Environmental, Climatic, and Production Perspectives in Agroforestry Research: Reassessing the Position of Rural Livelihoods Through Bibliometric Analysis

1. Introduction

Agroforestry is recognized as an integrated land-use strategy that deliberately combines woody vegetation with crops and/or livestock to generate ecological and economic interactions within the same production system [1]. Its importance extends beyond agricultural production, as tree-based production systems can contribute to carbon sequestration, biodiversity conservation, soil enrichment, water regulation, and other ecosystem services associated with multifunctional landscapes [2,3]. At broader spatial scales, tree cover on agricultural land also represents an important component of biomass carbon stocks, which strengthens the role of agroforestry in climate change mitigation and in the transition of agricultural landscapes toward more sustainable land uses [4]. These environmental functions are linked to rural development concerns, since agroforestry can also support food security, livelihood diversification, and landscape-scale synergies between agricultural and forestry agendas [5]. In this sense, agroforestry should not be understood only as an agronomic practice, but as a socio-ecological strategy within a framework of sustainable production, biodiversity conservation, climate resilience, and rural livelihoods.

The expansion of agroforestry research has been accompanied by a continuous diversification of its thematic structure. Bibliometric results indicate that, since 1990, agroforestry research has shifted from early emphases on intercropping, alley cropping, and multipurpose trees toward topics such as carbon sequestration, ecosystem services, and climate change, suggesting a broader and more heterogeneous scientific agenda[6]). Similarly, a systematic review focused on the Asia-Pacific region shows that agroforestry studies have expanded toward ecosystem services, plantation crop combinations, species habitats, biological control, genetic diversity, and climate- and biodiversity-related concerns, although it also reveals an uneven distribution of scientific evidence across services and geographical contexts [7]. However, this diversification has not necessarily produced a fully integrated knowledge structure. Evidence from high-income countries shows that agroforestry research has concentrated mainly on regulating ecosystem services, agricultural productivity, soil and water quality, carbon storage, and biodiversity outcomes, whereas evidence related to human well-being and policy interventions remains comparatively limited [8]. At the governance level, agroforestry implementation also appears to be affected by fragmented actors and institutional networks, in which conservation initiatives, agricultural actors, and local tree-management systems often operate through partially disconnected organizational structures [9]. Therefore, although agroforestry research has expanded toward environmental, productive, and socioeconomic concerns, its knowledge base appears to be organized through partially connected thematic streams.

From a conceptual perspective, this diversification can be organized around four analytical domains that are not mutually exclusive but reflect different research orientations within agroforestry studies. The biodiversity/conservation domain focuses on the role of agroforestry systems as managed habitats that can contribute to species conservation, ecological functions, and the provision of ecosystem services within agricultural landscapes [10,11]. The climate-carbon-soil domain centers on the capacity of tree-based systems to store carbon in biomass and soil, increase soil organic carbon stocks, and support climate change mitigation through changes in land-use management [12]. The production/silvopastoral/agronomic domain examines the biophysical interactions among trees, crops, pastures, and livestock, including competition, resource capture, forage dynamics, animal production, and productivity within mixed production systems [13,14]. Finally, the livelihoods/socioeconomic domain considers how agroforestry influences agricultural productivity, ecosystem services, household welfare, human well-being, and rural development, particularly in low- and middle-income contexts where smallholder systems remain central to land-use transitions [15].

Although rural livelihoods are frequently mentioned as part of the rationale for agroforestry, their conceptual position within the field remains less clear. Existing studies show that agroforestry can facilitate livelihoods through production diversification, food and income security, ecosystem-service incentives, and social-ecological resilience in smallholder systems, silvopastoral systems, coffee-based agroforestry systems, and home gardens [16,17,18,19]. However, the socioeconomic effects of agroforestry are not automatic, because adoption and livelihood outcomes depend on farmers’ perceptions, land tenure, market access, income sources, cooperative membership, technical assistance, policy recognition, and broader institutional conditions [20,21,22,23]. This indicates that rural livelihoods are not absent from agroforestry research, but they are often addressed through applied themes such as adoption, incentives, food security, income diversification, household welfare, and social-ecological resilience.

Existing bibliometric studies on agroforestry have mainly mapped publication trends, leading authors, institutions, countries, journals, collaboration networks, and broad thematic clusters. However, less attention has been given to the relative position of rural livelihoods within the conceptual structure of agroforestry research. The specific gap is that it remains insufficiently understood whether livelihood-related research forms an autonomous conceptual domain or appears as a less consolidated component embedded within socioecological, conservation-oriented, smallholder-related, or production-oriented research. This limitation restricts the capacity to evaluate how socioeconomic dimensions are positioned within agroforestry knowledge and whether they are structurally differentiated from the biodiversity/conservation, climate-carbon-soil, and production-oriented domains.

Based on this gap, the study addressed the following research questions: How has global agroforestry research evolved over time across environmental, climatic, production-oriented, and livelihood-related domains? How is the conceptual structure of agroforestry research organized, and what is the relative position of rural livelihoods within that structure? Do livelihood-related terms form an autonomous conceptual domain, or are they embedded within broader socioecological, conservation-oriented, smallholder-related, or production-oriented research? To answer these questions, this study evaluates the temporal evolution and conceptual structure of global agroforestry research, with particular attention to the positioning of rural livelihoods relative to biodiversity/conservation, climate-carbon-soil, and production-oriented domains. We hypothesized that agroforestry research is structurally organized around biodiversity/conservation, climate-carbon-soil, and production-oriented systems, whereas rural livelihoods are present within the literature but remain less autonomous within the conceptual structure of the field.

