Evaluation of soil quality in major grain-producing region of the North China Plain: Integrating minimum data set and established critical limits

Highlights

• Three MDSs were established for soil quality (SQ) assessment in the North China Plain.

• MDS derived from norm values of soil variables was more accurate in SQ assessment.

• Critical values of key indicators were established on farmers’ field conditions.

• Identified key soil attributes that influencing site specific crop productivity.

Abstract

Understanding of the nexus between soil quality (SQ) and productivity, the heterogeneity of SQ and the restricting indicators that affect crop productivity is essential for a specific region to encourage improving of SQ while contributing to profitable crop production. The objectives of this study were to develop minimum dataset (MDS) variable selection methods that reflects the total dataset (TDS) of soil properties for SQ assessment and crop yield prediction in Yucheng County, a major grain-producing region in the North China Plain. Critical values of soil properties for standardized scoring functions were established based on local crop yields in field condition and the key restricting indicators limiting crop productivity were evaluated. In this study, 99 topsoil (0–20 cm) samples were collected and 14 soil properties including pH, electrical conductivity, clay content, soil organic matter, cation exchangeable capacity, total nitrogen, phosphorus (P) and potassium (K), available P, K, Cu, Mn, Mo and Zn were analyzed. Three MDSs were created by different variable selection methods using principal component analysis (PCA) of soil properties. Among these MDSs, the indicator of norm values derived from the total principal component loadings was considered to be the most suitable method for evaluating SQ. The MDS-based soil quality index (SQI) had higher correlations with the TDS-based SQI and crop yield. The critical values of soil properties were identified by regressing soil variables with local yields of maize and wheat in the field condition. The SQIs calculated using measured critical values were more accurate than that based on published empirical values. The evaluation results revealed that the calculated SQI accounted for 28% of the variation in crop yields, and there probably are some strong limitations for crop yield through soil characteristics (soil texture, salinization, and nutrient supply, etc.). Thus, management practices and amelioration of the above soil-limiting factors should be comprehensively considered in agricultural production. The finding can enable policy making and policy implementation in relation to SQ in the future.

Introduction

A significant degradation in soil quality (SQ) is occurring worldwide through adverse effects on these soil physical, chemical and biological properties and inorganic and organic contamination caused by human activities (Arshad and Martin, 2002, Sánchez-Navarro et al., 2015, Nehrani et al., 2020). Hence, addressing this challenge requires a comprehensive understanding of SQ status, for the purpose of maintaining both considerable SQ level and crop production in agricultural regions.

Up to the present, many methods of SQ evaluation have been developed (Karlen et al., 1993, Nosrati, 2013, Pulido Moncada et al., 2014, Rahmanipour et al., 2014, Shukla et al., 2006), such as soil quality index (SQI) method (Doran and Parkin, 1994, Doran et al., 1996), dynamic variation of SQ models (Larson and Pierce, 1991), SQ cart design and test kit (Ditzler and Tugel, 2002), geostatistical methods (Sun et al., 2003) and multiple variable indicator kriging methods (Qi et al., 2009), to improve land use management and identify high risk areas for the sustainable use of land resources (Bindraban et al., 2000, Andrews et al., 2004, Li et al., 2019). Among these, the SQI method has become an effective way to evaluate the heterogeneity of SQ (Andrews et al., 2002, Askari and Holden, 2014, Brejda et al., 2000, Li et al., 2019). In the world, SQI has been broadly carried out to evaluate SQ under different agricultural management at different scales and in many regions (Chahal and Van Eerd, 2019, Cheng et al., 2016, Jahany and Rezapour, 2020, Qiu et al., 2019, Viaud et al., 2018, Wang et al., 2019). However, how to objectively integrate quantitative and qualitative SQ indicators (such as soil physical, chemical and biological properties) to calculate comprehensive SQIs is still the major challenge, especially for large-scale SQ assessemnent (Askari and Hloden, 2014; Guo et al., 2017, Obade and Lal, 2016, Yu et al., 2018), which is mainly due to the expensivelabour and capital needed to acquire enough data for accurate assessment in large-scale soil monitoring campaigns. Selecting the most representive indicators is therefore important to reduce costs and data redundancy (Lima et al., 2008, Raiesi, 2017, Bünemann et al., 2018, Wu et al., 2019), while providing enough data for a reliable and accurate assessment. Previous attempts, for example, Larson and Pierce (1991) first suggested a minimum data set (MDS) consisting of a number of variables to describe the quality of a soil. Some authors also suggested using the Delphi data set (DDS), which was selected according to the importance of given variables to overall SQ based on expert opinion, to reduce the number of variables (Herrick et al., 2002). However, Qi et al. (2009) found MDS to be more accurate than DDS due to the subjectivity of expert opinions in the establishment of DDS. Moreover, regardless of the method used, some useful information from the total dataset (TDS) will be lost. Therefore, it is necessary to select the most representative index as a MDS and quantify the degree of information loss when using the MDS.

