Background Hyperspectral imaging is emerging as a promising approach for plant disease identification. The large and possibly redundant information contained in hyperspectral data cubes makes deep learning based identification of plant diseases a natural fit. Here, we deploy a novel 3D deep convolutional neural network (DCNN) that directly assimilates the hyperspectral data. Furthermore, we interrogate the learnt model to produce physiologically meaningful explanations. We focus on an economically important disease, charcoal rot, which is a soil borne fungal disease that affects the yield of soybean crops worldwide. Results Based on hyperspectral imaging of inoculated and mock-inoculated stem images, our 3D DCNN has a classification accuracy of 95.73% and an infected class F1 score of 0.87. Using the concept of a saliency map, we visualize the most sensitive pixel locations, and show that the spatial regions with visible disease symptoms are overwhelmingly chosen by the model for classification. We also find that the most sensitive wavelengths used by the model for classification are in the near infrared region (NIR), which is also the commonly used spectral range for determining the vegetative health of a plant. Conclusion The use of an explainable deep learning model not only provides high accuracy, but also provides physiological insight into model predictions, thus generating confidence in model predictions. These explained predictions lend themselves for eventual use in precision agriculture and research application using automated phenotyping platforms.

Plant stress phenotyping is essential to select stress-resistant varieties and develop better stress-management strategies. Standardization of visual assessments and deployment of imaging techniques have improved the accuracy and reliability of stress assessment in comparison with unaided visual measurement. The growing capabilities of machine learning (ML) methods in conjunction with image-based phenotyping can extract new insights from curated, annotated, and high-dimensional datasets across varied crops and stresses. We propose an overarching strategy for utilizing ML techniques that methodically enables the application of plant stress phenotyping at multiple scales across different types of stresses, program goals, and environments. Understanding Plant Stress Is Crucial for Yield ProtectionPlant stress is a state of plant growth under non-ideal environmental conditions caused by various biotic (pathogen, insect, pest, and weed) and abiotic (temperature stress, nutrient deficiency, toxicity, herbicide) factors. Significant crop yield loss due to various plant stresses has the potential to threaten global food security [1]. Plant disease epidemics are a constant threat and continue to emerge owing to complex host-pathogen-environment dynamics [2,3]. Global climate change can exacerbate this situation because of the predicted increases in insect and pathogen pressure for major grains including rice (Oryza sativa), maize (Zea mays L.), and wheat (Triticum aestivum) [4]. Moreover, weather-related challenges such as drought, flooding, hail, and windstorms adversely affect crop production. Yield preservation and protection is a dynamic challenge for pathologists, entomologists, plant breeders, and crop producers globally. Understanding plant stress is crucial for improving yield protection to meet with the growing demand for food production [5]. In the past decade, significant advances in image processing and machine learning (ML; see Glossary) algorithms have been made to handle image-based stress datasets for automated data analysis and application of trained models [6][7][8]. We review the development and application of ML algorithms for image-based plant stress phenotyping at multiple scales ranging from leaf and canopy (plot) to field (production) scale. We discuss some of the major challenges in the practical application of ML algorithms, and list future efforts that will be necessary to make ML a more mainstream tool in plant stress phenotyping applications and usage. HighlightsPlant stress phenotyping is challenging to implement at multiple organizational scales (leaf, canopy, field).There is a need to improve the speed, accuracy, reliability, and scalability of stress phenotyping while allowing flexibility for highly variable program goals.

Humans draw maps when communicating about places or verbally describe routes between locations. Honeybees communicate places by encoding distance and direction in their waggle dances. Controversy exists not only about the structure of spatial memory but also about the efficiency of dance communication. Some of these uncertainties were resolved by studies in which recruits' flights were monitored using harmonic radar. We asked whether the two sources of vector information--the previously learned flight vector to a food source and the communicated vector--are represented in a common frame of spatial reference. We found that recruits redirect their outbound flights and perform novel shortcut flights between the communicated and learned locations in both directions. Guidance by beacons at the respective locations or by the panorama of the horizon was excluded. These findings indicate a spatial reference based on either large-scale vector integration or a common geocentric map-like spatial memory. Both models predict a memory structure that stores the spatial layout in such a way that decisions are made according to estimated distances and directions. The models differ with respect to the role of landmarks and the time of learning of spatial relations.

BackgroundCharcoal rot is a fungal disease that thrives in warm dry conditions and affects the yield of soybeans and other important agronomic crops worldwide. There is a need for robust, automatic and consistent early detection and quantification of disease symptoms which are important in breeding programs for the development of improved cultivars and in crop production for the implementation of disease control measures for yield protection. Current methods of plant disease phenotyping are predominantly visual and hence are slow and prone to human error and variation. There has been increasing interest in hyperspectral imaging applications for early detection of disease symptoms. However, the high dimensionality of hyperspectral data makes it very important to have an efficient analysis pipeline in place for the identification of disease so that effective crop management decisions can be made. The focus of this work is to determine the minimal number of most effective hyperspectral wavebands that can distinguish between healthy and diseased soybean stem specimens early on in the growing season for proper management of the disease. 111 hyperspectral data cubes representing healthy and infected stems were captured at 3, 6, 9, 12, and 15 days after inoculation. We utilized inoculated and control specimens from 4 different genotypes. Each hyperspectral image was captured at 240 different wavelengths in the range of 383–1032 nm. We formulated the identification of best waveband combination from 240 wavebands as an optimization problem. We used a combination of genetic algorithm as an optimizer and support vector machines as a classifier for the identification of maximally-effective waveband combination.ResultsA binary classification between healthy and infected soybean stem samples using the selected six waveband combination (475.56, 548.91, 652.14, 516.31, 720.05, 915.64 nm) obtained a classification accuracy of 97% for the infected class. Furthermore, we achieved a classification accuracy of 90.91% for test samples from 3 days after inoculation using the selected six waveband combination.ConclusionsThe results demonstrated that these carefully-chosen wavebands are more informative than RGB images alone and enable early identification of charcoal rot infection in soybean. The selected wavebands could be used in a multispectral camera for remote identification of charcoal rot infection in soybean.

We report a meta-Genome Wide Association Study involving 73 published studies in soybean (Glycine max L. [Merr.]) covering 17,556 unique accessions, with improved statistical power for robust detection of loci associated with a broad range of traits. De novo GWAS and meta-analysis were conducted for composition traits including fatty acid and amino acid composition traits, disease resistance traits, and agronomic traits including seed yield, plant height, stem lodging, seed weight, seed mottling, seed quality, flowering timing, and pod shattering. To examine differences in detectability and test statistical power between single- and multi-environment GWAS, comparison of meta-GWAS results to those from the constituent experiments were performed. Using meta-GWAS analysis and the analysis of individual studies, we report 483 peaks at 393 unique loci. Using stringent criteria to detect significant marker trait associations, 59 candidate genes were identified, including 17 agronomic traits loci, 19 for seed related traits, and 33 for disease reaction traits. This study identified potentially valuable candidate genes that affect multiple traits. The success in narrowing down the genomic region for some loci through overlapping mapping results of multiple studies is a promising avenue for community-based studies and plant breeding applications.