Adoption of Innovative Technologies for Sustainable Agriculture: A Scoping Review of the System Domain

Abstract

The agricultural sector is undergoing a profound transformation driven by the integration of innovative technologies and practices, but the adoption of these technologies remains uneven. Holistic approaches to the diffusion of innovative technologies in agriculture are seen as crucial for effective adoption and sustainable development. In this context, the systemic dimension of technology adoption is characterized by the interactions between actors that create knowledge and promote the process of technology adoption. Therefore, the overall objective of this study is to provide a comprehensive analysis of the current state of the art in relation to the systemic dimension of the process of technology adoption in developed countries. Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) extension protocol for scoping reviews, we examined the literature to capture the role of the systems dimension in the process of technology adoption. We conducted a two-analysis, bibliometric and content network analysis to identify the concepts and thematic clusters that define the systemic dimension and represent the main drivers of technology adoption for sustainable development in agriculture. The results show that the factors influencing the adoption of agricultural technologies are treated inconsistently in the literature, with a focus on technological and economic aspects rather than systemic elements such as governance and stakeholder interactions.

1. Introduction

The global agricultural sector faces the dual challenge of increasing productivity to meet the growing demand for food, while at the same time tackling pressing environmental issues [1,2]. To respond to these challenges, many regions around the world are exploring strategies to transition to more sustainable agricultural systems [3], adopting digital technologies to improve both productivity and environmental sustainability [4].

The adoption of alternative agricultural models characterised by technological innovation, diversification and sustainable agricultural practices has already been recognized as essential [5], especially in the context of regional challenges [6].

Innovative technologies play a key role in the modernization of agriculture by generating large amounts of valuable data through tools such as remote sensing, the Internet of Things and artificial intelligence [7,8,9]. These technologies, which go beyond traditional precision agriculture [10], have the potential to contribute to more sustainable agriculture by improving productivity, reducing the environmental footprint, and increasing resilience to climatic risks [11].

As innovations continue to gain traction, it is important to understand the factors that influence their adoption and diffusion [12,13].

The literature emphasizes the complexity of the adoption process [14] and identifies numerous factors that influence the decision-making process of farmers [15,16,17,18,19,20].

Farmers’ ability to integrate digital solutions into their workflows depends on several interrelated factors, including farming characteristics, peer networks, institutional support, technological characteristics, and attitudinal factors [13,19]. Among these factors, perceived economic benefits such as profitability and return on investment are the most important drivers of farmer engagement, emphasizing the importance of effectively communicating these benefits across systems [6]. Therefore, understanding and addressing these challenges is critical to promoting agriculture’s transition to a more sustainable system [21].

More recent perspectives emphasize the importance of a holistic approach that considers both individual farm-level decisions and broader systemic interactions [13].

Among others, some authors [22] point out that a possible way forward could be through farmer needs assessment and co-design initiatives, although these receive little attention in the literature.

In this context, the role of extension services and technical assistance is increasingly seen as crucial in facilitating knowledge sharing and technology adoption. Policies that provide such services are seen as key to overcoming knowledge gaps and promoting the diffusion of innovative technologies in agriculture [6]. Therefore, a systemic perspective is essential for understanding the broader determinants of agricultural technology adoption.

In outlining the systemic dimension, the literature shows a conceptually consistent understanding of the factors influencing the adoption of digital technologies. However, this is accompanied by some lexical variation in the way these elements are categorized, as well as a significant lack of procedures for operationalizing the concept.

We retain that the boundaries of the systemic dimension are defined by the actors interacting within the system and not by the technological elements that characterize the system itself. Basically, the systemic dimension is characterized by the relationships that generate and disseminate knowledge, promote the process of adoption and diffusion of technological innovations, and define its dimension. The result is a complex picture that encompasses the interactions between different actors (peers, providers, advisors, cooperatives, institutions, and the market) and infrastructures that influence the diffusion and integration of technological innovations. It encompasses technical, economic, social, and regulatory aspects in a dynamic and interconnected perspective [19,22,23,24].

Most research has focused on the adoption at the operational level and in the area of enabling technologies [25], while more recent literature emphasizes the need to engage stakeholders and co-develop solutions to improve adoption success [26,27,28,29,30].

Indeed, stakeholder engagement is increasingly valued as a key strategy for bridging the gap between technological development and practical application, particularly in the context of digital technologies that enable more resilient agricultural systems [11,22].

Recent studies emphasize that environmental and productivity goals dominate the discourse, while socio-economic and governance dimensions receive comparatively little attention despite being crucial for the successful adoption of agricultural innovations [31]. It is also emphasized that the sustainability of food systems is often constrained by structural challenges such as inefficient governance, market barriers, and conflicting policy objectives [31].

In their study, authors [19] found that the external drivers of technology diffusion, i.e., the external elements that influence operations beyond the boundaries of the organization, were the least studied category. However, several external drivers were identified, such as social influence and networking, government support, information sources, and societal and regulatory influence. In addition, issues of data sharing and ownership have come to the fore in relation to the adoption of agricultural technologies [32,33] but are not yet sufficiently explored in the existing literature—particularly from the farmer’s perspective [18].

These elements form the basis for farmers’ relationships with the various actors in the agri-technical system, such as advisors and contractors, who serve as sources of knowledge [34,35].

If the framework conditions for the flow of information and knowledge to farmers are not in place, external factors such as the lack of access to financial resources and limited availability of technical support for technology management act as barriers to technology adoption [19]. Research on agricultural technology adoption has been conducted predominantly by developed countries, with a focus on advanced technologies, while research in developing countries remains limited [13,19,22] and tends to focus on basic technologies associated with everyday activities and only indirectly related to the agricultural or industrial sectors.

Considering the structural differences between these contexts, e.g., in terms of infrastructure, financial resources, and policy framework, their combination could lead to a significant heterogeneity that makes the analysis incoherent. Given these conditions and the fact that the systemic dimension of adoption is still under-researched, this study was limited to industrialized countries where the institutional, economic, and technological conditions allow for a more comprehensive analysis.

As larger farms are generally more likely to adopt advanced technologies [36], studies on smallholders were also excluded to ensure greater analytical coherence and comparability.

Given the complexity of agricultural technology adoption and the diversity of factors involved, a scoping review is particularly well suited to capturing the breadth of existing research and the areas where knowledge in this field is still developing [37]. In contrast to systematic reviews, which focus on summarizing specific findings, scoping reviews aim to explore key concepts, identify gaps, and summarize a broad range of findings [38].

