Comprehensive Benefit Evaluation of Technological Models for Fertile Topsoil Restoration in Thin-Layer Black Soil Region: Evidence from Farmer Survey Data in the Southern Songnen Plain, China

Abstract

The severe degradation of thin-layer black soil in the Southern Songnen Plain threatens both regional agricultural sustainability and national food security. While various fertile topsoil restoration technologies have been proposed, a systematic evaluation of their comprehensive benefits is lacking, hindering effective policy and technology promotion. This study addresses this gap by employing an entropy weight–fuzzy comprehensive evaluation method to assess the economic, social, and ecological performance of four predominant restoration models—no-tillage, strip-tillage, deep-tillage, and indirect return—using survey data from 263 farmers. Results identify strip-tillage as the optimal model, achieving the highest integrated benefit score (8.153) by successfully balancing superior economic profitability and social acceptance with robust ecological performance. Although no-tillage excels in ecological benefits like moisture conservation (8.901) and pesticide reduction (8.524), its economic potential is constrained by higher management costs. Deep-tillage rapidly enhances soil fertility (8.628) but is limited by high operational costs, whereas the indirect model, despite high ecological sustainability (7.781), faces adoption barriers due to technical complexity and cost. The findings underscore the necessity of moving beyond one-size-fits-all approaches. We propose a targeted promotion system based on “categorized guidance and precision adaptation”, offering a practical framework for optimizing technology deployment to support both black soil conservation and sustainable agricultural development.

1. Introduction

As one of the most fertile soil types globally, chernozem possesses a thick humus layer and exceptional capacity for water and nutrient retention, granting it a crucial role in agricultural production [1]. Located in Northeast China, the black soil region constitutes one of the world’s four major chernozem belts and serves as a vital “ballast stone” for national food security. However, prolonged intensive agricultural practices and unsustainable tillage methods have led to a severe degradation crisis of the region’s soil resources [2]. Among these areas, the southern Songnen Plain stands out for its pronounced ecological vulnerability, particularly in its thin-layer chernozem zones (where topsoil thickness is less than 30 cm), which suffer from especially acute degradation [3]. Monitoring data indicate that the black topsoil layer in this region is being eroded at an annual rate of 0.3–0.5 cm. Compared to the period prior to land reclamation, soil organic matter has declined by more than 50%, resulting in a vicious cycle of “thinning–impoverishment–ecological function decline” [4,5]. This degradation trend poses a serious threat not only to the sustainability of regional agricultural development but also to the strategic goal of safeguarding national food security.

In response to the degradation crisis of black soil, fertile topsoil restoration technologies—such as straw return and organic fertilizer substitution—have been regarded as key solutions [6,7]. However, in practical application, these technologies exhibit significant variation in effectiveness. For instance, deep tillage with straw return can rapidly increase topsoil thickness in the short term, but excessive soil disturbance may damage soil aggregate structure and accelerate the loss of organic matter through mineralization [8]. In contrast, no-tillage straw mulching effectively reduces soil erosion, yet it faces challenges such as the slow decomposition of straw under low-temperature conditions [9]. Although organic fertilizer substitution has been shown to enhance soil fertility, its large-scale promotion is constrained by the uneven spatial distribution of the livestock industry and the high costs associated with technology adoption [10]. These context-specific limitations highlight the inadequacy of conventional single-dimensional evaluation frameworks. There is an urgent need to establish a multidimensional assessment system that balances short-term productivity gains with the long-term goal of soil fertility enhancement. Most existing studies focus on the agronomic effects of individual technologies [11,12,13], lacking a systematic evaluation of the comprehensive benefits of technological models from a multidimensional perspective encompassing economic, ecological, and social dimensions. Theoretically, agricultural technology evaluation has evolved from relying solely on yield-based indicators to adopting multidimensional comprehensive assessments. Traditional cost–benefit analysis (CBA) methods are often inadequate in fully capturing the ecological and environmental value of technologies [14], while emerging ecosystem service assessments tend to overlook the economic rationality of farmers [15]. The entropy weight–fuzzy comprehensive evaluation method, by integrating objective weighting and fuzzy mathematics, effectively addresses the uncertainty and incomplete information commonly found in agricultural systems [16], offering a novel methodological tool for evaluating the comprehensive benefits of such technologies. Practically, the southern Songnen Plain, as a representative thin-layer black soil region, with its unique geographical features and diverse household structures, provides an ideal case for examining the regional adaptability of different technological models and the adoption behaviors of farmers.

Based on the above, this study takes the fragile chernozem zone in the southern Songnen Plain as its research area. Drawing on micro-level household survey data, it employs the entropy weight–fuzzy comprehensive evaluation method to quantitatively assess the comprehensive benefits—which are defined as a multi-dimensional construct encompassing economic viability, social acceptability, and ecological sustainability—of four technological models: straw mulching with no-tillage return, straw mulching with strip-tillage return, straw mulching with deep-tillage return, and indirect straw return. The study aims to uncover the adaptation patterns between different technological models and the resource endowments of thin-layer black soil regions, thereby providing a theoretical basis for policy optimization and technology selection in black soil conservation and sustainable utilization.

2. Research Area and Overview of Technological Models

2.1. Overview of the Research Area

This study selects the typical thin-layer black soil region in the southern part of the Songnen Plain in northeastern China as the research area. Specifically, survey samples were collected from four representative counties: Nong’an County and Gongzhuling City in Changchun, and Qian’an County and Changling County in Songyuan (Figure 1). Geographically, the region spans from 43°33′ to 45°32′ N latitude and 122°49′ to 125°02′ E longitude. It belongs to a temperate semi-humid continental monsoon climate zone, with an average annual temperature of 4–6 °C, annual precipitation ranging from 400 to 600 mm, and a frost-free period of 130 to 150 days. The dominant soil types in this area are black soil and chernozem, with a cultivated topsoil thickness generally between 20 and 30 cm and organic matter content ranging from 1.5% to 3.0%, exhibiting the typical characteristics of thin-layer black soil. The total arable land across these four counties exceeds 1.2 million hectares, with maize accounting for over 70% of crop cultivation, making it a key commercial grain production base in Jilin Province. There is notable variation in farming scale among local households—ranging from moderately scaled operations of 3–5 hectares to a large number of smallholders managing less than 1 hectare—offering a diverse sample for analyzing adoption differences in technological models across farm sizes. The region is also rich in surface straw resources, with an average annual straw output of approximately 8 million tons, providing a solid material foundation for implementing various straw return-to-field technologies.

