Of immediate widespread concern is the accelerating transition from Holocene-like weather patterns to unknown, and likely unstable, Anthropocene patterns. A fell example is irreversible Arctic phase change. It is not clear if existing AOGCMs are adequate to model anticipated global impacts in detail; however, the GISS ModelE AOGCM can be used to locally compare and extend the PIOMAS Arctic ocean historical ice-volume dataset into the near future. Arctic Amplification (AA) mechanisms are poorly understood; to enable timely results, a simple linear, Arctic TOA grid-boundary energy-input is used to enforce AA, avoiding the perils of arbitrary modification of relatively well-studied parameterizations (e.g., restriction of cloud-top height to induce local warming). Only PIOMAS springtime/max and fall/min Arctic ice-volume decadal, linear trends were enforced. This temporally-broad grid-boundary modification produces a surprisingly detailed consonance with 10 out of 12 temporal profiles falling within 1-sigma of PIOMAS temporal data for the entire history modeled (2003 to 2021). The data are then integrated to 2050. The result is a zero-ice-volume, summer/fall half-year, beginning ca. 2035 (onset 1-sigma of ± ~5 years), with mean annual Arctic temperatures increasingly trending above freezing. Persistent, Arctic phase change follows this half-year transition about 20 years later. Also present in later stages, the 500 hPa height minimum is no longer nearly-coincident with the pole, suggesting jet stream disruption and its consequences. Hypothesized large clathrate-methane releases likely associated with Arctic temperature and phase change are also examined. A basic assumption is that the Arctic ice (i.e., temperature) must be preserved at all costs. This work establishes a reasonably detailed timeline for the Arctic phase change based on well-studied AOGCM physics, slightly tuned to decades of PIOMAS data. This result also points to the Arctic as a key, near-term site for localized, nondestructive intervention to mitigate Arctic phase change (e.g., Stjern ), thereby slowing the Holocene -> Anthropocene growing-season disruption. Although such an intervention cannot itself accomplish the requirements of the IPCC SP-15 , nor Planetary Boundaries theory, delaying the Arctic phase change will likely extend the time-window for accomplishing those critical tasks and ultimately to at least slow the rate of increase of climate emergencies.
Forest cover and streamflow are generally expected to vary inversely because reduced forest cover typically leads to less transpiration and interception. However, recent studies in the western US have found no change or even decreased streamflow following forest disturbance due to drought and insect epidemics. We investigated streamflow response to forest cover change using hydrologic, climatic, and forest data for 159 watersheds in the western US from the CAMELS dataset for the period 2000-2019. Forest change and disturbance were quantified in terms of net tree growth (total growth volume minus mortality volume) and mean annual mortality rates, respectively, from the US Forest Service’s Forest Inventory and Analysis database. Annual streamflow was analyzed using multiple methods: Mann-Kendall trend analysis, time trend analysis to quantify change not attributable to annual precipitation and temperature, and multiple regression to quantify contributions of climate, mortality, and aridity. Many watersheds exhibited decreased annual streamflow even as forest cover decreased. Time trend analysis identified decreased streamflow not attributable to precipitation and temperature changes in many disturbed watersheds, yet streamflow change was not consistently related to disturbance, suggesting drivers other than disturbance, precipitation, and temperature. Multiple regression analysis indicated that although change in streamflow is significantly related to tree mortality, the direction of this effect depends on aridity. Specifically, forest disturbances in wet, energy-limited watersheds (i.e., where annual potential evapotranspiration is less than annual precipitation) tended to increase streamflow, while post-disturbance streamflow more frequently decreased in dry water-limited watersheds (where the potential evapotranspiration to precipitation ratio exceeds 2.35).