In this context, a scoping review enables a comprehensive assessment of factors influencing adoption and goes beyond farm-level analysis to capture interactions across the entire agricultural innovation ecosystem.

Furthermore, the need for a scoping review arises from the fragmented nature of current research, in which studies often examine isolated determinants rather than consider how multiple factors interact within a broader framework [13,19]. Using this method, we seek to delineate the landscape of existing knowledge, identify unexplored areas, and gain insights that may be useful for future research and policy development.

This approach is also consistent with the call for an integrative perspective that combines individual adoption behavior with systemic and institutional dynamics [28] to develop resilient innovation frameworks [39,40,41,42].

By synthesizing evidence from multiple sources, this review will contribute to a more holistic understanding of how innovative agricultural technologies can be effectively implemented in the context of sustainable agricultural practices

The aim of this study is therefore to provide a comprehensive analysis of the current state of the science regarding the systemic dimension of the process of technology adoption as the key to the transition to sustainable agricultural development. Given the limited number of papers that take a systemic approach, we also use a network analysis to analyze the thematic relationships in the literature, to date.

To this end, the development and key dimensions of systemic factors are analyzed through a literature review to achieve the following specific objectives: (1) assess the extent and diversity of coverage of systemic factors in existing research, (2) identify the interactions between system characteristics and the process of technology adoption, and (3) identify gaps in the literature related to the systemic dimension of the process of agricultural technology adoption.

In order to achieve the objectives of the study, the following key questions were therefore formulated:

• To what extent and in what variety are systemic factors addressed in the existing research literature in developed countries? And how have they been defined and treated?
• What are the interactions between systemic characteristics and the process of technology adoption in developed countries?
• What are the key systemic determinants of innovative technology adoption in developed countries?
• What are the gaps in the literature on the systems dimension?

To achieve these objectives, a literature review was conducted on the system dimension of agricultural technology adoption in developed countries.

2. Methodology

A literature review was conducted on the systemic factors that influence the adoption of innovative technologies by farmers.

Following the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) extension protocol for scoping reviews by [43], we ensured a rigorous and reproducible process for the selection of suitable articles from the Scopus, Web of Science, Google Scholar, and Semantic scholar databases. This protocol allows the inclusion of academic publications such as scientific articles, book chapters, conference proceedings and reviews, as well as grey literature reports and documents. Given the different methodological approaches in the existing literature, it is important to ensure a comprehensive understanding of the topic [38].

To specifically capture the role of the system dimension in the process of technology adoption, we conducted a two-step analysis procedure, similar to other authors [44]. In a first step, a quantitative analysis of all selected documents (n = 151) was conducted using bibliometric network analysis and thematic analysis. The bibliometric network analysis allowed us to examine the frequency and co-occurrence of keywords in the selected articles in order to identify the main concepts and thematic clusters that represent the main drivers of technology adoption in agriculture. The thematic analysis made it possible to obtain descriptive information about the research design and methods used in the literature studied and to understand the context in depth.

We then selected the most important articles and analyzed them in detail using content analysis. In line with the aim of the study to explore the systemic dimension of agricultural technology adoption, the most relevant articles were selected based on the inclusion of stakeholders as study participants because this criterion reflects the systemic nature of the adoption process, which is characterized by the interaction of multiple actors. Therefore, only studies in which stakeholders were involved as participants were considered for the subsequent qualitative analysis, resulting in 28 articles (Supplementary Materials Tables S1 and S2) that were subjected to in-depth analysis. A comprehensive data extraction form (Supplementary Materials Questionaire S1) was developed to guide the content analysis of the full articles.

This approach allowed us to identify both the key determinants of technology adoption and the underlying relationships between individual decisions and overall system dynamics in agricultural systems. Finally, it also provided a comprehensive overview of how the systemic dimension has been formulated and developed in the literature over time.

2.1. Data Sources and Research Strategy

In order to systematically search the literature, a two-stage document search was carried out.

In the first search phase, keywords were selected on the basis of the previously reviewed literature, as described in the introduction to the article. The first bibliometric search was conducted from a systems perspective and focused on identifying keywords relevant to the research area, such as “system interactions”, or “holistic approach”, or “stakeholders’ involvement”, to effectively match the scope of the research.

The second search phase was based on the recommended broad PCC (participant/concept/context) framework (Figure 1) to properly narrow down the main concepts of the review questions, i.e., to map the system scope of factors influencing farmers’ adoption of innovative technologies [45]. This further review enabled the exclusion of documents that did not address the guiding questions.

The search strategy aimed to capture the systemic dimension of the technology introduction process. In order to properly outline the systemic dimension in the research context, we first examined the most relevant literature on the determinants of agricultural technology adoption in order to build a comprehensive body of knowledge.

The literature review was conducted in a systematic process to ensure a thorough and comprehensive approach. A set of terms corresponding to the systemic factors was established to be searched in scientific databases from February to March 2025.

The final validated search query used for the literature review was as follows (

Table 1. Key search terms.

Technological Context	Precision Agriculture OR Precision Farm* OR Digital Agriculture OR Smart Farm* OR Agriculture 4.0 OR Smart Agriculture
AND
System domain	training OR education OR infrastructure OR competitive pressure OR market OR financial support OR labor OR technology usability OR attractiveness OR appeal OR farming system OR cropping system OR stakeholders engagement OR stakeholders involvement OR events OR cooperation OR information sources OR institutions OR workshops OR regulations OR policy OR governance OR subsidies OR data privacy OR data security OR subjective norms OR technical support OR consultants OR advisors OR advisory service OR service sources OR extension services
AND
Process	tech* adoption* OR tech* diffusion* OR tech* uptake OR tech* implementation

N.B. The asterisk (*) is a character commonly used in databases to represent any group of characters following the root of a word.

):

2.2. Study Selection

According to the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) extension protocol for scoping review [43], shown in Figure 2, the study selection included four main phases.

Phase 1: Identification. The initial search in selected databases yielded a total of 5824 data records, which were then imported into the knowledge management software Zotero 7.0.11. The main sources on the introduction of agricultural technologies that had previously been used to develop the background and methodology of the article were also imported (13 documents). After the removal of 809 duplicates through automatic and manual deduplication techniques and one document withdrawn from publication, 5014 unique records remained for screening.

