Influence of Digital Literacy on Farmers’ Adoption Behavior of Low-Carbon Agricultural Technology: Chain Intermediary Role Based on Capital Endowment and Adoption Willingness

Abstract

Farmers are the key adopters of low-carbon agricultural technologies, and their adoption behavior is crucial for achieving the “dual carbon” goals. However, how digital literacy influences farmers’ technology adoption remains underexplored. Based on survey data from 742 farmers in Shandong Province, this study employs an ordered Logit model to examine the impact of digital literacy on the adoption of low-carbon agricultural technologies, as well as the mediating effects of capital endowment and adoption willingness, along with their heterogeneity. The results indicate that digital literacy significantly promotes farmers’ adoption of low-carbon agricultural technologies, but its effects vary across different technology types. Information acquisition literacy and security literacy have a greater impact on data-driven technologies (water-saving irrigation and soil testing-based fertilization), while content creation literacy and problem-solving literacy play a more significant role in knowledge-based technologies (integrated pest management). Mechanism analysis reveals that capital endowment and adoption willingness function as independent mediators, with a significant chain mediation effect between them. Furthermore, different dimensions of capital endowment exert heterogeneous influences on technology adoption: human and material capital primarily influence conservation tillage and water-saving irrigation, social capital facilitates integrated pest management, and economic capital plays a crucial role in water-saving irrigation adoption. Based on these findings, this study recommends enhancing farmers’ digital literacy, optimizing capital endowment structures, strengthening market mechanisms, and establishing demonstration bases to accelerate the widespread adoption of low-carbon agricultural technologies and contribute to the realization of the “dual carbon” goals.

1. Introduction and Literature Review

Global climate change has become a major challenge facing human society [1], and the increase in carbon emissions caused by human activities is an important cause of global warming [2]. Agricultural carbon emissions are an important part of global carbon emissions. The results of the fourth IPCC assessment show that 13.5% of global carbon emissions come from modern agricultural production activities [3], while scientific and reasonable low-carbon technologies can eliminate 80% of agricultural greenhouse gas emissions [4]. In 2024, The Third Plenary Session of the 20th Central Committee of the Communist Party of China proposed to improve the ecological civilization system, jointly promote carbon and pollution reduction and green expansion and growth, actively respond to climate change, and improve the green and low-carbon development mechanism. Therefore, adopting agricultural technologies and methods of low energy consumption, low emission, and low pollution in the process of agricultural production and management to develop green, low-carbon, and eco-friendly agriculture, are not only the “standard equipment” of an agricultural power but also an important way to help achieve the national “carbon peaking and carbon neutrality” goals from the perspective of agricultural carbon reduction [5].

Farmers are the main actors for adopting agricultural technology, and the effect of carbon sequestration and emission reduction caused by the application of existing agricultural technology is also the embodiment of the concentrated carbon effect of hundreds of millions of scattered farmers in the process of agricultural production [6]. Therefore, further strengthening and guiding farmers to adopt low-carbon agricultural technology is the key to realizing low-carbon agricultural production [7]. However, in the current situation, farmers’ awareness of low-carbon agriculture is not high, and their subjectivity is not in place. Traditional agricultural technology extension also has problems such as a single communication form, poor timeliness, and so on [8,9]. All these phenomena have reduced the efficiency of promoting low-carbon agricultural technology. With the accelerated penetration of new information technology into rural areas, the digital gap between urban and rural areas is narrowing, and the digital agricultural technology extension model has become an important engine to promote agricultural and rural modernization, which is helpful to encourage farmers to deepen their cognition and application of low-carbon agricultural production technology. However, whether the digital literacy of farmers in China is enough to accept the new agricultural technology extension model and what impacts digital literacy will have on farmers’ low-carbon agricultural technology adoption behavior are still topics that we need to discuss. Currently, academic circles have conducted in-depth discussions on farmers’ adoption of low-carbon agricultural technology and their influencing factors, focusing on farmers’ characteristics [2], capital endowment [8,10], cognitive traits [4,6], and institutional norms [11,12]. As digital rural development progresses, the information dividends of digital technologies in agriculture are gradually becoming evident. Consequently, some scholars have begun exploring the relationship between “digitalization” and farmers’ adoption of low-carbon production technologies. The results show that Internet use [13,14,15,16,17], digital technology use and digital literacy [18,19,20], digital agricultural technology extension [21,22], e-commerce participation [23,24], digital finance [25,26], etc. all promote the adoption of low-carbon production technology by farmers, which indicates that the application of digital technology can effectively promote the application of agricultural green and low-carbon production technology.

The above research examines the factors affecting farmers’ low-carbon production behavior from different angles, which provides a good theoretical basis for this paper, but there is still room for further refinement: in terms of the research field, most of the existing studies focus on carbon emission reduction in the industrial field, and rarely emphasize the important role of carbon neutralization in agriculture and rural areas or explore the new low-carbon development path of rural revitalization. From the research perspective, although the existing literature pays attention to the impact of digital literacy, capital endowment, and adoption willingness on farmers’ low-carbon production behavior, there is no precedent for bringing the three into the analysis framework at the same time. In terms of the research content, the existing research mainly analyzes the impact of the comprehensive level of digital literacy on farmers’ low-carbon production behavior, but lacks systematic research on the impact of multi-dimensional digital literacy on farmers’ low-carbon production behavior.

Based on this, this study attempts to explore the impact and mechanism of digital literacy on the adoption of low-carbon production technology by farmers and incorporates digital literacy, capital endowment, and adoption intention into the analysis framework. On this basis, survey data, the Logit model, and the chain intermediary effect model are employed for empirical testing, which can provide new understandings and pathways for how to further promote low-carbon green production technology in the digital landscape and build a long-term adoption mechanism of low-carbon agricultural production technology.

2. Theoretical Analysis and Research Hypothesis

2.1. The Influence of Digital Literacy on Farmers’ Adoption of Low-Carbon Agricultural Technologies

Previous studies have suggested that improving farmers’ awareness of climate change and increasing agricultural technologies extension services can significantly encourage farmers to adopt low-carbon emission reduction technologies [27]. However, farmers often face information barriers, technical knowledge gaps, and a lack of external support, making it difficult for them to access, understand, and adopt low-carbon agricultural technologies. Digital literacy can bridge these gaps by improving farmers’ ability to obtain information, receive technical training, and apply modern digital tools to optimize agricultural production.

Digital literacy plays a crucial role in overcoming these barriers through the following mechanisms:

(1) Enhancing information acquisition and data analysis: farmers with strong digital literacy can efficiently search, filter, and interpret agricultural data through digital platforms, allowing them to access reliable low-carbon farming knowledge and real-time market trends [28].

(2) Strengthening technical training and expert consultation: digital literacy expands farmers’ access to online training, remote guidance, and interactive knowledge-sharing platforms, enabling them to systematically master the operation of low-carbon technologies [18].

