Assessing the Alteration of Soil Quality under Long-Term Fertilization Management in Farmland Soil: Integrating a Minimum Data Set and Developing New Biological Indicators

Abstract

The key role of soil quality improvement in achieving sustainable agricultural development based on highly intensive use of farmland is increasingly being recognized, as is the ponderance of suitable evaluation of the soil quality. The overarching goal of this study was to determine an accurate assessment framework by the comparison of the scoring function (linear and non-linear) and integration method (area and weighted additive), which integrally evaluates the soil quality of an eleven-year field fertilization experiment (including CK, no fertilizer; CF, conventional fertilization; SF, formulated fertilization; SFO, SF with organic fertilizer). Thirty-three properties, including eighteen physiochemical-related and fifteen biological-related properties, associated with soil functions were measured as potential soil quality indicators, and the soil multifunctionality (SMF) was applied to validate the soil quality indices (SQIs). Principal component analysis and relationship analysis were used with indicators sensitive to management to determine a minimum data set (MDS). The results showed that the electrical conductivity, large macroaggregate-associated total nitrogen, small macroaggregate-associated organic carbon, carbon fixation, and enzyme activities of phenol oxidase and cellulase were chosen as the MDS. All the SQIs were significantly correlated with the SMF (p < 0.05). The fertilization strategies affected most indicators in different ways, and the index developed using the non-linear function and weighted additive integration method (SQI-NL) had the best sensibility and discriminability. The SQI value with the SQI-NL-MDS method was higher following the fertilization treatments than that of no fertilizer (p < 0.05), and the treatment of the organic fertilizer had the highest SQI value (0.66). Soil quality evaluation in long-term fertilized farmland suggested that the soil quality constraints between treatments of synthetic and organic fertilizer are related to the soil functions of nutrient cycling and sustain biological activity due to their higher contribution rates to the SQI in the organic fertilizer treatment, which provides insights into ways to reduce the gap in soil quality. The framework method can provide an accurate quantitative tool for the evaluation of soil quality from the target indicators by bridging management objectives and field-level actions.

1. Introduction

Soil quality is one of the basic support factors that, together with farmland management practices, determine crop productivity [1,2]. However, the intensive use of farmland has resulted in a weakening of soil functions and an increasing decline in soil quality [3]. Fertilizer input is commonly recognized as an effective farmland management practice in agricultural systems [4,5]. Unfortunately, owing to both the unreasonable application of synthetic fertilizer and the insufficient return on organic material, notable soil degradation, including the physical structure, organic matter content, and microbial community, has occurred [6,7,8]. In recent years, the improvement of soil quality has been increasingly regarded as a foundation for maintaining agricultural sustainability, and food insecurity could be even more exacerbated through the persistent degradation of farmland soil [9,10]. Therefore, the appropriate fertilization regimes with organic and inorganic fertilizers are important to promote soil functions and to improve soil quality in farmland. However, an accurate, robust, and composite evaluation of the impacts of different fertilizers on the soil quality based on long-term field trials remains a challenge.

Soil quality, an indicator of sustainable farmland productivity, can help sustain biological productivity, maintain or enhance environmental quality, and support plant and animal health, which integrates dynamic and inherent attributes of soil [11,12]. So, soil quality indicators are a prerequisite for quantifying and evaluating soil quality [13]. Numerous studies integrate a series of soil’s physical, chemical, and biological attributes into a single evaluation value for soil quality to explore the benefits of farmland management practices [14,15]. Saurabh et al. [16] used the total data set of three physical properties and four chemical properties, as well as three biological properties, to assess the impact of different residue management practices on the soil quality in rice–wheat-cropping systems. A total of 26 indicators, including 21 physicochemical properties and 5 biological properties, were selected for use in soil quality assessment in a smallholder-dominated agricultural system [17]. Obviously, two overarching soil characteristics (physical and chemical) have accounted for a large proportion of past studies on soil quality evaluation [18]. Although soil’s chemical and physical properties are all-important in relation to soil quality evaluation by supervisors and soil’s biological properties, which are associated with a series of underpinning soil processes and are more sensitive to disturbance, which could reveal the soil quality from a broader perspective [19,20]. Studies use a large number of indicators to describe soil quality, with microbial detection techniques and soil science providing opportunities for the development of new biological indicators [21]. Moreover, Guo, in 2021 [22], warned that information about any single overarching soil characteristic cannot holistically reflect variations in the soil quality or underlying causes. Whilst the necessity of assessing soil quality is broadly recognized, the broad application of biological indicator-based evaluation to soil quality assessment is lacking.

