The sex ratio of the rice stem borer, Chilo suppressalis (Walker), is important for assessing reproductive potential and outbreak risk, but automated sex classification in field trap images is hindered by small targets, variable postures, subtle morphological differences and complex backgrounds. An edge–cloud collaborative system was developed for fine-grained sex classification of the rice stem borer. The edge terminal performs image acquisition, target detection, ROI extraction, and foreground enhancement, while the cloud platform conducts sex classification and result management. To improve small-target localization, YOLO-CSNet was constructed by integrating a global attention mechanism and adaptive spatial feature fusion into YOLO11n. Within detection-constrained ROIs, Haar-like features and an AdaBoost discriminator were used to suppress tray textures, shadows and non-target regions. For sex classification, DRS-ViT combines convolutional patch embedding, a discriminative-region soft-weighting module and a KANsformer encoder to represent weak sex-related cues and model global structural relationships. YOLO-CSNet achieved precision, recall, mAP@0.5, and mAP@0.5:0.95 values of 94.36%, 93.19%, 95.32%, and 73.95%, respectively. DRS-ViT achieved accuracy, precision, recall, and F1-score values of 93.93%, 94.68%, 93.29%, and 93.91%, respectively. In a temporally independent field deployment conducted at one Shandong site from May to June 2026, the system achieved an image-upload success rate of 91.7% and an end-to-end accuracy of 87.9%. The field results support the feasibility of the workflow under the tested conditions. Future work will extend field validation across multiple sites and seasons to evaluate system generalizability.
Post-harvest field dew retting is a critical stage in flax fibre production because it facilitates the separation of fibre bundles from the surrounding stem tissues. In current agricultural practice, the optimum retting stop-point is still assessed largely using empirical criteria, motivating the development of quantitative indicators of retting progression. In this study, the mechanical properties of flax stems were investigated throughout a 91-day field dew retting campaign. Bending and lateral compression experiments were combined with a hollow-tube analytical model to determine equivalent flexural and compressive properties from force–deflection measurements. The equivalent flexural modulus and equivalent flexural yield strength showed an overall reduction during prolonged retting, whereas the equivalent compressive properties exhibited much less pronounced systematic variation. Additional tissue-removal experiments showed that the peripheral tissues make a substantial contribution to the whole-stem flexural response, while optical microscopy revealed increasingly evident separation and detachment of these tissues during the later stages of retting. Together, these results identify whole-stem flexural properties as promising quantitative indicators of retting progression and provide a basis for the future development of objective mechanical approaches for field retting assessment.
Tractor rollover is a persistent worldwide problem that does not yet have a fundamental solution. Adjusting the tractor’s attitude or configuration may enhance mobility under complex terrain conditions to prevent rollover. Therefore, in this study, an attitude adjustment method for offsetting the height difference between the uphill and downhill sides is proposed to adjust a tractor’s posture. Kinematic models are established for front-wheel, four-wheel, and articulated body steering modes. Steering mathematical models are developed for the three modes to describe the effects of posture change on tractor steering instability. This method predicts the steering stability by analyzing tire contact forces. Both the critical slope angle and steering speed are derived and used to predict instability while taking the steering radius into consideration. Considering a tractor’s attitude, simulations are conducted under two conditions; namely, attitude adjustment and level attitude. The results show that attitude adjustment is an effective method to enhance a tractor’s steering stability to avoid overturning. Furthermore, the models presented here provide theoretical references and optimization directions to prevent lateral overturning during tractor steering.
Oats (Avena sativa L.) are a popular crop in Poland, used primarily for animal feed, human food, and biomass in the energy sector. A three-year field experiment was conducted in Poland between 2020 and 2022. The experimental factors included hulled oat (Bingo) and naked oat (Maczo) varieties as well as growth regulators: chlormequat chloride (CCC) and trinexapac-ethyl (TE). The aim of this study was to investigate the effect of growth regulators on yield, yield structure components, selected physiological processes, and canopy architecture indices. This study found that the use of growth regulators resulted in an average 7.2% increase in oat grain yield. The yield of the Bingo variety increased after TE application, while that of the Maczo variety increased after CCC and TE application. Growth regulators also had a beneficial effect on increasing the weight and number of grains per panicle. Growth regulators, particularly CCC, caused the shortening of the main shoot and panicle. Both CCC and TE increased the leaf area index (LAI). Among the physiological parameters, the relative leaf chlorophyll content (CCI) was 2.6% higher after CCC treatment than after TE treatment, while the photosynthetic index (PI) was higher after TE treatment (a 2.2% increase compared to CCC). The Bingo cultivar showed a more favorable response to growth regulators than the Maczo cultivar. The weather conditions during the oat growing season significantly influenced some of the obtained results.