2.2. Overview of Fertile Topsoil Restoration Models

(1)

Straw Mulching with No-Tillage Return-to-Field Model

This model involves directly covering the soil surface with crop straw, without any tillage operations throughout the entire process. Sowing is conducted using a no-tillage seeder under the straw cover [17]. The core advantage of this approach lies in minimizing soil disturbance to preserve the native soil structure, while significantly improving soil moisture retention [18]. Straw mulching effectively reduces water evaporation, suppresses weed growth, lowers the risks of wind and water erosion, and facilitates the recovery of soil biodiversity and the natural accumulation of organic matter.

(2)

Straw Mulching with Strip Tillage Return-to-Field Model

Based on full straw mulching, this model employs strip tillage equipment to conduct localized tillage in the seeding zone (typically 30–40 cm wide and 15–20 cm deep), allowing for partial straw burial and localized soil loosening [19,20]. Its technical advantages include maintaining the ecological benefits of erosion control from straw mulching, while improving seedbed looseness, raising soil temperature, and promoting seed germination and seedling growth [18]. This model effectively addresses potential issues in no-tillage systems such as poor sowing quality and slow straw decomposition in low-temperature regions.

(3)

Deep Tillage Straw Return-to-Field Model

This model utilizes deep-tillage to mix straw with topsoil and bury it in the 20–30 cm subsoil layer, achieving full-volume straw incorporation. Its major advantage lies in rapidly improving the subsoil structure, breaking up the plow pan, promoting uniform straw decomposition, increasing soil organic matter content, and enhancing soil water and nutrient retention capacity [21]. This creates a favorable environment for root growth.

(4)

Indirect Tillage Straw Return-to-Field Model

This model includes several forms such as biogas slurry return, composted straw return, and digested manure application (through livestock excreta) [22,23]. Through biological or chemical transformation, straw is converted into organic fertilizers before being applied to farmland. Its characteristics include thorough organic matter decomposition, effective pathogen and weed seed elimination, low pest and disease transmission risk, significantly improved nutrient use efficiency, and more balanced nutrient supply [24]. This approach also enhances the structure of soil microbial communities.

3. Materials and Methods

3.1. Research Methods

3.1.1. Entropy Weight Method

The entropy weight method is an objective weighting technique grounded in information entropy theory. Its core principle involves calculating the entropy value of each indicator to assess the degree of data dispersion, thereby determining the relative importance of each indicator [25]. The greater the dispersion of an indicator, the more influence it has on the evaluation results, and consequently, the higher the weight it receives. Compared with subjective weighting approaches, the entropy weight method effectively avoids human interference and is particularly suitable for multi-indicator comprehensive evaluation studies [26]. In this study, the entropy weight method is employed to determine the objective weights of benefit evaluation indicators for various fertile topsoil restoration models. The specific steps are as follows:

(1)

Suppose there are n types of fertile topsoil restoration models and m benefit evaluation indicators. The original data matrix of evaluation indicators is constructed as $X$

$$X=\left[\begin{array}{ccc}{x}_{11}& \dots & x_{1,n}\\ ⋮& \ddots & ⋮\\ x_{m,1}& \dots & x_{m,n}\end{array}\right]$$

(1)

Let $X=X_{{ij}_{m\times n}}$, where $X_{ij}$ represents the value of the i (i = 1, 2, …, m) benefit evaluation indicator for the j (j = 1, 2, …, n) fertile topsoil construction model.

(2)

The data were standardized. Given the diversity of benefit evaluation indicators selected in this study, standardization is necessary to eliminate the influence of different dimensions and units, ensuring comparability across data. The selected indicators fall into two categories: (a) positive indicators, where higher values indicate better performance; and (b) negative indicators, where lower values are preferable. In this study, the negative indicators are transformed by taking their reciprocals prior to standardization. The standardized processing of raw data is conducted using the Z-SCORE method, following these steps:

Step 1: Based on the raw data, calculate the arithmetic mean (µ) and standard deviation (δ) for each indicator.

Step 2: Normalize the data using the Z-score method. The formula is $X_{ij}$ = ($X_{ij}-\mu$)/δ. In this process, $X_{ij}^{’}$ represents the normalized variable value. To ensure the positive orientation of all indicators, the absolute value of the inverse was applied for reverse indicators. For convenience, the normalized value $X_{ij}^{’}$ is denoted as $X_{ij}$ in the subsequent analysis.

(3)

The proportion of the j technology mode under the i benefit evaluation indicator, denoted as $P_{ij}$, is calculated using the following formula:

$$P_{ij}=X_{ij}/{\sum }_{j=1}^m{X}_{ij} (\mathsf{i}=1, 2, \dots , \mathsf{m}; \mathsf{j}=1, 2, \dots , \mathsf{n})$$

(2)

(4)

The entropy value of the i benefit evaluation indicator is calculated using the following formula:

$$E_j=-k{\sum }_{j=1}^n{P}_{ij}ln(P_{ij}) (\mathrm{k} > 0, \mathrm{k}=1/\ln (\mathrm{n}), {\mathrm{e}}_{\mathrm{j}}\ge 0)$$

(3)

(5)

The coefficient of variation for the ith benefit evaluation indicator is calculated using the following formula:

$$D_j=1-E_j$$

(4)

(6)

Deriving the weights using the following formula:

$$W_i=D_i{\sum }_{i=1}^m{D}_i, (1\le \mathsf{i}\le \mathsf{m})$$

(5)

(7)

Compute the overall scores of each model using the following formula:

$$S_i={\sum }_{i=1}^m{W}_j\times P_{ij}, (\mathsf{i}=1, 2, \dots \mathsf{m})$$

(6)

3.1.2. Fuzzy Comprehensive Evaluation Method

The fuzzy comprehensive evaluation method is a decision-making analysis approach based on fuzzy mathematics theory, which is suitable for addressing ambiguity and uncertainty commonly encountered in evaluation processes [27]. In this study, the method is employed to comprehensively assess the four technical models for fertile topsoil restoration. The specific steps are as follows:

