Journal of the Science of Food and Agriculture最新文献

Background: Accurate and real-time detection of tea leaf buds is a fundamental requirement for intelligent tea harvesting and smart agriculture. However, achieving high detection accuracy for small targets under complex tea plantation environments remains challenging, particularly for deployment on resource-constrained devices. Existing object detection models often suffer from excessive computational complexity or insufficient performance when applied to such scenarios. Therefore, it is necessary to develop a lightweight detection framework that balances detection accuracy and computational efficiency.

Results: To address these challenges, this study proposes a lightweight object detection model named TeaBudLiteNet. The model introduces a novel C2f_PConv module, which integrates the computational efficiency of PConv with the non-linear feature representation capability of the C2f module, achieving an effective trade-off between accuracy and efficiency. In addition, the SimAM_Slice attention mechanism is incorporated to enhance feature weighting across different scales, thereby improving small target detection. The Focaler-IoU_Inner regression loss function is further employed to dynamically optimize sample importance and accelerate model convergence, enhancing generalization and adaptability. Experimental results demonstrate that TeaBudLiteNet outperforms mainstream detection models in terms of accuracy, model size and inference speed. Specifically, the model achieves a precision of 90.47%, representing an improvement of 2.08% over the baseline. The parameter count is reduced to 1 912 947, achieving more than a 90% reduction in model size compared to conventional approaches, at the same time as maintaining a high inference speed of 227 frames per second.

Conclusion: The proposed TeaBudLiteNet effectively reduces model complexity at the same time as preserving high detection accuracy and real-time performance. Its lightweight architecture and superior efficiency make it highly suitable for deployment on resource-limited smart agricultural devices. By providing an efficient and robust solution for tea leaf bud detection in complex environments, this study demonstrates significant potential for automating tea harvesting and contributes to the advancement of intelligent agricultural systems. © 2026 Society of Chemical Industry.

Background: During the lyophilization (i.e., freeze-drying) of probiotics, the addition of a protective agent is essential for the survival of the frozen bacterial cells. Although many studies have explored the effects of exogenous protective agents during freeze-drying, whether these protective agents play an intracellular role requires study. Sorbitol is one of the most commonly used protective agents. In Lactiplantibacillus plantarum, srlD1/srlD2 is a key gene for sorbitol synthesis, and the pts-srlD gene cluster controls sorbitol transport. The objective of this study was to analyze the effect of sorbitol as an endogenous protective agent during the lyophilization of L. plantarum. Intracellular sorbitol levels were successfully controlled by the overexpression and knockout of sorbitol synthesis (srlD) or transport (pts-srlD) genes. Meanwhile, knockout of ldh will shift the metabolism of Fru-6P to the synthesis of sorbitol.

Results: In phosphate-buffered saline alone, the survival rate of the AR113Δldh-srlD2 strain, which had a large amount of intracellular sorbitol, reached 41.9% - a rate 3.2 times higher than that of AR113-srlD2. The addition of sorbitol decreased the survival of transport system-knockout strains compared with the wild-type strain, and the transport system-restored strain had a survival rate 3.1 times higher than that of the wild-type strain.

Conclusion: These results show that sorbitol plays an intracellular role in L. plantarum, and exogenous sorbitol is transported into the cell, where it exerts a protective effect. This study provides a theoretical basis for the application of protective agents. © 2026 Society of Chemical Industry.

