Progress 09/01/14 to 08/31/20
Outputs
Target Audience:The target audience was wheat breeders and plant breeders in general Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Over the five years the program has trained directly, or indirectly, 5 graduate students, and perhaps 20 undergrauates who have worked for us during the summer. The project has also participated in traing 2 post-docs and 2 visiting scholars How have the results been disseminated to communities of interest?The reuslts have been presented at several meetings and in publications What do you plan to do during the next reporting period to accomplish the goals?
Nothing Reported
Impacts
What was accomplished under these goals?
1) Determine the impact of up to five cycles Genomic Selection (GS) on target traits and breeding values (GEBVs) The average GEBV for yield was 0 for the training population. It increased to 4.8 (bu/acre) by cycle 5 of GS. The gain was 0.78 bu/ac per cycle. We noted that genetic variation corrected for inbreeding decreased from 16.7 in the training population to 1.7 by cycle 5. The decrease arose from the absence of F2 with low GEBVs for yield. The average GEBV for Fusarium Head Blight Index (FHB) was 0 in the training population. By cycle three the mean GEBV had decreased to -0.42, a decrease of 0.14 per cuycle. Several cycle 4 lines had better resistance than the best lines in the training population. As with yield, genetic variation decreased from 13.8 to 3.71 with the decrease, arising from a general absence of lines with susceptible scores. It was difficult to see a definitive trend for increased yield or FHB resistance over cycles among the lines that were field tested from each cycle. We did observe a positive correlation of GEBVs with observed trait value for yield and FHB. For yield, this correlation ranged from 0.11 to 0.40 when analyzed for phenotypes collected within individual years. It is not too surprising that the correlations are not greater as we observed strong genotype x year interactions for wheat grown in Ohio. For example, the correlation of yield between years is on average 0.09 for lines in stage-1 and stage-2 trials and 0.28 for stage-2 and stage-3 phenotypes. Thus, it is likely that the environments sampled when phenotyping the training populations differed from those sampled when testing the lines derived from the cycles of GS. The correlation of FHB GEBVs with observed FHB resistance of lines derived from GS ranged from 0.28 to 0.62 by individual years. FHB does not experience as much genotype x year interactions as does yield, in large part because the FHB data is collected from a nursery designed to promote high FHB disease pressure. The correlation of FHB phenotypes from stage-1 to stage-2 is 0.38 while it is 0.48 between stage-2 and stage-3 Conclusion: GS was quite effective at improving the mean GEBVs and elimination lines with undesirable GEBV for yield and FHB. While there was not strong trend for improve field performance over cycle of GS, we did note a positive correlation of GEBVs with actual phenotypes for both yield and FHB. It is likely that the preveance of genotype x year interactions in Ohio limited the effectiveness of GS. As in traditional breeding, GxY interactions are likely the greates limitation to the effectiveness of GS. 2) Assess the impact of GS on the diversity, allele frequencies, and linkage disequilibrium We assessed genetic diversity in the training population and within each cycle of GS by estimating the probability os sampling alleles that are identical by state (IBS) and polymorphism information content (PIC). There was an increase of the average IBS value from 0.68 in the training population to 0.74 among the cycle 5 F2: this is an increase of 0.012 per cycle. The average PIC value decrease from 0.33 in the training population to 0.26 by cycle 5, a decrease of 0.07 per cycle. We noted a mild increase in the percentage of marker that became fixe over the first 4 cycles but a large increase between cycles 4 and 5. We called a marker where the minor allele frequency was < 0.01 to be fixed. A total of 9,321 markers were scored in the training population and each of the 5 cycle. Of these, 0% were fixed in the training population while 7.48% were fixed by cycle 4 of GS. This percentage increased to 25.9% in cycle 5. It is unknown why the percentage of fixed loci increased dramatically from cycle 4 to cycle 5. We did not observe a discernable change in linkage disequilibrium over the cycle. Conclusion: The process of going through cycles of GS reduced molecular variation. The rate was steady when assessed by IBS and PIC values, by dramatic when assessing the percentage of fixed loci. 3) Develop genomic breeding strategies to design crosses to attain defined goals and utilize family selection. We assessed the impact of pedigree selection with cycles of GS. We selected superior F2 plants based on GEBVs for yield. These were selfed and we