Study of Phenotype Analysis of Maize Plant: Hyperspectral Image Perspective

Abstract

Maize, alongside rice and wheat, constitutes a trio of crops responsible for over 50% of the global calorie consumption. To address the rising food demand, enhancing the productivity and stress resilience of these crops becomes imperative. However, the progress of plant breeding initiatives is hindered by the cost and time constraints associated with acquiring plant phenotype data. To overcome these limitations and advance the field, there is a need for datasets that connect new forms of high-throughput phenotype data, gathered from plants, to the performance of identical genotypes in diverse agronomic settings and habitats. These datasets will pave the way for the development of innovative statistical and Machine Learning techniques, empowering researchers to expedite crop improvement efforts effectively.

Introduction

I. INTRODUCTION

The green revolution helped lessen hunger and drought across the world in the 1960s and 1970s even while population growth accelerated. It did this by increasing the yields of numerous significant crops.  The development of new major grain crop varieties with increased yield potential through traditional phenotypic selection was a crucial aspect of the green revolution. Since then, the demand for food has consistently grown, leading to significant endeavours in the public and private sectors to cultivate crop types with even higher yield potential. As the easily achievable advancements in productivity diminish, achieving further improvements in yield requires greater dedication of effort and resources. Latest research has found that harvest gains for some key grain crops have slowed or stopped in significant parts of the world. If the agricultural output is to keep growing to fulfil the requirements of a rising global population, new approaches to plant breeding must be devised.

Phenotype is a critical bottleneck in current plant breeding. There are two ways to use phenotyping. Initially, a plant breeder can identify lines with the highest yield potential and stress tolerance in a particular environment by extensively phenotyping a large number of lines. Second, enough detailed phenotyping measures from enough diverse plants, along with genotypic data, can be used to identify sections of a plant species' genome that contain advantageous or harmful alleles. After that, the breeder can generate new crop kinds with as many advantageous alleles as feasible while removing as many negative alleles as possible.

Phenotyping proves to be both time-consuming and costly. As breeders strive to uncover numerous alleles, each with a minor impact, the demand for phenotyping increases to achieve a specific rise in yield potential. However, plant phenotyping techniques integrated with high-throughput Machine Learning offer a promising solution to this bottleneck. These advanced tools allow precise quantification of even the smallest plant features, and their unit costs are likely to decrease with scale. In contrast, traditional phenotyping remains labor-intensive and does not benefit from the same cost reductions.

In several recent pilot experiments, various image-processing algorithms were employed to obtain phenotypic measurements from crop plants. Among the computer vision-based plant phenotyping approaches, the RGB (Red, Green, Blue) camera technology was predominantly used, as it is commonly found in the consumer market. Additionally, fluorescence and near-infrared (NIR) cameras have been utilized in high-throughput plant phenotyping endeavors, particularly when studying the response of plants to different abiotic stress conditions.

III. DATASET

Researchers commonly utilize the Panicoid phenome-1 dataset for phenotype analysis due to its comprehensive data, including RGB, Hyperspectral, Fluorescence, and Thermal Infrared images. This dataset is invaluable for conducting phenotyping research, specifically in areas like leaf counting, leaf alignment, leaf segmentation, leaf tracking, and 3D leaf reconstruction. It has become the standard dataset for such studies.

The Panicoid phenome-1 dataset consists of a substantial 485 GB of images, making it an excellent resource for evaluating the performance of proposed algorithms. To support and ensure consistent comparisons, we present the University of Nebraska-Lincoln Panicoid phenome-1 dataset, along with corresponding ground truth data.

IV. CHALLENGES

1. Sample Size and Diversity: The study might have a limited sample size or focus on specific crop varieties, which could restrict the generalizability of the results to other plant species or environmental conditions.
2. Data Quality and Preprocessing: The accuracy and reliability of high-throughput image analysis heavily depend on the quality of the acquired images and the effectiveness of preprocessing steps. Inaccuracies in segmentation or feature extraction could affect the validity of the conclusions.
3. Experimental Design: The paper might have potential issues related to experimental design, such as control treatments, randomization, or replication, that could impact the robustness of the conclusions.
4. Measurement Precision: Some phenotypic traits might be challenging to accurately measure using high-throughput imaging alone, leading to potential measurement errors or uncertainties.
5. Environmental Variability: The study may not fully account for the natural variability in environmental conditions, which could influence plant growth and drought responses.
6. Data Analysis Methods: The choice of data analysis methods and statistical approaches could influence the interpretation of results. Inappropriate or inadequate analysis methods might lead to biased or misleading conclusions.
7. Incomplete Understanding of Plant Physiology: While high-throughput imaging provides valuable insights, it might not capture all the underlying physiological mechanisms governing plant growth and drought responses.
8. Publication Bias: If the paper only reports significant results or positive outcomes, it may suffer from publication bias, potentially neglecting important negative findings.
9. Data Availability: The availability of the dataset used in the paper might not be fully open or accessible, limiting the reproducibility and transparency of the research.
10. Time Frame: As the paper was published in 2014, the research might not reflect the latest advancements in high-throughput image analysis or related technologies.