Phase 2: Screening. As part of the screening process, in-depth filtering was conducted to include the following eligibility criteria (

Table 2. Eligibility criteria.

Eligibility Criteria	Peer reviewed Literature	Grey Literature
Language	English	English
Documents Types	All	Journal articles, conference papers
Mandatory information	Authors, title, abstract	Authors, title, abstract
Study context	Developed countries Not smallholders	Developed countries Not smallholders

), which refined the scope and resulted in 501 documents that were analyzed.

Using Microsoft Excel, we first identified and removed the records whose language was not English (17 records) and which represented an invalid document type (1446 records).

Documents dealing with topics related to developing countries were then identified by searching for keywords in the title and abstract, using terms such as “developing country”, “developing countries”, “low income” and “emerging country”. These articles were marked and excluded, resulting in the removal of a further 835 records. In addition, a keyword-based search was conducted to find documents focusing on small-scale farming systems. Articles that contained the term “smallholder farmers*” in the title or abstract were reviewed and removed where appropriate, resulting in 304 additional records being excluded from the dataset.

Finally, documents where no authors were identified (24 records) and which did not contain an abstract (1887 records) were also excluded.

Phase 3: Eligibility. The titles and abstracts of the 501 documents were independently reviewed based on the definition of the key terms and the following predefined eligibility criteria:

• Study on the process of technology adoption
• Study on the agricultural sector
• Study that considers, among other aspects, the systemic dimension of the adoption process
• Study focusing on developed countries
• Study without reference

Similar to the approach of [22], references were excluded if they did not fulfil at least one inclusion criterion. After this phase, 350 studies were excluded due to irrelevance

Phase 4: Inclusion. 151 studies met the strict criteria for inclusion in the final quantitative synthesis (network analysis) and provided robust evidence for the research question under investigation. Finally, only the most relevant of the included studies (28 articles) were selected for in-depth content analysis.

2.3. Data Extraction

The bibliometric network analysis was conducted to identify the network of relationships between the systemic factors influencing the adoption of innovative technologies in agriculture.

This approach enabled a quantitative assessment of research trends, matching patterns and conceptual linkages in the literature, ensuring a structured understanding of the systemic dimension of technology adoption.

The publications screened according to the PRISMA-ScR protocol were carefully organized in Excel software to provide an overview of the selected papers, including details such as titles, authors, journals, and year of publication.

Subsequently, VOSviewer [46] was used to perform a single network analysis of title, abstract and keywords of the selected documents. VOSviewer is a powerful tool widely used in bibliometric analysis in various disciplines, including agricultural sciences. It is particularly valuable for building and visualizing bibliometric networks and is especially useful for mapping keywords to each other and for thematic clustering [47,48]. It enables researchers to map the relationships between authors, institutions, and research topics to identify important trends, influential publications, and emerging areas of interest in agricultural and social science research. The software’s ability to create intuitive visualizations helps to explore complex datasets and provides a deeper understanding of the dynamics in these fields [46].

In the analysis, a co-word analysis was conducted to identify major themes and concepts by examining the frequency of the co-occurrence of key words to identify groups of studies with similar conceptual underpinnings.

To ensure consistency, the dataset of selected articles was processed by normalizing keywords and author names. To ensure the accuracy of the results, the following settings were assigned to the software: “map based on bibliographic data”, “co-occurrence” as the type of analysis, “keywords” as the unit of analysis, “full count” as the counting method, and “minimum threshold” for the number of keyword co-occurrences at 6 keywords to highlight the most important nodes in the network. Finally, the “total number of elements” was defined as the number of terms to determine the full scope of information.

The network visualization feature in VOSviewer was then used to generate graphical representations where node size reflects the frequency of occurrence and edge strength reflects the strength of associations between terms. In addition, the density visualization function enabled the identification of research hotspots by displaying areas of intense keyword activity.

By applying this bibliometric approach, the underlying structure of research on systemic factors in agricultural technology adoption was revealed, enabling the identification of knowledge gaps and emerging trends. The application of these analytical techniques ensured a systematic, reproducible and objective synthesis of the literature and strengthened the methodological rigor of the study.

To extend the bibliometric network analysis, thematic clusters by author were also formed and analyzed to define the underlying co-authorship network.

On the other hand, the thematic analysis integrated the results of the bibliometric network analysis and thus contributed to a complete overview of the systemic dimension in the literature. The type of research design and methodology used in the study were identified for each selected article, capturing the following descriptive information: (1) technology/tool of reference; (2) theory/conceptual framework deployed; (3) methodology of the study; (4) type of participant involved in the research; (5) type of farming system considered; and (6) geographic area of the study.

Finally, the most relevant articles for the review were fully reviewed and subjected to an in-depth analysis based on the data extraction form (Supplementary Materials Questionaire S1) to examine how the systemic dimension in agricultural technology adoption has been defined and developed in the literature.

3. Results

3.1. Bibliometric Analysis

3.1.1. Overview of Studies

The development of publications shows that research interest has steadily increased over time. Between 2002 and 2014, the number of publications remained low and relatively stable, while a gradual increase began in 2017, followed by a sharp increase from 2019, with most (67%) published between 2022 and 2024 (Figure 3)

The data for 2025, which shows only two publications, reflects the first two months of the year and is not indicative of the general trend. Overall, the topic has gained increasing scientific attention, especially in recent years (from 2019 onwards), and the number of publications has risen noticeably.

In total, 33 countries represent the nationality of the authors. For clarity, only those countries that produced more than one study are shown in Figure 4. The geographical distribution of academic documents shows a strong concentration in a few countries, with Europe leading with 64 publications, especially Italy (19) and Germany (11), followed by the United States with 24 publications. This distribution reflects the strong scientific research infrastructure of developed countries, with other countries such as India (10), Australia (9), and China (6) reinforcing this pattern.

Most research on the adoption of agricultural technology was published in journal articles (79%), conference proceedings (15%), and books (6%) (Figure 5).

The distribution of publications among the journals that published at least two articles on this topic is shown in Figure 6. The journals with the highest number of articles are Precision Agriculture (23%) and Smart Agricultural Technology (11%). Other journals with significant contributions are Agricultural Systems, Sustainability (Switzerland), Agricultural, and Food Economics and Technology in Society. All other journals made a negligible contribution.

