Evaluation and Driving Force Analysis of Cultivated Land Quality in Black Soil Region of Northeast China

Cultivated land is an important natural resource to ensure food, ecological and economic security. The cultivated land quality evaluation (CQE) is greatly significant for protecting and managing cultivated land. In this study, 320 counties in the black soil region of Northeast China (BSRNC) represent the research units used to construct the CQE system measuring the soil properties (SP), cultivated land productivity (CLP), ecological environment (EE) and social economy (SE). The total of 19 factors were selected to calculate the integrated fertility index (IFI) and divided into grades. Simultaneously, we used the coupling coordination degree model to comprehensively analyze the spatial pattern of the cultivated land quality (CLQ) in the BSRNC, and use the structural equation model (SEM) to analyze the driving mechanism. The results show the following: 1) The CLQ of 262 counties in the BSRNC is in a state of coupling and coordination, and the coupling and coordination degree presents a spatial distribution pattern of ‘high in the southwest and northeast, low in the northwest and southeast’. The coordinated development degree of 271 counties is between 0.4 and 0.6, which is in a transitional state between coordination and disorder. 2) The CLQ in the BSRNC is generally good, with an average grade of 3. High-quality cultivated land accounts for 58.45% of all counties, middle- and upper-quality cultivated land accounts for 27.05%, and poor-quality cultivated land accounts for 14.49%. 3) The SEM analysis shows that the SP, CLP, EE, and SE all influence the CLQ. Among them, the SP has the largest driving force on the CLQ, while the SE has the smallest driving force on it. The results confirm that the main factors affecting the evaluation results are crop productivity level, normalized difference vegetation index, ratio vegetation index, difference vegetation index, and organic carbon content. When implementing protection measures in counties with a low CLQ, considering a balanced coordination of multiple systems and reasonably controlling the quality degradation are important. This study provides the current situation and driving factors of the CLQ in the BSRNC and will play an important role in black soil governance and utilization.


Introduction
The black soil region of Northeast China (BSRNC) is one of three black soil belts that are suitable for cultivation in the Northern Hemisphere (Wang et al., 2022) and has superior natural conditions for agricultural develop-ment. Cultivated land is one of the most valuable natural resources. Its quantity and quality play a vital role in the country's social stability and economic development (Li et al., 2018). Therefore, we should understand the current situation and characteristics of the BSRNC's cultivated land quality (CLQ), pay attention to its development, and objectively evaluate it in time. These are essential for black soil resource development and utilization, CLQ improvement, sustainable agricultural development, national food security, and ecological health.
The CLQ refers to cultivated land's status and conditions (Chen and Shi, 2020). Cultivated land quality evaluation (CQE) involves evaluating them (McBratney et al., 2014), establishes an evaluation index system and constructs a suitable evaluation model. Simultaneously, understanding and evaluating the CLQ are important for maintaining or improving soil quality and crop production, and it is also a main sustainable objective for ensuring food security (Foley et al., 2011). For a long time, researchers have conducted research and explored CQE from multiple perspectives and scales. The Cornell Framework for assessing soil health in the United States outlines that soil quality is based on soil chemical, physical, and biological processes to propose matching evaluation factors (Moebius-Clune et al, 2016). The United Kingdom and Canada mainly classify soil productivity, emphasizing the limitations of farmland changes on agricultural production (Huffman et al, 2000). Currently, China's CQE has evolved from a single assessment of cultivated land productivity and soil fertility to a comprehensive assessment of economic, social, and ecological security and sustainability. Traditionally, CQE is usually conducted through field measurements, which are time-consuming and expensive. Developing remote sensing (RS) and geographic information system (GIS) technology have improved the CQE to be multi-scale and multi-temporal and have wide coverage and high precision (Zhu et al., 2020).
