Investigating Published Research towards a Fossil-Energy-Free Agriculture Transformation

Abstract

The defossilisation of the agricultural sector is driven by intense worldwide academic research on non-fossil, renewable and energy-efficient agriculture, and the acknowledgment of the need for sustainable farming practices. For this purpose, not only technical transformations but also socio-technical system changes towards sustainability need to take place in a co-evolutionary manner. This paper investigates structural and qualitative characteristics of the knowledge produced by research on fossil-energy-free agriculture. We provide evidence on the worldwide research directions, as well as investigate whether academic research and publicly funded research projects foster knowledge creation for the desired transformation. Bibliographic maps are constructed using a query-based methodology as social networks to investigate the efficiency of the EU-funded research to achieve the goals set for the 2050 EU Green Deal. The H2020-funded papers are further analysed with dictionary-based text analysis to quantify the relative emphasis of different types of knowledge in the text. This approach is eventually used to relate transformational capacity to project profiles in the European Union, to evaluate past funding schemes and to improve the shape of future research programs on renewable and sustainable agriculture.

1. Introduction

The world is currently grappling with an unprecedented ecological crisis, influenced greatly by our dependence on fossil fuels. The relentless burning of coal, oil, and natural gas has unleashed a cascade of environmental challenges, from climate change and rising sea levels to biodiversity loss and air pollution. Burning fossil fuels adds billions of tons of CO₂ to the atmosphere each year, making it the main cause of anthropogenic climate change. To combat this crisis, a fundamental transformation is imperative—defossilisation. This process entails shifting away from our reliance on fossil fuels and unsustainable resource exploitation. It demands a transition to cleaner, renewable energy sources, efficient energy use, responsible land use, conservation efforts, and a commitment to sustainable living. Simultaneously, the need arises to reimagine whole economic sectors, such as transportation and agriculture, to reduce carbon emissions.

Global agriculture mainly relies on fossil resources for covering most of its energy needs and supporting agricultural productivity, contributing a significant share to global warming through greenhouse gas (GHG) emissions. Simultaneously, it is vulnerable to the adverse effects of climate change, which can disrupt food production systems, threaten food security, and harm ecosystems.

A series of mitigation and adaptation solutions for the defossilisation of the agricultural sector are being developed, including advanced digital technologies and regenerative practices. Academic research on non-fossil and energy-efficient agriculture has been gaining momentum worldwide, driven by the growing recognition of the need for sustainable farming practices. The European Union (EU) has been particularly proactive in funding and promoting research in this field through various programs, including Horizon 2020 (H2020 in the rest of the paper).

However, in the agricultural sector, there is still a significant gap between such developments and the actual adoption and use of the available tools and practices by the EU farmers. The defossilisation of agriculture is not merely a technological challenge [1]. It is a complex process that entails a transition towards a more sustainable socio-technical system, for which social and technical transformations need to take place in a co-evolutionary manner. For defossilisation endeavours to contribute to overcoming currently unsustainable practices, relevant research and innovations must involve more than alternative raw materials and new technologies. Radical systemic innovations are needed including new institutions, infrastructure, user practices, and mental models [2].

Scientists in the field of sustainability transformations identify the systems’ knowledge bases as strong leverage points for transition processes [3]. The knowledge needed for sustainability transitions is complex and transdisciplinary. It has been argued, though, that research and public policy primarily produce knowledge of a techno-economic nature, while lacking knowledge about systems, values, and genuine participation and learning of societies in transition processes. To induce transformative change in the face of wicked problems, a sustainability-dedicated knowledge base is necessary. To this end, three additional types of knowledge need to complement the prevailing techno-economic one: systems knowledge, normative knowledge, and transformative knowledge [4,5,6].

This paper intends to explore how and to what extent international- and European-level research outcomes attempt to transform global agriculture towards a defossilised era. For evaluating the contribution of research to the fossil-energy-free agriculture transition, the quality of the produced knowledge needs to be assessed as well. Based on this argument, the authors pose the following research questions: (RQ1) “How is research on fossil-energy-free agriculture progressing worldwide and in the EU, specifically?” and (RQ2) “Have academic research and publicly funded research projects fostered knowledge creation for the desired transformation?”

Guided by the above research questions, this work aims at investigating structural and qualitative characteristics of the knowledge produced by research on fossil-energy-free agriculture. To this end, we investigate Scopus indexed journal papers in the period from 2014 to 2020 using a query-based methodology, and bibliographic maps presenting the collected data are created using social network analysis (SNA) for illustration and interpretation purposes. This corpus of papers is then compared with a subset of papers which are funded by European H2020 projects. By focusing on the published outcome of projects funded under the H2020 framework, we investigate the efficiency of the EU-funded research to achieve the goals set for the 2050 EU Green Deal. The H2020-funded papers are further analysed with dictionary-based text analysis to quantify the relative emphasis of different types of knowledge in the text. Furthermore, we plotted the subset of the publications that have transformative character in the landscape of international research mapping in clusters derived from SNA to detect correlations of specific themes in the literature with more transformative potential.

The rest of the paper is organized as follows. Section 1 defines the subject matter, namely defossilisation technologies and strategies in agriculture, and innovative knowledge prerequisites for sustainability. Section 2 describes materials used, and the methodological steps for searching the literature and analysing the international and H2020-funded research corpus with SNA and dictionary-based text analysis, respectively. Next, the analysis results are presented and discussed in Section 3, while Section 4 presents the conclusions and further research directions.