(3) Improving decision-making and risk management: By utilizing AI-driven agricultural decision-making systems, smart agricultural machinery, and big data analytics, digitally literate farmers can optimize resource allocation, improve production efficiency, and reduce carbon emissions [29,30]. Farmers with higher digital literacy can utilize digital agricultural technologies, such as precision agriculture [31,32], agricultural IoT [33], AI-driven agricultural decision-making systems [34], smart agricultural machinery [35], and big data-based agricultural management, to set parameters that guide agricultural machinery for efficient operations and to strengthen management and resource allocation in the production process through digital monitoring technologies. As a supporting tool for low-carbon agricultural technologies, digital agricultural technologies enhance the efficiency of land, fertilizer, pesticides, and fuel utilization while reducing carbon emissions and environmental pollution, enabling low-carbon production without compromising soil integrity and the agricultural production environment [36].

Based on this, this paper puts forward the following hypothesis:

H1: 

Digital literacy promotes farmers’ adoption of low-carbon agricultural technology, and different dimensions of digital literacy have heterogeneous effects on this adoption.

Based on the above analysis, Figure 1 illustrates the conceptual framework of this study, depicting the pathways through which digital literacy influences farmers’ adoption of low-carbon agricultural technologies. Specifically, digital literacy can directly promote technology adoption (H1). Additionally, it enhances capital endowment (H2) and strengthens adoption willingness (H3), both of which serve as mediating factors. Furthermore, capital endowment influences adoption willingness, forming a chain mediating effect (H4) that ultimately promotes the adoption of low-carbon agricultural technologies. The following sections further elaborate on these mechanisms.

2.2. The Mediating Role of Capital Endowment

Capital endowment refers to all natural or acquired resources and capabilities owned by farmers. According to the sustainable livelihood capital theory, capital endowment is an important factor affecting farmers’ behavior and decision-making, and most farmers’ decisions are rational choices made after weighing the constraints of capital endowment. The improvement of farmers’ digital literacy will help alleviate the shortage of farmers’ capital endowment and the constraint of information occlusion, and inject a new impetus into farmers’ decision-making and low-carbon agricultural technology adoption behavior.

First, digital literacy has improved farmers’ knowledge reserve and understanding ability by enhancing the function of communication and education [37,38,39] to increase farmers’ human capital. This will help farmers break the traditional agricultural knowledge system and make them understand the scientific connotation of low-carbon agricultural technology more comprehensively, thus laying a theoretical foundation for the implementation of low-carbon agricultural technology [40].

Second, digital literacy improves the frequency of farmers’ instant messaging tools which makes it more convenient for farmers to maintain and expand their network and strengthens their interactions with peers, experts, cooperatives, and communities [41], and enhances farmers’ social capital.

Third, digital literacy can help farmers obtain financial services and market information more conveniently, enable them to better participate in land transfer [42,43], broaden financing channels [44], and enhance farmers’ natural capital and economic capital [45,46].

Generally speaking, the improvement of farmers’ digital literacy enhances farmers’ integration of their capital endowments and their ability to absorb external resources, and provides a scale effect and an agglomeration innovation effect for the adoption of low-carbon agricultural technology. Based on this, this paper puts forward the following assumption:

H2: 

Digital literacy promotes farmers’ adoption of low-carbon agricultural technologies by enhancing capital endowment, with different dimensions of capital endowment exhibiting heterogeneous mediating effects in this process.

2.3. The Mediating Role of Adoption Willingness

According to the Theory of Planned Behavior (TPB), behavior intention is manifested in perceived usefulness and perceived ease of use, which is the motivation and determination of individuals to implement a specific behavior based on their positive attitude, perceived social pressure, and self-efficacy in specific situations. Previous studies have repeatedly verified that farmers’ adoption willingness has a positive impact on adoption behavior [47,48,49]. On the one hand, the improvement of farmers’ digital literacy makes it easier for farmers to understand the problems, such as increased greenhouse gas emissions, waste of resources, and environmental pollution, caused by high-carbon agricultural production from the Internet [50], and realize the economic, ecological, and social benefits brought by low-carbon agricultural technology, thus shaping a positive low-carbon attitude, sense of responsibility, and environmental values. On the other hand, digital literacy helps farmers to participate in online digital training, receive online guidance and real-time feedback, and systematically master the operational specifications of low-carbon agricultural technology, which not only reduces the operational risks and psychological burdens of applying low-carbon agricultural technology [18], but also enhances farmers’ easy-to-use awareness of adopting low-carbon agricultural technology [51], and enhances their confidence in adopting low-carbon agricultural technology. Based on this, this paper puts forward the following assumption:

H3: 

Digital literacy promotes farmers’ adoption of low-carbon agricultural technology by enhancing their adoption willingness.

2.4. Chain Mediation Between Capital Endowment and Adoption Willingness

According to Bourdieu’s social practice theory, differences in social position lead actors to have varying perspectives on the same event, with their social position primarily determined by their capital endowment. From this perspective, differences in capital endowment can also influence farmers’ perceptions of adopting low-carbon agricultural technology. Additionally, resource endowment enhances farmers’ perceived ease of use by improving their cognitive abilities, agricultural production capacity, information acquisition skills, and risk tolerance, thereby strengthening their adoption willingness for new technologies [52]. Building on the previous analysis, digital literacy can influence farmers’ decision-making behavior through capability enhancement, capital accumulation, and increased willingness. Accordingly, this paper proposes the following hypothesis:

H4: 

Capital endowment and adoption willingness play a chain intermediary role between digital literacy and the adoption of low-carbon agricultural technology.

3. Research Design

3.1. Sample Selection and Data Sources

The data of this paper come from the survey of farmers’ adoption behavior of low-carbon production technology conducted by the research group in Shandong Province from June 2023 to January 2024. In this survey, 800 questionnaires were distributed to farmers. According to the research needs, 742 valid samples were finally obtained after eliminating some questionnaires with missing data and inconsistent information, and the effective rate of the questionnaires was 92.75%. The farmers questionnaires mainly use “one-on-one” interviews to investigate the heads of households or major family members involved in production decision-making, including basic information about farming individuals and families, digital literacy, capital endowment, adoption willingness for low-carbon agricultural technology, adoption behavior, and so on. To ensure the comprehensiveness and reliability of the measurement, the questionnaire was designed to systematically capture the core variables and control factors of this study. The detailed question settings and measurement scales for each variable are presented in

Table 1. Variable assignment and descriptive statistics.