CQE research mainly includes constructing the evaluation index system and methods and selecting the evaluation scales. 1) Index system construction. The evaluation systems constructed are different, such as the pressure state response (P-S-R) CQE model (Ma et al., 2016). Regarding index selection, many studies focused on soil fertility, health, and physical and chemical properties , while ignoring the other factors related to CLQ. However, the scientific selec-tion of factors should cover soil conditions, socioeconomic conditions, climatic and ecological factors, yield and other aspects. 2) Evaluation methods. Evaluation methods include the weighted summation method (Yu et al., 2010), fuzzy evaluation method (Wu et al., 2019;Kaufmann et al., 2009), gray correlation analysis method (Wang et al., 2016), technique for order preference by similarity to an ideal solution (TOPSIS) method (Cao et al., 2021), back propagation (BP) neural network model (Liu et al., 2022). Among these, the weighted summation method is probably the most used, because it has the advantages of simple construction and strong data compatibility (Shao et al., 2020). In determining weights, they can be divided into the subjective weighting method (Analytic Hierarchy Process and Delphi Method), objective weighting method (Entropy Weight Method), comprehensive weighting method (combination of Analytic Hierarchy Process and Entropy Weight Method), etc. (You et al., 2018). 3) Evaluation scales. The evaluation units are mainly divided by provincial, municipal, county or required study area boundary lines or by a grid or plot of a certain scale as the evaluation unit, as well as the land use map's irregular vector range . Different CQE scales have different evaluation factors, system frameworks, methods, and applications of results. Currently, the agricultural land classification and CQE in China are based on counties and implement national and provincial policies and plans. The county unit is the best scale for sustainable utilizing and managing cultivated land resources.
Coupling refers to the dynamic relationship between multiple different systems that affect and interact (Zameer et al., 2020). The coupling and coordination degree of harmony between systems in the development process, which reflects the system's trend from disorder to order. Many scholars have studied cultivated land using the coupling coordination degree, such as the multi-function utilization of cultivated land (Gong et al., 2022), low carbon utilization and urbanization of cultivated land (Zhang et al., 2022a), agricultural land circulation and green utilization of cultivated land (Zhou et al., 2022). Studies are increasingly evaluating the overall CLQ level by combining multiple systems. Most studies focus on the impact of a single system or different systems on it, and the understanding of its multi-system interaction is insufficient. However, the CLQ is the result of the coordinated development of multiple sys-tems, and there is little research on its coupling coordination degree's spatio-temporal pattern. CQE mostly uses the target, criterion, and indicator layers as the evaluation framework. Usually, the target layer is used to analyze the evaluation target, and the indicator layer is used to analyze each single indicator's impact mechanism. The criterion layer contains the intermediate links involved in achieving the objectives, and its important role in the evaluation framework can not be ignored. The coupling coordination degree model can be used to analyze the coupling coordination relationship between criterion layers, improving the CQE's robustness.
The existing CQE mainly serves for specific projects, and there is less research on CLQ's limiting factors (Yuan et al., 2018). Improving the driving factors directly affects the CLQ improvement. Therefore, determining various factors' driving roles is particularly important. In the face of different constraints and driving factors, targeted measures should be adopted to govern and improve cultivated land. At present, driving force analysis mostly uses methods such as principal component analysis (Olaniyi and Abdullah, 2021), the correlation coefficient method, multiple linear regression analysis (Zhang et al., 2022b). These methods can only calculate the single relationship between the independent and dependent variables, ignoring the complex interaction between the factors (Sha et al., 2017). The structural equation model can analyze the structural relationship between variables that can not be directly measured and the causal relationship between multiple variables and provide each relationship's strength (Bagozzi and Yi, 2012). Therefore, using structural equation modeling, we can not only analyze the multiple influence relationships between the CLQ and multiple influencing factors, but also obtain a specific influence's intensity. It resolves the limitation of analyzing the relationship between the CLQ and other single influencing factors in the past.
To summarize, this study uses the BSRNC as the research area, constructs the CQE system, analyzes its coupling development pattern and the CLQ's driving mechanism. The marginal contribution is as follows: first, using the county as the evaluation unit, a CQE system including four subsystems, soil properties (SP), cultivated land productivity (CLP), ecological environment (EE), and social economy (SE), was constructed. Comprehensive factor analysis and evaluation of the CLQ enrich the CQE research. Secondly, the internal interaction among its four subsystems is investigated by using the coupling coordination model, and the characteristics of the CLQ's spatial coupling development in the BSRNC are comprehensively analyzed. These clarify the interaction between its multiple systems and provide direction for cultivated land governance. Thirdly, a structural equation model is used to identify the key driving factors and clarify the driving mechanism between the CLQ and the four subsystems. It improves the traditional method based on the one-way influence relationship between independent variables and dependent variables, and further optimizes the influence path relationship between multiple elements. This study's significance is in providing a theoretical basis and technical support for scientifically quantifying the cultivated land level and proposing reasonable governance measures by exploring the BSRNC's CLQ.

