数字化与可持续农业：基于PRISMA的系统文献综述 涵盖62项2015-2026年实证研究

1. Introduction

Agriculture occupies a structurally irreplaceable position in the global economy. It contributes approximately 24% of global gross domestic product, employs around 1.3 billion people (roughly 16% of the world’s workforce), and underpins food security across approximately 1.2 to 1.5 billion hectares of cultivated land [1]. Yet the sector faces mounting pressure on multiple fronts: a growing global population, rising per capita food demand, accelerating climate change, and persistent rural poverty [2]. These intersecting challenges have renewed the urgency of a question posed by the International Assessment of Agriculture, Science and Technology for Development (IAASTD): how can hunger and poverty be alleviated, rural livelihoods enhanced, and equitable, environmentally, socially, and economically sustainable development advanced through the generation and application of agricultural knowledge and technology [3]?

Over the past decade, one of the most prominent responses to this challenge has been the rapid digitalisation of the agricultural sector. Digital technologies, including information and communication technologies (ICTs), precision agriculture tools, artificial intelligence, remote sensing, and digital platforms, are increasingly regarded as instruments of agricultural transformation. Existing research identifies a convergence of drivers: economic pressures that reward efficiency gains, environmental imperatives that demand more precise use of resources, and social objectives that require the inclusion of smallholder farmers and rural communities in the benefits of technological change [4]. Studies from diverse national contexts reflect these dynamics: they document productivity gains in the Chinese agricultural sector [5,6], report increased adoption of modern practices among Saudi Arabian farmers following access to digital information [7], and link digital agriculture to reductions in agricultural carbon emissions [8]. Finally, the literature highlights that the transformative potential of digital tools depends critically on inclusive, co-created knowledge transfer among farmers, consultants, and other stakeholders [9]. Despite this growing body of work, the literature remains fragmented. Most studies focus on a single technology, country, or dimension of sustainability, typically the economic dimension. Integrative analyses that assess the impact of digitalisation across all three dimensions of sustainable development remain scarce, and potential trade-offs between dimensions are insufficiently explored [10]. The present study addresses this gap by analysing the impact of digitalisation on the sustainable development of agriculture through a systematic synthesis of empirical evidence and by identifying directions for future research.

To achieve this aim, three objectives were set: to conduct a PRISMA-compliant systematic review of empirical literature on digitalisation and sustainable agriculture; to systematically identify and compare the positive and negative effects of digitalisation on the economic, environmental, and social dimensions of agricultural sustainability; and to identify research gaps and priority directions for further enquiry. The study is guided by two research questions: (1) Which geographical regions have been the focus of research on digitalisation and agricultural sustainability, and how has the volume of related publications changed from 2015 to 2026? (2) Do all studies report positive sustainability outcomes associated with digitalisation, or have adverse effects also been documented?

Theoretical Framework

The analytical structure of this review is grounded in the concept of sustainable development, as defined by the World Commission on Environment and Development (the Brundtland Commission), which characterised sustainability as development that “meets the needs of the present without compromising the ability of future generations to meet their own needs” [11]. This foundational definition has since been operationalised through the triple-bottom-line framework, which recognises three interdependent dimensions of sustainability: economic, environmental, and social [12]. The framework is now embedded in the United Nations’ 2030 Agenda and its 17 Sustainable Development Goals (SDGs) [13]. Several of these, including SDG 2 (Zero Hunger), SDG 8 (Decent Work and Economic Growth), SDG 13 (Climate Action), and SDG 15 (Life on Land), are directly relevant to agriculture.

In the agricultural context, these three dimensions can be defined as follows. Economic sustainability is the capacity of agricultural systems to generate productive output, maintain profitability, and support rural livelihoods over the long term, as reflected in indicators such as total factor productivity, crop yields, income, and market participation. Environmental sustainability is the capacity to preserve natural resources and ecological functions, particularly through reduced carbon emissions, efficient use of land and water, and improvements in biodiversity and soil health. Social sustainability concerns the conditions under which agricultural labour is performed, including the distribution of income gains, equitable access to technology, the quality of working conditions, and the protection of rural communities from displacement or marginalisation.

This three-dimensional framework directly informs the review’s organisational structure. The synthesis of empirical findings in Section 3 is organised around these three dimensions, enabling a systematic examination of whether digitalisation contributes to, detracts from, or produces uneven outcomes across economic, environmental, and social sustainability. Critically, the framework also highlights potential trade-offs between dimensions, for example, when gains in economic productivity come at the cost of labour displacement, or when environmental improvements require capital investments that disadvantage smallholders. Recognising these trade-offs is a central objective of this review and a distinctive contribution to the existing literature.