The practical applications of water-in-oil-in-water (W/O/W) double emulsions are limited by their instability. Whey protein isolate-flaxseed gum (WPI-FSG) conjugates have demonstrated excellent emulsifying capabilities in previous studies; however, their effectiveness in stabilizing double emulsions remains largely unexplored. To address this knowledge gap, a W/O/W double emulsion co-loaded with epigallocatechin gallate (EGCG) and quercetin was developed using a WPI-FSG conjugate as the stabilizer, and its stability was evaluated systematically under thermal, pH, and ionic strength stresses. RESULTS: A W₁/O/W₂ double emulsion co-loaded with hydrophobic quercetin and hydrophilic EGCG was stabilized by incorporating WPI-FSG conjugates into both aqueous phases simultaneously. The addition of 10 g L^-1 conjugate to the inner aqueous phase (W1) reduced the primary emulsion droplet size by 58.3%. Increasing the conjugate concentration to 50 g L^-1 in the outer aqueous phase (W2) raised the encapsulation efficiency of quercetin to 74.14% and that of EGCG to 96.15%. Bioaccessibility increased by 0.6-fold for quercetin and by 4.86-fold for EGCG. The use of the WPI-FSG conjugate resulted in a higher encapsulation rate and greater reagent loading than the addition of the WPI-FSG mixture under ambient conditions (pH 4.0-10.0, NaCl concentration = 0-200 mM; storage temperature = 4-40 °C for 7 days; heating temperature = 25-100 °C). CONCLUSION: The WPI-FSG conjugates showed significant potential for improving the stability and co-delivery performance of W₁/O/W₂ double emulsions. Their simultaneous use in the internal and external aqueous phases provides a robust strategy for encapsulating hydrophilic and hydrophobic bioactives, offering valuable insights for the development of functional food and pharmaceutical formulations. © 2026 Society of Chemical Industry.

Background: As a rapid detection method, the electronic nose exhibits enormous potential in food quality monitoring. However, the response data generated by electronic nose detection are highly complex time-series data. Traditional data analysis models struggle to fully resolve such long-sequence non-linear signals, leading to insufficient feature extraction and poor prediction accuracy.

Results: We propose a novel total volatile basic nitrogen (TVB-N) prediction method based on the attention improved encoder long short-term memory (AIE-LSTM) hybrid network with a dual-stream feature fusion architecture. A modified informer encoder captures temporal dependencies via multi-head attention, and a temporal down-sampling layer compresses sequence dimensions at the same time as preserving key trend features. In addition, a bidirectional LSTM network processes raw voltage sequences and manually designed physical features separately. Finally, a multi-level feature fusion mechanism integrates these two types of features through fully connected layers to output predictions. Experimental results demonstrated that the AIE-LSTM model achieved the optimal TVB-N prediction performance across nine batches of electronic nose datasets, with an average coefficient of determination (R²) of 0.960, as well as the lowest relative standard deviation of R², root mean square error (RMSE) and mean absolute error (MAE). Notably, the model exhibited the best performance in the fourth batch, where the R², RMSE and MAE between the predicted and actual TVB-N values reached 0.979, 0.988 and 0.589, respectively.

Conclusion: This model significantly improves the prediction accuracy of fish TVB-N content, providing a reliable technical solution for the rapid, non-destructive and accurate detection of fish freshness. © 2026 Society of Chemical Industry.

Background: Wheat germination increases α-amylase activity, leading to significant biochemical and structural changes that influence flour rheological properties. Understanding the correlation between internal kernel structure and flour rheology is essential for predicting processing performance and optimizing the use of germinated wheat in food applications.

Results: In this study, five selected Canada Western Red Spring (CWRS) wheat cultivars were germinated for up to 36 h. Germinated wheat kernels were used for microstructural analysis and whole wheat flour was used for rheological tests and enzyme assays. Germination induced structural degradation across all varieties, including crease widening and increased porosity after 24 h, with more pronounced changes at 36 h. α-Amylase activity increased significantly after 24 h, resulting in lower Falling Number and apparent viscosity values. Extended germination (36 h) also weakened gluten aggregation. Correlation analysis revealed strong links between microstructural and rheological properties, with outer layer thickness having the greatest influence. Total porosity was strongly negatively correlated with breakdown (r = -0.64) and peak maximum time (r = -0.64). Outer layer thickness showed even stronger negative correlations with peak maximum time (r = -0.84), peak viscosity (r = -0.82), and breakdown (r = -0.82).