1.1. Theoretical Context

1.1.1. Definition of Fossil-Energy-Free Agriculture and Fossil-Energy-Free Technologies and Strategies (FEFTSs)

Fossil-energy-free agriculture responds to a system where all energy needs for production of primary agricultural goods are derived from renewable sources. However, this system goes beyond just energy origin and attempts to rebuild the energy profile of intensive farms in Europe to reduce the energy consumption for the existing practices or implement new strategies that minimize the energy requirements to produce the same or even higher amounts of goods. The main agricultural subsectors, where energy is extensively used to maintain acceptable production levels, are arable farming, orchards, vineyards, open-field vegetables, greenhouses, and livestock facilities.

Today, such farms could be a reality, as industry and research entities have been developing novel technologies and strategies related to more sustainable energy production, its efficient use, and GHG emission reduction. Specifically for the EU, in the last 30 years, these technologies and strategies were applied in most economic sectors with very positive results, leaving agriculture out of scope because it was not considered a main energy consumer and GHG producer, and because the installation cost of such technologies was excessively high for small/medium farmers due to their low income and high prices of such innovative systems. The wide application of such technologies and strategies has increased their market share, allowing their optimisation for better efficiency and quality coupled with prices reduced to acceptable levels for most energy consumers. Therefore, agriculture has become one of the clients and technology has started to adapt to its needs.

In this paper, we introduce a new term, Fossil-Energy-Free Technologies and Strategies (FEFTSs) for farms, which refers to the tools required to address clean energy production and use in agriculture that can reduce both direct and indirect energy consumption. Renewable energy sources (RESs), biofuel production and use, agricultural machinery using biofuels, electrified implements, and others cover the direct energy reduction target by introducing locally produced energy for the farm’s needs. Energy efficiency technologies, agricultural constructions management systems, smart farming technologies, conservation agriculture methods, and best energy management practices for rational use of energy, on the other hand, cover the indirect energy reduction goals by increasing the energy use efficiency through a reduction in agricultural inputs (i.e., fertilisers, pesticides, irrigation water, animal feed) and the introduction of circular use of resources. The benefits of FEFTSs are related to cleaner and more efficient energy production and use, resulting in economic, agronomic, and environmental benefits.

1.1.2. Innovation and Knowledge Dedicated to Sustainability

In classical innovation management theory, any innovation is desirable and inevitable as a sign of progress and growth [7]. The unrestrained pursuit of material and economic growth, however, has long been problematised as a driver of climate change and other ecological degradation. Hence, this interpretation of innovation has inherent tensions with our stated desire for a more sustainable world. Today, sustainability scientists argue for innovation systems dedicated to sustainability [6], where, primarily, innovations that contribute to sustainability transformations are desirable [2,3].

The same reasoning applies to the knowledge needed for sustainability transformations. Sustainability scholars argue for sustainability-dedicated knowledge systems [8], which are validated through the new paradigm of transdisciplinarity [9]. In this paradigm, science is in a dialectic process with society to co-produce ‘socially robust knowledge’ [10]. The types of knowledge that are required for sustainability transformations have been debated in the literature [4,6,9]. According to a common typology [3,4,8,11], sustainability science needs to foster the creation of the following types of knowledge:

• Systems knowledge (SK) refers to the understanding of complex biological, economic, and social systems, and their interactions [11]. It involves recognizing the relationships, feedback loops, and interdependencies that exist within and between these systems, and considers the nonlinear interactions and unintended consequences that can arise from various interventions and changes within a system.
• Normative knowledge (NK) deals with the ethical and cultural aspects of sustainability. It explores questions related to values, beliefs, and norms of different social groups, and helps in understanding the ethical and cultural implications of different policies, practices, attitudes, or interventions related to sustainability.
• Transformative knowledge (TK) refers to knowledge that has the power to leverage significant and positive change towards sustainability goals. Achieving sustainability goals often requires shifts in perspectives, values, behaviours, and institutions, and transformative knowledge plays a crucial role in facilitating such shifts. Key strategies associated with the generation of transformative knowledge are participation, empowerment, education, and communication [4].
• It has been argued, however, that researchers and policymakers primarily produce and prioritize knowledge of a techno-economic nature, which, while necessary to some extent, is only one part of the knowledge base essential for sustainability transformations [2,3,6,8,12]. In this context, techno-economic knowledge (TEK) [3] refers to understanding the interconnections between technology and economics and is decisive for innovation production and introduction into the market.

It is the combination of the four aforementioned types of knowledge that is arguably necessary for successful interventions in socio-technical systems [3], such as transition processes towards the defossilisation of agriculture. Whether the funding efforts in research have assisted the creation of sustainability-dedicated knowledge beyond purely techno-economic knowledge requires, however, further investigation [12].

2. Materials and Methods

2.1. Data Collection

2.1.1. The Global Literature Corpus

A literature review was implemented to investigate international- and European-level research outcomes targeted towards the defossilisation of the agricultural sector by introducing FEFTSs. Peer-reviewed papers related to FEFTSs were found following a Scopus database search through queries from 2014 to 2023. Only papers that were published from 2014 (the starting date of H2020 projects) and onwards were taken into consideration as we aimed to determine if there is a correlation between the international research and H2020-derived papers.

The search aimed at finding articles related to FEFTSs (as explained in Section 2.1) that can (or could be) used by agricultural practitioners. This procedure entailed the creation of a pool of available FEFTSs by using specific keywords based on a categorisation consisting of four main categories: Energy User/Consumer, Clean Energy Supply, Energy Efficiency Improvement, and Soil Carbon Sequestration, divided into specific sub-categories. The hierarchical categorisation and Level 1 and 2 subcategories are presented in

Table 1. FEFTSs categories and subcategories.

