A Study on the Spatial–Temporal Evolution and Driving Factors of Non-Grain Production in China’s Major Grain-Producing Provinces

Food self-sufficiency in a large country with 1.4 billion people is very important for the Chinese government, especially in the context of COVID-19 and the Russian–Ukrainian conflict. The objective of this paper is to explore the spatial–temporal evolution and driving factors of non-grain production in thirteen major grain-producing provinces in China, which account for more than 75% of China’s grain production, using 2011–2020 prefecture-level statistics. In the present study, the research methodology included GIS spatial analysis, hot spot analysis, and spatial Durbin model (SDM). The findings of this study are as follows: (1) The regions with a higher level of non-grain production were mainly concentrated in the central and western regions of Inner Mongolia, the middle and lower reaches of Yangtze River and Sichuan, while the regions with a low level of non-grain production were mainly distributed in the Northeast Plain. The regions with a higher proportion of grain production to the national total grain production were concentrated in the Northeast Plain, the North China Plain, and the Middle and Lower Yangtze River Plain of China. The hot spot regions with changes in non-grain production levels were mainly distributed in the Sichuan region and Alashan League City in Inner Mongolia, and the cold spot regions were mainly distributed in Hebei, Shandong, Henan, and other regions. (2) An analysis of the SDM indicated that the average air temperature among the natural environment factors, the ratio of the sum of gross secondary and tertiary industries to GDP, the ratio of gross primary industry to the GDP of economic development level, the urbanization rate of social development, and the difference in disposable income per capita between urban and rural residents of the urban–rural gap showed positive spatial spillover effects. The grain yield per unit of grain crop sown area of grain production resource endowment, the total population of social development, and the area sown to grain crops per capita of grain production resource endowment all showed negative spatial spillover effects. The research results of this paper can provide a reference for the country to carry out the governance of non-grain production and provide a reference for China’s food security guarantee.


Introduction
Food security is an important issue related to social and economic stability and longterm national security. The development of urbanization and industrialization causes the transfer of agricultural labor to non-agriculture, the restructuring of rural industries, the decrease in the amount of high-quality cultivated land, the non-grain production of cultivated land, and other problems; thus, food security is constantly threatened [1][2][3][4][5][6].
As a country with a population of 1.4 billion people, the issue of food security has always been of concern to the Chinese government, which emphasizes that cultivated land is used for agricultural production and that high-quality cultivated land is used for grain improvement of crop varieties [44], regulating the use of transferred cultivated land [4 and improving grain-production conditions [46] and grain cultivation subsidies [47] c achieve stable grain production.
In summary, the following are some shortcomings of previous studies on non-gra production: (1) the research object has focused on the country or a particular provinc without considering the non-grain production situation among the main grain-producin provinces; (2) fewer studies have integrated temporal and spatial evolution character tics; and (3) fewer studies have considered spatial-temporal characteristics in the analys of driving factors. Therefore, this paper takes the thirteen major grain-producing pro inces in China as the study areas, uses spatial panel data from 2011 to 2020, and conduc a study on the spatial and temporal evolution trends and influencing factors of China non-grain production with the help of GIS spatial analysis, hot spot analysis, and spat Durbin model, to provide a reference for China's food security.

Study Area
In this paper, thirteen provinces of Heilongjiang, Henan, Shandong, Jiangsu, Anhu Sichuan, Jilin, Hebei, Hunan, Inner Mongolia, Hubei, Jiangxi, and Liaoning in China we used as the study areas ( Figure 1). The data from China's statistical yearbook show th these 13 provinces together account for more than 75 percent of China's total grain pr duction from 2011 to 2020. The structure of grain cultivation in these provinces has important impact on China's food security. Therefore, a study of non-grain production these provinces is very necessary for national food security.

Data Sources
The data were collected from the following sources: (1) administrative boundary data from the National Basic Geographic Information Center http://www.ngcc.cn/ngcc/, accessed on 5 June 2022; (2) several national and prefecture-level city statistical databases, including the "Statistical Yearbook,", the "National Economic and Social Development Statistical Bulletin", and the "Public Data of Government Departments"; and (3) DEM, temperature, and precipitation data from the Resource and Environmental Sciences and Data Center, Chinese Academy of Sciences http://www.resdc.cn, accessed on 5 June 2022). Among them, slope data were obtained using DEM data and Arc Gis 16.0 processing. Missing values were added using the neighboring values, neighboring mean values, and EXCEL-trend function.