Study area
The BSRNC is located at 115°24′00″E−135°00′12″E and 38°42′00″N−53°00′36″N. Its administrative divisions include Liaoning, Jilin and Heilongjiang provinces, Hulun Buir, Chifeng, Tongliao cities, and Hinggan League in the eastern part of the Inner Mongolia Autonomous Region (National Agricultural Technology Extension Service Center et al., 2017) ( Fig. 1). It belongs to the temperate continental monsoon climate. The summers are warm, rainy and short, with concentrated rainfall, and the winters are cold, long and dry. Northeast China is rich in light resources, sufficient precipitation and moderate temperatures in the planting season, which is suitable for producing many kinds of crops. Due to the cold climate and short land reclamation time, black or dark humus generally exists on the soil surface, forming a fertile and deep black soil layer; thus, the Northeast black soil has become one of the most fertile soils in the world (Xu et al., 2010). The grain-planting system is an annual crop rotation, mainly planting corn (Zea mays), soybean (Glycine max), rice (Oryza sativa), sorghum (Sorghum bicolor), millet (Setaria italica) and spring wheat (Triticum aestivum). The BSRNC is an important grain production base in China. In 2020, the commodity grain output of Northeast China's three provinces was about 138 million t (Qu et al., 2021). The total cultivated land area accounts for 17.63% of the national cultivated land area (Gu et al., 2018), which is mainly distributed in the Songnen Plain, Liaohe Plain, Liaoxi Hills, and Sanjiang Plain.

Data source and processing
The data include socio-economic, land use, meteorological, soil properties, and vegetation index data. Among them, the socio-economic data were obtained from the statistical yearbooks of Northeast China's provinces and cities (http://www.stats.gov.cn/). We obtained the 30 m annual land cover datasets and their dynamics in China from 1990 to 2020, and the 1 km monthly mean air temperature for China from January 1951 to December 2020 from Zenodo (http://zenodo.org). The precipitation data are from the monthly precipitation data set with a 1 km resolution in China from 1960 to 2020 downloaded from the Scientific Data Bank (http://cstr. cn/31253.11.sciencedb.01607). Using the Google Earth Engine (GEE) platform, the slope data are calculated with a 30 m resolution DEM, and the three vegetation indices data are extracted from Landsat 8 images. All the above data are dated 2020. The soil erosion data were obtained from the Resource and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn). This data set contains the soil erosion intensity in 1995, which was used as the an-nual erosion rate to evaluate the future erosion change. The data of the soil texture, water content, organic carbon content and pH value were obtained from the global 250 m resolution soil data set on GEE, which was released in 2018 (https://earthengine.google.com/). Due to the slow change in the soil data, the gap between 2018 and 2020 is only two years, so the soil data used in this paper can be used to assess the 2020 CLQ situation.
We used the American soil texture classification method for the soil texture data, which divides them into 12 types. The 12 soil texture types were divided into five grades (Table 1). We calculated each indicator's average value within each county. Additionally, all the data were standardized using the range transformation method to realize the dimensionless processing of different variables types. After screening and sorting the data, 320 counties in Northeast China were selected as evaluation units, which basically cover the BSRNC, with strong representativeness.
The flow chart of this study is shown in Fig. 2

Evaluation method 2.3.1 Construction of evaluation index system
The CLQ refers to cultivated land's ability to support agricultural production under the comprehensive effects of nature and social economy. Its evaluation factors should comprehensively and objectively reflect the factors affecting it, such as natural, social-economic, and environmental health. The selection of the CQE factors should follow scientific principles and accurately reflect its main contents. It must measure the relationship between the CLQ and its regional soil, nature, society and environment from a macroscopic perspective to obtain the correct assessment results. The CQE can divide the CLQ into multiple subsystems, define it as a complex composed of different quality subsystems, and use multiple factors to measure the different subsystems (Sheng et al., 2021). Based on the above principles and fully referencing the existing relevant research, our research constructs a CQE system from four aspects: SP, CLP, EE, and SE. Simultaneously, attention should be paid to the system's hierarchical nature and the interrelationship and coordination among the elements. Among them, the SP is the basis for developing the cultivated land production potential and is the most direct evaluation index of the CLQ (Shao et al., 2007). The CLP can directly reflect the cultivated land production capacity. The EE affects the CLQ and causes it to continuously change. The SE influences the CLQ through human activities (Jiang et al., 2017). These four factors constitute a complete CQE system. The formation and development of the CLQ in any region is the result of the comprehensive action of all parts of the whole system, rather than a single part and its contents.