Category	Variable Name	Variable Definition	Mean Value	Standard Deviation
Explained Variable: Adoption of Low-carbon Agricultural Technology	Conservation Tillage Technology (Y1)	Was it adopted before? Never adopted = 1, Occasionally adopted = 2, Partially adopted = 3, Mostly adopted = 4, Always adopted = 5	2.872	1.389
	Integrated pest management (IPM) technology(Y2)		3.434	1.276
	Water-Saving Irrigation Technology (Y3)		2.900	1.477
	Soil Testing and Formulated Fertilization Technology (Y4)		2.718	1.312
Core explanatory variable	Digital Literacy (DL)	Digital Information Acquisition Literacy (DL1): Can you use network tools to obtain the information you need?	2.629	1.27
		Digital Content Creation Literacy (DL2): Can you use web tools to share your agricultural experience with others?	3.135	1.353
		Digital Security Literacy (DL3): Can you judge whether the information obtained through the network is true or false?	2.916	1.186
		Digital Problem-Solving Literacy (DL4): Do you think the information obtained through the Internet is helpful to your work and life?	3.495	0.995
Intermediary Variable	Capital Endowment (CE)	Natural Capital: Actual cultivated area (mu)	7.471	8.060
		Human Capital: Have you participated in agricultural technical training? 1 = Yes; 0 = No	0.222	0.416
		Material Capital: What is the situation of agricultural machinery and facilities owned by your family? 1 = Almost none; 2 = less; 3 = General; 4 = More; 5 = a lot	3.539	0.906
		Social Capital: How often do you contact relatives, friends and acquaintances? 1 = No connection; 2 = a little connection; 3 = General; 4 = More; 5 = a lot	4.236	0.779
		Economic Capital: What is the average annual income of families in recent 3 years? 1 = less than 25,000; 2 = 25,001~50,000; 3 = 50,001~100,000; 4 = 100,001~200,000; 5 = 200,001 and above	2.214	0.995
	Adoption Willingness (WILL)	Are you willing to adopt low-carbon agricultural technology? 1 = Very reluctant; 2 = Less willing; 3 = General; 4 = More willing; 5 = Very willing	3.969	0.855
Control Variable	Sex (SEX)	Male = 1, female = 0	0.606	0.489
	Age (AGE)	Actual value of survey (years)	54.972	12.342
	Education (EDU)	1 = Never attended school; 2 = Primary school; 3 = Junior high school; 4 = High school or technical secondary school; 5 = College or higher vocational education; 6 = Undergraduate; 7 = Graduate and above	2.702	1.131
	Political Identity (VIL)	Is it a village cadre? Yes = 1, No = 0,	0.098	0.298
	Health Status (HEALTH)	1 = Very poor; 2 = Comparatively poor; 3 = General; 4 = Better; 5 = Very good	3.799	0.962
	Landform (LAND)	Most of the terrain of the land you operate is: 1 = Mountains; 2 = Hills; 3 = Plains; 4 = Other	2.787	0.56
	Distance from Village to Town (DISTANCE)	How many miles is your home from the nearest town? 1 = within 5 miles; 2 = 6~10 miles; 3 = 11~20 miles; 4 = 21~30 miles; 5 = 31 miles and above	1.575	0.761
	Number of Labor Force (LABOR)	Number of family members engaged in agricultural labor all the year round?	2.336	0.936
	Years of Farming (EXP)	Actual time engaged in agricultural labor (years)?	17.327	5.617
	Low Carbon Cognitive (LCA)	Have you ever heard of low-carbon farming practices? 1 = Never heard of it; 2 = Hear a little; 3 = General; 4 = Frequently heard; 5 = Always hear	2.996	1.204
	Cooperative Organization (COO)	Do you join a cooperative organization? Yes = 1, No = 0	0.109	0.312

.

3.2. Measurement of Variables

3.2.1. Dependent Variable: Low-Carbon Agricultural Technology Adoption Behavior (Y)

Based on existing studies [2,7,11,23,53], this paper identifies low-carbon agricultural technologies (In this study, low-carbon agricultural technologies are measured as technology packages, meaning that each technology category (Y1–Y3) consists of several specific techniques. A farmer is considered to have adopted a particular low-carbon agricultural technology if they have implemented at least one of the included techniques. This classification ensures that adoption measurement aligns with real-world farming practices and allows for a more accurate assessment of technology uptake) with significant carbon sequestration and emission reduction effects, including conservation tillage technology (Y1), integrated pest management (IPM) technology (Y2), water-saving irrigation technology (Y3), and soil testing and formulated fertilization technology (Y4). Each technology is defined as a five-category variable (never adopted = 1, occasionally adopted = 2, partially adopted = 3, mostly adopted = 4, always adopted = 5). As a comprehensive technology package [27], low-carbon agricultural technologies encompass a series of agricultural production practices aimed at reducing carbon emissions and enhancing carbon sequestration capacity. The application of these practices varies across different production processes and crop types [12,54].

Conservation tillage technology (Y1) mainly includes no-till, minimum tillage, and strip tillage:

(1)

No-till technology is primarily used for wheat and maize, as it reduces soil disturbance, increases soil organic matter content, and enhances soil carbon sequestration capacity, thereby reducing carbon emissions [55].

(2)

Strip tillage technology is more commonly applied in maize production, as it helps retain soil moisture and reduces both wind and water erosion [56].

(3)

Minimum tillage technology is suitable for various crops, improving land-use efficiency while minimizing carbon emissions caused by mechanical plowing.

Integrated pest management technology (Y2) is a comprehensive pest management strategy that, based on the actual adoption situation in the surveyed area, integrates multiple approaches, including agricultural, biological, physical, and chemical control measures. These measures aim to reduce the use of chemical pesticides, enhance crop resistance to pests and diseases, minimize pest-related losses, and promote green and low-carbon agricultural development. The key measures include the following:

(1)

Agricultural control: disease-resistant variety selection and seed treatment for pest prevention.

(2)

Physical control: insect-proof net technology, pheromone traps, and insecticidal lamps.

(3)

Biological control: conservation and utilization of natural enemies, biological pesticide technology.

(4)

Chemical control: deep plowing and water irrigation for pest suppression.

In addition, water-saving irrigation technology (Y3) is widely used in grain production in arid and semi-arid regions, particularly sprinkler irrigation, drip irrigation, and subsurface drip irrigation technologies, which effectively reduce carbon emissions from agricultural irrigation [57,58].

Furthermore, soil testing and formulated fertilization technology (Y4) optimize nutrient application and reduce excessive fertilizer use, promoting low-carbon agricultural practices. For instance, in Henan Province, the promotion of this technology has successfully reduced fertilizer consumption while increasing crop yields.

It can be seen from Table 1 that farmers have different acceptance and application degrees of low-carbon agricultural technologies in different links of production. The average adoption degree of integrated pest control technology is the highest, the average adoption degree of soil testing and formula fertilization technology is the lowest, and the average adoption degree of conservation tillage technology and water-saving irrigation technology is close, which has not yet reached the level of half.

3.2.2. Core Explanatory Variable: Digital Literacy (DL)

Gilster (1997) first introduced the term “digital literacy” and defined it as the ability to acquire, memorize, understand, solve problems, and critically synthesize information [59]. In 2013, the European Union released the first version of the Digital Competence Framework (DigComp 1.0), which identified five domains of digital literacy: “information”, “communication”, “content creation”, “safety”, and “problem-solving”. Subsequent updates were introduced in 2016 (DigComp 2.0), 2017 (DigComp 2.1), and 2022 (DigComp 2.2), continuously refining and expanding the scope of these five domains [60]. The United Nations Educational, Scientific and Cultural Organization (UNESCO) defines digital literacy as the ability to safely and appropriately access, manage, understand, integrate, communicate, evaluate, and create information [61]. Currently, there is no universally accepted standard for measuring digital literacy. For instance, Du et al. (2024) developed an evaluation system for farmers’ digital literacy based on digital information literacy, digital social literacy, and digital financial literacy, aiming to assess the overall level of farmers’ digital literacy [19]. Similarly, Yang Yuzhen and Zhang Xueke (2024) constructed an index system for smallholder farmers’ digital literacy by considering general digital literacy, social literacy, creative literacy, safety literacy, and professional literacy, exploring the theoretical framework of how digital literacy influences smallholder farmers’ integration into modern agriculture [41]. Furthermore, Yang Jiali and Wu Congliang (2023) assessed farmers’ digital literacy levels by focusing on digital information communication literacy, digital content creation literacy, digital security literacy, and digital problem-solving literacy [62].