(1)

Determining the Evaluation Factor Set

The comprehensive benefit evaluation indicators for fertile topsoil restoration are divided into three categories: economic benefit indicators $X_1$, social benefit indicators $X_2$, and ecological benefit indicators $X_3$. The evaluation factor set is denoted as U = {${\mathrm{X}}_1$, ${\mathrm{X}}_2$, ${\mathrm{X}}_3$}, where ${\mathrm{X}}_1$ = {${\mathrm{X}}_{11}$, ${\mathrm{X}}_{12}$, ${\mathrm{X}}_{13}$, ${\mathrm{X}}_{14}$, ${\mathrm{X}}_{15}$, ${\mathrm{X}}_{16}$}; ${\mathrm{X}}_2$ = {${\mathrm{X}}_{21}$, ${\mathrm{X}}_{22},{\mathrm{X}}_{23}$, ${\mathrm{X}}_{24}$, ${\mathrm{X}}_{25}$, ${\mathrm{X}}_{26}$}; ${\mathrm{X}}_3$ = {${\mathrm{X}}_{31}$, ${\mathrm{X}}_{32}$, ${\mathrm{X}}_{33}$, ${\mathrm{X}}_{34}$, ${\mathrm{X}}_{35}$, ${\mathrm{X}}_{36}$}.

(2)

Constructing the Evaluation Set

Let the evaluation set be V = {Excellent, Good, Fair, Poor, Very Poor}, corresponding to the scores F = {10, 7.5, 5, 2.5, 0}. Following the classification criteria proposed [28], the definitions are as follows: Excellent: Indicator value ≥ top 20th percentile; Good: Between the 40th and 20th percentiles; Fair: Between the 60th and 40th percentiles; Poor: Between the 80th and 60th percentiles; Very Poor: Below the 80th percentile.

(3)

Constructing the Membership Function

Different types of indicators are evaluated using differentiated membership functions: For positive-effect indicators, an ascending semi-trapezoidal membership function is adopted (Equation (7)); for negative-effect indicators, a descending semi-trapezoidal membership function is used (Equation (8)).

$$\mu positive(x)=\left\{\begin{array}{c}1, x\gg b_1 \\ {\displaystyle \frac{x-b_2}{b_1-b_2}}\\ 0, x\ll b_2\end{array}\right., b_2<x<b_1$$

(7)

$$\mu negative(x)=\left\{\begin{array}{c}1,x\ll a_1\\ {\displaystyle \frac{a_2-x}{a_2-a_1}}\\ 0, x\gg a_2\end{array}\right., a_1<x<a_2$$

(8)

where $b_1, { b}_2$ and $a_1,{ a}_2$ are the grading thresholds, which are determined based on the actual data distribution.

(4)

Constructing the Fuzzy Relation Matrix

For each sample value $X_{ij}$, the degree of membership $r_{ij}^k (k=1, 2, \dots , 5)$ to each evaluation grade is calculated to form the fuzzy relation matrix $R$.

(5)

Calculating the Comprehensive Evaluation Value

Combining the entropy-based weights ${\omega }_j$, the comprehensive evaluation value $B_i$ is obtained using a weighted average operator.

$$B_i={\sum }_{j=1}^m{\omega }_j·R_{ij}$$

(9)

The comprehensive benefit level of each technical model was ultimately determined based on the principle of maximum membership.

3.1.3. Methodological Selection Justification

The entropy weight–fuzzy comprehensive evaluation method was selected over alternative multi-criteria approaches due to its unique capacity to address both objective data variability and subjective ambiguity inherent in agricultural technology assessments. Unlike the Analytic Hierarchy Process (AHP), which relies on subjective expert judgments for weighting, the entropy weight method determines indicator weights objectively based on the dispersion of empirical survey data, thereby minimizing arbitrariness. In contrast to Data Envelopment Analysis (DEA), which focuses on efficiency evaluation, our method accommodates multi-dimensional non-efficiency criteria (economic, social, and ecological) essential for a holistic sustainability assessment. Furthermore, the fuzzy comprehensive evaluation component explicitly handles qualitative uncertainties—such as farmers’ perceptions of satisfaction or soil fertility—through membership functions, a feature lacking in traditional multi-criteria analysis. By integrating objective entropy weighting with fuzzy set theory, this method provides a robust framework for evaluating complex technological trade-offs in heterogeneous agricultural systems.

3.1.4. Mitigation Strategies for Social Desirability Bias

Recognizing the potential for social desirability bias in farmer-reported data—where respondents may over-report positive behaviors or under-report negative ones—this study implemented a multi-faceted strategy throughout the research process to enhance data validity. The approach was informed by established methodologies for mitigating respondent bias in survey-based environmental and agricultural studies [29].

(1)

Assurance of Anonymity and Confidentiality. Prior to each interview, respondents were explicitly and repeatedly assured of the complete anonymity and confidentiality of their responses. They were informed that their answers were solely for academic research purposes, would not be shared with local authorities or any third party, and could not be linked to their personal identity. This fundamental measure aimed to reduce respondents’ motivation to provide socially desirable answers.

(2)

Neutral Questionnaire Design. The questionnaire was carefully designed with neutral, non-leading wording to minimize interviewer influence. For instance, rather than asking, “Don’t you think this technology is excellent?” the survey used neutral phrasing such as, “Please rate your overall satisfaction with this technology on a scale of 1 to 5.” This approach allowed farmers to express genuine opinions without perceived pressure to conform to a specific response.

(3)

Data Triangulation via Multi-dimensional Indicators. The evaluation framework incorporated multiple indicator types to cross-validate subjective responses. Specifically, within the social benefit dimension, subjective perceptions (e.g., farmer satisfaction) were combined with more objective behavioral measures (e.g., commodification rate, crop yield per unit area, frequency of technical training). The consistency observed across these different indicator types reinforces the validity of the results and mitigates the risk of perception bias unduly influencing the overall evaluation.

