With increasing demands for sustainable food production, expediting innovation within the development of agricultural products is paramount for Bayer and similar companies. We used a novel gene-editing protocol that generates numerous events to increase our gene-editing capacity. TREDMIL was used create 800 distinct edits in soybeans at three Dt1 gRNA targets in over 1500 events sites distributed across 100 soy lines. With the ability to produce great numbers of edit events, our phenotypic testing had to evolve to keep pace. Using a hypothesis-based approach, we have refined phenotypic testing to measure relevant plant traits that impact yield. Our in-field phenotyping of the target set of plant traits feeds a machine learning model that adjusts small plot yield. Testing in both corn and soy has demonstrated small plots are predictive of large-scale yield testing (84% agreement) and that modeling with additional traits improved the predictive capacity to 93%.

In past global dust storms, no long lasting anomalies in the pressure cycle had been observed. The Global Dust Storm of Mars Year 34 (MY34), however, left behind an average surface pressure lower than what was expected based on the the values recorded on previous years by the Rover Environmental Monitoring Station (REMS) on Curiosity. The main signal contribution to the daily average surface pressure is the CO2 cycle, which is controlled by the Polar ice sublimation and freezing cycles. We used REMS and Mars Climate Sounder (MCS) data to search for correlations between the REMS anomaly and anomalies in the circulation compared to MCS observations from previous years. The findings include an early start of the retreat season for the Northern Polar cap, followed by the longest period of growth for the Southern Polar (SP) cap ice expansion since Curiosity had landed and then, during the dust storm, the longest retreat season of the Southern Polar cap. We also find a larger Northern Polar Cap extension after the storm, suggestive of a larger deposition of CO2 ice. The changes in length of the SP growth and retreat seasons might be consequence of the response of the zonal mean circulation to the dust storm. Changes in the structure of the zonal mean circulation compared to previous years are found in MCS data and presented. The combination of these anomalies constraint what physical processes may have caused this response in surface pressure after the dust storm.

In the realm of reservoir engineering, the application of machine learning has emerged as a transformative force, offering unprecedented insights into reservoir parameter characterization. In this study, we present a comprehensive analysis of four distinct machine learning models, namely Bagging, Extra Tree Regressor, XGBoost, and Ridge, to elucidate their efficacy in predicting permeability, a critical parameter for reservoir characterization. Our findings reveal an understanding of each model's performance. The Bagging model, while demonstrating an impressive trained accuracy of 0.99, exhibits some uncertainty in high permeability predictions, casting slight shadows on its applicability for reservoir characterization. In contrast, the Extra Tree Regressor model outshines the Bagging model with a trained accuracy of 100% and a prediction accuracy of 99.8%. It boasts lower absolute and absolute percentage errors, reinforcing its ability in permeability prediction. However, the XGBoost model takes a unique approach by emphasizing the density-corrected log over gamma-ray and sonic logs. Despite achieving remarkable trained and predicted data accuracy exceeding 99%, its reliance on the corrected density log introduces a mean absolute percentage error above 10, warranting closer scrutiny. In contrast, the Ridge model struggles, evident from its high AIC reading, signifying its limited compatibility with permeability prediction. Joint plots and LMplot analyses further showcases model behaviors. The Extra Tree model exhibits a 99% confidence interval, underscoring its reliability with minimal underpredictions. Conversely, the Bagging and Ridge models show susceptibility to high uncertainties in permeability predictions, particularly at extreme values. Our study concludes that the Extra Tree Regressor model excels in permeability prediction, with potential applications in reservoir interval assessments. The XGBoost model, while competent in sandstone reservoir prediction, bears a higher uncertainty burden. The Bagging and Ridge models, due to their uncertainty challenges, are less suitable for non-reservoir and sandstone reservoir interval predictions. High permeability correlations with elevated porosity, reduced water saturation, and lower gamma ray readings highlight the reservoir intervals' distinct characteristics. These observations underscore the reliability of our models and their potential contributions to reservoir engineering practices.