Current soil C inventories focus on surface layers although over half of soil C is found below 20 cm. Recent and ongoing changes in agricultural management, crop productivity, and climate in Midwest US cropland may influence subsoil C stocks. The objectives of this study were to determine how surface soil and subsoil organic C stocks have changed in croplands of Iowa and Illinois and to evaluate mechanisms to explain the observed subsoil organic C changes. Using resampling studies from Iowa and Illinois, we found that subsoil (20-80 cm) organic C increased at a rate of 0.31 Mg C ha-1 yr-1 between the 1950s and early 2000s despite C losses of similar magnitude in the top 20 cm (0.26 Mg C ha-1 yr-1). Based on this analysis, we estimated a subsoil C storage rate of up to 11.8 Tg C yr-1 for Iowa and Illinois, which equates to 12% of annual US greenhouse gas emissions from crop cultivation if surface C losses and non-CO2 greenhouse gases are controlled. We also measured changes in soil organic C stocks from two long-term cropping systems experiments located in Iowa, which demonstrated similar rates of subsoil C changes for both historical and contemporary crop rotations. Using publicly available crop yield data, we determined that changes in crop productivity likely contributed minorly to observed changes in subsoil organic C. The accumulation of organic C in subsoils may be attributed to regional climate change, which has led to greater precipitation and wetter subsoils that inhibit transformation of soil organic C to CO2. Because farmers may respond to increasing soil wetness by expanding and intensifying artificial drainage infrastructure, there is an urgent need to further assess subsoil C stocks and their vulnerability to drainage system changes.
Recent Innovations in Precision Agriculture (PA) are driven by Computer Vision and Data Processing systems to quantify plant parameters. Quantitative analysis of Plant Phenotyping in PA and monitoring morphological traits is a protracting process, precluding the objective and phenotyping pipeline. Greenhouses growing Genetically Modified (GM) crops need to be maintained at constant environmental and simulated conditions. Multiple parameters have to be controlled and regulated inside a greenhouse for effective growth of crops and yield maximisation. Not at all times are these factors derived and so, yield maximisation in greenhouse is an experimental approach to new varieties. For deduced environmental parameters and conditions for certain crops, few other biotic and abiotic factors can hinder or affect growth in certain ways that are not always factored in during calculating parameters conducive for plant growth. Such factors may not always be affecting parametral calculations, but transpose visual cues on plant growth environment such as spectral change in soil values, or minute changes like leaf reflectance or visible changes in plant stimuli to biotic factors. Plant growth is inclusive of multiple environmental variables, and yield maximisation approaches are experimental to finding the optimum derived value for these variables. Computer Vision provides a catalytic approach to predicting optimum parameters for yield maximization in phenomics. Computer Vision and Generative Adversarial Networks (GAN)’s offer a catalytic approach to the time-consuming process, providing a solution to the phenotyping bottleneck. This research proposes a concept of curating data of plant growth over time to predict conditional growth and responsive stimuli of the plant under different situations and how this can affect crop yield. The method proposed here is a non-invasive approach to the existing destructive biomass estimation methods and Frameworks. This methodology of the research focuses on utilizing image parameters modelled using a time series Progressively Growing Generative Adversarial Networks PGGAN to map plant growth patterns and progressive variance in biomass of plant in the Spatio-Temporal Domain. These Generative networks evaluate and predict based on merely raw pixel input excluding dependence on further constraints, feature vectors or parameters influencing data.
Roots are the interface between the plant and the soil and play a central role in multiple ecosystem processes. With intensification of agricultural practices, rhizosphere processes are being disrupted and are causing degradation of the physical, chemical, and biotic properties of soil. Improvement of ecosystem service performance is rarely considered as a breeding trait due to the complexities and challenges of belowground evaluation. Advancements in root phenotyping and genetic tools are critical in accelerating ecosystem service improvement in cover crops. Here I will present root phenotyping approaches for assessing ecosystem service in a prospective cash cover crop; pennycress (Thlaspi arvense L.). In development is a large format mesocosm system that will allow 3D root system architecture analysis of multiple plants. Using this system, we will be assessing how variation in pennycress root system architecture can affect ecosystem service and abiotic stress tolerance with the plant to scale from single plant to canopy level traits.
