Progress 05/01/23 to 04/30/24
Outputs
Target Audience:At this intermediate point in this research, our primary target audience is other researchers working in this area. Presentations were given during this reporting period to communicate findings with the target audience of other researchers who are interested in the prevention of post-harvest loss due to Fusarium graminearum and other fungal pathogens. These presentations consisted of both oral presentations and student posters. Abstracts have been submitted for presentations that will take place in Fall 2024. Changes/Problems:All changes to the project this year were mostly positive. Due to the generosity of collaborators such as Beck's Hybrids and the Genomes to Fields Initiative, we have been able to expand the scope of this project immensely to include two field seasons. This highlights the importance of collaboration among industry-academia and the widespread collaboration offered through the Genomes to Fields Initiative. Both are a testament to the amazing community of scientists in this research sector. The second change was an improvement to the protocols used to quantify the phytochemicals and mycotoxins of interest as part of this project. Although this resulted in a very minor delay, the impact this will have on the scientific industry is much greater because we have ensured the protocols we are using are accessible to most scientists. There are no major problems or other changes to report. What opportunities for training and professional development has the project provided?Through this project, a total of 4 graduate students and 10 undergraduate students have been trained. These students have received experience conducting and managing a research project, overseeing undergraduate students (in the case of graduate students), and presenting their results at professional conferences and scientific meetings. One student successfully defended her MS thesis shortly following the end of this project period, and another student successfully passed her PhD preliminary examination during this project period. All students have gained skills within their area of desired expertise, whether that be plant pathology, genomic prediction, analytical chemistry and biochemistry techniques, or data science. Additionally, most undergraduates who have participated in this project have gained experience working in both supervised research settings and, eventually, within an independent study environment. Most undergraduate students have also elected to present their research at their home institution's undergraduate research symposium. How have the results been disseminated to communities of interest?Results are being disseminated to communities of interest as they become available. The results generated to date have been presented at professional conferences such as the ASA-CSSA-SSSA annual meeting. Results have been presented as both oral presentations and student poster presentations. It is anticipated that as the final year of phenotypic data collection comes to a close, that our dissemination strategies will move from oral presentations and conference presentations to a greater volume of written publications. In anticipation of this event, we have prepared a literature review that captures the breadth of information available regarding Fusarium graminearum in maize and wheat. It is anticipated that this will be submitted in Summer 2024. What do you plan to do during the next reporting period to accomplish the goals?Goal 1 The 2024 field season is the final phenotyping year planned for this project. The panel of 108 hybrids are to be grown at three locations by UIUC personnel, and these will again be inoculated. Two additional locations will be grown in partnership with Beck's Hybrids. A subset of 9 genetically representative hybrids will be grown at 9 locations through the Genomes to Fields Initiative. All locations will consist of trials that are replicated twice. In each location, the same set of phenotypic traits will be collected as in the 2023 field season. These traits include GER ratings and mycotoxin contamination levels at all locations. Goal 2 Flavonoid concentration measurements are underway and are the final remaining set of phenylpropanoid phytochemicals to be analyzed for this project. Now that the protocol has been optimized to better serve the scientific community, it is anticipated that this phenotypic data collection will be completed during the Fall or early Winter of 2024. Following the completion of this dataset, final correlations will be calculated, and statistical models will be completed. Goal 3 Following the completion of the phenotyping stage of this project, we anticipate a major pivot to the data science and machine learning portion of the project in the upcoming project period. A student has been hired to begin testing different machine learning methods (namely, decision tree, logistic regression, ensemble methods such as random forest, and neural network methods such as multi-layer perceptrons) during Summer 2024. More complex genomic modeling techniques will be explored upon the completion of phenotypic data measurements, although the preliminary models suggest that categorical response variables and multi-trait genomic prediction are needed. With this information in hand, models will be built and evaluated that combine the findings from the initial genomic and agronomic model findings.
Impacts
What was accomplished under these goals?