3.1.2. Network Analysis

The aim of network analysis was to explore and visualize the underlying structure of relationships in literature. By applying network analysis techniques, the aim was to identify patterns, clusters, and key units that are central to the area under investigation.

As shown in Figure 7, the coincidence network map consists of nodes (terms), which represent important concepts discussed in the selected articles, and edges (links), which indicate the strength of the relationship between the terms based on the frequency of their co-occurrence.

The size of the nodes reflects the frequency of a term: larger nodes correspond to more frequently mentioned concepts, and the color of the nodes groups terms into thematic clusters that indicate closely related research topics. Finally, the distance between two nodes indicates their correlation. The co-occurrence map was created by including terms that appeared more than five times in the title, abstract, and keywords of the selected articles.

The analysis reveals a complex and structured research landscape on technology adoption in agriculture. The key concepts that emerge most frequently include “precision agriculture” (with 55 occurrences), “technology adoption” (with 46 occurrences), “agriculture” (with 31 occurrences), “adoption” (with 29 occurrences), “agricultural technology” (with 24 occurrences), “innovation” (with 20 occurrences), and “smart farming” (with 18 occurrences), highlighting the growing interest and relevance of technology innovation within agricultural systems.

Furthermore, the keywords with the most and stronger interconnected links were “precision agriculture” (total link strength 208), “technology adoption” (total link strength 185), “adoption” (total link strength 122), “agricultural technology” (total link strength 121), “agriculture” (total link strength 118), “innovation” (total link strength 100), “farming system” (total link strength 89), and “management” (total link strength 88).

Notably, the term “farmers”, despite scoring less frequency and connection, appears in a central position, emphasizing the crucial role that farmers play in adopting new technologies.

The overview of the keyword co-occurrence mapped clearly highlights the centrality of the technological dimension within the adoption process. However, a general underestimation of the systemic component is outlined. Indeed, terms that refer to the systemic domain, such as “systems”, “stakeholders”, and “agricultural policy” are positioned at the edges of the map, pointing out a gap in the literature.

As can be seen in Figure 7, three clusters were formed, each represented by a specific color.

The first cluster comprised 6 elements, the second cluster was the largest with 13 elements, and the third comprised 10 elements.

The first cluster, highlighted in blue, is concerned with precision agriculture and the processes of technology adoption, in particular agricultural systems and agricultural labor with regard to the integration of new technologies into agricultural production systems. In this context, there is a close link between “precision farming” and “technology adoption”, suggesting that research in this area primarily investigates the factors facilitating the adoption of precision technologies.

The second area, highlighted in red, concerns the system elements of the process of introducing agricultural technology. It looks at the barriers and drivers to technology adoption, as well as the broader implications for sustainability and climate change. Key terms such as “stakeholder”, “agricultural policy”, “decision making”, “drivers”, “barriers”, “sustainability”, and “climate change” indicate a strong interest in the socio-economic and environmental aspects influencing the dissemination of innovation in the agricultural sector. The presence of “agricultural policy” and “decision making” specifically suggests that public policies and business strategies play a key role in shaping technology adoption decisions. Moreover, the outer position of the words “stakeholders” and “system” highlights a lack of valuation of the role of systemic factors within the literature.

In addition, the weak connection between “sustainability”, “climate change” and “adoption” highlights the need to align technological advancements with ecological and resilience-oriented goals.

The third, green-colored cluster deals with new technologies and digitalization and contains terms such as “smart agriculture”, “big data”, “Internet of Things (IoT)”, “technology”, and “user acceptance.” This set of concepts fully identifies the technological dimension of the adoption process, reflecting the increasing research focus on connected and advanced digital tools enhancing farming efficiency. The strong presence of “big data” and “IoT” suggests that digital technologies are becoming fundamental in transforming traditional agricultural practices, making them more precise and data driven. Notably, the presence of “user acceptance” in the third cluster highlights the importance of behavioral factors in the adoption of agricultural technologies. However, “user acceptance” is weakly or not at all connected to system-related elements such as “system”, “stakeholders”, and “decision making” suggesting that research in this area primarily explores interactions of farmers with digital innovations, focusing on usability, perceived benefits, and technological integration rather than governance or policy-driven adoption strategies. The positioning of the term and its distant associations may suggest that more research is needed on farmer engagement in technology development process co-creation initiatives to encourage adoption and wider diffusion.

Figure 8 illustrates the evolution over time of key research topics in the area of agricultural technology adoption, highlighting shifts in focus over recent years. The color gradient, which ranges from purple (older) to yellow (newer), shows how different topics have gained or lost importance over time.

Originally, until 2021, research focused mainly on technology adoption in precision agriculture, with an emphasis on the management dimension of the technological process and little attention paid to system aspects, such as “stakeholder”, “system”, and “agricultural policy”, suggesting that the early studies focused on the institutional and policy framework that shapes agricultural innovation.

Starting from 2022, particular attention to the decision-making process faced by farmers has emerged. Keywords such as “barriers”, “drivers”, “decision making”, “user acceptance”, and “farmers” were central during this period, reflecting a shift from technology to factors influencing the integration of new technologies into farming practices.

As the research progressed through 2022, the focus shifted to the digital transformation in agriculture, with increased attention being paid to the following aspects: “big data”, “alternative agriculture”, and “smart farming”. This phase shows a growing interest in the practical applications of digital tools and automation in agricultural systems.

By 2023, the research landscape evolved further, with a focus on connectivity and the role of digital solutions in modern agriculture. Keywords such as “digital agriculture”, “Internet of Things” (IoT), and “Internet” are among the latest trends that point to a clear shift towards data-driven decision-making processes and the use of advanced digital infrastructures. At the same time, terms such as “sustainable development”, “climate change”, “sustainability”, and “food security” indicate that technological innovations are increasingly being viewed in terms of long-term environmental and economic viability.

The density map in Figure 9 provides a visual representation of the concentration and relevance of the most important research topics for the introduction of agricultural technology. The color intensity indicates the frequency of occurrence of terms, with yellow areas representing the most frequently discussed topics, while green and blue areas correspond to less frequently occurring terms.

The most conspicuous clusters are “technology adoption”, “precision agriculture”, and “adoption”, which appear in the brightest yellow areas. This shows that these topics are at the center of the research landscape and underline the importance of a meaningful understanding of the process of agricultural technology adoption.