Accurate and stable soil quality assessment results depend on the applicable quantitative approaches [23]. Different approaches have been developed for soil quality assessment [6,24,25]. The soil quality index approaches are most generally used based on integrating several soil quality indicators in contact with soil functions into an index, which can be used specifically for objective decision-making [26]. However, this approach requires a combination of mathematical and statistical methods to accurately and reasonably provide the weights and scores of the indicators. Recently, the soil quality index approach has also been used in some studies due to its simple calculation processes [24]. The approach that chooses a suitable quantitative model of soil quality assessment is essential [23]. Therefore, practices are required that demonstrate the advantages of a soil quality assessment approach before it is extensively employed.

Fertilization management induces a series of variations in soil’s chemical, physical, and biological properties and the crop yield [27,28]. It is widely recognized that a large amount of synthetic fertilizers once achieved higher crop yields, but the functionality and quality of the soil are deteriorating [29,30]. However, in previous studies on the formation of evaluation methods, the increase in crop yield was usually used as a criterion to verify soil quality improvement under fertilization practices, which is unreasonable. In consideration of retaining and increasing soil quality for the sustainable utilization of farmland, the superiority of using soil multifunctionality review quantification methods of soil quality in the present study. In this study, we focused on understanding the changes in soil quality and its precise evaluation methods using the unique soil biological indicators, complemented by different quantitative models of soil quality, using an eleven-year long-term inorganic and organic fertilized field experiment. The specific objectives were to (1) evaluate how the organic fertilizer affects soil quality and potential properties based on soil multifunctionality; (2) identify the most appropriate and accurate quantitative models of soil quality; and (3) establish a minimum data set with biological significance for quantifying soil quality under long-term fertilizer use in subtropical regions of China. Given the sensitivity of soil microorganisms to disturbance, we hypothesize that combining new soil biological indicators from functional genes associated with C and N biogeochemical cycling can improve the accuracy of soil quality evaluation.

2. Materials and Methods

2.1. Site Description and Experiment Design

A long-term field experiment was initiated in 2009 in Dongyang, Zhejiang Province, China (29.29° N, 120.24° E), characterized by a typical subtropical monsoon climate. The region has an annual temperature of 11.2 °C and an annual precipitation of 779 mm. The soil type is yellow–red soil (Alfisols, USDA soil taxonomy), and the soil contained 13.8 g kg⁻¹ of organic matter, 0.5 g kg⁻¹ of total nitrogen, and 5.42 pH in tillage depth at the start of the field experiment.

The field experiment encompassed a single-cropping rice (Oryza sativa) system and followed a randomly repeated design with four treatments, including no fertilization (control, CK), conventional fertilization (chemical fertilizer, CF), formulated fertilization (chemical fertilizer, SF), and SF with organic fertilizer (chemical fertilizer + spent mushroom substrate, SFO). Each treatment had 3 replicates, which were applied to 12 plots (33 m² for each). For CF treatments, urea (210 kg N per ha), superphosphate (60 kg P per ha), and potassium chloride (70 kg K per ha) were applied according to the local fertilizer application protocol. For SF treatments, the optimal amount and an appropriate proportion of inorganic fertilizer input (including 225 kg N per ha, 85 kg P per ha, and 105 kg K per ha) is given based on the actual situation of local soil before the start of the experiment. The fertilizer of spent mushroom substrate was produced by Kingenta Ecological Engineering Co., Ltd. (Linyi, China), which contained 1.8% total N, 2.2% total K, 1.6% total P, and 45.0% total C. The chemical and organic fertilizers were incorporated into the soil by the tillage as basal amendments before planting upland rice. Other management practices were the same throughout the experimental period according to the local farming management.