Goal 1 During Summer 2023, we worked with the Genomes to Fields Initiative, Corteva, and Beck's Hybrids to expand this research to a diversity of environments nationwide. The full panel was grown in inoculated conditions in Urbana-Champaign in two fields with varying disease history. The full panel of inbreds was also grown at these locations under these conditions. A third uninoculated trial was grown in Western Illinois and included the full inbred and hybrid panels. All other trials used hybrids only and were uninoculated. Beck's Hybrids grew two replicated trials of the 108 hybrids included in this study. One trial was grown in Central Iowa, and the other in Central Indiana. Corteva grew seven trials, spread across Illinois, Indiana, Iowa, Minnesota, and Wisconsin. These seven trials also included all 108 hybrids. Finally, through the Genomes to Fields Initiative, 9 representative hybrids were selected and planted in replicated trials at each of 9 locations, spanning New York to Georgia to Texas in reach. GPS coordinates were provided at each of the locations so that weather data could be downloaded from the closest weather station. GER ratings were collected at all three UIUC locations, along with yield and test weight, and ears were analyzed at SIUE for mycotoxin contamination levels. Beck's Hybrids collected yield, moisture, and test weight data, and also harvested ears so that GER ratings and mycotoxin analysis could be conducted by UIUC and SIUE researchers, respectively. Corteva collected test weight, yield ratings, and other opportunistic agronomic traits. Genomes to Fields sites harvested and shipped ears back for GER and mycotoxin analysis. New mycotoxin protocols that examine the presence of a diversity of mycotoxins, not only those that are associated with GER, have been validated and will be employed early in the next reporting period to understand the variation in disease and mycotoxin presence based on geography and weather conditions. Goal 2 Hydroxycinnamic acid concentration measurements were finalized during the reporting period. During the reporting period, it was noted that the original flavonoid protocol could not be employed without using mass spectroscopy (MS). However, while many laboratories possess HPLC, they are much less likely to possess HPLC-MS. Thus, to ensure that the research conducted as part of this project is impactful and is in direct service to stakeholders, all quantitation methods (mycotoxins, hydroxycinnamic acids, and flavonoids) were optimized such that measurements could take place without the use of MS. Flavonoid extraction is ongoing, and both extraction and quantitation should be finished in the first half of the next reporting period. Correlations among phyotchemicals, including the data collected via NIR during the 2022-23 project period, were calculated to provide basic insight into the relationship among these compounds in inoculated and uninoculated conditions. Statistical modeling techniques have been employed to explore the data preliminarily, and it appears that generalized linear mixed modeling techniques will be most appropriate for analyzing the data based on the distribution of the estimates of the error. At this time, disease resistance appears to be directly related to hydroxycinnamic acid content, but the relationship between deoxynivalenol and zearalenone concentration relative to hydroxycinnamic acid concentrations appears less certain. It is anticipated that as the mycotoxin data becomes more complete, that we will be able to state with confidence whether there is or is not a relationship between these mycotoxins and the hydroxycinnamic acids of interest within our germplasm. Based on the relationships seen thus far, it is anticipated that machine learning models that dichotomize GER resistance and mycotoxin contamination may be needed. Goal 3 Traditional genomic models and multi-trait genomic models were trained using the preliminary datasets. Genome-wide association studies (GWAS) identified a small number of candidate genes to be explored in further detail upon final compilation of the dataset, which should take place in early 2025. Single trait genomic prediction models indicated a high selective potential for GER and a moderately high selective potential for one of the hydroxycinnamic acids under study. Interestingly, although multi-trait genomic prediction was less accurate for GER, the precision across all traits included in the model improved. As such, the multi-trait genomic prediction models suggested that GER and both of the primary hydroxycinnamic acids under study could be successfully selected for using genomic prediction techniques. Interestingly, these preliminary models used linear response variables, as opposed to dichotomized or categorized variables. Additional modeling techniques will be undertaken to study whether genomic prediction may be improved by altering the assumptions of the distribution of the response variable. To explore the weather aspects of the model, a student has been hired to begin testing different machine learning methods (namely, decision tree, logistic regression, ensemble methods such as random forest, and neural network methods such as multi-layer perceptrons). This work will take place in Summer 2024. Upon completion of this work, as well as compilation of the final phenotypic datasets in the upcoming project period, much of the work in the upcoming year will shift from data collection to modeling and machine learning.