Aiming at the path tracking oscillation problem caused by sideslip of navigation agricultural machinery in complex farmland environments, this paper takes the Yanmar YR-70D transplanter as the research object and proposes an adaptive control method based on deep belief network (DBN) sideslip identification. Based on the preview tracking model, this method dynamically adjusts the preview distance by real-time identifying the sideslip status via DBN. In this research, the Optuna framework is utilized to optimize the DBN model, and the SHAP framework is introduced to quantify the feature contribution. The results show that, considering both real-time performance and accuracy, when the combined variable of “lateral deviation + heading deviation” with a data length of 20 is adopted, the prediction accuracy of the model reaches 88.04%, among which heading deviation (HD) has the highest contribution with a SHAP value of 0.957. Comparative experiments indicate that the accuracy of DBN is comparable to that of mainstream models such as LSTM and Transformer, but its forward propagation speed (0.0487 ms) is better than other models’. Upland field experiments verify that under different test factors of vehicle speed and slope angle, the line-on time (1.83–5.60 s), line-on distance (1.46–5.07 m), and maximum overshoot (0.32–15.75 cm) of the adaptive control group are all superior to those of the fixed-parameter control group. Paddy field experiments further confirm that under the working condition of 1.0 m/s vehicle speed and 15° slope, the above three indicators of the experimental group are 0.85–3.1 s, 0.75–3.14 m and 0–7.44 cm, respectively, all of which are better than those of the control group. The proposed method demonstrated superior path tracking performance under the operating conditions of the tested rice transplanting machines, indicating its potential application in lateral slip perception adaptive navigation control. However, further verification is still needed with different types of transplanting machines, field environments, and operating conditions to assess its general applicability.
Plant wearable sensors have emerged as a transformative technology for precision agriculture and plant phenotyping, enabling in situ, real-time, and continuous acquisition of physiological signals from plant surfaces or internal tissues. However, existing reviews have organized the literature by monitoring targets, sensing functions, or material platforms, without systematically comparing technologies from the fundamental dimension of the degree of intervention imposed on plants. Drawing on representative studies identified through a structured literature search, this review establishes a three-tier classification framework—invasive, minimally invasive, and non-invasive—and conducts a head-to-head comparison across six dimensions: signal characteristics, plant disturbance, long-term stability, manufacturing complexity, field deployability, and biosafety. The results reveal that invasive sensors (nanobionic probes, implantable microelectrodes, and organic electrochemical transistors) achieve nM–pM detection limits, yet wound responses generally limit their effective monitoring duration to the order of days; non-invasive sensors (flexible patches, strain sensors, and multimodal platforms) support weeks-to-months of continuous monitoring and are amenable to scaled deployment, but the indirectness of surface signals confines detection limits to the μM level; minimally invasive technologies (microneedle arrays and ultra-thin microelectrodes) offer a compromise between the two extremes. On this basis, a decision framework based on three-layer selection is proposed to guide technology selection across laboratory research, field deployment, and controlled environment agriculture. Future efforts should focus on standardized performance evaluation protocols, biodegradable self-powered systems, and the integration of invasive–non-invasive hybrid sensing networks with plant digital twins.
Soil is the foundation of resilient agricultural systems, yet its assessment remains a complex challenge due to its inherent variability across scales and management systems [...]
Accurate prediction and energy-efficient control of pig-house environments remain challenging because temperature, relative humidity, NH3, and CO2 are nonlinear, strongly coupled, and affected by ventilation and seasonal conditions. This study proposes an intelligent prediction-control framework integrating hybrid deep learning prediction with nonlinear model predictive control for multivariable pig-house environmental regulation. Several hybrid deep learning models were compared, including CNN-LSTM-standard transformer, CNN-LSTM-lightweight transformer, and the proposed CNN-LSTM-linformer-style model. The CNN-LSTM-lightweight transformer achieved the highest overall prediction accuracy, whereas the proposed CNN-LSTM-linformer-style model provided the most compact structure by using separable convolution, global average pooling, and linformer-style attention, reducing training time and memory usage by approximately 53% compared with the CNN-LSTM-standard transformer. The prediction model was integrated with FLC, NMPC, and ANMPC for closed-loop ventilation control. ANMPC adjusts control weights online according to environmental deviations to balance environmental regulation and energy use under disturbances. In the nominal closed-loop simulation, NMPC and ANMPC reduced ventilation energy consumption by approximately 44% compared with FLC, while ANMPC achieved a 4.35% lower NH3 steady-state error and a 3.5% faster NH3 recovery response than NMPC under disturbance conditions. In 24-h pre-field verification, NMPC and ANMPC reduced energy consumption by 35.8% and 27.4%, respectively, while maintaining pollutant safety.