2.2. Soil Sample Collection and Analytical Determinations

Soil samples were obtained at the ripening stage (November 2021) of upland rice. Five soil cores (3 cm diameter) in each experimental plot (0–20 cm) were collected and mixed thoroughly as a composite sample. Each fresh soil sample was manually sieved (2 mm) to remove plant matter and stones and subsequently divided into two subsamples: one part was stored at −80 °C for DNA extraction, while the other was air-dried at room temperature to determine chemical properties.

The detailed procedures of analysis for soil physical, chemical, and biochemical properties were described by Ying et al. [31]. Briefly, the standard protocols are used to measure soil samples (Lu, 1999) [32]: (1) soil pH (1:2.5 soil–water suspension), (2) total nitrogen (TN, Kjeldahl, and Digestion), (3) available nitrogen (AN, NaOH hydrolysis), (4) total phosphorus (TP, spectrophotometer detection), (5) available phosphorus (AP, colorimetric analysis), (6) electrical conductivity (EC, soil/water 1:2.5), (7) available potassium (AK, flame photometer analysis), (8) organic carbon (SOC, potassium dichromate oxidation), and (9) aggregate distribution (comprising three classes: large macroaggregates (LA, >2 mm), small macroaggregates (SA, 2–0.25 mm), and microaggregates (MA, <0.25 mm), wet sieving method as described by Six et al. [33]). Activities of extracellular enzymes, including phenol oxidase (E-PPO), peroxidase (E-POD), β-1,4-glucosidase (E-GC), cellulase (E-CL), β-Nacetyl-glucopyranoside (E-NAG), and glutaminase (E-GLS), were measured using the modified universal buffer (MUB)-linked model substrates method [34]. Total DNA was extracted from 0.50 g fresh soil using a Fast DNA Spin kit (MP Biomedical, Santa Ana, CA, USA) in accordance with the manufacturer’s instructions. High-throughput sequencing was used to uncover the diversity (SHA) and composition (BC) of bacteria in the DNA samples based on the examination of V3-V4 hypervariable regions of the 16S rRNA gene [7]. The sequencing was carried out by Allwegene Company in Beijing, China utilizing an Illumina MiSeq platform. Twenty-seven functional genes associated with C and N biogeochemical cycling in soil were quantified using Quantitative Microbial Element Cycling [35] and subsequently categorized as seven functional processes (regarded as new biological indicators) based on gene function: methane metabolism (G-MM), carbon degradation (G-CD), carbon fixation (G-CF), nitrogen fixation (G-NF), ammonium oxidation (G-AO), denitrification (G-DE), and ammonification (G-AM). A detailed introduction to the functional gene is in Table S2.

2.3. Soil Quality Assessment

2.3.1. Determining the Minimum Data Sets (MDSs)

Sustainable farmland uses and soil multifunctionality were the main objectives of assessing soil quality regarding the cropping system in the present study (Figure 1). Thirty-three indicators (considered as TDS) from soil biological, chemical, and physical properties can cover a wide range of soil functions (such as physical stability and support (F1), nutrient cycling (F2), water relations (F3), and sustaining biological activity (F4)), and, meanwhile, can predict sensitively variation of soil quality. The soil functions are sufficiently independent and yet together reflect the complex multidimensionality of soil quality. Considering the information redundancy among soil indicators (high relationship), it is necessary to select representative indicators to accurately evaluate soil quality. A minimum data set of indicators (MDS) can be determined through combined principal component analysis (PCA) and correlation analysis. Briefly, the variables with absolute values within 10% of the highest factor loading in each retained principal component (PCs, eigenvalues ≥ 1) were defined as highly weighted variables. When multiple variables are found in a PC, only variables with the highest correlation sum are determined as indicators of MDS. The detailed indicator screening process can be found in our previous studies [6]. In addition, we also constructed minimum data sets 1 and 2 to evaluate the suitability of new biological indicators.