The Angstrom-Prescott (A-P) model is widely suggested for estimating solar radiation (Rs) in areas without measured or deficiency of data. The coefficients of this model must be locally calibrated, to calculate evapotranspiration (ET) correctly. The aim of this research was calibration and validation of the coefficients of the A-P model at six meteorological stations across arid and semi-arid regions of Iran. This model was improved by adding the air temperature and relative humidity terms. Besides, the coefficients (’a’ and ‘b’) of the A-P model and improved models was calibrated using some optimization algorithms including Harmony Search (HS) and Shuffled Complex Evolution (SCE). Performance indices, i.e., Root Mean Square Error (RMSE), Mean Bias Error (MBE), and coefficient of determination (R2) were used to analyze the models ability in estimating Rs. The results indicated that the performance of the A-P model had more precision and less error than improved models in all the stations. In addition, the best results were obtained for the A-P model with the SCE algorithm. The RMSE varies between 0.82 and 2.67 MJ m-2 day-1 for the A-P model with the SCE algorithm in the calibration phase. In the SCE algorithm, the values of RMSE had decreased about 4% and 7% for Mashhad and Kerman stations in the calibration phase compared to the HS algorithm, respectively. In other words, the highest decrease of RMSE is related to Kerman station.
The United Nations 2030 Agenda brings a holistic and multi-sectoral view on sustainability via the Sustainable Development Goals (SDGs). However, a successful implementation of this agenda is contingent on understanding the multiple, complex interactions among SDGs, including both synergies and trade-offs, for informing planning for sustainability at the local level. Using a case study in the Goulburn-Murray region in Victoria, Australia, we prioritised global goals and targets for the local context, characterised the interactions between them, analysed the main synergies and trade-offs, and identified potential policy solutions to achieve local sustainability. We identified the five highest priority SDGs for the region as clean water and sanitation (SDG 6), agricultural activities (SDG 2), economic growth (SDG 8), climate action (SDG 13), and life on land (SDG 15). Across these five priority SDGs and their 45 targets, we found 307 potential interactions, of which 126 (41%) were synergistic, 19 (6%) were trade-offs, and 162 (53%) were benign. We highlight the most salient trade-offs, particularly how unsustainable agricultural practices could negatively affect water resources, the environment, and sustainable economic growth. Also, critical ongoing uncertainties like climate change, local policies on environmental water recovery, international markets, and emerging new technologies could pose risks for the future of agriculture and the economy. Our results provide important insights for local and regional sustainability policy and planning across multiple sectors. Our methodology is also broadly applicable for prioritising SDGs and assessing their interactions at local scales, thereby supporting evidence-based policy-making for the SDGs.
Canola has a prominent floral signature and requires careful consideration when selecting spectral indices for yield estimation. This study evaluated several spectral indices derived from high-resolution RGB images. A small plot (2.75m x 6m) experiment was conducted at Kernen Research Farm, Saskatoon, where canola was grown under varying row spacings and seeding rates (192 plots). The canopy reflectance was imaged during the flowering period and seed yield was obtained at physiological maturity. Indices were evaluated for accuracy in quantifying canola flowers in high-resolution RGB imagery with within-canopy shadow pixels. Digitalized flower number from the peak flowering date was used to test and validate a non-linear three-parameter asymptotic regression model to simulate canola seed yield. 70 % of the data was used to develop the model, and 30 % was used to validate the model. Model performance was tested with Pseudo-R2, r, MAE, and RMSE. HFI (High-resolution Flowering Index) and MYI (Modified Yellowness Index) were able to accurately identify flowering pixels with the least amount of error pixel. The yield simulation model resulted in a pseudo-R2 value of 0.11 for the tested model and, correlation of 0.88 for validation with RMSE and MAE of 343.1 and 265.3, respectively. Our results indicate that HFI and MYI indices are a better indicator of yield potential compared to NDYI as the metric is well capable of handling within canopy shadows. Further studies are necessary to evaluate the performance of HFI and MYI vegetation indices for medium resolution-UAV and satellite imagery.