Non-Grain Production Ratio
In combination with the research basis of its predecessors, we used previous research results to measure the level of non-grain production by the proportion of non-grain crops planted to the total area of crops sown, where non-grain crops are based on relevant Chinese laws and regulations and are non-grain crops other than cereals, soybeans, and tuber crops [15,48,49]. The formula can be written as follows: In Equation (1), R is the level of non-grain production, B is the area sown to grain crops, and A is the total sown area of crops.

Share of Grain Production
We used the share of grain production of a region in the total national grain production as the degree of contribution of the region to the national grain production. The formula is as follows: where G is the share of grain production in a region in the total national grain production, A is the total national grain production, and P is the grain production of a region.

Hot Spot Analysis
The hot spot analytical method can identify high-or low-value clustering areas of non-grain production changes, and it can effectively represent the spatial distribution characteristics of non-grain production changes [50,51]. In this paper, based on the hot spot analytical tool of the spatial analysis module of Arc Gis10.6 software, we used Getis-Ord Gi * to analyze the hot and cold spot distribution characteristics of statistical significance in the changes in non-grain production. Please refer to Arc Gis 10.6 software for the details of the formula.

Spatial Durbin Model
According to the first law of geography, "Everything is related to everything else, but near things are more related to each other" [52]. Traditional OLS models do not consider spatial characteristics. Spatial panel models consider both the temporal and spatial characteristics of the samples simultaneously. Therefore, they are helpful for capturing the spatial-temporal heterogeneity of the statistical relationships between variables [53]. Spatial economic models mainly include the spatial lag model (SLM), the spatial error model (SEM), and the spatial Durbin model (SDM) [54]. Spatial lag models introduce the spatial lag terms of the explained variable as explanatory variables, and spatial error models introduce the spatial lag terms of the error term as independent variables [55]. While Spatial Durbin models introduce both the spatial lag terms of the dependent variables and the spatial lag terms of the error as independent variables [56]. Therefore, we used the spatial Durbin model to investigate the relationship between non-grain production and its influencing factors. The model is as follows: where y is the explained variable; X is the vector of the explanatory variable; W is the spatial weight matrix; Wy is the spatial lagged term of the explained variable; β is the coefficient of the explained variable; ρ and λ are the spatial autoregressive coefficients; WX is the explanatory variable spatial lag term; θ is the coefficient of the spatial lag term of the explanatory variables; and ε is the residual term. When θ = 0, the spatial Durbin model reduces to the spatial lag model; when θ + ρβ = 0, the spatial Durbin model reduces to the spatial error model. The spatial Durbin model does not adequately respond to the effect of explanatory variables on the explained variables and does not take into account feedback effects. LeSage and Pace [57] proposed a partial differential approach to measure the spatial spillover effects of the variables, or the direct and indirect effects in their terms. Therefore, our study also used this approach. The spatial weight matrix was obtained using GeoDa 1.20 software, and the spatial Durbin model was calculated using Stata 16.0 software.
In this study, the non-grain production ratio was used as the explanatory variable. Concerning existing studies and data availability, 15 indicators were selected from natural conditions, resource endowment of grain production, agricultural science and technology level, urban-rural gap, agricultural production efficiency, social development, and economic development to perform the analysis of influencing factors (Table 1) [27,[58][59][60]. In this study, both the explanatory variables and the explained variables were treated as natural logarithms to avoid the interference of the unit choice of economic variables on the test results. The value of the regression coefficient at this point indicates the elasticity of the explained variables to the explanatory variables, and the economic significance is the percentage change of the explained variables when the explanatory variables change by 1% [61].

