Mapping the global knowledge structure of artificial intelligence applications in horticultural production: A bibliometric analysis

Artificial intelligence (AI) is increasingly important for precision horticulture, yet research on its applications remains fragmented across crops, technologies, and production challenges. This study uses bibliometric analysis to map the recent (2020–2025) knowledge structure of AI applications in horticultural production and assess Africa's position within it. Bibliographic records were retrieved from the Dimensions database using artificial-intelligence terms (“artificial intelligence,” “machine learning,” “deep learning,” and “computer vision”) combined with horticulture-related terms, yielding 7519 documents across 1070 sources. Because only 6.8% of these records contain terms from both query blocks in their title or abstract, all headline indicators were also computed for a validated subset of 726 documents and for two further corpus definitions. Output grew at a compound annual rate of 54.3% between 2020 and 2025, with a sharp rise in 2024 and slower growth in 2025, driven primarily by China and India, followed by the United States, Italy, and Australia. Agronomy, Frontiers in Plant Science, and Agriculture were the most productive sources, while China Agricultural University and Northwest Agriculture and Forestry University led institutional output; authorship is highly dispersed, with no author exceeding 18 documents. Co-word analysis of title and abstract terms separated an image-based detection and diagnostics theme from crop production, management, and sustainability themes, although finer cluster boundaries were not stable across runs. International co-authorship accounted for 27.93% of publications. African countries accounted for 8.09% of country–document occurrences (6.93% under fractional counting; 6.4–8.4% across corpus definitions), with Egypt, Nigeria, South Africa, Ethiopia, and Morocco as the leading contributors, while 16 African countries were absent from the dataset. This share exceeds Africa's share of global scientific publications (about 3.5%) but is well below its population share (about 16–18%); African-affiliated documents are present in the highly cited core and are cited at least as often as other documents, and their co-authorship links extend to Asia, the Middle East, Europe, and North America. The findings indicate that Africa's gap lies mainly in leadership, institutional depth, and scale rather than absence from the field, and they support targeted funding, local datasets, interdisciplinary capacity building, and stronger South–South research collaboration to advance AI-enabled horticulture across the continent.