2.3.2. Indicator Scoring

To avoid the impact of unit differences among soil indicators, non-linear and linear scoring functions were used to assign a score (unitless value) between 0 and 1 for each indicator. The suitable scoring algorithm was selected by interpreting the soil functions of each indicator based on the objective of soil sustainability, as described below.

For non-linear scoring (NL), the following equation is used:

$${\mathrm{S}}_{\mathrm{N}\mathrm{L}}={\displaystyle \frac{a}{1+{\left(\frac{x}{x_0}\right)}^b}}$$

(1)

where S_NL is the score of the soil indicator; x is the measured value of the indicators; x₀ is the mean value of the soil indicators; a is the maximum score which was equal to 1 in this study; and b is the slope assumed to be −2.5 for “more is better” and 2.5 for “less is better” [6,36].

For linear scoring (L), two score function equations (“more is better”: Equation (2); “less is better”: Equation (3)) were used as follows:

$${\mathrm{S}}_{\mathrm{L}}=\left\{\begin{array}{c}0,x\le L\\ {\displaystyle \frac{x-L}{H-L}}\\ 1,x\ge H\end{array}\right.,L<x<H$$

(2)

$${\mathrm{S}}_{\mathrm{L}}=\left\{\begin{array}{c}1,x\le L\\ {\displaystyle \frac{H-x}{H-L}}\\ 0,x\ge H\end{array}\right.,L<x<H$$

(3)

where S_L is the score of indicators; x is the measured value of the soil indicators; and L and H are the lowest and highest values of soil indicators, respectively [37].

2.3.3. Developing and Validating Soil Quality Indices

The scores of indicators were integrated into soil quality indices using the area approach (Equation (4)) [38] and the weighted additive (Equation (5)) [39] method.

$${\mathrm{S}\mathrm{Q}\mathrm{I}}_{\mathrm{a}\mathrm{r}\mathrm{e}\mathrm{a}}=0.5\times \sum _{i=1}^n{S}_i^2\times sin({\displaystyle \frac{2\pi }{n}})$$

(4)

$$\mathrm{S}\mathrm{Q}\mathrm{I}=\sum _{i=1}^n{W}_i\times S_i$$

(5)

where n is the number of indicators integrated into the index; S_i is the indicator score (linear or non-linear); W_i is the weighting value of each indicator, which was determined by the ratio of its communality with the sum of communalities of all indicators. Four soil quality indices based on two methods of soil indicator scoring and soil quality quantification were compared in this study, as follows: SQI-NL, SQI-L, SQI_area-NL, and SQI_area-L (Figure 2). Soil quality index value may be divided into five different classes according to Zhang et al. [40]: very high (>0.80), high (0.60–0.80), medium (0.40–0.60), low (0.20–0.40), and very low (<0.20).

2.4. Data Analyses

The function indicators, including TN, C/N, TP, AN, AP, AK, SHA, pH, SOC, E-PPO, E-POD, E-GC, E-CL, E-NAG, E-GLS, and functional genes were Z-score transformed and then were integrated to obtain a soil multifunctionality index using an averaging approach. Different functional genes were devoted to representing different soil bio-processes (including G-MM, G-CD, G-CF, G-NF, G-AO, G-DE, and G-AM) based on the first principal components from the PCA results, respectively.

All statistical analyses (Pearson correlation, analysis of variance, and principal component analysis) were performed using SPSS 18.0 (SPSS, Chicago, IL, USA). There were significant differences (p < 0.05) among all the treatments for all the soil indicators, followed by the LSD test. Pearson correlation coefficients were used to test the relationships between soil quality indices and soil multifunctionality index. All figures were generated with R software version 4.1.0.