Publications
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2024
Citation:
Lipps S, C Hasty, Z Hill, IP Osikoya, CJ Butts-Wilmsmeyer, MO Bohn, and T Jamann. The maize kernel and its unexploited potential mechanisms in resistance to Fusarium graminearum. ASA-CSSA-SSSA International Annual Meeting.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Hasty CA, S Lipps, V Zerda-Pinto, IP Osikoya, U Eleonu, TM Jamann, MO Bohn, KR Tucker, RP Dixon, and CJ Butts-Wilmsmeyer. The Microscopic Shield: Can Phenylpropanoid Concentrations Provide Natural Resistance to Ear Rots and Mycotoxin Contamination, and Can We Predict Enhanced Resistance Using Genomic Techniques? ASA-CSSA-SSSA International Annual Meeting. St. Louis, MO. October 2023.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Benware M-M, CA Hasty, S Lipps, V Zerda-Pinto, F Romo, KR Tucker, RP Dixon, MO Bohn, TM Jamann, and CJ Butts-Wilmsmeyer. Genome-Wide Association Study of Resistance to Mycotoxin Production in Maize (Zea mays L.). ASA-CSSA-SSSA International Annual Meeting. St. Louis, MO. October 2023. 2nd Place Student Poster.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Lipps S, CA Hasty, CJ Butts-Wilmsmeyer, MO Bohn, and TM Jamann. Its Quantitative! Resistance to Fusarium graminearum, Deoxynivalenol, and Zearalenone in Maize. ASA-CSSA-SSSA International Annual Meeting. St. Louis, MO. October 2023.
- Type:
Journal Articles
Status:
Other
Year Published:
2024
Citation:
Lipps S, J Rutkoski, CJ Butts-Wilmsmeyer, MO Bohn, S Mideros, and T Jamann. Resistance to Fusarium graminearum in maize: A comparison with wheat. In preparation.
|
Progress 05/01/22 to 04/30/23
Outputs
Target Audience:At this midway point in the research, our primary target audience is other researchers working in this area. Presentations were given during this reporting period to communicate findings with the target audience of other researchers who are interested in the prevention of post-harvest loss due to Fusarium graminearum and other fungal pathogens. These presentations consisted of both oral presentations and student posters. Abstracts have been submitted for both student presentations that will take place in Fall 2023. Changes/Problems:The most significant problem encountered during this project period pertained to changes that had occurred during restructuring events within our industry partners. These restructuring events meant that it would no longer be possible for some of our industry partners to provide seed for mycotoxin sampling or to provide Gibberella ear rot ratings, which, while understandable, posed an issue for a project that relies on multiple environments for its models. To overcome this issue, it was decided that the yield and test weight values provided by the industry collaborator may still provide useful data, but additional locations were needed. One additional site was added at Monmouth, Illinois that is a University of Illinois location, and two additional sites were added that were through a collaboration with another industry partner. Additionally, multiple sites were added through the Genomes to Fields Initiative that examined a select set of 9 hybrids across the United States, so as to obtain additional scenarios that should add power to the final predictive model. Thus, this challenge has largely been resolved. A smaller issue that was encountered is that the LC-MS equipment in the Shimadzu SPARK Laboratory space at SIUE that is used for this project was down for maintenance for a few months because of a technical issue. The director of the laboratory space worked with Shimadzu to restore the equipment, and it is now running and fully functional. This did cause a slight delay in the timeline for the analysis of the mycotoxins, hydroxycinnamic acids, and flavonoids, but it is anticipated that all of the laboratory analyses for all of the 2021 and 2022 samples should be completed by the end of 2023. Thus, while causing a slight delay in the original timeline for both the laboratory analyses and algorithm development, this is also largely resolved. What opportunities for training and professional development has the project provided?Through this project, a total of 4graduate students and 6undergraduate students have been trained during this project period. These students receiveexperience conducting and managing a research project, receive experience overseeing undergaduate students, and also receive experience presenting their results at professional conferences and scientific meetings. One student successfully defended her MS thesis in December, and another student successfully passed her PhD preliminary examination during this project period. Additionally,undergraduate students have gained experience working in supervised research settings, and most have elected to present their findings at the undergraduate research symposium at their home institution. How have the results been disseminated to communities of interest?Results have begun to be disseminated to communities of interest. These results have been presented at professional conferences such as the ASA-CSSA-SSSA annual meeting. Results have been presented as both oral presentations and student poster presentations. What do you plan to do during the next reporting period to accomplish the goals?Goal 1 During Summer 2023, we will be partnering with the Genomes to Fields Initiative and Corteva to expand this experiment to a diversity of environments nationwide. Full inbred panels will be grown at two locations in Urbana-Champaign and another location in Western Illinois. The full panel of 108 hybrids will be grown at these three locations plus 2 additional locations in partnership with Becks and 7 additional locations in partnership with Corteva. GER ratings will be collected at all three University of Illinois locations, along with yield and test weight. Plots will be sampled, and representative ears will be shipped to SIUE for mycotoxin analysis. Becks will collect yield, moisture, and test weight, and will harvest seed so that mycotoxin analysis can be conducted by SIUE. Corteva will collect test weight and yield ratings. Genomes to Fields sites will harvest ears and ship ears back for mycotoxin analysis. Goal 