Bulk Selection Methods

The Basic Scheme

• Grow out the F2 generation. Allow natural selection to occur. Harvest seed from remaining plants.

• Grow out the F3 generation. Allow natural selection to occur. Harvest seed from remaining plants.

• Continue for several generations. Select single plants, grow progeny rows, harvest seed for yield testing.

Florell (1929)

• 19 wheat crosses grown as bulks• First selections in F5 or F6• Selected 9-49 heads per cross• Planted head rows, kept best (looking) 45

• Increased seed for yield test• 33 of 45 beat the check. Top yielding line was 55 bu/a vs. 37 bu/a for check

A procedure for inbreeding a segregating population until the desired level of homozygosity is achieved. The seed used to grow each inbreeding generation is a sample of that harvested from plants the previous generation. The bulk method is used predominately with inbred species. It is totally unsuited for fruit crops and most vegetables. Unlike the mass selection method, no human imposed selection is made during successive generations of inbreeding in bulk.

Bulk selection was developed in Sweden in the early 20th century. It was trying to handle segregating generations of a winter wheat hybrid that attempted to combine winterhardiness and high yield. The bulk, grown over several years in Swedish winters, was influenced by natural selection during the approach to homozygosity. Natural selection increased the frequencies of winter-hardy types in the population.

Because of the importance of natural selection in this method, the breeder should carefully choose environments in which natural selection is likely to favor the desired genotypes in the population, i.e., a population segregating for disease resistance should be grown in the presence of the pathogen in order to reduce the productivity of susceptible plants. Because of this concern for the environment, the population undergoing the bulk method would rarely be grown in off-season nurseries.

The central issue in bulk population breeding is the nature of this correlation : “Is the ability to survive in competition related to agricultural worth?”

Genetic ConsiderationsGenotypic frequencies in a population inbred by the bulk method are determined by four variables associated with natural selection in a heterogeneous population.

1) genetic potential of a genotype for seed productivity

2) competitive ability of a genotype

3) influence of the environment on the genotype expression

4) sampling of genotypes to propagate the next generation

There is no way to know if certain F2 plants have progeny in F3 generation, or later generations. There is also no way to predict the genetic variability for any character in any generation. If variables favor the desired genotype, the frequency of desired genotypes will increase; if not, you’re out of luck!

Subdivision/types of Bulk MethodPractical short-term breeding : Up to F2 to F6 generation. Survival among pure line genotypes Survival of different plant types in hybrid - derived populations

Long-term Evolutionary breeding Composite cross method

Long-term Evolutionary breeding Composite cross method

The two methods are separate entities. If one selects out of F4, F5, or F6, having bulked in F2, the probably impacts of natural selection will be small in most cases. This does not mean insignificant, however. On the other hand, if one selects out of a bulk propagated for 15 to 20 generations, the impact of natural selection may be much greater.

Survival among pure line genotypes : Harlan, J., and Martini, 1938. J. Agric. Res. 57:189-199.

They reported on the effects of natural selection in cultivated plants. Mixed 11 pure lines of barley and grew them for 4-12 years at Ag. Experiment Stations from Davis, CA to Ithaca, NY. The change in composition was monitored at each location.

% of Excellent competitorspop Intermediate Poorer of 2 eliminated

in straight line; other Intermediate type in humped curve

Poor competitors At least 1 line wasEliminated rapidly atEach station.

Results1 or 2 lines became dominant very quickly at each location. Natural selection in barley is a significant force at all locations and a force of great magnitude at some locations.

Similar experiments were conducted by Suneson and Wiebe (1942) with four varieties of California barley. They mixed equal amounts of the varieties and a census was taken annually over a 9 year period of propagation. They also measured the mean agronomic performance in adjacent plots. ‘Atlas’ quickly dominated the bulk population. ‘Hero’ and ‘Vaughn’ were virtually eliminated. But differences in productivity of the cultivars was small.

The experiment was not conclusive in establishing a relationship between yielding ability and survival. ‘Vaughn’, the highest yielder, was the poorest competitor; so there was not a very strong positive association between these characteristics. The marked differences in competitive ability must have depended on other characteristics of competitive ability besides yield. Later Josh Lee (1960) provided information that might explain superiority of ‘Atlas’ over ‘Vaughn’. ‘Atlas’ accounted for 55% of spikes in mixed stands, but in a pure stand it has 40% fewer spikes than ‘Vaughn’. He found that ‘Atlas’ also produced a larger root system then ‘Vaughn’.

An important feature of these experiments, which may not necessarily relate to the bulk population method, is that these experiments all involved inbred homozygous lines. But the predominant use of the bulk method is to “self-in-bulk” until homozygosity is reached. Segregation will occur in bulk hybrids with the result that the competing genotypes are not expected to be constant from one generation to another. In previous studies, the populations investigated were very simple in terms of components – pure lines. It was not possible to assay for the survival of an allele or allelic combination per se because that allele or allelic combination was always associated with all other alleles in the genotype.Only at the time of homozygosity (F6 to F8) will the situation begin to represent that of variety mixtures.

Plant height segregated in this cross in a monofactorial basis. Samples of the segregating generations were from the F2 to the F6 generation. One sample was grown with zero Nitrogen and a second with high N.

