A 30 to 80% flower drop in soybeans grown across different regions in the US is an unresolved and
persisting bottleneck that has limited soybeans ability to achieve the full genetic yield potential. The major
challenge has been the lack of robust, field-based high throughput phenotyping and analysis tools to
capture temporal variation in flower abortion and pod retention across large genetically diverse
germplasm. The multi-regional (KS, MO, TN and TX) and trans-disciplinary team will develop an imagebased
field phenotyping system, integrated with deep-learning tools to capture large genetic variation in
flower abortion and pod retention under different soil and climatic conditions. A genetically diverse set of
50 genotypes including late group III and early group IV will be tested under natural dryland conditions
in MO and KS, and under irrigated and severe drought and heat stress conditions in TX and TN. Using
contrasting lines from information generated from year 1, molecular mechanism that control flower
abortion and pod retention will be determined. This fundamental knowledge will help discover molecular
switches to enhance flower and pod retention, and thereby enhance yield potential under diverse
environmental conditions. The proposed project will address - Tools and Technology for Soybean
Improvement and utilizing these to induce Extreme Weather Resiliency. In summary, the overall goal is
to increase flower and pod retention by 20 to 30%, with a potential to enhance yields by 10 to 15%,
ultimately translating to an additional 400 million dollars to the national soybean industry.

• Continue to explore the genetic diversity in flower abortion under different soil moisture and climatic conditions using a diverse set of landraces and elite genotypes
• Improve the image-based field phenotyping system and deep-learning tools to document temporal dynamics in flower abortion and pod retention in diverse soybeans grown under field conditions
• Identify molecular mechanisms controlling flower abortion under heat and drought conditions using contrasting genotypes that differ in proportion of flower abortion

• Range in phenotypic variation associated with flower abortion and pod retention in different maturity groups of soybean grown under different soil, moisture and climatic conditions determined.
• Field image-based phenotyping protocols established to track flowers and pods across all four participating institutes, with different rates of flower abortion captured
• Deep learning tool developed can analyze images and acquire temporal changes in flower numbers with minimal human interference, from field-based images collected across all four locations
• Candidate genes identified for flower abortion under favorable and drought and heat stress conditions.

Updated April 25, 2024:
Project title - Field phenotyping using machine learning tools integrated with genetic mapping to address heat and drought induced flower abortion in soybean.

Participating institutions – Texas Tech University, Kansas State University, University of Missouri, and University of Tennessee.

Goals & Objectives

Long-term Goal – Develop soybean cultivars with 20 to 30% lower flower abortion under favorable to challenging environmental conditions, leading to about 10-15% increase in yield potential.

Objectives (Year 2)
• Continue to explore the genetic diversity in flower abortion under different soil moisture and climatic conditions using a diverse set of landraces and elite genotypes.
• Improve the image-based field phenotyping system and deep-learning tools to document temporal dynamics in flower abortion and pod retention in diverse soybeans grown under field conditions.
• Identify molecular mechanisms controlling flower abortion under diverse climatic conditions.

Objective 1 - Explore the genetic diversity in flower abortion under different soil moisture and climatic conditions using a diverse set of landraces and elite genotypes
A diverse selection of 50 soybean genotypes, classified within groups III and IV, were meticulously chosen based on considerations of maturity and lodging scores from Year 1 multi-location studies. Seeds were provided by the University of Missouri for implementation of trails across multiple locations, namely Kansas State University (KS; Manhattan), Texas Tech University (TX; Lubbock), and The University of Tennessee (TN; Jackson), marking the beginning of the second experimental year of our project. All locations are currently in the preparatory phase for planting, scheduled in May.

This year, a significant advancement in our methodology involves the integration of QR codes (Figure 1) onto all plot labels across all locations. This technological enhancement streamlines imaging processing and storage procedures, facilitating efficient data management.

Each genotype will be planted in four row plots with a spacing of 30” between them. Within each row, seeds will be planted at a density of 8 seeds per foot. Furthermore, to ensure robustness and reliability of results, each experimental setup will be replicated three times within each location.

At Lubbock - TX we will incorporate two distinct irrigation regimes -100% ET and 50% ET leveraging the region's natural hot summers to induce heat stress. These irrigation treatments will be implemented via a sub-surface drip irrigation system. Drought stress (50% ET) will specifically start and persist throughout the flowering stage.

Meanwhile, in Jackson - TN, a greenhouse experiment will be conducted under stress conditions, focusing on four lines exhibiting low flower abortion rates and another four lines with high flower abortion percentages, based on year 1 findings. This targeted investigation aims to understand soybean response mechanisms that result in differential flower abortion under control and stress conditions.

Objective 2 – Improve the image-based field phenotyping system and deep-learning tools to document temporal dynamics in flower abortion and pod retention in diverse soybeans grown under field conditions.
At each location, uniform platforms for imaging are being constructed, ensuring consistency across all sites (Figure 2). The imaging process will start using a single GoPro Hero 11 camera, with additional cameras added as the crop matures, if necessary. Parameters for camera setup will remain consistent with those utilized in the previous year, ensuring continuity and comparability of data.

To streamline the data acquisition process, a Python program for generating QR codes and label IDs (Figure 1) has been developed for common use across all locations. Additionally, a program for detecting QR codes within the recorded videos has been successfully developed and will be integrated into the model in subsequent stages.

The forthcoming task involves the development of a program designed to extract videos from the cameras and organize them into location-specific folders, incorporating field specifications such as row numbers and the number of cameras per row. This program will greatly simplify the process of video collection and enhance data management efficiency for the machine learning-based flower and pod tracking algorithms.

Furthermore, significant effort has been made in refining the accuracy of the pod and flower tracking models through rigorous training processes, aimed at maximizing precision and reliability in data analysis.

Objective 3 - Identify molecular mechanisms controlling flower abortion under diverse climatic conditions.
To identify the molecular mechanism involved in flower abortion, we have selected the most contrasting accessions from FY23 multi-location trials for bulk transcriptomics analysis under controlled greenhouse conditions. Eight accessions, four lines exhibiting low flower abortion rates and another four lines demonstrating high flower abortion rates, will be evaluated under controlled greenhouse conditions to eliminate the variation associated with environmental factors.

Following tissues will be carefully harvested from 3rd internode to ensure the developmental stage and flower position. Tissue samples including pedicel (stalk connecting to stem), flower bud, partially opened, fully opened, and flower after anthesis (initiation of senescence) will be collected periodically. Following the sample collections, RNA will be extracted from at least 3 replicates/stage/sample and will be subjected to RNA sequencing using commercial vendor. Approximately, 144 samples (8 genotypes X 6 stages X 3 replications) will be collected. RNAseq data will be analyzed using the standard pipeline to identify differentially expressed genes as detailed in our publication Chen et al. (2016; PMID: 27486466).

View uploaded report

Retaining even a proportion of 30% to 80% of flowers aborted under well-watered and stressful conditions, respectively, will allow for 10 to 20% increase in yield for the soybean producers in the US. This advantage can be extended to different soil and water available conditions, to support a wide range of soybean producers and is the major rationale for embarking on testing this hypothesis across four different soybean growing states with a focus on maturity groups III to IV. The advantage proposed through this collaboration, will allow the soybean producers to gain additional economic return at the same level of investment i.e., with same seed cost, fertilizer level and management. With changing climate leading to an increase in temperature and lesser water available scenarios, the proportion of flower drop would increase proportionally, further lowering yield and producer profits. Hence, germplasm, breeding populations, novel QTL/genes and CRISPR edited lines developed with increased flower retention would help enhance the yield potential under current climates and retain the advantage even under future warmer and drier environments.