3.2. Data Sources and Sample Description

3.2.1. Data Sources

The data used in this study were obtained from a household questionnaire survey conducted by the research team in November 2024 in the core thin-layer black soil region of the southern Songnen Plain. A multi-stage stratified random sampling method was employed. First, according to the degree of black soil degradation and the structure of agricultural cultivation, four representative counties, including Nong’an County in Changchun City, were selected in the study area. Second, three townships were randomly selected from each county, followed by three villages from each township. Finally, eight households were randomly interviewed in each village, yielding a total of 288 household interviews. After eliminating 25 questionnaires due to missing key responses, 263 valid questionnaires were retained, resulting in a response rate of 91.3%.

The questionnaire consisted of three sections: The first section captured basic household characteristics, including the gender, age, and education level of the household head; the number of agricultural laborers in the household; and the total cultivated land area. The second section focused on the household’s adoption and implementation of fertile topsoil restoration techniques. The third section involved the household’s overall evaluation of the different fertile topsoil restoration models.

3.2.2. Sample Description

The sample characteristics are presented in

Table 1. Descriptive statistics of household heads and family farming characteristics.

Type	Item	Frequency	Percentage
Gender	Male	234	89.0%
	Female	29	11.0%
Age	30 or below	9	3.4%
	31–40	36	13.7%
	41–50	87	33.1%
	51–60	75	28.5%
	Over 60	56	21.3%
Education level	Secondary school or below	141	53.6%
	High school	82	31.2%
	Junior college	31	11.8%
	Undergraduate degree and above	9	3.4%
Household agricultural labor	1 or below	20	7.6%
	2–3	217	82.5%
	4 and above	26	9.9%
Cropland area	Below 1 hm2	46	17.5%
	1–3 hm2	95	36.1%
	3–5 hm2	87	33.1%
	Above 5 hm2	35	13.3%

. In terms of household head gender, males accounted for a significantly higher proportion (89.0%) compared to females (11.0%). Regarding age distribution, household heads were primarily concentrated in the 40–60 age range, representing 61.6% of the total sample, while those over 60 accounted for 21.3% and those under 30 made up only 3.4%. In terms of education level, individuals with a secondary school education or below constituted the largest group (53.6%), followed by high school graduates (31.2%), while those with a junior college degree or above accounted for only 15.2%. As for the household agricultural labor structure, the average number of agricultural laborers per household was 2.4. Households with 2–3 laborers represented the vast majority (82.5%), whereas those with one or fewer accounted for 7.6%, and those with four or more made up 9.9%. The distribution of cropland area showed that medium-scale farmers (1–5 hm²) formed the majority (69.2%), followed by small-scale farmers (less than 1 hm²) at 17.5%, and large-scale farmers (more than 5 hm²) at 13.3%.

Figure 2 illustrates the adoption rates of the four technical modes among farmers: the no-tillage model is the most favored (35.4%), followed by the strip-tillage model (32.3%), the deep-tillage model (20.9%), and the indirect-tillage model (11.4%).

3.3. Construction of the Evaluation Index System

Based on the characteristics of the thin-layer black soil region in the Songnen Plain and the attributes of the farmer survey data, and with reference to previous academic research [30,31,32,33,34,35,36], this study establishes a three-dimensional evaluation index system encompassing economic, social, and ecological dimensions, integrating 18 quantifiable indicators, as shown in

Table 2. A comprehensive evaluation index system for fertile topsoil restoration technology models in the thin-layer black soil region.

Target Level	Criterion Level	Indicator Level	Attribute
Comprehensive Benefit ($\mathrm{X}$)	Economic Benefit (${\mathrm{X}}_1$)	Output value per unit area of crops (${\mathrm{X}}_{11}$)	+
		Profit per unit area of crops (${\mathrm{X}}_{12}$)	+
		Output–input ratio $({\mathrm{X}}_{13}$)	+
		Net output value per unit of labor (${\mathrm{X}}_{14}$)	+
		Net land productivity $({\mathrm{X}}_{15}$)	+
		Unit price of agricultural products (${\mathrm{X}}_{16}$)	+
	Social Benefit $({\mathrm{X}}_2$)	Income increase driven by technology adoption (${\mathrm{X}}_{21}$)	+
		Commodification rate of products $({\mathrm{X}}_{22}$)	+
		Crop yield per unit area (${\mathrm{X}}_{23}$)	+
		Farmer satisfaction $({\mathrm{X}}_{24}$)	+
		Frequency of technical training for farmers (${\mathrm{X}}_{25}$)	+
		$\mathrm{Crop} \mathrm{growth} \mathrm{cycle} ({\mathrm{X}}_{26}$)	−
	Ecological Benefit $({\mathrm{X}}_3$)	Pesticide use per unit area (${\mathrm{X}}_{31}$)	−
		Fertilizer use per unit area (${\mathrm{X}}_{32}$)	−
		Soil fertility $({\mathrm{X}}_{33}$)	+
		Moisture retention effect $({\mathrm{X}}_{34}$)	+
		Seedling emergence rate $({\mathrm{X}}_{35}$)	+
		Effective straw conversion rate $({\mathrm{X}}_{36}$)	+

Note: Among the indicators, crop growth cycle (${\mathrm{X}}_{26}$), pesticide use per unit area (${\mathrm{X}}_{31}$), and fertilizer use per unit area (${\mathrm{X}}_{32}$) are negative indicators, where lower values are preferred; the remaining indicators are positive indicators, where higher values indicate better performance.

. The design of the indicator system features three notable characteristics: First, it adopts a combination of positive and negative indicators (e.g., pesticide usage is treated as a negative indicator) to fully reflect the impacts of each technology. Second, it balances objective quantitative indicators (such as: output value per unit area) with subjective evaluation indicators (such as: satisfaction level). Third, it incorporates context-specific indicators like: effective straw conversion rate, which are designed to capture the core benefits of straw-return technologies in the thin-layer black soil region.

4. Results and Analysis

4.1. Determination of Entropy-Based Weights

Using the entropy weight method, objective weights were assigned to the comprehensive benefit evaluation indicators for fertile topsoil construction technologies in thin-layer chernozem areas. The resulting indicator weights are presented in

Table 3. Entropy-based weighting results.