2. Materials and Methods

Bibliometric analysis was used to examine the temporal evolution, conceptual structure, and thematic organization of global agroforestry research [24,25]. This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [26]. No preregistered protocol was developed for this review. However, the methodological workflow was internally predefined through a structured procedure established prior to evidence selection.

2.1. Eligibility Criteria

This review included journal articles written in English and published as final documents between 1976 and 2025. Eligible records were required to address agroforestry in relation to at least one of the analytical domains considered in this study: biodiversity/conservation, climate-carbon-soil, production/silvopastoral-agronomic systems, or livelihoods/socioeconomic dimensions. Only peer-reviewed journal articles were retained. Conference papers, discussion papers, book chapters, editorials, notes, and other non-article document types were excluded.

2.2. Information Source and Search Strategy

Records were obtained from the Scopus database, which provides curated bibliographic metadata suitable for longitudinal analyses in bibliometric research [27,28]. The search covered records published between 1976 and 2025. The most recent search was executed on June 9, 2026. No additional databases, websites, registers, grey literature sources, or direct contact with authors were used to identify supplementary records.

A Boolean search string was designed to retrieve peer-reviewed journal articles aligned with agroforestry and at least one of the analytical domains considered in this study: biodiversity/conservation, climate-carbon-soil, production/silvopastoral-agronomic systems, or livelihoods/socioeconomic dimensions. The complete search strategy applied in Scopus was as follows: ( TITLE ( ( agroforest* OR silvopast* OR agrosilvopast* OR "alley cropping" OR "forest farming" OR "home garden*" OR "tree-crop system*" OR "tree crop system*" OR "multistrata system*" OR "shade-grown" OR "shade coffee" OR "cocoa agroforest*" OR "cacao agroforest*" ) ) AND TITLE-ABS-KEY ( ( biodivers* OR conservation OR "species richness" OR "species diversity" OR "ecosystem service*" OR habitat* OR restoration OR "climate change" OR climate OR carbon OR "carbon sequestration" OR "carbon stock*" OR biomass OR mitigation OR adaptation OR livestock OR cattle OR pasture* OR grazing OR silvopast* OR "animal production" OR "livestock production" OR productivity OR "crop yield" OR "soil fertility" OR livelihood* OR income OR poverty OR household* OR wellbeing OR "well-being" OR welfare OR "food security" OR "rural development" OR adoption OR smallholder* OR socioeconomic OR "socio-economic" ) ) ) AND ( EXCLUDE ( PUBYEAR , "2026" ) ) AND ( LIMIT-TO ( DOCTYPE , "ar" ) ) AND ( LIMIT-TO ( LANGUAGE , "English" ) ) AND ( LIMIT-TO ( SRCTYPE , "j" ) ) AND ( LIMIT-TO ( PUBSTAGE , "final" ) ).

The initial search identified 8,200 documents in Scopus. After restricting the time span to publications from 1976 to 2025, 7,758 records remained. The dataset was further refined by retaining only English-language journal articles published at the final publication stage. After applying all filters, the final corpus comprised 5,713 articles included in the bibliometric analysis (Figure 1).

2.3. Data Charting and Data Items

Bibliographic metadata were exported from Scopus and inspected to verify the avail- ability of the fields required for analysis, including titles, abstracts, Author Keywords, authors, affiliations, country information, publication year, document type, citation counts, and document identifiers. Derived analytical variables included standardized keywords, domain assignments, continents, country income groups, and MCA cluster membership. No additional information was requested from the authors, as the study re- lied exclusively on bibliographic metadata available in Scopus.

2.4. Critical Appraisal of Individual Sources of Evidence

A critical appraisal of individual sources of evidence was not conducted. This decision was consistent with the bibliometric scope of the study, which focused on the structure, evolution, and thematic organization of scientific production rather than on assessing the methodological quality or risk of bias of primary empirical studies.

2.5. Data Analysis

-

Bibliometric indicators

Bibliometric and statistical analyses were conducted in R using the Bibliometrix package, complemented with custom scripts for data cleaning, classification, statistical testing, and visualization [29]. Descriptive indicators included annual scientific production and citation trends. Annual values were grouped into historical periods to examine temporal changes. Differences among periods were evaluated using Kruskal-Wallis tests followed by Dunn’s post-hoc tests with adjusted p-values.

-

Keyword cleaning and semantic normalization

Before conducting the conceptual analyses, Author Keywords were standardized through a controlled semantic cleaning procedure. This procedure included the correction of repeated terms, spelling variants, singular and plural forms, hyphenation differences, and equivalent expressions. A stop-word list was used to remove non-discriminant terms, including generic bibliometric expressions, broad agroforestry-related labels, geographic names, taxonomic names, and terms that did not contribute to conceptual differentiation. A synonym list was then applied to harmonize equivalent expressions across the corpus. Conceptually related but non-equivalent terms were not merged, in order to preserve analytical resolution among domains such as biodiversity, conservation, restoration, ecosystem services, climate change, carbon sequestration, production systems, and rural livelihoods. The complete stop-word list and the controlled dictionary containing the synonym mappings are provided in Tables S1 and S2, respectively.

