July 9th, 2020

1. Introduction to the problem

Crop by region

CIAT EA Beans

Problem specification

Selection uses a base index that could be improved in accuracy by taking into consideration the genetic covariance among traits (i.e. Smith-Hazel index).

Breeding strategy component tackled

Selection

Breeders’ equation terms tackled

r

Hypothesis

Increasing accuracy will increase the rate genetic gain.

2. Materials and methods

Treatments

Treatment Description
BASEINDEX Current scheme, selecting parents at PYT (20), VEF (5), OF (10) using a base index with weights provided by Bodo Raatz.
SHINDEX Current scheme, selecting parents at PYT (20), VEF (5), OF (10) using a Smith-Hazel index considering genetic correlation and weights provided by Bodo.

 

Simulation procedure

A 20 year burn-in period was modeled using the current breeding scheme. The burn-in was followed by a 40 year evaluation period to measure rates of genetic gain for all treatments. Genetic gain was measured by assessing changes in genetic merit in F9 lines. Genotype-by-year interaction variance was assumed to be equivalent to genetic variance (based on average correlation between locations being equal to 0.5). All evaluations were conducted using 15 replications.

We simulated 4 complex and 3 simple traits to be behind the genetic merit and inferred through a selection index.

What other simulations have shown?

Do simulations show that picking transgressive individuals is a good method to increase genetic gains?

T1: use an index to pick the best for total merit (10%)

T2: pick the best for yield and then best for zinc (31% > 31% = 10%)

T3: pick the best individuals for each trait (top 5% in each = 10%)

What other simulations have shown? (continuation)

  1. Selecting for individual traits ignoring the others is a terrible approach.

  2. A selection index outperforms Independent culling (multiplicative index) for recycling given a fixed set of weights.

3.0 Results for 2 traits (Network trials with low accuracy)

By year 40, the use of a Smith-Hazel index generated 1.44 (95% CI: 1.06,1.96) times more gain, than the base index.

3.1 Results multi-trait pleiotropy

By year 40, the use of a Smith-Hazel index generated 0.98 (95% CI: 0.93,1.02) times more gain, than the base index.

3.2 Results multi-trait with provided G

By year 40, the use of a Smith-Hazel index generated 1.02 (95% CI: 0.97,1.07) times more gain, than the base index.

4. Conclusion

Considering genetic correlations to create more accurate selection indices has a positive impact in the rate of genetic gain only when the traits are highly correlated and accuracy of selection (and H2) is low for one of the traits.

Unless selection is applied under these conditions we do not recommend to prioritize an improvement plan for using a Smith-Hazel index instead of a base index since current selection occurs in MET and multi-rep trials.