3. Results

3.1. Soil Properties and Indicators Scores

Long-term fertilization improved most of the soil properties, except for soil bacterial composition (Figure 3A). Higher soil EC, TP, AN, AP, AK, LA, LA-C, E-GLS, G-CF, and G-DE were detected in SFO treatment than in other treatments. The application of organic fertilizers increased almost all soil indicator scores regardless of the method of non-linear and linear scoring (Figure 3B,C). However, the polygons formed by the score of soil indicators in each treatment using non-linear scoring functions are more regular than those using linear scoring.

3.2. Minimum Data Sets

Given the significant difference in soil properties among all treatments, thirty-two soil indicators except for BC were chosen for subsequent analysis (Figure 3A). The principal component analysis results based on thirty-two soil quality indicators displayed that the first five PCs with eigenvalue >1 explained 91.71% of the cumulative variation (

Table 1. Results of principal component analysis of significant variables under different fertilizer practices in Zhejiang, China. Abbreviations as described in Table S1.

PCs	PC1	PC2	PC3	PC4	PC5
Eigenvalue	19.68	4.14	2.81	1.63	1.09
Percent %	61.50	12.95	8.78	5.08	3.41
Cumulative percent %	61.50	74.44	83.22	88.30	91.71
Eigenvectors
pH	0.59	0.51	0.31	0.25	0.13
EC	0.95	−0.19	−0.21	−0.01	−0.03
SOC	0.68	−0.69	0.12	−0.12	0.08
TN	0.85	0.10	−0.01	−0.49	0.01
C/N	−0.41	−0.75	0.07	0.41	−0.04
TP	0.93	−0.06	−0.23	0.07	0.06
AN	0.92	−0.30	−0.22	0.13	−0.02
AP	0.92	−0.34	−0.15	−0.05	0.07
AK	0.91	−0.25	−0.31	−0.06	−0.04
LA-C	0.74	−0.52	0.38	0.09	−0.01
LA-N	0.67	−0.35	0.59	0.12	0.19
SA-C	0.84	−0.13	0.11	0.01	−0.45
SA-N	0.58	0.17	0.50	0.17	−0.04
MA-C	0.82	0.32	−0.12	0.26	0.30
MA-N	0.68	0.50	0.06	0.09	−0.39
SHA	0.63	−0.16	−0.51	0.42	−0.30
LA	0.82	−0.08	−0.43	−0.11	−0.02
SA	0.87	−0.32	0.01	−0.05	−0.10
MA	−0.93	0.25	0.18	0.08	0.08
E-PPO	0.40	0.11	0.55	0.57	−0.07
E-POD	0.40	0.74	−0.15	0.30	−0.20
E-CL	0.28	0.76	−0.41	0.12	0.28
E-GC	0.86	−0.31	−0.33	0.16	0.12
E-NAG	0.86	0.10	−0.02	−0.20	0.36
E-GLS	0.98	0.06	−0.02	0.03	0.07
G-MM	0.80	0.17	0.49	−0.08	0.09
G-CD	0.83	0.09	0.45	−0.16	0.07
G-CF	0.91	0.18	0.24	−0.14	−0.09
G-AO	0.64	−0.01	−0.27	0.40	0.35
G-DE	0.94	0.26	0.11	−0.09	−0.11
G-AM	0.86	0.32	−0.05	−0.16	−0.18
G-NF	0.90	0.23	0.04	−0.25	−0.01

). The PC1 explained 61.5% of the variation and displayed ten highly weighted variables within 10% of the highest factor loading, including EC (0.95), TP (0.93), AN (0.92), AP (0.92), AK (0.91), MA (−0.93), E-GLS (0.98), G-CF (0.91), G-DE (0.94), and G-NF (0.90). They were highly correlated, and EC was chosen as a representative indicator in the PC1 to be used in the minimum data set (MDS) due to its highest correlation sum (

Table 2. Pearson correlation coefficients and correlation sums for highly weighted variables under principal components with multiple high factor loadings. Abbreviations as described in Table S1. ^a The correlation coefficient sums.