FEFTSs Category	Level 1 Sub-Category	Level 2 Sub-Category
Energy User/Consumer	Agricultural technology applications	heating and cooling of buildings process heat/cold lighting agricultural field practices vehicles tools energy sales to external consumers
Clean Energy Supply	Renewable Energy Sources	solar, wind, hydro geothermal bioenergy free energy
	Energy types	heating cooling electricity mechanical energy chemical energy
	Energy Technologies	photovoltaics solar thermal windmills hydropower heat pumps geothermal solid biomass conversion biogas/biomethane production liquid biofuels production
	Energy Storages	heat storage electricity storage cold storage intermediate bioenergy carriers
Energy Efficiency Improvement	Energy savings	efficient buildings efficient vehicles efficient tools precision agriculture precision livestock farming conservation agriculture
Carbon sequestration	Carbon sequestration	soil organic cover tillage (conservation agriculture + CTF) nutrient management crop diversification soil and water conservation techniques fire management grassland management

.

With each addition of a new term, the results were better refined. The majority of the queries contained the word “agriculture” and then, based on the four main categories, corresponding keywords were used (i.e., the Renewable Energy Sources category contained the word “energy”). It should be highlighted that the queries searched for the aforementioned keywords in the title, keywords, and abstract of the paper only. The full list of queries used in this review is given in Table S1. Through this query-based search, initially, 6621 scientific publications were collected. This global literature corpus is onwards referred to as the “global corpus”.

2.1.2. The European Literature Corpus

In order to perceive the impact of H2020-funded projects on agricultural defossilisation knowledge through scientific papers, first, we identified such projects in the CORDIS database. The methodology used for the identification of these projects is similar to the search queries methodology used for the scientific papers. By using the categorisation in Table 1, queries for the CORDIS database were created. The full list of queries used for the research projects is given in Table S2. Regarding the selection made amongst the available projects found, it was performed in relation to the total number of projects found from each query. Queries regarding solar energy and biomass yielded the majority of results, whereas aerothermal and hydro energy-related projects were scarce. For the purpose of our study and due to limited resources, we selected around 10% of the number of projects found per query, after reading their titles and short descriptions in the CORDIS database, in order to include the most relevant projects to all FEFTS types.

Additionally, we only kept research projects that were funded under the H2020 scheme and had published at least one peer-reviewed article. In total, 345 papers derived from those projects were collected. This corpus of papers derived from H2020 projects is onwards referred to as the “European corpus”. The next sub-sections present how a combination of methods were employed to analyse the global and European corpora of papers on fossil-energy-free agriculture. Our methodological procedure is summarised in Figure 1.

2.1.3. Social Networks for Bibliographic Analysis

Analysing the international literature corpus of FEFTSs may offer a bird’s eye view on the evolution and dissemination of knowledge in this field. Bibliometric analysis techniques enable researchers to analyse such large volumes of research from the literature, mainly through the application of statistical indicators on bibliographic data. Bibliometric analysis provides valuable insights into the productivity, impact, structure, relationships, and evolution of research activities in various fields [13].

Bibliometric network analysis specifically focuses on analysing the relationships and patterns within bibliographic data (such as citations, co-authorships, co-citations, and keywords) to gain insights into the structure and dynamics of scholarly communication networks. It involves identifying national and international networks, as well as mapping the development of new transdisciplinary fields of science and technology through statistical indicators from the scientific literature on the productivity of individuals, groups, institutions, and countries. Bibliometric network analysis is widely used in academic research, policymaking, and information retrieval to identify influential papers, track research trends, evaluate the impact of research, and make strategic decisions related to research funding and collaboration. By visualizing and quantifying relationships between scholarly publications, researchers can gain valuable insights into the evolution and dissemination of knowledge within specific fields of study [14].

A bibliometric network is defined as the visual representation, by means of nodes and links, of complex meanings with multi-level influences that allows us to transform the quantitative bibliometric information into qualitative conclusions. There are different types of analysis that are used to determine the relatedness of the network’s terms [15], including citation analysis, co-citation analysis, bibliographic coupling, co-occurrence analysis, and co-authorship analysis [16].

Specifically, the present investigation concerns the following:

• A keyword’s co-occurrence network to identify thematic areas and the evolution in time of the international research on FEFTS. Analysing keywords assigned to publications can help identify emerging trends, topics, and the evolution of research areas over time. Nodes represent keywords, and their connections indicate co-occurrence in documents. The relatedness of items is determined based on the number of documents in which they occur together.
• A funding organisation’s co-authorship network to see which organisations co-fund research on FEFTSs. Studying co-authorship patterns helps visualize collaborations among researchers. Nodes represent authors, and edges represent co-authorship relationships. The relatedness of items is determined based on their number of co-authored documents. Analysing this network can reveal important research collaborations and trends.

These networks are constructed with the help of the open-source software VOSviewer 1.6.20 (www.vosviewer.com, accessed on 4 March 2024) [14,15]. The global literature corpus of FEFTSs in csv format is inserted into the software, which creates the maps. VOSviewer 1.6.20 creates distance-based networks, meaning that the distance between two nodes approximately indicates relatedness. In VOSviewer 1.6.20, the weight of each node can be set to represent different variables, e.g., the node’s number of occurrences, citations, or links. The constructed networks can be viewed with different visualisations to support different interpretations. We have used two types of visualisations:

• the network visualisation, whereby nodes are grouped in clusters of different colours based on their relatedness, in order to identify thematic areas of the literature;
• the density visualisation, whereby each point has a colour that indicates the density of items at that point, in order to emphasize network areas with greater productivity (e.g., of authored documents).