The cleaned Author Keywords field was used as the primary textual input for the conceptual analyses. This decision was made to prioritize author-defined conceptual content and reduce the influence of database-generated terms. Index Keywords were not used in the main Multiple Correspondence Analysis because they may include broader controlled-vocabulary terms that can distort the interpretation of conceptual proximity.

-

Domain classification

Records were classified into four non-exclusive analytical domains: biodiversity/conservation, climate-carbon-soil, production/silvopastoral-agronomic systems, and livelihoods/socioeconomic dimensions. Classification was based on a controlled dictionary applied to titles, abstracts, and Author Keywords after semantic normalization.

Because domain assignment was non-exclusive, a single article could be assigned to more than one domain. Domain frequencies and percentages were calculated using the total number of articles in the corpus, period, or category as the denominator, depending on the analysis. Therefore, percentages were not expected to sum to 100%. For the global comparison of domain contributions, each article was coded as a binary observation for each domain. Domain proportions were estimated with 95% binomial confidence intervals. Differences among domains were tested using Cochran’s Q test, followed by pairwise McNemar tests with adjusted p-values [30].

-

Conceptual structure and thematic evolution analysis

The conceptual structure of global agroforestry research was examined using Multiple Correspondence Analysis based on cleaned Author Keywords [31]. MCA was used to identify associations among recurrent keywords and to visualize the relative position of major conceptual groups within the bibliographic corpus. Terms retained for MCA were selected after semantic normalization and frequency filtering, ensuring that the resulting conceptual map represented recurrent and analytically informative keywords rather than isolated or excessively generic terms. The interpretation of MCA clusters was based on keyword proximity, cluster membership, centroid position, and the conceptual coherence of representative terms.

Thematic evolution across historical periods was analyzed using a Sankey mapping approach based on normalized keywords [32]. Prior to mapping, keywords were cleaned to remove duplicates, spelling variants, singular/plural inconsistencies, and non-informative terms, and synonym standardization was applied to harmonize equivalent expressions while preserving conceptually distinct terms. Taxonomic names were removed when they did not contribute to thematic differentiation. The Sankey map represented nodes as normalized keywords within each period and flows as thematic connections across consecutive periods, enabling the visualization of continuity, transformation, and emergence of themes.

-

Geographical and economic association analysis.

Associations between analytical domains and geographical or economic categories were evaluated using chi-square tests of independence. Standardized residuals were calculated to identify domain-category combinations that were overrepresented or underrepresented relative to expected frequencies [33]. Positive residuals indicated combinations with higher-than-expected representation, whereas negative residuals indicated lower-than-expected representation. Residual heatmaps were used to complement the chord diagrams and provide statistical support for the interpretation of geographical and economic asymmetries in agroforestry research. Geographical and economic patterns were assessed by linking domain-classified records with country affiliation data. Countries were assigned to continents and World Bank income groups [34].

When articles included affiliations from multiple countries, all represented countries were retained. Domain-continent and domain–income-level co-occurrence matrices were used to generate chord diagrams. Because domains were non-exclusive and all countries in multi-country affiliations were retained, these analyses were interpreted as affiliation-based research presence rather than mutually exclusive article counts. Associations between analytical domains and geographical or economic categories were evaluated using chi-square tests of independence. Standardized residuals were calculated to identify combinations with higher- or lower-than-expected representation [33] and were displayed in residual heatmaps to support the interpretation of geographical and economic asymmetries.

2.6. Synthesis of Results

Results were synthesized according to the research question and specific objectives of the study. First, temporal trends in publication output, citation patterns, and domain participation were summarized across historical periods. Second, the conceptual structure of agroforestry research was examined through cleaned Author Keywords, Multiple Correspondence Analysis, and thematic evolution. Particular attention was given to the position of rural livelihood-related terms in relation to biodiversity/conservation, climate-carbon-soil, and production-oriented domains. Third, geographical and economic distributions of research domains were assessed using domain-continent and domain-income-level matrices, chord diagrams, chi-square tests, and standardized residual heatmaps. This analytical sequence allowed the study to connect temporal growth, conceptual organization, and spatial asymmetries in global agroforestry research.

3. Results

3.1. Temporal Evolution of Agroforestry Research

-

Annual scientific production and citation trends

Global agroforestry research showed a temporal increase between 1979 and 2025 (Figure 2). Research output was distributed across four periods: Period I (1979-1989), with 100 articles; Period II (1990-1999), with 428 articles; Period III (2000-2012), with 1,050 articles; and Period IV (2013-2025), with 4,133 articles. Annual scientific production increased from an average of 9.09 articles per year in Period I to 42.8 in Period II, 80.8 in Period III, and 318 articles per year in Period IV. The annual publication trajectory followed an increasing polynomial trend, particularly after 2013. The fitted model for annual articles indicated a sustained expansion of agroforestry research output during the most recent period (Figure 2A). Citation counts also increased over time. Annual citations were lowest during Period I and increased in subsequent periods, reaching their maximum value during Period IV (Figure 2B). Differences among historical periods were statistically significant for both annual articles and annual citations.

The Kruskal-Wallis’s test indicated significant period-level differences in annual scientific production (H = 42.3, df = 3, p = 3.53 × 10⁻⁹) and annual citations (H = 38.3, df = 3, p = 2.39 × 10⁻⁸). Dunn’s post-hoc comparisons showed that annual scientific production differed significantly between Period I and Period III, and that Period IV differed significantly from all previous periods. For annual citations, significant differences were only observed between Period I and Period III, Period I and Period IV, and Period II and Period IV.