Non-destructive, real-time monitoring of root development can be helpful to farmers in improving crop resilience while minimizing resource use (Mervin et al., 2022). However, it is still an unexplored frontier in understanding root responses efficiently. In this study, we employed three in-soil fiber Bragg grating (FBG) based fiber sensors to generate root phenotyping data and developed an automated method using the deep learning architecture ResNet to monitor underground root development. In our preliminary study, we conducted a simulation experiment using two metal rods with diameters of 1mm and 5mm to mimic plant's roots. These rods were inserted to a depth of 15 cm in two different scenarios, 6 and 11 minutes, with the three in-soil FBG sensors continuously collecting data-two FBGs placed on the sides, and one placed at the bottom. The sensor data was preprocessed, resulting in 3228 samples for root diameter and 477 for root depth prediction models. We used an 80/20 split for training and testing the ResNet models to predict the artificial root diameter and ten different depth levels. The achieved accuracy was 0.95 for depth and 0.91 for diameter prediction. Overall, our study demonstrates the potential of ResNet architectures to accurately predict root depth and diameter with fiber optics-based sensors. Therefore, non-destructive root phenotyping in agricultural applications might be possible. Future work will involve evaluating these models in field experiments to assess their real-world performance.

Agriculture utilizes large quantities of freshwater resources to maintain crop production to a level that meets global demand. As the world population expands at a rapid rate, it is critical that we find more efficient ways to manage freshwater resources. The Plant DiTech Phenotyping Platform (Plant-DiTech LTD, Yavne, Israel) is a dynamic plant screening system that allows researchers to create controlled environments and rapidly collect a large quantity of physiological traits in real time. This platform opens the door to hundreds of potential experiments to explore growth trait responses to future environmental conditions and irrigation practices. Here, we test its utility on cultivated rice and upland cotton in a small pilot experiment. Cultivated rice (Oryza sativa ) and upland cotton (Gossypium hirsutim ) are both C3 crop species that are grown throughout the U.S. and many other countries and are critical for food and fiber production, respectively. Cotton is typically produced in the warm and dry regions of southern U.S. (USDA-ERS) where it heavily relies on irrigation practices to remain productive, while rice requires flooded conditions to maintain high productivity. Using these two test species for a preliminary experiment, a series of growth trait measurements in conjunction with system-recorded data by the Plant DiTech phenotyping platform are analyzed to test the linkages between organ- and canopy-level traits.