2.1.4. Dictionary-Based Text Analysis of Scientific Papers

The corpus of publications from H2020-funded projects is analysed with dictionary-based text analysis to investigate the relative emphasis of different types of knowledge in these publications. Dictionary-based text analysis is a natural language processing technique used to analyse and extract information from text using predefined dictionaries or lexicons [17]. These dictionaries contain lists of words or phrases, often categorized into different themes or sentiments. The goal of dictionary-based text analysis is to identify and classify words or phrases in a text document based on their presence in these dictionaries. The following steps are typically involved: (1) dictionary creation, (2) text pre-processing, (3) matching words, and (4) analysis and interpretation.

To quantify the relative emphasis of each type of knowledge in these publications, we use lexicons of indicator words available in the literature. Abson et al. [4] have identified vocabulary related to different aspects of sustainability based on an ‘expert’, top-down consensual process. They thus compiled lexicons for normative knowledge (NK), systems knowledge (SK), and transformative knowledge (TK). Following this publication, Bogner and Dahlke (2022) complemented the three lexicons with one for techno-economic knowledge (TEK), by analysing techno-economic papers and identifying word co-occurrence with the word “knowledge”. They then used the four lexicons to assess the types of knowledge present in bioeconomy strategy papers.

In this paper, we use the four lexicons to assess the types of produced knowledge in fossil-energy-free agriculture papers. Specifically, the SK, TK, and NK dictionaries have been directly applied in our analysis without modifications. We decided, however, to validate the TEK dictionary because it had been compiled with a different methodology than the other three dictionaries and comprised significantly more keywords. We used the following approach:

(1)

For this purpose, a sample of techno-economic papers was identified within the global corpus of papers on fossil-energy-free agriculture. This way, 57 papers that included the term “techno-economic” in their title were collected.

(2)

These papers and the original TEK dictionary were entered in content analysis software (WordStat 9—https://provalisresearch.com/products/content-analysis-software/ accessed on 4 March 2024). For each keyword in the TEK dictionary, the software counted the number of papers in which it was present.

(3)

We eliminated less significant keywords by testing two different thresholds; keywords that were present in at least the threshold number of papers were kept in the dictionary. Eventually, a threshold of 10 papers was applied. The authors eliminated a few extra keywords which were deemed as too generic, thus incapable of signifying any specific type of knowledge. In the end, 97 out of 174 keywords were kept in the modified TEK dictionary.

The four dictionaries that have been used in our analysis—including the modified TEK dictionary—are presented in Tables S3–S6. Next, we used the open-source content analysis software Yoshikoder v0.6.4 (https://www.yoshikoder.org/ accessed on 4 March 2024) to perform the dictionary-based text analysis. The four dictionaries, as well as the 345 H2020 papers, were entered in the software. To improve the accuracy of the text analysis, the following parts of the texts have been deleted: references, affiliations, and the journal’s generic information.

2.1.5. Analysing Transformative Papers and Projects

Next, in order to check the correlation between the worldwide and the European research, we proceeded to characterize the EU-derived papers based on thematic areas of the global literature. To this end, we reviewed the titles, keywords, and abstracts of the most ‘transformative papers’ (those with the highest percentage of TK) and matched them to the previously presented SNA thematic clusters (or their sub-sections) to understand whether specific themes among the clusters relate better to the production of transformative knowledge. We also performed a concordance check with the Yoshikoder software to exclude any papers that were rated with high transformative knowledge but the keywords identified as ‘transformative’ do not have the intended meaning (for example, “machine learning” is not the intended form of learning to justify transformative knowledge production).

Based on the number of ‘transformative’ papers produced, the most ‘transformative’ projects also emerged. These projects were investigated for correlations that justify more transformative potential. The analysis was performed based on the available data regarding the project coordinator, partners, activity type of the coordinator, amount of the partners’ budgets, funding scheme, and programme. Detailed characteristics of the analysed projects according to the above criteria is presented in Section 3.2.3.

3. Results and Discussion

3.1. Analysing and Mapping International Research on Fossil-Energy-Free Agriculture

3.1.1. Statistical Indicators

The total outcome of the queries used to obtain the global literature corpus included 6621 papers. In Figure 2, the distribution of the number of scientific papers on FEFTSs found for each Level 2 query is presented. It is evident that, regarding energy use in agriculture, biomass (including liquid biofuels and biogas/biomethane) and solar technologies are the dominating renewable energy sources, whereas heat pumps, precision agriculture, and e-storage provide evidence on the emphasis given to energy efficiency/saving technologies. It is interesting to point out that conservation agriculture is highly ranked, as it is a standalone innovative strategy that acts as an auxiliary pillar to assist defossilisation in both direct (reduction in fossil fuels consumed by field machinery) and indirect (reduction in agricultural inputs) energy consumption. This supports the fact that, in the past decade, there has been dedicated research directed to energy-related issues covering most important FEFTS types.

3.1.2. Keywords Co-Occurrence Network

Through the network visualisation in VOSviewer 1.6.20, the keywords of the international literature are visualized in clusters, indicating thematic areas of research (Figure 3). An in-depth description of the observed clusters follows.

1.

Blue Cluster: Biomass Conversion into Energy and circular by-products

This cluster of keywords refers to scientific articles related to the use of biomass of any kind for energy production, mainly heat and secondly power. However, since the clustering has been performed on recent papers of the last decade, the cluster also contains a lot of research activities that attempt to gain valuable by-products. More particularly, biomass remains the core of the cluster, but anaerobic digestion and biogas (acting as almost synonymous) play a significant role as large nodes of the network, also connected to bioenergy as a separate node. This is expected, since

• livestock facilities have identified their waste as a valuable feedstock for biogas production, leading researchers to work a lot with different waste qualities to optimize the biogas production and composition, and recently even with new technologies to upgrade biogas into biomethane (for direct use as fuel or injection into the natural gas grids), and
• agriculture has shifted significantly into using agricultural residues (e.g., straw) or other products (e.g., grass) as new feedstock for biogas production and a lot of research is conducted on the optimum conditions for existing digesters to host such feedstock.