then predicted the yield of the F3 plants derived from the superior F2 plants. The same F2s were also crossed among them selves to produce the next cycle of GS. We compared the GEBVs of F3 advanced from cycle-n by pedigree to the average GEBV of F2 in cycle-n+1. The average yield GEBV of the cycle-n derived F3s was on average 12% greater than the average GEBVs of the cycle-n+1 F2. We determined that one did not really need to modify POPVAR to attain our goals. POPVAR produces a simulated population from each cross and calculates a GEBV for each. One can obtain the standard deviation of the GEBVs of the population. The standard deviation can be used to calculate the probability of an individual with a GEBV that is above the average of the parents, or superior to the value of the best parent (a transgressive segregant), or some other defined value. Given that such an individual occurs with a probability of x, then you can calculate the size population you need to evaluate to be 95% certain of producing such an induvial as ln(1-0.95)/ln(1-p). For example, if the probability of a desired type in 0.005 the you need to evaluate 597 individual from that cross to be 95% certain such an individual will exist to the evaluation population. If p=0.01 in another cross then you need only evaluate 297 individuals to be 95% certain. Conclusion: Pedigree selection can be effective within F2:3 families that are derived from superior F2 plants. A simple extension of POPVAR can not only identify the most promising crosses but also the size population you need to grow to attain some specific objectives.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2020
Citation:
Russell, B, G Brown-Guedira, C Sneller, and M Mohammadi. 2020. Transferability of Marker Trait Associations in Wheat is Disturbed Mainly by Genotype x Year Interaction. Crop Breeding Genetics and Genomics 2020;2(3):e200013. https://doi.org/10.20900/cbgg20200013
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Ward BP, Brown-Guedira G, Kolb FL, Van Sanford DA, Tyagi P, Sneller CH, et al.. (2019) Genome-wide association studies for yield-related traits in soft red winter wheat grown in Virginia. PLoS ONE 14(2): e0208217. https://doi.org/10.1371/journal.pone.0208217
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Gaire R, M Huang, C Sneller, C Griffey, G Brown-Guedira, and M Mohammadi. 2019. Association analysis of baking and milling quality traits in an elite soft red winter wheat population. Crop Sci. 59:1-10 doi:0.2135/cropsci2018.12.0751
- Type:
Journal Articles
Status:
Published
Year Published:
2018
Citation:
Huang M, N Mheni, G Brown-Guedira, A McKendry, C Griffey, D Van Sanford, J Costa, C Sneller. (2018). Genetic analysis of heading date in winter and spring wheat. Euphytica 214:128. https://doi.org/10.1007/s10681-018-2199-y
- Type:
Journal Articles
Status:
Published
Year Published:
2017
Citation:
Huang M, B Ward, C Griffey, D Van Sanford, A McKendry, G Brown-Guedira, C Sneller. 2017. The prediction accuracy of genomic selection between environments and populations for wheat traits. Crop Sci. 58: doi: 10.2135/cropsci2017.10.0638
|
Progress 09/01/18 to 08/31/19
Outputs
Target Audience:Plant breeders in the USA and in Africa Changes/Problems:We are hopeful that we will collect useful data from the 2019-2020 field trials after several frustrating years of poor field trials. What opportunities for training and professional development has the project provided?5 undergraduates were hired to work part time in the summer. The project also was associated with 1 part-time graduate student and 1 part-time post-doc How have the results been disseminated to communities of interest?Some results have been presented at workshop and meetings What do you plan to do during the next reporting period to accomplish the goals?Field trials have been planted in the fall of 2019 to produce data in 2020. We will develop the PopVar+ package and will finalize the changes in genome structure from cyccles of GS.
Impacts
What was accomplished under these goals?
We have determined that the population has shifted through cycles of genomic selection with the progeny in the fifth cycle deviating significantly from the initial training population. We have noted a decrease in diversity through the cycles of GS and the loss of alleles. We planted the progeny from cycles of genomic selection at three Ohio locations. The trials were lost from two location due to excessive water logging throughout much of April, May and June and 2019. Data was collected from the third location though yields were low. The field trials were planted at the same three locations again in the Fall of 2019 to produce data in 2020. The trials were also planted in a Fusarium head blight nursery for evaluation in June of 2019. Again excess water logging caused many plots to die prior to collecting FHB data. These trials were replanted in the fall of 2019 to produce data in June of 2020.
