Front. Sustain. Food Syst. Frontiers in Sustainable Food Systems Front. Sustain. Food Syst. 2571-581X Frontiers Media S.A. 10.3389/fsufs.2025.1656562 Original Research A scientometric retrospective of the livestock long shadow report Duluins Océane 1 * Validation Formal analysis Investigation Conceptualization Writing – review & editing Writing – original draft Project administration Goutsmedt Aurélien 2 3 Visualization Validation Investigation Writing – review & editing Methodology Software Vandevoorde Noé 1 Data curation Conceptualization Writing – review & editing Baret Philippe V. 1 Conceptualization Supervision Writing – review & editing SYTRA, UCLouvain, Earth and Life Institute, Louvain-la-Neuve, Belgium ICHEC Brussels Management School, Brussels, Belgium ISPOLE, Universite catholique de Louvain, Louvain-la-Neuve, Belgium Correspondence: Océane Duluins, oceane.duluins@uclouvain.be 06 02 2026 2025 9 1656562 30 06 2025 24 11 2025 26 11 2025 Copyright © 2026 Duluins, Goutsmedt, Vandevoorde and Baret. 2026 Duluins, Goutsmedt, Vandevoorde and Baret https://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

Concerns over the sustainability of livestock systems have been at the forefront of environmental discourse for decades. A pivotal moment came with the FAO’s 2006 report Livestock’s Long Shadow (LLS), which framed livestock as a major contributor to global environmental challenges, including greenhouse gas emissions, land degradation, and biodiversity loss. Despite criticism, it has become a seminal work, cited over 3,000 times as of 2024 on Scopus. Using a scientometric approach, this paper examines how the scientific community has engaged with the LLS report and explores the extent to which it is linked to the emerging discourse around the protein transition, a shift away from animal-based products toward more sustainable alternatives. We pursue three objectives: (1) to map the research communities citing LLS, (2) to investigate the connections between LLS-related and protein transition literature, and (3) to assess whether academic treatments of production and consumption remain siloed or integrated. Using bibliographic coupling and topic modeling, we identify seven thematic clusters spanning livestock emissions, nutrient pollution, climate mitigation, land use, biodiversity, sustainable consumption, and food innovation. Notably, three of these clusters align with the major narratives of the protein transition. However, our findings point to a continued divide. Livestock-related research largely focuses on environmental and production-side concerns, while protein transition literature is predominantly framed around consumption, ethics, and health.

 livestock long shadow bibliometric and network analysis scientometric protein transition livestock systems alternative proteins climate change The author(s) declare that no financial support was received for the research and/or publication of this article. section-at-acceptance Land, Livelihoods and Food Security Introduction

The role of livestock in future food systems has become a central question in both societal and academic debates on sustainability (Beal et al., 2023). Livestock systems sit at the intersection of multiple environmental, social, and ethical challenges and opportunities. On one side, a vast body of research highlights their negative impacts, including greenhouse gas emissions, land degradation, biodiversity loss, and animal welfare concerns (Poore and Nemecek, 2018; Steinfeld and Gerber, 2010). On the other, numerous studies emphasize the positive contributions of livestock to food systems, such as supporting rural livelihoods, maintaining grassland ecosystems, recycling nutrients, converting non-edible biomass into food, and providing culturally and nutritionally valuable products (Leroy et al., 2022). Increasingly, scholars call for a balanced perspective that recognizes both the benefits and costs of livestock production in the broader pursuit of sustainable food systems (Herzon et al., 2023).

Parallel to these debates, concerns about the consumption of animal-based products have gained prominence (Rockström et al., 2025; Willett et al., 2019). Excessive intake of red (processed or not) and processed meats has been linked to multiple health issues, including obesity, cardiovascular diseases, type 2 diabetes, and other non-communicable diseases (Boada et al., 2016; Qian et al., 2020). These concerns have fueled the rise of a “protein transition,” which advocates for a shift toward more plant-based, alternative, and diversified protein sources (Duluins and Baret, 2024; Jenkins et al., 2024; Lumsden et al., 2024). Together, these debates reflect growing societal and scientific attention to how both production and consumption practices must evolve to address sustainability challenges. Despite the need to address production- and consumption-related issues as interlinked dimensions, this interconnection is often overlooked in scientific research (Duluins and Baret, 2024).

Published in 2006 by Steinfeld et al. (2006) the FAO report Livestock’s Long Shadow (LLS) marked a pivotal moment in global awareness of the environmental consequences of animal agriculture. Appearing at time when climate change was rapidly gaining prominence on the international policy agenda, following milestones such as the enforcement of the Kyoto Protocol in 2005 (European Environment Agency, 2005), the report offered one of the first comprehensive assessments of livestock’s multifaceted environmental impacts. It notably estimated that the livestock sector was responsible for nearly 18% of global greenhouse gas emissions, surpassing those from the entire transport sector (Steinfeld et al., 2006). Beyond climate change, the report also linked livestock to land degradation, deforestation, water and air pollution, overuse of natural resources, and biodiversity loss (Steinfeld et al., 2006). The report’s core message, that farmed animals are major contributors to anthropogenic greenhouse gas emissions, created a shockwave across scientific, political, and public arenas (Glatzle, 2014; Kristiansen et al., 2021). It was widely covered in global media and catalyzed responses from government agencies, environmental groups, and animal welfare organizations, many of which began calling for a reassessment of meat production and consumption practices on both environmental and ethical grounds (Brown, 2020; Vergunst and Savulescu, 2017; WWF International, 2022).

Despite subsequent criticism of its methodology and emissions estimates (Glatzle, 2014; Neslen, 2023; Pitesky et al., 2009), the LLS report nonetheless catalyzed global discussions on plant-based diets and alternative proteins, notably influencing dietary guidelines (de Boer et al., 2014; Westhoek et al., 2014; Willett et al., 2019). Nearly two decades later, it remains one of the most cited FAO reports, with more than 3,000 citations on Scopus as of 2024. This extensive citation record reflects the emergence of multiple research communities, including, among others: environmental scientists who expanded the report’s carbon accounting methods (Poore and Nemecek, 2018); animal scientists who sought to refine or contest its emissions estimates (Holden, 2020); social scientists and political ecologists who examined the discursive and institutional consequences of the report’s framing (Bristow, 2011; Morris and Jacquet, 2024); and nutritionists who linked its insights to broader debates on dietary change (Kingston-Smith et al., 2010). Yet, despite this wide-ranging engagement, there has been no comprehensive effort to examine how these research communities relate to one another or to evolving paradigms in sustainable food systems research.

This study addresses that gap by systematically mapping the intellectual landscape shaped by LLS, looking at how the report has influenced scientific communities and research agendas over time. We adopt a scientometric approach, which applies quantitative methods like network analyses to data on scientific publications (especially citations and abstracts) to trace how ideas, concepts, or research fields evolve. By mapping citation patterns and examining thematic orientations, we provide a systematic and evolutionary overview of how the LLS, through its citations, has shaped different strands of research.

Importantly, the legacy of LLS lies not only in raising awareness of the environmental impacts of livestock but also in serving as a foundational reference for the emergence of multiple research domains. Although these domains remain linked by their common reference to the report, they have progressively diverged, giving rise to distinct disciplinary and thematic “silos.” The protein transition provides a particularly illustrative example of this process. While it draws part of its justification from LLS, it has also evolved in parallel, shaped by broader debates on nutrition, innovation, and consumer behavior. By examining this case and using the citations of Livestock’s Long Shadow as a point of departure, this paper explores how sustainability challenges have been framed in recent research and identifies the narratives that guide both scholarly inquiry and policymaking. In doing so, it positions LLS as a cornerstone in the intellectual evolution of sustainable food systems research, shedding light on how scientific narratives diversify and endure across paradigmatic shifts.

This paper makes two main contributions to the field of sustainable food systems research. First, it provides a bibliometric and topic-based mapping of the research communities citing LLS, identifying how distinct intellectual communities have emerged and evolved. Secondly, by taking the protein transition as case study, it demonstrates how consumption and production-related issues are still treated separately in scientific literature, showing how different research communities engage with these issues and the solutions they propose.

