47.65% and 52.35% respectively over the years. Figure 5 Carbon uptake of crop production system

Analysis of carbon uptake of crops production
The annual grain carbon uptake was presented in Figure 6. The grain carbon uptake tended to be stable from 2000 to 2003. But grain carbon uptake increased rapidly between 2003 and 2012, with an average annual growth rate of 12.28%, reaching a maximum value of 41.9 billion kg in 2012. After that, grain carbon uptake declined with an average annual decline rate of 3.10%. Moreover, the carbon uptake of rice, maize and soybean accounted for 54.47%, 21.29% and 10.68% of grain carbon uptake, which were the main components. And maize surpassed soybeans in 2005 and became the second-largest crop in carbon uptake. However, the carbon uptake of maize decreased significantly since 2015, with an annual average decline rate of 28.02%, while soybean carbon uptake increased significantly, with an average annual growth rate of 128.43%.    Figure 10 demonstrated the annual net carbon sequestrations per unit of area of the crops production system in HLRA between 2000 and 2017, in which carbon sequestrations were more than the carbon emissions. The net carbon sequestrations per unit of area increased from 2000 to 2012, with an average annual growth rate of 8.85%, reaching a maximum value of 21 800 kg/hm 2 in 2012. After that, the net carbon sequestrations per area declined with an average annual decline rate of 4.04%. Figure 10 Net carbon sequestrations per unit of area According to these results, the carbon sequestrations of crop production system were greater than the carbon emissions, which showed carbon sequestration benefits.
Moreover, carbon emissions per unit of area and the carbon emissions per unit of gross production value declined in recent years. And net carbon sequestrations per unit of area increased. It indicated that crop production in HLRA not only developed rapidly but also improved economic, environmental benefits.

Simulating optimization result of planting structure
3.4.1 Simulation analysis of carbon emissions per unit of gross production value Figure 11 indicates that with the decrease of carbon emission per unit of gross production value, the planting area of rice decreased, the planting area of maize increased, the planting area of soybean remained the same, and farmers' incomes decreased. While the carbon emission per unit of gross production value decreased by 5%, the planting area of rice, soybean, and maize were 1.236×10 6 hm 2 , 0.774×10 6 hm 2 and 0.769×10 6 hm 2 , respectively. Compared with the planting area before, the planting area of rice decreased by 20.7% and the planting area of maize increased by 72.1%, mainly because the carbon emissions per unit of gross production value of rice (0.3232 kg/RMB) > soybean (0.3085 kg/RMB) > maize (0.1577 kg/RMB). Under the constraint that the carbon emission per unit of gross production decreased, the planting area of crops with low carbon emission per unit of gross production value should increase, which led to the results that the planting area of maize increased and the planting area of rice decreased. However, farmers' incomes decreased, mainly since the economic return per unit of gross production value of rice was higher than maize. This simulation indicated that if carbon emission per unit of production value of the planting production system was reduced, the planting area of rice would reduce, and at the same time, farmers' incomes declined. 3.4.2 Simulation analysis of total soybean yield Figure 12 shows that with the growth of total soybean production, the planting area of soybeans increased rapidly, while the planting area of rice and maize decreased simultaneously. While the total soybean production increased by 50%, the planting area of soybean increased by 3.8×10 6 hm 2 . Due to the total planting area remaining the same, the planting area of rice decreased by 2.0×10 6 hm 2 , and the planting area of maize decreased by 1.8×10 6 hm 2 . The reason why the reducing area of maize and rice were similar was that the economic return per unit area of rice was higher and the carbon emission per area of maize was lower. As the total soybean production increased, farmers' incomes gradually decreased, mainly because the economic return per unit area of soybean was lower than rice or maize. It suggested that if soybean production was increased in HLRA, the planting area of rice and maize would be reduced and the problem of farmers' incomes decreasing occurred. Figure 11 Trends of planting area of three major grain crops accompanied by the decrease of carbon emissions per unit of gross production value Figure 12 Trends of planting area of three major grain crops accompanied by the increase of total soybean production