Target Level	Criterion Level	Weight	Indicator Level	Weight
$\mathrm{X}$	${\mathrm{X}}_1$	0.352	${\mathrm{X}}_{11}$	0.168
			${\mathrm{X}}_{12}$	0.201
			${\mathrm{X}}_{13}$	0.158
			${\mathrm{X}}_{14}$	0.175
			${\mathrm{X}}_{15}$	0.142
			${\mathrm{X}}_{16}$	0.156
	${\mathrm{X}}_2$	0.328	${\mathrm{X}}_{21}$	0.192
			${\mathrm{X}}_{22}$	0.165
			${\mathrm{X}}_{23}$	0.181
			${\mathrm{X}}_{24}$	0.208
			${\mathrm{X}}_{25}$	0.125
			${\mathrm{X}}_{26}$	0.129
	${\mathrm{X}}_3$	0.320	${\mathrm{X}}_{31}$	0.178
			${\mathrm{X}}_{32}$	0.162
			${\mathrm{X}}_{33}$	0.185
			${\mathrm{X}}_{34}$	0.176
			${\mathrm{X}}_{35}$	0.159
			${\mathrm{X}}_{36}$	0.140

. The results show that the economic benefit dimension holds the highest weight (X₁ = 0.352), with indicators such as crop profit per unit area (X₁₂ = 0.201) and net labor productivity (X₁₄ = 0.175) ranking highest—reflecting farmers’ sensitivity to economic outcomes in the Songnen Plain, a major grain-producing region [37]. In the social benefit dimension (X₂ = 0.328), farmer satisfaction (X₂₄ = 0.208) and income-increasing effect (X₂₁ = 0.192) stand out, confirming that agricultural technology promotion must be grounded in enhancing farmers’ sense of gain [35]. For the ecological benefit dimension (X₃ = 0.320), soil fertility (X₃₃ = 0.185) and pesticide use per unit area (X₃₁ = 0.178) carry high weights, aligning with the core goal of “preserving ecology and improving soil fertility” in thin-layer chernozem zones [38]. Notably, although the moisture retention effect (X₃₄ = 0.176) has a slightly lower weight than other ecological indicators, its critical role in enhancing water-use efficiency, improving drought resistance, and stabilizing crop yields should not be overlooked. To test the robustness of the weight results, the study re-calculated the weights by eliminating individual indicators and found no significant changes in the rankings of the technology models (variation < 5%), indicating a stable and reliable evaluation outcome.

4.2. Robustness and Sensitivity Analysis of Weights

To ensure the reliability of the evaluation results, a comprehensive sensitivity analysis was conducted to examine the robustness of the entropy-based weights before proceeding with the final comprehensive evaluation.

4.2.1. Ranking Stability Analysis

A Monte Carlo simulation with 10,000 iterations was performed by randomly perturbing each indicator’s weight within a ±20% range while preserving the unity constraint of the weight vector. The ranking probability distribution (Figure 3) demonstrates high stability across all technology models. The strip-tillage model maintained the top ranking position with a probability exceeding 0.95, confirming that its superior performance is robust against weight uncertainties. The no-tillage, deep-tillage, and indirect-tillage model also showed consistent ranking patterns, each preserving their respective positions in over 90% of simulations. This high degree of ranking stability provides strong evidence for the reliability of the evaluation outcomes.

4.2.2. Global Sensitivity Analysis

A variance-based global sensitivity analysis using the Sobol’ method was conducted to identify the weights that contribute most significantly to output uncertainty. The results (Figure 4) reveal that ranking uncertainty is primarily driven by a small subset of key indicators. The weight for ‘Profit per unit area (X₁₂)’ emerged as the most influential factor, accounting for the largest proportion of output variance. This was followed by ‘Farmer satisfaction (X₂₄)’ and ‘Soil fertility (X₃₃)’, which together explained a substantial portion of the remaining variance. Most other weights showed negligible sensitivity indices, with individual contributions typically below 5% of the total uncertainty.

In summary, these analyses collectively verify that the subsequent evaluation results are not sensitive to minor fluctuations in the majority of weights, thereby underscoring the robustness of the ranking derived from the entropy weight method.

4.3. Fuzzy Comprehensive Evaluation Results Analysis

4.3.1. Comprehensive Evaluation Results

Using the fuzzy comprehensive evaluation method, this study assessed the comprehensive benefits of four fertile topsoil restoration technology models. The results are presented in

Table 4. Evaluation scores of fertile topsoil construction technology models in thin-layer black soil regions.

Evaluation Grade		No Tillage	Strip Tillage	Deep Tillage	Indirect Tillage
Goal Level	$\mathrm{X}$	7.538	8.153	6.892	6.832
Criterion Level	${\mathrm{X}}_1$	7.633	8.259	6.794	6.233
	${\mathrm{X}}_2$	7.406	8.457	6.754	6.483
	${\mathrm{X}}_3$	7.574	7.743	7.129	7.781
Indicator Level	${ \mathrm{X}}_{11}$	8.151	8.767	7.664	6.419
	${ \mathrm{X}}_{12}$	7.421	9.095	6.575	5.447
	${ \mathrm{X}}_{13}$	7.563	8.421	6.114	5.269
	${ \mathrm{X}}_{14}$	8.062	8.114	6.857	6.173
	${ \mathrm{X}}_{15}$	7.346	7.822	6.517	5.306
	${ \mathrm{X}}_{16}$	7.255	7.334	7.039	8.783
	${ \mathrm{X}}_{21}$	8.043	8.955	6.447	5.532
	${ \mathrm{X}}_{22}$	7.569	8.283	7.011	7.546
	${ \mathrm{X}}_{23}$	8.123	8.594	8.244	7.626
	${ \mathrm{X}}_{24}$	7.598	9.105	6.534	5.428
	${ \mathrm{X}}_{25}$	5.451	7.553	6.473	5.525
	${ \mathrm{X}}_{26}$	7.649	8.253	5.816	7.242
	${ \mathrm{X}}_{31}$	8.524	7.556	6.598	6.177
	${ \mathrm{X}}_{32}$	7.519	7.054	6.196	8.902
	${ \mathrm{X}}_{33}$	6.933	7.585	8.628	8.814
	${ \mathrm{X}}_{34}$	8.901	8.226	6.573	7.269
	${ \mathrm{X}}_{35}$	7.534	8.625	6.554	6.573
	${ \mathrm{X}}_{36}$	6.033	7.412	8.227	8.951

Note: The negative indicators have been converted accordingly; thus, higher scores indicate lower actual values and better performance.

