Search Results (3,032)

Impervious surfaces are an important land surface indicator of urbanization level and human activity intensity, playing a crucial role in urban development monitoring and ecological environment assessment. However, in complex high-altitude regions such as the Qinghai–Xizang Plateau, the identification accuracy of existing medium-resolution impervious surface products remains limited at the regional scale due to complex land surface backgrounds, sparse distributions of impervious surfaces, and their generally small spatial extent. To address this challenge, this study proposes a Seed-Driven Grid Adaptation (SDGA) framework for large-scale impervious surface mapping over the Qinghai–Xizang Plateau. The proposed method uses the Google Satellite Embeddings (GSE) dataset as the primary input features and incorporates a 10 m impervious surface prior (P10) derived from a 2 m high-resolution impervious surface product to provide spatial constraints. Based on this prior information, a Prior-guided Hybrid Active Sampling (PHAS) strategy is developed to automatically construct high-value training samples through uncertainty-based positive sample mining and cluster-based negative sample mining. The framework first builds an initial seed knowledge base in the Lhasa seed area and subsequently performs local adaptive expansion within a 2° × 2° grid system, enabling automated impervious surface mapping across the Qinghai–Xizang Plateau. Experimental results show that, with only a small number of initial samples, the PHAS strategy significantly improves model performance, increasing the F1 score for impervious surface identification in the Lhasa seed area from 65.02% to 82.22%. During the grid-level adaptation stage, approximately 67% of the grids achieved improved accuracy, with an average F1 score increase of 0.1109 across the study area. Ultimately, the SDGA framework produced a 10 m resolution impervious surface product for the Qinghai–Xizang Plateau (SDGA-ISC10m), achieving an overall F1 score of 0.8223. Compared with seven existing medium-resolution impervious surface datasets, the proposed method demonstrates improved recognition performance under complex plateau environments, particularly in detecting sparsely distributed and small-scale impervious surfaces. The results indicate that integrating remote sensing embedding features with active learning strategies can effectively reduce the need for manual annotation and provide a new technical pathway for large-scale impervious surface mapping in complex regions. Full article

This study demonstrated the effects of dietary supplementation with 1% raspberry (RO) or 1% strawberry (SO) seed oil from 5 to 12 weeks of age (n = 6/group) on folliculogenesis, hormonal parameters, the ovarian fatty acid profile, and the expression of related genes in juvenile rabbits. After slaughter, ovaries and blood were collected. Ovaries were used for histology, fatty acid profiling, and gene expression analysis, while plasma was used to measure progesterone (P4), testosterone (T), estradiol-17β (E2), follicle-stimulating hormone (FSH), and anti-Müllerian hormone (AMH) concentrations. Both RO and SO reduced the number of primary follicles (p = 0.04), whereas RO increased the number of antral follicles (p = 0.04) compared with the control. In both supplemented groups, FSH (p = 0.04 and p = 0.035) and AMH (p = 0.04) concentrations were higher. RO increased P4 and E2 (p = 0.03 and p = 0.013) concentrations, while SO only increased P4 (p = 0.02) levels. SO altered the ovarian fatty acid profile, increasing selected monounsaturated fatty acids and reducing polyunsaturated fatty acids, likely by increasing the expression of the converting enzyme, stearoyl-CoA desaturase 5 (p = 0.038). Overall, both oils influenced folliculogenesis through hormonal changes, and SO modified ovarian fatty acid composition, which may affect ovarian function in juvenile rabbits. Full article