Discussion
The carbon emissions of the planting production system in HLRA were analyzed, including carbon emissions from agricultural inputs, N 2 O emissions from managed soils, CH 4 emissions from rice cultivation, and straw burning emissions.
First, straw burning emissions accounted for an average of 43.36% over the years, which were the largest sources of carbon emissions from the planting production system in HLRA (Figure 2). Crop straws were supposed to fix carbon ( Figure 5), but their large straw burning significantly increased carbon emissions and polluted the atmosphere. Assuming straw burning was converted to comprehensive utilization in 2017, disregarding the economic benefits, for the time being, 39.03% dropped in the simulated carbon emissions per unit area and 22.59% dropped in net carbon uptake per unit area. So, it was urgent to achieve crop straw comprehensive utilization and reduce straw burning, which was conducive to agriculture low-carbonization development [61] . Through interviews with farmers, they also wanted to utilize crop straws in multiple ways. But the bottleneck was the straw collection, storage and transportation. During the crop harvest period in autumn, such as the lack of agricultural machinery operation time, the lack of straw balers and the production cost increases, the straws were little bundled and left the field quickly. Therefore, it was suggested that the local governments should encourage research institutes and farms to work together to optimize more appropriate technology or farm equipment for crop straw harvesting, storage and transportation, to realize the purpose of low cost.
Secondly, CH 4 emissions from rice cultivation were the second-largest source of carbon emissions in HLRA, accounting for an average of 23.91% over the years (Figure 2). This was mainly due to the anaerobic environment formed by flooding, which could be reduced by improved irrigations. Recent studies had shown that improved irrigations such as controlled irrigation [62,63] , leaf-age mode irrigation [63] , and alternate wet and dry irrigation [64,65] could significantly reduce carbon emissions from rice. Through field surveys, conventional irrigation and controlled irrigation were the main rice irrigation modes in HLRA [66] . And the current purpose of improving irrigation was to save water and increase yields. However, in the need of promoting green agriculture, the ways to achieve low carbon emissions in rice cultivation need to be considered in the future. Therefore, it was recommended that the issue of carbon emissions should be considered at the same time when promoting irrigation technology in HLRA.
Furthermore, the average percentage of carbon emissions from agricultural inputs reached 23.81% over the years, which was slightly smaller than the CH 4 emissions from rice cultivation. It was mainly composed of carbon emissions from nitrogen fertilizers, agricultural machinery, diesel, and electricity. Carbon emissions from nitrogen fertilizers accounted for the highest percentage, reaching 24.19% on average (Figure 2). It was the application of nitrogen fertilizers that guaranteed food security in the agricultural production process. Through previous researches, it was found that soil nitrogen in farmland soils in HLRA was in a balanced state, with nutrient utilization efficiency of 51.03% [67] . Compared with developed countries such as the United States [68] , there was still plenty of room for improvement. Therefore, nutrient management techniques could be adopted, such as side-deep fertilization [69] , to improve nutrient utilization efficiency and moderately reduce nitrogen fertilizer application, and then cut down carbon emissions from nitrogen fertilizers and N 2 O emissions from managed soils.
In the simulation and optimization of cropping structure, the simulation analysis of two scenarios was carried out in the planting production system. In the scenario of decreasing carbon emission per unit of gross production value, since the carbon emission per unit of gross production value of maize was almost half of rice, the optimization results showed that the planting area of rice reduced and the planting area of maize increased, and the problem of farmers' incomes declining was brought about. Enriching farmers' income had always been a priority work of the Chinese government. Therefore, in the future, if the cropping structure in HLRA was adjusted according to carbon emissions, some methods were needed considering such as increasing farmers' production subsidies to solve the problem of farmers' incomes declining. Besides, new rice varieties were also beneficial to carbon emission reduction [70] .
In the scenario of increasing the total soybean production, it was found that the planting area of soybean grew, while the planting area of rice and maize declined rapidly. The underlying reason was the low soybean yield of 2467.5 kg/hm 2 . Compared to the soybean yield of 2748.8 kg/hm 2 in Brazil, there was a large gap. In that case, the low soybean yield required more land to plant soybean. Therefore, further exploration of soybean yield potential was needed, from various aspects such as soybean variety selection and breeding [71] , agronomic practices [72] and mechanized production to improve soybean yields and enhance soybean market supply in the future.

Conclusions
This study evaluated and analyzed the carbon footprint of the planting production system in HLRA from 2000 to 2017, of which the total carbon emissions and carbon uptake showed a "stable-growth-decline" trend. In 2017, the total carbon emissions of HLRA were 2.84×10 10 kg, and the total carbon uptake was 7.49×10 10 kg. In terms of carbon emissions, the carbon emissions from agricultural inputs, N 2 O emissions from managed soils, CH 4 emissions from rice cultivation and straw burning emissions accounted for 23.81%, 8.91%, 23.91% and 43.36% on average over the years respectively, and the proportions changed little. The straw burning emissions were the largest source of carbon emissions in HLRA. There were still technical difficulties in the low-cost straw leaving the field, but this category had the greatest potential for emission reduction. In terms of carbon uptake, seed carbon uptake and straw carbon uptake accounted for 47.65% and 52.35% of the total carbon uptake on average over the years, and the proportion changed little. On the whole, the carbon uptake was greater than the carbon emissions of the planting production system in HLRA. The carbon emissions per unit area and carbon emissions per unit of gross production value both showed a decreasing trend, which indicated that the rapid development had been accompanied by an increase in economic and environmental benefits, of the planting production system. To further reduce carbon emissions in HLRA, it was suggested that the local governments should increase the comprehensive utilization level of straw resources to reduce straw burning emissions. Further, something more could be done, such as promoting controlled irrigation to reduce CH 4 emissions from rice cultivation; fertilization methods improvement and nitrogen fertilizer application reduction to low down carbon emissions from agricultural inputs and N 2 O emissions from managed soils.
By simulating and optimizing the cropping structure of the three major crops in HLRA, it was found that a decrease in carbon emissions from the planting production system might require a reduction in the planting area of rice and a moderate increase in the planting area of maize and soybeans, which led to a decrease in farmers' incomes. The low soybean yields required more land to plant soybean. Therefore, it was recommended that the local governments should deeply exploit the soybean yield potential from variety selection and breeding, techniques improvement, and compensate for the decline in farmers' incomes by increasing their production subsidies.
At this stage, only two aspects, farmers' incomes and carbon emissions, were considered in this paper to simulate and optimize the cropping structure of the planting production system. In future studies, if soil health or water consumption were included in the simulation and multi-objective optimization was carried out, it would be possible to provide more comprehensive suggestions for cropping structure optimization and agricultural green development in HLRA.