, showing that each model demonstrates varying performances across the economic, social, and ecological dimensions. In terms of overall scores, the strip-tillage model ranked highest with a total score of 8.153, outperforming other models in both economic benefits (8.259) and social benefits (8.457), while also maintaining a relatively high level of ecological benefits (7.743). The no-tillage model excelled in the ecological dimension (8.131), particularly in moisture conservation (8.901) and pesticide reduction (8.524). The deep-tillage model showed remarkable performance in soil fertility improvement (8.628), but due to its high operational cost, it scored relatively lower in economic (6.794) and social (6.754) benefits. The indirect model performed well in fertilizer substitution (8.902) and straw conversion rate (8.951), yet it received the lowest overall score due to limitations in technical complexity and farmer adoption. It is worth noting that all models achieved scores above 7 in the ecological dimension, indicating that current mainstream fertile topsoil restoration technologies effectively support black soil conservation. However, significant differences remain in their economic feasibility and social acceptance.

4.3.2. Analysis of Evaluation Results by Category

(1)

Economic Benefit Analysis

As shown in Figure 5, the four technical models exhibit clear gradient differences in economic feasibility within the thin-layer black soil region of the Songnen Plain. The strip-tillage model achieved the highest economic benefit score of 8.259, demonstrating the most favorable overall economic performance. Its advantage is mainly reflected in two core indicators: crop profit per unit area (9.095) and output-input ratio (8.421). These results validate the economic viability of the “localized tillage + straw mulching” technical approach: strip tillage reduces mechanical energy consumption during sowing (15–20% lower than no-tillage), while straw mulching effectively suppresses weed growth and reduces herbicide usage, thereby achieving the dual goal of cost reduction and efficiency improvement [39]. The no-tillage model presents a “low-cost, medium-return” economic pattern with clear cost-control advantages. It performs well in terms of net labor productivity (8.062) and output value per unit area (8.151), largely due to savings in machinery energy related expenses and pesticide costs from the no-tillage operation. However, slow straw decomposition under low-temperature conditions limits crop yield (unit yield score of 8.123, 5.8% lower than that of strip tillage), thus constraining the model’s potential for further economic improvement [40]. The deep-tillage model is constrained by high mechanical operation costs, resulting in a relatively low score for output-input ratio (6.114). However, through deep tillage and deep incorporation, this model significantly enhances soil organic matter content and water-holding capacity [41], leading to a crop yield per unit area score (8.244) comparable to that of the no-tillage model. The indirect model, limited by high technical conversion costs (e.g., composting and transportation) and organic fertilizer application costs, shows weaknesses in output-input ratio (5.269) and profit per unit area (5.447), which are key constraints on its economic performance [42]. Nevertheless, its high agricultural product unit price (8.783) reflects a market premium for organic products. Overall, the four technical models demonstrate a clear economic gradient: strip-tillage straw return ranks highest, followed by no-tillage, with deep-tillage in the middle and indirect at the bottom. Among them, the strip-tillage model stands out as the most economically viable option for the thin-layer black soil region, owing to its remarkable cost-reduction and efficiency-enhancement advantages.

(2)

Analysis of Social Benefit Evaluation

As shown in Figure 6, the four technical models exhibit significant differences in terms of farmer acceptance and potential for technical dissemination. The strip-tillage model achieved the highest social benefit score (8.457), a finding primarily driven by its exceptional farmer satisfaction (9.105). This high satisfaction stems from the model’s ability to meet several critical needs at once—securing a stable harvest, managing economic risk, and maintaining operational practicality. It secures stable and high yields (8.594) while mitigating the agronomic risks and management challenges inherent to no-tillage, such as poor seeding quality and demanding weed control. Furthermore, it remains more economically accessible than the high-cost deep-tillage alternative. By effectively resolving these critical trade-offs, the model fulfills the core farmer demand for “stable yield and increased income” [31], thereby cementing its status as the most socially acceptable option. The no-tillage model performs well in per unit yield (8.123) and income improvement (8.043), but its scores in training frequency (5.451) and farmer satisfaction (7.598) are relatively lower, reflecting initial adaptation challenges. Farmers often face technical barriers in seeding quality control and weed management during early application of the no-tillage technique. The deep-tillage model, while achieving a decent per unit yield (8.244), suffers from low scores in farmer satisfaction (6.534) and income enhancement (6.447), possibly due to the labor-intensive nature of the operation and limited short-term yield improvements. The indirect model is constrained by complex operational procedures and higher technical thresholds, leading to weaker performance across all social benefit indicators—particularly in technical training needs (5.525), which highlights a significant gap between required knowledge and actual farmer mastery. In summary, the social benefit evaluation reveals an “adaptation gap” in technology promotion. Models that combine agronomic adaptability (e.g., strip tillage) with operational simplicity (e.g., no-tillage) are more likely to be favored by farmers.

(3)

Ecological Benefit Evaluation Analysis

As shown in Figure 7, the four technological models demonstrate differentiated characteristics in terms of ecological benefits. Among them, the indirect model achieves the highest overall ecological sustainability score of 7.781. Its core strengths lie in fertilizer substitution (8.902) and straw conversion efficiency (8.951), as the model utilizes biological transformation to efficiently recycle straw resources. This significantly reduces chemical fertilizer input and improves nutrient utilization, aligning with the concept of circular agriculture, and is especially suitable for promotion in integrated farming areas [23]. The no-tillage model excels in moisture retention (8.901) and pesticide reduction (8.524), highlighting the advantages of no-tillage techniques in soil and water conservation and non-point source pollution control. However, it has a relatively low straw conversion rate (6.033), reflecting the technical bottleneck of slow straw decomposition under low-temperature conditions. The strip-tillage model presents a well-balanced ecological performance [43]. Its technical approach—localized tillage combined with full straw coverage—preserves surface coverage benefits while enhancing the soil microenvironment through strip-tillage. This leads to synergistic improvements in moisture retention (8.226), soil fertility (7.585), and pesticide reduction (7.556), making it particularly suitable for hilly areas prone to soil erosion. In contrast, the deep-tillage model can rapidly enhance soil fertility (8.628), but due to intensive disturbance of the soil structure, it performs less well in moisture retention (6.573) and pesticide reduction (6.598) [44]. This model exhibits a trade-off between “rapid fertility enhancement” and “ecological disturbance,” and requires optimization in tillage depth and frequency to mitigate its adverse impacts on soil ecological functions. Overall, each model demonstrates unique ecological features, offering diversified technical solutions tailored to different ecologically vulnerable areas.