The paper is organized as follows: Section 2 describes the objectives and methods used; Section 3 presents the results of the bibliometric and topic analyses; Section 4 discusses the implications of these findings, with a particular focus on the link between livestock and protein transition discourses; and Section 5 concludes.

 Objectives and methods

This paper investigates how the FAO’s 2006 report Livestock’s Long Shadow has shaped academic discussions on livestock and environmental sustainability. Specifically, it addresses three key objectives:

1) Map the research communities that have cited the report in scientific literature, to identify distinct clusters or intellectual groupings based on shared citation patterns and automated text analysis of articles abstract.

2) Investigate the link between communities related to LLS and protein transition communities.

3) Assess whether issues of production and consumption are treated separately in the academic literature referencing LLS and determine the degree of siloing or openness among the thematic communities involved.

To meet these objectives, we adopt a scientometric approach and apply two complementary analytical methods to a unified database. First, we use bibliographic coupling network analysis to examine how publications citing LLS are connected through shared references. This method enables the identification of research communities, i.e., groups of papers that rely on similar sources, thereby revealing the intellectual structure and dynamics of the field.

Second, we run a topic model on the abstracts of these publications to uncover dominant themes discussed within each research community. Topic modeling allows us to observe how topics are related to each other and group them by their proximity.

The methodology is further described by distinguishing three different phases:

1) Phase 1: Database creation

2) Phase 2: Quantitative analysis: Bibliographic coupling networks and topic modelling

3) Phase 3: Qualitative analysis: Community analysis

Each step of the methodology is synthetized in Figure 1.

Methodological steps separated in three successive phase: (1) Database creation, (2) Quantitative analysis and (3) Qualitative analysis.

Flowchart depicting research methodology divided into three sections: 2.1 Database creation, detailing database extraction into Level-1 and Level-2 documents; 2.2 Quantitative analysis, with steps from temporal networks construction to principal component analysis; 2.3 Qualitative analysis, outlining community profile characterization to thematic grouping. Includes a legend differentiating processes and outputs. Phase 1: Database creation

The database consists of two types of documents: Level-1 documents, which are documents citing LLS, and Level-2 documents, which are the references of Level-1 documents. Consequently, some Level-2 documents are also found to be Level-1.

Level-1 collection and cleaning

We first searched Scopus and Google Scholar1 (on August 5, 2024) using the keywords “livestock’s long shadow,” restricted to reference lists for Scopus. For each result, we extracted the abstract, subject categories, source of publication (especially journals), authors’ affiliations, and reference list using the rscopus package in R (Muschelli, 2019). This initial search returned 4,793 documents for Scopus and 7,130 for Google Scholar (see Appendix A in Supplementary file 1 for a more complete version).

We applied the following filters:

Remove duplicates.

Retain only records with a DOI.

Retain only articles, reviews, and book chapters (exclude books, preprints, editorials, etc.).

Require that documents have a title, year, and journal information.

Include only records with English abstracts, identified using the fastText package in R (Mouselimis, 2024).

After cleaning, 3,638 documents remained for Scopus, and 0 for Google Scholar. Using Citation Chaser (Haddaway et al., 2021), we retrieved missing abstracts, adding 61 more papers for Scopus, and 70 for Google Scholar for a total of 3,769 Level-1 documents (Table 1).

Level 1 and 2 references included from Scopus and Google Scholar by applying a set of including/excluding criteria.

Entries Scopus Google Scholar Total
Level 1 Level 2 Level 1 Level 2
Raw CC(1) Raw Raw CC(1) CC(2)
Initial entries 4,793 144 465,522 7,130 1,472 56,003
Final entries (unique papers) 3,638 61 136,449 0 70 2,185
Final entries (level1 contribution) 96.5% 1.6% 0.0% 1.9% 3,769
Final entries (level2 contribution) 99.1% 1.6% 138,634
Final entries (total contribution) 2.57% 0.04% 96.47% 0.00% 0.05% 1.56% 142,403
Level-2 references

We then extracted Level-2 from all Level-1 documents (Level-2). The same quality filters were applied, except for the last one (presence of an abstract). Scopus provided 136,449 Level-2 references, and Google Scholar and Citation Chaser further added 2,185 references, for a total of 138,634 Level-2 documents.

 Final dataset

The dataset comprises 142,403 records collected from Scopus and Google Scholar, consisting of 3,769 Level-1 and 138,634 Level-2 documents used to build the bibliographic coupling network (Table 1).

 Phase 2: Building the bibliographic coupling networks and running topic modeling

Regarding the bibliographic coupling method, we construct a series of temporal networks, where nodes represent citing documents (or Level-1 documents) and edges are “weighted” links between these nodes, based on the references they share (Goutsmedt and Truc, 2023), i.e., Level-2 documents. The process involved Four key steps described below and synthetized in Figure 1.

Step 1: Temporal networks construction

Scientific literature often prioritizes recent contributions, which influences how bibliographic networks form over time. As a result, a bibliographic coupling network covering two decades is likely to show temporal clustering: documents tend to group together based on their publication period with newer papers clustering around shared recent references, and older ones around earlier citations. To mitigate this temporal bias in academic publication networks, we construct “temporal networks” using a moving five-year window (2007–2011, 2008–2012, 2020–2024). This method, and the resulting manipulation of temporal networks, were implementing thanks to the R networkflow package (Goutsmedt and Truc, 2024). To verify the robustness of our results, we have produced the results for different sizes of the moving window (see Appendix B in Supplementary file 1). Once nodes are filtered by publication year, edges between nodes are established based on shared references. The construction of edges follows three criteria.

Minimum common reference threshold: Two nodes are linked only if they share at least two references (edge threshold > 1). This prevents the inclusion of weak connections based on a single shared reference, thereby reducing noise in the network. We also constructed networks without applying the edge threshold parameter, for comparison (See Appendix B).

Bibliography length: The length of a document’s bibliography affects the likelihood of shared references. Longer bibliographies naturally increase the probability of common citations. Therefore, a shared reference is considered more significant when it appears in documents with shorter bibliographies. For example, if two articles with short bibliographies share two references, the weight of the edge connecting them will be greater than that of an edge connecting two articles with the same number of shared references but longer bibliographies (see Appendix C in Supplementary file 1).

Overall citations of shared references: If a reference shared by two articles is highly cited across the whole corpus, it is considered less significant than a rarely cited reference, which is more likely to indicate a meaningful connection between the articles. When two articles share two references, the less frequently cited reference contributes more to the edge weight between the two articles (see Appendix C).

To account for both the length of an article’s bibliography, and the overall citation frequency of shared references, we use the “coupling similarity” measure (Shen et al., 2019). For comparison, we also examined the results using a simpler measure that considers only bibliography length when constructing edges, that is the coupling angle (see Appendices B, C).

 Step 2: Inter-temporal community detection and visualization

Once the temporal networks were constructed, we used a community detection algorithm to identify distinct clusters of thematically or intellectually connected articles. For each temporal network, we applied the Leiden algorithm (Traag et al., 2019) which aims to find the best partition of the network to obtain dense connections among nodes within the same community, but sparse connections between nodes in distinct communities. By grouping nodes densely connected, such partition of the network allows to identify communities of articles talking about similar themes, using similar methods, data, or theory, etc.

In a second step, we seek to assess the persistence of certain communities across temporal networks. To do so, we compare all communities in pairs between two consecutive temporal networks. Two communities are considered the same “inter-temporal” community if they share more than 55% of their nodes in both directions—that is, if over 55% of the nodes in community i at time t are also in community j at time t + 1, and vice versa.

Finally, we use an alluvial diagram to visualize the evolution of inter-temporal communities over time (Figures 2, 3). Each vertical bar represents a temporal network and is divided into segments corresponding to the inter-temporal communities that compose it, with segment size reflecting the number of nodes.

Alluvial diagram depicting bibliographic coupling communities identified using the coupling similarity method (Threshold = 2, Window = 5, Resolution = 0.6). The diagram illustrates the evolution and thematic shifts in literature citing Livestock’s Long Shadow, with labels positioned at the points where communities emerge over time. Seven different main research strands (represented by colors) were identified, grouping communities (represented by the different shades for each color) that shared common research thematics.

