Grape leaf diseases threaten viticulture by reducing yield and quality, while manual scouting remains inconsistent and can drive unnecessary chemical use. Although deep learning enables automated diagnosis, standard models trained on curated datasets are computationally heavy and prone to overconfident errors on out-of-distribution (OOD) vineyard imagery. We present VitiNet, a lightweight, OOD-aware grape leaf disease classifier built on a ResNet18 backbone and optimized through staged fine-tuning, an engineered ‘Other’ class for explicit OOD rejection, and a risk-mitigating dual-threshold deployment protocol. Evaluated on a challenging, OOD-skewed test set, VitiNet achieved 98.2% accuracy and a 0.982 weighted F1-score, with exceptional OOD performance (99.6% precision for the ‘Other’ class). The primary limitation of this model was found to be the confusion between Bacterial_rot and OOD images (Bacterial_rot F1 = 0.807). Grad-CAM analyses showed that the model focuses on symptomatic regions but can be overly sensitive to necrotic textures that resemble Bacterial_rot in non-disease images. To ensure safe and reliable operation, we suggested a dual-threshold policy for deployment: T-disease = 0.96 (chosen to achieve approximately 95% recall for disease detection) and T-other = 0.99, directing ambiguous predictions to human review. VitiNet delivers fast, accurate, and non-invasive decision support for grape growers and provides a robust foundation for future domain adaptation with real-world field data, including eventual on-device deployment for in-field use.

Enhancing photosynthetic efficiency via biostimulant application could offer a sustainable pathway to improve alfalfa forage yield, quality traits and resilience against unfavourable environmental conditions. This current study was conducted at the University of Debrecen to evaluate the impact of two biostimulants on gas exchange parameters and chlorophyll fluorescence in alfalfa under field conditions. The experiment was arranged in a randomized complete block design with three treatment levels, namely, Tricho Immun (T1) (21.6 g/3 L), a biostimulant containing Si, MTU^®, and pidolic acid (T2) (2 ml/3 L) and a control (T3) (no biostimulant). Parameters measured include net photosynthetic rate (A), transpiration rate (E), intercellular CO₂ concentration (Ci), ambient CO₂ concentration (Ca), water use efficiency (WUE) and quantum yield of PSII (ΦPSII). Data gathered were subjected to analysis of variance using Genstat statistical software edition 18, where significant means were separated at a 5% probability level using the least significant difference, and DMRT was used to rank the performance of the individual treatment on measured traits. The results show that biostimulant application significantly improved net photosynthetic rate, transpiration rate, intercellular CO₂ concentration, water use efficiency, and quantum yield of PSII, indicating improved photochemical efficiency. However, no significant changes were observed in ambient CO₂ concentration. Our findings indicated that T2 significantly increased ΦPSII by 43.9%, net photosynthetic rate by 37.8%, and transpiration rate by 97.6%, while T1 also increased ΦPSII, net photosynthetic rate, and transpiration rate by 11.7%, 9.8%, and 39.5%, respectively, compared to the control (T3). These results suggest that biostimulants can enhance photosynthetic performance and resource-use efficiency in alfalfa under field conditions, offering a promising approach for sustainable forage production.

Barley (Hordeum vulgare L.) is a crop of vital importance for human and animal food security in Algeria. However, in the high plateau and mountainous regions, its yields are severely affected by low-temperature stress and winter frosts.

To address this limitation, a controlled laboratory experiment was conducted to evaluate the cold tolerance of three local barley genotypes (Saida, Tihert, and Oued el Maleh) and to assess the potential role of salicylic acid (SA)—a phytohormone known for its involvement in abiotic stress tolerance—in enhancing plant resilience. Barley seeds were germinated under low-temperature conditions (0–4 °C) for 10 days. Morphological and physiological parameters, such as germination rate and root length, were measured and statistically analyzed using Statistica software.