Conclusion

We conclude that using the Image processing techniques we easily analyze the plant phenotype. We learn about dataset and use various techniques for Extraction, Clustering and Classification. In this publication, we attempted to conduct a critical assessment of studies conducted by researchers in various parts of the world.

References

[1] Hartmann, Anja, Tobias Czauderna, Roberto Hoffmann, Nils Stein, and Falk Schreiber. \"HTPheno: an image analysis pipeline for high-throughput plant phenotyping.\" BMC bioinformatics 12, no. 1 (2011): 1-9. [2] Topp, Christopher N., Anjali S. Iyer-Pascuzzi, Jill T. Anderson, Cheng-Ruei Lee, Paul R. Zurek, Olga Symonova, Ying Zheng et al. \"3D phenotyping and quantitative trait locus mapping identify core regions of the rice genome controlling root architecture.\" Proceedings of the National Academy of Sciences 110, no. 18 (2013): E1695-E1704. [3] Zurek, Paul R., Christopher N. Topp, and Philip N. Benfey. \"Quantitative trait locus mapping reveals regions of the maize genome controlling root system architecture.\" Plant physiology 167, no. 4 (2015): 1487-1496. [4] Zheng, Zihao, Stefan Hey, Talukder Jubery, Huyu Liu, Yu Yang, Lisa Coffey, Chenyong Miao et al. \"Shared genetic control of root system architecture between Zea mays and Sorghum bicolor.\" Plant physiology 182, no. 2 (2020): 977-991. [5] Dadon, Alon, Moshe Mandelmilch, Eyal Ben-Dor, and Efrat Sheffer. \"Sequential PCA-based classification of mediterranean forest plants using airborne hyperspectral remote sensing.\" Remote Sensing 11, no. 23 (2019): 2800. [6] Andrade-Sanchez, Pedro, Michael A. Gore, John T. Heun, Kelly R. Thorp, A. Elizabete Carmo-Silva, Andrew N. French, Michael E. Salvucci, and Jeffrey W. White. \"Development and evaluation of a field-based high-throughput phenotyping platform.\" Functional Plant Biology 41, no. 1 (2013): 68-79. [7] Zhang, Xuehai, Chenglong Huang, Di Wu, Feng Qiao, Wenqiang Li, Lingfeng Duan, Ke Wang et al. \"High-throughput phenotyping and QTL mapping reveals the genetic architecture of maize plant growth.\" Plant physiology 173, no. 3 (2017): 1554-1564. [8] Van Overbeek, Leo, and Jan Dirk Van Elsas. \"Effects of plant genotype and growth stage on the structure of bacterial communities associated with potato (Solanum tuberosum L.).\" FEMS Microbiology Ecology 64, no. 2 (2008): 283-296. [9] Choudhury, Sruti Das, Vincent Stoerger, Ashok Samal, James C. Schnable, Zhikai Liang, and Jin-Gang Yu. \"Automated vegetative stage phenotyping analysis of maize plants using visible light images.\" In KDD workshop on data science for food, energy and water, San Francisco, California, USA. 2016. [10] Scharr, Hanno, Massimo Minervini, Andrew P. French, Christian Klukas, David M. Kramer, Xiaoming Liu, Imanol Luengo et al. \"Leaf segmentation in plant phenotyping: a collation study.\" Machine vision and applications 27 (2016): 585-606. [11] Taghavi Namin, Sarah, Mohammad Esmaeilzadeh, Mohammad Najafi, Tim B. Brown, and Justin O. Borevitz. \"Deep phenotyping: deep learning for temporal phenotype/genotype classification.\" Plant methods 14, no. 1 (2018): 1-14. [12] Großkinsky, Dominik K., Jesper Svensgaard, Svend Christensen, and Thomas Roitsch. \"Plant phenomics and the need for physiological phenotyping across scales to narrow the genotype-to-phenotype knowledge gap.\" Journal of experimental botany 66, no. 18 (2015): 5429-5440. [13] Jansen, Marcus, Frank Gilmer, Bernhard Biskup, Kerstin A. Nagel, Uwe Rascher, Andreas Fischbach, Sabine Briem et al. \"Simultaneous phenotyping of leaf growth and chlorophyll fluorescence via GROWSCREEN FLUORO allows detection of stress tolerance in Arabidopsis thaliana and other rosette plants.