Prof. Roberto Grobman Keywords: skin, genetics, algorithms, biomarkers, wrinkles, aging, artificial intelligenceAbstractIntroduction: Skin, being the largest organ system in the body, is of utmost importance when it comes to timely diagnostics and treatment of skin conditions. Diagnostics, in history, have been dependent on symptoms and the doctor’s experience. Today, with advances in technology, is it possible to diagnose skin conditions more accurately and early. Skin imaging and deep learning have contributed immensely in very early diagnosis and hence a better prognosis. Artificial intelligence (AI) techniques have been applied in clinical genomics to identify genetic markers for predisposed conditions such as melanoma, psoriasis etc.Methods and results: Research and analysis of three studies were performed to obtain collective data on the current trends in skin disease diagnosis and mapping of genetic markers. AI shows a lot of promise in prediction of skin conditions and early treatment.Conclusion: Skin disease prognosis has been improved by the use of skinomics, microarray and AI techniques for accurate diagnostics and treatment.IntroductionThe skin is the largest organ of the body, composed of epidermis, dermis, and subcutaneous tissues, containing blood vessels, lymphatic vessels, nerves, and muscles, which can perspire, perceive the external temperature, and protect the body. Covering the entire body, the skin can protect multiple tissues and organs in the body from external invasions including artificial skin damage, chemical damage, adventitious viruses, and individuals’ immune system . Skin diseases have a big impact on everyday life and detecting underlying issues at the earliest is gaining importance. It is necessary to develop automatic methods in order to increase the accuracy of diagnosis for multitype skin diseases.Skin diseases and conditions are extremely prevalent, yet diagnostics are based on symptoms and the experience of the doctor. These are, often, not fool-proof and sometimes require a trial-and-error approach to diagnosis. Over the past few years, the image processing technique has achieved rapid development in medicine . A great example, the skin disease varicella was detected by Oyola and Arroyo through image processing technique’s colour transformation, equalization as well as edge detection, and the image of varicella was eventually collected and classified through Hough transform. The final empirical results demonstrated that a better diagnosis was received in terms of detection on varicella, and preliminary test was also conducted on varicella and herpes zoster on that basis. Sumithra et al. proposed a novel approach for automatic segmentation and classification of skin lesions by using SVM and k-nearest neighbor (k-NN) classifier. Kumar and Singh [20] established the relationship of skin cancer images across different types of neural network. Then, medical images were collected into this skin cancer classification system for training and testing based on the matlab image processing toolbox .Bioinformatics is a research field that uses computer‐based tools to investigate life sciences questions, employing “big data” results from large‐scale DNA sequencing, whole genomes, transcriptomes, metabolomes, populations, and biological systems, which can only be comprehensively viewed in silico. The epidermis was among the earliest targets of bioinformatics studies because it represents one of the most accessible targets for research. Consequently, bioinformatics methods in the fields of skin biology and dermatology generated a large volume of bioinformatics data, which led to origination of the term “skinomics.” Skinomics data are directed toward epidermal differentiation, malignancies, inflammation, allergens, and irritants, the effects of ultraviolet (UV) light, wound healing, the microbiome, stem cells, etc. Cultures of cutaneous cell types, keratinocytes, fibroblasts, melanocytes, etc., as well as skin from human volunteers and from animal models, have been extensively experimented on . We are presenting some combined research information on diagnostic imaging and application of bioinformatics in skin diseases through this article.Methods and resultsBioinformatics is an interdisciplinary field of knowledge that combines computer science, biology and biomedical sciences and statistics. Bioinformatics is oriented to the application and development of new computational methods to expand biological, biomedical or epidemiological knowledge.We used a data set provided by Transceptar Technologies/FullDNA, from Israel. The algorithm developed by Transceptar Technologies TRCPR18 has AI-based technology and allows the analysis of millions of data in a few seconds, taking into account the orientation of the gene and proceeding with various types of predisposition calculations. The Transceptar / FullDNA algorithm analyzes more than 61 skin-related conditions and this information was used to confirm previous research.Recent developments in high-speed technologies have led to a major revolution in biological and biomedical research and where today bioinformatics plays an increasingly central role in the analysis of large amounts of data.Literature from three studies were researched to summarise modern advances in skin disease diagnostics using Artificial Intelligence (AI), bioinformatics, skin imaging and machine learning.Imaging and deep learning applications:A study conducted by Patnaik et al. researched an approach to use various computer vision based techniques (deep learning) to automatically predict the various kinds of skin diseases. The system uses three publicly available image recognition architectures namely Inception V3, Inception Resnet V2, Mobile Net with modifications for skin disease application and successfully predicts the skin disease based on maximum voting from the three networks. The study approach involved development of a widespread plan to test the special features and general functionality on a range of platform combination, initiated by the test process. The method involves use of pre-trained image recognizers with modifications to identify skin images. The use of deep learning and ensembling features, results showed higher accuracy rate along with identification of more diseases. Previous models reported a maximum of six skin diseases with an accuracy level of 75% compared to as many as twenty diseases with an accuracy of 88%, in the study conducted by Patnaik et al. This proves that deep learning algorithms have a huge potential in the real world skin disease diagnosis .Microarray and skinomics applications:The most commonly used and highly preferred methodology in skinomics is DNA microarray technology, such as Affymetrix and Illumina. DNA microarrays are a perfect medium as they simultaneously measure the expression of the entire genome . Printed cDNA arrays, originated by Brown at Stanford , are often homemade, inexpensive, and can compare two samples on the same chip. Commercial alternatives such as oligonucleotide microarrays  are available too, but a little expensive. These techniques offer personalized medication and find broad applications in the future. Microarray technology can be applied in skin ageing studies, UV damage studies, transcriptional studies in melanoma and wound healing studies. Genome‐wide association studies, GWAS, comprise examination of many common DNA polymorphisms in a large population cohort to detect association of polymorphisms with a given disease. Such polymorphisms can point to the genes where disease‐causing mutations may map. GWAS are particularly useful in the analysis of diseases, such as psoriasis, which are common and with a strong genetic component .Artificial intelligence in clinical genomics:Most artificial intelligence techniques have been adapted to address the various steps involved in clinical genomic analysis—including variant calling, genome annotation, variant classification, and phenotype-to-genotype correspondence—and perhaps eventually they can also be applied for genotype-to-phenotype predictions . AI has proven to be highly effective in the following areas:Variant Calling : The clinical interpretation of genomes is sensitive to the identification of individual genetic variants among the millions populating each genome, necessitating extreme accuracy. Standard variant-calling tools are prone to systematic errors that are associated with the subtleties of sample preparation, sequencing technology, sequence context, and the sometimes unpredictable influence of biology such as somatic mosaicism . AI algorithms can learn these biases from a single genome with a known gold standard of reference variant calls and produce superior variant calls .Phenotype-to-genotype mapping : The molecular diagnosis of skin disease often requires both the identification of candidate pathogenic variants and a determination of the correspondence between the diseased individual’s phenotype and those expected to result from each candidate pathogenic variant. AI algorithms can significantly enhance the mapping of phenotype to genotype, especially through the extraction of higher-level diagnostic concepts that are embedded in medical images and EHRs .Genotype-to-phenotype prediction : The ultimate purpose of clinical genetics is to provide diagnoses and forecasts of future disease risk. Although, not many successful predictions have been made in literature yet, this shows promise in the fact that a few simple studies have shown to accurately predict conditions .Conclusion:AI systems have surpassed the performance of state-of-the-art methods and have gained FDA clearance for a variety of clinical diagnostics, especially imaging-based diagnostics. The availability of large datasets for training, together with advances in AI algorithms is driving this surge of productivity. Deep-learning algorithms have shown tremendous promise in a variety of clinical genomics tasks such as variant calling, genome annotation, and functional impact prediction. It is possible that more generalized AI tools will become the standard in these areas, especially for clinical genomics tasks where inference from complex data is a frequently recurring task .The application of AI in medicine is a burgeoning area of development in light of the major impact it could potentially have on healthcare provision. The application of machine learning in medical imaging on skin lesions  has been the most impactful, and demonstrates the potential for this technology in medical practice .