2 Hydroxycinnamic acid and flavonoid concentrations will be finalized in Fall 2023. Correlations among phytochemicals, including the data collected via NIR during the 2022-23 project period, will be calculated to provide basic insight into the relationship among these compounds. Correlations among all phytochemicals will be calculated to provide basic insight into the relationship among these compounds. Statistical modeling techniques (e.g. mixed models or generalized linear mixed models) will be used to explore the relationship among the phytochemicals and disease resistance in more detail. The relationships identified will help determine the algorithms used in future machine learning models. Goal 3 As the second year of field and laboratory is generated and datasets become more complete, traditional genomic models (e.g. GWAS and genomic prediction models) will be built. As the dataset becomes more complete, multi-trait genomic prediction models will also be built. Additionally, relationships between agronomic, disease ratings, mycotoxin contamination, metabolic, and weather data, as well as genomic data, will continue to be explored as the datasets become more complete. These relationships will make it possible to decide among potential candidate machine learning algorithms, and we anticipate being to test some of these algorithms more thoroughly during this coming project period. Additionally, with the full inbred dataset having been completed by the end of Fall 2023, it is anticipated that preliminary models may be built.
Impacts
What was accomplished under these goals?
Goal 1 The 432 maize inbreds included in this study were evaluated in Summer 2022 under inoculated field conditions. Inoculated trials were conducted at two field sites. This builds upon the dataset collected in 2021. This complete dataset has two years of data for each of the 432 maize inbreds. These inbred cultivars represent a genetically diverse set of maize that will, following the statistical and laboratory analyses currently underway, provide the genetic information necessary to select the best-suited cultivars for minimizing post-harvest loss. All inbreds were evaluated for susceptibility to Gibberella ear rot. Heritability estimates across the two environments have been calculated and are estimated to be approximately 77%. It is important to note that this is under inoculated conditions. Under non-inoculated conditions, disease expression was much less consistent. 108 hybrid crosses were created. These 108 crosses make use of a combination of Resistant, Moderately Resistant, and Susceptible varieties. Additionally, crosses were designed such that relatively equal numbers of Resistant, Moderately Resistant, and Susceptible varieties, as well as crossing types (e.g. Resistant x Resistant, Resistant x Susceptible, etc.) were used to the extent possible while also considering flowering date and genetic clustering assignments. These hybrid crosses were increased such that they will be evaluated at two inoculated sites at Urbana-Champaign in 2023, 1 uninoculated site in Western Illinois that belongs the University of Illinois and has a history of disease pressure, and across 7 uninoculated sites in collaboration with Corteva. Two additional sites will be planted by Becks. Additionally, 9 hybrids were selected that represent the genetic and phenotypic diversity of the parents used in this study. These 9 hybrids will be grown at multiple locations in 2023 through the Genomes 2 Fields Initiative. Mycotoxin quantitation protocols using LC-MS were optimized for efficient phenotyping of samples. Representative samples from all plots grown in 2021 were phenotyped for mycotoxin content. The heritability for DON and zealernone were, respectively, 95.4% and 79.2%. While these estimates may change as more environments are added to the results, these preliminarily high estimates are promising. Seed was packaged for shipment for mycotoxin analysis at the end of the project reporting period and will be analyzed during Summer and early Fall 2023. Goal 2 Quantitation protocols for the flavonoids were optimized using LC-MS. All hydroxycinnamic acids from the 2021 field season have been extracted and are in the process of being prepared for LC-MS analysis. Extraction of the hydroxycinnamic acids from the samples from the 2022 field season will be completed during Summer and early Fall 2023. All samples for flavonoid extraction from the 2021 field season have been weighed and are slated for analysis in Fall 2023. Analysis of the flavonoids from the 2022 field season will directly follow. All samples from the uninoculated trials in 2021 were scanned using NIR. Baseline starch, protein, ash, and oil concentrations are available for all of these samples. Goal 3 Although most work this year has focused on Goals 1 and 2, a portion of this year was also dedicated to preparing the data for large-scale analysis and the exploration of algorithms and platforms that could be used for analysis. This work has built off of work that started in the previous project period and continued through this project period. Briefly, the most promising algorithms currently make use of feature extract techniques, rather than "blind" modeling approaches that make use of all data irrespective of growing conditions or plant physiology. These algorithms will provide insight as to the data manipulations, feature extraction, and potential algorithmic pipelines necessary for incorporating weather, genetic, and metabolic data into models for predicting plant performance, and ultimately Gibberella ear rot disease pressure and potential loss due to mycotoxin contamination. Additionally, initial single-trait GWAS and genomic prediction models have indicated: Trend-level significant peaks for zealernone, but not for DON, after a single year. This will be verified following the collection of the second year of mycotoxin data. Single-trait genomic prediction confirmed the GWAS results, with very low predictive ability for DON (17.6%), but low to moderate predictive ability for zealernone (29.8%). While preliminary, this suggests that genetic gains may be made for varieties resistant to zealernone accumulation.