4.4. Stratified Analysis by Farm Management Scale

To identify the optimal technology models for different types of farming entities, this study further stratified the sample households into three groups based on management scale: small-scale (<1 hm²), medium-scale (1–5 hm²), and large-scale (>5 hm²). The entropy weight–fuzzy comprehensive evaluation method was then applied separately to each subgroup to determine the optimal technology model for each scale-based cohort. The results of this stratified evaluation are presented in

Table 5. Stratified Analysis of Technology Models by Farm Size.

Farm Size Category	Sample Proportion	Preferred Technology Model	Comprehensive Benefit Score
Small-scale Farmers	17.50%	No-tillage	7.65
Medium-scale Farmers	69.20%	Strip-tillage	8.18
Large-scale Farmers	13.30%	Strip-tillage	8.41

.

Analysis of the results indicates significant differences in optimal technology choices among farmers of different management scales. For small-scale farmers, the No-tillage model demonstrated the most favorable comprehensive benefits. This is primarily attributable to the model’s minimal machinery investment costs and lower labor requirements, allowing it to exhibit optimal adaptability within the smallholder economy characterized by dual constraints of capital and labor.

For both medium-scale and large-scale farmers, the strip-tillage model proved to be the optimal choice. As the dominant group in this study (comprising 69.2% of the sample), the performance of the Strip-tillage model among medium-scale farmers stems from its superior output-input ratio and profit per unit area. It successfully integrates the ecological advantages of no-tillage with the agronomic convenience of localized tillage, perfectly aligning with this cohort’s core demand for “stable yield and increased income”. For large-scale farmers, the advantages of the Strip-tillage model are further amplified by economies of scale, resulting in the highest comprehensive benefit score (8.41) across all groups. This is particularly evident in the economic benefit and farmer satisfaction indicators. This finding confirms that the Strip-tillage model is a key technology for achieving the goal of “reducing costs and increasing efficiency” in large-scale operations.

5. Discussion

5.1. Discussion of Research Findings

This study employed the entropy-weighted fuzzy comprehensive evaluation method to systematically assess the comprehensive benefits of four typical fertile topsoil construction models in the thin black soil region from the perspectives of economic, social, and ecological dimensions. The results showed that the strip-tillage model achieved the highest overall benefit, particularly excelling in the economic and social dimensions. This finding can be interpreted from the perspective of the compatibility between technical characteristics and regional resource endowments. The strip tillage model, through the combination of “localized tillage + full straw coverage,” integrates the advantages of both no-tillage and deep-tillage. On the one hand, strip tillage improves seedbed compaction and soil temperature, effectively addressing common problems in no-tillage practices such as poor sowing quality and slow straw decomposition in cold conditions. On the other hand, full straw coverage helps retain soil moisture and prevents erosion. This conclusion is consistent with the findings of Xu et al. [16] on maize cultivation models in the Songnen Plain, which also pointed out the significant advantages of strip tillage in reducing costs and improving efficiency. However, unlike Ao et al. [3], who emphasized the absolute advantage of no-tillage in controlling wind erosion, this study finds that in the thin black soil region, strip tillage is more acceptable to farmers due to its better agronomic adaptability and economic feasibility. This reflects the need for technology promotion to balance ecological benefits with socio-economic realities.

The no-tillage model demonstrates notable ecological benefits (8.131 points), particularly in terms of soil moisture conservation (8.901 points) and pesticide reduction (8.524 points). However, its relatively lower economic benefit score (7.633 points) can be attributed to two main factors. First, with the reduction in chemical pesticide use, the management of pests, diseases, and weeds increasingly depends on mechanical or manual methods, resulting in higher non-chemical control costs. Second, the slow decomposition of straw under low-temperature conditions may negatively impact sowing quality and increase field management expenses. These findings align with those of Zhang et al. [45], who observed that although pesticide reduction under no-tillage mulching systems in Northeast China supports ecological protection, it significantly increases mechanical and manual weed control costs. Meanwhile, Hang et al. [46] also found that the low spring temperatures in the Songnen Plain significantly delay the decomposition rate of surface straw, thereby affecting seed germination and increasing subsequent tillage or replanting costs, which corroborates the findings of this study. The deep-tillage model demonstrates significant effectiveness in rapidly enhancing soil fertility (8.628 points), yet its economic feasibility is relatively low (6.794 points) due to high mechanical operation costs. This result is consistent with the findings of Schneider et al. [47] who analyzed the cost–benefit of deep-tillage in the Northeast black soil region and pointed out that although this technique improves soil quality and fertility, mechanical input costs can account for over 35% of direct production costs, significantly reducing net income for farmers. The indirect model exhibits the best ecological sustainability (7.781 points), but its widespread adoption is hindered not only by high technical complexity and cost but also by a lack of accessible institutional frameworks and supporting infrastructure, such as centralized composting facilities, biogas plants, and efficient straw collection and redistribution networks. This aligns with the “high willingness–low adoption” phenomenon reported by Farahbakhsh et al. [48], reflecting the compound challenges of technical, capital, and institutional constraints faced by smallholder farmers.

Our evaluation results reveal a complex web of trade-offs and synergies among the economic, social, and ecological dimensions. A prominent trade-off was observed in the no-tillage model, where high scores in ecological indicators were offset by lower economic performance due to increased management costs and relatively lower farmer satisfaction, creating a tension between environmental goals and livelihood realities. Conversely, the strip-tillage model demonstrated a notable synergy. Its design successfully bridged the gap between the ecological advantages of no-tillage and the agronomic/social needs for a favorable seedbed and higher yields, thereby achieving high scores across all three dimensions and validating its top-ranked comprehensive benefit. These findings underscore that the ‘comprehensive benefit’ is not a linear sum but a context-dependent equilibrium, where the optimal technology depends on which set of synergies are most valued and which trade-offs are most acceptable in a specific socio-ecological context, as evidenced by our farm-scale stratification analysis (Table 5).