The selected factors can not only analyze, compare and evaluate the CLQ status, but also judge its development and change. This paper determines the evaluation factors from the following aspects. First, the decisive factors affecting CLQ, as well as the factors in national, industry and local industry standards; for example, the rules for Grading Agricultural Land Quality (GB/T 28407−2012), cultivated land quality grade (GB/T 33469−2016), soil environmental quality standard (GB 15618−2008). Second, the factors are derived from the CLQ assessment and early warning, agricultural land classification and other relevant documents. Third, the sele-ction of factors can reflect the practicability and pertinence of the BSRNC's large-scale characteristics. Fourth, there are sufficient means to obtain various factors. The study shows that the CLQ is determined by soil characteristics, climatic conditions, production capacity and technical conditions. In addition to the above factors, this study also added soil erosion to the evaluation system, thus highlighting the harm to cultivated land. Simultaneously, combined with remote sensing, the multiple vegetation indices are used to reflect the cultivated land's status. Therefore, we selected 19 single evaluation factors that support the four subsystems (Table 2). These evaluation factors are quantitative analysis factors under the subsystem category, and various evaluation contents reflecting the CLQ system are selected.

Integrated fertility index calculation
The Analytic Hierarchy Process (AHP) is a research method that combines qualitative and quantitative analysis to solve complex multi-objective decision-making problems (Ozdemir, 2005). It is based on experience and knowledge, combined with objective judgments, to establish a paired comparison matrix between various indicator factors (Li et al., 2018). A hierarchical model is constructed based on the CLQ indicator system, and the weight values reflecting the factors' relative order of importance are calculated from the judgment matrix constructed using the 1-9 level comparison scale. The entropy weight method is important for assigning index weights and constructing a measurement model based We choose a combination of subjective and objective weighting methods to determine the weight values of multiple factors. The formula is as follows: where w j and are the weight values of each index determined using the AHP and entropy weight method, respectively, and j is the number of factors.
Calculating the Integrated Fertility Index (IFI) is a key link in constructing the CQE system ) that can reflect the CLQ level in Northeast China. The cumulative method is used to calculate each evaluation unit's IFI. The calculation formula is as follows: where C ij is the score of each evaluation factor in the CQE indicator system, i is the number of the evaluation unit, j is the number of factors, W j is the weight value determined by each evaluation factor, and n is the number of evaluation factors in the indicator system. Finally, the CLQ is divided into ten grades using the equidistant method. The first-class land is the best, while the tenthclass land is the worst.

Coupling coordination degree model
Coupling and coordination is a phenomenon from system theory, which emphasizes the close relationship and complex interaction between different subsystems, and can explain the system's sustainable development (Fang et al., 2016). The coupling coordination degree (C) can reflect the coordination degree among CLQ systems. The coordination development degree (D) considers both the interaction strength between the four subsystems and their respective development levels, which can Notes: Crop productivity level = actual productivity/potential productivity. The actual output represents the actual productivity; potential production is expressed by net primary productivity (NPP) of natural vegetation. Habitat quality index = (0.35 × Forest land + 0.21 × Grassland + 0.28 × Water area wetland + 0.11 × Cultivated land + 0.04 × Construction land + 0.01 × Unused land)/area. -means that indicator is dimensionless reasonably express the measurement of coupling and coordination and the development levels between systems.
Regarding the coupling and coordination degrees of the CLQ, when the four subsystems of the SP, CLP, EE, and SE are well matched, it is the high-quality state of CLQ. Any unreasonable change in either party will affect the four subsystems' coupling and coordination (Zhang et al., 2019). We use the modified coupling model proposed by Wang et al. (2021). The model disperses the C in [0,1] as much as possible, which increases the discrimination of the C values and has a higher use validity. The formula is as follows: (3) where U i (i = 1, 2, 3, 4) is the comprehensive index of the four subsystems: SP, CLP, EE and SE. α 1 , α 2 , α 3 , α 4 is the weight of the four subsystems, C is the coupling coordination degree, and D is the coordinated development degree. T is the comprehensive evaluation index of the SP, CLP, EE, SE in the study area. The closer the C value is to 1, the better the coupling degree of the four subsystems of the CLQ is; the closer the D value is to 1, the closer the four subsystems are to order and coordination (Li and Gao, 2022). The four subsystems' weights are determined using the comprehensive weighting method, and the comprehensive index is calculated using the cumulative method according to the index system mentioned above. In this study, the C and D values are classified into ten grades and three types, and then the coupling and coordinated development degree of the four subsystems of the CLQ are judged (Table 3).