Publications
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2022
Citation:
Lipps S, MO Bohn, CJ Butts-Wilmsmeyer, and TM Jamann. Dissecting the Complex Relationship of Host Resistance in Maize to Fusarium graminearum. ASA-CSSA-SSSA International Annual Meeting.
- Type:
Conference Papers and Presentations
Status:
Accepted
Year Published:
2022
Citation:
Ayegbidun OM, et al. Marker-Assisted Selection for Stress Resistance in Maize. ASA-CSSA-SSSA International Annual Meeting.
|
Progress 05/01/21 to 04/30/22
Outputs
Target Audience:
Nothing Reported
Changes/Problems:The most significant problem encountered during this project period pertained to winter seed increase. The COVID-19 Pandemic changed our plans because the company we had originally planned to work with could no longer perform this task. We quickly found an alternative and secured a potential vendor. However, the legal and purchasing teams at both of our institutions would not allow the work to proceed because the company identified, headquartered in South America, very understandably could not agree to some of the legal and purchasing requirements that must be followed by public institutions in the State of Illinois. Although we attempted for months to work with all parties involved, a resolution was not met. Unfortunately, this means that seed increase must take place during the Summer 2022 growing season as opposed to the 2021/2022 Winter Nursery, and the project is one year behind schedule. As a result, we will almost undoubtedly request a one-year no-cost extension, please. The second issue that we have encountered is a significant supply chain disruption that has resulted in shipping delays for some of our laboratory chemicals and other supplies. Depending on the supplier, some of our supplies are taking up to one year to arrive after ordering. While we try to anticipate these supply chain disruptions and order ahead as much as possible, and we also try to find other vendors, some shortages have not been easy to address due to lack of suppliers, unexpected shortages of items, and widespread shortages across multiple vendors. These shortages have primarily caused a delay among laboratory analyses. However, all grain samples are being stored in a cool, dry location until sufficient supplies are available for a particular analysis. Therefore, we do not expect any issues beyond a delay in laboratory operations. What opportunities for training and professional development has the project provided?Three graduate students, two of whom were funded by other sources and one of whom was funded partially by another source, had the opportunity to work on this project this year. One graduate student was involved in the field inoculation trials as well as the plant breeding nursery and seed increase. As part of her work, she also developed an inoculation protocol. One graduate student was involved in the development of the extraction protocols needed for phenotyping. The third graduate student was involved with the creation and optimization of quantitation methods using LC-MS that will be needed for phenotyping. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?Goal 1 Although we encountered issues with winter seed increase, as described in the "Changes and Problems" section, we will still be continuing both field and seed increase operations in Summer 2022. Specifically, the seed required for the last field season will be produced during Summer 2022 in Champaign, IL. Given this limitation, one benefit is that we will also be able to use this as an opportunity to collect an additional year of field data on the 432 maize inbred lines under inoculated trials in Champaign, IL. Mycotoxin contamination continues to be quantified using HPLC protocols, and we intend to complete this process pending any additional supply chain issues. Goal 2 Hydroxycinnamic acids and flavonoids continue to be quantified for each maize sample. We intend to complete the quantification of these phytochemicals during the next project period, pending supply chain issues. In addition to the proposed work, we have also secured access to a scanning (non-destructive) NIR and will also be quantifying the amount of protein, starch, and oil in each sample during this project period. Correlations among all phytochemicals will be calculated to provide basic insight into the relationship among these compounds. Statistical modeling techniques (e.g. mixed models or generalized linear mixed models) will be used to explore the relationship among the phytochemicals and disease resistance in more detail. The relationships identified will help determine the algorithms used in future machine learning models. Goal 3 As data continues to be collected pertaining to Goals 1 and 2, preliminary genomic models (e.g. traditional genome-wide association studies (GWAS) and genomic selection models) will be built. Additionally, as the complete dataset is built, relationships between agronomic, disease ratings, mycotoxin contamination, metabolic, and weather data, as well as genomic data, will continue to be explored. These relationships will make it possible to decide among potential candidate machine learning algorithms, and we anticipate beginning to test some of these algorithms during this project period, at least preliminarily.