Temporal Evolutionary Characteristics of Non-Grain Production
From 2011 to 2020, as Figure 2 shows, Heilongjiang, Henan, and Shandong provinces were the major grain-producing provinces in China with the top three grain production ratios to the national grain production, with mean ratios of 11.20%, 9.75%, and 7.98%, respectively. Hubei Province, Jiangxi Province, and Liaoning Province had the lowest three grain production ratios to the national grain production, with mean ratios of 4.16%, 3.34%, 3.42%, respectively. Anhui Province, Jilin Province, Hebei Province, and Inner Mongolia showed an upward trend in fluctuations in the ranking of the ratio of grain production to total national grain production, while Jiangsu, Sichuan, and Hunan provinces showed a downward trend in fluctuations in the ranking of the ratio of grain production to total national grain production, and the remaining provinces showed a flat trend in the ranking of the ratio of grain production to total national grain production.   The share of grain production to total national production during 2011-2020. Figure 2. The share of grain production to total national production during 2011-2020. Table 2 shows that, from 2011 to 2020, the non-grain production ratio of China showed an upward trend in fluctuations, and the non-grain production ratios ranged from 28.69% to 30.28%; two of these years, 2019 and 2020, had non-grain production ratios exceeding 30%.  Figure 3 shows that, from 2011 to 2020, the non-grain production ratios of the thirteen major grain-producing provinces showed a general downward trend, the non-grain pro-7 of 16 duction ratios of Sichuan and Hunan showed an upward trend, and the average non-grain production ratio in the southern provinces was higher than that in the northern provinces.   The non-grain production ratios in the southern provinces of Hubei, Hunan, Jiangxi, and Sichuan were always above 30%; the average non-grain production ratios in Hubei Province and Hunan Province were higher than 40%; and the average ratios of non-grain production in Jiangsu, Henan, Shandong, Inner Mongolia Autonomous Communities, Hebei, Anhui, and Liaoning provinces were 27.89%, 26.71%, 25.98%, 22.19%, 21.03%, 19.00%, and 18.03%, respectively. From 2011 to 2020, Heilongjiang and Jilin provinces ranked as the last two in terms of non-grain productivity: Heilongjiang's average non-grain productivity was at 4.11%, with the highest non-grain production ratio at 6.68% in 2011 and the lowest non-grain productivity at 2.92% in 2019, showing a decreasing trend. Jilin's average non-grain productivity was at 8.55%, with the highest non-grain productivity at 10.02% and the lowest non-grain productivity at 7.63 in 2020, showing a decreasing trend. Hubei Province and Hunan Province ranked alternately first and second in 2015 for non-grain production ratio; Jiangxi Province and Sichuan Province ranked alternately third and fourth in 2015 for non-grain production ratio except for 2012; Jiangsu Province, Henan Province, Inner Mongolia, and Liaoning Province ranked in a fluctuating upward trend for non-grain production ratio; and Shandong Province, Hebei Province, and Anhui Province ranked in a fluctuating downward trend for non-grain production ratio.
It could be observed that there was no significant relationship between the proportion of grain production to total national production and the ratio of non-grain production in the thirteen major grain-producing provinces during 2011-2020 (Figures 2 and 3). For example, Heilongjiang had a high proportion of total national grain production and a low ratio of non-grain production; Henan and Shandong Province had a high proportion of total national grain production and a high ratio non-grain production; Jiangxi Province had a low proportion of total national grain production and a low ratio non-grain production; Hubei, Hunan, and Sichuan Provinces had a low proportion of total national grain production and a high ratio non-grain production. Figure 4 shows that the spatial distribution characteristics of the proportion of grain production to total national production and the non-grain production ratios in the thirteen major grain-producing provinces at the prefecture-level city scale were clearer. Due to space limitations, this paper only includes the spatial distribution of the proportion of grain production to the total national production and the ratios of non-grain production in 2011, 2016, and 2020. To better analyze, the proportion of grain production to total national production and the ratios of non-grain production are divided into six categories according to their patterns. a diametrically opposite trend. Figure 4b shows the ratios of non-grain production for the years 2011, 2016, and 2020. The spatial distribution characteristics of the non-grain production ratios were clearer. The regions with high non-grain production ratios were distributed in the central and western regions of Inner Mongolia and the middle and lower reaches of the Yangtze River, and the regions with high non-grain production ratios were distributed in the Northeast Plain. There were many regions where non-grain productivity was above 50%, especially in the regions such as Jingzhou City, Wuhan City, Huanggang City, Enshi Tujia, Miao Autonomous Prefecture, Ezhou City, Hubei Province, Dongying City Shandong Province, Nanjing City, Jiangsu Province, Bayannur City, and Inner Mongolia in 2011 ( Figure  4b2011); Alashan League, Bayannur City, Inner Mongolia Autonomous Region, Wuhan City, Huanggang City, Ezhou City, Yichang City, Huangshan City, Xianning City, Hubei Province, Xiangxi Tujia, Miao Autonomous Prefecture, and Hunan Province in 2016 (Figure 4b2016); and Alashan League, Bayannur City, Inner Mongolia, Wuhan City, Ezhou City, Xianning City, Huanggang City, Huangshi City, Hubei Province, Zhangjiajie City, and Hunan Province in 2020 (Figure 4b2020).  Figure 4a shows the national grain production for the years 2011, 2016, and 2020. The main grain-producing regions in China with a high proportion of national grain production were concentrated in the Northeast Plain, the North China Plain, and the Middle and Lower Yangtze River Plain, etc. Harbin, Qiqihar, and Suihua accounted for more than 14‰ of the national grain production, and Changchun and Jiamusi accounted for more than 14‰ of the national grain production in 2016 and 2020. In 2011, 2015, and 2020, the number of cities in the northeast region with a high proportion of total national grain production was on an increasing trend, while the number of cities in the middle and lower reaches of Yangtze River with a high proportion of total national grain production was in a diametrically opposite trend. Figure 4b shows the ratios of non-grain production for the years 2011, 2016, and 2020. The spatial distribution characteristics of the non-grain production ratios were clearer. The regions with high non-grain production ratios were distributed in the central and western regions of Inner Mongolia and the middle and lower reaches of the Yangtze River, and the regions with high non-grain production ratios were distributed in the Northeast Plain. There were many regions where non-grain productivity was above 50%