Structural equation model
The Structural Equation Model (SEM) is used to establish, estimate and test various causal relationships between different variables. The SEM integrates multiple regression, factor analysis, path analysis and other statistical analysis models (Liu et al., 2015). It includes not only the measurement variables that can be directly observed, but also the latent variables that are difficult to directly observe. The SEM analyzes the causal relationships between measurement variables and latent variables and between different latent variables by establishing hypothesis models and combining indicator systems (Ben Nasr et al., 2021). The SEM includes measurement and structural models.
The measurement model describes the latent variable measured using the observed variable, in the following form: where η is an endogenous latent variable, which is affected by other variables in the model, is an exogenous latent variable, which can only explain other variables in the model, X is the observation index of , Y is the observation index of η, Λx is the coefficient matrix of x on , which is composed of the factor load of x on , and describes the relationship between the exogenous explicit variables and the exogenous latent variables, Λy is the coefficient matrix of y on η, which is composed of the load of y on η and describes the relationship between Notes: C refers to coupling coordination degree, and D refers to coordinated development degree the endogenous and explicit variables and the endogenous variables, δ is the measurement error term of x and ε is the measurement error term of y. The structural model is used to analyze the relationship between latent variables that cannot be directly measured or observed. The form is as follows: where B is the coefficient matrix of the endogenous latent variable, describing the mutual influence between endogenous latent variables η, Г is the coefficient matrix of the exogenous latent variable, which describes the influence of the exogenous latent variable on the endogenous latent variable η, and is a random interference term, reflecting the part of endogenous potential variable η that cannot be explained in the equation.
The partial least squares structural equation model (PLS-SEM) can reduce the requirement for data to meet the requirements of normal distribution, better solve the problem of multicollinearity among indicator factors, and has advantages in exploratory research with less samples and more complex models (Hair et al., 2019). Our research builds a CLQ driving mechanism analysis model based on the PLS-SEM. The partial least squares method is used to estimate the parameters, analyze the driving relationship between the CLQ and multiple indicator factors, explore the factors that affect it, and analyze the influence's path and extent. Meanwhile, the CLQ and four subsystems are used as latent variables, and 19 individual factors are used as measurement variables (Table 4).

Spatial pattern of CLQ coupling and coordinated development degree
Due differences in the natural environment, economic development, soil state and other conditions, the C and D of each BSRNC county also significantly differed.  According to the basic connotation of the coupling coordination degree model, the C and D of the SP, CLP, EE and S of each city are calculated using formulas (3)-(6). On this basis, the CLQ's coordination and development status in the BSRNC is evaluated. According to the division standard (Table 3), the C and D are classified and displayed ( Fig. 3 and Fig. 4). The coupling coordination degree calculation results for the 320 counties are classified into ten types. In general, the four subsystems of the CLQ in the BSRNC counties are in a coupling coordination state, while a few counties are in a disordered state. From a spatial perspective, the CLQ coupling coordination degree's regional differences in Northeast China are significant, which is related to the regional geographic location to a certain extent. The C presents a spatial distribution pattern of 'high in the southwest and northeast, low in the northwest and southeast', with obvious spatial differentiation characteristics of decreasing from the central region to the north-south direction. Specifically, the middle and south of Inner Mongolia Autonomous Region, the west of the Liaoning Province, and the north of the Jilin Province have the best coupling and coordination degree, and obvious aggregation trends. The regions with poor coupling are mostly far away from the middle regions, such as the north of Inner Mongolia Autonomous Region, the south of the Jilin Province, and the southeast of the Liaoning Province, the south of the Jilin Province, and the southeast of the Liaoning Province.
From the coordinated development degree (D) calculation results, it can be concluded that most counties' Ds in the BSRNC are between 0.4 and 0.6. The D is in the transitional stage of development, dominated by the ND and RC. The D is less than 0.7 in all regions, and is generally lower than the C. This shows that although the four subsystems of the CLQ in Northeast China are well coupled, the overall coordinated development level is not high. Moreover, its coordinated development speed is relatively slow, and the elasticity of its development and change are poor. From a spatial perspective, the spatial distribution of the D in the BSRNC is characterized by 'coordination in the east, disorder in the south, north and west'. Additionally, its east-west direction shows a weak E shape trend. Specifically, the coordinated development regions are mainly concentrated in the middle and east of the Heilongjiang Province, the middle of the Jilin Province, and the middle of Inner Mongolia Autonomous Region. In addition, it shows a decreasing trend from east to west. The disorder areas are mainly concentrated in the Liaoning Province, the southeast of the Jilin Province, and the south and north parts of Inner Mongolia Autonomous Region.