The remaining digestate has also been considered as an important asset for farmers, as it can be further processed to produce bio-fertilisers and soil amendments (e.g., compost) and be returned into the land that originally produced the feedstock for direct (straw and grass) and indirect (feed for livestock) biogas production.

In smaller but significant nodes, the cluster contains other forms of biomass conversion into energy. Biofuels, mainly in their liquid form (e.g., biodiesel, bioethanol, bio-oil) have been playing an important role in the agricultural input of renewable energy, and research continues to identify ways to produce and use biofuels in a more sustainable way (e.g., biorefineries) through combining extra by-products (e.g., natural colorants). Gasification comes as a technology of huge potential to convert biomass of different kinds into gaseous fuels (syngas), and it has been continuously researched for decades, due to the fact that a single recipe with gasification specifications does not exist, leading to different solutions for other biomass types, including the idea of torrefaction to produce a middle product of unified characteristics to be later gasified. By-products, such as biochar, are also a part of the gasification process, increasing the research conducted in this sector.

2.

Green Cluster: Conservation Agriculture and Its Impacts

Conservation agriculture is the key node of this network cluster. It is a well-known term that refers to an agricultural production system which favours minimum mechanical soil disturbance and, therefore, redirects agricultural practices from conventional mechanised farming. It follows three main pillars: (i) application of zero tillage; (ii) continuous soil coverage with either cover crops or residues; and (iii) crop rotation that allows soil to recover from intensive agriculture and remain healthy and fertile. These principles are given in the cluster image with many small nodes that refer to the principles themselves, which are very closely related to each other (e.g., no-tillage, conventional tillage, conservation tillage, zero tillage, etc.) and their impacts (e.g., nitrogen use efficiency, energy use efficiency, soil degradation, soil quality, crop yield, etc.). Research on conservation agriculture is continuous and, especially in Europe where the application of this production system is relatively low, there is much room for applied research to prove the benefits of its application.

An important keyword node, closely connected to conservation agriculture, is soil carbon sequestration, which is surrounded by smaller nodes concerning soil health, crop residues, tillage systems, and nutrients. This comes as a result of the huge potential of soils to act as carbon pools and naturally absorb a portion of the anthropogenic CO₂ to assist in the battle against climate change and adaptation and, at the same time, help soils to improve their characteristics and maintain their productivity for the years to come with less fertilising requirements (indirect energy input). Therefore, the fact that research is directed to this sector seems to be natural.

Finally, it is interesting to point out that ecosystem services (e.g., carbon farming) seem to have a significant role in recent research output. They offer a very good financial incentive for farmers, encouraging them to shift to conservation agriculture to simultaneously reduce direct (machinery fuels) and indirect (fertilisers reduction) energy consumption, and assist their soil to recover its structure and organic matter through the sustainability of underground flora and fauna.

3.

Light Blue Cluster: Climate Change and Food Security

This cluster attempts to connect research for climate change and food security. Climate change is indeed directly connected to food security due to extreme climatic events (e.g., long drought periods or instant heavy rain, hail etc.) that are difficult to be confronted, especially in agricultural production systems that are not technologically advanced, like the ones in developing countries (e.g., sub-Saharan Africa). Therefore, research into climate change’s impact on agriculture in these countries has significantly evolved, and sustainable agriculture and intensification are considered as important nodes of research to cover the gap between current production systems and future intense climatic conditions. Adaptation, resilience, and resistance close the loop of research actions to avoid food crises. It is interesting to stress the very close connection to conservation agriculture (green cluster), which proposes a sustainable production system that reduces human disturbance to the agricultural environment.

4.

Orange Cluster: Life Cycle Assessment for Agricultural Systems

The process of life cycle assessment (LCA) has been applied in all production systems and tries to use data from all processes in the production line to identify the most important environmental impacts (including energy consumption) that accompany a product in its life cycle. Recently, LCA is complemented by life cycle costing (LCC) and social impact analysis, in order to have a holistic analysis of a product’s impact. Agriculture has also been using this tool and research has grown significantly in this sector. This cluster refers to the application of LCA in agriculture, and the main outputs that interest researchers are the measurement of global warming potential and greenhouse gas emissions.

5.

Purple Cluster: Energy Efficiency Through Smart Agricultural Technologies

This cluster connects energy efficiency in agricultural production systems with smart farming. Precision farming, Internet of Things (IoT), and wireless sensors are tools to receive a continuous data flow from a field during a production season in order to identify commonalities and differences on a spatial and temporal basis and assist in decision making regarding agricultural practices. Therefore, smart farming indeed has a huge potential to increase energy efficiency of a farm, as it helps reduce the number of agricultural practices (machinery fuel reduction) and more importantly the agricultural inputs (fertilisers, pesticides, water) applied in the field. Therefore, a lot of research is conducted in this direction, having energy efficiency in the core of the activity.

6.

Red Cluster: Heat Pumps and Solar Technologies in Agriculture

Solar technologies have been applied in agricultural systems to a large extent, which can be clearly seen in this cluster. It is expected that optimisation of these systems is constantly performed, and research plays its role in this regard, to become more efficient and to reduce the energy consumption of farms and the environmental impact of agriculture. However, heat pumps and ground source heat pumps, which are the largest nodes of this cluster, are not considered so closely related to agriculture, and yet this result shows that probably most agricultural constructions (greenhouses, livestock buildings, silos, etc.) are gaining interest from engineering research to apply heat pumps for climate control (heating, ventilation, air conditioning). For both solar and heat pumps, the cluster shows an interest in research using simulation tools (e.g., TRNSYS) and computational fluid dynamics, but also in important issues to help or hinder adoption, which are energy, environmental, and economic evaluation. The solar technologies are in close vicinity to the yellow cluster, as they are a part of the renewable energy technologies.