Drawing on relevant literature and considering the objectives of this study, we define farmers’ digital literacy as a comprehensive capability to acquire, understand, create, evaluate, and disseminate information through the effective use of digital tools and technologies in daily life and work. In addition, we assess farmers’ digital literacy across four key dimensions relevant to low-carbon agricultural technology adoption.

(1) Digital Information Acquisition Literacy—a prerequisite for using digital tools and engaging in digital life, enabling farmers to access more low-carbon agricultural production information.

(2) Digital Content Creation Literacy—enhances technical exchanges among farmers and strengthens public opinion supervision.

(3) Digital Security Literacy—improves farmers’ ability to manage risks associated with technology adoption.

(4) Digital Problem-Solving Literacy—helps farmers optimize production decisions, enhance the efficiency of low-carbon technology applications, and improve their adaptability to the digital agricultural environment.

These dimensions were chosen as they correspond to the specific needs of farmers in adopting low-carbon agricultural technologies, thus facilitating a comprehensive evaluation of digital literacy in this context.

Each of these dimensions is measured using a five-point Likert scale (1 = very difficult, 2 = difficult, 3 = neutral, 4 = basically able, 5 = fully capable). The entropy method is employed to standardize these four dimensions and calculate their respective weights, thereby objectively determining the levels of each dimension of digital literacy. The overall digital literacy score is then derived accordingly. The specific questionnaire design and descriptive statistical results are presented in Table 1.

3.2.3. Mediation Variable

Combined with existing studies [10,37,52,63], capital endowment is mainly measured across five dimensions: natural capital, human capital, material capital, economic capital, and social capital. The scale of farmland represents the amount of land resources available to farmers, which is an important component of natural resources. Agricultural technical training reflects the accumulation and improvement of farmers’ agricultural knowledge and skills. The level of agricultural mechanization reflects farmers’ investment in machinery, production tools, production services, and other material capital. Income level directly reflects farmers’ economic strength and their ability to support agricultural production. Social networks directly indicate farmers’ ability to receive support and mutual assistance in social relationships, resource sharing, and information acquisition. The entropy method is used to standardize the above indicators, calculate their weights, and then derive the comprehensive score of capital endowment (CE). As can be seen from Table 1, adoption willingness is a variable that is difficult to observe directly. Drawing lessons from existing studies [8,48,64], we ask farmers whether they are willing to adopt low-carbon agricultural technologies to measure farmers’ adoption willingness low-carbon technologies (WILL).

3.2.4. Control Variable

Combined with the related studies of farmers’ low-carbon production behavior [2,7,8,11,65,66], this paper selects the factors that affect farmers’ low-carbon production technology adoption behavior, such as gender, age, education, political identity, health status, landform, distance from village to town, number of labor force, years of farming, low-carbon cognition, cooperative organization, etc., as control variables.

3.3. Modeling

(1) In order to explore the influence of digital literacy (DL) on farmers’ low-carbon agricultural technology adoption behavior (Y1, Y2, Y3, Y4), this paper first constructs a multivariate Logit benchmark model to verify the research hypothesis H1, as shown in Equation (1).

$$\mathrm{l}\mathrm{o}\mathrm{g}\mathrm{i}\mathrm{t}({\mathrm{Y}}_{\mathrm{i}\mathrm{j}})={\mathsf{\alpha }}_{1\mathrm{j}}{\mathrm{D}\mathrm{L}}_{\mathrm{i}}+{\mathsf{\alpha }}_{\mathrm{j}}{\mathrm{X}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{j}}$$

(1)

${\mathrm{Y}}_{\mathrm{i}\mathrm{j}}$ represents the choice of i farmer in the j kind of low-carbon agricultural technology, ${\mathrm{D}\mathrm{L}}_{\mathrm{i}}$ is the core independent variable, indicating the digital literacy level of i farmer, ${\mathrm{X}}_{\mathrm{i}}$ is a set of control variables, and ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{j}}$ is the error term.

(2) Based on the benchmark model, this paper constructs a mediating effect model to verify the research hypothesis H2, as shown in Equations (2) and (3).

$${Mid}_{ij}={\_cons}_{ij}+{\beta }_{1j}{DL}_i+{\beta }_j{X}_i+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{j}}$$

(2)

$$\mathrm{l}\mathrm{o}\mathrm{g}\mathrm{i}\mathrm{t}({\mathrm{Y}}_{\mathrm{i}\mathrm{j}})={\_\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{s}}_{\mathrm{i}\mathrm{j}}+{\mathsf{\gamma }}_{1\mathrm{j}}{\mathrm{D}\mathrm{L}}_{\mathrm{i}}+{{\gamma }_{2j}Mid}_{ij}+{\mathsf{\gamma }}_{\mathrm{j}}{\mathrm{X}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{j}}$$

(3)

In Equation (2), ${Mid}_{ij}$ represents the values of the mediating variables, including capital endowment (CE) and adoption willingness (WILL).

(3) Drawing lessons from the multiple mediating effect analysis method proposed by Wen Zhonglin and Ye Baojuan (2014), and based on Equations (1) and (2), this paper constructs a chain mediating effect model to verify the research hypothesis H4, that is, the path of “digital literacy → capital endowment → adoption willingness → low-carbon agricultural technology adoption behavior”, as shown in Equations (4) and (5).

$${WILL}_{ij}={\_cons}_{ij}+{\theta }_{1j}{DL}_i+{\theta }_{2j}{CE}_i+{\theta }_j{X}_i+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{j}}$$

(4)

$$\mathrm{l}\mathrm{o}\mathrm{g}\mathrm{i}\mathrm{t}({\mathrm{Y}}_{\mathrm{i}\mathrm{j}})={\_\mathrm{c}\mathrm{o}\mathrm{n}\mathrm{s}}_{\mathrm{i}\mathrm{j}}+{\mathsf{\mu }}_{1\mathrm{j}}{\mathrm{D}\mathrm{L}}_{\mathrm{i}}+{{\mu }_{2j}CE}_{ij}+{{\mu }_{3j}WILL}_{ij}+{\mathsf{\mu }}_{\mathrm{j}}{\mathrm{X}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{j}}$$

(5)

4. Empirical Analysis

4.1. Test of Direct Effect and Independent Mediating Effect

(1) Direct effect test

The results of the multicollinearity test show that the tolerances of all independent variables are greater than 0.5, and the values of the variance expansion factor (VIF) are all below 2, which shows that there is no serious multicollinearity problem among all independent variables, and further regression analysis can be carried out.

From

Table 2. Estimation results of direct effect of farmers’ adoption behavior of low-carbon agricultural technology.