-

Temporal contribution of analytical domains

The temporal contribution of analytical domains varied across historical periods. During Period I (1979-1989), production/silvopastoral/agronomic systems represented the largest proportion of classified articles, accounting for 70% of the period-specific corpus. Climate-carbon-soil research accounted for 47%, followed by livelihoods/socioeconomic dimensions with 44%, whereas biodiversity/conservation showed the lowest contribution with 19%.

A similar pattern was observed in Period II (1990-1999), where production/silvopastoral/agronomic systems remained the dominant domain, reaching 72.7% of articles. Climate-carbon-soil research represented 44.9%, livelihoods/socioeconomic dimensions accounted for 36.7%, and biodiversity/conservation showed a moderate increase to 21.7%. In Period III (2000-2012), production-oriented research still showed the highest contribution, although its relative proportion declined to 61.4%. Climate-carbon-soil research increased slightly to 48.9%, while biodiversity/conservation rose to 35.9%. Livelihoods/socioeconomic dimensions remained stable, representing 37.0% of articles during this period.

Period IV (2013-2025) showed a more balanced distribution among domains. Climate-carbon-soil research reached the highest contribution, with 57.3% of articles, followed by production/silvopastoral/agronomic systems with 54.9% and biodiversity/conservation with 48.9%. Livelihoods/socioeconomic dimensions also increased, reaching 43.6% of the period-specific corpus. Across the full temporal sequence, biodiversity/conservation showed the strongest relative increase, rising from 19% in Period I to 48.9% in Period IV. Production/silvopastoral/agronomic systems remained important, although their relative contribution decreased after Period II. Climate-carbon-soil research increased in the most recent period, while livelihoods/socioeconomic dimensions showed a stable trajectory, with a moderate increase in Period IV.

-

Thematic evolution across periods

The Sankey map showed changes in the composition and continuity of recurrent author keywords across the four historical periods (Figure 3). During Period I (1979-1989), the thematic structure was represented by a small set of terms, mainly intercropping, alley cropping, home gardens, soil fertility, land tenure, and mixed farming. These keywords indicated an early concentration of agroforestry research around production systems, farm-level management, and soil-related attributes.

In Period II (1990-1999), the keyword structure became more diversified. Home gardens, soil fertility, alley cropping, intercropping, silvopastoral systems, competition, Leucaena leucocephala, crop yield, and conservation appeared as recurrent terms. Several links connected the early production-oriented terms from Period I with intercropping, alley cropping, and soil fertility in Period II. At the same time, silvopastoral systems and conservation emerged as additional thematic components.

During Period III (2000-2012), the thematic map incorporated a broader set of terms, including soil fertility, alley cropping, competition, biodiversity, adoption, silvopastoral systems, Grevillea robusta, and nitrogen. Soil fertility, alley cropping, competition, and silvopastoral systems maintained visible links with terms from previous periods. Biodiversity and adoption also appeared in this period, indicating the incorporation of ecological and socioeconomic terms into the temporal keyword structure.

In Period IV (2013-2025), the recurrent terms included carbon sequestration, silvopastoral systems, climate change, and biodiversity. Silvopastoral systems showed continuity from Period II and Period III into the most recent period. Biodiversity also persisted from Period III to Period IV, while carbon sequestration and climate change appeared as relevant terms in the final period.

3.2. Conceptual Structure of Agroforestry Research

-

MCA based on cleaned Author Keywords

The Multiple Correspondence Analysis (MCA) based on cleaned Author Keywords revealed a differentiated conceptual structure in global agroforestry research (Figure 4). The first two dimensions separated three main keyword groups associated with production/silvopastoral/agronomic systems, socioecological and conservation-oriented research, and carbon/biomass/soil-related research. Dimension 1 mainly differentiated the socioecological-conservation space from the production and carbon-soil groups, whereas Dimension 2 separated the production-oriented cluster from the carbon-biomass-soil cluster.

The production/silvopastoral/agronomic cluster was positioned in the upper-left area of the MCA. This cluster grouped keywords related to productive systems, crop-livestock interactions, agronomic performance, and farm-level management. It included terms such as silvopastoral systems, pasture, grazing, livestock, yield, alley cropping, intercropping, competition, forage, soil fertility, soil properties, microclimate, maize, and shading.

The carbon/biomass/soil cluster was located in the lower-left area of the map. This cluster corresponded to a distinct biophysical axis associated with carbon storage, biomass accumulation, and soil carbon dynamics. It was represented by carbon sequestration, carbon stocks, soil organic carbon, biomass, and climate change mitigation.

The largest cluster occupied the central and right areas of the MCA space and grouped socioecological, conservation, and livelihood-related terms. This cluster included terms such as biodiversity, conservation, ecosystem services, restoration, species diversity, and agroecology. Rural livelihoods were positioned within this broader socioecological-conservation cluster, close to terms related to smallholders, food security, home gardens, ecosystem services, and local knowledge.

The relative contribution of analytical domains showed uneven representation across the global agroforestry research corpus (Figure 5). Climate-carbon-soil research showed the highest contribution, with 59.1% of articles assigned to this domain (n = 3,377). Production/silvopastoral/agronomic systems represented the second largest domain, accounting for 53.1% of the corpus (n = 3,034). Biodiversity/conservation accounted for 44.7% of articles (n = 2,553), whereas livelihoods/socioeconomic dimensions showed the lowest relative contribution, with 42.2% of articles (n = 2,408).

