Legume crops are an important part of global food and nutritional security, sustainable agriculture and climate resilient farming systems. Although grain legumes are becoming more relevant, there is limited information about the global research landscape of grain legumes. The current study aimed to assess the intellectual structure, research productivity, collaboration profile and emerging thematic trends in legume research using a scientometric analysis based on the publications published in the Scopus database from 2016 to 2025. A total of 19,996 articles from 1,016 sources were analysed using bibliometrix and science-mapping approaches. The results showed a positive increase in legume research, 19,996 publications, with an annual growth rate of 16.92%. Soybean and chickpea emerged as the most extensively studied crops, as these crops are globally important for food, feed and sustainable agricultural systems. Rajeev K. Varshney and Yang Wenyu were recognized as the most prolific authors and Tamil Nadu Agricultural University, Iowa State University, ICAR-Indian Agricultural Research Institute and Punjab Agricultural University were identified as important institutional hubs. Legume Research, Frontiers in Plant Science and Agronomy were among a few highly productive journals. A shift in conventional breeding and genetic improvement research to sustainability, adaptation to climate change, precision agriculture, artificial intelligence, deep learning and microbiome-based approaches were identified in recent years through trend topic analysis. International collaboration accounted for 23.11% of publications, highlighting the growing globalization of legume research.
Background: Soybeans are primarily cultivated in open fields. In such environments, it is very difficult for experts to be deployed in real-time to diagnose pests. While deep learning offers promising diagnostic capabilities, standard convolutional neural networks often place heavy computational burdens on resource-constrained agricultural edge devices. Methods: In this paper, we propose an Advanced-ResNet-20 model that combines channel and spatial attention (CBAM) with the ResNet-20 structure. The model was evaluated on an independent stratified test partition (n = 641, seed 42) with data augmentations applied strictly on-the-fly to training batches only. Result: The model achieved a test accuracy of 89.70% (91.11% with test-time augmentation), representing a 5.14% p improvement over the retrained baseline ResNet-20 (84.56%). With only 309,541 parameters (1.18 MB in FP32) and 1.01 GFLOPs, the proposed architecture provides a practical lightweight solution suitable for prospective deployment on mobile and drone hardware.
Background: Low and unstable seed yield remains a major constraint on alfalfa (Medicago sativa L.) seed production in China, increasing production costs and limiting domestic seed supply. Methods: A four-factor, three-level L9(34) orthogonal array was used over two consecutive growing seasons to evaluate cultivar, cutting time, compound fertilizer rate and early-spring mulching time. The established field was divided into three adjacent, equal-area cultivar zones. Within each cultivar zone, the three L9 treatment combinations corresponding to that cultivar were randomized in three blocks, giving three replicate plots per treatment combination and 27 plots in total. The same established plots received the same assigned treatment combinations in 2024 and 2025 and were therefore followed repeatedly across years. The orthogonal array was used to estimate main effects only; interactions were not independently estimable. Within each year, factor-level differences were assessed separately by one-way ANOVA using replicate plot-level observations; the two years were analyzed separately. Seed yield, yield components, agronomic traits and treatment-level economic returns were evaluated in three alfalfa cultivars. Result: Among the nine tested treatment combinations, Treatment 2 (Zhongmu No. 10, Jingzhe mulching, branching-stage cutting and 450 kg ha-1 compound fertilizer) produced the highest observed seed yield in both years, reaching 549.7 kg ha-1 in 2024 and 529.6 kg ha-1 in 2025. Range analysis showed that fertilizer rate had the largest spread among factor-level means in both years, followed by cutting time and mulching time; this descriptive ranking does not quantify each factor’s contribution to total variation. Treatment 2 also gave the highest observed partial-budget return, with RMB 18,405.8 ha-1 in 2024 and RMB 17,635.4 ha-1 in 2025. The theoretical 2025 combination A3B2C2D2 identified by range analysis was not included in the L9 array and therefore was not field-tested. Overall, Treatment 2 was the highest-yielding and highest-return observed combination across the two years.
Background: Field pea (Pisum sativum L.) productivity is severely affected by terminal heat stress, particularly under late sown conditions. Therefore, understanding genetic variability for yield and its contributing traits is important for developing heat-tolerant and high-yielding genotypes. Methods: The experimental material comprised 55 genotypes, including 10 parental lines and 45 F1 hybrids developed through a half-diallel mating design. The material was evaluated under timely- and late-sown environments during two Rabi seasons using RBD with three replications. Eleven agronomic traits were recorded and the data were analyzed for genetic variability, heritability, correlation and path coefficients to identify traits association with seed yield. Result: The analysis of variance showed significant differences among the evaluated genotypes for all measured traits under both sowing conditions, confirming the existence of ample genetic variation. PCV were consistently greater than GCV, indicated that environmental factors influenced trait expression. Biological yield/plant, seed yield/plant, pods/plant and secondary branches/plant showed high heritability with substantial genetic advance, suggesting good prospects for selection. Seed yield/plant showed favourable associations with biological yield/plant, pods/plant, seeds/pod and seed index, whereas days to flowering and days to maturity were negatively associated with seed yield/plant. Path analysis highlighted biological yield/plant (timely sowing) and pods/plant (late sowing) as the direct contributors to seed yield/plant.
Background: Waterlogging is a major abiotic constraint limiting pigeon pea productivity, particularly in regions receiving high rainfall during early growth stages or in crop lands with poor drainage. Exposure to prolonged soil saturation during the early vegetative stage disrupts normal physiological processes, resulting in reduced plant growth, impaired metabolism, and yield losses. Deciphering of the adaptive responses associated with waterlogging stress is crucial for devising crop breeding and selection strategy for developing waterlogging-resilient pigeon pea cultivars. Methods: Thirty pigeon pea genotypes, comprised test genotypes as well as tolerant and susceptible checks, were evaluated under controlled waterlogging conditions during two consecutive growing seasons. Waterlogging stress was imposed for 24, 48 and 72 hours at 30 days after sowing while a parallel set of plants was maintained under normal conditions. Morphological responses, survival percentage, adventitious root formation and aerenchyma development were recorded. Physiological parameters viz. photosynthetic rate, stomatal conductance, transpiration rate, chlorophyll content, nitrogen balance index, flavonols content, anthocyanin content and NDVI were recorded using standard analytical procedures. The biochemical parameters namely, antioxidant activity of enzymes and soluble sugar concentration were also estimated using standard protocol to assess biochemical responses associated with waterlogging stress. Result: Considerable variation in waterlogging tolerance was observed among the evaluated genotypes. IPAC-79, ICPL-87091, ICPL-20241, ICP-2376, IPAC-212, IPAWL-22-7 and IPAWL-22-8 consistently exhibited superior tolerance by maintaining higher survival, improved photosynthetic performance, greater chlorophyll retention and enhanced antioxidant activity under stress. These genotypes also developed abundant adventitious roots and well-defined aerenchyma tissues, facilitating adaptation to oxygen-deficient conditions. Moderately tolerant genotypes displayed intermediate responses, whereas susceptible genotypes exhibited severe chlorosis, wilting, poor survival and substantial reductions in physiological and biochemical traits in response to prolonged waterlogging stress.