	(1)	(2)	(3)	(4)
	Y1	Y2	Y3	Y4
DL	1.54 ***	1.55 ***	1.81 ***	2.42 ***
	(4.68)	(4.63)	(5.35)	(7.00)
SEX	0.07	0.05	0.12	0.03
	(0.53)	(0.34)	(0.82)	(0.23)
AGE	0.01 *	0.02 ***	0.01	0.004
	(1.66)	(3.15)	(1.35)	(0.65)
EDU	0.22 ***	0.19 ***	0.02	0.09
	(2.88)	(2.59)	(0.22)	(1.19)
VIL	0.4 *	0.25	0.29	1.1 ***
	(1.78)	(1.04)	(1.28)	(4.46)
HEALTH	0.08	0.17 **	0.33 ***	0.02
	(1.04)	(2.15)	(4.07)	(0.28)
LAND	0.35 ***	0.25 **	0.42 ***	0.12
	(2.71)	(1.96)	(3.29)	(1.00)
DISTANCE	−0.01	−0.1	−0.01	0.06
	(−0.12)	(−1.17)	(−0.14)	(0.66)
LABOR	0.11	0.15 **	0.15 **	−0.04
	(1.51)	(2.04)	(2.01)	(−0.57)
EXP	0.03 **	0.005	0.02 *	0.03 **
	(2.24)	(0.36)	(1.78)	(2.3)
LCA	0.33 ***	0.18 ***	0.28 ***	0.33 ***
	(5.06)	(2.76)	(4.37)	(5.08)
COO	−0.08	0.98 ***	1.32 ***	0.77 ***
	(−0.32)	(3.83)	(5.09)	(3.15)
Log likelihood	−1097.73 ***	−1070.60 ***	−1057.41 ***	−1054.62 ***
LR Chi2 (15)	159.23	139.36	231.18	228.87
N	742	742	742	742

Note: The values in parentheses represent t-values. *, **, and *** represent passing the significance test at the levels of 10%, 5%, and 1%, respectively.

, the direct effect estimation model of farmers’ low-carbon agricultural technology adoption behavior (1)~(4), it can be seen that farmers’ digital literacy (DL) can significantly and positively affect farmers’ adoption of different low-carbon agricultural technologies (Y1, Y2, Y3, Y4). The higher the farmers’ digital literacy, the higher the farmers’ adoption level of low-carbon agricultural technologies, which supports the research hypothesis H1. The endogenous test results show that the above models have not reached the significant level after Durbin and Wu–Hausman tests, and there is no endogenous problem.

4.2. Mediation Effect Analysis

In this paper, the Process plug-in of the SPSS 26.0 software is used, and the Bootstrap method is used to test the mediation effect. The results are shown in

Table 3. Analysis of bootstrap mediation effect.

Category	Path	Effect Value	Standard Deviation	95% Confidence Interval
				Lower Limit	Upper Limit
Direct Effect	DL → Y1	0.7692 ***	0.2216	0.3342	1.2042
Mediating Effect (CE)	DL → CE → Y1	0.1501 ***	0.0534	0.0537	0.2651
Mediating Effect (WILL)	DL → WILL → Y1	0.0934 ***	0.0419	0.0253	0.1873
Chain Mediating Effect	DL → CE → WILL → Y1	0.0068 ***	0.0045	0.0006	0.0177
Direct Effect	DL → Y2	0.7263 ***	0.2084	0.3172	1.1354
Mediating Effect (CE)	DL → CE → Y2	0.0863 ***	0.0346	0.0276	0.1623
Mediating Effect (WILL)	DL → WILL → Y2	0.1301 ***	0.0508	0.0405	0.2406
Chain Mediating Effect	DL → CE → WILL → Y2	0.0094 ***	0.0055	0.0009	0.0224
Direct Effect	DL → Y3	0.9935 ***	0.2319	0.5383	1.4486
Mediating Effect (CE)	DL → CE → Y3	0.0964 ***	0.0419	0.0283	0.1918
Mediating Effect (WILL)	DL → WILL → Y3	0.0798 ***	0.0425	0.0118	0.1756
Chain Mediating Effect	DL → CE → WILL → Y3	0.0058 ***	0.0040	0.0003	0.0155
Direct Effect	DL → Y4	1.1473 ***	0.2009	0.7529	1.5417
Mediating Effect (CE)	DL → CE → Y4	0.1507 ***	0.0519	0.0546	0.2585
Mediating Effect (WILL)	DL → WILL → Y4	0.0659 ***	0.0368	0.0097	0.152
Chain Mediating Effect	DL → CE → WILL → Y4	0.0048 ***	0.0032	0.0002	0.0123

Note: *** represent passing the significance test at the levels of 10%, 5%, and 1%, respectively.

. The indirect effect of digital literacy (DL) on the adoption of Y1 by conservation tillage low-carbon agricultural technology through the intermediary effect of capital endowment (CE) is 0.1122, and the 95% confidence interval does not include 0, which indicates that the intermediary effect of capital endowment (CE) is significant. The indirect effect of digital literacy (DL) on Y1 adoption of low-carbon agricultural technology in conservation tillage through the mediation of adoption willingness (WILL) is 0.2072, and the 95% confidence interval does not include 0, which indicates that the mediation effect of adoption willingness (WILL) is significant. The indirect effect of digital literacy (DL) on Y1 adoption of conservation tillage low-carbon agricultural technology through the chain intermediary effect of capital endowment (CE) and adoption willingness (WILL) is 0.0185, and the 95% confidence interval does not include 0, which indicates that the chain intermediary effect of capital endowment (CE) and adoption willingness (WILL) is significant. Similarly, verifiable capital endowment (CE) and adoption willingness (WILL) have significant independent intermediary and chain intermediary effects on integrated pest control technology Y2, water-saving irrigation technology Y3, and soil testing and formula fertilization technology Y4, respectively, and the research hypotheses H2, H3, and H4 have been verified (This study employs stepwise regression to analyze the independent mediating effects and chain mediating effects of capital endowment and adoption intention between digital literacy and the adoption of low-carbon agricultural technologies, yielding conclusions consistent with those in Table 3. Due to space limitations, detailed data and analysis processes are not presented here. For further information, please contact the author to obtain the complete dataset and analysis results).

4.3. Robustness Analysis

This paper mainly tests the robustness by the following three methods. First, replace variables, drawing on previous research [67], this paper measures digital literacy across six competency domains: CA1 digital equipment, CA2 data acquisition, CA3 digital communication, CA4 digital creation, CA5 digital security, and CA6 digital problem-solving. After standardizing the above six indicators by the entropy method, calculate the weight, objectively determine the level of each dimension of digital literacy, and further calculate the digital literacy. After replacing variables and controlling variables, orderly Logit regression is carried out. Due to space limitation, the robustness test only reports the regression results of the chain intermediary effect of soil testing and formula fertilization technology Y4, as shown in

Table 4. Estimation results of robustness test of farmers’ low-carbon agricultural technology adoption behavior.