Bar charts illustrate various countries, journals, and scientific disciplines' contributions to key environmental and food-related topics. Categories include emissions modeling, land use, livestock nutrition, and climate change mitigation. Each chart compares percentages of contribution or focus area, with variations in shading representing different domains or countries. Topics such as novel protein innovation, sustainable consumption practices, and socio-technological aspects are also highlighted across several panels.

Alluvial diagram of bibliographic coupling communities using the coupling similarity method (Threshold = 2, Window = 5, Resolution = 0.6). The diagram illustrates the evolution and thematic shifts in literature citing Livestock’s Long Shadow, with labels positioned at the points where communities emerge over time. The figure highlights communities associated with the protein transition, including those related to sustainable consumption practices (Thematic group 3), socio-technological (Thematic group 6), emergence of novel proteins and food innovations thematic groupings (Thematic group 7).

Sankey Diagram showing the proportion of research network strands from 2007 to 2024. The chart illustrates themes like sustainable consumption practices, emergence of novel protein and food innovation, and socio-technological aspects. Key topics include entomophagy, sustainable diets, cultured meat, and political economy. Each theme is color-coded: blue for sustainable consumption, orange for novel protein, and pink for socio-technological factors. The chart highlights how these themes evolve over time.

The flows between vertical bars depict the trajectories of communities across time. Specifically, they show the proportion of nodes from a community i at time t that transition into various communities at time t + 1. In this way, the alluvial diagram reveals both the structure of each temporal network and how this structure evolved across successive periods.

 Step 3: Inter-temporal community description and labeling

Once the inter-temporal communities have been identified, the next step is to generate a series of complementary indicators to characterize their thematic content (see Appendix D in Supplementary file 1). These include: (i) Term Frequency-Inverse Document Frequency, which highlights terms that are common within a given community but rare across others; (ii) Citation-based metrics, used to identify key documents, influential nodes, and patterns of connectivity; (iii) Topic prevalence, capturing the dominant themes associated with each community (see Step 5: Topic Modelling).

Together, these indicators provide an empirical basis for understanding the thematic contours of each inter-temporal community, supporting the assignment of preliminary labels. In the next phase of the analysis (Section 2.3), we draw on these metrics to guide a more in-depth qualitative interpretation and refinement of community labels.

 Step 4: Siloing and connectivity across thematic groups analysis

To assess the degree of siloing or openness among thematic communities citing LLS, we combined diversity indicators applied across multiple domains and network-based connectivity analysis using bibliographic coupling.

We first evaluated how each thematic group distributes its publications across three variables: countries of authors’ affiliation, subject categories, and journals. For each variable, we computed two diversity indicators:

1) Shannon entropy: capturing both richness (number of categories used) and evenness (distribution of outputs across categories). Higher values means that a group displays a larger diversity of levels (countries, subject categories or journals).

2) Herfindahl–Hirschman Index (HHI): capturing concentration and dominance. A high value means that a few countries, categories or journals, dominate the group.

To enable comparison and combination, both indicators were min–max normalized across variables of analysis (countries, subject categories, and journals Figure 4). Shannon entropy was then inverted so that higher values consistently indicate stronger siloing. For each thematic group, we averaged the rescaled Shannon and HHI values across all domains to obtain a composite silo index, where higher scores reflect greater overall siloing.

Thematic groups distribution by geography (on top), by journals (middle), and subject categories (at the bottom).

Bar charts show the top contributors to each thematic group across geographic, subject-area, and journal dimensions. Bars represent each contributor’s share within the group, and points mark their overall global share. Bars taller than the point indicate over-representation relative to the global baseline, allowing comparison of how regions, countries, disciplines, and journals differ across thematic groups.

In parallel, we assessed structural connections between thematic groups, and we also rely on the links constructed for bibliographic coupling networks, where the weight between two documents reflects the number of references they share (see Phase 2). We aggregated these weights at the thematic group level and for all time window to obtain a group-by-group connectivity matrix. The Matrix was row-normalized so that each row sums to 1, providing the relative distribution of a group’s connections to all others (i.e., which proportion of a group’s links goes to itself and to other groups). Visualized as a heatmap, darker diagonal cells signal strong within-group coupling (siloing), while darker off-diagonal cells indicate cross-group interactions and openness Figure 5.

Proportion of thematic groups’ links going to all thematic groups. We take the sum of weighted links for each temporal network and normalize them to calculate the proportion of links going to each thematic group. Then, we average these values on all temporal networks.

Heatmap illustrating the interrelations between the various tehmatic groups. Colors range from light yellow to dark brown, indicating normalized weights from zero to one. Darker hues suggest stronger relationships.

Together, the diversity-based silo indices and the network-derived mixing matrices provide complementary perspectives on how production- and consumption-related themes interact in the academic literature referencing the LLS.

 Step 5: Topic modelling and analysis of topic proximity

Complementing the bibliometric analysis, this textual analysis helps clarify the substantive focus of different communities—such as climate mitigation, sustainable diets, or livestock systems. A topic model identifies k latent themes within a corpus—in this case, the abstracts of our Level-1 documents. The output of a topic model consists of two components:

Topics as mixtures of words: Each topic is represented as a distribution over words from the corpus vocabulary, i.e., the set of unique terms in the corpus. For each topic, the model estimates the probability that a given word belongs to that topic.

Documents as mixtures of topics: Each document is represented as a mixture of topics. For each document, the model estimates the probability that a given topic is present — this is referred to as topic prevalence.

We implement topic modelling using the stm R package (Roberts et al., 2013). Topic modeling serves two main purposes in this study. First, it provides additional information about the inter-temporal bibliographic communities identified earlier (see Step 3). For each inter-temporal community, we can assess which topics are most prevalent, enriching our understanding of their thematic content. In this way, topic modeling complements the bibliographic coupling analysis by focusing on semantic content rather than citation patterns. When both approaches converge on similar themes, such as livestock emissions and related terms like “methane” or “feed efficiency,” it strengthens the reliability of our interpretation (see Appendix E in Supplementary file 1 for more information on topic modeling implementation).

Second, topic modelling allows us to address the third research objective—evaluating whether production and consumption issues are treated as separate domains within the academic literature referencing LLS. To do so, we explore the similarity between topics. If documents are a mixture of topics, then topics can likewise be represented as mixtures of documents. Each topic is thus represented as a vector of length equal to the corpus size, with each value indicating the prevalence of the topic in a given document. We assume that if two topics are prevalent in the same documents, they are likely to share intellectual similarities.

We then performed a principal component analysis (PCA) to reduce the dimensionality of these topic vectors. PCA summarizes the variance across all topics into a smaller number of orthogonal components, allowing us to visualize and interpret topic similarity in a reduced-dimensional space (see Appendix F in Supplementary file 1). We applied the k-means clustering algorithm to the full set of PCA components to identify groups of topics that are thematically related. Finally, we projected these clusters onto a two-dimensional space defined by the first two principal components (Figure 6). This projection enables us to examine which topics are grouped together and how they are positioned relative to one another along the two principal axes. These axes can be interpreted as latent dimensions that capture the most significant differences among the topics, thereby offering insight into the underlying structure of thematic variation, such as whether production- and consumption-related topics are conceptually separated.

Principal Component Analysis (PCA) projection of the 40 topics onto the first two principal components, with colors indicating cluster assignments for k = 2 Dot sizes reflect tipic prevalence, with larger dots indicating greater prevalence.

Bar chart comparing the Herfindahl-Hirschman Index and Shannon (inverted) for several categories, such as "Emergence of novel protein and food innovation" and "Sustainable consumption practices." The x-axis represents the Silo index from zero to sixty percent, where zero is open and one is siloed. Higher values indicate more siloing. Phase 3: Qualitative analysis of the results Step 1: Community profiles characterization

A qualitative analysis was conducted for each community depicted in the alluvial diagram. This analysis focused on the origins of the community, key structural nodes (i.e., those with strong internal cohesion and limited external links), and main thematic trends identified through topic modeling. This approach clarified community formation and thematic focus, enabling the manual assignment of labels using both automated topic modelling and abstract content. The resulting labels and descriptions for each community are compiled in Appendix G in Supplementary file 1.