Spatial-Temporal Characteristics of Changes in Non-Grain Production
This paper uses the difference in the proportion of non-grain production between the two periods before and after to characterize the change in non-grain production. Due to space limitations, the changes in the non-grain production ratios from 2011 to 2015, from 2015 to 2020, and from 2011 to 2020 are shown graphically using the hot spot analysis.
Hot regions indicate high-value clusters for changes in non-grain production ratios and cold regions indicate low-value clusters for changes in non-grain production ratios. As shown in Figure 5, during the three study periods, the Sichuan region and Alashan League city in Inner Mongolia were hot spots of non-grain production rate changes, and Hebei, Shandong, and Henan regions were cold spots of non-grain production rate changes.

Spatial-Temporal Characteristics of Changes in Non-Grain Production
This paper uses the difference in the proportion of non-grain production between the two periods before and after to characterize the change in non-grain production. Due to space limitations, the changes in the non-grain production ratios from 2011 to 2015, from 2015 to 2020, and from 2011 to 2020 are shown graphically using the hot spot analysis. Hot regions indicate high-value clusters for changes in non-grain production ratios and cold regions indicate low-value clusters for changes in non-grain production ratios. As shown in Figure 5, during the three study periods, the Sichuan region and Alashan League city in Inner Mongolia were hot spots of non-grain production rate changes, and Hebei, Shandong, and Henan regions were cold spots of non-grain production rate changes.

Driving Factors of Non-Grain Production
In this section, the first step is to determine whether there is a spatial correlation in the ratios of non-grain production, and the commonly used methods are Moran's I index method, Geary index method, and Getis-Ord index method, etc. We used the widely used global Moran's I index method to test whether there was a spatial correlation in the ratios of non-grain production, and the results are reported in Table 3. During the period from