Severe transients in nuclear power plants (NPPs) are strongly coupled and highly nonstationary, which makes reliable short-horizon multivariate forecasting difficult for conventional sequence models. To address this challenge, this study develops a hybrid LSTM-Transformer forecasting framework for severe nuclear accident time series and uses the Sparrow Search Algorithm (SSA) as a task-oriented joint hyperparameter optimization tool for nuclear accident forecasting. In this framework, the self-attention mechanism captures long-range temporal dependencies and cross-variable interactions, while the LSTM component strengthens the modeling of short-term dynamics and local temporal memory. SSA is employed as a task-oriented joint hyperparameter optimization tool to adapt key model settings, including the number of attention heads, encoder depth, model dimension, LSTM hidden units, and dropout rate, for severe nuclear accident forecasting. In addition, a regularized training strategy combining dropout and validation-based early stopping is adopted to alleviate overfitting and improve training stability. The main comparison results are reported as mean ± standard deviation over 20 independent runs with the same data split and different random seeds. Experiments on high-fidelity PCTran/APR1400 simulations covering LOCA, LACP, and SLBIC scenarios, together with a severity-shifted LOCA test, demonstrate strong and statistically stable predictive performance. Across the three representative accident scenarios, the proposed framework achieves mean R² values of 0.943 ± 0.009, 0.951 ± 0.007, and 0.946 ± 0.010, while maintaining about 30% lower mean nRMSE and nMAE than the strongest LSTM-Transformer baseline. A 2 × 2 ablation study shows that regularization mainly improves training efficiency, reducing the required epochs by a range of about 36–41%, whereas SSA primarily improves predictive accuracy through better hyperparameter selection. Their combination provides the best overall generalization. Cross-severity LOCA evaluation further confirms the robustness of the proposed model, yielding mean R² = 0.885 ± 0.017 and mean nRMSE = 0.100 ± 0.010. The model also achieves low inference latency (P50 = 7.6 ms per sample), indicating its computational potential for near-real-time multivariate forecasting in safety-critical transient monitoring. Full article

Controlled-release fertilizer (CRF) improves fertilizer-use efficiency through sustained nutrient release, but its rate-dependent effects on the growth and physiology of Paeonia delavayi seedlings remain unclear. In this study, germinated seeds of P. delavayi with radicles 3–4 cm in length were grown under container nursery conditions with four CRF application rates: (CK, 0 kg·m⁻³), treatment 1 (T1, 0.6 kg·m⁻³), treatment 2 (T2, 1.2 kg·m⁻³), and treatment 3 (T3, 2.4 kg·m⁻³). Morphological traits, root characteristics, biomass accumulation, physiological parameters, and chlorophyll fluorescence were evaluated, and Pearson correlation and fuzzy membership analyses were used to compare overall treatment performance within the tested range. CRF significantly promoted seedling height, leaf number, petiole length, and biomass accumulation, although the promoting effect did not increase continuously with fertilizer rate. By June, seedling height in T2 was 160% greater than that in CK, while aboveground biomass increased by 552% and 574% in T2 and T3, respectively. Root morphological traits were not significantly affected, suggesting that CRF primarily promoted aboveground development and biomass production. Medium and high CRF rates increased leaf superoxide dismutase (SOD) activity by 42% and 103%, respectively, and peroxidase (POD) activity by 163% and 250%, respectively. Aboveground starch content was 45% higher in T2 than in CK. In contrast, photosynthetic pigment contents and the chlorophyll a/b ratio were not significantly affected by CRF. Chlorophyll fluorescence analysis showed that Fv/Fm remained stable among CRF treatments (0.78–0.82) and was significantly higher than that in CK (0.65), whereas the actual quantum yield of PSII [Y(II)] did not differ significantly among treatments. Relative to CK, tthe quantum yield of non-photochemical quenching [Y(NPQ)] increased from 0.20 to 0.40 in T2, while the quantum yield of non-regulated energy dissipation in PSII [Y(NO)] decreased from 0.37 to 0.24–0.22 in T2–T3. Pearson correlation and fuzzy membership analyses ranked the treatments as T2 > T3 > T1 > CK, indicating that T2 performed most favorably within the tested range, although its advantage over T3 was small. Overall, an appropriate CRF rate promoted P. delavayi seedling growth and was associated with changes in biomass accumulation, antioxidant enzyme activity, carbon assimilate storage, and chlorophyll fluorescence parameters. Full article