Significant genotypic differences were observed. In the absence of SA, the Tihert genotype achieved a germination rate of 65%, compared with 20% for Saida and 5% for Oued el Maleh. The application of SA (0.5 mM) markedly enhanced post-germination growth and resilience in the sensitive genotypes, increasing germination rates to 55% for Saida and 25% for Oued el Maleh, while its effect remained limited in Tihert, which already exhibited high natural tolerance.

Hierarchical clustering analysis based on Euclidean distances confirmed Tihert as the most cold-tolerant genotype. The observed improvement in the sensitive genotypes (Saida and Oued el Maleh) suggests that SA may activate antioxidant defense systems and membrane-stabilizing mechanisms.

These findings highlight the importance of genetic diversity among local barley varieties and demonstrate the potential of salicylic acid as a physiological inducer of cold tolerance. Genotypes such as Tihert represent valuable genetic resources for breeding programs aimed at developing barley varieties better adapted to Algeria’s harsh climatic conditions.

Abstract:

Introduction: Early detection of plant stress is critical for sustainable agriculture. Conventional techniques, including threshold-based sensors and image recognition, face limitations such as environmental interference, high costs, and dependency on visual data. To address these challenges, we propose a novel sensor-to-text approach that leverages natural language processing (NLP) for plant health monitoring.

Methods: A dataset containing soil properties, environmental variables, and nutrient concentrations was prepared with labels (Healthy, Moderate Stress, High Stress). A rule-based algorithm generated descriptive symptom statements from sensor readings (e.g., “Soil is too dry, risk of dehydration”). These narratives were processed using two pipelines: (i) TF-IDF with Random Forest and Support Vector Machine classifiers, and (ii) a transformer-based DistilBERT model fine-tuned for multi-class classification.

Results: Baseline models with TF-IDF and traditional classifiers achieved 85–88% accuracy, with an average F1-score of 0.86. In comparison, the DistilBERT model significantly outperformed them, reaching 95% accuracy, with precision 0.94, recall 0.95, and F1-score 0.94. The transformer approach also showed particular strength in distinguishing subtle differences between moderate and high stress conditions.

Conclusions: This study presents a novel pipeline that converts agricultural sensor data into natural language descriptions for classification using transformer-based NLP. The results highlight the potential of this method to improve plant disease detection, provide interpretable feedback, and support scalable AI-driven advisory systems for farmers.

Fruit identification and location technology plays a significant role in modern agriculture, especially in smart agriculture and automated picking systems. Due to factors such as irregular fruit distribution, branch and leaf obstruction, and environmental light, traditional manual or ground positioning methods face many challenges in large orchards. Through current research on the phased achievements of fruit recognition and positioning based on unmanned aerial vehicles, the key technologies of fruit recognition and positioning are deeply analyzed.

Fruit identification and location technology is a key link in smart orchards. The positioning technology based on unmanned aerial vehicles (UAVs) significantly improves the positioning accuracy in complex environments through high flexibility and multi-sensor fusion. Research shows that machine vision and multi-sensor fusion models have made progress in the positioning of fruits such as citrus and strawberries, thanks to the three-dimensional spatial flexibility of multi-rotor platforms, the efficiency of cluster operations, and their adaptability to orchards. However, this technology still faces many technical bottlenecks in practical applications: the complex orchard environment has a significant impact on the flight stability of unmanned aerial vehicles and the accuracy of sensors, and the humid climate also shortens the service life of the equipment. The positioning accuracy and real-time performance of existing algorithms are difficult to balance in scenarios with severe fruit occlusion and dense overlap, and there is a contradiction between lightweight models and high-precision requirements. Meanwhile, the equipment cost is high, the cluster control and data fusion technology is still not mature, and the standardization degree of orchard planting is insufficient. Enhancing environmental adaptability through multi-sensor fusion and intelligent algorithm optimization, developing lightweight and modular recognition models, ensuring positioning efficiency, and promoting the standardization of orchard planting will be the key directions for breakthroughs in fruit positioning technology based on unmanned aerial vehicle (UAV) platforms.

