Chenyao Yang

The traditional methods of field-based phenotypic data collection for crop germplasm resources are often inefficient and highly subjective. As the foundation for breeding innovation, these resources require precise identification of phenotypic traits for effective evaluation and utilization. Therefore, efficient and standardized management of germplasm data is critical during the breeding process. To address this, we have developed an intelligent recognition and management system focused on the crop’s organ characteristics. The system consists of a web client for overall project management and data download, and a WeChat Mini Program for data collection and uploading. Both components are integrated with image analysis models. Using a soybean variety screening experiment as a case study, we have constructed multiple high-definition datasets for soybean phenotypic traits, and employed YOLOv11 series models for object detection, image classification, instance segmentation, and pose estimation to build analytical models for each of these traits. All models achieved a mean average precision (mAP@0.5) exceeding 94%, along with a top1_accuracy of 0.999. In practical evaluations, all models took between 0.71 and 3.03 s to make predictions for 100 images, achieving an accuracy rate of over 98%. This system delivers a comprehensive solution for field phenotypic identification of crop germplasm resources, substantially enhancing the efficiency and objectivity of data collection and analysis. It serves as a valuable decision-support tool for precision breeding and digital agriculture.

The accurate detection of tea buds in outdoor environments is crucial for the intelligent management of modern tea plantations. However, this task remains challenging due to the small size of tea buds and the limited computational capabilities of the edge devices commonly used in the field. Existing object detection models are typically burdened by high computational costs and parameter loads while often delivering suboptimal accuracy, thus limiting their practical deployment. To address these challenges, we propose TeaBudNet, a lightweight and robust detection framework tailored for small tea bud identification under outdoor conditions. Central to our approach is the introduction of Weight-FPN, an enhanced variant of the BiFPN designed to preserve fine-grained spatial information, thereby improving detection sensitivity to small targets. Additionally, we incorporate a novel P2 detection layer that integrates high-resolution shallow features, enhancing the network’s ability to capture detailed contour information critical for precise localization. To further optimize efficiency, we present a Group–Taylor pruning strategy, which leverages Taylor expansion to perform structured, non-global pruning. This strategy ensures a consistent layerwise evaluation while significantly reducing computational overhead. Extensive experiments on a self-built multi-category tea dataset demonstrate that TeaBudNet surpasses state-of-the-art models, achieving +5.0% gains in AP@50 while reducing parameters and computational cost by 50% and 3%, respectively. The framework has been successfully deployed on Huawei Atlas 200I DKA2 developer kits in real-world tea plantation settings, underscoring its practical value and scalability for accurate outdoor tea bud detection.

Grasslands are key elements of the global agricultural system, covering around two-thirds of all agricultural areas and playing an important role in biodiversity conservation, food security, and balancing the carbon cycle. Climate change is a growing challenge for the agricultural sector and may threaten grasslands. To address these challenges, it is vital to conduct in-depth climate studies to understand the vulnerability of grasslands. In this study, machine learning was used to build an advanced model able to evaluate the future impact of climate change on grassland vigour. The objective was to identify the most vulnerable grassland areas, analyse the interaction between climate and grassland performance, and outline management strategies against the detrimental implications of climate change. A Random Forest (RF) regression was used to model the Normalised Difference Vegetation Index (NDVI) using the Standardised Precipitation-Evapotranspiration Index (SPEI). The model explained 76% of the NDVI variability. The foremost significant predictors of grassland vigour are the SPEI with temporal lags of 1, 4, and 12 months. These findings suggest that the vegetative status of grasslands exhibits high sensitivity to short-term drought while also being influenced by the memory of past climatic events over longer periods. Future projections indicate an overall reduction in grassland vigour, mostly in RCP8.5. The results indicate that negative effects will be more pronounced in mountainous regions, which currently host the most vigorous grasslands. Dry lowlands in the north should continue to have the lowest vigour in the future. A substantial reduction in vigour is expected in autumn, with an effect on grassland phenology. The development of grasslands in winter, favoured by increasing temperatures and precipitation, can advance the harvesting of grassland (cutting) and the grazing of livestock. To ensure that vigour is maintained in less favourable zones, adaptation measures will be needed, as well as more efficient management of highlands to provide an adequate level of production.

Vine phenology modelling is increasingly important for winegrowers and viticulturists. Model calibration is often required before practical applications. However, when multiple models and optimization methods are applied for different varieties, it is rarely known which factor tends to mostly affect the calibration results. We mainly aim to investigate the main source of the variability in the modelling errors for the flowering timings of two important varieties of vine in the Douro Demarcated Region (DDR) of Portugal; this is based on five phenology model simulations that use optimal parameters and that are estimated by three optimization algorithms (MLE, SA and SCE-UA). Our results indicate that the main source of the variability in calibration can be affected by the initially assumed parameter boundary. Restricting the initial parameter distribution to a narrow range impedes the algorithm from exploring the full parameter space and searching for optimal parameters. This can lead to the largest variation in different models. At an identified appropriate boundary, the difference between the two varieties represents the largest source of uncertainty, while the choice of algorithm for calibration contributes least to the overall uncertainty. The smaller variability among different models or algorithms (tools for analysis) compared to between different varieties could indicate the overall reliability of the calibration. All optimization algorithms show similar results in terms of the obtained goodness-of-fit: the RMSE (MAE) is 5–6 (4–5) days with a negligible mean bias and moderately good R² (0.5–0.6) for the ensemble median predictor. Nevertheless, a similar predictive performance can result from differently estimated parameter values, due to the equifinality or multi-modal issue in which different parameter combinations give similar results. This mainly occurs for models with a non-linear structure compared to those with a near-linear one. Yet, the former models are found to outperform the latter ones in predicting the flowering timing of the two varieties in the DDR. Overall, our findings highlight the importance of carefully defining the initial parameter boundary and decomposing the total variance of prediction errors. This study is expected to bring new insights that will help to better inform users about the importance of choice when these factors are involved in calibration. Nonetheless, the importance of each factor can change depending on the specific situation. Details of how the optimization methods are applied and of the continuous model improvement are important.

The present study is devoted to climate change impact assessment on the phenological development and ripening of cv. Touriga Nacional in the Dão Wine Region, Portugal. For this purpose, the dates of the three main phenological stages (budbreak, flowering and veraison) and two maturity stages are projected for two future periods (2041–2070 and 2071–2100), under two anthropogenic radiative forcing scenarios (RCP4.5 and RCP8.5), and compared against a baseline period (1991–2020). The phenological and maturity stages are simulated using phenological development models (PDMs) and a temperature‐based ripeness model (TRM), respectively. An overall advancement in both phenology and ripening stages are identified under future warmer climates, though site‐dependent. Furthermore, the advancements in phenology are more pronounced (a) for stages between the beginning of veraison and end of ripening than for the earlier stages, (b) for the long‐term future period (2071–2100) under RCP8.5 and (c) for the vineyard site “Viseu.” These changes are due to the combination of budbreak advancement with a shortening of some phenophases. The strongest shortening is found in the ripening period, while no significant changes in flowering timings and duration of the berry development period are projected. The advancement and the shortening of the grapevine growing season will shift ripening to the warmest part of the year. This twofold climate change impact of the air temperature on ripening may affect the sugar and organic acid balance, as well as the colour of the must. The current findings can be used by the regional winemaking sector in planning and implementing suitable climate change adaptation to enhance its climate resiliency and sustainability. Subsequent studies for this wine region should be carried out to assess the climate change impacts on late frost risk, on climatic viticultural zoning, on yield and berry quality at harvest.