Driving Factors of Non-Grain Production
In this section, the first step is to determine whether there is a spatial correlation in the ratios of non-grain production, and the commonly used methods are Moran's I index method, Geary index method, and Getis-Ord index method, etc. We used the widely used global Moran's I index method to test whether there was a spatial correlation in the ratios of non-grain production, and the results are reported in Table 3. During the period from 2011 to 2020, the global Moran's I results showed that there was a significant positive spatial correlation between the non-grain production ratios of the thirteen major grain-producing provinces in China; all Moran's I indices were positive and all passed the 1% significance test. Therefore, we could use the spatial econometric models for driver estimation. The second step is to determine which of the SLM, SEM, and SDM models to use. We used the Maximum likelihood (ML) method for estimations ( Table 4). The statistic obtained using the Hausman test is 301.06, p-value = 0.000 (Table 5). Table 5 shows the results tested using Wald spatial lag and LR spatial lag, and all p-values are at the 5% level of significance. According to the above test results, the fixed-effect SDM is the final analytical model. In Table 6, the direct effects on the non-grain production of a region are those exerted from the changes in the explanatory variables of the said region. The indirect effects on non-grain production are those exerted from the changes of explanatory variables in neighboring regions. The total effects are the sum of the direct and indirect effects. Table 6. Spatial spillover effects of variables.

Factors Explanatory Variables Direct Effect Indirect Effect Total Effect
Natural environment Among the natural environmental determining factors, average height, average slope, average air temperature, and average rainfall were used to quantify the effects of terrain and climate, respectively, on non-grain production. The non-grain production of the regions was positively affected by average altitude, with no significant spatial spillover effects, and a significant weak positive direct effect (0.15321) and total effect (0.18885). Average slope had a substantial weak negative total effect (−0.16109) on the non-grain production of the regions, with no significant spatial spillover effects, and a significant weak negative direct effect (−0.14312) and total effect (−0.16109). The significant direct and total effects of average elevation and average slope showed that the former promoted regional non-grain crop production while the latter promoted regional grain crop production. This suggests a trend of grain crop production and non-grain crop production "up the hill," similar to the changing pattern of cultivated land change in China. The average air temperature had a significant strong positive indirect (23.78788) and total effect (27.94924). This indicates a strong positive effect of average air temperature on non-grain production and a strong spatial spillover effect. This is because it is related to the spatial distribution characteristics of China's temperature and non-grain production, which are low in the north and high in the south. Average rainfall had significant negative total effects (−0.22194). The significant negative total effect indicates that average precipitation promotes grain production.
The ability of a region to produce grain was gauged by its resource endowment for grain production. The grain production capacity of a region was measured using area sown to grain crops per capita and grain yield per unit of grain crop sown area. Area sown to grain crops per capita had significant negative direct (−0.53242), indirect (−0.28612), and total (−0.81854) effects on non-grain production. Grain yield per unit of grain crop sown area had a significant positive direct effect (0.17654), negative indirect effect (−0.49140), and negative total effect (−0.31486) on non-grain production. There were significant positive direct effects, negative indirect effects, and negative total effects of grain yield per unit of grain crop sown area. The above data illustrate that higher resource endowment for grain production has a significant spatial spillover effect on non-grain production. In other words, this indicates that regions with greater potential for grain production will encourage nearby regions to do the same. Of course, this applies to every region as well. However, the direct effect of grain yield per unit of grain crop sown area is to promote non-grain production in the region.
The total power of agricultural machinery per unit crop sown area was used to characterize the level of agricultural science and technology, which was used to indicate the degree to realize agricultural mechanization and scale. The significant direct (−0.21141) and total effects (−0.22252) showed that the region's grain production benefited more from agricultural science and technology advancements than non-grain production.
The difference in disposable income per capita between urban and rural residents was used to characterize the degree of the urban-rural gap. The calculation results showed that the urban-rural gap had more significant direct (0.39398), indirect (0.23858), and total (0.63256) effects on non-grain production. This suggests that farmers in their region, in the neighboring regions, and in all regions are encouraged to produce non-grain crops because of the high profits they can generate.
The gross output value of farming was used to measure agricultural production benefits. The significant direct (0.50878) and total effects (0.52320) showed that a region's and all regions' non-grain production was positively impacted by agricultural benefits. This is due to China's severe price regulations on grains, which are in contrast to those that apply to other agricultural products.
Urbanization rate and total population were used to characterize the degree of social development. Urbanization rate had significant positive indirect (0.47215) and total (0.51685) effects. The direct, indirect, and total effects of the total population were significant. This implies that the impact of urbanization rate on inter-regional non-grain production has spatial spillover effects. In other words, it means that an increase in urbanization rate in a region will boost non-grain production in the surrounding areas to meet the demand for non-grain products in the region. The total population had significant negative direct, indirect, and total effects, implying significant direct and spatial spillover effects on non-grain production in both a region and its neighboring regions. This indicates that the concentration of the population will have a great demand for grain production, which effect spills to the surrounding areas to increase the incentive to produce grain. Taken together, both the increase in urbanization rate and population agglomeration will have spatial spillover effects on grain production and non-grain production. In other words, this promotes grain production and non-grain production in the surrounding areas. Among the factors influencing the level of economic development, there was no significant effect of GDP. However, the ratio of gross primary industry to GDP and the ratio of the sum of gross secondary and tertiary industries to GDP had significant positive direct (0.29213, 2.44747), indirect (0.52132, 1.75502), and total (0.81345, 4.20250) effects. In the context of strict control of grain prices, the ratio of the gross primary industry to GDP had a significant positive effect on direct and spatial spillover effects. In the context of urbanization and industrialization, farmers who remain in rural areas tend to grow non-grain crops of higher economic value to increase their income. Spatially, the ratio of gross primary industry to GDP and the ratio of the sum of the gross secondary and tertiary industry to GDP not only positively affected non-grain production in a region but also positively affected its neighboring regions to generate non-grain production. In other words, economic development in a region provides incentives for farmers in and around the region to engage in high-income non-farm work and to grow non-grain crops of high economic value.