The comparison of domain proportions showed significant differences among the four categories (Cochran’s Q = 390.87, df = 3, p = 2.11 × 10⁻⁸⁴). Pairwise McNemar tests with Holm-adjusted p-values indicated significant differences among all domain pairs. The climate-carbon-soil domain showed a higher proportional contribution than biodiversity/conservation, production/silvopastoral-agronomic systems, and livelihoods/socioeconomic dimensions. In turn, production/silvopastoral-agronomic systems showed a higher proportion than biodiversity/conservation and livelihoods/socioeconomic dimensions, whereas biodiversity/conservation recorded a significantly higher proportion than livelihoods/socioeconomic dimensions. Although the livelihoods/socioeconomic domain was present in more than two-fifths of the corpus, it showed the lowest relative contribution among the four domains.

-

Position of rural livelihoods within the conceptual structure

To better characterize the conceptual organization identified by the MCA, the keyword clusters were summarized according to their centroid position, representative terms, interpreted domain, and the location of rural livelihoods within the MCA space (Table 1). Three conceptual clusters were identified. The first was associated with production, silvopastoral systems, and agronomic management; the second grouped socioecological, conservation, and livelihood-related terms; and the third corresponded to a compact carbon/biomass/soil domain.

Rural livelihoods appeared only within the socioecological/conservation cluster, together with terms related to smallholders, food security, home gardens, biodiversity, ecosystem services, and local knowledge. This result indicates that rural livelihoods were present in the conceptual structure, but within a broader socioecological-conservation domain, rather than positioned as an independent cluster in the MCA.

3.3. Geographical and Economic Distribution of Research Domains

-

Distribution of analytical domains across continents

The distribution of analytical domains across continents indicated geographical variation (Figure 6). The chord diagram showed that all four research domains were represented across continents, although the strength of domain-continent associations varied among regions. Europe, Asia, Latin America and the Caribbean, Africa, and North America showed visible connections with the main analytical domains, whereas Oceania contributed less to the overall domain-continent structure (Figure 6A).

Standardized residuals from the chi-square analysis showed patterns of overrepresentation and underrepresentation across continents (Figure 6B). The biodiversity/conservation domain was overrepresented in Europe (standardized residual = 5.21) and underrepresented in Oceania (-3.63). Climate-carbon-soil research was overrepresented in Asia (4.63), whereas it was underrepresented in Europe (-2.76). Production/silvopastoral/agronomic systems were strongly overrepresented in Latin America and the Caribbean (11.27), with additional positive residuals in Oceania (3.09) and North America (2.48). In contrast, this domain was underrepresented in Africa (-6.11) and Asia (-6.63).

Livelihoods/socioeconomic dimensions showed the highest positive residual in Africa (8.79), followed by Asia (2.51). The same domain was underrepresented in Latin America and the Caribbean (-9.89), whereas residuals for Europe, North America, and Oceania were close to zero or slightly negative. The residual structure indicated that geographical differences were not uniformly distributed across domains: The strongest positive residuals were observed for livelihoods/socioeconomic research in Africa, climate-carbon-soil research in Asia, biodiversity/conservation in Europe, and production/silvopastoral/agronomic systems in Latin America and the Caribbean.

-

Distribution of analytical domains across country income levels

The distribution of domains according to country income level showed differences in the associations between economic income categories and thematic areas of research. The chord diagram showed that the four domains were related to the income levels, although the strength of these associations varied among high-income, upper-middle-income, lower-middle-income, and low-income countries (Figure 7A).

Standardized residuals from the chi-square analysis showed patterns of overrepresentation and underrepresentation (Figure 7B). In high-income countries, the biodiversity/conservation domain showed slight overrepresentation (standardized residual = 2.12), whereas climate-carbon-soil research was underrepresented (-2.21). The remaining domains showed representation close to the expected values.

In upper-middle-income countries, production/silvopastoral/agronomic systems were overrepresented (4.25), whereas the livelihoods/socioeconomic domain showed strong underrepresentation (-7.45). In this context, agroforestry research was more strongly associated with productive and agronomic approaches than with perspectives centered on rural livelihoods.

In lower-middle-income countries, the livelihoods/socioeconomic domain showed clear overrepresentation (6.05), whereas biodiversity/conservation (-4.49) and production/silvopastoral/agronomic systems (-3.12) were underrepresented. In low-income countries, the livelihoods/socioeconomic domain also showed strong overrepresentation (6.32), whereas production/silvopastoral/agronomic systems showed marked underrepresentation (-5.49).

From a rural livelihood’s perspective, this pattern indicated that agroforestry research related to livelihoods/socioeconomic dimensions was mainly concentrated in lower-middle-income and low-income contexts. In these income groups, the domain was more strongly associated with topics such as smallholders, food security, income, rural welfare, adoption, and local development. In contrast, the underrepresentation of livelihoods/socioeconomic dimensions in upper-middle-income countries indicated that, in these contexts, the domain distribution showed stronger representation of productive, silvopastoral, and agronomic systems than livelihoods/socioeconomic dimensions.

4. Discussion

4.1. Expansion and Thematic Reorientation of Agroforestry Research

The results show that global agroforestry research experienced sustained expansion from 1979 to 2025, with a marked increase during the 2013-2025 period. This pattern indicates not only sustained growth in publication output, but also a progressive broadening of agroforestry research toward more diverse agendas. In the early periods, the thematic structure was more closely associated with productive and farm-level management practices, such as intercropping, alley cropping, home gardens, soil fertility, and mixed farming. In contrast, the most recent periods incorporated terms more strongly associated with biodiversity, adoption, silvopastoral systems, carbon sequestration, and climate change. This shift suggests that agroforestry ceased to be represented mainly as an agronomic practice and began to occupy a broader position within debates on sustainability, climate resilience, ecosystem services, and socioecological systems.