2. Materials and Methods

2.1. Overview and Design

This study employs a systematic literature review in line with the PRISMA 2020 guidelines [14], and the checklist could be found in the Supplementary Material. The PRISMA framework was selected for its transparent, reproducible protocol for identifying, screening, and synthesising empirical evidence, thereby minimising selection bias and enhancing the reliability of conclusions. The review adopts a combined deductive–inductive analytical approach: deductive methods were used to map the geographical and temporal distribution of research (RQ1), while inductive thematic analysis was applied to synthesise substantive findings across sustainability dimensions (RQ2). Data analysis drew on both quantitative methods, aggregating numerical indicators such as yield changes and productivity growth, and qualitative thematic coding, in which the 62 included studies were systematically categorised according to the economic, environmental, and social dimensions of the theoretical framework described in the “Theoretical Framework” section. A formal meta-analysis was not conducted because preliminary analysis revealed substantial heterogeneity across the included studies in outcome measures, research designs, analytical methods, and reported effect estimates. Consequently, calculating comparable pooled effect sizes and applying statistical methods for quantitative synthesis, including a funnel plot and Egger’s test to assess publication bias, were not methodologically justified.

2.2. Data Sources and Search Strategy

A systematic search was conducted in two stages: the first stage was conducted on 29 January 2025 using a single electronic database; Scopus and Web of Science (WoS) was incorporated only in the second stage, the 2026 update (see Section 2.4).

Scopus was selected as the primary database for its extensive multidisciplinary coverage and strong performance in systematic search retrieval. WoS was included as a complementary source to broaden disciplinary coverage and reduce the risk of database-specific indexing gaps, consistent with best practice in systematic reviewing [14].

The search string actually executed in Scopus (Article title, Abstract, Keywords field) was the free-text phrase “digitalisation of agriculture”, which Scopus parsed as digitalisation AND of AND agriculture. The term “precision agriculture” was not used as a separate search term, as the review aimed to examine the broader phenomenon of agricultural digitalisation rather than a single technological approach. Precision agriculture was therefore treated as one form of agricultural digitalisation within the review’s broader conceptual scope. It is important to note that studies focused on precision agriculture and related technologies were identified through the search strategy and are represented among the 62 studies included in the final sample.

The following filters were then applied sequentially, with the resulting record counts: publication years 2015–2025 (n = 1353); subject areas Computer Science, Environmental Science, Agricultural and Biological Sciences, Social Sciences, Business, Management and Accounting, Economics, Econometrics and Finance, Decision Sciences, and Multidisciplinary (n = 1233); document language limited to English (n = 1186); source/document type limited to Journal and Conference Proceeding (n = 927); the Environmental Science subject category was then excluded to avoid duplication with other categories (n = 717).

2.3. Eligibility Criteria

Inclusion and exclusion criteria were developed collaboratively by all four authors (Zh.S., U.S., A.M., G.D.) before screening and are presented in Table 1. The criteria were designed to include only empirical studies that directly examined the impact of digitalisation on agricultural sustainability outcomes, while excluding review articles, theoretical or conceptual papers, and studies focused on non-agricultural sectors.

Table 1. Inclusion and exclusion criteria applied in the systematic literature review.

2.4. Study Selection

After database searches and deduplication, 739 records were retained for title and abstract screening. Screening was conducted independently by three authors (Zh.S., A.M., G.D.) using the eligibility criteria outlined in Section 2.3. Discrepancies between screeners were recorded: Zh.S. identified 42 eligible articles, A.M. 45, and G.D. 42. All discrepancies were resolved through discussion and consensus arbitration by the fourth author (U.S.). A total of 40 studies were included in the final synthesis. The full selection process is documented in the PRISMA 2020 flow diagram (Figure 1).

Figure 1. PRISMA flowchart illustrating the study selection process for the first search round (Round 1, Scopus, January 2025). See Table 2 for the second search round (Round 2, 2026 update). Source: (Page et al., 2021) [14].

Table 2. PRISMA-style flow for the second search round (Round 2, 2026 update), combining Scopus and Web of Science.

Figure 1. PRISMA flowchart illustrating the study selection process for the first search round (Round 1, Scopus, January 2025). See Table 2 for the second search round (Round 2, 2026 update). Source: (Page et al., 2021) [14].