7.

Yellow Cluster: Renewable Energy for Sustainable Agriculture

This cluster occupies the central position in the network, connecting all the rest, because the scientific articles selected for this analysis focus mainly on clean energy production in agriculture. Therefore, it can be seen that there are numerous small nodes connected to the three main nodes of renewable energy, sustainability, and agriculture. These, in combination, explain the research attempts to change agricultural production to a more sustainable system in terms of energy and water use, and also to reduce the greenhouse gas emission share of agriculture, as this main node of the red cluster is exactly in the centre. Hence, all renewable energy applied in agriculture is always researched in terms of being environmentally friendly.

3.1.3. Funding Organisations Co-Authorship Network

Through the density visualisation in VOSviewer 1.6.20, the funding organisations of the international literature are visualized in a coloured network, whereby the red colour indicates areas with a high concentration of publications (Figure 4).

Following the National Natural Science Foundation of China, the European Union (represented mainly by the nodes “European Commission”, “Horizon 2020 framework program”, and “7th framework program”) concentrates a large number of funded publications on fossil-energy-free agriculture. Since the EU arises as a key actor in funding research on fossil-energy-free agriculture, in the next phase, we focus our analysis on the papers produced by H2020-funded projects to investigate the quality of the produced knowledge in these projects.

3.2. Focus on EU Results

3.2.1. Research Project Statistics

Our search in the CORDIS database according to the aforementioned methodology yielded 156 research projects, out of which 37 were funded under the H2020 framework and had published peer-reviewed articles. The results are presented in detail in

Table 2. Analysis of the selected research projects.

Type of Research Project	Number
Total number of selected research projects acquired from CORDIS database	156
H2020 Projects	95
H2020 Projects that produced papers (peer-reviewed and conference papers)	40
H2020 Projects that produced peer-reviewed papers	37

https://ec.europa.eu/research/participants/docs/h2020-funding-guide/grants/applying-for-funding/find-a-call/what-you-need-to-know_en.htm accessed on 4 March 2024.

.

Different action types specify H2020 objectives, including Research and Innovation Actions (RIA), Innovation Actions (IA), Coordination and Support Actions (CSA), European Research Council (ERC), Marie Skłodowska-Curie Actions, SME Instruments, and Fast Track to Innovation (FTI). Regarding the action type of the 37 analysed projects, most are funded by RIA (14) and by IA (13), followed by seven CSA, one SME-2, one ERC-STG, and one ERA-NET—Cofund. This result is presented in Figure 5. The full list of research projects identified in CORDIS is available in Table S7.

By examining the scientific papers produced by the aforementioned projects, we acquired 345 peer-reviewed articles to be analysed further. The final numbers are presented in detail in

Table 3. Analysis of the scientific papers derived from the chosen research projects.

Type of Paper Derived from Projects	Number
Papers derived with funding number and project name from Scopus	507
Irrelevant papers found	78
Conference papers	81
N/A papers	3
Peer-reviewed papers to be analysed	345

.

In their majority, these papers stem from RIA (220), followed by IA (85), CSA (25), ERA-NET—Cofund (13), ERC-STG (2), and SME-2 (1) projects. The results are presented in Figure 6. In Table S8, the full list of the research papers identified from the 37 research projects is presented. This result seems reasonable, as indeed RIA projects are dedicated to basic and applied research trying to bring novel ideas to prototyping, while IA projects focus more on demonstrating existing technologies, optimising them and bringing them closer to commercialization. On the other hand, CSA projects do not focus directly on research results but attempt to develop the conditions for specific scientific themes to become more known and integrated into real life, by developing mainly policy and data analysis papers. ERA-NET, ERC-STG, and SME-2 funding programmes are not represented with a large number of projects, which leads to a small number of published papers as well.

3.2.2. Lexicographic Focus to Explore the Quality of the Produced Knowledge

Next, the results of the dictionary-based text analysis of the European literature corpus are presented. For each paper, the content analysis software counted how many keywords from each dictionary are present. This resulted in a percentage allocated to each type of knowledge (normative, techno-economic, transformative, or systems knowledge) within each paper, as well as an average percentage among all papers. This analysis, presented in Figure 7, shows that research papers on fossil-energy-free agriculture are dominated by techno-economic knowledge, which averagely accounts for 58.5% of all dictionary words found in the papers and, therefore, outperforms the other categories altogether. Systems knowledge averagely accounts for 20.7%, normative knowledge for 12.5%, and transformative knowledge for 8.3% of the detected dictionary words.

A deeper examination of the content analysis results further reveals the dominance of TEK, even horizontally across the other knowledge types. Each knowledge type comprises four sub-categories, which represent different aspects of the main theme. For example, transformative knowledge consists of four elements: policy and decision making; participation; motivation; communication and education. Figure 8 shows in detail the relative presence of the sub-categories in the analysed papers. We may observe that, for each knowledge type, the sub-categories that are more relevant to the dominant techno-economic paradigm are the most prominent: “efficiency” in NK, “market” and “technology” in TEK, “policy and decision making” in TK, and “function and process” in SK. On the contrary, the least numerous sub-categories (less than 2% of total dictionary words) are “justice and ethics” in NK, “resilience” in SK, “motivation” in TEK, and “communication and education” in TK. Thus, the most transformative dimensions of each knowledge type are the most under-represented.

This outcome suggests that the EU-funded research products in fossil-energy-free agriculture have favoured the production of techno-economic knowledge and have hardly embraced a more transdisciplinary and transformative paradigm. While a sustainability-dedicated knowledge base requires all the four aforementioned knowledge types, our analysis shows that transformative and normative knowledge are rather poorly represented. This finding suggests that the produced knowledge, while improving technological status and economic performance, may not be transformative enough to trigger transitions to alter the regime.