High-throughput plant phenotyping is increasingly implemented in a wide array of experimentation and presents challenges both logistically and analytically. Phenotype data are often longitudinal and proper modeling of plant growth requires sophisticated modeling techniques to account for the intra-plant correlations and changing variation over time (heteroskedasticity). For this reason, plant growth is often analyzed by comparing only single time points or the start and end points for inference with no regard for the trends themselves. Single time point analysis can be sufficient for simple biological comparisons, but modeling has the potential to unlock additional insights by utilizing all the information at hand. Current plant growth modeling strategies do account for intra-plant correlations but are still limited to constant variance assumptions and therefore perform sub-optimally. Here we propose a Bayesian hierarchical approach as an alternative method for plant growth modeling by demonstrating the utility of heteroskedastic sub-model parameterizations. We show that accounting for heteroskedasticity greatly improves model accuracy and subsequent inference. Additionally, Bayesian methodologies inherently lend themselves to near real-time model updating and we propose integration with Clowder to facilitate adaptive experimental designs. We show by example the utility of Bayesian updating and how it relates to experimental decision making.
Mung bean [Vigna radiata (L.) Wilczek] is a drought-tolerant, short-duration crop, and a rich source of protein and other valuable minerals, vitamins, and antioxidants. The main objectives of this research were (1) to study the root traits related with the phenotypic and genetic diversity of 375 mung bean genotypes of the Iowa (IA) diversity panel and (2) to conduct genome-wide association studies of root-related traits using the Automated Root Image Analysis (ARIA) software. We collected over 9,000 digital images at three-time points (days 12, 15, and 18 after germination). A broad sense heritability for days 15 (0.22–0.73) and 18 (0.23–0.87) was higher than that for day 12 (0.24–0.51). We also reported root ideotype classification, i.e., PI425425 (India), PI425045 (Philippines), PI425551 (Korea), PI264686 (Philippines), and PI425085 (Sri Lanka) that emerged as the top five in the topsoil foraging category, while PI425594 (unknown origin), PI425599 (Thailand), PI425610 (Afghanistan), PI425485 (India), and AVMU0201 (Taiwan) were top five in the drought-tolerant and nutrient uptake “steep, cheap, and deep” ideotype. We identified promising genotypes that can help diversify the gene pool of mung bean breeding stocks and will be useful for further field testing. Using association studies, we identified markers showing significant associations with the lateral root angle (LRA) on chromosomes 2, 6, 7, and 11, length distribution (LED) on chromosome 8, and total root length-growth rate (TRL_GR), volume (VOL), and total dry weight (TDW) on chromosomes 3 and 5. We discussed genes that are potential candidates from these regions. We reported beta-galactosidase 3 associated with the LRA, which has previously been implicated in the adventitious root development via transcriptomic studies in mung bean. Results from this work on the phenotypic characterization, root-based ideotype categories, and significant molecular markers associated with important traits will be useful for the marker-assisted selection and mung bean improvement through breeding.
For field workers around the world, wheat trials are often synonymous with wheat heads counting: a tedious but important task to measure this important yield component. Deep Learning has been a promising solution to automate the acquisition of wheat head density from a high-throughput phenotyping system, but it has been shown to be sensitive to changing acquisition conditions, also known as “domain change.” In response, an international collaboration built the “Global Wheat Head Dataset” in 2020 and 2021, a collection of 6515 images acquired during 47 different acquisition sessions in 12 countries. In addition to these datasets, two data competitions were held in 2020 (Kaggle, over 2,200 competitors) and 2021 (AIcrowd, over 400 competitors). The winning solutions are expected to be usable in plant phenotyping pipelines to robustly assess wheat spike density. We tested this hypothesis by evaluating the 2021 winning solution on an independent dataset consisting of images measured both in the field and in the image by a human, taken with the same acquisition protocol. We use triple collocation analysis to demonstrate that the predicted density appears to be more reliable than the human density measured in the field and in the image. Furthermore, we demonstrate that Global Wheat Head Dataset can be used to estimate wheat ear density from a drone.