 Step 2: Thematic grouping

To address this paper’s core objectives—namely, exploring the connections between communities referencing LLS and those focused on the protein transition, a targeted classification process was undertaken allowing to assign communities to larger thematic groups.

The first step involved identifying which communities were associated with the protein transition. To this end, we adopted the framework proposed by Duluins and Baret (2024) which outlines three core narratives, each representing a distinct protein transition thematic group: the consumer-oriented group, focused on consumer behavior and dietary shifts aimed at reducing animal protein consumption and increasing alternative proteins in diets; the techno-centered group focused on innovation and technology-driven development of novel protein sources to replace or supplement animal proteins; and the socio-technological group envisioning systemic transformation of the entire food and protein regime.

Communities that did not align with this framework were classified using an inductive approach. These were grouped based on recurring topics and thematic patterns observed in the dataset, allowing us to capture shared areas of focus and organize the communities into coherent thematic categories.

This classification was conducted manually through a close reading of the labels and descriptions found in Appendix G, which served as the principal reference for assigning each community to the most appropriate thematic group (see Appendix H in Supplementary file 1).

 Results

The results are structured around two main analyses. First, the intertemporal bibliographic coupling analysis, represented through an alluvial diagram illustrating the evolution of different research communities over time (Figure 2), called “thematic groups,” with a focus on protein transition communities (Figure 3) (Section 3.1). Second, an assessment of how open or siloed these groups are (Section 3.2), based on bibliometric indicators (Figures 5, 7). Finally, using textual data from the documents’ abstracts, we apply a topic model and a PCA (Figure 6) to examine how text-based clusters align with the identified thematic groups.

Aggregated indices of openess/siloing based on Shannon entropy and Herfindahl–Hirschman Aggregated indices of openess/siloing based on Shannon entropy and Herfindahl–Hirschman for each thematic group. The x-axis represents the Silo index in percentage, where zero is open and 100% is siloed. Higher values thus indicate more siloing.

Heatmap illustrating the interrelations between various factors related to sustainable consumption practices. The axes represent categories such as "Land use, biodiversity and ecosystem services," "Socio-technological," and "Emergence of novel protein and food innovation." Colors range from light yellow to dark brown, indicating normalized weights from zero to 0.8. Darker hues suggest stronger relationships, primarily along the diagonal. Mapping the evolution of research communities over time

An alluvial diagram illustrates the temporal evolution of research communities over successive five-year windows, based on bibliographic coupling analysis (Figure 2). Each vertical bar represents the research network structure at a given time, with individual communities depicted as blocks proportional to their number of nodes (i.e., publications). The connecting flows indicate the movement of nodes between communities, capturing how communities emerged, persisted, merged, or dissolved over time. Communities sharing a significant proportion of nodes across time windows are defined as part of the same inter-temporal community (see Step 2 of section 2.2). Each community is numbered based on its order of appearance and labeled according to the themes identified in in Phase 3).

In total, 80 inter-temporal communities were identified, each with distinct thematic and temporal patterns. However, we focus on a subset of 29, selecting those that comprised more than 5% of the nodes of a network in at least one time window. This subsample actually represents more than 95% of all the nodes in the networks. Following the thematic grouping described in Section 2.3. (Step 2), these 29 communities were organized into seven thematic groups (indicated with colors in Figure 2), each representing a major research trajectory:

1) Emissions modeling and nutrient pollution communities (Thematic group 1)

2) GHG emissions and climate change mitigation communities (Thematic group 2)

3) Sustainable consumption practices related communities (Thematic group 3)

4) Land use, biodiversity and ecosystem services communities (Thematic group 4)

5) Livestock nutrition, microbiome and emission reduction strategies communities (Thematic group 5)

6) Socio-technological related communities (Thematic group 6)

7) Emergence of novel protein and food innovation communities (Thematic group 7)

At the country level, the United States of America (USA) are the top contributor in five of the seven Thematic groups: GHG emissions and climate change mitigation communities (Thematic group 2), Sustainable consumption practices related communities (Thematic group 3), Land use, biodiversity and ecosystem services communities (Thematic group 4), Socio-technological related communities (Thematic group 6) and Emergence of novel protein and food innovation communities (Thematic group 7; see Figure 4). While this dominance is not unexpected, it is noteworthy that the United States is disproportionately represented—relative to the overall corpus—in thematic groups 3 and 6. The United Kingdom is also significantly over-represented in Sustainable consumption practices related communities (Thematic group 3) and Socio-technological related communities (Thematic group 6), for which it is the second contributor behind the USA. Denmark is a leading contributor to publications within the Emissions modeling and nutrient pollution thematic group (Thematic group 1), while India contributes the most to Livestock nutrition, microbiome and emission reduction strategies (Thematic group 5). The Netherlands ranks as the second major contributor of publications within Emissions modeling and nutrient pollution communities (Thematic group 1), GHG emissions and climate change mitigation communities (Thematic group 2) and Emergence of novel protein and food innovation communities (Thematic group 7; Figure 4).

Each thematic group is associated with distinct subject areas and publication venues. For example, Thematic group 1 (Emissions modelling and nutrient pollution) aligns primarily with environmental chemistry, animal science and zoology, with most articles being published in journals such as the Journal of Environmental Quality, and Environmental Science and Technology. In contrast, Thematic group 7 (Emergence of novel protein and food innovation) is rooted in food and insect science, and industrial and manufacturing engineering, with top-leading journals including the Journal of Insects as Food and Feed, Critical reviews in Food Science and Nutrition, and Frontiers in Nutrition Figure 4.

For each thematic group, the following paragraphs trace the chronological and thematic evolution of the associated research communities, emphasizing key developments and shifts over time.

We present the thematic groups starting by reading Figure 2 from left to right, thus presenting the thematic groups as they appear in time.

1. Thematic group 1: Emissions modeling and nutrient pollution communities

Between 2007 and 2018, research communities made important advances in understanding and mitigating emissions and nutrient pollution from livestock systems. Early efforts concentrated on spatial modeling of pollutant flows, particularly nitrogen compounds such as ammonia and nitrous oxide, as well as methane. These studies examined the geographic distribution and environmental impact of emissions, linking them to broader issues of climate change, air quality, and eutrophication (intertemporal community 5). In parallel, technical research focused on methane mitigation through improved manure management practices, including slurry separation, composting, and optimized storage techniques (Community 6).

Beginning around 2011, attention expanded to include the microbial and nutritional drivers of methane production. Research explored how dietary interventions, microbial inoculants, and feed composition influence rumen fermentation and the chemical properties of slurry, highlighting interactions between livestock nutrition, microbial activity, and manure emissions (Community 43).

From 2012 onward, an integrated approach to livestock sustainability emerged, combining emissions reduction with energy efficiency and waste management. Studies assessed the environmental performance of different housing and production systems, including organic dairying and traditional livestock practices, aiming to develop more efficient and climate-resilient models of livestock husbandry (Community 57).

2. Thematic group 2: GHG emissions and climate change mitigation communities

Between 2007 and 2024, research on greenhouse gas emissions and climate change mitigation in agriculture has moved from global assessments to applied, region-specific strategies. Initial work between 2007 and 2011 critically examined the environmental impacts of livestock systems through life cycle assessments, specifically looking at the GHG footprint of livestock production systems, and the role of land use in carbon dioxide emissions. In parallel, studies explored the trade-offs between expanding biofuel production, livestock management, land-use change, and disruptions to the nitrogen cycle, emphasizing the interconnectedness of global food, energy, and environmental systems (Community 4 & Community 3).

Between 2008 and 2013, research shifted to focus on practical mitigation approaches, focusing on soil carbon sequestration, pasture management and restoration, and methane reduction using dietary interventions and improved manure management. These efforts aimed at developing farming systems that support both productivity and environmental sustainability (Community 12 & Community 15).

From 2009 to 2018, a broader sustainability perspective took shape, linking nutrient use efficiency and land-use governance with global climate mitigation strategies. This phase emphasized the integration of ecological science, policy modeling, and economic instruments to advance resilient, efficient agri-food systems (Community 18).

In parallel, since 2014, a more applied and regionally grounded body of research has focused on climate adaptation in livestock systems. This includes the use of agricultural waste biomass for renewable energy, the adoption of silvopastoral practices in the Amazon, and biodiversity-focused approaches to livestock sustainability (Community 95).