Furthermore, situating our findings within a global context enriches the discussion and underscores the importance of region-specific technology adaptation. The challenges of black soil degradation are not unique to the Songnen Plain but are a global concern across the world’s major chernozem belts. However, the optimal solution has significant regional differences. For instance, in the warmer and more extensive steppes of Ukraine, the classic no-tillage model is often more widely applicable and effective [49]. In the North American prairies, while conservation tillage systems including no-till and strip-till are also widely implemented, their adoption patterns and implementation scales are significantly influenced by different farm structures, market mechanisms, and policy frameworks [50]. These international comparisons underscore a fundamental principle: while the core conservation objective of minimizing soil disturbance remains universal across chernozem regions, its practical implementation must be carefully calibrated to local conditions. The specific agro-climatic constraints of the Songnen Plain—characterized by its thin soil layers and cold springs—necessitate technological adaptations that may differ from solutions optimal in other global chernozem regions.

5.2. Theoretical and Practical Contributions

In terms of theoretical contribution, this study establishes a comprehensive evaluation index system encompassing economic, social, and ecological dimensions, thereby addressing the shortcomings of traditional agricultural technology assessments that often overlook social acceptance and ecological sustainability. By introducing the entropy weight method to determine index weights, the study avoids subjective bias and objectively reflects the high sensitivity of black soil region farmers to economic returns (economic dimension weight: 0.352), as well as their core demand for “improving soil fertility and preserving ecology” (ecological dimension weight: 0.320). Furthermore, by integrating fuzzy mathematics theory, the study effectively addresses the ambiguity and uncertainty inherent in expert evaluations and farmers’ perceptions within agricultural systems, offering a methodological reference for the multi-dimensional evaluation of green agricultural technologies. This research validates the applicability of the entropy-weighted fuzzy comprehensive evaluation method in the field of agricultural technology assessment, providing a novel methodological tool for future related studies.

On the practical level, this study provides empirical evidence and decision-making references for the application and promotion of black soil conservation technologies. Through the quantitative evaluation of four technical models, it clearly identifies the performance and limiting factors of each model under real production conditions, offering reliable comparative bases for agricultural extension agencies, emerging agricultural business entities, and smallholder farmers. Additionally, the research findings can serve as a scientific basis for governmental investment decisions, thereby enhancing the precision and implementation efficiency of agricultural subsidy policies and promoting a shift in black soil conservation efforts from “emphasis on implementation” to “emphasis on effectiveness.”

5.3. Research Limitations and Future Directions

This study is subject to several limitations that point to valuable avenues for future research. First, while the findings are grounded in data from typical counties of the southern Songnen Plain, extending this research to other black soil subregions with diverse soils, climates, and farm management structures would enable insightful cross-regional comparisons. Second, although our evaluation incorporated multi-dimensional indicators, the assessment of ecological benefits relied on proxies rather than direct measurements of core metrics like long-term soil carbon sequestration, biodiversity, and biological soil quality. Future work should integrate specialized soil monitoring or life cycle assessment (LCA) to close this gap. Third, the reliance on cross-sectional data restricts our understanding of the dynamic impacts of these technologies. Long-term positioning experiments and panel studies are essential to verify sustainability and establish causal pathways. Methodologically, despite robustness checks, the absence of formal statistical significance testing for score differences and the lack of control for confounding farmer characteristics present opportunities for refinement in subsequent analyses. Finally, this study did not explore the synergistic potential of combining different technological practices to overcome individual model limitations, nor did it fully account for external factors such as detailed policy costs and institutional barriers.

To address these limitations, a multi-faceted research agenda is proposed. Future efforts should not only broaden the geographical and ecological scope of sampling but also leverage remote sensing and field monitoring to build dynamic databases for real-time evaluation. Adopting participatory approaches can embed farmers’ local knowledge into the technology adaptation process. Furthermore, researching integrated technological synergies and simulating quantified policy scenarios—factoring in subsidies, infrastructure, and training—are critical steps to verify economic feasibility and optimize large-scale promotion strategies. Through these methodological and contextual refinements, the theoretical depth and practical utility of comprehensive agricultural technology assessments can be significantly advanced.

6. Conclusions

Based on 263 household survey samples collected from the thin black soil region in the southern Songnen Plain, this study applied the entropy weight–fuzzy comprehensive evaluation method to systematically assess the comprehensive performances of four tillage layer construction models from economic, social, and ecological dimensions. The results reveal significant differences among the models in terms of their overall and dimensional effectiveness. The strip-tillage model demonstrates the best overall performance, with outstanding economic and social benefits and relatively high ecological performance, indicating strong adaptability and high promotion potential. The no-tillage model shows remarkable ecological advantages—particularly in moisture conservation and pesticide reduction—but its economic performance is relatively constrained due to the higher costs of pest, disease, and weed management. The deep-tillage model significantly enhances soil fertility in a short period but suffers from limited economic feasibility due to high mechanical operation costs. The indirect model excels in nutrient recycling and ecological sustainability but faces major adoption challenges due to its technical complexity and high implementation costs.

Based on the systematic evaluation of the comprehensive benefits of different technology models, this study proposes the establishment of a “categorized guidance and precision adaptation” technology extension system to synergistically advance black soil conservation and agricultural sustainable development. Specifically, the strip-tillage with straw mulching model, which demonstrated the optimal comprehensive benefit, should be established as the dominant regional model and prioritized for promotion among medium- and large-scale farmers. The no-tillage with straw mulching model, notable for its outstanding ecological benefits, should be positioned as a targeted model for promotion in areas such as water erosion-prone zones or among small-scale farmers, supported by ecological compensation measures. A cautious promotion strategy is recommended for the deep-tillage model, which rapidly enhances soil fertility; its application should be focused on areas with shallow topsoil and be coupled with subsidies for agricultural operations. Meanwhile, the indirect-tillage model should be systematically advanced in integrated crop-livestock regions through the development of a comprehensive “collection–transformation–application” socialized service system. This differentiated and systematic extension network aims to achieve the most efficient match between technological advantages and regional resource endowments.