(1) The growing interest in the use of natural and sustainable ingredients highlights the investigation of vegetable oils in dermato-cosmetic applications. In this context, the vegetable oil obtained from walnut (Juglans regia L.) is of actual interest due to its composition rich in unsaturated fatty acids. The aim of the present study was to investigate and characterize walnut oil from a physicochemical, structural, and rheological point of view. (2) The oil was obtained by a cold pressing process from walnut seeds, with a yield of about 51.03 ± 1.41%, and subsequently analyzed by complementary methods. (3) The results show an acceptable physicochemical profile, characterized by appropriate values of density, pH, and spreadability. The oxidative stability indicated a moderate resistance to degradation, specific to oils rich in polyunsaturated fatty acids. Fourier infrared transform spectrometry (FTIR) analysis confirmed the presence of functional groups characteristic of triglycerides, without indications of advanced oxidation, and atomic force microscopy (AFM) investigations revealed a heterogeneous morphology. The rheological properties indicated a pseudoplastic behavior, favorable for topical application. The determination of heavy metals confirmed the safety of the raw material for the intended dermato-cosmetic use. While arsenic levels were slightly above the strict Codex Alimentarius limits for foodstuffs, all values remained within the safety ranges established for cosmetic ingredients. A total of six fatty acids were found in cold-pressed walnut oil, determined using GC-MS methods. The number of compounds identified in the silylated sample was found to be 17. The antioxidant activity determined using DPPH and ABTS methods was generally considered good and relatively stable over time. The measured sun protection value (SPF) demonstrates a favorable capacity to act as a photoprotective ingredient against ultraviolet (UV) radiation. (4) Overall, the results demonstrate that walnut oil presents adequate physicochemical and structural properties, supporting its further use as a potential cosmetic raw material. Full article

Encrypted transport increasingly limits the visibility required by intrusion detection systems (IDS), motivating payload-free learning from flow statistics and protocol metadata. We introduce GCP, a graph-contrastive pretraining framework that casts flows as nodes in a sparse graph and learns transferable node embeddings via an InfoNCE-style objective with graph-specific augmentations. The learned encoder is evaluated through frozen-embedding linear probing and cross-dataset out-of-domain (OOD) transfer, within a fully scripted pipeline that freezes run manifests and artifacts to make every reported number traceable and reproducible. Experiments cover enterprise IDS and encrypted DNS/DoH traffic using CICIDS2017, UNSW-NB15, and DoH-Combined at three label granularities (L1/L2/L3), for both binary detection (y) and finer-grained targets ($y_{\mathrm{multi}}$), aggregated over five fixed split seeds with 95% confidence intervals. Results show that GCP yields a pronounced in-domain advantage on UNSW-NB15 for y (Macro-F1 $\approx 0.993$) while substantially reducing false-alarm rate (FAR $\approx 0.013$) compared with strong tabular baselines. In feature-separable regimes (CICIDS2017 and DoH L1/L2), boosted-tree and supervised baselines remain difficult to surpass, but ablations confirm that graph structure alone is insufficient without contrastive pretraining. OOD transfer is strongly source–target dependent, with the most reliable transfer within closely related DoH domains, highlighting dataset shift as a first-class evaluation criterion for encrypted-traffic IDS. Full article

The geographical origin of aniseed (Pimpinella anisum L.) represents a key quality determinant, as it directly influences the chemical composition and commercial value of its essential oil. Agronomic traits of aniseed (plant height, umbel diameter, number of umbels per plant), productivity-related traits (number of seeds, thousand-seed weight, yield per plant, plant biomass, harvest index, yield per hectare, essential oil content and yield), and physiological traits (germination energy and total germination) exhibit variations depending on geographical origin. The study proposes an integrated framework for accurate classification by combining agronomic, productivity, and physiological data with GC-MS profiles and advanced machine learning (ML) techniques. A total of 144 samples were analyzed, based on a factorial design including three locations, six fertilizer treatments, two years, and four replications. trans-Anethole was the dominant compound in all samples (89.508–101.441%). Several classification models, including artificial neural networks, random forests, MARSplines, boosted trees, interactive trees, naïve Bayes, and support vector machines, were evaluated to discriminate samples by geographical origin using agro-meteorological and GC-MS data. The results indicate that AI and ML approaches effectively captured complex non-linear relationships. Overall, the multi-model framework highlights the strong potential of machine learning for agro-food authentication, supporting improved traceability, site-specific decision-making, and quality control. Full article

Discrete memristive chaotic maps are promising for secure communications due to their digital compatibility, yet existing designs face limitations, including narrow hyperchaotic ranges and a single type of chaotic attractor. This paper proposes a family of 2D hyperchaotic maps by coupling a discrete exponential memristor with four 1D seed maps. Theoretical analysis reveals that the exponential memristor induces non-hyperbolic fixed points and periodicity with respect to the memristor’s initial charge, enabling controlled coexistence of both homogeneous and heterogeneous multistable attractors. Numerical simulations show two positive Lyapunov exponents (LEs) and broad hyperchaotic regions; the memristor-coupled Sine map achieves a maximum LE of 0.4963 and spectral entropy (SE) of 0.8915, outperforming representative cosine- and quadratic-based benchmarks. A pseudorandom number generator (PRNG) passes all National Institute of Standards and Technology (NIST) SP 800-22 tests. STM32F407-based hardware experiments confirm physical realizability, and an image encryption application demonstrates near-ideal entropy (7.9883) and strong differential attack resistance. These results establish the discrete exponential memristor as an effective nonlinearity for enriching chaos complexity and hardware-oriented security primitives. Full article