3.2.3. Focus on Transformative Papers and Projects

Among the H2020-derived papers, which have been analysed with dictionary-based text analysis, we further examined those with the highest percentage of transformative knowledge. Specifically, 60 papers with the highest TK percentage were selected, ranging from 16.7% to 61.1% of transformative knowledge content. Through the concordance check performed with Yoshikoder, four of these papers have been excluded as irrelevant. The remaining 56 papers were analysed to identify potential correlations that explain their increased content of TK compared to the rest of the papers. The aforementioned papers are highlighted with yellow in Table S8.

An initial analysis shows that these papers are not concentrated in specific journals; rather, they are published in a variety of journals. The publishing year does not seem to be an important factor, as there is no tendency in more recent papers to contain higher (or lower) percentages of transformative knowledge. Unlike what the recent literature suggests [18], our analysis shows that research on sustainable agriculture has not yet shifted away from the techno-economic focus.

Instead, there seemed to be specific H2020 research projects that have produced a large number of ‘transformative’ papers. To further investigate this correlation, we associated each of these transformative papers with a corresponding H2020 research project. It should be mentioned that, from our analysis, we came across eight research projects that were not included in our original list of projects but were under the H2020 framework, as each of these projects has produced a paper jointly with another project that was already included in our analysis. The projects were removed from our analysis as they did not meet the criteria set in our methodology and the papers produced from them were already counted and analysed. These projects are GENIALG (two papers with the MacroFuels project), LANDSUPPORT (one paper with the SOILCARE project), LANDMARK (one paper with the SOILCARE project), CONSOLE (one paper with the LIFT project), INVADE (one paper with the SET-Nav project), TRANSrisk (one paper with the REEM project), and SolACE (one paper with the SOILCARE project). Furthermore, the NEFERTITI project co-produced one paper together with two other projects on our list (IOF2020 and SmartAgriHubs). For the purposes of our analysis, this paper was removed both from the NEFERTITI and IOF2020 projects, with the latter being chosen as it produced more papers than SmartAgriHubs. As a result, a list has been created with H2020 projects sorted by the high-TK papers they have generated, which is presented in

Table 4. H2020 projects with high TK% papers (TK ≥ 25%).

Project Name 1	Number of Papers	Project’s High-TK/Project’s Total # of Papers
LIFT	8	42.11%
SOILCARE	7	15.91%
IOF2020	5	42.11%
BioEcon	3	23.08%
SET-Nav	3	21.43%
WiseGRID	2	100.00%
MacroFuels	2	22.22%
REEEM	2	7.14%
inteGRIDy	1	6.25%
ISAAC	1	50%
Smart-AKIS	1	25%

¹ All of these projects are represented in this table with their acronyms but detailed information about them is given in Table S9.

.

The subjects of all projects with the largest number of TK papers and their characteristics are presented in Table S9. As this investigation is a fortiori policy-oriented, we meant to relate TK outcomes of projects to the action types that characterize H2020 projects, and the respective funding modes. In

Table 5. Number of research projects per action type for the different TK thresholds.

Action Type	TK ≥ 25%	TK ≥ 20%	TK ≥ 16.7%	% of Total Papers
RIA	5	5	6	2.7
CSA	3	3	4	16
IA	3	4	4	4.7

, it can be seen that the majority of the projects with a high TK are RIA projects, followed by IA and CSA. We observe that CSA projects have a higher percentage of transformative papers among their total produced papers (16%), while this percentage for RIA and IA projects is 3% and 5%, respectively.

This can be explained by the fact the CSA funding primarily covers accompanying measures concerning policy dialogues and mutual learning exercises and studies, as well as complementary activities of strategic planning, networking, and coordination between programmes in different countries. Thus, research and innovation outcomes from various projects are compiled in the frame of CSA funding for dissemination, raising awareness, and communication to ultimately support policy making. On the other hand, RIA and IA projects aiming at producing new knowledge/technology and new or improved products, respectively, have resulted in a higher absolute number of TK papers, along with a much higher number of total produced papers compared with CSA projects.

Therefore, scientific papers with high TK can be correlated with RIA projects as these projects’ main goals are to establish new knowledge and, at the same time, explore the feasibility of new/improved technology. Although the large part of research outcomes of RIA projects pertains to knowledge types such as systems knowledge, pointing out complex system interactions, and techno-economic knowledge, evaluating economic feasibility of technology as a prerequisite for innovation, strong concern for transformative knowledge can also be observed. In fact, RIA projects first undertake a thorough analysis to assess the status of the studied subject, and then new methodologies are produced, so as to improve the situation and introduce new technologies. For example, the LIFT project, evaluated as one of the most transformative projects in terms of its produced papers, investigated the potential benefits of the adoption of ecological farming in the EU through 30 case studies and then promoted the performance and sustainability of such systems across the EU.

Finally, we checked the correlation between the worldwide and the European research by characterizing the most ‘transformative’ EU-derived papers based on the previously identified clusters of the global literature. To this end, we reviewed the titles, keywords, and abstracts of these papers and matched them to the previously presented thematic clusters (or their sub-sections) to understand whether specific themes among the clusters relate better to the production of transformative knowledge. The papers were categorised according to the closest relation to the clusters, as shown in

Table 6. Number of research papers related to the clusters defined from the international literature review.