This reorientation is consistent with recent studies that have described a broadening of the agroforestry field toward more integrative approaches. For example, the systematic review by Shin [7] on agroforestry and ecosystem services in the Asia-Pacific region shows that recent literature has shifted part of its attention toward multiple ecological functions, environmental benefits, and ecosystem service assessment. Similarly, Gallo [35] through a bibliometric analysis of agroforestry, adaptation, and resilience in agricultural landscapes, show that current discussion has moved toward frameworks in which agroforestry is interpreted as a strategy for responding to climatic pressures and land-use change. This interpretation is consistent with our results, in which climate change, carbon sequestration, and biodiversity appeared as more visible terms in the most recent scientific literature.

Nevertheless, this diversification does not imply a complete replacement of the initial production-oriented approaches. Rather, the results suggest that silvopastoral systems and agronomic practices remain relevant components, while climate, biodiversity, and ecosystem service agendas have gained greater presence in recent literature. This inter- pretation is important because it avoids reading the evolution of the field as a linear tran- sition from production toward sustainability. The evidence points to a more complex re- organization: production-oriented approaches persist, but they now coexist with broader climatic, ecological, and socioecological frameworks. Recent studies on climate-smart ag roforestry, such as that of Ntawuruhunga [36] illustrate this convergence by integrating spatial characterization, agroecological conditions, and climate adaptation into the analysis of agroforestry systems.

In this context, the temporal expansion observed in this study supports the idea that agroforestry has become consolidated as an interdisciplinary research field. However, this expansion also produced heterogeneous domains, in which some thematic areas appear more clearly consolidated than others. The expansion of scientific literature, therefore, should be understood as a process of thematic diversification, but also as a potential source of internal asymmetries within its knowledge domains.

4.2. Consolidation of Climate-Carbon-Soil and Production-Oriented Domains

The results indicate that the climate-carbon-soil and production/silvopastoral-agronomic systems domains are two of the most consolidated components within the global agroforestry literature. This consolidation is reflected not only in their high contribution to the corpus, but also in the way both domains were differentiated within the conceptual structure. The climate-carbon-soil axis was associated with terms such as carbon sequestration, carbon stocks, soil organic carbon, biomass, and climate change mitigation, whereas the production-oriented axis was organized around silvopastoral systems, pasture, grazing, livestock, yield, forage, soil fertility, and agronomic practices. This separation indicates that both domains respond to distinct scientific questions: the former is oriented toward biogeochemical and climatic functions, whereas the latter is more closely related to productive performance, agronomic management, and system efficiency.

The strong organization of the climate-carbon-soil domain can be partly explained by the availability of quantifiable and comparable indicators across systems, regions, and management conditions. Recent literature has accumulated evidence through meta-anal- yses and comparative studies focused on soil organic carbon, biomass, carbon storage, and climate change mitigation [12,37,38].

However, the strength of the climate–carbon–soil axis should not be interpreted as an exclusively environmental agenda. Part of its relevance derives from its capacity to articulate carbon, soil, functional biodiversity, productivity, and sustainability [39,40] . These studies allow a more nuanced interpretation of this domain, showing that although the carbon–soil cluster appears conceptually differentiated, its practical implications often extend toward productivity, food security, income, and local resilience.

The second consolidated domain corresponds to productive, silvopastoral, and agronomic systems. This research line maintains continuity with the early phases of the field. In recent literature, production-oriented systems are linked to agronomic efficiency, sustainable intensification, animal welfare, pasture management, crop yield, and livestock system performance. Amadu [41] for example, showed that agroforestry can act as a pathway to positive agricultural yield impacts within climate-smart agriculture investments in Malawi. In livestock systems, Lemes [42] demonstrated that silvopastoral systems can improve thermal comfort, animal welfare, and productive performance in beef production systems. Similarly, Smith [43] analyzed benefits, challenges, and management activities reported by producers and natural resource professionals in silvopastoral systems in the United States.

The scientific literature on tropical agroforestry systems further reinforces the connection between productivity, management, and environmental functions. Niether [44] in a multidimensional meta-analysis, compared cocoa agroforestry systems with monocultures and showed that the evaluation of these systems requires simultaneous consideration of yield, carbon, biodiversity, microclimate, and productive diversification. Suárez [45] in turn, found that cacao agroforestry systems can improve soil fertility compared with degraded pastures in the Colombian Amazon. These studies are relevant because they show that the production-oriented domain is not limited to yield or primary production; rather, it also incorporates soil properties, tree-cover management, and effects on the ecological quality of the system. Nevertheless, this conceptual structure remains differentiated from the climate-carbon-soil axis because its central questions are focused on agronomic performance, productive viability, and system management.

Overall, the findings suggest that the climate-carbon-soil and production/silvopastoral-agronomic systems domains are more consolidated, possibly because of the greater methodological traceability of their variables. Carbon stocks, soil organic carbon, biomass, yield, pasture productivity, animal performance, and soil fertility are variables that support comparisons, meta-analyses, and quantitative assessments. By contrast, the lower thematic consolidation of socioeconomic dimensions does not relate to their relevance, but rather to how their indicators are conceptualized and measured. Whereas biophysical and productive domains tend to be organized around more standardized variables, rural livelihoods depend on more context-specific indicators, such as adoption, income, food security, wellbeing, local knowledge, and household strategies.