Soil salinization is a significant environmental factor limiting agricultural production. Developing salt–alkali-tolerant microbial resources is important for the improvement of saline–alkali land. Plant growth-promoting rhizobacteria stimulate crop growth by producing the plant growth hormone indole-3-acetic acid (IAA), but their fermentation process under salt stress still needs optimization. Single-factor experiments and response surface methodology (RSM) were used to systematically optimize the fermentation conditions of the salt–alkali-tolerant Vreelandella titanicae J113. Key influencing factors were screened using the single-factor experiment design, and optimal process parameters were determined using the Box–Behnken design. IAA production and cell biomass were used as evaluation indicators to study the interactions of carbon sources, nitrogen sources, inorganic salts, temperature, cultivation time, and inoculum size. The optimal fermentation process was obtained: starch concentration 17.5 g/L, NaCl concentration 32.5 g/L, yeast extract 5 g/L, cultivation temperature 30 °C, inoculum size 3%, and cultivation time 144 h. After optimization, IAA production reached 23.02 μg/mL, an increase of 115% compared with before optimization. Salt stress experiments showed that the strain could still maintain high IAA production under 3% NaCl, demonstrating good salt tolerance. Maize seed germination experiments demonstrated that the optimized fermentation broth significantly promoted seed germination and seedling growth under salt stress conditions, with root length, fibrous root number, and fresh weight increasing by 61–86%, 137–200%, and 25–57%, respectively, compared to the control group. This study established an efficient IAA fermentation process for the salt–alkali-tolerant Vreelandella titanicae J113, providing technical support for developing microbial plant growth regulators suitable for saline–alkali land. The optimized strain exhibits excellent growth-promoting potential under salt stress conditions, offering favorable application prospects. Full article

The rapid growth of the Internet of Things (IoT) has introduced significant cybersecurity challenges due to the heterogeneity, scale, and limited protection capability of connected devices. Although machine learning has been widely adopted for IoT intrusion detection, many existing studies still rely primarily on closed-world evaluation settings, unequal baseline comparison budgets, fixed decision thresholds, and limited integration of explainability into model assessment. To address these issues, this paper proposes an explainable XGBoost-based framework for IoT attack detection with unseen attack family evaluation using the large-scale CICIoT2023 dataset. In the proposed framework, IoT traffic is formulated as a binary classification task that distinguishes benign from malicious flows. The study integrates two complementary evaluation protocols: (1) closed-world stratified 10-fold cross-validation for in-distribution performance assessment and (2) unseen attack family evaluation, in which one malicious family is excluded from training and used only for testing under a zero-day-like but single-dataset condition. A fair-budget experimental design is adopted to compare seven representative models under the same training budget, including default XGBoost, optimized XGBoost, Random Forest, LightGBM, CatBoost, Logistic Regression, and a simple multilayer perceptron. To improve reproducibility and operational validity, the revised framework further reports the sampling strategy, split-overlap audit, XGBoost hyperparameter search protocol, repeated unseen-family evaluation, validation-based threshold calibration under fixed-FAR constraints, cost-sensitive threshold analysis, and XGBoost-native SHapley Additive exPlanations (SHAP) compatible feature contribution analysis. The closed-world results show that tree-based ensemble methods clearly outperform the linear and shallow neural baselines. Random Forest achieves the highest closed-world macro-F1 of 0.9713, followed by LightGBM with 0.9602 and optimized XGBoost with 0.9566. In the fair-budget unseen-family setting under the default threshold, Random Forest again obtains the highest mean macro-F1 of 0.8433 and the lowest false negative rate (FNR) of 0.0712, but it also produces a substantially higher false alarm rate (FAR = 0.0536). By contrast, optimized XGBoost provides a lower-FAR default operating point, achieving a mean macro-F1 of 0.8194, Matthews correlation coefficient (MCC) of 0.7067, FAR of 0.0086, and FNR of 0.2996. Repeated unseen-family experiments over five random seeds confirm the same trade-off: Random Forest provides stronger recall-oriented detection, whereas optimized XGBoost provides a lower-FAR default operating point. After validation-based threshold calibration at an approximate FAR target of 0.01, Random Forest achieves the strongest calibrated recall-oriented performance, with macro-F1 of 0.8754, MCC of 0.7757, FNR of 0.2000, and attack recall of 0.8000. Optimized XGBoost remains competitive at the same FAR target, with macro-F1 of 0.8323, MCC of 0.7193, FNR of 0.2760, and attack recall of 0.7240. The explainability analysis indicates that the optimized XGBoost detector relies mainly on TCP control-flag, temporal, and packet-statistical features, with rst_count, IAT, urg_count, Tot size, Number, Header_Length, and Magnitude among the most influential variables. Local contribution tables for representative true-positive, false-positive, false-negative, and true-negative cases further improve the readability of the explanation results and confirm that native pred_contribs reconstructs the model margin with negligible numerical error. Overall, the results show that the most appropriate model depends on the deployment objective: Random Forest is preferable when minimizing missed attacks under a calibrated FAR constraint is prioritized, whereas optimized XGBoost remains a strong primary model for an explainable low-FAR XGBoost-based framework that emphasizes scalability, operational conservativeness, and native contribution-based interpretation. Full article

