Agroforestry is widely promoted as a socioecological strategy that can integrate biodiversity conservation, climate change mitigation, sustainable production, and rural development, yet the relative position of rural livelihoods within the scientific structure of the field remains insufficiently understood. This study evaluated the temporal evolution, conceptual organization, and geographical and economic distribution of global agroforestry research, with particular attention to the autonomy of livelihood-related research. A bibliometric analysis was conducted using 5,713 Scopus-indexed journal articles published between 1976 and 2025. Records were classified into four non-exclusive analytical domains: biodiversity/conservation, climate-carbon-soil, production/silvopastoral-agronomic systems, and livelihoods/socioeconomic dimensions. The analysis combined temporal bibliometric indicators, semantic normalization of Author Keywords, Multiple Correspondence Analysis, Sankey mapping, domain proportion tests, chord diagrams, and chi-square tests. Agroforestry research expanded substantially and became thematically more diversified. Climate-carbon-soil and production/silvopastoral-agronomic systems showed the highest proportional contributions, whereas rural livelihoods were visible but did not form an autonomous conceptual domain. Instead, livelihood-related terms were embedded within a broader socioecological/conservation cluster. Livelihoods/socioeconomic research was more strongly represented in Africa, Asia, and lower-middle- and low-income contexts. These findings indicate that rural livelihoods are present in agroforestry research but remain less conceptually autonomous than environmental, climate-related, and production-oriented domains.