In recent years, the use of field-based high-throughput phenotyping (FHTP) has surged across diverse disciplines. Particularly, it has gained significant traction in agricultural research, enabling scientists to efficiently gather extensive data for a deeper understanding of plant biology in the context of plant growth dynamics. This abstract aims to demonstrate potential applications of data obtained through high-throughput phenotyping in the fields of plant biology and predictive plant breeding.The study utilized temporal phenotype data derived from repetitive drone flights equipped with various sensors. These data were incorporated into a novel mixed model, providing insights into the temporal genetic effects on different genotypes/plants. Gaussian or Lorentzian peak models, as well as Functional Principal Component analysis, were employed to characterize the growth patterns of various genotypes in diverse environments. The research revealed that Temporal Effect Sizes of Quantitative Trait Loci (QTLs) influence growth differently across time points, highlighting the dynamic nature of plant development. Furthermore, the study uncovered time-dependent associations between genotypes and their environments based on temporal phenotype values.The predictive capability of temporal phenomic data was found to surpass that of genomic data in predicting complex traits in maize. However, the combination of phenomic and genomic data consistently yielded the most accurate predictions for complex traits. By analyzing drone flights at specific growth stages, the study quantified physiological traits such as senescence progression across multiple time points. This analysis led to the calculation of new traits, including days to senescence and grain filing period, providing valuable insights into plant development and growth dynamics.