In contrast, terms such as “user acceptance”, “big data”, “internet”, and “internet of things” are found in less dense areas. This indicates that although these topics are part of the research landscape, they have not yet achieved the same status as core topics such as adoption and precision agriculture, which is consistent with the previous analysis of temporal development. Their positioning on the periphery is indicative of emerging trends, particularly in relation to digital transformation and connectivity in agriculture.

In addition, environmental and sustainability-related terms such as “climate change”, “sustainability”, and “sustainable development” are present, but in lower density. This suggests that the link between technology adoption and sustainability, while recognized, has not been as thoroughly explored as the more immediate concerns of adoption and precision agriculture.

Overall, the density map shows a strong focus on challenges and opportunities related to technology adoption, while also showing a growing interest in digitalization and sustainability. The lower density of terms related to user adoption and connectivity indicates potential research gaps and highlights the need for further study on how farmers are using and integrating these technologies into their practice.

Co-authorship analysis is an advanced method of bibliometric network analysis that aims to form social groups in order to investigate and analyze extensive collaboration. The analysis was carried out using VoSviewer software 1.6.19, which enables the identification of author partnerships in a research field. In total, 622 authors from 151 selected publications over a 10-year period contributed to the study of the systemic dimension in the process of technology adoption (

Table 3. Distribution of the number of publications among authors.

Documents	Authors Number	Proportions
1	581	93.4%
2	33	5.3%
3	4	0.6%
4	2	0.3%
5	1	0.2%
6	1	0.2%
Total	622	100.0%

).

The VOSviewer representation consists of nodes representing authors and links indicating relationships, e.g., collaboration between authors. The strength of these relationships is quantified by the Total Link Strength (TLS), which reflects the cumulative weight of an author’s connections with other authors in the network. A higher TLS value indicates stronger collaboration within the network. The degree of collaboration between authors varies from 0 to a maximum of 40 links, with most authors at the lower end of the scale (

Table 4. Distribution of the Total Link Strength among authors.

TLS	Authors	Proportions (%)
0	6	1.0%
1	38	6.1%
2	109	17.5%
3	94	15.1%
4	60	9.6%
5	54	8.7%
6	82	13.2%
7	42	6.8%
8	56	9.0%
9	5	0.8%
10	32	5.1%
12	2	0.3%
13	12	1.9%
15	1	0.2%
16	16	2.6%
18	8	1.3%
25	2	0.3%
17	2	0.3%
44	1	0.2%
Total	622	100.0%

).

As part of the mapping process, thresholds could be set for the number of documents to be included in the analysis to ensure the relevance and interpretability of the visualizations.

However, this option was not used due to our distribution of the number of publications, and the threshold of at least one document was retained. Nevertheless, we limited the analysis to the authors with the most connections (more than 10), with the top 76 authors accounting for almost 12% of the total.

Figure 10 illustrates the general collaboration between authors in the systemic area of agricultural technology adoption. The network of collaborations shown in different colors identifies eight interconnected clusters with 409 connections, giving a total connection strength of 479.

Clusters are useful for identifying authors and their academic relationships. In our collaboration framework, the clusters are isolated, with the exception of the purple cluster containing the most influential author in the field, Fountas Spyros, which acts as a link to the red and green clusters. The size of the nodes represents the number of publications of each author and makes it possible to identify the most influential authors. The most important authors include: Fountas, S.; Vecchio, Y.; Eastwood, C.; Masi, M.; Adinolfi, F.; De Rosa; Gemtou, M.; and Gómez-Barbero, M.

The results show that most authors have not yet established fruitful collaborations outside their research groups. However, two prominent authors have made efforts to expand their social networks. For example, Fountas, S. has 23 connections, 14 of which are to external clusters, and Soto, I. has 25 connections, 9 of which are to external clusters.

3.2. Thematic Analysis

Empirical studies accounted for about 60% of the selected documents, together with 6.1% books, 28.8% reviews, and finally 6.8% that were not identified because there was no complete access to the document.

Research methodology. The empirical studies predominantly used quantitative methods (53.2%), followed by qualitative (23.4%), and mixed methods (14.3%). Only a few studies dealt specifically with case studies (7.8%) and co-design approaches (1.3%). Quantitative studies were mostly based on surveys, while qualitative research included several methods, mainly individual interviews, focus groups, and the Delphi method.

Regarding the theories and models that served as the basis for the empirical studies, the most striking feature is the absence of reference theories in almost 80% of the documents analyzed. Among the theories used, the mixed approach was the most frequently used, while single theories AKIS—Agricultural Knowledge and Innovation System, DOI —Diffusion of Innovations, and TPB—Theory of Planned Behavior (2.27%); and TAM—Technology Acceptance Model, UTAUT—Unified Theory of Acceptance and Use of Technology, and ADOPT—Adoption and Diffusion Outcome Prediction Tool (7.41%) were the only ones relevantly mentioned. It is interesting to note that most of the studies without theoretical references originate from quantitative methodology.

With regard to the type of research participants, only 3.9% of the empirical studies did not provide any information or could not be verified as access to the full document was not possible. Farmers were the most represented at 54.55%, followed by other individual participant categories such as cooperatives, researchers, and students, which were only marginally represented. Conversely, studies involving multiple stakeholders achieved a share of 33.77%.

The integration of digital technologies on farms is influenced by many factors, including the agricultural and technological context of reference.

About 60% of the empirical studies analyzed do not refer to a specific agricultural system, with arable farming (62.5%) being the most represented, followed by livestock farming (18.75%), and fruit growing (15.63%). The prominent area of organic and agroecological farming was only examined in 3.13% of cases, while the area of hydroponics and vertical farming was completely absent, with the exception of overview documents.

Moreover, in terms of the type of technology examined, about half of the papers examined referred to a general technology landscape, such as smart farming or digital agriculture. In contrast, the most frequently examined technologies were precision farming and climate-smart agriculture (15.25%), Internet of Think (10.7%), and blockchain, unmanned aerial vehicles (UAV) and decision support systems (8.47%).

3.3. Content Analysis

Content analysis was conducted on the most relevant articles identified through the data extraction form (Supplementary Materials Questionaire S1) to investigate how the systemic dimension in the adoption of agricultural technologies was defined and developed in the literature. Of the 28 articles selected, 4 were not accessible for a full read and were therefore excluded from the analysis. In addition, articles cited by China were also included in the analysis as China is the dominant country leading the global supply chain and is considered the largest manufacturing economy in the world [49].