Deep learning (DL) methods have transformed the way we extract plant traits – both under laboratory as well as field conditions. Evidence suggests that “well-trained” DL models can significantly simplify and accelerate trait extraction as well as expand the suite of extractable traits. Training a DL model typically requires the availability of copious amounts of annotated data; however, creating large-scale annotated dataset requires non-trivial efforts, time, and resources. This has become a major bottleneck in deploying DL tools in practice. Self-supervised learning (SSL) methods give exciting solution to this problem, as these methods use unlabeled data to produce pretrained models for subsequent fine-tuning on labeled data, and have demonstrated superior transfer learning performance on down-stream classification tasks. We investigated the application of SSL methods for plant stress classification using few labels. Plant stress classification is a fundamentally challenging problem in that (1) disease classification may depend on abnormalities in a small number of pixels, (2) high data imbalance across different classes, and (3) there are fewer annotated and available plant stress images than in other domains. We compared four different types of SSL methods on two different plant stress datasets. We report that pre-training on unlabeled plant stress images significantly outperforms transfer learning methods using random initialization for plant stress classification. In summary, SSL based model initialization and data curation improves annotation efficiency for plant stress classification tasks.
Crop pest detection and mitigation remains an extremely challenging task for the farmers. Majority of the pest classification and detection techniques rely on supervised deep learning frameworks that require significant human intervention in labeling the input data, thereby making the down-stream tasks tedious. Therefore, this study presents a self-supervised learning (SSL) approach to classifying 12 types of agricultural insect pests from 9549 RGB images, by leveraging the Bootstrap your own latent (BYOL) algorithm. SSL uses minimal labeling and is indifferent to data augmentations or distortions. Hence, latent representations from pretrained SSL networks could be generalized well for downstream tasks like classification or object detection. For desirable classification of the insect images, the greatest challenges observed were: i) large intra-class variation (the same insect was found with different colors and patterns), and ii) complex background with inconspicuous foreground. Hence, to overcome these issues and aid generalizability of the representations learned through BYOL, entropy-guided segmentation (segments based on texture not color), is proposed as input to the SSL network in this study. Both raw and segmented images were separately fed to two independent BYOL SSL networks, i.e., with ResNet18 and ResNet50 architectures as the backbone. The efficacy of the latent representations for downstream applications was assessed using linear evaluation, and subsequently compared with traditional transfer learning outcomes from ResNet18 and ResNet50. The results indicated that, while ResNet50 backbone intuitively performed better in all cases, SSL aided with entropy-based segmentation led to ~94% classification accuracy compared to raw images (with ~90% maximum accuracy).
The development of new phenomics approaches to image and process data from the subcellular to ecosystem-scale has accelerated over the past decade. Many of these tools are produced “in-house” within a single lab or a group of collaborating labs, making it hard to keep up with the state-of-the-art for phenomics hardware and software development. The Plant Cell Atlas Phenomics Committee is creating a collaborative space that connects phenomics developers with each other and with the greater plant science community, with the goal of facilitating wide-reaching collaborations. To do this, we will be hosting a video series called “How We Built It” where developers provide a short tour of their inventions and relevant biological applications. We will follow the series with a more in-depth networking event where plant scientists can connect with the inventors and discuss collaborative opportunities. Our goal is to streamline the invention of new phenotyping tools and broaden the application of existing tools.
