Breeding for Yield PLS 664 Spring 2007

Issues How do we build yield potential into a cross? How do we select for yield in the generations prior to yield testing? How do we select for yield in the first generations of yield testing?

Yield Potential - where does it come from? Good x good Dandan’s data tanksley

Yield Potential - how to breed for it in very early generations Evauating vigor

Microplots Hill plots Honeycomb design Single row plots

The first yield test - 1 location, 1 replication How meaningful can it be?

Preliminary Wheat Test Logan Co. 2006

Next Step Multiple Environments Why? G x E

Genotype x Environment Interaction Two sources –Differences in scale of genetic variance –Genotype rank changes Genotype rank changes are the real problem

Env. 1Env. 2 V 1 V 2 Yield Genotype Rank Change - The Bane Of The Plant Breeder’s Existence

Yield Potential vs. Yield Stability Locations Yield

Yield Potential vs. Yield Stability Locations Yield

Yield Potential vs. Yield Stability Locations Yield We want it all!

Heritability: Proportion of phenotypic variance That can be explained by genetic effects

The most important function of heritability is its predictive role. Traits with high heritability values can be improved with rapidity and with less intensive evaluation than traits with low heritability (Nyquist, 1991). For example, we evaluate tens of thousands of F 3:4, F 4:5, and F 5:6 lines in short 4-foot rows annually. We are successful in identifying lines with appropriate plant height and maturity in these unreplicated plots because of the high heritability of these two traits. Selection for grain yield, on the other hand, a low heritability trait, begins with F 5:7 lines, grown in 55 square-foot plots in replicated tests at several locations. This intensive evaluation continues over several years for each new set of F 5:7 lines.

Definitions of Heritability Nyquist (1991) pointed out two contexts within which a trait is hereditary: 1) the context of the trait being determined by the genotype of the individuals being observed, This context gives rise to what we call Broad-Sense Heritability (H), which is that portion of the phenotypic variance that is genetic in origin, = = It reflects the correspondence between phenotypic values and genotypic values. If the proportion is high, the genotype plays a large role in determining the observed phenotype; if the proportion is low, then the alternative is true

The second context gives rise to what we call Narrow-Sense Heritability (h 2 ), which is that portion of the phenotypic variance that is due to variance in breeding values among the individuals in a population, or the ratio:. = In this equation refers to additive genetic variance Or the variance of breeding values.

Using our own breeding program as an example, narrow- sense heritability for plant height and maturity must be high in our F 3:4 wheat lines, because we find that the short, early maturing lines we select will generally produce short, early maturing F 4:5 progeny. Narrow-sense heritability gets to the core of the predictive role of heritability. It expresses the degree to which the phenotype is determined by the alleles transmitted from parent to offspring--i.e., is what we see in this generation what we will get in the following generation? It gives a practical interpretation to heritability. It is an estimate of the phenotypic difference between lines which one expects to recover in the progeny.

We can estimate two forms of broad-sense heritability based on the data that this experiment has provided: 1. Per-plot basis 2. Entry mean basis

Heritability on a Per-Plot Basis This experiment involved evaluation of genotypes in a two-rep experiment grown at two locations in two years- common scenario. In order to have some common base-line of comparison, the heritability can be reported on a per-plot basis--the plot being the lowest unit of observation, or selection. H (per-plot basis) = =(19.8%)

Heritability on an Entry Mean Basis While the per-plot estimate of heritability described previously has value as a common baseline comparison, the phenotypic means that are available to us from the experiment analyzed in Table 10.2 were based on data averaged over eight plots (2 replications at 2 locations in each of 2 years). Such data would be far more precise than single plot data not only because we have two replications at each location, but because  2 GY,  2 GL and  2 GLY were all significant sources of variation in the analysis of variance.of variation also.

Having done the work, we can estimate the heritability of the trait based on the Entry Means over the eight values for each line. The formula is H (entry mean basis) = = (or 49.5%). =

Thus, having exposed our material to the vagaries of  2 GL,  2 GY,  2 GLY, and  2, the phenotypic means we have calculated are closer approximations to the true genotypic value than the single plot estimates. As one scans over the yields of the lines based on entry mean data, it can be inferred that for every one unit difference in observed phenotypic means, 49.5% of the difference is due to differences at the DNA level. The remaining 50.5% is due to the blurring of the true genotypic value by environmental and experimental error effects. As we would expect, the performance of the progeny of a line selected on the basis of its entry mean would likely be a closer approximation to the eventual performance of the progeny than the performance of the progeny of a line selected on a per-plot basis.

Heritability on a Single Plant Basis Single Plants Within a Plot or Grid Pattern Selection on a single plant basis can be conducted on plants within plots of, say, S 0:1 lines or half-sib families. Alternatively, one could grow a heterogeneous population on a large area of the nursery and subdivide the area into grids or plots. The superior plant(s) within each grid would be selected without consideration of plants in other grids. Variance among plots Variance Within plots

Heritability on a single plant basis would be estimated as: For convenience we utilize the variances from the previous example. Heritability on a per plant basis is estimated as: h 2 = = 0.12 (or 12%).

Under reasonable management conditions, we would assume that plot-to-plot variation measured over a large number of plants per plot would be less than plant-to-plant variation within plots (  2 w ). Heritability on a per-plant basis is estimated as: Or 11 percent.