Plant Phenomics最新文献

Separating wood and leaf components in tree point clouds is one of the key tasks for achieving automated forest inventory and management. To obtain accurate wood‒leaf separation results, traditional methods typically rely on large amounts of annotated point cloud data to train supervised semantic segmentation networks. However, point wise annotation is not only extremely labor-intensive but also time-consuming and costly, which greatly limits the widespread application and adoption of supervised learning methods in wood‒leaf separation tasks. To eliminate the dependence on annotated point clouds, this study explores the feasibility of wood‒leaf separation under completely unsupervised conditions. To this end, we propose an unsupervised semantic segmentation network that is capable of directly extracting wood and leaf components in 3D point clouds. The network adopts a sparse convolutional neural network as the backbone and incorporates two custom-designed modules: the dual point attention (DPA) module and the point cloud feature convolutional integrator (PFCI) module, for enhanced feature fusion and extraction. Semantic classification is then achieved by generating pseudolabels via super point clustering. Based on large-scale public datasets containing coniferous and broadleaf forests, in addition to our self-constructed dataset, our proposed network achieved an overall accuracy (oAcc) of 67.583%, a mean classification accuracy (mAcc) of 50.249%, and a mean intersection over union (mIoU) of 38.512%, and in wood and leaf separation at the tree level, it attained an oAcc of 80.856%, a mAcc of 64.013%, and a mIoU of 49.695%. Across both the forest and tree scenarios, our network outperforms the current state-of-the-art methods, namely, GrowSP and PointDC. Ablation experiments further confirm that each of the proposed modules contributes significantly to improving the segmentation accuracy, and in addition, our segmentation network demonstrates strong robustness even under high occlusion rates and exhibits excellent generalization capability.

Fusiform rust, caused by the pathogen Cronartium quercuum (Berk.) Miyabe ex Shirai f. sp. fusiforme, is the most important disease of loblolly pine (Pinus taeda L.) in the U.S., causing millions of dollars in damage each year. Using resistant genotypes has proven a successful strategy to limit the disease, but resistance selection still relies on visual inspection for symptoms, which can lead to misclassification due to human error and the presence of 'escaped susceptibles' (i.e., susceptible individuals with no visible symptoms due to either an extended asymptomatic phase of the disease or the lack of adequate disease pressure to become infected). Here, we propose the use of near-infrared (NIR) spectroscopy and chemometrics to improve the accuracy of how phenotypes are rated. We collected and analyzed phloem and needle spectra from 34 non-related families replicated across eight stands in three states in the southeastern region of the U.S. using a portable, handheld NIR spectrometer. We also used a benchtop Fourier-transformed mid-infrared (FT-IR) spectrometer to analyze phloem phenolic extracts of the same samples, as this phenotyping approach has proved successful in other pathosystems. Our results show a moderate association between the phloem spectra and resistance, and models built with NIR spectra were able to classify extremes (i.e., very resistant or very susceptible) with up to 69 ​% testing accuracy. This study provides a framework for using NIR spectroscopy for phenotyping loblolly pine resistance against pathogens and advocates for using alternative technologies in forestry.

Canopy nitrogen content (CNC) and canopy phosphorus content (CPC) of vegetation in wetlands are key physiological traits, which can be associated with the process of wetland ecosystems. Because of the spectral signals obscured by pigments and water content, it is challenging to accurately estimate CNC and CPC of vegetation species in wetlands using multispectral images. Therefore, we developed the constrained PROSAIL-PRO spectra matching (CPSM) approach to extend multispectral reflectance of unmanned aerial vehicle measurements to 400 ∼ 2500 ​nm. We verified the matched accuracy and spectral reliability of CPSM's spectra from two aspects of reflectance and vegetation spectral characteristic based on field-measured spectral data. We proposed a novel hybrid retrieval strategy to achieve the high-precision estimation of CNC and CPC for seven karst wetland vegetation species. Finally, we evaluated the applicability of combining CPSM with our strategy to estimate CNC and CPC for two typical species in mangrove wetlands. Our results proved that CPSM-based spectra had good consistency with original reflectance of UAV images (R² ​= ​0.82 ∼ 0.86), and they could maintain similar spectral characteristics to measured spectra. Besides, this study found that the optimal spectral features of CNC and CPC were distributed near the red edge position and water-absorption valley of vegetation spectra. We obtained high-precision estimation of CNC and CPC in karst wetland using CPSM and our hybrid retrieval strategy (R² ​= ​0.60 ∼ 0.98, MRE ​= ​5.91 ​% ​∼ ​26.25 ​%). The approach also showed a better transferring performance in estimating CNC and CPC of mangrove species (R² ​= ​0.77 ∼ 0.89, MRE ​= ​9.65 ​% ​∼ ​16.87 ​%). The CPSM approach is effective to achieve high-precision estimation of vegetation CNC and CPC.