To identify the key causative factors and their characteristics across different types of chemical accidents, text mining techniques were first applied to extract causative factors from accident investigation reports. The extracted factors were then classified according to an improved Human–Machine–Environment–Management (HMEM) framework, which incorporates an additional government influence layer. To address data imbalance, a random undersampling method was employed. Specifically, sampling was repeated 30 times using different random seeds, and association rule mining was conducted for each sampled dataset. On this basis, a hybrid analytical framework integrating the Apriori algorithm and complex network theory was developed to examine the topological characteristics of the causation network. The results indicate that the network exhibits both small-world and scale-free properties, with strong interconnections among causative factors and a limited number of key nodes playing important bridging roles. PageRank centrality analysis further reveals that nodes associated with all accident types are located in the core region of the network, although differences exist in the associated causative factors across different accident types. In addition, the comprehensive importance analysis indicates that D6 (illegal production organization), B5 (pipeline rupture or blockage), and D12 (unsafe work practices) are the top three most important causative factors. These findings provide a theoretical foundation and practical insights for chemical accident prevention and the improvement of safety management. Full article

Effective weed management is crucial for optimizing agricultural productivity and minimizing environmental impacts. Weeds are most effectively managed during their seedling or early growth stages, which can be achieved with the aid of tools for predicting seedling emergence. However, many persistent weed species exhibit dormant seedbanks, thus complicating prediction attempts. The number of seedlings emerging in these species is closely tied to seedbank dormancy levels, which are influenced by seasonal variations. Thus, predictive population-based threshold models incorporate seedbank dormancy regulation to accurately forecast seedling “window” emergence. These models use the functional relationship between environmental cues (i.e., temperature, light, alternating temperatures, and soil water content) and seed dormancy behavior. Considering that these environmental signals vary among microsites in the field, these tools can be adapted to predict weed emergence in both temporal and spatial dimensions, thus making them suitable for site-specific weed management. The aim of this review is to synthesize existing modeling approaches and present a conceptual framework for dynamic, site-specific weed emergence predictions, supported by case-study-based applications. The illustrative application shows that incorporating soil water content into dormancy dynamics modifies emergence timing and magnitude, restricting emergence to specific topographic zones and potentially reducing herbicide use by up to 60–70%. This approach can improve the efficiency of herbicide applications and other control measures, reducing costs and environmental impact while enhancing crop yields. This work underscores the potential of integrating environmental cues into sophisticated modeling approaches to address the complexities of weed emergence in diverse agricultural landscapes. Full article