Publications
- Type:
Journal Articles
Status:
Under Review
Year Published:
2020
Citation:
Borrenpohl, D, Huang, M., Olson, E.,and Sneller, C. 2019. The value of early stage phenotyping in the era of genomic selection. Theor Appl Genet (submitted)
|
Progress 09/01/17 to 08/31/18
Outputs
Target Audience:The target audiences are other plant breeders, quantitative geneticists, and wheat growers. Changes/Problems:We have had problems getting good yield data on lines derived from cycles of GS in 2015 (water damage), 2016 (water damage), and 2017 (bunt). Thus we have added field trials in the 2018-19 season. We are requesting a no-cost extension so additional field tirlas can be conducted in the 2019-2020 season to insure that adequate field data is obtained. We had to rescore the genotypes of all training population and cycle 1-5 individuals. This required us to learn the biouinformatic pipeline, acquire access to the Ohio Super Computer, and then call all the alleles using a newer reference genome. What opportunities for training and professional development has the project provided?The project trained one post-doc, one PhD student, and 4 undergraduates (part timers). How have the results been disseminated to communities of interest?The PI, post-doc, and graduate student havegiven presentations based on the results in the USA, Ivory Coast, Nigeria, Uganda, Kenya, Zimbabwe, and India. Publications are being prepared. What do you plan to do during the next reporting period to accomplish the goals?We are currently finalizing the analysis of genome changes over five cycles of GS. Wehave planted yield trials and FHB trials that will provide data in June and July of 2019. A post-doc is beginning to work on the crossprediction program. We are behind schedule in several areas due to issues with phenotyping and the need to rescoreall genotypes. Thus we are requesting a non-cost extension so we can complete all objectives by the end of August 2020.
Impacts
What was accomplished under these goals?
Objetive 1: The GEBVs for yield increased from an average of 0.0 in the training population to 4.9 bu/ac in cycle 5, a gain of about 1 bu/ac per cycle of GS and 1 bu/ac per year. This is similar to yield gains from phenotypic selection in the OSU program. In the Fusarium Head Blight (FHB) population, the average GEBV for resistance decreased from 0.0 in the taining population to -0.42 in the cycle 3 progeny, a rate of -0.15 per cycle and per year. Genetic variance for yield and FHB resistance decreased sharply after one cycle of GS. The decrease was due to the absense of individuals with undesirable GEBVs. Objective 1: We phenotyped progeny from cycles 1, 2, and 3 in the field for yield in 2018 and for FHB in 2015, 2016, 2017 and 2018. The correlation of observed and predicted yield was 0.40 in 2018. The 10% of the lines with the best GEBV yielded 16% above average. Previous attempts to obtain good yield data in 2015, 2016, and 2017 had failed due to water damage and disease. For FHB the correlation of observed and predicted resistance ranged from 0.20 to 0.61 over four different evaluation trials. Objective 2: We assessed gentic diversity of the progeny of five cycles of genomic selection for yield. The probability of sampling alleles that were identical-by-state increased from 0.68 in the training population to 0.72 in the cycle 5 progeny. The average Polymorphism Information Content (PIC) valuewas 0.32 in the training population and 0.26 in the cycle 5 progeny. Objective 3: We conducted pedigree selection using GS and compared the average GEBVs of F3 individualsadvanced by pedigree selection (eg selecting F3 plants derived from the F2 plants with the best GEBVs) to those advanced by recurrent selection based on GEBVs (eg selecting best F2s, crossing them, and deriving the next F2 population). F3 individuals advanced by the pedigree method had 10-14% higher GEBVs for yield than F2 populations advanced by recurrent selection. Objetive 3: We have not yet modified the programs to predict the value of crosses.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2019
Citation:
Ward, B. P., G. Brown-Guedira, P. Tyagi, F. L. Kolb, D. A. Van Sanford, C. H. Sneller, and C. A. Griffey. 2019. Multi-environment and Multi-trait Genomic Selection Models in Unbalanced Early-Generation Wheat Yield Trials. Crop Sci. 0. doi:10.2135/cropsci2018.03.0189
|
Progress 09/01/16 to 08/31/17
Outputs
Target Audience:The target audience in wheat breeders and breeders in general Changes/Problems:Due to the difficulties of getting useful phenotypic data from the 2016 and 2017 field trials, we would like to propose that we continue the project for one additional year to shore up the field data portion of the project. What opportunities for training and professional development has the project provided?The project has trained 1 post-doc, 1 PhD, 1 MS, 1 intern, and 5 undergrauate labores in the past year How have the results been disseminated to communities of interest?Results have been presented at local, state, regional, national, and international meetings. Publications are being prepared What do you plan to do during the next reporting period to accomplish the goals?We are behind in our phenotyping plans due to a heacvy infestation of smut (2017) and water damage (2016) that precluded collecting yield data. Thus we plan field trials in the 2018-19 season and hopefully in the 2019-20 season if we can get a no-cost extension. We will propare publication on the genome changes, and the field trials for FHB. We will finalize the coding for a program to expand on POPVAR.