Conclusions
This paper examines the spatial and temporal evolution and driving factors of nongrain production in thirteen major grain-producing provinces in China, using GIS spatial analysis, hot spot analysis, and spatial Durbin model. The main findings of this paper are as follows: (1) Non-grain production showed an upward trend from 2011 to 2020 in the thirteen major grain-producing provinces of China. From 2011 to 2020, Heilongjiang, Henan, and Shandong provinces were the top three provinces in the thirteen major grainproducing provinces of China in terms of grain production, with an average share of 11.20%, 9.75%, and 7.98% of the national grain production, respectively. Hubei, Jiangxi, and Liaoning provinces were the last three provinces in the thirteen major grain-producing provinces of China in terms of grain production, with an average share of 4.16%, 3.34%, and 3.42% of the national grain production, respectively. From 2011 to 2020, Hubei, Hunan, Jiangxi, and Sichuan provinces had a level of non-grain production above 30%, while Jilin and Heilongjiang provinces had a level of non-grain production below 10%. The share of grain production in the thirteen provinces was not significantly related to the level of non-grain production. (2) The regions with high ratios of grain production to total national production were concentrated in the Northeast Plain, the North China Plain, and the Middle and Lower Yangtze River Plain of China. The areas with high non-grain production were mainly concentrated in the central and western regions of Inner Mongolia, the middle and lower reaches of the Yangtze River, and Sichuan, while the areas with low non-grain production were mainly distributed in the Northeast Plain. The hot spot areas of nongrain production changes were mainly distributed in the Sichuan region and Alashan League City in Inner Mongolia, and the cold spot regions were mainly distributed in Hebei, Shandong, Henan, and other regions. (3) There was a significant spatial spillover effect of influencing factors in China's major grain-producing regions, with regions being influenced by non-grain production in their neighboring regions. The analysis of the SDM indicates that the average air temperature of natural environmental factors, the ratio of the sum of gross secondary and tertiary industries to GDP, the ratio of gross primary industry to the GDP of economic development level, the urbanization rate of social development, and the difference in disposable income per capita between urban and rural residents of the urban-rural gap showed positive spillover effects and affected the non-grain production of neighboring areas. In addition, the grain production resource endowment of grain yield per unit of grain crop sown area, the total population of social development, and the area sown to grain crops per capita of grain production resource endowment all but significantly and negatively affected the non-grain production of neighboring areas.  Data Availability Statement: The DEM, temperature, and precipitation data were obtained from http://www.resdc.cn. The data were also obtained from several national and prefecture-level city statistical databases, including the "Statistical Yearbook,", the "National Economic and Social Development Statistical Bulletin", and the "Public Data of Government Departments" (too many sites to list). The data that support the findings of this study are available from the first author upon reasonable request.