Cluster Name	Number of Papers Mainly Related to the Cluster
Blue Cluster: Biomass conversion into energy and circular by-products	4
Green Cluster: Conservation agriculture and its impacts	8
Light Blue Cluster: Climate change and food security	17
Orange Cluster: Life cycle assessment for agricultural systems	1
Purple Cluster: Energy efficiency through smart agricultural technologies	6
Red Cluster: Heat pumps and solar technologies in agriculture	0
Yellow Cluster: Renewable energy for sustainable agriculture	18
Non-Related	3

.

It can be observed that the majority of TK papers devoted to the subject of fossil-energy-free agriculture refer to renewable energy systems (Yellow Cluster) and the cluster of biomass conversion into energy (Blue Cluster), which in reality is a sub-category of the former as they both refer to direct energy production systems in farms. These categories of FEFTSs cover the direct energy impact of agriculture, which was expected to occupy the interest of the TK papers under consideration. Climate change (Light Blue Cluster) follows, and conservation agriculture (Green Cluster) comes next, both of which are also closely interrelated, as they refer to strategies and practices of sustainable and ecological agricultural production. Smart agricultural technologies (Purple Cluster) close the analysis, showing that the three latter clusters cover all FEFTSs concerning a reduction in indirect energy consumption in agricultural practices. Life cycle assessment of agricultural systems (Orange Cluster) and heat pumps (Red Cluster) have negligible participation in the selected papers, showing that environmental impact is still not in the core of agricultural transformation and that heat pumps are very innovative and specific to buildings’ FEFTSs, making them still non-relevant for the immediate FEFTS integration into European farms.

The exercise discussed above proves it is difficult to relate with accuracy these 56 most transformative papers of the H2020 projects under analysis to the seven clusters identified in the international literature. This may be a result of concurrently scattered efforts from various thematic areas to produce more transformative research, without, however, more organized endeavours. Papers of high TK are not always related to a specific topic, but they are wider in perspective, trying more to analyse the communicational, educational, motivational, participatory, and policy-making components of knowledge.

4. Conclusions

Research related to the defossilisation of agriculture shapes a global nexus interlinking major topics to prevailing technologies. In this work, a social network analysis of the global corpus revealed climate change, food security, conservation agriculture, and biomass conversion as major topics, as well as technology focusing on smart agriculture, heat pumps, and renewable energy for sustainable agriculture. Life cycle assessment is a crucial methodological tool to environmentally evaluate any technology or strategy that declared an impact on the defossilisation of agriculture, leading to forming a separate cluster.

Europe is one of the global funding pillars of research on defossilisation of agriculture along with China and the USA, with other actors remaining of minor importance. Based on this, this paper proceeded with further analysis of the European research outcome corpus to provide evidence of the effectiveness of such research efforts undertaken by the EU administration and taxpayers.

As a first step, the authors examined the nature of knowledge created out of the research activity and, more specifically, the outcome of EU-funded research since 2014. Published material in peer-reviewed journals were classified based on the knowledge type they provide as systemic, normative, transformative, or techno-economic, while having in mind the overall goal to achieve sustainability. Ideally, all these aspects should be present to build sustainability knowledge; nevertheless, only the presence of transformative knowledge ensures effectiveness and tangible results for the transition to sustainability. Our analysis showed that techno-economic knowledge prevails in research outputs, while transformative and normative knowledge are rather poorly represented. This does not necessarily mean that research is performed to produce more research, as significant research outcomes support instrumental knowledge that integrates into the incumbent system, improving technological status and economic performance, although it is not able to trigger transitions to alter the regime. It leads to incremental changes rather than fundamental transformation of the landscape.

We observe, though, that the transformative research outcome is significant in relative terms in CSA funding schemes and in absolute terms in certain RIA- and IA-funded projects. These latter have some common characteristics (Table S9): (i) they are coordinated by major academic actors in the European Research Area, (ii) they pronounced clear vision statements (LIFT: “how socio-economic and policy drivers impact on the development of ecological approaches to farming”, SET-Nav: “support strategic decision making in Europe’s energy sector”), (iii) they explicitly adopted transdisciplinary approaches and team work (SoilCare), (iv) they are focused on social acceptance (ISAAC: “raising social awareness and identifying barriers to wider biogas and biomethane production by means of a participatory model with the aim to involve multiple actors”), (v) they are conducted through stakeholder analysis (IOF2020: “network of farmers, food industry stake-holders, technology providers and research institutions, … inclusion and interaction of various actors involved in the transformation process…”, SmartAgriHubs: “multi-stakeholder approach and covered a broad value-chain network across all EU member states”), and last but not least, (vi) they are explicitly oriented to strategy development (REEM: “focused not only on technology research, development and innovation, but also impact on society and environment based on specific case studies”, BIO-ECON).

Limitations of the analysis concern the detection of knowledge types, as content analysis was performed by means of software due to the large volume of information to examine. While dictionary-based text analysis provides a quick and interpretable way to analyse textual data, it does have limitations. For instance, it might not consider the context of words in sentences, leading to potential inaccuracies in the analysis. Nonetheless, it serves as a valuable tool for extracting insights from large volumes of text data.

The obtained results do, however, constitute a basis for further discussions on the right approach to achieve the desired transformation through academic research and publicly funded research projects on defossilisation of agriculture. This analysis can be seen as a step forward to a better understanding of the correlation between research funding in Europe for agricultural defossilisation and the creation of transformative results in this sector. It could be beneficial to follow this research with a further analysis of other EU research funding programmes, such as Horizon Europe, to see what trends in generating transformative knowledge prevail, and what are the most prominent actors and regions in the network of European research on fossil-energy-free agriculture, thus identifying the right conditions for successful development of transformative research. Such analyses would contribute to making sure the European knowledge network generated by the EU research funding programmes creates a favourable environment for transformative knowledge production and diffusion to help EU agriculture defossilise at a faster pace.