In this section, the results are presented according to the research questions defined below.

Research question 1. What types of research design and methodologies have been employed?

Half of the contributions used a qualitative research methodology, with interviews being the most common instrument, followed by workshops, the Delphi method and, in only one case, contributions in social media. Participatory methods were almost completely absent, apart from one case where a co-design workshop with stakeholders was conducted. Quantitative methods were used in a quarter of cases, followed by mixed methods (20%), where surveys and interviews were often combined.

In terms of the theories guiding the research, the majority of studies were not based on any theory (65%). Of the studies that do rely on theories, a proportion used a mixed framework (12.5%), with Roger’s Diffusion of Innovation theory being the most frequently used both as a single theory (8.3%) and among the mixed frameworks. The remaining part is covered by the following theories: ontological security, innovation system theory of Malerba, and Adoption Diffusion Outcome Prediction Tool (ADOPT).

In the vast majority (70%), mixed categories of participants were involved in the studies, with farmers, advisors, technology providers, researchers, cooperatives, and policy maker among the most represented. Among the categories of individual participants, on the other hand, experts are the most common (12.5%), followed by farmers, extension professionals, and web users with the same frequency (4.2%)

The most common technology context in the studies analyzed (n = 24) was general digital technologies (66%). The technology categories analyzed were Artificial Intelligence (AI), Laser-based autonomous weed control, Blockchain, Climate-smart agriculture (CSA), and Internet of Think (IoT). In two cases, the authors referred to precision agriculture and innovations that can be attributed to the general category of digital technologies.

In terms of the agricultural system studied, the majority of studies (37.5%) did not refer to a specific system. Among the individual agricultural systems, arable farming was the most studied (25%), followed by livestock and fruit growing (8.3%), and viticulture (4.2%). The mixed systems studied were arable and fruit growing (8.3%), livestock and arable (4.2%), and organic and agroecological farming (4.17%).

The most important geographical area was Europe, where 62.5% of the studies were conducted, with Germany being the most strongly represented with 16.7%. Other important countries were Australia and New Zealand with 12.5% of the studies, the USA (8.3%), and Canada and China (4.3%). The proportion of global studies, on the other hand, was 83.

Research question 2. How have systemic dimensions in agricultural technology adoption been defined?

The purpose of this section is to examine how the systemic dimension of the adoption of agricultural technologies has been defined, comparing it with the definition given in the introduction, as well as providing an overview of the main levels of analysis considered and the main systemic adoption factors.

We have characterized the systemic dimension by the relationships that generate and disseminate knowledge, promote the process of adoption and diffusion of technological innovations, and define its dimension. Its characterizing element is therefore the interactions between different actors (peers, providers, advisors, cooperatives, institutions, and market) and infrastructures that influence the diffusion and integration of technological innovations, including technical, economic, social, and regulatory aspects in a dynamic and networked perspective [19,22,23,24].

Another part of the studies (8.3%) only partially corresponds to our definition of the systemic dimension, as they only analyze the role of the most important actors in the diffusion of agricultural technologies. Study [50] focuses on mid-level actors such as agricultural advisors and agronomists who play a crucial role in addressing farmers’ uncertainty to promote the adoption of smart farming technologies. Study [28] (Supplementary Materials Tables S1 and S2) analyses the role of agricultural extension in disseminating technologies and interacting with producers to promote their knowledge and decision-making processes. The remaining part of the studies (33.3%) is not based on the systemic dimension as it includes individual categories of actors, such as study [51] (Supplementary Materials Tables S1 and S2) on practitioners, studies [52,53,54] (Supplementary Materials Tables S1 and S2) on key users, and study [55] (Supplementary Materials Tables S1 and S2) on advisors.

The determinants of adoption identified in the studies examined were grouped into meaningful categories useful for identifying the reference actors. As shown in

Table 5. Factors distribution among categories (source: our elaboration).

Categories of Factors	Factors Proportion (%)
Access to funding or economic incentives	21%
Governance and coordination structures	17%
Institutional factors	15%
Social interactions	29%
Technical support and advisory services	17%
Total	100%

, the distribution of the 52 factors identified across the categories was fairly balanced, with social interactions (29%) and access to finance (21%) being the most important categories.

The full list of factors can be found in Supplementary Materials Table S3.

The category of social interactions can be traced back to the relationships between the end users of the technology and the actors who promote it, through which knowledge is disseminated and the benefits of the technology are demonstrated. The factors most frequently mentioned in the studies were the dissemination of results, the provision of information, and social support. The category of actors to which social interactions can be related covers the whole spectrum of actors and consequently, the ability of the system to create networks between stakeholders comes into question.

The category of access to finance or economic incentives reflects the economic component of the process of agricultural technology adoption. Its prominence in the studies analyzed underlines its central importance, with public financial instruments, such as the EU agricultural policy, and private institutions (banks) playing a key role. The factors most frequently represented in the analysis were financial support, flexible financing, financial incentives and economic benefits, and feasibility.

The category of technical support and advice refers to the sphere of meso-actors who play a key role in bridging the gap between technology and farmers and actively contribute to the dissemination of technology. The most frequently mentioned factors in this category were technical support, increased technical advice, and extension services, indicating the advisory rather than educational component.

The category of institutional factors represents the interaction between the public and economic dimensions of the process. It is interpreted by the public institutions that support the sector with regulatory and incentive measures. The factors that attracted the most interest were policy measures, funding programs, environmental incentives, and government regulations and compliance, with the environmental dimension added as facilitating technology diffusion.

Governance and coordination structures refer to the administrative procedures of the technological system, which sometimes has value but struggles to be disseminated, verified and integrated into existing agricultural systems. The factors mentioned in the literature, such as the importance of cooperatives, coordination between actors, and the clear management of projects, suggest that the interpreters of this category are organized subjects that act as coordinators and mediators of the system, managing its normative, economic, and attitudinal dimensions. Governance and coordination structures thus play a central role in the integration of technologies into local agricultural systems.

Finally, the analysis of adoption determinants revealed that the interactions between the different factors were only investigated in 12.5% of the studies examined. Nevertheless, these studies illustrate the complexity of the relationships between the individual issues and show that the treatment of one factor can have cascading effects on others.

Research question 3. How has systemic dimension been addressed/developed? What approach?