Impacts
What was accomplished under these goals?
Goal 1 The 432 maize inbreds included in this study were evaluated in Summer 2021 under both inoculated and natural field conditions. Furthermore, inoculated and non-inoculated trials were conducted at two field sites. These inbred cultivars represent a genetically diverse set of maize that will, following field experiments, provide the genetic information necessary to select the best-suited cultivars for minimizing post-harvest loss. All inbreds were evaluated for susceptibility to Gibberella ear rot. Preliminary heritability estimates have been calculated and are estimated to be approximately 74.7%. While these estimates may change once more environments are included, these high preliminary numbers are very encouraging. In addition to visual ratings of susceptibility to Gibberella ear rot, standard agronomic information (e.g. flowering dates) and weather information were also collected. 108 hybrid pairs were identified based on a combination of genetic clustering information and the Gibberella ear rot ratings. Crosses make use of a combination of Resistant, Moderately Resistant, and Susceptible varieties. Additionally, crosses were designed such that relatively equal numbers of Resistant, Moderately Resistant, and Susceptible varieties, as well as crossing types (e.g. Resistant x Resistant, Resistant x Susceptible, etc.) were used to the extent possible while also considering flowering date and genetic clustering assignments. There are a total of six hybrid crosses between each of the possible 3 female x 6 male = 18 cluster combinations. Mycotoxin extraction protocols have been finalized. Mycotoxin quantitation protocols using LC-MS have been developed and are in the final stages of being optimized. Seed was shipped for mycotoxin analysis in March 2022. Since then, approximately 350 of the roughly 1600 mycotoxin samples have been extracted, although quantitation must still take place; supply chain disruptions have delayed the delivery of necessary reagents and standards for the quantitation portion of this work. This work is projected to be completed by Fall 2022, provided that supply chain issues do not cause delays in the receipt of reagents. To date, the disease expression is much more consistent in inoculated trials than in non-inoculated field trials, which is ideal because it will enable us to obtain baseline levels of the metabolic compounds of interest in this study. Goal 2 All extraction protocols (hydroxycinnamic acids and flavonoids) have been finalized and optimized. Quantitation protocols for the hydroxycinnamic acids have been fully optimized. Quantitation protocols for the flavonoids have been developed and are in the final stages of being optimized (pending supply chain delays). All protocols have been developed such that they will enable us to quickly phenotype the maize samples generated in this study, thereby understanding another facet of the biochemistry that underlies disease response. In addition to the originally proposed metabolic panel, we have been able to secure access to a non-destructive scanning NIR for use in this study. This will allow us to quantify the baseline levels of starch, protein, oil, ash, fiber, and moisture in healthy grain samples in addition to the baseline levels of the hydroxycinnamic acids and flavonoids. Goal 3 Although most work this year has focused on Goals 1 and 2, we have also spent a portion of this year focusing on the development of preliminary algorithms using open-source data. The most promising algorithms currently make use of feature extraction techniques, rather than "blind" modeling approaches that make use of all data irrespective of growing conditions or plant physiology. These algorithms will provide insight as to the data manipulations, feature extraction, and potential algorithmic pipelines necessary for incorporating weather, genetic, and metabolic data into models for predicting plant performance, and ultimately Gibberella ear rot disease pressure and potential loss due to mycotoxin contamination.