The sequestration and release of carbon in soil is a crucial aspect of agricultural production studies, involving numerous small-plot trials and modelling processes. Small-scale heterogeneity in soil properties can influence measured carbon dioxide (CO2) fluxes. This study aimed (i) to compare CO2 emission from two neighbouring sandy soil plots managed with identical agricultural practices; (ii) to identify the key factors influencing soil CO2 emissions; and (iii) to examine the effect of soil water content (SWC) and soil temperature (Ts) on the results. Two plots located approximately 30 m apart and differing in terms of their humus depths and contents were selected for investigation. Continuous SWC and Ts measurements were taken. Portable devices were used to determine CO2 emissions, penetration resistance (PR) and vegetation cover. The humus layer in plot A was 55 cm thicker than in plot B, while the soil organic carbon (SOC) content was 18% and 163% higher in the 0–30 cm and 30–90 cm soil layers, respectively. Vegetation cover was nearly twice as high in plot A, and the mean soil CO2 emissions were 36% higher than those measured in plot B. SWC showed an opposite trend, with plot B exhibiting values that were 9.7% and 17.7% higher than those of plot A in both the top and deepest soil layers, respectively. These findings emphasize the importance of including small-scale spatial heterogeneity when parameterizing or interpreting biogeochemical models, particularly when model inputs are based on limited soil measurements from specific locations.
Reliable joint detection of mango fruits and stems is an essential upstream perception task for robotic harvesting, but remains challenging because stems are small, slender, frequently occluded, and visually degraded by illumination variation. This study proposes MangoNET, a YOLOv11n-based framework for joint mango fruit and stem detection in complex orchard environments. A P2 high-resolution detection head preserves fine spatial information for small targets, while SPPF-ELAN aggregates local and contextual features for partially visible objects. SENet recalibrates channel responses under illumination variation, and WIoU v3 regulates bounding-box samples with different localization qualities. A dataset containing 1782 original images of Tainong and Jinhuang mangoes was collected from two orchards and data augmentation was applied only to the training set, increasing its size from 1172 to 2886 images through rotation, contrast adjustment, and Gaussian noise addition. MangoNET achieved fruit and stem F1-scores of 0.920 and 0.916, respectively, with mAP50 and mAP50–95 values of 0.941 and 0.690. Compared with YOLOv11n, mAP50 and mAP50–95 increased by 1.6 and 2.9 percentage points, respectively, while stem recall increased from 0.877 to 0.906. Source-image-independent five-fold cross-validation yielded mean mAP50 and mAP50–95 values of 0.944 and 0.711, respectively. Pilot evaluations using images acquired by a UAV and an RGB-D camera in a geographically distinct orchard suggested that MangoNET could maintain detection performance in a different orchard environment. MangoNET supplies fruit and stem candidate regions for subsequent association, harvesting-point localization, and robotic manipulation.
Financial vulnerability in agriculture is shaped by both enterprise-specific factors and the agricultural systems in which firms operate, challenging financial distress prediction models based exclusively on firm-level financial metrics. This study assesses the predictive performance of logistic regression (LR), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), and category boosting (CatBoost), examines whether country-level agricultural environment characteristics improve prediction beyond traditional financial indicators, and uses Shapley additive explanations (SHAP) to analyze predictor contributions. The empirical analysis is based on 28,745 firm-year observations of agricultural enterprises in the Visegrad Group (V4) countries over 2019–2024. Adding agricultural environment variables produced a modest but consistent improvement in predictive performance, increasing AUC from approximately 0.854–0.855 in firm-level boosting models to 0.864–0.865 in integrated specifications. All three gradient boosting algorithms outperform conventional LR, although differences among XGBoost, LightGBM, and CatBoost remain statistically insignificant. SHAP analysis identifies liquidity, leverage, profitability, and firm size as the primary predictors while revealing nonlinear and threshold-dependent relationships not captured by conventional linear models. Integrating agricultural environment characteristics with explainable machine learning (ML) therefore supports more comprehensive, transparent, and context-aware early-warning systems for agricultural financial distress.