The systemic dimension in the analyzed literature was developed through different approaches, each emphasizing different perspectives of the innovation adoption process, including methodological orientations, perceptions of stakeholders, and interactions between adoption determinants.

First, several studies [51,55,56,57,58,59] (Supplementary Materials Tables S1 and S2) examined the systemic view by analyzing stakeholder perceptions with the aim of identifying differences between stakeholder groups. This approach typically aimed to highlight key actors that could act as facilitators or barriers to enable targeted interventions or improve collaboration between actors within the innovation system.

Another group of studies, still relying on the assessment of actors’ perceptions, shifted the analytical focus to the relationships between the adoption factors rather than just the actors themselves. The studies that took this approach emphasized how takeover factors are interrelated and stressed the importance of looking at combinations of factors together. For example, Study [28] (Supplementary Materials Tables S1 and S2) demonstrated that the identified factors influencing the adoption of climate-smart agriculture in Europe are interlinked, highlighting the need for future research to explore these interactions in order to effectively promote behavioral change towards sustainable food systems along value chains.

A third perspective was to examine the perceptions of stakeholders in order to explicitly define the critical factors for adoption. In contrast to the relational approach, this perspective focused primarily on categorizing and characterizing the system by individual determinant factors in order to provide clarity on specific elements that influence adoption behavior in complex innovation environments.

Finally, the fourth perspective identified pursued a methodological orientation that focused on a participatory approach. This included the active involvement of farmers and stakeholders through co-creation and co-innovation processes during the design phase of specific tools or solutions. In these cases, the innovation process was driven by the needs and experiences of stakeholders, which enabled the development of tailored and contextualized solutions.

However, several studies did not take a full systemic perspective, as the analytical focus was limited to specific groups of actors. Nevertheless, these contributions are still highly relevant due to the strategic role that these groups of actors play as intermediaries between technological innovations and end-users.

Study [60] (Supplementary Materials Tables S1 and S2) focused exclusively on innovation intermediaries (Italian operating groups) and examined their crucial role as drivers of innovation adoption. Although systemic relationships were not comprehensively addressed, these studies applied an interactive innovation model that emphasizes co-decision and co-design processes involving stakeholders, thus making an important contribution to the understanding of governance mechanisms that facilitate the diffusion of technologies among farmers and foresters.

Another part of studies [61,62] (Supplementary Materials Tables S1 and S2) looked specifically at the role of meso-level actors, such as agricultural extension agents and agronomists, who act as key intermediaries in technology diffusion. Although these analyzes did not take a fully systemic perspective, they provided important insights by examining how meso-actors’ perceptions of technology uncertainty influence their effectiveness in promoting innovation adoption, and by exploring the evolving skills and roles of advisors in data-driven smart farming contexts.

Finally, another study [63] (Supplementary Materials Tables S1 and S2) analyzed the experiences of large-scale pilots (LSPs), focusing exclusively on adoption processes in specific case studies. By examining end-user challenges and barriers from a process-oriented perspective, these studies provided practical insights into real-world implementation contexts, even if they lacked full systemic integration of other actors or adoption factors.

4. Discussion

This paper examines the systemic dimension of factors involved in the adoption of innovative agricultural technologies by showing how the literature defines and addresses this specific area.

The bibliometric analysis shows an exponential increase in publications, with a clear concentration in the recent past. The outlined trend confirms the urgency of understanding the factors influencing the adoption and diffusion of innovative technologies [12,13], and justifies the growing academic interest in this field of science. However, this rapid growth also promotes heterogeneity and fragmentation. Therefore, regular systematic reviews are needed to reduce the potential impact of this risk, avoid redundant research streams, and enable cumulative knowledge. Geographically, the majority of the corpus comes from industrialized countries, with Europe, particularly Italy and Germany, dominating, confirming the results of previous studies [13,19,22]. This bias also highlights the lack of research in emerging countries, where infrastructure, data management frameworks and support mechanisms differ significantly from those in developed countries. Tailored research programs could expand collaborative projects that co-create technologies adapted to the capital and connectivity constraints of middle-income countries.

The main findings of the bibliometric analysis show that, although the systemic dimension is recognized in the literature, the focus is predominantly on technological and economic aspects, while governance, institutional frameworks, and behavioral dimensions receive relatively little attention [19,31].

The results show that technology adoption and precision agriculture are the most central themes, confirming the solid research base on the technical feasibility of innovation in agriculture [11,12]. However, systemic components such as stakeholders, agricultural policy, and decision-making appear to play only a minor role. This suggests that the studies we reviewed, while recognizing their importance [13], have not comprehensively integrated systemic factors and barriers into the analysis of technology adoption processes.

This finding is at odds with previous literature [26,27,28,29,30], calling for holistic, multi-actor approaches that emphasize the key role of collaborative patterns in promoting technology adoption.

Methodologically, the majority of empirical studies are based on standardized surveys; only 1% used participatory methods, and 80% were conducted without an explicit theoretical framework. These data contradict the focus of recent literature [22] on farmer participation and needs assessment. Future research should therefore adopt mixed methods that incorporate living labs and other participatory approaches to broaden the target audience and capture the interactions between key features of the system. To encourage such methodological changes, competitive funding programs could include a certain budget share for co-design workshops and pre- and post-implementation impact evaluations.

In general, policies need to take into account the broader socio-economic context and apply participatory governance models that involve different stakeholders to ensure an equitable distribution of the benefits of digital technologies across farms of different sizes and regions [6,64].

With regard to the adoption process, several determinants were identified, with social interactions and access to financial incentives being the most important categories. While this finding confirms the emphasis on advisory networks and dedicated funding in the introduction [6,19], only a minority of the literature examines how the categories of factors interact.

In addition, the review identified several key systemic determinants that affect technology adoption, including policy, management, and sustainability. However, the weak relationships between these components and the adoption process indicate a lack of evaluation of their role. This finding is consistent with the literature that emphasizes the critical role of policy and management in promoting the alignment of innovation models with sustainable agricultural development [11,31]. Developing integrated incentive packages (funding, advice, open data platforms) and monitoring their synergies through common performance indicators across agriculture, environment, and innovation would clarify the cascading effects on uptake. Patterns of collaboration between researchers exacerbate these issues. Indeed, the network of co-authorship is highly dispersed, with the majority of authors publishing only once and exhibiting weak overall connectivity, with a handful of nodes bridging clusters. Fragmented collaboration hinders theory building and data sharing. Transdisciplinary consortia would promote replicability, and funding evaluations that reward network additionality rather than individual productivity could further incentivize cross-cluster collaboration.