PC1 Variables	EC	TP	AN	AP	AK	MA	E-GLS	G-CF	G-DE	G-NF
Correlation coefficients
EC	1.00	0.93	0.98	0.97	0.99	0.97	0.91	0.78	0.83	0.81
TP	0.93	1.00	0.93	0.93	0.91	0.92	0.96	0.76	0.83	0.78
AN	0.98	0.93	1.00	0.97	0.97	0.95	0.89	0.69	0.75	0.71
AP	0.97	0.93	0.97	1.00	0.96	0.96	0.90	0.72	0.75	0.73
AK	0.99	0.91	0.97	0.96	1.00	0.96	0.87	0.70	0.77	0.77
MA	0.97	0.92	0.95	0.96	0.96	1.00	0.88	0.79	0.80	0.81
E-GLS	0.91	0.96	0.89	0.90	0.87	0.88	1.00	0.86	0.93	0.85
G-CF	0.78	0.76	0.69	0.72	0.70	0.79	0.86	1.00	0.96	0.94
G-DE	0.83	0.83	0.75	0.75	0.77	0.80	0.93	0.96	1.00	0.94
G-NF	0.81	0.78	0.71	0.73	0.77	0.81	0.85	0.94	0.94	1.00
Correlation sums a	9.16	8.95	8.85	8.89	8.89	9.04	9.04	8.20	8.57	8.33
PC2 variables	SOC	C/N	E-POD	E-CL
Correlation coefficients
SOC	1.00	0.24	0.30	0.36
C/N	0.24	1.00	0.55	0.65
E-POD	0.30	0.55	1.00	0.69
E-CL	0.36	0.65	0.69	1.00
Correlation sums	1.89	2.44	2.54	2.71

). Because PC1 accounted for data variation of more than 50%, we also selected the indicator with the lowest correlation sum, G-CF, to demonstrate the variation within this group. The PC2 explained 13.0% of the variation and displayed four highly weighted variables from SOC (−0.69), C/N (−0.75), E-POD (0.74), and E-CL (0.76) (Table 1). E-CL was entered into the MDS because of its highest factor loading and correlation sum (Table 2). For the PC3 (8.8% of variation), LA-N and E-PPO were highly weighted variables. LA-N was chosen for the MDS because of its higher factor loading. The PC4 (5.1% variation) and PC5 (3.4% variation) each have only one indicator that is a highly weighted variable, and thus E-PPO and SA-C were also entered into the MDS. Therefore, the final refined soil quality indicators of MDS for calculating SQI-MDS were the following: EC, LA-N, SA-C, E-PPO, E-CL, and G-CF.

3.3. Calculation and Comparison of Soil Quality Indices

Eight soil quality index models, including SQI-NL-MDS, SQI-NL-TDS, SQI-L-MDS, SQI-L-TDS, SQIarea-NL-MDS, SQIarea-NL-TDS, SQIarea-L-MDS, and SQIarea-L-TDS, were calculated with non-linear and linear scoring of indicators (Figure 2). Considering the characteristics of the soil quality indicators reflecting soil function, total indicators were used in the “more is better” type for scoring, except for C/N and MA (“less is better”). For SQI with the weighted additive method, the soil indicator weights from TDS and MDS were assigned by PCA as shown in Table S3. The comparative eight soil quality index models are shown in Figure 4 and Figure 5. The correlation results of the soil quality index from scoring functions and SMF showed that the R-value of the non-linear scoring function was higher than those of linear scoring, regardless of the SQI method from the area or weighted additive (Figure 4). Further, the positive correlations (p < 0.01) between soil quality indices from MDS and TDS were found for two SQI methods (Figure 5). Similarly, the soil quality indices showed that the weighted additive method was more sensitive than the area method because of the higher correlation coefficient, regardless of non-linear or linear scoring functions (Figure 4 and Figure 5).

3.4. Evaluation of Soil Quality under Different Fertilizer Treatments

Based on the SQI-NL-MDS method, we evaluated the long-term influence of different fertilizers on soil quality. Overall, the soil quality index was significantly increased by fertilization practices (Figure 6). The value of the soil quality index under the four treatments ranged from 0.25 to 0.66, which was the greatest in the SFO treatment (reaching a “high” soil quality level). Compared to CF treatment, the soil quality index value in treatments of SF and SFO increased by 11.0% and 45.6%, respectively. Gaps 1 and 2 between CF, SF, and SFO are mainly due to low scores from soil function of F2-4 and F1-3, respectively.