3. Thematic group 3: Sustainable consumption practices related communities

Research on food consumption has played a central role in the debate on transforming food systems, developing through distinct intellectual currents that reflect evolving concerns about health, ethics, sustainability, and public policy.

The first research community emerged between 2007 and 2012, focusing on plant-based diets, public health, and climate change. This line of inquiry explored how reducing the consumption of animal products benefits both human and planetary health. It also highlighted the ethical motivations underpinning vegetarianism and veganism and examined the role of medical professionals in promoting dietary transitions (Community 2).

By 2009, this community evolved to encompass a stronger ethical and political dimension, emphasizing individual moral responsibility and intergenerational justice. Researchers increasingly examined how policy instruments such as food labeling and consumption restrictions could guide sustainable dietary choices (Community 17).

From 2015 onwards, the focus shifted to more institutional and systemic approaches. One key research community looked at public procurement practices and the role of advocacy in reshaping food environments. Scholars explored how institutional food services could serve as levers to reduce meat consumption, often mobilizing environmental arguments to support dietary change (Community 96).

4. Thematic group 4: Land use, biodiversity and ecosystem services communities

Research on land use and ecosystem services has progressively shifted from ecological mechanisms to more integrated approaches addressing sustainability and climate adaptation. Early work focused on agro-ecosystems and pasture management in relation to soil carbon sequestration and greenhouse gas mitigation (Community 12).

Around 2012, the focus shifted to the ecological roles of large herbivores and carnivores, exploring their impact on carbon cycling and ecosystem stability (Community 60). This trajectory deepened with studies on the functional and biodiversity effects of grazing by native versus introduced herbivores, highlighting how grazing intensity shapes ecosystem responses in drylands and semi-arid environments (Community 72).

From 2015 onward, research increasingly addressed the sustainability of livestock systems in tropical and forested regions, particularly through silvopastoral practices and landscape-level approaches to climate resilience (Community 129 & Community 95). These communities focused on improving livestock welfare and reducing environmental impact while strengthening food and nutritional security. In parallel, attention turned to the role of agricultural waste, carbon footprints, and sustainable intensification strategies in the Amazon and other vulnerable landscapes (Community 95).

5. Thematic group 5: Livestock nutrition, microbiome, and emission reduction strategies communities

Between 2008 and 2022, research communities made significant strides in developing integrated strategies to reduce greenhouse gas emissions from livestock by targeting nutrition and microbial processes. Early investigations focused on dietary interventions such as incorporating tannin-rich forages and adjusting crude protein levels to lower methane and ammonia emissions while enhancing feed efficiency and manure quality, meaning its nutrient content and suitability for use as fertilizer (Community 13). Parallel studies explored the environmental benefits of utilizing feed byproducts to further reduce methane emissions and improve manure chemistry, referring to the chemical composition of manure (e.g., nitrogen forms, pH, and carbon content) in ways that lower its contribution to eutrophication and global warming potential (Community 20).

Beginning in 2011, attention shifted towards the impact of feed additives, microbial inoculants, and dietary modifications on rumen fermentation and slurry emissions, underscoring the critical connections between nutrition, microbial activity, and environmental outcomes (Community 43).

From 2016 onward, a genomic and microbiological perspective emerged, investigating the rumen microbiome, host–microbe interactions, and heritable microbial traits—highlighting how genetic and dietary factors jointly shape emissions and feed conversion efficiency (Community 115).

6. Thematic group 6: Socio-technological related communities

Building on a broader shift in research focus, two distinct research communities have emerged since 2017, both reflecting a socio-technological perspective moving beyond individual behavior to explore systemic transformations involving evolving norms, regulatory frameworks, and coordinated action across the food system.

The first research community, developed between 2017 and 2022, explores how plant-based diets challenge entrenched social norms, particularly within institutional settings like healthcare. Here, veganism is often stigmatized and framed as a deviant or fringe practice. This research also examines the influence of media and digital activism—such as the Finnish Vegan Challenge and documentaries like Cowspiracy—in shaping public perceptions. Additionally, it addresses ongoing challenges around food labeling and the competition between alternative and animal-based protein foods (Community 123).

The second research community, active from 2019 through 2024, focuses on the political economy of meat and the psychological and social factors that shape meat consumption. It investigates how consumers navigate tensions between ethical or environmental concerns and everyday eating habits. Tools like the Swedish Meat Guide are examined for their role in enabling more informed choices, while attention is also given to how institutional norms may continue to reinforce barriers to change (Community 144).

7. Thematic group 7: Novel protein and food innovation communities

Research into novel proteins began to emerge as a distinct field as early as 2009 when scholars began exploring the potential of insects as food and feed. These studies emphasized the nutritional and ecological benefits of entomophagy and discussed the socio-cultural and regulatory challenges of integrating insects into Western diets (Community 19 & Community 47).

From 2016 onward, this research evolved to encompass a broader vision of food innovation—one that includes safety and nutritional assessments, consumer acceptance, and the ethical and marketing dimensions of alternative proteins. This evolution marked the rise of a more integrated perspective, where novel proteins are situated within a larger agenda of structural transformation aimed at tackling global challenges such as food security, environmental sustainability, and public health (Community 110).

Meanwhile, work on lab-grown and cultured alternatives gained prominence from 2014 onward. Researchers examined technologies such as cultured meat, clean milk, and advanced plant-based proteins, along with their implications for biotechnology, animal welfare, and market dynamics. These investigations also considered the political economy of cellular agriculture, and the cultural and ethical changes needed to reimagine food production (Community 86 & Community 131).

The three thematic groups (Sustainable consumption practices related communities, Socio-technological related communities and Novel protein and food innovation communities) reflect key research communities associated with the protein transition, evolving from broad concerns about sustainable consumption to more targeted investigations into alternative protein sources—first insects, then cultured meat—and gradually moving toward a systemic understanding of the socio-structural factors shaping dietary choices (Figure 3).

 Structural connectivity and fragmentation in the thematic landscape

This section assesses how open or siloed thematic groups are using bibliometric data such as journal, subject, country of authors affiliations, and citation links between communities. To strengthen the analysis of separation between groups, we also use textual data from abstracts. We apply topic modeling and a PCA to examine how text-based topic clusters align with the bibliographic communities.

Siloing and connectivity across thematic groups Silo indices

Across all indicators and domains, GHG emissions and climate change mitigation (Thematic group 2) emerges as the least siloed thematic group (Figure 7). It displays both high Shannon diversity and low HHI concentration, indicating that its contributions are broadly and evenly distributed across countries, subjects, and journals.

At the opposite end of the spectrum, Livestock nutrition, microbiome and emission reduction strategies (Thematic group 5) and Emissions modeling and nutrient pollution (Thematic group 1) are the most siloed groups (Figure 7). Their outputs rely heavily on a narrow set of categories, with strong dominance by a few specific subjects or locations. Emergence of novel protein and food innovation (Thematic group 7) also shows substantial siloing, driven less by low richness than by the dominance of one or two central categories.

Two thematic groups, Sustainable consumption practices (Thematic group 3) and Socio-technological (Thematic group 6), display intermediate and mixed profiles (Figure 7). Sustainable consumption spans many categories but with marked dominance patterns, while Socio-technological engages fewer categories overall but distributes its contributions relatively evenly. These contrasting forms of diversity explain their differing ranks between Shannon and HHI and position both Thematic groups in the mid-range of the combined silo index.

Taken together, the combined silo indices show a clear divide: a small number of thematic areas are relatively open and cross-cutting, while several others remain distinctly siloed, with limited cross-domain engagement.

 Aggregated mixing matrix

Three groups exhibit particularly strong within-group coupling (Figure 5): Land use, biodiversity and ecosystem services (Thematic group 4), Livestock nutrition, microbiome and emission reduction strategies (Thematic group 5), and Novel protein and food innovation (Thematic group 7). In other words, the articles of these groups tend to cite mainly references cited by the other documents of the group and share few references with the other groups.

Here again, GHG emissions and climate change mitigation (Thematic group 2) appears as the most opened group. It acts notably as a bridge between production-oriented group like Emission modelling and nutrient pollution (Thematic group 1) on one hand, and protein-transition groups that are more consumption-focused, such as Sustainable consumption practices (Thematic group 3) and Socio-technological (Thematic group 6).