Wheat is an important primary crop that nourishes billions of people worldwide. Wheat diseases, particularly Fusarium head blight (FHB) disease, often have a severe effect on wheat yield in terms of both quantity and quality posing potential threats to the health of humans and livestock. Traditional methods such as field surveys for monitoring and assessing wheat diseases are time-consuming, costly and inefficient. In recent years, remote sensing approaches, particularly aerial imaging using Unmanned Aerial Vehicles (UAV), have become invaluable tools for rapid field scouting at larger scale, as well of crop growth and health status. This study aims to investigate the potential of combining high-resolution UAV multispectral imagery with machine learning (ML) methods for the estimation of FHB disease severity. Two experimental wheat fields were established at Volga, South Dakota, USA, in 2022. The severity of FHB disease was assessed and rated in the fields; and synchronous UAV flights were conducted to collect multispectral imagery. Canopy spectral and texture features were derived from the UAV multispectral imagery and used as input variables for ML models to predict FHB disease severity levels. Both classification and regression approaches were applied to estimation FHB severity using ML models such as Random Forest, Support Vector Machine, and Deep Neural Networks. The results show that both canopy spectral and texture features are important indicators for monitoring the severity of FHB wheat disease. Furthermore, the use of UAV remote sensing, combined with ML-based modeling, is a sustainable approach for rapid and accurate detection of wheat FHB disease severity.

Unoccupied / Unmanned / Uncrewed Aerial Systems (UAS, also known as drones) are tools that can provide field-based phenotyping and phenomics derived insights into plant breeding, biology, genetics and agronomy. There are many important, yet disparate agricultural UAS activities, occurring in silos which with better communication across research, education, and extension, could create transformative change for all stakeholders. The goal of this USDA-NIFA and AG2PI supported project is to advance knowledge and activities through promoting UAS data collection, processing, analysis, and community discussions. The objectives include: 1) Encourage collaboration pertaining to best practices between university, industry, and personal stakeholders who are currently developing and using UAS tools, ranging from beginner to experts. 2) Process Genomes to Fields (G2F) datasets (2017 to 2023, and up to eleven locations with ~6TB of data) into widely accessible and usable end products the community can directly use. And 3) Institute a user-centered webpage so that contributors and interested parties can access information about UAS based HTP conveniently. To accomplish the goals of the project, a UAS Project Coordinator will work to identify existing use of UAS in agriculture user groups, listen to seasoned advice and newcomer needs, work as a liaison to conduct the sharing of wisdom, and summarize their knowledge and share personal connections across discipline, institutions, and species; making this information easier to access in the future. The project is working to ease the learning process of beginners using UAS tools, as well as sharing the advancements of experienced users on the forefront of innovation and discovery.

Phosphorus (P) is a vital macronutrient for building up essential biomolecules in plants, and its accurate quantification can guide effective crop management and increase crop growers’ profit. Traditional chemical reaction-based methods for measuring P levels in plants are destructive and complex. Hyperspectral imaging offers a real-time, non-destructive avenue for assessing crop nutrient status. While these images are rich in both spatial and spectral information, limitations in current devices and analytical algorithms have led most studies to concentrate solely on the spectral features. In this study, a novel algorithm to combine features in spatial and spectral domains is proposed and implemented to differentiate phosphorus deficiency symptoms in corn plants. At the V6 vegetative stage, leaf-level hyperspectral images from three P levels and two leaf positions were collected using the handheld proximal hyperspectral imager, LeafSpec. Spatial and spectral features that exhibit significant differences between the P treatments were generated by integrating pre-designed spatial partitions with spectral index maps. The correlation coefficient between the P content and elected spatial and spectral features was treated as the standard to further refine the mining results. The spatial and spectral joint effects showed superior ability than spectral indices in differentiating P deficiency at both leaf positions, especially for medium P and sufficient P. Related visualization maps also gave out a preliminary insight into the differences in P deficiency symptoms. This study highlights the great potential and effectiveness of combining spatial and spectral features in differentiating P levels at corn’s early vegetative stage.