	Substitution Variable				Ordered Probit Model Substitution				Winsorize
	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)	(16)
	Y4	CE	WILL	Y4	Y4	CE	WILL	Y4	Y4	CE	WILL	Y4
DL					1.339 ***	0.146 ***	0.51 ***	1.172 ***	2.428 ***	0.145 ***	0.447 ***	2.113 ***
					(6.91)	(3.3)	(3.49)	(5.96)	(7.03)	(3.29)	(3.23)	(6.01)
CA	1.769 ***	0.125 ***	0.421 ***	1.508 **
	(5.36)	(2.88)	(2.92)	(4.53)
CE			0.268 **	1.659 ***			0.254 **	0.951 ***			0.498 ***	1.59 ***
			(2.2)	(5.96)			(2.09)	(5.88)			(2.93)	(5.69)
WILL				0.247 ***				0.139 ***				0.216 **
				(2.85)				(2.78)				(2.48)
Control Variable	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Log Likelihood	−1065 ***			−1042 ***	−1059 ***			−1036 ***	−1054 ***			−1034 ***
LR chi2	207.32			254.04	219.56			264.31	228.99			270.02
F		11.38 ***	6.55 ***			11.63 ***	6.86 ***			11.62 ***	7.29 ***
Adj_R2		0.1439	0.0888			0.1469	0.0933			0.1467	0.0994
N	742	742	742	742	742	742	742	742	742	742	742	742

Note: The values in parentheses represent t-values. **, and *** represent passing the significance test at the levels of 10%, 5%, and 1%, respectively.

(5)~(8). Secondly, replace the econometric model, using the ordered Probit model to replace the original ordered Logit model, and the results are shown in Table 4 (9)~(12). Thirdly, the Winsorize method is selected to eliminate the outliers of samples, and the continuous variables (digital literacy (DL), capital endowment (CE), adoption willingness (WILL), age (AGE), planting scale (SCALE), labor force (LABOR), and farming years (EXP)) in all explanatory variables are scaled down by 1% and 99% quantiles, and the processed data are regressed again. The specific results are shown in Table 4 (13)~(16). The three robustness test results all show that the coefficient symbols and significance levels of digital literacy, capital endowment, and adoption willingness have not changed significantly, which verifies the robustness of the analysis results.

5. Heterogeneity Analysis

5.1. Analysis of the Influence of Different Dimensions of Digital Literacy on Farmers’ Adoption of Low-Carbon Agricultural Technology

In order to further explore the impact of different dimensions of digital literacy on farmers’ adoption of low-carbon agricultural technology, this paper uses the Process plug in the SPSS software and the Bootstrap method to test the mediating effects of digital information acquisition literacy (DL1), digital content creation literacy (DL2), digital security literacy (DL3), and digital problem-solving literacy (DL4), respectively.

According to the direct effect test results in

Table 5. Influence of different dimensions of digital literacy on the adoption of low-carbon agricultural technologies.

Category	Path	Effect Value	Standard Deviation	95% Confidence Interval
				Lower Limit	Upper Limit
The direct effect of DL1 on low-carbon agricultural technology adoption	DL1 → Y1	0.0890 ***	0.0430	0.0047	0.1733
	DL1→ Y2	0.1080 ***	0.0406	0.0283	0.1876
	DL1 → Y3	0.1795 ***	0.0448	0.0915	0.2675
	DL1 → Y4	0.1787 ***	0.0391	0.1019	0.2555
The direct effect of DL2 on low-carbon agricultural technology adoption	DL2 → Y1	0.1257 ***	0.0423	0.0427	0.2088
	DL2 → Y2	0.1197 ***	0.0400	0.0412	0.1982
	DL2 → Y3	0.0812 ***	0.0447	0.0065	0.1689
	DL2 → Y4	0.1429 ***	0.0388	0.0666	0.2191
The direct effect of DL3 on low-carbon agricultural technology adoption	DL3 → Y1	0.1474 ***	0.0439	0.0612	0.2336
	DL3→ Y2	0.1100 ***	0.0414	0.0287	0.1913
	DL3 → Y3	0.1818 ***	0.0460	0.0914	0.2722
	DL3 → Y4	0.2232 ***	0.0399	0.1449	0.3016
The direct effect of DL4 on low-carbon agricultural technology adoption	DL4 → Y1	0.0001 ***	0.0508	0.0998	0.0996
	DL4 → Y2	0.1230 ***	0.0477	0.0101	0.2167
	DL4 → Y3	0.0534 ***	0.0534	0.0513	0.1582
	DL4 → Y4	0.0446 ***	0.0467	0.0472	0.1363

Note: *** represent passing the significance test at the levels of 10%, 5%, and 1%, respectively.

, all dimensions of digital literacy have a significant positive impact on farmers’ adoption of low-carbon agricultural technologies. This indicates that higher digital literacy enables farmers to more effectively acquire, understand, and apply low-carbon agricultural technologies, thereby increasing the likelihood and enthusiasm for adoption. However, the direct effects of the different dimensions of digital literacy on low-carbon technology adoption exhibit heterogeneity.

Water-saving irrigation technology (Y3) and soil testing and formulated fertilization technology (Y4) are primarily influenced by digital information acquisition literacy (DL1) and digital security literacy (DL3), suggesting that these technologies strongly rely on precise data analysis, sensor data processing, and risk management. Integrated pest management technology (Y2) is significantly affected by digital content creation literacy (DL2) and digital problem-solving literacy (DL4), indicating that the adoption of this technology depends on farmers’ ability to exchange technical knowledge, share information, and respond to agricultural production challenges. Meanwhile, conservation tillage technology (Y1) is less influenced by all dimensions of digital literacy, suggesting that its adoption is more dependent on external promotion policies and farmers’ accumulated production experience.

From the results of the intermediary effect test in

Table 6. Influence of different dimensions of digital literacy on the adoption of IPM low-carbon agricultural technology.

Category	Path	Effect Value	Standard Deviation	95% Confidence Interval
				Lower Limit	Upper Limit
Direct effect	DL1 → Y2	0.1080 ***	0.0406	0.0283	0.1876
Mediating effect CE	DL1 → CE → Y2	0.0143 ***	0.0065	0.0031	0.0287
Mediating effect WILL	DL1 → WILL → Y2	0.0200 ***	0.0092	0.0042	0.0400
Chain mediating effect	DL1 → CE → WILL → Y2	0.0017 ***	0.0011	0.0001	0.0042
Direct effect	DL2 → Y2	0.1197 ***	0.0400	0.0412	0.1982
Mediating effect CE	DL2 → CE → Y2	0.0148 ***	0.0065	0.0038	0.0290
Mediating effect WILL	DL2 → WILL → Y2	0.0178 ***	0.0087	0.0031	0.0369
Chain mediating effect	DL2 → CE → WILL → Y2	0.0018 ***	0.0010	0.0002	0.0042
Direct effect	DL3 → Y2	0.1100 ***	0.0414	0.0287	0.1913
Mediating effect CE	DL3 → CE → Y2	0.0121 ***	0.0063	0.0013	0.0263
Mediating effect WILL	DL3 → WILL → Y2	0.0174 ***	0.0092	0.0017	0.0381
Chain mediating effect	DL3 → CE → WILL → Y2	0.0015 ***	0.001	0.0001	0.0037
Direct effect	DL4 → Y2	0.1230 ***	0.0477	0.0101	0.2167
Mediating effect CE	DL4 → CE → Y2	0.0173 ***	0.0068	0.0055	0.0322
Mediating effect WILL	DL4 → WILL → Y2	0.0466 ***	0.0136	0.0224	0.0754
Chain mediating effect	DL4 → CE → WILL → Y2	0.0016 ***	0.001	0.0001	0.0039

Note: *** represent passing the significance test at the levels of 10%, 5%, and 1%, respectively.