Cross-cluster interactions are also evident within the protein-transition–related groups. For example, even if showing a particularly high intra-group connectivity (see above), Novel protein and food innovation (Thematic group 7) displays a notable connectivity with Socio-technological (Thematic group 6) and Sustainable consumption practices (Thematic group 3).

 Principal component analysis

With the PCA, four clusters were identified based on all principal components, Figure 6 displays the first two principal components, which explain 10 and 7% of the variance, respectively. Although the explained variance is relatively low, indicating that much of the variation among topics is captured by higher-order components, these first two axes still capture meaningful distinctions in the overall structure of topic similarity.

The horizontal axis reveals a clear gradient. On the right side, topics cluster around consumption-oriented themes (e.g., Topic 9 and 36) and alternative proteins (e.g., Topic 28 and 39). On the left side, topics are more aligned with livestock production systems and their environmental impacts, including greenhouse gas emissions (Topic 7) and livestock waste management (Topic 19).

The vertical axis appears to capture a scale from specific to systemic focus. Topics at the top deal with more targeted elements, such as diets (Topic 39), meat consumption (Topic 36), or animal welfare (Topic 23). In contrast, the lower part of the plot features broader, system-level themes, such as climate change assessments (Topics 7 and 12), global agricultural land use (Topic 35), and crop production systems (Topic 27).

While PCA helps reveal the underlying structure of topic similarity and allows us to visualize broad thematic gradients within the literature, it does not directly assess how these topic clusters correspond to the thematic groupings identified through bibliographic analysis. To explore this alignment more systematically, we analyzed the distribution of topic clusters identified through topic modeling in relation to our predefined bibliographic communities, examining whether specific themes were disproportionately represented within each group. This comparison provides insight on how the conceptual structure of the literature (as captured by topic modeling) mirrors its citation-based structure.

To quantify this relationship, we computed the log-ratio comparing the observed and expected co-occurrence of documents in each cluster-thematic group pair (see Appendix I in Supplementary file 1 for more details). Positive log-ratios suggest that a topic cluster is more prominent in a bibliographic thematic group than would be expected by chance, while negative values point to under-representation. These relationships are visualized in a heatmap (Figure 8). The results reveal that Clusters 1 and 2 identified through the PCA are notably associated with thematic groupings related to the protein transition (that is thematic groupings 6, 3 and 7: Socio-technological, Sustainable consumption practices and Emergence of novel protein and food innovation). In contrast, Cluster 4 is more aligned with bibliographic communities focused on livestock production systems, including among others Thematic groups 1 (Emissions modeling and nutrient pollution), 4 (Land use, biodiversity, and ecosystem services), and 5 (Livestock nutrition, microbiome, and emission reduction strategies). Finally, Cluster 3 shows a less contrasting profile but leans toward system-level topics, such as greenhouse gases emissions and climate change Figure 8.

Log-ratio heatmap comparing topic clusters (K-Means) with bibliographic thematic groupings. Each cell shows the log₂-ratio between observed and expected co-occurrences of documents in a given topic cluster and thematic group. Positive values indicate over-representation (greater overlap than expected by chance), while negative values indicate under-representation.

Positive values indicate over-representation (greater overlap than expected by chance), while negative values indicate under-representation.

Taken together, this comparison reinforces the PCA findings: themes linked to the protein transition are concentrated in topic clusters emphasizing consumption-related dynamics, while other thematic communities are more aligned with system-level and production-oriented concerns.

 Discussion Key takeaways of the results in light of the research questions

This study offers the first bibliometric analysis of the Livestock’s Long Shadow (LLS) report, a landmark publication in global environmental discourse. Through the combined use of bibliographic coupling and topic modeling, we identified and characterized seven distinct thematic groups each containing different research communities that have cited the LLS in scientific literature. These Thematic groups, reflect distinct areas of inquiry, ranging from emissions modelling and nutrient pollution to climate change mitigation, sustainable consumption, land use and biodiversity, and the emergence of novel proteins and food innovations.

Among the seven thematic groups identified in this paper, three are strongly aligned with the narratives of protein transition identified in Duluins and Baret (2024). In their paper, the authors identified three main narratives of the protein transition, embodying different visions of change for future food systems, namely, the consumer narrative (corresponding to the sustainable consumption research strand in this research), the techno-centred narrative (corresponding to the protein-food innovations in this paper), and the socio-technological narrative (corresponding to the socio-technological research strand of this paper) (Duluins and Baret, 2024). Focusing on the first two thematic groups, the consumer narrative emphasizes reducing unsustainable consumption patterns by encouraging dietary change and studying emerging trends such as vegetarianism and veganism (Deckers, 2009; Joyce et al., 2008) (Intertemporal Community 2). The core idea is to lower meat consumption by substituting animal proteins with plant-based or other alternative sources (Meyer and Bäumer, 2020; van Huis and Oonincx, 2017) (Community 47 & Community 123). In this narrative, consumers are the main agents of change; transformation is expected to result from individual choices and behavioral shifts (Springmann et al., 2016; Stehfest et al., 2009; Westhoek et al., 2014) (Community 17 & Community 96). The techno-centered narrative, in contrast, highlights the inefficiencies of current protein production systems, often illustrated by the large quantity of crops required to produce a much smaller amount of animal protein. This narrative promotes the development of new, more resource-efficient production methods that generally exclude animals from future food systems (Heidemann et al., 2020) (Community 131). The focus lies on research and innovation, particularly the development and scaling of alternative protein technologies such as lab-grown meat or plant-based analogues (Cornelissen and Piqueras-Fiszman, 2023; Ruzgys and Pickering, 2020) (Community 131). Together, these narratives show how LLS not only shaped citation networks but also framed sustainability challenges, legitimized specific solutions (consumer behavior change or alternative protein development), and reinforced shared narratives that continue to structure contemporary sustainability research (Duluins, 2025).

Narratives matter not only because they shape how problems are defined and which solutions are legitimized, but because they also influence political priorities and research agendas (Béné et al., 2019; Béné and Lundy, 2023; Duluins and Baret, 2024). A consumer-focused narrative channels policy and funding toward dietary change through education campaigns, fiscal measures, and behavioral research, while a techno-centered narrative directs resources to developing and commercializing alternative protein technologies (Duluins and Baret, 2024). Together, these narratives determine whether sustainability efforts by research and funding target consumption patterns or technological innovation, and where public and private money should be spent, both in terms of political and research priorities.

Analyses of bibliometric clusters, principal component analysis and the combined analyses of diversity indices and bibliographic coupling all revealed a pronounced division between production- and consumption-oriented thematic communities within the LLS literature. Consumption-oriented topics such as meat intake reduction, plant-based diets, and ethical eating cluster distinctly from those related to production systems, like livestock emissions, manure management, and land use. This structural separation suggests that academic discourse continues to treat these domains in parallel rather than in an integrated, systems-oriented fashion. Moreover, the PCA also revealed a secondary axis distinguishing specific interventions (e.g., dietary change, consumer behavior) from broader systemic concerns (e.g., climate change, global land use). This suggests that much of the literature still oscillates between micro-level behavioral studies and macro-level environmental modeling, with relatively few studies bridging both scales.

While some thematic groups, notably GHG emissions and climate change mitigation, act as cross-cutting connectors linking otherwise siloed areas, several production-focused groups remain highly insular. Intermediate thematic groups display mixed profiles, underscoring that siloing is multidimensional: richness and dominance jointly shape the openness of a thematic group. Structurally, the bibliographic coupling network confirms that strong within-group cohesion does not preclude outward connectivity, yet persistent gaps between production and consumption research highlight ongoing challenges for integrated knowledge development.

Connecting LLS with the protein transition is theoretically significant because it provides an opportunity to understand why certain conceptual and institutional “silos” persist, despite the interdependence between production and consumption systems. While some overlap exists, our findings suggest that the communities focusing on livestock systems and their impacts at large and those focused on the protein transition remain relatively siloed scientific communities.

Rather than interpreting these silos as the result of deliberate exclusion, they can be understood as an emergent outcome of disciplinary evolution: as the initial LLS concept diversified, specialized clusters developed, resulting in parallel and partially disconnected research streams.