Multispectral imaging with unmanned aircraft systems (UAS) is a promising high-throughput phenotyping technology that has been shown to help understand the causal mechanisms associated with crop productivity. This imaging technology can accurately predict complex agronomic traits like grain yield within a given generation, creating the potential to fast-track selections in plant breeding and increase genetic gains. The objective of this study was to determine the effectiveness and efficiency of prediction on grain yield in an abnormal drought year across locations within a breeding program. Eleven spectral reflectance indices (SRI) including NDRE, NWI, NDVI, and percent canopy cover were used to evaluate Washington State University winter wheat breeding lines between 2018 and 2021. Data was collected using a DJI Inspire 2 drone, equipped with a Sentera Quad Multispectral Sensor, and collected at the heading date. Lines were observed from single location, single replication preliminary yield trials to multi-location, replicated advanced yield trials. Lines advanced in the breeding program were evaluated across 13 different location-year trials. The calculated SRIs and canopy cover were used individually and in combination as fixed effects in mixed model prediction for grain yield under drought conditions. Models were independently validated with 2021 data. Across locations, SRIs are shown to improve the prediction performance for grain yield under abnormal drought conditions by as much as 40% in the case of NDRE. This research is vital for plant breeders to understand the utility of UAS imaging in variety improvement when dealing with abnormal growing seasons.
The field is not always easy for plant phenotyping using conventional phenotyping platforms due to the limited accessibility and regulated aviation area. Smartphone-triggered ground images were collected on wheat field that has a limited access to monitor growth conditions of four wheat varieties, Shinyoung (SY), Joseong (JS), Taewoo (TW), and Cheongwoo (CW). For field mapping during the growing season, six sets of the raw RGB images were acquired by a smartphone camera in an oblique view angle and processed to transform into nadir view images. A series of algorithms were developed to process the skewed tile images to straighten into the nadir images, align the deskewed images, and stitch them into a field image by detecting crop rows using Hough Transformation. Open-source software, iStitch, was developed to automate the algorithms in a batch process. Plot-level metrics were extracted to analyze plant growth of the wheat varieties using a gridding method for vegetation and leaf area indexes. The processed images resulted in the successful transformation and consistency of algorithms on image alignment and stitching. Plot-level analysis indicated that SY variety performed superior to the other varieties in plant quality and quantity and significantly different from TW variety in canopy coverage. The proposed approach of the stitching and gridding was applied on the skewed images acquired by a smartphone camera but can be directly used for other applications of plant phenotyping on images acquired by a camera on a mobile platform or a grid of stationary cameras in greenhouse or outdoor fields.
Canopy imaging is a non-invasive phenotyping approach to quantify parameters such as canopy size, stress symptoms, and pigment concentrations. Unlike destructive measurements, canopy imaging is fast and easy. However, analysis of the images can be time consuming. To facilitate large-scale use of imaging, the cost of imaging systems needs to be reduced and the analysis needs to be automated. We developed low-cost imaging systems using a Raspberry Pi microcomputer, equipped with a monochrome camera and filter, at a total hardware cost of ~$500. The latest version of our imaging system takes images under blue, green, red, and infra-red light, as well as images of chlorophyll fluorescence. The system uses a Python-based program to collect and analyze images automatically. The multi-spectral imaging system separates plant from background using the chlorophyll fluorescence image and generates normalized difference vegetation index (NDVI) and anthocyanin content index (ACI) images and histograms, providing quantitative, spatially-resolved information. We verified that these indexes correlate strongly with leaf chlorophyll and anthocyanin concentration. The low cost of the system can make this imaging technology widely available.
Cosegmentation is a recent and rapidly emerging and rapidly growing extension of segmentation, which aims to detect the common object(s) in a group of images. Current cosegmentation methods are ideal and effective only for certain dataset types with limited features that still produce errors making it difficult to obtain detailed metrics of object parts. We propose to build a unified, trainable framework that incorporates multiple features of a high-throughput dataset’s segmented images from multiple algorithms using cosegmentation. Specifically, we propose a novel Cosegmentation for Plant Phenotyping Network (CoPPNet) that utilizes a Fully Convolutional Neural Network with a K-Means Clustering feedback loop for optimal temporal loss. The results from this study will set the benchmark for a novel advancement in computer vision segmentation accuracy and plant phenomics to better understand a plant’s environmental interactions for maximal resilience and yield.