Search Update (Round 2)

On 5 February 2026, the Scopus search was re-run, retrieving 278 additional records published in 2026. On 17 February 2026, Web of Science was searched for the first time, covering 2015–2026 and retrieving 259 records. The authors judged that a single-database search was insufficient for a comprehensive review and therefore incorporated WoS at this stage. The combined total of 537 records (278 Scopus + 259 WoS) was screened for duplicates; 25 duplicate records were identified and removed, leaving 512 distinct records (278 Scopus + 234 WoS) for screening against the eligibility criteria in Table 1. Applying these criteria to the 512 records identified 22 additional eligible studies, which were added to the 40 studies retained from the first search round, yielding a final total of 62 studies included in the review.

2.5. Data Extraction and Quality Assessment

Data from the 62 included studies were extracted by three authors (Zh.S., A.M., G.D.) using a standardised extraction template. For each study, the following information was recorded: author(s) and year; country/region of study; research design and data collection method; digital technology or intervention examined; sustainability dimension(s) addressed; key quantitative or qualitative findings; and direction of effect (positive, negative, or mixed). The full extraction is presented in the Appendix A.

No formal quality appraisal tool (such as the Mixed Methods Appraisal Tool, MMAT [15], or the Joanna Briggs Institute checklists) [16] was used in this review, given the heterogeneity of study designs and methodological approaches across the 62 included studies. This is acknowledged as a limitation of the review (see Section 4.5).

2.6. Synthesis Approach

Given the heterogeneity of study designs, geographic contexts, and outcome measures across the included studies, a narrative synthesis was adopted rather than a meta-analysis. Synthesis proceeded in two stages. First, a descriptive analysis mapped the distribution of studies by geographical region, publication year, and sustainability dimension addressed (addressing RQ1). Second, a deductive–inductive thematic analysis synthesised the substantive findings within each sustainability dimension (addressing RQ2). Initial codes were derived deductively from the three-dimensional theoretical framework (economic, environmental, and social sustainability. Sub-themes were then developed inductively from patterns emerging within each dimension. Coding was conducted by three authors (Zh.S., A.M., G.D.) and reviewed for consistency by the fourth (U.S.).

3. Results

This section is organised in accordance with the study’s four guiding research questions. The first question provides contextual information on the digitalisation of agriculture, whereas the subsequent three questions focus on a detailed analysis of how digitalisation affects the sustainable development of the agricultural sector.

3.1. RQ1. Which Geographical Regions Have Been the Focus of Research on the Impact of Digital Technologies on Agricultural Sustainability, and How Has the Volume of Related Publications Evolved in Recent Years?

All studies included in this review were conducted in countries across three continents, Europe, Asia and Africa. However, the geographic distribution of studies was uneven across these regions (see Figure 2). Among the 37 studies conducted in Asia (59.7% of the studies), the overwhelming majority, 26 studies or 42% of the total sample, were conducted in China, which is not surprising. China is one of the world’s largest producers of agricultural products and has been striving not only to expand production volumes but also to improve product quality. The Chinese government has strengthened agricultural land protection and implemented various incentives to promote agricultural production [17]. In addition, the country continues to liberalise trade in agricultural products [18]. As a result of these combined factors, China has effectively integrated digital technologies into both the production and trade sectors of its agriculture. Following China in the ranking are India (4.84%) and Kazakhstan (3.2%), both of which are major producers of staple crops such as rice and wheat. Other contributing countries include Indonesia (1.6%), Uzbekistan (1.6%), Vietnam (3.2%), Azerbaijan (1.6%) and members of the South Asian Association for Regional Cooperation (SAARC), including Bangladesh, Nepal, Pakistan, Bhutan, the Maldives, and Afghanistan (1.6%).

Figure 2. Geographical distribution of studies. Source: Author’s elaboration using Bing Maps.

Europe accounts for 22.58% of the studies (14 articles), making it the second most represented continent. Germany stands out as the most researched country in Europe (6.45%). As one of the leading agricultural producers in the European Union [19], German researchers, citing previous studies, argue that nearly 50% of agricultural professional activities in the country could be replaced by digital technologies [20]. These trends have contributed to Germany’s heightened interest in the digitalisation of its agricultural sector.

Africa ranks third in terms of the number of studies, accounting for 17.7% of the total. In Africa, the greatest interest in the impact of digitalisation on agriculture was shown by researchers in Ghana (6.45%) and Sub-Saharan African countries (4.84%). Additional studies were carried out in Tanzania (1.6%), Nigeria (1.6%), Somali (1.6%) and other African nations (1.6%). The focus on digital agriculture in Africa is lar