Recognizing the existence and evolution of these communities serves three important purposes. First, it highlights the diversity and overall configuration of the research landscape, enabling scholars to situate their work within a broader system. Second, it supports efforts to foster collaboration across communities—for example, through interdisciplinary approaches, integrated modeling, or policy research that links production- and consumption-side outcomes (e.g., combining life-cycle assessment with behavioral research on dietary change). Third, it strengthens systemic awareness by underscoring how consumer behavior is shaped by food environments, which themselves are influenced by subsidies, trade policies, and production decisions.

Overall, the LLS report has influenced citation networks and the formation of distinct research communities. By framing environmental problems and legitimizing specific solutions, these communities shape research agendas, funding priorities, and policy discourse. Understanding silos as an emergent feature of research evolution allows scholars and policymakers to appreciate the diversity of approaches while remaining aware of the broader system dynamics at play.

 Limitations of this study

A first limitation concerns the data used for topic modeling. Specifically, we relied on abstracts, which although they provide a concise summary of article content—are inherently limited in depth. As short texts, abstracts constrain the expressiveness and granularity of thematic analysis, potentially oversimplifying the content and obscuring less prominent themes. This may reduce the capacity of topic modelling to fully capture the richness and complexity of scholarly debates within the corpus.

A second limitation regards to the network clustering method used, namely, the Leiden algorithm, which suffers from the classic “resolution limit” problem associated with modularity-based algorithms (Traag et al., 2011). Such algorithms tend to overlook smaller communities when optimizing modularity, instead favoring larger communities that contribute more to the overall modularity score. As a result, thematically coherent but relatively small groups of documents may be merged into broader communities, potentially obscuring finer-grained intellectual distinctions within literature. This is likely to happen in our case, notably with the largest communities (Community 18, for instance). To minimize the impact of this issue, we varied the resolution parameter of the Leiden algorithm, which controls the number of communities identified. We also experimented with different edge weighting measures and adjusted the size of the time window used to construct temporal networks (see Appendix B). These variations produced different community partitions within each temporal network, and so different inter-temporal communities. The goal was to compare our chosen set of parameters—and the interpretation derived from them—with alternative results, thereby ensuring that the observed trends are robust and not artifacts of specific parameter choices.

 Theoretical and practical implications

By identifying the main research communities that have cited the LLS, this study provides a foundational step toward understanding how scientific communities on livestock sustainability have evolved and fragmented over time. Methodologically, the study offers a replicable scientometric approach that can be applied to other landmark publications to trace the diversity and interconnections of research communities. Conceptually, it offers an entry point for an epistemological reflection on how sustainability problems are framed and legitimized across disciplines, taking the specific case of the protein transition as an example.

From a practical perspective, the findings clarify how research communities working on livestock sustainability and the protein transition are structured and where their research focus and priorities diverge. This mapping can help organizations, policymakers, and interdisciplinary networks identify areas where expertise is fragmented and where closer collaboration could strengthen system-oriented approaches to food-system sustainability.

 Future research directions

This paper lays the groundwork for a more integrated understanding of how foundational reports like LLS shape scientific discourse over time. Future research could deepen this analysis by incorporating full-text semantic analysis or citation context analysis to capture not just that LLS is cited, but how it is referenced across domains.

Additionally, future work could explore how more recent high-impact publications (e.g., IPCC reports, EAT-Lancet Commission, etc.) interact with or displace LLS in shaping research trajectories. There is also scope for investigating the translation of scientific discourse into policy and public narratives, examining whether the silos observed in academia mirror or diverge from those in governance and advocacy.

Ultimately, bridging the production–consumption divide in both research and policy will be critical for achieving sustainable food system transitions.

 Conclusion

This paper has examined how the 2006 FAO Livestock’s Long Shadow (LLS) report shaped academic discourse on livestock and environmental sustainability. By mapping the research communities that cite the report and analyzing the thematic content of their work, we identified seven major thematic groupings of research communities, ranging from emissions modeling and land use to sustainable consumption and food innovation. Among these, three thematic groupings align closely with the field of protein transition, highlighting a growing interest in plant-based diets, socio-technical change, and alternative proteins.

Our findings point to a persistent fragmentation in the literature with research communities focusing on livestock production and their environmental impacts remaining largely distinct from protein transition communities focusing on consumption and alternative proteins. The principal component analysis confirms this divide, revealing a structural separation between production- and consumption-focused topics, with protein transition communities focusing on the latter, as well as between specific interventions and broader systemic concerns.

 Data availability statement

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author. Programming scripts used to produce the data analysis and figures can be found at: https://github.com/agoutsmedt/livestock_long_shadow.

 Author contributions

OD: Validation, Formal analysis, Investigation, Conceptualization, Writing – review & editing, Writing – original draft, Project administration. AG: Visualization, Validation, Investigation, Writing – review & editing, Methodology, Software. NV: Data curation, Conceptualization, Writing – review & editing. PB: Conceptualization, Supervision, Writing – review & editing.

 Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The author(s) declared that they were an editorial board member of Frontiers, at the time of submission. This had no impact on the peer review process and the final decision.

 Generative AI statement

The authors declare that no Gen AI was used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

 Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

 Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fsufs.2025.1656562/full#supplementary-material