The SQI method typicallycomprises a four–step procedure: choosing appropriate variables, converting variable values to scores, determining variable weights, and integrating variable scores into an SQI (Rezaei et al., 2006, Yao et al., 2013). During the scoring of indicators, standardized scoring functions (SSFs) are often used to score soil variables and normalize the measured variable values (Diack and Stott, 2001) before they are integrated into an SQI (Glover et al., 2000, Masto et al., 2007, Qi et al., 2009, Askari and Holden, 2014). Linear and non-linear methods are usually applied to transform the key soil properties (Rahmanipour et al., 2014, Yu et al., 2018). For all SSFs, the shape of the SSF curve is determined by function types and critical values. Therefore, the accurate estimation of critical values is crucial for SQ assessments using SSF. Unfortunately, the establishment of critical values of indicators is difficult for a given study area. Critical values are soil property values where the crop yield reaches high productivity when the measured soil property is at an optimal level or has low productivity when the soil property is at an unacceptable level (Glover et al., 2000, Merrill et al., 2013, Biswas et al., 2017). In many cases, critical values based on expert opinion, published values, or measured values observed under near-ideal soil conditions for a specific site and special crop species were directly extended to large areas (Bhaduri and Purakayastha, 2014, Manna et al., 2005, Li et al., 2013). However, previous studies comparing some indices under different standardized methods and found the existing models that performed better to be much site-specific (Rahmanipour et al., 2014, Obade and Lal, 2016, Guo et al., 2017). Given the complexities of crop yield response to critical values, perhaps, the best we can do is to build these values of key soil indicators based on local crop yield.

As one of the most important agricultural regions in China, the North China Plain accounts for approximately 19% of the nation's total agricultural area (Liu et al., 2001). Wheat-Maize double cropping system played a very important role in agriculture production in this area, supplying more than 75% and 32% of the nation's total wheat and maize, respectively (Chang et al., 2020). Crop yield is considered here as a suitable integrator of the SQI because it is a primary concern to farmers in most cases (Merill et al., 2013; Bai et al. 2018). The crop yield in specific condition (ranging from low to high) could be useful in developing a guideline for interpretation of individual key soil indicator (Lopes et al., 2013). Thus, a reliable and accurate envaluation of SQ and an exploration of restricting indicators based on crop yields could provide a theoretical basis for further developing a win–win strategy combining profitable crop production with improvement of SQ. Since SQ evaluation based on the establishment of critical limits of indicators in the farmers' field conditions is limited in North China Plain. Here we developed a framework to evaluate and compare SQ in Yucheng County in the North China Plain using a TDS and some MDSs of physical and chemical soil variables. The specific objectives of this study were: (a) to objectively identify the MDS consisting of key soil variables; (b) to obtain critical values for SSFs based on local crop yields in the real field condition; (c) to quantify SQI according to the MDS and critical values; and (d) to evaluate the restricting indicators that affect crop productivity. We hope this study can improve our understanding of the SQ status in North China Plain and provide an underlying basis to select indicators for future monitoring of SQ and production potential in other agricultural areas. And the finding can enable policy making and policy implementation in relation to SQ in the future.