Publications
|
Progress 05/01/20 to 04/30/21
Outputs
Target Audience:
Nothing Reported
Changes/Problems:At this point, everything is going very well with the project. Even with the challenges posed by the COVID-19 pandemic, the project team is on track with the proposed timeline of milestones and deliverables. One concern that is emerging, however, is that our company collaborator may not be able to commit as many resources as previously intended in Winter 2021 and Summer 2022, due to COVID-19 budget cuts. We are currently seeking out support from additional industry collaborators, as a contingency plan. What opportunities for training and professional development has the project provided?Four graduate students, all of whom were funded by other sources, had the opportunity to work on this project this year. One graduate student was involved with the plant breeding nursery and seed increase. Two graduate students were involved with the creation of phenotyping protocols. A fourth graduate student was responsible for creating the weather dataset, as described above. How have the results been disseminated to communities of interest?
Nothing Reported
What do you plan to do during the next reporting period to accomplish the goals?Goal 1 All 432 genetically representative inbreds will be grown in both inoculated and non-inoculated conditions at two locations this field season. One location has a history of disease pressure, while the other does not. Visual ratings and standard agronomic information will be collected (e.g. flowering dates, yield, test weight, etc.). This information will be analyzed using SAS and R, and the first-year of field study information will be available next year. This information, along with a genetic clustering analysis based on the genotype-by-sequence data, will be used to identify 27 inbred parents and 108 hybrid crosses for further field study in 2022. Preliminary information will be available regarding the heritability of Gibberella ear rot resistance in maize and the variability in resistance and susceptibility measures following the completion of this year's field season. Mycotoxin contaminationwill be quantified using HPLC protocols. Goal 2 Hydroxycinnamic acids and flavonoids will be quantified for each maize sample. Correlations will be calculated to understand the relationship between these metabolites and disease resistance. Although correlations provide basic insight, more advanced modeling techniques (e.g. generalized linear mixed models) can be used to explore this relationship in more detail. The relationships identified will help determine the algorithms used in future machine learning models. Goal 3 Preliminary genomic models (e.g. preliminary traditional genome-wide association studies (GWAS) and genomic selection) will be built using the data provided by Goals 1 and 2. Although these models are preliminary, the datasets and scripts created during this process can be reused during later, more complex model-building phases of the project. Relationships between agronomic, disease ratings, mycotoxin contamination, metabolic, and weather data, as well as genomic data, will be explored.
Impacts
What was accomplished under these goals?
Goal 1 Seed of 432 inbreds increased so that experiments can take place in summer 2021. These inbred cultivars represent a genetically diverse set of maize that will, following experiments in 2021 and 2022, provide the genetic information necessary to select the best-suited cultivars for minimizing post-harvest loss. Mycotoxin protocols were identified and are under development. Expected completion of protocol development is August 2021. Mycotoxin quantification is vital because mycotoxin contamination is a significant source of postharvest loss, representing a potential source of loss for famers and grain distributors alike. In order to prepare for the large scale screenings in Year 1/2 of the project, large-scale GER inoculation protocols were developed and tested in the field using known resistant and susceptible lines. Goal 2 Laboratory protocols for the quantification of hydroxycinnamic acidswere modified for current laboratory conditions and new equipment. These protocols will enable us to quickly phenotype the maize samples generated in this study, thereby understanding another facet of the biochemistry that underlies disease response. Laboratory protocols for the quantification of flavonoids were identified and are currently under development. These protocols will undergo final development during Summer 2021. In addition to hydroxycinnamic acids, flavonoids are putative deterrents of Gibberella ear rot and mycotoxin deposition. Therefore, understanding how the concentrations of both flavonoids and hydroxycinnamic acids are related to disease severity and mycotoxin contamination is vital when developing new models for crop performance and the minimization of post-harvest loss. Goal 3 Current open-source databases require significant restructuring before they can be used for the creation of algorithms. We have taken the open-source data provided by the Genomes-to-Fields Initiative (https://www.genomes2fields.org/; data available for download through Cyverse) and transformed it such that the data can now be used for the construction of preliminary modeling algorithms which consider crop development and key points of agronomic stress. These algorithms will provide insight as to the data transformations and preliminary algorithms necessary for incorporating weather data into models for predicting plant performance, and ultimately Gibberella ear rot disease pressure and potential loss due to mycotoxin contamination.
Publications
|
|