Plant energy use is fundamental to plant survival and growth. However, we still lack effective means to quantify plant energy use strategies. This study introduced a concept quantifying the light level at which photochemical and non-photochemical energy use in plants are in equilibrium - the photochemical compensation point (PCCP) which can be determined with chlorophyll fluorescence measurements. We used winter wheat as a test case to explore the dynamics of PCCP and its physiological and biochemical regulations. Winter wheat PCCP decreased significantly across growth stages from jointing to grain filling. Long-term nitrogen and phosphate (NP) fertilization significantly increased PCCP, whereas potassium (K) and manure (M) fertilizer supplementation had negligible effects. PCCP exhibited significant positive correlations with leaf thickness, leaf P and sulfur (S), and stomatal conductance (g _s ) across all growth stages. All manure-amended treatments exhibited positive correlations of PCCP with leaf N, P, K and g _s , and negative correlations with leaf calcium (Ca). Random forest analysis revealed that g _s was the most significant predictor of PCCP variation, followed by leaf P, iWUE, and leaf thickness across all treatments. We suggest that plant energy use strategies are strongly coupled with plant water use strategies and nutrient availability through a complex interplay of effects on physiological and biochemical traits.

Three-dimensional tree modeling is crucial for forest ecological applications. However, building accurate individual tree models still faces unresolved challenges, such as wrongly connected branches within the canopy and poor quality modeling results when dealing with tree points containing data gaps. To address these issues, this paper proposes an in-novation method for individual tree modeling based on skeleton graph optimization and fractal self-similarity. In this paper, the skeleton points are initially extracted through the Laplacian-based contraction and the farthest distance spherical sampling. To centralize the extracted skeleton points within each point set, a method for skeleton points adjusting and optimization is presented, which helps achieve centralized skeleton points, particularly in cases with incomplete branch points. Additionally, instead of using Euclidean distance or its square as edge weight, the paper proposes a novel edge weight definition, which ensures the construction of correctly connected skeleton lines, especially for branches within the canopy. To improve fidelity and robustness against outliers, fractal self-similarity is first applied in this paper to refine individual tree models and achieve better modeling results. The effectiveness of the pro-posed method is evaluated using 29 individual trees of different structure characteristics with known harvest volumes. Experimental results demonstrate that this method achieves tree volumes closest to the referenced values, with a relative mean deviation of 0.01 ​% and a relative root mean square error of 0.09 ​%. Moreover, the concordance correlation coefficient of the proposed method is 0.994, outperforming two classical individual tree modeling methods, TreeQSM (Quantitative Structure Model) and AdQSM, based on five accuracy indicators.

The growth of the global population has increased the demand for fruits and vegetables, while high harvesting labor costs severely constrain industry development. Currently, relevant personnel primarily utilize 2D object detection technology to facilitate automated harvesting, aiming to reduce labor costs. However, 2D detection technology is limited to providing planar information and cannot meet the requirements of scenarios that need 3D spatial data, whereas 3D object detection technology can effectively address these needs, including point cloud-based methods and monocular-based methods. Since point cloud-based object detection methods require expensive equipment, they are not suitable for low-cost agricultural harvesting scenarios. In contrast, monocular 3D object detection methods have the advantage of only requiring a camera and being easy to deploy. However, there is a lack of specialized monocular 3D object detection datasets and algorithms suited for natural scenes in the agricultural field, which limits the application and development of this technology in agricultural automation. To address this, we construct a 3D object detection dataset for wax gourds and propose a network called TPDNet, which aims to capture the 3D information of objects from a single RGB image for fruits and vegetables in fields. Specifically, we construct a depth estimation and enhance module that introduces depth information into the model with the help of depth auxiliary labels, and improves the representation of depth information by utilizing weight information across spatial and channel dimensions. Meanwhile, since depth features and image features are heterogeneous, we design the phenotype aggregation and phenotype intensify module to capture the correspondence between image and depth features, promoting the effective fusion of image and depth information. The experimental results show that our method significantly outperforms others in terms of mAP _3D and mAP _BEV metrics, demonstrating the effectiveness and validity of our proposed method. We open our code and dataset at: http://tpdnet.samlab.cn.

Freeze injury during the seedling stage significantly impacts wheat growth and yield, making the development of freeze-tolerant varieties crucial for ensuring stable yields. To identify key genetic factors for wheat freeze tolerance, an accurate assessment of freeze tolerance is necessary. However, traditional methods, such as visual inspection, are subjective and can vary significantly among observers. In this study, we developed FreezeNet, a lightweight deep learning model designed to accurately quantify freeze injury using an image-based phenotyping method. Freeze tolerance traits, including vegetation area (VA), green vegetation area (GVA), yellow vegetation fraction (YVF), and mean hue value (mHue), were extracted for freeze tolerance assessment. We captured standardized images with a smartphone and used FreezeNet to extract the freeze tolerance traits for 220 wheat accessions. These traits were strongly correlated with traditional injury scores estimated through visual inspection. Moreover, they presented relatively high heritability. Using these traits, we conducted genome-wide association studies (GWASs) to identify genetic loci associated with freeze tolerance. Eleven significant QTLs associated with freeze tolerance were identified, including 8 novel loci. By integrating four of these loci into a wheat germplasm that lacked any of the 11 QTLs, we significantly enhanced its freeze resistance, demonstrating the practical application of these genetic loci in breeding for improved freeze tolerance. Our results highlight FreezeNet as an advanced tool for assessing wheat freeze injury and identifying the genetic factors responsible for freeze tolerance, with the potential to guide breeding efforts toward the development of more resilient wheat varieties.

