September 18th, 2020
The East Africa Pipeline
1. Introduction to the problem
Crop by region
CIP-Sweetpotato-ECA
Problem specification
The number of parents, crosses and progeny have an impact on genetic gain. Too large or too small can affect genetic gain
Breeding strategy component tackled
Crossing, Evaluation, Selection
Breeders’ equation terms tackled
\(\sigma_g\), i
\(\Delta_g = (i * \sigma_g * r)/L\)
Hypothesis
Finding the optimal number of parents, crosses and progeny will help increase genetic gains.
2. Materials and methods
Treatments
| Treatment | Description |
|---|---|
| Grid1500 | Expanding a grid restricted at 1500F1 (nParents = 15,60,5), (nCrosses = 5,250,5), and (nProgeny=5,50,5) |
| Grid4000 | Expanding a grid of scenarios restricted to between 1500-4000F1 (nParents = 15,60,5), (nCrosses = 10,150,5), (nProgeny = 5,200,5). |
Simulation procedure
A 20 year burn-in period was modeled using the baseline. The burn-in was followed by a 30 year evaluation period to measure rates of genetic gain for all treatments. Genetic gain was measured by assessing changes in genetic merit at AT. Genotype-by-year interaction variance was assumed to be equivalent to genetic variance (based on average correlation between locations being equal to 0.5). 20 replications done. We considered 6x ploidy and 0.2 multiallelic recombination. The hybrids were restricted to 6000 according to the baseline.
3.0 What we know from previous simulations
The number of testers between 2-5 is good. We stuck with the baseline (nTesters = 3)
3.1 Results comparing best and worst scenarios
Best and worst scenarios @1500 and @1500-4000
There was no big difference for best and worst scenarios @1500 restriction. We proceeded with further analysis with the @1500-4000 grid
3.2 Optimal nParents
The lower the nParents (15-30), the better
3.3 Optimal nCrosses
The higher the nCrosses, the better
3.4 Optimal nProgeny
The lower the nProgeny compared to nCrosses, the better
3.5 Looking at best and worst scenarios
Appears that the current 150 crosses can be reduced by half and the nProgeny (10) be increased 2x to better sample the value of each cross
4. Conclusion
There appeared to be no difference in scenarios when the F1 were restricted a @1500 as per the baseline. Increasing the range to @1500-4000 showed clear differences between best and worst scenarios. Optimal number of parents ranged between 15-30. Results show the need for better sampling of diversity among parents using a higher nCrosses. Although the current numbers (nParents = 20, nCrosses = 150 and nProgeny = 10) seems good, results suggest that this nCrosses can further be reduced to accommodate a bit more nProgeny in order to better represent the value of each cross.