Whether the extracts of Fomes fomentarius, Ganoderma applanatum and Trametes versicolor can inhibit the growth of Fusarium graminearum in vitro, reduce the severity of Fusarium head blight (FHB) and reduce fungal colonization of wheat seeds artificially infected with this pathogen were the questions that initiated this research. The mycelial growth of F. graminearum was significantly reduced after treatment with ethanol extracts of the studied white-rot fungi, from 43.1% with F. fomentarius to 53.2% with T. versicolor. Likewise, these extracts significantly reduced disease development in artificially infected wheat seeds. Disease severity was quantified using the Disease Severity Index (DSI). In fact, after artificial infection with F. graminearum of lines carrying effective resistance (Fhb7 locus) from an alien source and their treatment with mushroom extracts, disease severity was negligible, and seedling fitness was very high. The highest efficacy of 99.49% was recorded in the resistant line R6-1, carrying the Fhb7 allele from Thinopyrum elongatum (7E) following treatment with the T. versicolor extracts. Slightly lower efficacy values were obtained after treatment with G. applanatum (97.83%) and F. fomentarius (97.47%). The obtained results showed that the selected mushroom extracts can be used in F. graminearum control and support successful wheat cultivation in FHB-prone areas.
Fixed full-boundary headlands create redundant non-working space in irregular fields, while working direction and swath offset jointly affect coverage quality and inter-swath turning cost. This study proposes a dynamic-headland-aware coverage path planning framework that allocates headland space according to swath endpoints and turning demand and generates parallel working swaths and Bézier U-turns for each direction–offset candidate. Field-specific empirical cumulative distribution function (ECDF)-midrank normalization converts turning distance, coverage error, and headland ratio into relative quality scores, and DQN-SHADE searches the resulting non-smooth discrete evaluation landscape. In geometric simulations on 24 actual field boundaries, DQN-SHADE achieved the lowest mean gap to the discrete reference optimum (0.0034) and the highest threshold success rate, SR5×10−3, of 80.83% among six stochastic optimizers under a common candidate-evaluation budget. Relative to fixed full-boundary headlands, dynamic headland allocation reduced the mean headland ratio from 13.30% to 7.52%, increased retained working area by 16,378.23 m2 per field, and maintained 98.16% mean coverage. The proposed framework improves field-space utilization while maintaining coverage quality and geometric feasibility under the evaluated simulation conditions.
Automatic reseeding in spoon-chain potato planters requires reliable classification of miss-seeding, normal, and multiple-seeding events. Oblique viewing and continuous spoon motion cause scale changes, boundary truncation, blur, and occlusion, which destabilize single-frame predictions. The proposed system treats each seed-spoon passage as one control event. It combines lightweight detection, track-level learning, and multi-frame classification with air-blow clearing and missed-seed reseeding. Based on the target-size distribution, the detector removes the P3 head from YOLOv8n and restores shallow details through space-to-depth rearrangement and gated addition. Frame quality combines boundary distance, sharpness, adjacent-box intersection over union, and box-area stability. High-quality frames form a track prototype, while low-quality frames receive stronger constraints. A quality-gated temporal network classifies each spoon as containing 0, 1, or at least 2 seed potatoes. Each model was trained with three random seeds under the same data split. Compared with YOLOv8n, the detector reduced the parameter count and computational cost by 35.8% and 48.3%, respectively. Track-level learning improved event accuracy by 3.67 percentage points over frame-level training. The seven-frame model achieved 93.41% event accuracy and 93.80% macro-F1. At 0.5 m/s, the closed-loop recognition, air-blow, and reseeding success rates were 93.7%, 96.4%, and 97.1%, respectively. These results provide a practical basis for real-time, event-level planting-quality control in spoon-chain potato planters.