Results of cultivated land quality evaluation
The weights of the 19 index factors are determined using the analytic hierarchy process and the entropy weight method (Table 5). To analyze the CLQ's spatial differences in different counties, the CLQ spatial distribution pattern in the BSRNC is obtained according to the calculation results (Fig. 5). Overall, the CLQ of the BSRNC counties is distributed from the first to the tenth grades, with an average grade of three. This shows that the CLQ in the BSRNC is at a high level. From a spatial perspective, it presents a spatial pattern of 'low in the southwest, high in the east, south and north', with a certain degree of aggregation. Among them, from first to third grade lands are high-quality cultivated land, ac-  Table 4 counting for 59.37% of the total number of counties. They are mainly distributed in the south and east of Northeast China. The CLQ of from fourth to sixth grade land is at the middle and upper level, accounting for 29.38% of the total number of countries. They are mainly distributed in the north, middle and south of Northeast China. The CLQ of from seventh to tenth grades land is poor, accounting for 11.25% of the total number of counties. They are mainly distributed in the west of Northeast China.

Results of structural equation model
The bootstrapping method is commonly used to verify the model's significance (Hair et al., 2014). A large subsample from the original sample is used to estimate the model and determine its significance by randomly replacing the measurements. A suitable model usually requires P < 0.05 (Mendes et al., 2022). Simultaneously, the coefficients of determination (R 2 ) and cross validated redundancy (Q 2 ) are selected to judge the predictive ability of the model and evaluate the model's explanatory ability. R 2 > 0.75 indicates that the model has a strong explanatory ability, R 2 > 0.5 indicates that the model has a medium or above explanatory ability, and R 2 > 0.25 indicates that the model has a weak explanatory ability (Streukens et al., 2016). When Q 2 assesses the prediction correlation of the internal model, it is generally required that Q 2 > 0. The greater the Q 2 , the stronger the model's prediction ability. Several factors confirm that the quality of the PLS-SEM reached the standard ( Table 6). The R 2 of the CLQ is 0.775, and the R 2 of the CLP is 0.823, thus indicating that the model has a strong explanatory ability. The R 2 of the SP is 0.599, thus indicating that the model has a medium or higher interpretation capacity. The Q 2 of the CLQ is 0.789, the Q 2 of the CLP is 0.405, and the Q 2 of the SP is 0.232, which are all greater than zero. The model has good prediction ability. Fig. 6 shows the results of the driving mechanism ba-   Fig. 6 Total effect of cultivated land quality driving relationship in the black soil region of Northeast China. The solid line represents the total effect, and the dotted line represents the indirect effect. The direct effect is equal to the total effect minus the indirect effect. * means P < 0.10, ** means P < 0.05, and *** means P < 0.01. The abbreviations are consistent with Table 4. → represents the path through which the former affects the latter sed on the PLS-SEM. The results show that there are complex interactions among the CLQ system drivers, but the driving forces among different factors are also different. The results of the PLS-SEM analysis show that there are complex driving relationships between the CLQ and the SP, CLP, EE, and SE. The SP has the largest overall impact on the CLQ (0.736). The second is the EE's impact on the CLQ (0.651), the third is the CLQ's impact on the CLP (0.435). Finally, the SE has the least impact on the CLQ (0.095). Simultaneously, the EE also has a positive effect on the CLP and SP, and the influence intensity is 0.774 and 0.767, respectively. We also analyzed the influence of the SP and SE on the CLP. However, the results were not significant, thus indicating that they had no significant direct impact on the CLP.
In addition, we obtained the indirect interaction's intensity between the latent variables. Among them, the indirect effect of the EE → SP → CLQ path is 0.564, the indirect effect of the EE → SP → CLP path is 0.320, and the indirect effect of the SE → CLQ → CLP path is 0.042. The EE has a greater indirect impact on the CLQ.