Conclusion: These results showed that extended germination altered wheat kernel structure and functionality, as evidenced by increased porosity, enhanced α-amylase activity, and reduced gluten aggregation capacity and flour viscosity. Among the structural features, outer layer thickness emerged as a key factor influencing flour rheology. © 2026 The Author(s). Journal of the Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Background: Almond hulls (AH) are the main by-product of almond processing for human consumption and contain bioactive compounds that can improve meat quality. Although AH are used as feed for dairy cows in some countries, information on their potential effects on meat quality is limited. This study evaluated the effects of partly replacing conventional feeds with AH in the concentrate of light lambs on carcass traits and meat quality.

Results: Thirty Manchega lambs (15 females and 15 males) were divided into three homogenous groups according to body weight and sex, and each was fed a concentrate containing 0, 60 or 120 g AH kg ^-1. Lambs were slaughtered at approximately 23.0 kg of body weight and carcass traits, chemical composition, pH and fatty acid (FA) profile of meat, and changes in color and lipid oxidation of meat over 6 days storage were analyzed. Inclusion of AH in the concentrate did not affect either carcass weight and conformation or meat pH and chemical composition. However, feeding AH significantly improved the meat FA profile by increasing (P < 0.05) its polyunsaturated FA (PUFA) content, which may be related to modifications of ruminal FA biohydrogenation. No significant effects of AH on meat color or lipid oxidation over the storage period were observed. Sex-related differences were minimal, but males showed higher PUFA content and lower intramuscular fat than females.

Conclusion: Replacing conventional feeds with up to 120 g AH kg ^-1 in the concentrate of light lambs can enhance the FA profile of their meat without compromising carcass characteristics or meat composition. © 2026 The Author(s). Journal of the Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Background: Italy harbors one of the richest grapevine biodiversities worldwide, yet the sensory identity of wines from many native cultivars remains poorly defined despite their relevance on the market at regional, national, or international levels. This study provides a systematic sensory characterization of 18 Italian monovarietal white wines, analyzed across 246 commercial samples, including wines never investigated before by sensory analysis. Analysis of variance, hierarchical cluster analysis, and principal component analysis were applied to Rate-All-That-Apply (RATA) data performed by trained panelists to define a lexicon and identify the sensory attributes characterizing and discriminating the 18 wine types.

Results: A statistically based lexicon comprising 29 olfactory and seven taste/mouthfeel descriptors was defined. Multivariate statistics showed that the 18 monovarietal wine types belong to four main olfactory dimensions, labeled as fruity-balsamic, thiolic-mineral, floral-sweet, and toasty-dried. A three-dimensional space was defined along the four olfactory directions. Müller Thurgau, Gewürztraminer, Albana, and Falanghina emerged as the most representative wines in these directions, outlining the vertices of a spatial framework within and around which the other wines are distributed. Sensory wheels representing structured visual synthesis of the most relevant attributes (odor, taste, mouthfeel) were developed as 'identity models', providing systematic tools for defining wines' varietal typicality.

Conclusion: Results are relevant for enologists and wine sellers as standardized reference sensory models for production and communication, as benchmarks for PDO/PGI quality control and disciplinary improvement, for researchers to advance modeling of wine sensory quality through sensometabolomic wine studies. Results also support the international recognition and valorization of Italian grapevine biodiversity according to Agenda 2030 sustainable development goals. © 2026 The Author(s). Journal of the Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Background: Antibiotic-induced gut dysbiosis poses significant challenges to microbial homeostasis, necessitating effective prebiotic interventions. Given the increasing interest in dietary polysaccharides for modulating microbial imbalance, this study systematically investigates the prebiotic potential of native tamarind seed polysaccharide (NTSP) and its enzymatic hydrolysates (ETSP1, ETSP2) in restoring clindamycin-disrupted intestinal microbiota in mice, with a focus on the impact of molecular weight on structure-activity relationships.