Phosphorus (P) is one of the main limiting factors for sugarcane establishment and longevity in tropical soils, requiring biotechnological strategies to improve fertilizer use efficiency. This study evaluated the effects of inoculation with Bacillus velezensis UFV 3918, alone or combined with reduced monoammonium phosphate (MAP) doses, on the morphological, nutritional, and root nutrient uptake responses of sugarcane grown in a dystrophic Red Latosol under greenhouse conditions. The experiment was conducted in a completely randomized design with six treatments and four replicates: absolute control (AC, without MAP), commercial control (CC, recommended MAP dose), B. velezensis alone (Bv), and Bv combined with 1/3, 2/3, or the full recommended MAP dose. Plant growth responses varied across treatments and stages, with Bv promoting greater leaf area than the CC at 120 and 180 DAP, whereas Bv + 1/3 MAP maintained leaf area comparable to the CC throughout the evaluation period. Principal component analysis (PCA) revealed a clear separation among treatments, with Bv exhibiting the most distinct plant response, primarily associated with greater stalk diameter, root and stalk biomass, higher concentrations of potassium and phosphorus in stalks, sulfur and magnesium in roots, and greater boron-use efficiency. Bv + 1/3 MAP displayed an intermediate multivariate profile, mainly associated with phosphorus uptake per unit root length and nutritional attributes related to phosphorus acquisition under reduced fertilizer supply. Correlation analysis further revealed strong positive associations among growth-related traits, biomass accumulation, and phosphorus-related variables, indicating a close relationship between P acquisition and plant responses to bacterial inoculation. Overall, inoculation with B. velezensis UFV 3918 improved early sugarcane growth and nutrient acquisition, and its combination with 1/3 of the recommended MAP rate maintained plant performance. These findings indicate the potential for reducing mineral P inputs during early sugarcane establishment; however, long-term field trials are required to determine whether these responses can be sustained throughout the crop cycle and under commercial production conditions.
Climate variability can affect forage production through changes in temperature, precipitation, and the frequency of stressful weather conditions. We examined climatic variability and alfalfa (Medicago sativa L.) productivity in three Romanian development regions—South-Muntenia, South-East, and North-East—over 2006–2024. The climatic variables included mean, maximum, and minimum air temperature, annual precipitation, and maximum 24 h precipitation. Regional relationships between climatic variables and alfalfa yield were examined using descriptive statistics, Kendall’s rank-based trend analysis with Sen’s slope, Pearson correlation, and multiple linear regression with diagnostic testing. We also developed an exploratory Alfalfa Climate Resilience Index (ACRI) to compare relative climate-resilience profiles of regional alfalfa production systems. The index combines two performance-related components, normalized mean yield and yield stability, with two climatic-context components, normalized precipitation and inverse normalized mean temperature. Its sensitivity to alternative component weights was also evaluated. Mean annual temperature increased significantly in all regions, with Sen’s slopes ranging from +0.071 to +0.100 °C year−1, whereas annual precipitation showed no significant monotonic trend. Yield slopes were negative in all regions, but only North-East showed a statistically significant decline (τ = −0.794, p < 0.001). Mean annual temperature was negatively correlated with yield in all three regions (r = −0.617 to −0.749), whereas the precipitation indicators showed no significant bivariate relationships with yield. The regional multiple regression models accounted for 64.3–85.1% of the observed variation in annual yield (R2 = 0.643–0.851), with the relative importance and statistical significance of climatic predictors differing among regions. ACRI ranked North-East first (0.760), South-Muntenia second (0.469), and South-East third (0.295), and this ordering remained unchanged across the alternative weighting scenarios examined. These scores represent relative differences among the three regional production systems within the study dataset and should not be interpreted as absolute measures of intrinsic alfalfa resilience. The findings show that climate–yield relationships vary among regions and that temperature was more consistently associated with yield than the precipitation indicators considered here. ACRI is therefore presented as an exploratory, dataset-dependent framework for within-dataset comparison of regional alfalfa production-system climate resilience and requires validation using broader reference datasets, common normalization criteria, and additional management and environmental variables before wider application.
Biochar application is widely proposed to improve crop yield and fertilizer productivity, yet its performance varies across climatic, soil, and management contexts. This global meta–analysis study of about 705 database entries retrieved from 93 studies analyzed responses across climatic classification, biochar feedstock and properties, soil properties, fertilizer and co–application regimes, and experimental types. Overall, biochar responded at 33%, 34%, and 32% for yield, PFPN, and PFPP. The strongest response occurred at the tropical, humid agro–ecological zone under moderate temperature regimes. Wood-derived biochar with lower pyrolysis temperatures (<400 °C), combined with moderate soil carbon content and nitrogen, produced a greater improvement. Organic + biochar co–applied fertilizer application rates of ≤50 kg N ha–1 and 50–100 kg P2O5 ha–1 increased the response, while excessive fertilizer rates (>200 kg ha) showed reduced response. Economic benefit analysis also showed the average overall highest maximum yield at ≤40–80 t ha–1 biochar rates demonstrating simultaneous improvement in yield and fertilizer productivity under appropriate management.