4.3. Biodiversity/Conservation as a Bridge Between Ecological and Socioecological Agroforestry Research

The results show that the biodiversity/conservation domain occupies a particular position in agroforestry research. Unlike the climate-carbon-soil and production/silvopastoral-agronomic systems domains, which are organized around more compact and measurable variables, biodiversity/conservation functions as a space of connection between ecological processes, ecosystem services, and the social dimensions of rural systems. In the MCA, this domain did not appear as an isolated ecological category, but rather as part of a broader socioecological cluster associated with ecosystem services, restoration, species diversity, agroecology, home gardens, smallholders, food security, and local knowledge. These results suggest that biodiversity can be interpreted as a conceptual bridge linking conservation, production, local management, and rural wellbeing.

Recent scientific literature supports the view that biodiversity in agroforestry systems is no longer discussed only in terms of species richness, habitat conservation, or the maintenance of biological communities. Rather, it is increasingly analyzed as a functional basis for ecosystem services, landscape resilience, ecological restoration, and the sustainability of diversified agricultural systems [46,47]

From this perspective, biodiversity/conservation should be interpreted as a domain with an articulating function. Its position within the socioecological cluster indicates that it connects ecological dimensions with local practices, diversified production, restoration, food security, and territorial knowledge. This interpretation helps explain why rural livelihoods appear close to terms such as smallholders, home gardens, food security, ecosystem services, and local knowledge. Livelihoods are not absent from the conceptual structure, but their visibility is mediated by broader socioecological categories. Thus, biodiversity/conservation acts as a bridge that integrates ecological functions with rural systems, while at the same time it may contribute to livelihood-related dimensions being incorporated within conservation and ecosystem-service approaches rather than differentiated as an autonomous conceptual domain.

4.4. Rural Livelihoods as a Visible but Less Autonomous Component of Agroforestry Knowledge

The results of this study indicate that rural livelihoods are not absent from the agroforestry literature, but their conceptual position is not autonomous when compared with the climate-carbon-soil and production/silvopastoral-agronomic systems domains. The livelihoods/socioeconomic domain was present in a substantial proportion of the corpus, but it recorded the lowest relative contribution among the four domains and did not form an independent cluster in the MCA. Instead, livelihood-related terms appeared within the socioecological/conservation cluster, close to smallholders, food security, home gardens, ecosystem services, adoption, and local knowledge. This proximity suggests that research on rural livelihoods is mediated by categories linked to conservation, adoption, smallholders, and rural development.

This conceptual structure can be explained by the multidimensional and context-dependent nature of livelihoods. Unlike variables such as carbon stocks, soil organic carbon, biomass, yield, or pasture productivity, rural livelihoods are not expressed through a single standardized indicator. Their assessment usually involves income, food security, household wellbeing, poverty, access to resources, productive strategies, local knowledge, farmer perceptions, institutional capacity, and market conditions. Miller [15] and Castle [48] similarly showed that evidence on agroforestry extends across agricultural productivity, ecosystem services, and human wellbeing, indicating that socioeconomic objectives are frequently integrated with environmental and productive dimensions.

Adoption represents one of the main pathways through which socioeconomic dimensions are incorporated into agroforestry research. Decisions to adopt agroforestry practices depend on interacting with technical, economic, institutional, and household-level conditions, particularly when benefits are delayed or uncertain. Tranchina et al. [49], for example, identified technical–agronomic, socioeconomic, and policy-legislative barriers that constrain agroforestry adoption. Evidence also indicates that agroforestry may support food security, income diversification, and household resilience, although these outcomes vary according to local resources, management strategies, and institutional contexts [19,20].

4.5. Geographical and Economic Asymmetries in Agroforestry Research Priorities

The results show that agroforestry research priorities are unevenly distributed across regions and income groups. Biodiversity/conservation, climate-carbon-soil, production/silvopastoral-agronomic systems, and livelihoods/socioeconomic dimensions exhibited differentiated geographical and economic patterns.

The positive association of livelihoods/socioeconomics with Africa and Asia, as well as with lower-middle-income and low-income countries, indicates that research on livelihoods tends to gain greater visibility in contexts where agroforestry is directly linked to rural poverty, food security, dependence on natural resources, technological adoption, and household resilience. In sub-Saharan Africa, this orientation is reflected in studies addressing land tenure, gender arrangements, institutional conditions, and access to resources [50]. In Asia, agroforestry has similarly been examined as a strategy for combining environmental benefits, productive diversification, and livelihood improvement [51]. However, the region was also positively associated with climate-carbon-soil research, consistent with geospatial assessments linking agroforestry potential to productivity, food security, and climate resilience in South Asia [20].

A different pattern was observed for production/silvopastoral–agronomic systems, which were overrepresented in Latin America and the Caribbean and in upper-middle- income countries. In these contexts, agroforestry research is more frequently structured around livestock management, perennial crops, silvopastoral practices, agronomic efficiency, scaling, and economic viability. This orientation is illustrated by research on traditional livestock systems in Mexico [52]. Evidence from the Ecuadorian Amazon further shows that production-oriented systems may simultaneously integrate governance, environmental integrity, economic resilience, social wellbeing, car- bon conservation, tree diversity, and rural livelihoods [53,54].