Finally, our content analysis shows that the systemic dimension was developed according to four recurring systemic perspectives: (i) stakeholder perception studies, (ii) mapping of factor interrelationships, (iii) simple cataloguing of determinants, and (iv) participatory co-creation represented by only one article. Although the need for systemic thinking is recognized in the literature [26,27,28,29,30], truly integrative frameworks remain rare. Future research should foster permanent centers for co-innovation where technology providers, agronomists, policy makers, and farmers iterate through cycles of large-scale design and testing; develop multi-level models linking on-farm decisions and extension networks; and validate system-level indicators across countries. Public programs could fund living labs within European digital innovation hubs and introduce specific calls for integrated system coordination that include governance, data trust, and coordinated advisory services.

To conclude our discussion, we emphasize that our findings are consistent with the review conducted by [65]. The authors highlight the role of key actors: farmers, consumers (not considered in our study), and public institutions. Farmers are the main adopters and implementers of new technologies, with their decisions influenced by economic considerations, perceived risks, and the support from education and social networks. At the same time, consumer acceptance is crucial as market demand ultimately determines the success and longevity of technological innovations in agriculture; emotional responses and risk perceptions are significant barriers for consumers. Public institutions play an enabling role by providing incentives, training and advisory services, and by shaping regulatory frameworks that facilitate both adoption and social acceptance. The authors [65] also state that networks and collective engagement further enhance the diffusion of innovations, fostering trust, knowledge exchange, and peer learning among both farmers and consumers. Effective strategies for advancing sustainable agriculture must integrate economic, institutional, and social dimensions, ensuring that technological solutions are not only technically and economically viable, but also widely accepted and supported across the entire agri-food system.

To enrich the contribution of our review, it is essential to recognize its limitations, as it may reproduce some of these biases. Despite its scope, no review can completely escape the influence of prevailing academic paradigms or publication filters. Therefore, transparency in the selection criteria and efforts to include underrepresented perspectives can enhance the credibility of the findings. In summary, dealing with bias in thematic reporting is not just an academic problem, but a fundamental problem that affects the effectiveness, equity, and legitimacy of policy interventions. A more inclusive and critical research environment is therefore not only desirable but necessary to ensure that policies are based on a diverse and representative evidence base.

5. Conclusions

This study highlights the crucial role of systemic factors in the adoption of innovative agricultural technologies in industrialized countries.

The literature on the adoption of innovative technologies in agriculture has experienced exponential growth since 2019, with a clear concentration in recent years. However, this increase has not been matched by stronger collaborative networks. Most authors publish in isolation, which hinders the development of cumulative theories and the sharing of data. This fragmentation, combined with the rapid pace of publication, highlights the urgent need for regular systematic reviews to consolidate methods and support coherent knowledge accumulation.

In this regard, researchers should increasingly focus their work on interdisciplinary consortia that integrate agricultural, economic, and social expertise. Policy makers, in turn, could encourage such synergies through funding calls that explicitly reward the additionality of networks. Extension services could also play a more active role in research processes by acting as co-authors and facilitators of exchanges between academic institutions and practitioners.

Although this study aimed to comprehensively review the literature and subsequent grey literature, it must be acknowledged that it may have overlooked more informal or locally published forms of collaboration. This represents an important limitation to the generalizability of the findings, including to cultural and geographical contexts that produce results from studies published in a language other than English.

The literature continues to prioritize the technological dimension, and indeed technology adoption and precision agriculture are the most frequently occurring semantic poles, while systemic factors such as stakeholders, policy, and decision-making appear at the margins. This reflects the persistent difficulty of integrating governance, institutions, and behavior into adoption models. Researchers should therefore enrich their designs with multi-level analyses that link farm-level decisions with institutional dynamics and public policy. Policy makers should include systemic impact indicators in the criteria for research funding, while extension services could act as intermediaries between regulatory frameworks and practical application by translating policy guidelines into actionable knowledge for farmers. Furthermore, the situation is complicated by methodological shortcomings. Most studies lack a theoretical framework and participatory methods are still significantly underrepresented, so they are not in line with recent scientific recommendations for co-design and assessment of farmers’ needs. To overcome these limitations, researchers should use mixed methods that combine qualitative interviews, participant observation, and field experiments. Policy makers could support this transition by mandating budget allocations within research calls for co-design processes and impact evaluations. Extension services should be systematically involved as intermediaries between research and practice, providing sound evidence from the field. However, it must be acknowledged that the present content analysis is based on a limited number of full-text articles available in English, which may have led to an underestimation of the qualitative or participatory methods documented in local publications.

Among the 52 factors of adoption identified, social interactions and access to financial incentives stand out, with only a few studies exploring the interrelationships between these categories. This points to the need for dynamic studies capable of modelling such relationships. In this sense, future research could use simulation modelling and network analysis based on data collected in living labs to observe how different combinations of incentives and support mechanisms affect technology adoption. Policy makers, for their part, should design integrated policy packages that coordinate financial instruments, technical services, and digital governance, and assess their impact using cross-sectoral performance indicators. Extension services could facilitate access to these tools and help farmers navigate complex innovation ecosystems.

From a conceptual perspective, four recurring systemic approaches to developing the systemic dimension can be found in the literature, with the stakeholder perception approach being the most widespread and the participatory approach being largely underrepresented. Although the academic discourse increasingly recognizes the importance of systemic thinking, it still struggles to translate this recognition into robust methodologies and common theoretical frameworks. Therefore, it is critical that researchers commit to developing integrative theoretical frameworks that bridge different levels of analysis and that policy makers invest in co-innovation infrastructures, such as regional hubs and permanent labs, where system-level impacts can be co-designed and monitored over time. Finally, extension services should be trained and involved as structural facilitators in participatory processes to ensure that digital technologies are not only more accessible, but also better aligned with the actual conditions and needs of end users.

Addressing these gaps could contribute to more effective policies and a deeper understanding of the influence of system dynamics on the diffusion of technologies. By summarizing existing knowledge and identifying critical research gaps, this review highlights the importance of a more integrated approach to understanding the adoption of agricultural technologies. Strengthening the link between technological innovation, governance, and sustainability will be key to ensuring that the digital and environmental transitions in agriculture contribute effectively to long-term sustainable development goals.