Stem development determines plant architecture and mechanical support, directly affecting yield, lodging resistance, and ornamental quality. Microtubules are essential components of the plant cytoskeleton, but their role in stem development in herbaceous peony (Paeonia lactiflora Pall.) remains largely unexplored. Here, a β-tubulin gene, PlTUB1, was identified from P. lactiflora. Its coding sequence was 1353 bp in length, encoding a 450-amino-acid protein that localized to microtubules, and its transcript levels declined progressively during stem development. Transient virus-induced gene silencing of PlTUB1 in P. lactiflora, with a silencing efficiency of approximately 55%, enhanced stem development, with increased stem diameter, elevated cellulose accumulation, and thickened secondary walls. Given that cellulose deposition and secondary wall thickening were closely associated with cortical microtubule organization, the role of PlTUB1 in microtubule regulation was further examined. Heterologous overexpression of PlTUB1 in Arabidopsis thaliana disrupted transverse microtubule alignment, reduced cellulose deposition, and impaired secondary wall and xylem development, thereby confirming the mechanistic link between PlTUB1 and microtubule organization. Collectively, these findings demonstrated that PlTUB1 negatively regulates stem development by modulating microtubule organization and secondary wall deposition, providing new insights into the cytoskeletal regulation of stem development in P. lactiflora.
Soil degradation constrains cropland productivity in Northeast China, where conservation tillage has been widely adopted to improve soil structure and sustain maize (Zea mays L.) production. However, previous studies have focused mainly on single soil types or regional-scale assessments, and the applicability of conservation tillage across contrasting cultivated soil types remains inadequately characterized. From 2021 to 2023, field experiments were conducted across geo-ecological zones within mid-temperate sub-humid and mid-temperate sub-arid climatic regions, encompassing seven typical cultivated soil types: Aeolian sandy soil, Bielic, Black soil, Chernozem, Histosol, Inceptisol, and Mollisols. A strip-tillage with straw return treatment (ST) was compared against no-tillage with straw mulching treatment (CK) to systematically evaluate the effects of strip-tillage on (i) soil physical properties (bulk density and porosity), (ii) soil chemical properties (organic matter and total nutrient contents), and (iii) maize yield components. Across the full dataset, ST significantly reduced soil bulk density and increased total porosity relative to CK; however, soil-type-specific comparisons showed that significant responses were detected only in some soil types. By contrast, soil chemical properties were significantly influenced by the treatment × soil type interaction: SOM and STN contents increased significantly in Inceptisol under ST, STP increased significantly in Mollisols, and STK increased significantly in Histosol. A PCA-derived composite physicochemical score, based on six standardized soil indicators, was higher under ST than CK in six of the seven soil types relative to CK, with the most pronounced improvements occurring in Aeolian sandy soil (+1.23), Bielic (+0.87), and Black soil (+0.65); significant maize yield increases under ST were detected in Aeolian sandy soil and Black soil, with mean increases of 7.07% and 10.94%, respectively. The associations between maize yield components and grain yield differed between ST and CK, with 1000-kernel weight showing a stronger association with grain yield under ST. Collectively, these findings indicate that ST can improve soil physical structure, but its effects on soil chemical properties, maize yield, and yield-component associations are soil-type dependent.
Sloping farmland erosion in cold-region black soil zones threatens global agriculture, yet regional-scale quantitative evidence on the erosion mitigation performance and spatial suitability of control measures remains lacking. This study integrated 959 field observation datasets from 132 peer-reviewed publications, and used meta-analysis and Boosted Regression Tree (BRT) models to evaluate tillage, biological, engineering, and combined conservation measures in Northeast China. Results showed that all measures significantly reduced erosion, achieving an average runoff reduction of 73.2% and sediment reduction of 83.6%. Specifically, engineering measures showed the highest runoff reduction, while combined measures achieved more than 95% sediment reduction. Machine learning revealed that runoff reduction was primarily regulated by precipitation, whereas sediment reduction was mainly controlled by soil clay content and bulk density. Spatially, tillage, biological, and engineering measures showed the highest suitability in continuous cultivated plains, plain–hill transition zones, and low hilly and gully regions, respectively. This study moves erosion-control assessment beyond comparisons of average runoff and sediment reduction rates toward environmentally matched spatial allocation, providing a basis for targeted black soil conservation and resilient grain production in cold-region agricultural landscapes.