, capital endowment and adoption willingness can play a significant independent intermediary effect in the process of influencing farmers’ low-carbon agricultural technology adoption in all dimensions of digital literacy. The test results of the chain intermediary effect show that capital endowment and adoption willingness can play a significant chain intermediary effect in the process of influencing farmers’ low-carbon agricultural technology adoption in all dimensions of digital literacy, which shows that farmers’ digital literacy in information acquisition, content creation, digital security, and problem-solving is an important factor to promote the adoption of low-carbon agricultural technology. By exerting the direct effect of digital literacy and the independent intermediary and chain intermediary effect of capital endowment and adoption willingness, farmers’ use of low-carbon agricultural technology can be effectively promoted (Due to space constraints, Table 6 only presents the direct and mediating effects of the different dimensions of digital literacy on farmers’ adoption of integrated pest management (IPM) low-carbon agricultural technology).

5.2. Intermediary Effect Analysis of Different Dimensions of Capital Endowment on Farmers’ Adoption of Low-Carbon Agricultural Technology

In the process of digital literacy affecting farmers’ adoption of low-carbon agricultural technology, in order to further analyze the differences of different dimensions of capital endowment, this paper takes natural capital (NC), human capital (HUM), material capital (MC), social capital (SC) and economic capital (ECO) as intermediary variables in the intermediary effect model for intermediary effect analysis. First, the mediating effect of natural capital on the adoption of low-carbon agricultural technologies by farmers is not significant. This may be due to the weak dependence on the adoption of conservation tillage techniques, integrated pest and disease control techniques, and soil testing and formula fertilization techniques on the land scale, while the adoption of water-saving irrigation techniques requires supporting infrastructure support, and it is difficult for natural capital alone to independently promote technology adoption. Secondly, the mediating effect of human capital on farmers’ adoption of low-carbon agricultural technology is significant. The reason may be that human capital promotes farmers’ technical knowledge, skills, and innovation consciousness, makes it easier to understand and master low-carbon agricultural technology, and thus increases their willingness and ability to adopt low-carbon agricultural technology, and finally promotes farmers’ adoption of low-carbon agricultural technology. Further regression analysis shows that human capital and adoption willingness play a chain mediating effect in the influence of digital literacy on farmers’ adoption of low-carbon agricultural technology. Third, material capital has a significant intermediary effect on the adoption of conservation tillage technology and water-saving irrigation technology because conservation tillage and water-saving irrigation technology need specific mechanical equipment and infrastructure support, while other low-carbon technologies (such as pest control, soil testing and formula fertilization, etc.) rely less on material capital. Fourthly, the mediating effect of social capital on conservation tillage technology, integrated pest control technology, and soil testing and formula fertilization technology is significant, while the mediating effect on water-saving irrigation technology is not significant. At the same time, economic capital only has a significant mediating effect on water-saving irrigation technology. The reasons may be that conservation tillage, integrated pest control technology, and soil testing and formula fertilization technology depend on information sharing and technology popularization, and are easy to spread through social networks, while the adoption of water-saving irrigation technology depends more on supporting infrastructure and capital investment. Only with a certain economic capital can farmers bear the construction cost, equipment maintenance, and continuous operation costs of water-saving irrigation system. Therefore, economic capital plays a significant intermediary role in the adoption of water-saving irrigation technology. (Due to space constraints,

Table 7. Analysis of the influence of different dimensions of capital endowment on the mediating effect of water-saving irrigation technology adoption.

	Direct Effect	Natural Capital		Human Capital		Material Capital		Social Capital		Economic Capital
	(17)	(18)	(19)	(20)	(21)	(22)	(23)	(24)	(25)	(26)	(27)
	Y3	NC	Y3	HUM	Y3	MC	Y3	SC	Y3	ECO	Y3
DL	1.81 ***	0.32	1.81 ***	0.14**	1.76 ***	0.27 *	1.77 ***	0.35 ***	1.77 ***	1.27 ***	1.63 ***
	(5.35)	(0.26)	(5.36)	(1.99)	(5.18)	(1.75)	(5.22)	(2.62)	(5.20)	(7.78)	(4.62)
NC			0.02 *
			(1.78)
HUM					0.43 **
					(2.53)
MC							0.14 * (1.70)
SC									0.13 (1.42)
ECO											0.15 *
											(1.94)
Control	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Variable
Log Likelihood	−1057 ***		−1055 ***		−1054 ***		−1056 ***		−1056 ***		−1056 ***
LR chi2	231.18		234.44		237.61		234.07		233.21		234.93
F		27.01 ***		8.47 ***		6.18 ***		5.23 ***		12.58 ***
Adj_R2		0.2964		0.1079		0.0774		0.0641		0.1579
N	742	742	742	742	742	742	742	742	742	742	742

Note: The values in parentheses represent t-values. *, **, and *** represent passing the significance test at the levels of 10%, 5%, and 1%, respectively.

only presents the mediating effects of different capital endowments on the adoption of water-saving irrigation technology).

6. Conclusions

6.1. Research Conclusions

Based on the questionnaire data of 742 farmers in Shandong Province, this paper empirically tests the impact of digital literacy on farmers’ adoption of low-carbon agricultural technology by using the ordered Logit model, and explores the intermediary role and chain intermediary role of capital endowment and adoption willingness. This study considers low-carbon agricultural technologies as a “technology package”, which includes conservation tillage technology (Y1), integrated pest management (IPM) technology (Y2), water-saving irrigation technology (Y3), and soil testing and formulated fertilization technology (Y4). The main research conclusions are as follows:

1. Digital literacy significantly promotes farmers’ adoption of low-carbon agricultural technologies. However, due to differences in the application methods and carbon reduction mechanisms of these technologies, farmers’ reliance on different dimensions of digital literacy varies when adopting them. As a result, the impact mechanisms of various digital literacy dimensions on technology adoption exhibit significant heterogeneity. Information acquisition and security literacy have a greater influence on technologies that rely on data analysis and precision management, such as water-saving irrigation, soil testing, and formulated fertilization. In contrast, content creation and problem-solving abilities play a more prominent role in the adoption of knowledge-sharing and practice-based technologies, such as integrated pest management. Comparatively, the adoption of conservation tillage technologies is more influenced by policy support and farmers’ practical experience, with digital literacy playing a relatively smaller role. After a series of tests, such as changing the measurement method of farmers’ digital literacy, introducing an ordered Probit model, re-regression analysis after 1% censoring of variables, and endogenous analysis by an instrumental variable method, the conclusions are still valid.