Grape embryo rescue technology is currently the primary method for breeding new seedless grape cultivars. The timing of berry sampling directly impacts the efficacy of this technique. Therefore, achieving efficient, accurate, and non-destructive determination of the optimal sampling time for seedless grape embryo rescue breeding has long been a challenge. This study collected near-infrared spectral data and data on 19 physiological indicators from 2940 grape berries of six grape cultivars at six sampling times to construct a baseline dataset. Remarkably, it was discovered for the first time that pericarp puncture hardness (PPH) is closely associated with the embryo development rate of seedless grape. Subsequently, the optimal sampling times for 'Flame Seedless', 'Ruby Seedless', and 'Jingzaojing' were determined when their PPH reached 720 ​± ​20 ​g, 990 ​± ​20 ​g and 633 ​± ​20 ​g, respectively. Then, a total of 840 models for PPH recognition were established and assessed based on their coefficient of determination (R ²) and root mean square error (RMSE). The optimal recognition models for three seedless grape cultivars suitable for embryo rescue-'Flame Seedless', 'Ruby Seedless', and 'Jingzaojing'-were identified as follows: D1+PLSR (R ² ​= ​0.94, RMSE ​= ​42.26), D1+MLR (R ² ​= ​0.79, RMSE ​= ​66.31) and D1+PLSR (R ² ​= ​0.93, RMSE ​= ​47.9). Utilizing the established D1+PLSR or D1+MLR models for PPH, a non-destructive and precise method for sampling seedless grapes during embryo rescue was introduced for the first time. This approach led to a notable increase in the embryo development rate by 15 ​% and enhanced the plantlet rate by 14 ​%. Overall, our proposed strategy provides new perspectives for accelerating the breeding process of new seedless grape cultivars.

Assessing forest loss from snow and ice storms is vital for disaster evaluation and sustainable management. Traditional optical remote sensing methods, which focus on horizontal canopy changes, struggle to capture vertical stand alterations caused by snow and ice storms. This study introduces the LiDAR Forest Structure Change Index (LFSCI), a novel index that employs bitemporal unmanned aerial vehicle (UAV) LiDAR point data to comprehensively evaluate changes in the vertical distribution of forest stands. Following ice storms in Shizishan, Wuhan, China in early 2024, research was conducted to compare the performance of LFSCI with traditional metrics, such as canopy cover (CC), Leaf Area Index (LAI), and tree height (TH), across two spatial scales (grid and individual tree). LFSCI was evaluated at nine point densities (5-177 ​pt/m²). Through validation with field-measured stand volume changes from 43 plots, LFSCI showed superior correlation (R ² ​= ​0.64 for grids, 0.59 for trees) in comparison to CC (R ² ​= ​0.52), LAI (R ² ​= ​0.38), and TH (R ² ​= ​0.16). Higher point densities enhanced accuracy, with 50 ​pt/m² recommended for effective snow and ice storm impact detection. Pure broad-leaved forests were more susceptible to loss in comparison to mixed conifer-broadleaf forests, mixed broadleaf forests, and needle forests. Additionally, stands characterized by greater tree heights, steeper slopes, and shaded conditions were more vulnerable to damage than those in other environments.

Accurate assessment of plant leaf disease severity is crucial for implementing precision pesticide application, which in turn significantly enhances crop yields. Previous methods primarily rely on global perceptual learning, often leading to the misidentification of non-lesion regions as lesions within complex backgrounds, thereby compromising model accuracy. To address the challenge of background interference, we propose a location-guided lesion representation learning method (LLRL) based on image generation to assess the severity of plant leaf diseases. Our approach comprises three key parts: the image generation network (IG-Net), the location-guided lesion representation learning network (LGR-Net), and the hierarchical lesion fusion assessment network (HLFA-Net). IG-Net is designed to construct paired images necessary for LGR-Net by utilizing a diffusion model to generate diseased leaves from healthy ones. First, the LGR-Net facilitates the network's focus on the lesion area by contrasting paired images: healthy and diseased leaves, obtaining a pre-trained dual-branch feature encoder (DBF-Enc) that incorporates lesion-specific prior knowledge, providing focused visual features for HLFA-Net. Second, the HLFA-Net, which shares and freezes the DBF-Enc, further fuses and optimizes the features extracted by DBF-Enc, culminating in a precise classification of disease severity. In addition, we construct an image dataset containing three plant leaf diseases from apple, potato, and tomato plants, with a total of 12,098 photos, to evaluate our approach. Finally, experimental results demonstrate that our method outperforms existing classification models, with at least an improvement of 1 ​% in accuracy for severity assessment, underscoring the efficacy of the LLRL method in accurately identifying the severity of plant leaf diseases. Our code and dataset are available at http://llrl.samlab.cn/.