This study examined the economic viability of various colored shade nets on mango production (Mangifera indica L. cv. Keitt) in terms of climatic conditions, growth and productivity, nutrient status, and water use efficiency. The study investigated the effects of different screen net colors on plant growth over two seasons (2023 and 2024), focusing on minimum and maximum air temperatures and relative humidity under various screen net colors compared to open-field conditions. All shade nets resulted in elevated relative humidity and lower maximum temperatures. The black screen net and open-field treatments produced the lowest mango yields. In contrast, the use of white or yellow screen nets significantly increased both the number of fruits per plant and overall yield per plant compared to other treatments. The results indicate that white and yellow nets provided the most favorable environment for mango cultivation in Egypt, with yields of 3663 kg/2000 m² and 2997 kg/2000 m², respectively, in the first season. The black greenhouse net cover produced the lowest yield, at 1998 kg/2000 m², in the first season. Over the two seasons, the open-field treatment had the lowest macro- and micronutrient content. Furthermore, various screen net covers demonstrated higher water use efficiency throughout both seasons, with white screen nets outperforming yellow screen nets.

Lettuce is a leafy vegetable consumed fresh or processed in ready-to-eat products, supporting lifestyle trends favouring quick, healthy, and convenient meals. Morphological traits impact yield, processing efficiency, shelf life, and product quality. Biofertilisers are gaining attention as eco-friendly, sustainable alternatives that enhance soil fertility, increase microflora diversity, stimulate plant growth, and reduce pollution in all agricultural systems. This study aimed to investigate the effect of biofertilisers on lettuce morphology traits for fresh consumption and processing. Six lettuce cultivars ('Kiribati', 'Murai', 'Aquino', 'Gaugin', 'Aleppo', and 'Carmesi') were cultivated over three consecutive seasons (autumn, winter, and spring) under greenhouse conditions. Experiments were conducted in two distinct soil types: Mollic Gleysol (Calcaric)-GL and Hortic Anthrosol (Terric, Transportic)-AT. Two biofertilisers, EM Aktiv and Vital Tricho, were applied alone and combined to the soil and foliar via a battery sprayer. Green cultivars, especially 'Kiribati' and 'Aleppo', had the highest levels in most morphological parameters, while red 'Gaugin' had the highest core ratio (stem to rosette height) in both soils. In GL soil, all biofertilisers increased rosette diameter, with Vital Tricho boosting dry leaf weight, and EM Aktiv and combined fertilisers raising core ratio. In contrast, in AT soil, all fertilisers significantly reduced fresh leaf weight and stem diameter, while Vital Tricho and combined treatment lowered rosette diameter and core ratio. Generally, morphological traits were higher in winter and spring trials than in autumn. Pearson correlation coefficients showed that fresh leaf weight was positively linked to all other traits in both soils. An increased core ratio negatively impacts processing quality; thus, maintaining it below 0.5 is essential. Although fertiliser did not influence all parameters, the interaction among all factors was significant, except for dry leaf weight in both soils, indicating the need to optimise biofertiliser application according to soil type.

A robotic system that efficiently uses water and pesticides for tomato cultivation is not widely available in sustainable agriculture practice. This study proposes a design for a new AI-powered robot that automates pesticide and irrigation applications on small- to medium-sized tomato farms. The system operates based on water and pesticide requirements, turning on weekly to follow a predefined zigzag path in the field to reach every plant. The robot uses ultrasonic sensors and small reflective markers on the plant support sticks to accurately identify each tomato plant. The system then takes a picture of the plant and runs the data through a pretrained YOLOv8-based disease classifier, which helps find plants that need treatment. The robot skips plants that are healthy, which saves pesticides, resources, and the environment. The robot sprays pesticides on the affected area and, simultaneously, applies irrigation water. The irrigation water demand is calculated based on crop water needs and the amount of water that evaporates from the air, ensuring efficient water use. The system hasn’t been tested in the field yet, but an early estimate of the development cost is USD 300, suggesting a low initial price for small- to medium-sized farms. The combination of plant-level monitoring, selective treatment, and combined irrigation will help lower input costs, have less impact on the environment, and enable farmers to use more sustainable methods.