 References BealT. GardnerC. D. HerreroM. IannottiL. L. MerboldL. NordhagenS. . (2023). Friend or foe? The role of animal-source foods in healthy and environmentally sustainable diets. J. Nutr. 153, 409–425. doi: 10.1016/j.tjnut.2022.10.016, 36894234 BénéC. LundyM. (2023). Political economy of protein transition: battles of power, framings and narratives around a false wicked problem. Front. Sustain. 4:1098011. doi: 10.3389/frsus.2023.1098011 BénéC. OosterveerP. LamotteL. BrouwerI. D. de HaanS. PragerS. D. . (2019). When food systems meet sustainability – current narratives and implications for actions. World Dev. 113, 116–130. doi: 10.1016/j.worlddev.2018.08.011 BoadaL. D. Henríquez-HernándezL. A. LuzardoO. P. (2016). The impact of red and processed meat consumption on cancer and other health outcomes: epidemiological evidences. Food Chem. Toxicol. 92, 236–244. doi: 10.1016/j.fct.2016.04.008, 27106137 BristowE. (2011). Global climate change and the industrial animal agriculture link: the construction of risk. Soc. Anim. 19, 205–224. doi: 10.1163/156853011X578893 BrownN., 2020. 7 reasons why meat is bad for the environment Greenpeace UK. Available online at: https://www.greenpeace.org.uk/news/why-meat-is-bad-for-the-environment/ (Accessed December 9, 2024). CornelissenK. Piqueras-FiszmanB. (2023). Consumers’ perception of cultured meat relative to other meat alternatives and meat itself: a segmentation study. J. Food Sci. 88, 91–105. doi: 10.1111/1750-3841.16372, 36374214 de BoerJ. SchöslerH. AikingH. (2014). Meatless days or “less but better”? Exploring strategies to adapt Western meat consumption to health and sustainability challenges. Appetite 76, 120–128. doi: 10.1016/j.appet.2014.02.002, 24530654 DeckersJ. (2009). Vegetarianism, sentimental or ethical? J. Agric. Environ. Ethics 22, 573–597. doi: 10.1007/s10806-009-9176-3 DuluinsO., 2025. Scientific narratives for the protein transition: will systemic change be possible? UCL - Université Catholique de Louvain, Louvain-La-Neuve, Belgium. DuluinsO. BaretP. V. (2024). A systematic review of the definitions, narratives and paths forwards for a protein transition in high-income countries. Nat. Food 5, 28–36. doi: 10.1038/s43016-023-00906-7, 38172295 European Environment Agency, 2005. The Kyoto Protocol entered into force 16 February 2005. Available online at: https://www.eea.europa.eu/highlights/Ann1108541351 (Accessed May 15, 2025). GlatzleA. (2014). Questioning key conclusions of FAO publications ‘livestock’s long shadow’ (2006) appearing again in ‘tackling climate change through livestock’ (2013). Pastoralism 4:1. doi: 10.1186/2041-7136-4-1 GoutsmedtA. TrucA. (2023). An independent European macroeconomics? A history of European macroeconomics through the lens of the European economic review. Eur. Econ. Rev. 158:104559. doi: 10.1016/j.euroecorev.2023.104559 GoutsmedtA. TrucA., 2024. Networkflow: functions for a workflow to manipulate networks. doi: 10.5281/zenodo.10458040 HaddawayN.R. GraingerM.J. GrayC.T., 2021. Citationchaser: an R package and shiny app for forward and backward citations chasing in academic searching. HeidemannM. S. MolentoC. F. M. ReisG. G. PhillipsC. J. C. (2020). Uncoupling meat from animal slaughter and its impacts on human-animal relationships. Front. Psychol. 11:1824. doi: 10.3389/fpsyg.2020.01824, 32849075 HerzonI. MazacR. ErkkolaM. GarnettT. HanssonH. KaljonenM. . (2023). A rebalanced discussion of the roles of livestock in society. Nat. Food 4, 926–927. doi: 10.1038/s43016-023-00866-y, 37821624 HoldenL. A. (2020). Understanding the environmental impact of global dairy production. J. Anim. Sci. 98:skz365. doi: 10.1093/jas/skz365 JenkinsW. M. N. TrindadeL. M. PyettS. van MierloB. WelchD. van ZantenH. H. E. (2024). Will the protein transition lead to sustainable food systems? Glob. Food Secur. 43:100809. doi: 10.1016/j.gfs.2024.100809 JoyceA. W. DixonS. ComfortJ. HallettJ. (2008). The cow in the room: public knowledge of the links between dietary choices and health and environmental impacts. Environ. Health Insights 1:EHI.S914. doi: 10.4137/EHI.S914, 21572845 Kingston-SmithA. H. EdwardsJ. E. HuwsS. A. KimE. J. AbbertonM. (2010). Plant-based strategies towards minimising ‘livestock’s long shadow. Proc. Nutr. Soc. 69, 613–620. doi: 10.1017/S0029665110001953, 20682089 KristiansenS. PainterJ. SheaM. (2021). Animal agriculture and climate change in the US and UK elite media: volume, responsibilities, causes and solutions. Environ. Commun. 15, 153–172. doi: 10.1080/17524032.2020.1805344, 33688373 LeroyF. AbrainiF. BealT. Dominguez-SalasP. GregoriniP. ManzanoP. . (2022). Animal board invited review: animal source foods in healthy, sustainable, and ethical diets – an argument against drastic limitation of livestock in the food system. Animal 16:100457. doi: 10.1016/j.animal.2022.100457, 35158307 LumsdenC. L. JägermeyrJ. ZiskaL. FanzoJ. (2024). Critical overview of the implications of a global protein transition in the face of climate change: key unknowns and research imperatives. One Earth 7, 1187–1201. doi: 10.1016/j.oneear.2024.06.013 MeyerD. BäumerT. (2020). “Less meat, less heat—the potential of social marketing to reduce meat consumption” in Innovations for metropolitan areas: Intelligent solutions for mobility, logistics and infrastructure designed for citizens. eds. PlaningP. MüllerP. DehdariP. BäumerT. (Berlin, Heidelberg: Springer). MorrisV. JacquetJ. (2024). The animal agriculture industry, US universities, and the obstruction of climate understanding and policy. Clim. Chang. 177:41. doi: 10.1007/s10584-024-03690-w MouselimisL. 2024 fastText: Efficient Learning of Word Representations and Sentence Classification using R MuschelliJ. 2019 rscopus: Scopus Database “API” Interface NeslenA. 2023 ‘The anti-livestock people are a pest’: how UN food body played down role of farming in climate change. The Guardian PiteskyM. E. StackhouseK. R. MitloehnerF. M. (2009). “Chapter 1 - clearing the air: livestock’s contribution to climate change” in Advances in agronomy, advances in agronomy. ed. SparksD. L. (Boston: Academic Press). PooreJ. NemecekT. (2018). Reducing food’s environmental impacts through producers and consumers. Science 360, 987–992. doi: 10.1126/science.aaq0216, 29853680 QianF. RiddleM. C. Wylie-RosettJ. HuF. B. (2020). Red and processed meats and health risks: how strong is the evidence? Diabetes Care 43, 265–271. doi: 10.2337/dci19-0063, 31959642 RobertsM.E. StewartB.M. TingleyD. AiroldiE., 2013. The structural topic model and applied social science. Presented at the International Conference on Neural Information Processing. Avilable online at: https://www.semanticscholar.org/paper/The-structural-topic-model-and-applied-social-Roberts-Stewart/f72719e1771603fb902d6b14f8e086282f91f922#citing-papers RockströmJ. ThilstedS. H. WillettW. C. GordonL. J. HerreroM. HicksC. C. . (2025). The EAT-lancet commission on healthy, sustainable, and just food systems. Lancet 406, 1625–1700. doi: 10.1016/S0140-6736(25)01201-2 RuzgysS. PickeringG. J. (2020). Perceptions of cultured meat among youth and messaging strategies. Front. Sustain. Food Syst. 4:122. doi: 10.3389/fsufs.2020.00122, 41307009 ShenS. ZhuD. RousseauR. SuX. WangD. (2019). A refined method for computing bibliographic coupling strengths. J. Informetr. 13, 605–615. doi: 10.1016/j.joi.2019.01.012 SpringmannM. GodfrayH. C. J. RaynerM. ScarboroughP. (2016). Analysis and valuation of the health and climate change cobenefits of dietary change. Proc. Natl. Acad. Sci. 113, 4146–4151. doi: 10.1073/pnas.1523119113, 27001851 StehfestE. BouwmanL. van VuurenD. P. den ElzenM. G. J. EickhoutB. KabatP. (2009). Climate benefits of changing diet. Clim. Chang. 95, 83–102. doi: 10.1007/s10584-008-9534-6 SteinfeldH. GerberP. (2010). Livestock production and the global environment: consume less or produce better? Proc. Natl. Acad. Sci. 107, 18237–18238. doi: 10.1073/pnas.1012541107, 20935253 SteinfeldH. GerberP. WassenaarT. CastelV. RosalesM. de HaanC., 2006. Livestock’s long shadow. Rome: Food and Agriculture Organization of the United Nations (FAO) TraagV. A. Van DoorenP. NesterovY. (2011). Narrow scope for resolution-limit-free community detection. Phys. Rev. E 84:016114. doi: 10.1103/PhysRevE.84.016114, 21867264 TraagV. A. WaltmanL. van EckN. J. (2019). From Louvain to Leiden: guaranteeing well-connected communities. Sci. Rep. 9:5233. doi: 10.1038/s41598-019-41695-z, 30914743 van HuisA. OonincxD. G. A. B. (2017). The environmental sustainability of insects as food and feed. A review. Agron. Sustain. Dev. 37:43. doi: 10.1007/s13593-017-0452-8 VergunstF. SavulescuJ., 2017. Five ways the meat on your plate is killing the planet. The Conversation. Available online at: http://theconversation.com/five-ways-the-meat-on-your-plate-is-killing-the-planet-76128 (Accessed December 9, 2024). WesthoekH. LesschenJ. P. RoodT. WagnerS. De MarcoA. Murphy-BokernD. . (2014). Food choices, health and environment: effects of cutting Europe’s meat and dairy intake. Glob. Environ. Change 26, 196–205. doi: 10.1016/j.gloenvcha.2014.02.004 WillettW. RockströmJ. LokenB. SpringmannM. LangT. VermeulenS. . (2019). Food in the Anthropocene: the EAT–lancet commission on healthy diets from sustainable food systems. Lancet 393, 447–492. doi: 10.1016/S0140-6736(18)31788-4, 30660336 WWF International, 2022. The 2022 Living Planet Report. Available online at: https://livingplanet.panda.org/en-US/ (Accessed December 9, 2024).

Edited by: Idowu Oladele, Global Center on Adaptation, Netherlands

Reviewed by: Francesca Galli, University of Pisa, Italy

Thales Botelho De Sousa, Federal Institute of São Paulo, Brazil

We also searched PubMed, CAB, and Web of Science, but these added little value (under 150 documents pre-cleaning).