Impacts
What was accomplished under these goals?
In this year we have rescored all the marker data from 7000 lines which total close to 75 million data points. We have begun analysis of the diversity over five cycles of GS. We have observed a modest decline in diversity over the five cycles (probability of identical by state raising from 0.68 to 0.72). The genetic distance between the training population and the F2s from cyclce of GS have increased through the five cyclces: that last PCA graph shows the cycle 5 F2s occupying a different genetic space than the training population lines. Also the proportion of marker loci with a minor alleles frequency < 0.05 has increased from 0.01% to 25% and a significant number of marker loci have become fixed. The average predicted yield has increased over the five cycles, from 0 in the training population to 4.8 in cycle five. Still no cycle five F2 has a superior predicted yield than the best lines in the training population: this could be because F2s are very heterozygous cannot express their full potential. Some F2s have nearly equal yield as the best training population lines. We were able to get good yield data in 2018 after several rough season of trying to get phenotypic data due to smut and water damage. We assessed the yield of 173 F2-derived lines. The correlation of the observed yield and predicted yield was 0.41. We have also analyzed data from three cycles for GS for Fusarium Head Blight resistance. Over 4000 F2 plants have been genotyped over the three cycles. After three cycles the average predicted FHB score has dropped from 0 to -0.45 and F2s with better resistance than the best training population lines have been noted in each cycle. These lines have been phenotyped and the correlation of the observed phenotyped of the F2-derived lines and their prdicted values ranges from 0.28 t0 0.62 with an avareg of 0.48.
Publications
|
Progress 09/01/15 to 08/31/16
Outputs
Target Audience:Plant breeders and geneticists Changes/Problems:We had issues with re-calling alleles. This was solved by developing the capacity to conduct the bioinformatics pipeline at Ohio State University. The other major isses was the heavy infection of smut in the 2017 plots that severly compromised the quality of the 2017 yield data. This issue was address by treating all seed prior to planting in the fall of 2017. The trials that were to be analyszed using 2017 data have been replanted to obatin their data in 2018. What opportunities for training and professional development has the project provided?The project has trained one post-doc, two PhD students, one MS student, and one visiting scholar. How have the results been disseminated to communities of interest?Results have been presented at a research conference. Two publications have been submitted What do you plan to do during the next reporting period to accomplish the goals?Now that we have re-called all the alleles we can quickly move along on all the objectives. We will first predict the value of all F2s from all cycles of both populations, corrlate those with currently obtained phenotypes, estimate the correlation of the LD matrices between all cycles, and assess changes in alleles frequencies over the cycles.
Impacts
What was accomplished under these goals?
Cycles 4 & 5 of GS in the Yield +FHB populations have been completed. Cycles 2 & 3 of GS in the FHB population have been completed. We have now re-genotyped all the material from the "Yield+FHB" population useing the GBS system developed by J Poland at Kansas State University. Our first attempts to call alleles failed due to errors in the key file. It was then determined that it would be better for Ohio State University personnel to run the bioinformatics than for the KSU staff to devote further time to that endeavor. The OSU staff learned how to run the bioinformatics pipeline and we have now re-called alleles in the entire "Yield+FHB" population (eg traing population and cycle 1, 2, 3, 4 and 5 F2s). This has also been done in the FHB populationn for the training set and the cycle 1, 2, and 3 F2s. Field plots were first established to assess yield gains from GS in 2015. Stands were compromised due to a wet spring. Additional plots were established in 2016 and acceptable quality yield data was obtained. The plots were established again for a 2017 harvest but smut severly compromised the quality of the yield data at all locations. Seed was treated and plots for the 2018 season have been planted at three locations in the fall of 2017, for 2018 harvest, with good stands. Data for FHB has been obtained in 2016 and 2017 and a trial established in the fall of 2017 for 2018 data collection. We have not analyzed data from the field trial as we have just re-called the alleles for all of the lines from the Yield+FHB and FHB populations. We will soon predict the yield and FHB resistance of all F2 deirved lines from all cycles of both populations and then assess how the predict values correlate with the obtained phenotypes. The diversity of the Yield_FHB population has decreased slightly through four cycles of GS with the average relationship among individuals increasing by 8.7% from the TP to cycle 4.