Phenotyping plays a crucial role in parameterizing, calibrating, and evaluating process-based plant models, which can be used to understand and predict crop behavior under various conditions. This, in turn, allows for more informed decision-making in agriculture and provides insights into how crops may respond to changing environmental factors. Nevertheless, several potential disparities between current plant models and phenotyping methods exist, which have the potential to undermine the precision and applicability of these models. These discrepancies encompass differences in data resolution and scale (both spatial and temporal) between what is feasibly gathered during phenotyping efforts and what is required by the models. Additionally, issues related to compatibility between the traits measured during phenotyping campaigns and the parameters required for models are prevalent. Furthermore, the representation of genetic and environmental variability in both models and phenotyping data is often limited, resulting in a gap between these two components.To address these discrepancies, we are developing a new “minimum” plant model. This model has been designed with strong consideration of what data can be measured in phenotyping campaigns across multiple genotypes. While the core model remains straightforward, it has the capability to simulate essential processes to represent fundamental plant physiology. Moreover, it has the potential to be integrated into larger, more comprehensive models or parameterized to suit specific genotypes. This approach aims to bridge the gap between phenotyping and crop models, thus enhancing their effectiveness in addressing the challenges that emerge by utilizing them in broader agricultural applications.

Deep learning is a central tool in plant phenotyping. Proficiency in artificial intelligence and programming remains essential for designing a deep learning model. The reuse of deep models can be challenging for non-coding end-user such as plant biologists. Currently, very few tools facilitate collaboration between computational infrastructure, non-coding end-users, and deep learning project manager. Consequently, this limits the deployment of multi centric large scale deep learning initiatives (such as the notable exception of the Globally Wheat Challenge) in plant phenotyping. We propose the scheme of Fig. 1 to allow non-coding end-users to test existing deep learning models with a homemade software capable of running any segmentation, classification or object detection deep learning model. This software is an open source plugin coined MANINI (https://github.com/hereariim/manini) which runs under the north American initiative NAPARI (https://napari.org/). The plugin also enables the manual correction of the inference. The corrected inferences can then be used to retrain or fine-tune the model on a large-scale infrastructure (in our case the European Grid Inrastructure EGI) via the DEEPaaS API. The resulting new model is then accessible on a public repository. We illustrate the interest of this scheme on various plant phenotyping use cases.

Comparing the genomes of two species offers a robust approach to unveil potentially overlooked genes that wield substantial influence on observed agronomic traits. Extensive phenotypic data from maize and sorghum were previously utilized in genome-wide association studies. This project aims to harness the potential of comparative genomics to strengthen confidence in these marker-trait associations and to suggest previously unexplored relationships. To achieve this goal, insights from studies on the genetically diverse Sorghum Association Panel and the maize Wisconsin Diversity Panel were leveraged to classify candidate genes as either shared between the two species or unique to one or the other. Candidate orthologous genes found in both species, associated with shared phenotypes, enhance the reliability of the associations within each species. Additionally, genes unique to each species provide parameters to inform future predictive models. Finally, given the ancient tetraploidy of maize and the biased loss of genes over time, the candidate genes identified in sorghum provide valuable information for understanding the orthologs found in the maize subgenomes.

Enhancing photosynthesis for increased sorghum grain yield has become a key focus in sorghum breeding efforts. Phenotyping, involving the measurement of various morpho-physiological and physical traits associated with photosynthesis and grain yield, is a time-intensive process. However, the potential of non-invasive leaf-level hyperspectral imaging to swiftly detect plant performance, optimizing photosynthesis and grain yield, is promising. This study aimed to evaluate the feasibility of utilizing hyperspectral reflectance in the 350–950 nm range for the rapid estimation of these traits in intact sorghum leaves. Multiple machine learning regression algorithms were developed using leaf-level hyperspectral reflectance data from nearly 400 sorghum accessions within an association panel. The best-performing prediction models were then considered as potential methods for constructing a prediction model targeting multiple other physiological and yield traits in sorghum accessions. The results indicate that this approach enables the early detection of leaf photosynthetic and yield traits through leaf-level hyperspectral reflectance without the need for a full-range, high-cost leaf spectrometer.