Deep learning has demonstrated high efficiency in histopathological image analysis, particularly in lung cancer classification. However, the stability of these models with image corruption and cross-dataset validation remains an important practical concern. In this study, we explored the potential of adding spectral information derived from the discrete wavelet transform (DWT) and spatial convolutional representations to enhance the robustness of multi-class lung cancer classification between Normal, Adenocarcinoma and Squamous cell carcinoma. The lightweight ResNet18 backbone was used to obtain spatial features, and spectral descriptors were obtained through wavelet sub-bands and integrated through early feature-level fusion. The models were trained and evaluated using the LC25000 dataset. Subsequently, it was tested under controlled perturbations, such as Gaussian noise and Gaussian blur. Three random seeds were used to assess performance variability, and paired t-tests were conducted as an indicative statistical measure of the results. Under clean conditions, the spatial and hybrid models were nearly saturated, and there was no significant difference between them (spatial: 99.85 ± 0.26; hybrid: 99.72 ± 0.22; p = 0.1217). The hybrid model exhibited higher robustness when Gaussian noise (σ = 0.05) was added, which resulted in 84.89% ± 4.52% accuracy versus 74.99% ± 7.20% of the spatial baseline (p = 0.0443) with an observed effect size (Cohen’s d = 2.64), noting that these estimates are based on a limited number of runs and should be interpreted with caution. The same behavior was observed in Gaussian blur perturbations, where the hybrid representation was slightly more stable. We also investigated a simplified adaptive gating mechanism process and found that the learned gate parameter also tends to converge towards spatial feature dominance with a model trained with clean data. Finally, cross-dataset validation with LungHist700 showed a slight increase in the balanced accuracy of the hybrid model (0.5158) over the spatial baseline (0.4722). These results indicate that spectral and spatial features can be used to enhance robustness to image corruption and still yield high classification accuracy, indicating that spectral–spatial representations can improve robustness under controlled perturbations, whereas their impact on cross-dataset generalization remains limited. The results further indicate that robustness improvements are strongly influenced by training strategies, such as noise augmentation, whereas the contribution of fusion is comparatively moderate. Full article

This study assessed phenotypic variation among open-pollinated half-sib families from a single relict population. Autochthonous silver fir (Abies alba Mill.) preserved in the Tisovik Reserve of Białowieża Forest represents the northeasternmost isolated relict population of the species in Europe. To secure its genetic resources and evaluate its breeding potential, a conservation plantation of open-pollinated half-sib families was established in the Hajnówka Forest District outside the natural species range. This study assessed the effects of half-sib family affiliation on the growth and phenotypic performance of almost two thousand 28–31-year-old trees representing 20 half-sib families and compared them with age-matched managed stands in the State Forests of Poland. Significant within- and among-family variation was observed for diameter at breast height (DBH) and height (H), while environmental factors had only marginal influence under the uniform site conditions of the plantation. Several half-sib families produced disproportionately high numbers of individuals with exceptional phenotypic performance, including DBH values exceeding 25 cm and height values surpassing those of managed stands. Based on a combined assessment of qualitative traits, selection differential, and 95th percentile values, 30 prototype plus trees were selected as sources of scions for establishing a future seed orchard. The outstanding growth parameters of these individuals correspond to stand ages of 40–65 years according to yield tables, despite their biological age of only 28–31 years. The results confirm the high breeding value and substantial genetic variability of the Tisovik population and demonstrate its potential for producing genetically diverse planting material adapted to lowland sites under changing climatic conditions. Full article

Soybean (Glycine max L.) yield is critically influenced by the number of nodes on the main stem (MSN), which serves as the primary site for pods and seeds. To elucidate the genetic mechanisms underlying MSN, we conducted a multi-omics analysis integrating bulk segregant analysis sequencing (BSA-seq), phytohormone, and transcriptome profilings in a soybean mutant, LSD914, which exhibits a significantly increased MSN number compared to its wild-type parent, HN48. BSA-seq of an F2 population identified 27 candidate genomic regions spanning 2.92 Mb, primarily on chromosome 18. Within these regions, 149 genes harbored non-synonymous SNPs and 26 genes contained frameshift InDels, with functional enrichment pointing to pathways in plant hormone signal transduction and developmental regulation. Phytohormone profiling revealed a distinct shift in LSD914, characterized by down-regulation of jasmonates, salicylates, and auxins, alongside specific accumulation of cis-zeatin. Integrative transcriptome analysis identified Glyma.18G259400, a gene encoding a gibberellin-regulated protein (GmGASA32), which was consistently and significantly down-regulated in LSD914 across all developmental stages and tissues. This finding contrasts with previous reports of its overexpression promoting plant height, suggesting a nuanced, context-dependent regulatory role. Our integrated approach identifies a key set of candidate genes and highlights GmGASA32 as a pivotal node in a hormone signaling network that orchestrates soybean node number, providing valuable targets for breeding high-yield soybean varieties with optimized plant architecture. Full article