4. Discussion

4.1. Fertilization Effects on Soil Quality Indicators

In this study, the soil quality indicators were selected by focusing on soil functions and sustainable farmland productivity to reflect the effect of different fertilization strategies. Thirty-three indicators, including eighteen physiochemical-related and fifteen biological-related properties, in response to four treatments with noticeable changes, uncovered that the influences of optimizing fertilization on soil quality indicators were positive and related to multiple soil functions. Previous reports have illustrated that additional application of organic fertilizers undoubtedly could enhance organic matter, improve aggregate structure, and promote microbial activity in farmland soil [41,42]. This was also consistent with our results, which revealed that organic fertilizer treatment had a higher effect on the turnover process for microorganisms in the soils where they improved nutrient status and physical-related properties (Figure 3) [7]. Further, the principle by Pareto in 1964 [43] indicated that four to five variables may support a description of soil information that is eighty percent as comprehensive if twenty variables can comprehensively describe soil properties. Hence, considering that redundant information caused by high collinearity among soil properties, a minimum data set (MDS), consisting of EC, LA-N, SA-C, E-PPO, E-CL, and G-CF, was established for soil quality evaluation. It is understood that almost all MDS indicators have been identified as crucial soil quality indicators in previous studies [3,44,45]. Pivotal indicators are considered one of the core factors determining the accuracy of soil quality evaluation and can effectively distinguish the gap in soil quality among different fertilization treatments [40]. The physical preservation of soil organic carbon and total nitrogen is based on the positive associations between soil structure and organic matter, then they are very important soil quality indicators [46,47]. Liu et al. [48] found that organic fertilizer can advance soil organic carbon and promote enzyme activity in the soils involved in the C, N, and P cycling, which is consistent with the findings of our study. The activity of soil enzymes and microorganisms is vital for soil quality because of their role in soil organic matter decomposition, nutrient turnover, and physical structure stabilization [49,50].

4.2. Comparison of Different Methods for Calculating Soil Quality Index

In order to be a practical and robust tool for evaluating soil quality, a scoring function and integrating method must be able to detect differences as a result of variations in farmland management practices [51,52]. The scoring function of the non-linear method was recommended as a proper approach for quantitative soil quality [3,53,54], although the linear method is considered uncomplicated and utility [55]. This was supported by our results that SQI-NL-TDS (or SQI_area-NL-TDS) have a higher R value of correlation with SMF than the SQI-L-TDS (or SQIarea-L-TDS) (Figure 4). Further evidence of the correlation between SQIs from TDS and MDS also confirms this result (Figure 5). This may be due to the greater variance of indicator scoring using linear equations (Figure 3), then it might be assumed that the soil quality indexing using non-linear equations is more representative of system functionality than the linear indices [56]. In terms of integrating method, SQI-TDS showed stronger correlations with SMF and SQI-MDS than the SQIarea-TDS regardless of the scoring functions (Figure 4 and Figure 5), meaning that SQI has a more discriminate and sensitive effect than SQI_area.