Income-level patterns provide a complementary interpretation. In high-income countries, the stronger association with biodiversity/conservation may reflect greater emphasis on restoration, ecosystem services, multifunctional landscapes, and agroecological transitions. In upper-middle-income countries, the predominance of production-oriented re-search and the underrepresentation of livelihoods/socioeconomic dimensions suggest that social issues may be addressed indirectly through commercial, productive, or management frameworks. Conversely, in lower-middle-income and low-income countries, agroforestry is more frequently studied as a response to rural vulnerability and dependence on smallholder systems. Evidence from Ethiopia also links agroforestry with climate-related livelihood resilience and household income [55,56].

These geographical and economic asymmetries help explain why rural livelihoods are visible but less conceptually autonomous within the global literature. Livelihood-related research is concentrated in contexts where socioeconomic vulnerability and small- holder dependence are particularly prominent, whereas in other regions it may be incorporated into conservation, climate, or production-oriented framework.

4.6. Knowledge Integration and Rural Development Implications

The results of this study show that agroforestry research has moved toward greater thematic diversification, but with an imbalance in the integration of its domains. The climate-carbon-soil and production/silvopastoral-agronomic systems domains appear more consolidated, whereas biodiversity/conservation functions as a socioecological bridge and rural livelihoods appears as a less autonomous dimension. This structure of thematic organization has important implications for agroforestry research and rural development agendas. If agroforestry is promoted as a strategy capable of integrating climate change mitigation, biodiversity conservation, sustainable production, and rural wellbeing, then scientific literature needs to move from the coexistence of domains toward analytical frameworks that assess their interdependencies, without reducing livelihoods to a consequence of environmental or productive benefits.

A first implication is the need to strengthen integrative approaches capable of simultaneously analyzing benefits, trade-offs, and synergies among thematic domains. McDonald [57] demonstrated that carbon benefits and local ecosystem services may operate either as synergies or as trade-offs. Complementarily, Mathieu [58] showed that agroforestry can improve the provision of ecosystem services and biodiversity.

A second implication relates to governance and the landscape approach. Integration among conservation, production, and rural development does not depend only on measuring more variables, but also on connecting actors, institutions, and decision-making scales. Zinngrebe [9] argued that agroforestry governance can be organized through a landscape approach by connecting conservation and agricultural actors. Therefore, a knowledge integration agenda should consider not only biophysical and productive variables, but also institutional arrangements, land tenure, market access, and territorial governance.

A third implication concerns research with the adoption and scaling of agroforestry practices. The lower conceptual autonomy of rural livelihoods suggests that future studies should avoid treating the adoption of agroforestry practices as an exclusively technical outcome. Tranchina [49], through a global systematic review, identified technical, socioeconomic, and policy-legislative barriers that hinder agroforestry adoption. Consequently, agroforestry-based rural development should not be formulated only as the transfer of practices or technologies, but as a situated process that depends on local capacities, economic incentives, social organization, and institutional stability.

Studies conducted in Ecuador offer useful examples of this need for integration. Castañeda-Ccori [59], in their analysis of cacao agroforestry systems in Amazonian communities, showed that the sustainability of these systems depends on the interaction among institutional conditions, land access, community organization, commercial production, and food self-sufficiency. Complementarily, Torres [53] showed that governance, environmental integrity, economic resilience, social wellbeing, and livelihood capitals should be analyzed jointly.

From a rural development perspective, our results suggest that agroforestry research

should strengthen the measurement of social and economic outcomes without separating them from environmental and productive domains. Indicators of income, food security, wellbeing, equity, and household resilience should be incorporated with the same conceptual clarity as carbon, biomass, soil, or yield. Ntawuruhunga [36], in their evaluation of climate-smart agroforestry practices in Rwanda, showed that effects on income and food security can be analyzed together with agroecological conditions and productive strategies.

This study should be interpreted within the scope of its bibliometric design. The analysis focused on Scopus-indexed, English-language journal articles published as final documents and therefore does not capture the full diversity of local, non-English, technical, policy-oriented, or grey literature on agroforestry and rural livelihoods. Similarly, the use of a controlled dictionary and cleaned Author Keywords provided a reproducible basis for identifying analytical domains and conceptual structures, although keyword-based approaches represent the visible semantic organization of the corpus rather than the complete content of each article. The geographical and income-level analyses were based on author affiliation metadata; therefore, they indicate patterns of research presence and institutional participation rather than mutually exclusive study sites.

5. Conclusions

This bibliometric analysis of 5,713 Scopus-indexed journal articles showed that global agroforestry research expanded substantially between 1976 and 2025 and became thematically more diversified over time. The results indicate that climate-carbon-soil research and production/silvopastoral-agronomic systems are the most consolidated domains, while biodiversity/conservation operates as a socioecological bridge connecting ecological functions, ecosystem services, local management, and rural dimensions. Rural livelihoods were clearly visible within the corpus, but they did not form an autonomous conceptual domain; instead, they appeared embedded within a broader socioecological/conservation cluster associated with smallholders, food security, home gardens, ecosystem services, adoption, and local knowledge. Geographical and income-level patterns further showed that livelihoods/socioeconomic research was more strongly represented in Africa, Asia, and lower-middle- and low-income contexts, whereas other domains followed different regional and economic distributions. The findings indicate that the main issue is not the absence of rural livelihoods in agroforestry research, but their limited conceptual autonomy within a field more structurally organized around environmental, climate-related, and production-oriented perspectives.