Smallholder farmers in developing countries generate rich, structured agricultural knowledge through social media platforms every day — crop diagnostic descriptions, peer-validated price information, technique queries, and market intelligence — without knowing they are doing so, without owning what they produce, and without governance frameworks that protect their rights over this data or ensure they benefit from its potential use in artificial intelligence systems. This paper examines farmer-generated social media data from an original survey of 720 smallholder farmers across all 20 administrative blocks of Prayagraj district, Uttar Pradesh, India, and applies the Responsible AI and Data Justice frameworks to evaluate the ethical conditions under which AI agricultural advisory systems could legitimately build on this data. We find that the data farmers generate constitutes a structurally valuable resource for training localised AI advisory systems — including labelled diagnostic image-text pairs, real-time hyperlocal price signals, and agricultural information demand data — but that current governance conditions are systematically inadequate: genuine informed consent is absent, community data ownership does not exist, language representation is ignored, and algorithmic accountability mechanisms are non-existent. We argue that the choice for agricultural AI development in the Global South is not between using farmer-generated data or not, but between using it extractively and using it responsibly. Realising responsible AI agricultural advisory requires not technical solutions but governance ones: community data trusts, participatory consent mechanisms, language justice requirements, algorithmic bias auditing, and public institutional architecture that aligns AI development with farming community interests rather than technology corporate extraction.
Buckwheat ( Fagopyrum sp.) is a nutritionally rich and climate-resilient crop; however, its genetic potential for consistent performance under variable environmental conditions is still underexploited. To address this gap, the present study assessed genotype-by-environment (G×E) interaction effects on seed yield across 15 buckwheat genotypes evaluated in three distinct environments using AMMI and GGE biplot models. Combined ANOVA revealed significant contributions of genotype (G), environment (E), and their interaction (G×E) to yield variability, indicating strong differential responses among genotypes. AMMI analysis identified G5, G15, and G14 as the highest-yielding genotypes, while G8, G10, G9, and G7 exhibited remarkable stability across environments. The GGE biplot, which accounted for 95.9% of total variation through the first two principal components, provided clear insights into genotype performance and interaction patterns. The “which-won-where” analysis delineated three mega-environments, with G5 dominating in E1, G13 in E2, and G15 in E3. Among the test sites, E3 emerged as the most discriminating and representative environment for genotype evaluation. Considering mean performance, genotypes G15, G5, and G14 were superior yielders, while G14, G8, and G7 demonstrated broad adaptability and high stability. Overall, G15 was identified as the ideal genotype for favourable conditions, whereas G14 combined high yield with wide environmental adaptability. These results offer valuable genetic resources and insights for developing resilient buckwheat varieties suited to diverse and challenging agro-climatic conditions.
Sowing time, irrigation scheduling and nitrogen management jointly determine wheat productivity, and crop simulation models such as DSSAT-CERES-Wheat offer a route to extend two-year field results to a wider range of conditions. This study evaluated the pooled (2019-20 and 2020-21) effect of these three factors on growth, yield and profitability of wheat, and calibrated/validated the DSSAT-CERES-Wheat model (v4.7) for the Vindhyan agro-climatic zone. A split-plot field trial with three replications was conducted over two rabi seasons at the Crop Research Farm, SHUATS, Prayagraj, with two sowing dates (20 November, 20 December), three irrigation schedules and three nitrogen levels (80, 120, 160 kg N ha⁻¹) as eighteen treatment combinations. First-year data were used to calibrate genetic coefficients of cultivar HD 2967, and second-year data were used for independent validation of phenology, leaf area index (LAI) and yield. On a pooled basis, 20 November sowing, four irrigations (CRI, tillering, booting, milking) and 160 kg N ha⁻¹ produced the tallest plants, highest dry matter, LAI and yield attributes, with grain yield of 3.44 and biological yield of 7.94 t ha⁻¹ and the highest net return and benefit-cost ratio, the D₁I₃N₃ combination giving the best economics. The calibrated model predicted days to anthesis and maturity within 1–2 days of observation (nRMSE < 1.5%, d > 0.98) and reproduced LAI closely (nRMSE 7.8% calibration, 6.2% validation; d > 0.95). Grain and biological yields were simulated with r² of 0.92–0.96 in calibration and 0.93–0.95 in validation, confirming good-to-excellent model accuracy. Timely sowing, adequate irrigation and higher nitrogen consistently improved wheat productivity and profitability, and the DSSAT-CERES-Wheat model, once calibrated for cultivar HD 2967, reliably reproduced phenology, LAI and grain yield for the Vindhyan Zone, supporting its use as a decision-support tool for regional wheat management.