Results: Enzymatic depolymerization selectively reduced the molecular weight (M_w from 5.36 × 10⁵ to 4.05 × 10⁴ g mol^-1) and enhanced chain rigidity while preserving the galactoxyloglucan backbone, as confirmed by monosaccharide composition, nuclear magnetic resonance, and high-performance size-exclusion chromatographic analyses. In vivo, both NTSP and ETSPs ameliorated clindamycin-induced intestinal dysbiosis via suppression of pathogenic genera (e.g., Escherichia-Shigella, Klebsiella) and enrichment of beneficial taxa. Notably, the low-M_w ETSP2 preferentially promoted Lactobacillus and Paludicola, whereas moderate-M_w ETSP1 enhanced Bacteroides, Flavonifractor, and unclassified_f_Lachnospiraceae, and significantly increased short-chain fatty acid production, particularly of acetic acid and valeric acid, as quantified by gas chromatography-mass spectrometry.

Conclusion: These findings highlight the critical role of molecular weight in determining prebiotic efficacy, offering insights into the rational design of structure-function optimized polysaccharide-based therapeutics to combat antibiotic-associated dysbiosis. © 2026 Society of Chemical Industry.

Background: Quinoa is widely recognised for the high biological value of its proteins. However, its outer layer contains saponins, an antinutrient that must be removed for consumption. The process of dry desaponification of grains is known as scarification and produces a residual powder that is usually discarded.

Results: The quinoa scarification residue (QSR) contained 5 g kg^-1 of saponins. Thus, washing with water, homogenisation-assisted washing (HQSR) and sonication-assisted washing (SQSR) methods were evaluated. The assisted washing methods used 50% less water, although all of them reduced the residual saponin content (~0.5 g kg^-1) and significantly concentrated the protein content of the samples (~270-290 g kg^-1), compared to the control sample (~210 g kg^-1). The electrophoretic profile in sodium dodecyl sulfate-polyacrylamide gel electrophoresis showed the presence of the structural protein chenopodin, with its acidic (30-40 kDa) and basic (20-25 kDa) subunits, as well as protein macromolecules higher than 100 kDa. Fragmented particles were also observed in the HQSR and SQSR samples using low-vacuum scanning electron microscopy, although the latter also presented porous structures. Both a decrease in oil/water interfacial tension - mainly in HQSR (19.22 mN m^-1) - and a significant increase in interfacial viscosity in SQSR (1.5 E^-5 Pa s) were observed. Emulsions formulated with the sonicated sample showed greater stability against coalescence, creaming and sedimentation processes.

Conclusion: Ultrasound-assisted desaponification proved to be efficient, improving the functional characteristics of this residue and opening new possibilities for its use in food systems. © 2026 Society of Chemical Industry.

Background: Gamma-aminobutyric acid (GABA) is a naturally occurring bioactive component in plants. Our previous study demonstrated that ultrasound could increase the activity of glutamate decarboxylase (GAD) and effectively enhance GABA accumulation in coffee leaves. However, the mechanism underlying this action has yet to be explored. In this study, we investigated how ultrasound promotes GABA accumulation in coffee leaves by analyzing the relative contribution of the two GABA synthesis pathways, as well as changes in the key enzymes, signal molecules, and transcriptomes in coffee leaves under ultrasound treatment.

Results: The mechanical extraction had a negligible effect on GABA levels in coffee leaves, and the substrate sodium glutamate was essential for GABA accumulation under ultrasound stress. Aminoguanidine pretreatment reduced GABA content by 31.02% under ultrasound treatment. Fluorescent imaging revealed increased intracellular Ca²⁺ and H⁺ levels, potentially contributing to enhanced GAD activity in ultrasound-treated leaves. Transcriptomic analysis identified 1053 differentially expressed genes associated with multiple metabolic pathways, including carbohydrates, amino acids, flavonoids, and other primary and secondary metabolite biosynthesis. Further analysis indicated that ultrasound may enhance GABA accumulation by modulating Ca²⁺, reactive oxygen species, mitogen-activated protein kinase, and phosphatidylinositol signaling pathways, as well as the glyoxylate and dicarboxylate metabolic pathways.

Conclusion: This study revealed that ultrasound promoted GABA accumulation mainly through the GABA shunt pathway, with mechanical extraction playing a minimal role. Ultrasound may enhance GAD activity by increasing intracellular Ca²⁺ levels and lowering pH, while also regulating the expression of genes related to GABA biosynthesis. © 2026 Society of Chemical Industry.