Fusarium wilt caused by Fusarium oxysporum f. sp. cubense represents one of the most destructive banana diseases worldwide. With tropical race 4 (TR4) threatening global production and limited resistant varieties available, innovative biocontrol strategies are essential. Following promising in vitro results, this study evaluated the protective efficacy of chitosan-based nanocarriers loaded with Rubia tinctorum and Uncaria tomentosa extracts under controlled greenhouse conditions.

Banana plantlets were treated with nanocarrier formulations and subsequently inoculated with Foc. Disease severity was assessed using a 0-4 scale over 79 days, and the area under disease progress curve (AUDPC) was calculated. Additionally, physiological parameters including stomatal conductance, transpiration rate, and chlorophyll fluorescence were monitored using porometry and fluorometry.

Results demonstrated significant disease reduction with R. tinctorum-loaded nanocarriers, achieving 34.9% control efficacy compared to untreated infected plants (AUDPC: 134.5 vs 206.6). This treatment showed statistically significant differences in disease severity from 31 days post-inoculation onwards, with final severity scores of 2.6 compared to 3.6 in infected controls. In contrast, U. tomentosa formulations provided limited protection (7.8% efficacy, AUDPC: 190.5) with no significant differences from the infected control. Physiological analyses revealed enhanced stomatal conductance and transpiration rates in R. tinctorum-treated plants. Growth parameters demonstrated significant improvements in R. tinctorum-treated plants, including 10-15% increases in leaf width, enhanced leaf length and fresh weight (16% increase), alongside 36-43% higher chlorophyll content compared to controls. DUALEX measurements confirmed elevated flavonoid levels (621.2 vs 318.1) and anthocyanin content (231.0 vs 168.4) in R. tinctorum-treated plants, indicating enhanced stress tolerance.

These findings validate the transition from laboratory to greenhouse conditions, demonstrating that R. tinctorum-based nanocarriers offer substantial disease suppression potential. The integration of nanotechnology with natural bioactive compounds presents a promising sustainable approach for Fusarium wilt management, contributing to the development of environmentally friendly alternatives for banana crop protection in commercial production systems.

Introduction: Plant biostimulants are pivotal to developing more sustainable agricultural systems, enhancing crop performance, and reducing reliance on synthetic inputs. Among them, botanical extracts remain underexplored, especially those derived from wild plant species. In this study, Silene inflata, a spontaneous herbaceous plant typical of Mediterranean grasslands, was investigated for its biostimulant potential on lettuce.

Methods: Leaves and roots of S. inflata were collected from greenhouse-grown plants and stored at - 80°C. Both tissues were biochemically characterized for polyphenols, ascorbate, proteins, saponins, chlorophylls, and carotenoids. Aqueous phosphate-buffered extracts from leaves (LES) and roots (RES) were prepared via ultrasound-assisted green extraction. A pilot study showed that irrigation was more effective than foliar spraying, and this method was used in the main experiment. Three-week-old Lactuca sativa L. (cv. Canasta) seedlings were transplanted into 10 cm pots and grown in a greenhouse under natural light (15 h day/9 h night; 7-25 °C). Plants were irrigated daily with 20 mL of tap water. On treatment days, 10 mL of tap water was applied, followed 1 h later by 10 mL of LES or RES at 1 or 10 mL/L once per week for three weeks. Controls received water or PBS.

Seven days after the last treatment, plants were harvested for fresh/dry weight measurements and biochemical analyses.

Results: RES showed high saponin levels, while LES had five times more polyphenols. Ascorbate and soluble proteins were higher in leaves (17 mg/100 g FW and 0.5 mg/g FW, respectively). LES at 10 mL/L significantly improved biomass, while LES at 1 mL/L enhanced chlorophylls, carotenoids, and polyphenols. Protein content was reduced in LES 10-treated plants. Overall, LES had stronger biostimulant effects in a dose-dependent manner. Additionally, S. inflata was successfully cultured in vitro, suggesting its suitability for sustainable biomass production.

Conclusions: These findings reveal S. inflata as a promising biostimulant, highlighting the value of wild plant biodiversity for sustainable agricultural innovation.