Publications
- Type:
Journal Articles
Status:
Submitted
Year Published:
2017
Citation:
Cabrera A, M Huang, A McKendry, F Kolb, D Van Sanford, M Sorrells, J Tanaka, E Olson, J Poland, C Sneller. 2017. Genome-wide Analysis of Resistance to Fusarium Head Blight in Soft Winter Wheat: Association analyses. Submitted to Crop Science, December 2017.
- Type:
Journal Articles
Status:
Submitted
Year Published:
2017
Citation:
Cabrera A, J Isidro, M Huang, A McKendry, F Kolb, D Van Sanford, M Sorrells, J Tanaka, E Olson, J Poland, C Sneller. 2017. Genome-wide Analysis of Resistance to Fusarium Head Blight in Soft Winter Wheat: Genomic Selection and Optimizing the Training Population. Submitted to Crop Science, December 2017.
|
Progress 09/01/14 to 08/31/15
Outputs
Target Audience:Plant breeders Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?The project is training one research associate and one PhD student 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?The project is mostly on schedule. We are a bit behind on regenotyping the F2s from cycles 1 and 2, but that does not delay progress in other areas. The FHB phenotyping of cycle-1 F2-derived families failed in 2015 and will be repeated as soon as seed is avaialble. All other tasks are on schedule.
Impacts
What was accomplished under these goals?
1) Determine the impact of up to five cycles Genomic Selection (GS) on target traits and breeding values (GEBVs) In the Y&F population we genotyped the training population (n=470) and 1,821 plants: 988 cycle 3 F2; 452 cycle 2 F3; 381 cycle 1 F4. The GBS system provided about 8,000 scored markers per plant. GS models were created for all traits (YLD in Wooster, YLD over all environments, FHB resistance) using genotypic and phenotypic data from the training population. We completed 126 crosses among the plants (F2, F3, F4) with the best GEBV for yield traits. The resulting F1 were grown and F2 seed will be harvested and replanted in September of 2015. An additional 17 crosses were made between the best plants for FHB resistance. In the FHB population we genotyped the training population (n=641) and 1,045 cycle 1 F2 plants derived from 52 crosses among the best training population lines. We obtained scores for about 4,600 markers and used those to model their GEBVs . We made 155 crosses among the F2s with the best GEBV for FHB resistance. We note that the mean GEBV for yield of the F2 (etc) from cycles of GS are greater than those of the training population, while the GEBVs for FHB are lower. These indicate that GS is effective at changing GEBVs. The increase in the mean occurs primarily by eliminating individuals with very poor trait values. Just a few individuals have GEBVs that are equal to the GEBV of the best training population line. We attempted to phenotype 47 training population lines and cycle 1 F2-derived families for FHB in the 2014-15 season. Unfortunately disease pressure was too low to produce useful phenotypic data for analysis. We yield tested 89 F2-derived families from cycle 1 and 19 training population lines for yield in 2014-15. The F2 derived lines were planted in just 1 replication whereas the TP lines were replicated 2-3 times. This data has not been analyzed yet. We increased seed of the best/worst F2 derived lines from cycle 2 and these will be placed in FHB and yield trials (limited reps) in the 2015-16 season. 2) Assess the impact of GS on the diversity, allele frequencies, and linkage disequilibrium We are in the process of isolating DNA from the cycle 1 (n=923) and cycle 2 (n=1002) F2 so the diversity and LD analyses can be executed. We had proposed doing these isolations in March and May of 2015 but we were not able to identify a graduate student for this project until August of 2015. We are writing R scripts that will be used to analyze this data so they will be ready when the GBS data is available for cycles 1,2,3 and 4 F2 in December of 2015. 3) Develop genomic breeding strategies to design crosses to attain defined goals and utilize family selection We are investigating methods of reducing the number of markers to use in prediction to one less than the number of lines in the training population. We are using significance from association analyses and clustering based on an LD matrix. To date we have found that if we use only marker that are significantly associated with a trait at p <0.05 that we can obtain a GS accuracy that is the same or higher than if we use all markers. We genotyped 381 F4 plants derived from some of the best cycle-1 F2s (eg sampled within an F2:4), and 452 F3 plant derived from the best cycle-2 F2 (eg sampled within an F2:3). For yield over all locations, the average GEBVs of the F4 was just 88% of the average GEBVs of the cycle-3 F2 while the average GEBVs of the F3s was 108% of the GEBV of the cycle-3 F2. For yield in Wooster, the average GEBVs of the F4 was just 69% of the average GEBVs of the cycle-3 F2 while the average GEBVs of the F3s was 119% of the GEBV of the cycle-3 F2.
Publications
|
|