\" Functional Plant Biology 36, no. 11 (2009): 902-914. [14] Su, Yanjun, Fangfang Wu, Zurui Ao, Shichao Jin, Feng Qin, Boxin Liu, Shuxin Pang, Lingli Liu, and Qinghua Guo. \"Evaluating maize phenotype dynamics under drought stress using terrestrial lidar.\" Plant methods 15 (2019): 1-16. [15] Das, Abhiram, Hannah Schneider, James Burridge, Ana Karine Martinez Ascanio, Tobias Wojciechowski, Christopher N. Topp, Jonathan P. Lynch, Joshua S. Weitz, and Alexander Bucksch. \"Digital imaging of root traits (DIRT): a high-throughput computing and collaboration platform for field-based root phenomics.\" Plant methods 11, no. 1 (2015): 1-12. [16] Scharr, Hanno, Massimo Minervini, Andrew P. French, Christian Klukas, David M. Kramer, Xiaoming Liu, Imanol Luengo et al. \"Leaf segmentation in plant phenotyping: a collation study.\" Machine vision and applications 27 (2016): 585-606. [17] Liang, Zhikai, Piyush Pandey, Vincent Stoerger, Yuhang Xu, Yumou Qiu, Yufeng Ge, and James C. Schnable. \"Conventional and hyperspectral time-series imaging of maize lines widely used in field trials.\" Gigascience 7, no. 2 (2018): gix117. [18] Grassini, Patricio, Kent M. Eskridge, and Kenneth G. Cassman. \"Distinguishing between yield advances and yield plateaus in historical crop production trends.\" Nature communications 4, no. 1 (2013): 2918. [19] Chen, Dijun, Kerstin Neumann, Swetlana Friedel, Benjamin Kilian, Ming Chen, Thomas Altmann, and Christian Klukas. \"Dissecting the phenotypic components of crop plant growth and drought responses based on high-throughput image analysis.\" The plant cell 26, no. 12 (2014): 4636-4655. [20] Ge, Yufeng, Geng Bai, Vincent Stoerger, and James C. Schnable. \"Temporal dynamics of maize plant growth, water use, and leaf water content using automated high throughput RGB and hyperspectral imaging.\" Computers and Electronics in Agriculture 127 (2016): 625-632. [21] Fahlgren, Noah, Maximilian Feldman, Malia A. Gehan, Melinda S. Wilson, Christine Shyu, Douglas W. Bryant, Steven T. Hill et al. \"A versatile phenotyping system and analytics platform reveals diverse temporal responses to water availability in Setaria.\" Molecular plant 8, no. 10 (2015): 1520-1535. [22] Al-Tamimi, Nadia, Chris Brien, Helena Oakey, Bettina Berger, Stephanie Saade, Yung Shwen Ho, Sandra M. Schmöckel, Mark Tester, and Sónia Negrão. \"Salinity tolerance loci revealed in rice using high-throughput non-invasive phenotyping.\" Nature communications 7, no. 1 (2016): 13342. [23] Chen, Dijun, Rongli Shi, Jean-Michel Pape, Kerstin Neumann, Daniel Arend, Andreas Graner, Ming Chen, and Christian Klukas. \"Predicting plant biomass accumulation from image-derived parameters.\" GigaScience 7, no. 2 (2018): giy001. [24] Campbell, Malachy T., Avi C. Knecht, Bettina Berger, Chris J. Brien, Dong Wang, and Harkamal Walia. \"Integrating image-based phenomics and association analysis to dissect the genetic architecture of temporal salinity responses in rice.\" Plant physiology 168, no. 4 (2015): 1476-1489. [25] Breiman, Leo. \"Random forests.\" Machine learning 45 (2001): 5-32. [26] Wheeler, Lucas C., and Stacey D. Smith. \"Computational modeling of anthocyanin pathway evolution: biases, hotspots, and trade-offs.\" Integrative and comparative biology 59, no. 3 (2019): 585-598. [27] Honsdorf, Nora, Timothy John March, Bettina Berger, Mark Tester, and Klaus Pillen. \"High-throughput phenotyping to detect drought tolerance QTL in wild barley introgression lines.\" PloS one 9, no. 5 (2014): e97047.

Copyright

Copyright © 2023 Neha B. Chavan, Dr. Ramesh Manza, Diksha R. Pawar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.