Comprehensive analysis of evaluation results and coupling coordination
In the southwest of the BSRNC, the C and CLQ levels are opposite, and these regions have the problem of having a high C but a low CLQ, which is very unfavorable to the cultivated land's future development. In the middle of the Heilongjiang Province and the Jilin Province, the C and CLQ grades are both high, indicating that this region's CLQ is balanced and coordinated, and the state is the best. In the west of Hulun Buir, both the C and CLQ grades are low, indicating that this region's CLQ coordination is poor. Even if one of the subsystems has a high level of quality, the overall quality is still poor. Among them, there are three counties where the C belongs to extreme disorder, namely the Jalai Nur District and New Barag Left Banner in the Inner Mongolia Autonomous Region, and Ji'an City in Jilin Province. Jalai Nur District is adjacent to the New Barag Right Banner, both of which are traditional pastoral areas. The composite factors of the EE and CLP are extremely low, and the SE is also far less than that of other regions. The four subsystems' development of the CLQ is unbalanced, and the coupling and coordination degree is extremely poor. The two regions are not suitable for agricultural industrialization due to their geographical location, climate conditions and other factors. Ji'an City is in the southeast border of Jilin Province, and the calculated SE is extremely low, which is quite different from the comprehensive index of other three subsystems. Among them, this region's PCC and LRR are much smaller than those in other regions, which is the main reason for the low C. Fuxin City and Xinmin City in the Liaoning Province are high-quality coordination units, and their geographical locations are adjacent, which shows that their four CLQ subsystems have the characteristics of balanced coordination and similar comprehensive levels. It is worth noting that although the coupling and coordination of the four CLQ subsystems of the two are strong, the IFI is not high, being lower than the average IFI value in Northeast China. To summarize, except for the middle of Heilongjiang Province and Jilin Province, the CLQ level in other regions is opposite to the C's trend. Although the CLQ is high, the C is very low. Therefore, CLQ improvement in these areas can start from the C, and targeted measures can be adopted according to its four subsystems' coupling statuses to comprehensively manage and improve them.
In terms of the D, Tongliao City, Chifeng City, and western Hulun Buir City in Inner Mongolia, and the north of the Jilin Province have low CLQ and D levels and certain spatial similarities. This indicates that the development level of low CLQ regions will also be poor. However, the CLQ in the Jilin Province, the southern Liaoning Province and eastern Hulun Buir is higher, but the D is lower. Therefore, we should pay attention to effective cultivated land management in these areas to prevent CLQ degradation. Among them, the counties with serious and extreme disorder are, respectively, the Jalai Nur District, New Barag Right Banner, and Kailu County, all within the Inner Mongolia. The C and D of the three are at low levels. The Kailu County has low SP, high soil erosion intensity, large pH, and poor soil texture, which are the important reasons for the low coordinated development level. In conclusion, the coordinated development level of the BSRNC's CLQ requires further improvement. In this process, the disordered development of any kind of subsystem will lead to the decline of subsystem coordination, so all subsystems need to work together to improve the CLQ.
Spatial agglomeration characterizes both the C and D, which reflects the similarity of adjacent areas in terms of the soil, cultivated land productivity, natural environment, agricultural economic development level, etc. This indicates that there is strong interaction and coordination between adjacent regions. The coupling and coordinated development attributes of the four CLQ subsystems depend on their respective development levels. Improving the comprehensive development level of a single subsystem can not promote the benign and coordinated development of the CLQ. In addition, regarding the CQE, we should not only consider the coupling and coordination degree between the CLQ subsystems, but also pay attention to the comprehensive evaluation level and judge the CLQ level in the BSRNC. Only when the IFI, C and D all reach a high level, will there be a high-quality state and high-quality balanced coordination of the cultivated land.

Driving mechanism of cultivated land quality
The CLQ changes mainly depend on the comprehensive changes in the SP, CLP, EE and SE, and its four subsystems are ultimately influenced by each index factor. Among them, the SP has the largest driving force, so we should pay attention to their dominant role in the evaluation. It is also the most closely related to and important factor for the CLQ. Specifically, the SOC is the biggest driving force factor in the SP, which indicates that its impact on the CLQ is extremely important. The SOC is the effective expression of organic matter content in soil and is the main source of nutrients required for crop growth (Xing et al., 2021). The SMC and ST also have a strong driving effect on the SP. The ST is closely related to soil aeration, water retention, and fertilizer retention. The SMC is the main influencing factor of nutrient transport in soil. To ensure normal plant transpiration progress, crop growth and yield depend on the availability of water resources (Yang et al., 2018). The S has a negative influence on the SP, being a soil topographic factor. It affects the soil quality by influencing differences in the surface water and heat distribution, as well as the distribution, direction, and rate of movement of the material accumulation. The SED also affects the CLQ change. With serious water and soil loss, the CLQ will decline. The pH has a negative impact on the SP. With the increase in the pH, the salinization degree of cultivated land will increase, which is detrimental to crop production.