4.3. Impacts of Long-Term Fertilization on Soil Quality

Soil quality evaluation has widely been reported to determine sustainable practices of farmland management and large fluctuations in soil quality are caused by long-term interactions between crops and soils as well as strongly disturbed by field management [57,58]. Our study demonstrates that long-term organic fertilization treatment is superior to synthetic fertilizer treatments for soil quality (Figure 6). Consistent with many previous studies, sustainable agricultural intensification from organic input could prevent the decline of soil quality [6,59,60]. This may be because of the comprehensive benefits of organic fertilizer for soil structure, nutrient release, capacity to preserve moisture and fertility, and microbial activity [48,61]. Compared to no fertilization treatment, the magnitude of the improvement of soil quality was significant with synthetic fertilizer treatments (Figure 6). However, Fallahzade and Hajabbasi, 2012 [62] suggested that despite these soils exhibiting instant enhancement in soil quality by chemical fertilization in existing conditions, these inputs will not have an impact in the long term on sustainable soil quality. Our results showed that the soil quality index gap between CK (0.25) and SF (0.50) treatment is mainly due to the low contribution of electrical conductivity and biological indicators (Figure 6). Moreover, the gap between treatments of synthetic and organic fertilizer is mainly due to the performance of soil functions of nutrient cycling and sustain biological activity that reflected by MDS indicators. Similar results were found by Li et al. [3], who emphasized a better soil function for soil quality under green manure addition. Consequently, nutrient and microbial activity management was one of the crucial farmland practices to affect soil quality levels [40]. Overall, the researcher needs to focus on the feasibility of practitioners and the interpretability of soil quality index outputs when evaluating existing soil quality indices, which is important for better farmland soil management.

4.4. Evaluation of the Suitability of New Biological Indicators

Increasing biological indicators can greatly promote soil quality evaluation due to the core role of soil organisms in soil functioning [63]. In this study, seven new biological indicators were developed using twenty-seven functional genes to evaluate soil quality [31,64]. To evaluate the suitability of new biological indicators, MDS1 and MDS2 were also established to form soil quality indices (

Table 3. Different minimum data sets.

Sets	Soil Quality Indicators
MDS	EC	LA-N	SA-C	E-PPO	E-CL	G-CF
MDS1	EC	SA	SHA	E-PPO	E-CL
MDS2	EC	LA-N	SA-C	E-PPO	E-CL	SOC

). MDS1 is constructed by screening soil properties other than new biological indicators using the same method as establishing MDS (Tables S4 and S5); MDS2 is constructed by replacing the new biological indicator (G-CF) in MDS with SOC, which is mainly due to the widespread use of SOC as a key soil quality indicator in previous studies [57]. Compared to SQI-MDS, the lower coefficient of variation of SQI-MDS1 among treatments (Figure 7A) uncovers that the SQI developed by conventional indicators is less sensitive to fertilization interferences. Furthermore, the lower R-value from the correlation between SQI-MDS1 and SMF confirms that SQI-MDS1 has weak accuracy compared to SQI-MDS (Figure 7B). It should be noted that this finding is not entirely valid because MDS (six indicators) contains more soil information than MDS1 (five indicators) [3]. Bünemann et al. [57] pointed out that using additional or novel soil quality indicators in the MDS is meaningful. For this reason, we used the equal quantity indicator data set (MDS vs. MDS2) for comparison and found that SQI-MDS has a higher coefficient of variation and R-value (relative to SMF) compared to SQI-MDS2 (Figure 7). In total, the results supported our hypothesis that combining new biological indicators to evaluate changes in soil quality under long-term fertilization management has better discriminability and accuracy.

5. Conclusions

The soil quality index developed using a non-linear scoring function and weighted additive integrates method had the best sensibility and discriminability based on the long-term fertilization management regimes. The integrating method of the weighted additive represented a satisfactory balance in terms of the accuracy and discrimination of calculating the soil quality index, although its simplicity is inferior to the area method. In the present study, new biological indicators were developed to integrate associated soil functions. Six soil indicators including EC, LA-N, SA-C, E-PPO, E-CL, and G-CF were chosen as a minimum data set, which can be applicable to assess soil quality for long-term organic fertilization. The values of the soil quality index of treatments followed the order of SFO, SF, CF, and CK. Organic fertilization has made an evident contribution to improving soil quantity by regulating soil functions of nutrient cycling and sustaining biological activity. Moreover, the soil quality indices were positively correlated with soil multifunctionality. These new biological indicators from functional genes associated with C and N biogeochemical cycling have implications for our understanding of soil quality changes with long-term nutrient management. Finally, due to the complexity of soil organisms and their functions, more soil biological indicator studies are needed, which could contribute to a deeper understanding of soil quality, and then promote sustainable farmland use.