Plot extraction for field trials plays a foundational role in supporting various research and development related to agricultural applications, including the classification of crop types, estimation of crop yields, and monitoring the health of crops. Traditional methods employed to define these plots have a trade-off between the substantial human labor involved and the precision of the delineated plots. In our research, we introduce a semi-automatic framework for plot extraction that requires only two essential inputs: the width and height of the plot. Our framework leverages the Segment Anything Model (SAM) for image segmentation, producing masks that are subsequently converted into polygons. These generated polygons are then filtered based on the user input. We refine the positions and orientations of these polygons by maximizing their overlap with the actual plot field. Experiment results were evaluated by comparing the extracted boundaries with manually digitized ground truth data. The results of our study demonstrate the successful extraction of individual plots across various fields characterized by diverse crop types. It is our expectation that this framework will significantly reduce the manual labor required for precise plot extraction, thus enhancing the ease and efficiency of this critical prerequisite task.

Source-Sink Regulated Senescence (SSRS) in maize is a complex trait involving sugar sensing and carbohydrate partitioning. While the yield effects of SSRS are still under investigation, mapping work in Midwest commercial germplasm has found evidence for QTL on several chromosomes. In this study, we seek to understand the variation of this trait in global maize germplasm and map the genetic loci linked to this senescence response using the Maize Nested Association Mapping population. Wide variation was seen in expression of the senescence phenotype including several lines that are potentially anti-senescent in response to sink removal. Phenotypes of NAM RIL populations support the possibility of anti-senescent or senescence suppressing genetic factors. Joint linkage mapping identified QTL for the SSRS trait on three chromosomes which are distinct from those previously reported. By understanding and mapping the global diversity of this trait, we can better understand the physiology of maize senescence and integrate insights into breeding programs.

Recent progress in proximal remote sensing has elevated both the spatial and temporal resolution of data acquisition, expanding the accessibility of these technologies for digital agriculture applications. These advanced sensors enable the gathering of extensive and novel datasets, proving instrumental in accurately characterizing phenotypes and parameterizing models for crop growth. Despite the distinctive structural, spatial, and spectral information embedded in these data streams, they have predominantly been utilized in isolation. Thus, this research aims to integrate these disparate data sources to improve estimations of agronomically important crop traits, such as yield. Deep learning methods, such as autoencoders, will be used to extract latent phenotypes, which will be used to characterize manually measured traits. We focus on multispectral images (MSIs) collected by unoccupied aerial vehicles and lidar scans collected by unoccupied ground vehicles. MSIs capture canopy-level spectral information, including the red, green, blue, red edge, and near infrared bands. Lidar scans are converted to point clouds to construct the three-dimensional sub-canopy architecture of maize plants. Data were collected on maize hybrids as part of the Genomes to Fields project, from 2018 to 2022, in Aurora, NY. Autoencoder model training on MSIs shows that latent phenotypes are effective image representations, containing relevant and sufficient information to generate image reconstructions. The latent codes are also predictive of the image date and normalized difference vegetation index values. Latent phenotypes were extracted from the lidar point clouds as well, and the prediction accuracies of models using these measurements separately and jointly will be compared.

Forest ecosystems are the largest terrestrial carbon sink and monitoring them effectively, particularly in the context of global change, requires rapid and accurate determination of tree functional traits that indicate forest health. Hyperspectral reflectance has the capacity to predict leaf traits non-destructively using multivariate statistical approaches. The ability of hyperspectral data to estimate tree physiochemical responses is affected by wavelength range selection and the influence of wavelength on accuracy of trait estimation is not well known, especially for more complex physiological processes. To bridge this knowledge gap, this study examined chemical and physiological responses of one-year-old black walnut (Juglans nigra L.) and northern red oak (Quercus rubra L.) to a combination of biotic and abiotic stress events, including pathogen inoculation, water stress, nutrient deficiency, and salt deposition. Leaf photosynthetic-related, water-related, and chemical traits were paired with hyperspectral measurements spanning 350−2500 nm. A total of 100 different wavelength ranges were evaluated using PLSR to determine the variation prediction accuracy. Key findings indicated that incorporating short infrared wavelength ranges (1300−2500 nm) significantly enhanced trait prediction accuracy. In addition, this study also demonstrated that hyperspectral data can detect tree stress responses at fine-scale chemical and physiological levels that are in agreement with reference measurement responses to the different stressors. We suggest that hyperspectral reflectance can potentially offer a solution for monitoring forest health in multi-stress environments with an increase in the accuracy of trait estimations and an expansion of the classes of traits estimated.