The CLP indirectly reflects the CLQ. In the CLP, the CPL's driving force is larger than that of the GY. In addition, the CLQ and EE also have significant positive effects on the CLP. The EE has a greater impact on the CLP, and it is the most powerful influencing factor. The CLQ also has an important positive effect on the CLP. The SP has no significant effect on the CLP, but the SP indirectly affects the CLP by affecting the CLQ. Therefore, our research found that some factors may play an important role in the soil quality, but have little impact on the yield. There is fertile soil in the BSRNC, and the SP may have fully managed the yield limiting factors to achieve a high yield. This is consistent with the study by Chen et al (2013).
The EE has a great driving effect on the CLQ, which indirectly affects the CLP through the SP. Climate factors such as temperature and precipitation have positive driving forces, thus driving the soil material transport and affecting crop growth. Climate conditions play a controlling role in crop production. The NDVI, DVI, and RVI have relatively large driving forces. The NDVI can reflect the cultivated land's fertility, DVI reflects the soil moisture, and RVI, as an important indicator reflecting the soil degradation degree, shows the degree of stress suffered by the CLQ. The HQI reflects the cultivated land ecosystems' stability and has a positive driving force for the CLQ. Among the potential variables of the SE, the GRC is the main driving factor. In the economic development and urban construction processes, the cultivated land area is constant changing. On the one hand, a large amount of cultivated land is occupied due to new urban areas being constructed and the increase in land used for transportation and construction. On the other hand, with agricultural mechanization and intelligence developing, a large amount of uncultivated land has been transformed into cultivated land. The cultivated land's size plays an important role in the agricultural ecological stability. In addition, although the PCC's factor load is small, the population factor can not be ignored, and the increasing population will inevitably cause pressure on the food demand. The LRR and PCG reflect cultivated land's importance and the agricultural development degree in the evaluation unit, while the MCI reflects the proportion of effective planting areas of cultivated land, which has a positive driving role.

Conclusions
In this study, a CQE system based on county units was constructed to evaluate and classify the CLQ using the BSRNC as the study area. Meanwhile, a coupling coordination degree model was constructed by integrating the SP, CLP, EE and SE subsystems to judge the CLQ's coupling coordination status, and the driving mechanism was analyzed using the PLS-SEM. The main conclusions are as follows: (1) The CLQ's coupling coordination degree in most BSRNC counties is at a coordinated level, but the coordinated development degree of most counties is at the edge of coordination and disorder. In general, the coupling coordination degree in the Northeast is higher than the coordinated development degree. We should pay attention to the balance and coordination of multiple subsystems in cultivated land management, which is the high-quality state of CLQ development.
(2) The quality of the BSRNC's cultivated land is generally good, with an average grade of 3, which is at a high level. The spatial pattern of 'low in the southwest, high in the east, south and north' is presented. For regions with high CLQ, we should adopt appropriate management methods to prevent their degradation. For regions with low CLQ, focus on improvement and governance in combination with various systems and factors is necessary.
(3) There is a complex driving relationship between the SP, CLP, EE, SE and CLQ, in which the driving forces from large to small are: SP > EE > CLP > SE. The CPL, NDVI and SOC are the main driving factors. Among them, the SP has the main driving role. In addition to the SOC, SMC and ST are also very important. Regarding the ecological environment, climate factors are closely related to the CLQ. In addition, the cultivated land area and productivity, economic population, and other factors also impact it. Therefore, we can combine the results of this study's results to effectively improve the CLQ against different influencing factors.
The black soil grades' spatial distribution in Northeast China has a significant aggregation feature, which provides a scientific basis for black soil management. To maintain or improve the level of cultivation and utilization of the cultivated land, it is necessary to implement various land management measures. The southwest of Northeast China is a low-value area of CLQ distribution. Therefore, further improving the cultivated land's productivity by implementing some control and governance measures is necessary. The evaluation index system constructed in this paper is systematic and representative to a certain extent, and it can objectively reflect the CLQ's diversity and functions. This study provides a new evaluation method for agricultural land classification, CLQ improvement and sustainable use, and it can also be used for reference by other similar countries and regions. Moreover, this study's results are only the results of one evaluation. However, the CLQ is a continuously changing concept. Therefore, in subsequent research, the CLQ should be dynamically monitored, and its change pattern should be analyzed spatially and temporally. It facilitates a more real-time understanding of the cultivated land's status, as well as protection and management.