2. Mechanism analysis indicates that capital endowment and adoption willingness play both independent and chain-mediating roles in the impact of digital literacy on farmers’ adoption of low-carbon agricultural technologies. On the one hand, an improvement in capital endowment provides the necessary resources for farmers to adopt low-carbon technologies, while an increase in adoption willingness enhances farmers’ recognition and acceptance of these technologies. On the other hand, digital literacy not only directly promotes the adoption of low-carbon agricultural technologies but also enhances capital endowment, enabling farmers to more effectively acquire and utilize information, strengthen their technical understanding, and improve their risk tolerance. This, in turn, further reinforces their willingness to adopt low-carbon technologies, ultimately facilitating their widespread application.

3. Heterogeneity analysis shows that different dimensions of digital literacy significantly promote the adoption of low-carbon agricultural technology by farmers, and capital endowment and adoption willingness play a significant chain intermediary role. The different dimensions of capital endowments play a differentiated intermediary role in the process of digital literacy affecting farmers’ adoption of low-carbon agricultural technologies, among which human capital and material capital have significant intermediary effects on conservation tillage and water-saving irrigation technologies, social capital plays an active role in the adoption of technologies with strong information dependence, and economic capital has a key impact on the adoption of water-saving irrigation technologies.

6.2. Policy Implications

1. Implement multiple measures to enhance farmers’ digital literacy:

(1) Digital information acquisition literacy: The government, in collaboration with large-scale grain farmers, agricultural cooperatives, agricultural technology extension stations, and Internet enterprises, should regularly conduct low-carbon agricultural technology promotion training programs in rural areas. These programs should educate farmers on the applications of low-carbon agricultural technologies. Additionally, a localized agricultural information app should be developed to provide real-time updates on low-carbon agricultural techniques, market trends, and policy information, ensuring that farmers can efficiently and conveniently access the latest agricultural information.

(2) Digital content creation literacy: Implement a “Farmer Short Video + Livestreaming Empowerment Program” to encourage farmers to document and share agricultural production experiences. Farmers can use platforms such as Douyin and Kuaishou to promote low-carbon agricultural technologies. Outstanding content creators should receive financial incentives to further motivate participation.

(3) Digital security literacy: Integrate anti-fraud alerts and security guidelines into agricultural apps. Additionally, village-level cybersecurity awareness workshops should be promoted, incorporating real-life fraud cases to enhance farmers’ ability to recognize and prevent cyber threats.

(4) Digital problem-solving literacy: Establish rural low-carbon agricultural technology consultation stations, utilizing a “Hotline + App + Livestreaming + AI Smart Customer Service” model to provide remote technical support. This ensures that farmers can access precise and efficient low-carbon agricultural technology guidance anytime and anywhere, enhancing their digital problem-solving skills in agricultural production.

2. Strengthening farmers’ capital endowment to enhance the resource guarantee for low-carbon technology adoption:

(1) Human capital: Implement a “Technology Specialist Rural Residency Program” in agricultural technology extension stations. Agricultural experts should be assigned to rural areas to conduct quarterly field training sessions and integrate with rural revitalization projects to form a sustainable support mechanism. For instance, Zhejiang Jiaxing’s technology specialist model has successfully promoted efficient water-saving irrigation technologies, providing a valuable reference.

(2) Social capital: Promote a “Farmers’ Cooperatives + Agricultural Insurance” model, where the government takes the lead in encouraging farmers to invest in low-carbon agricultural projects through mutual aid funds. Cooperatives should facilitate group purchasing discounts for low-carbon agricultural equipment. For example, the Nanzhun Qianjin Yonggen Ecological Fishery Cooperative in Huzhou, Zhejiang established a mutual aid fund to support freshwater aquaculture farmers struggling with high initial costs, helping members overcome financing difficulties.

(3) Material capital: Provide financial subsidies for low-carbon agricultural equipment, such as establishing a farm machinery trade-in subsidy program to promote the adoption of water-saving irrigation and precision fertilization technologies. A reference model is the farm machinery sharing program implemented by agricultural cooperatives in Shouguang, Shandong, where the government subsidizes cooperatives to set up farm machinery rental stations, thereby reducing individual farmers’ equipment acquisition costs.

(4) Economic capital: Banks should collaborate with government subsidy programs to introduce “Low-Carbon Agricultural Special Loans”, reducing farmers’ financing costs. Furthermore, referring to the “Contract Farming + Futures + Finance” model piloted in Shaoyang, Hunan, agricultural enterprises or cooperatives should sign contract farming agreements with farmers to lock in product prices. Meanwhile, futures markets can be used to hedge price fluctuations, stabilizing farmers’ income and mitigating the financial risks associated with adopting low-carbon agricultural technologies.

3. Enhancing farmers’ willingness to adopt low-carbon technologies and stimulating their intrinsic motivation:

(1) Establishing low-carbon technology demonstration bases: Build low-carbon agricultural demonstration villages or farms in key agricultural areas to provide farmers with on-site learning and practical opportunities. For instance, the “Water-Saving Agriculture Demonstration Park” in Xinxiang, Henan, follows a government-led + enterprise-invested + cooperative-promoted model. It offers farmers open access to demonstration areas for water-saving irrigation and precision fertilization, allowing them to experience the benefits of low-carbon technologies firsthand and improving their acceptance.

(2) Implementing economic incentive policies: Introduce a “Low-Carbon Subsidy” or “Low-Carbon Credit Points” system, allowing farmers to exchange accumulated credits for agricultural inputs, access low-interest loans, and enjoy insurance premium reductions. For example, Jiangsu Province has launched a “Low-Carbon Agricultural Credit Points + Credit Loan” model, which has benefited over 2000 farmers, reducing financial barriers to low-carbon technology adoption.

(3) Strengthening market protection mechanisms: Promote the “Cooperative + Farmer + Enterprise” model to expand contract farming, facilitate low-carbon agricultural product certification, and support cooperatives in connecting with e-commerce platforms. These measures will broaden sales channels for low-carbon agricultural products, enhance market stability, and reduce market risks for farmers.

6.3. Limitations and Prospects for the Future

(1) Extension of data time span:

This study utilizes cross-sectional data, which accurately captures farmers’ adoption behavior of low-carbon agricultural technologies at a specific point in time and incorporates key variables in the research design to reduce omitted variable bias. However, cross-sectional data cannot track dynamic behavioral changes over time. Future research could extend the data dimension by adopting panel data or time-series data to capture the dynamic evolution of farmers’ behavior and explore the long-term impact mechanisms of low-carbon agricultural technology adoption.

(2) Expansion of external influencing factors:

This study focuses on the impact of digital literacy, capital endowment, and adoption willingness on the adoption of low-carbon agricultural technologies using cross-sectional data. Key variables, including farmers’ characteristics and production environment, have been controlled to minimize the interference of external factors. However, considering the dynamic changes in the policy environment and market conditions, future research could integrate panel data or time-series analysis to track shifts in policy environments, market conditions, and other external factors, thereby providing a more comprehensive understanding of their long-term influence on farmers’ adoption behavior.

(3) Validation of regional applicability:

This study is based on survey data from farmers in Shandong Province, a region with a strong representation in agricultural development, making its findings valuable for reference in other areas. However, differences in agricultural development models, policy environments, and market conditions may exist across regions. Future research could expand to areas with significantly different agricultural development models, such as regions dominated by facility agriculture or ecological agricultural demonstration zones, to further enhance the generalizability of the research findings.