Understanding extreme precipitation intensification under progressive global warming is essential for flood risk assessment and climate adaptation planning in vulnerable monsoon-dependent basins. This study investigates projected changes in very wet day precipitation (R95p—annual sum of daily precipitation exceeding the 95th percentile) across three global warming levels (1.5°C, 2°C, and 3°C) in the Rapti River Basin using four CMIP6 climate models: ACCESS-CM2, ACCESS-ESM1-5, EC-EARTH3, and MPI-ESM1-2-HR. Employing the Expert Team on Climate Change Detection and Indices (ETCCDI) framework, we analyze spatial distributions, temporal trends, and inter-model variability in extreme precipitation contributions under Paris Agreement-aligned temperature scenarios. Results reveal pronounced intensification of very wet day precipitation across all warming levels, with ACCESS-ESM1-5 projecting the most substantial R95p increases among the model ensemble. This model's enhanced sensitivity—likely reflecting interactive biogeochemical feedbacks and vegetation-climate interactions—produces R95p values that consistently exceed other models by 15–30% across warming thresholds. At 1.5°C warming, R95p contributions range from 800-2,200 mm across the basin, escalating to 900-2,600 mm at 2°C and 1,000–2,800 mm at 3°C in ACCESS-ESM1-5 projections. Trend analysis identifies 2°C as a critical threshold where statistically significant intensification emerges in three of four models, marking a transition from natural variability dominance to detectable forced climate signals. Striking spatial heterogeneity characterises all projections—northern orographic regions experience R95p contributions 2–3 times higher than southern plains, with disparities widening under higher warming scenarios. The disproportionate contribution of very wet days to total annual precipitation increases progressively, suggesting fundamental regime shifts toward more extreme-dominated precipitation patterns. An increasing trend in R95p was noted across most models and warming levels, indicating systematic intensification of the most intense precipitation events. These findings underscore urgent requirements for enhanced flood protection infrastructure, updated hydrological design standards, and adaptive water management strategies capable of accommodating increasingly concentrated precipitation delivery. The multi-model analysis provides critical evidence demonstrating that limiting warming to 1.5°C versus 3°C produces substantially different extreme precipitation futures with profound implications for flood risk, agricultural water management, and infrastructure resilience in the Rapti River Basin.
Understanding the spatial distribution of salinity and plant species is crucial to manage and maintain sustainability and to reclaim areas affected by salinity. In this study, we modeled the global distribution of soil salinity using inverse distance weighted (IDW) interpolation on the WOSiS salinity database. The predicted global map was used to extract salinity values and confidence intervals at species locations, through which 243 species were predicted in median locations affected by high to severe salinity. The Amaranthaceae family contributed the highest number of salt-tolerant species with 30 species (upto 12%), followed by Poaceae (21 species), Fabaceae (20 species), with 191 out of the 243 listed species having some salt tolerance trait and 96 having root microbiome interaction. Species such as Atriplex portulacoides, Eryngium maritimum, Dittrichia viscosa, Phragmites australis, and Anagallis arvensis have high chances of establishment in such harsh environments, given the presence of salt-tolerant traits and root microbiome interaction assisting in nutrient availability. We conclude that spatial interpolation techniques can be a useful tool to shortlist species for the economic bioremediation and reclamation of highly saline areas around the world, which is in correspondence with the goals set by the UN-Decade of Ecosystem Restoration.