Obviously, the short plants were at a disadvantage in competition with tall ones in this bulks in both cultures. However, the disadvantage was intensified under high N conditions. In a pure stand the short plants outyielded the tall by a mean of 35%. In this study, natural selection was counterproductive to the goal of breeding high yielding rice cultivars. Tall genotypes had vigorous vegetation which gave them a decided competitive advantage for light interception. Such a high differential in reproductive rates caused the demise of the short, high yielding genotypes very rapidly.

Peta (TT) x Taichung Native I (tt) = F1 selfed segregating F2

(tall, leafy) (dwarf)

% DwarfExpected % Dwarf Low N High N

F2 25 24.9 24.9F3 37.5 31.2 22.4F4 43.8 33.5 18.9F5 46.9 28.3 14.2F6 48.5 23.7 8.8

Jennings recommends roguing the population to remove the good competitors in order to eliminate the bad effects of competition in segregating populations. But Allard concludes that agronomically poor types are generally poor competitors. So differences in opinion exist.

Composite Cross Method : Suneson, C. 1956. Agron. J., 48:188-190.

The composite cross method can be called “evolutionary plant breeding” because it uses composite crosses and natural selection. The rationale is survival ability. Survival ability is sometimes detrimental to productivity in a pure stand but has special significance for determining agricultural fitness.

Develop Composite Cross II : crossed together 28 diverse cultivars of barley (378 crosses). Grown in bulk from F2 to F29 without selection.

Periodically compare yielding ability of CCII with ‘Atlas’ using remnant seeds from various generations.

% AtlasF3 to F4 67.6F7to F8 85.0F11 to F14 88.0F15 to F20 106.0F21 to F23 113.8F24 135.7

The composite bulk was conspicuously inferior to ‘Atlas’ in yielding ability and general agronomic appearance in early generations. There was a gradual improvement, however, in both yielding ability and agronomic type until, by F15 the bulk equaled ‘Atlas’ and then continued to improve. So natural selection did cause significant increased in yield.

After 15 generations :1) Continued natural selections for significant gain in yield.2) Cyclic hybrid recombination alternated with natural selection.3) Conventional selection and testing to identify best lines.

Evolutionary plant breeding in barley had produced numerous cultivars. It can be thought of as a “stretched” bulk method, resting on the dynamics of natural selection. It is very unique in the length of its development period. It is theoretically and practically sound, and its greatest value would be as an adjunct breeding program. Because of the long wait before selection can begin, it is not feasible as a single method breeding program. The department and your farmers might will not be very happy if it takes you over 20 years to develop a cultivar!! Perhaps a compromise would be to treat all crosses by the short-term bulk method, forming derived lines in the F4 – F6 generations for the breeding project. If the population is judged useful for the future, it could be continued as an evolutionary breeding entity.

Pros

• Automatic increase in proportion of homozygous plants with each generation

• May increase mean yield via natural seln.

• Final purification of lines simplified• Selection among crosses can occur before selection within crosses begins, thus the elite crosses are the ones that remain in the program

Cons

• Plants of one generation not all represented by progeny the next generation

• Genotypic frequencies and genetic variability cannot be defined readily

• Bulk method not suited to greenhouses, off season nurseries

• Environment must be suitable to reinforce breeding objectives

Early Generation Testing

• Objective: identify those populations that are likely to contain superior lines

• Strategy: eliminate those populations of low potential from the inbreeding process

• Goal: maintain and develop lines from populations with high genetic potential

Jenkins, 1935

• Usual method of estimating combining ability in maize was to inbreed lines, then mate them to a common tester

• Jenkins saved seed from S0:1 lines through many selfing generations, then crossed them to common tester

• Found that combining ability was already determined in S0:1 lines

Self-Pollinated Crops

• Determine the generation for testing

• If it is to be the F2, you will have to grow the F1 in an environment which favors seed production

• A more common choice would be F2:3 lines

Self-Pollinated Crops

• Harvest seed from individual F2 plants

• Plant seeds in F2:3 progeny rows

• Identify the superior rows• Harvest all seed in each selected row in bulk

• Grow replicated tests of F2:4 lines

• Grow replicated tests of F2:5 lines

Self-Pollinated Crops

• Harvest selected F5 plants individually

• Grow F5:6 lines in headrows

• Test F5 - derived lines extensively

Breeder’s Decisions

• Generation to test• Number of reps, locations and years - tradeoff between early and late generation testing

• Separate program for inbreeding or not

• Selected lines can be advanced by pedigree, bulk, or SSD

• Number of plants chosen from each hetergeneous line may vary

Genetic Considerations

• Recall that there is all of the additive variance among F2:3 lines and one-half of the additive variance within F2:3 lines

• In later generations of F2 derived lines, there is still all of the additive variance among lines, and an increasing amount within lines, as inbreeding progresses

Genetic Considerations

• Therefore, one may need to take a large number of heads to adequately sample the variation within the F2 - derived line

• Now one must decide how to allocate resources

• Should you sample more lines or more selections within lines?

Pros

• Inferior individuals and crosses are discarded early in the process

• One hetergeneous line may yield more than one cultivar

Cons

• When you commit a lot of resources to early generation testing, you cannot devote as much to thorough evaluation of more inbred material

• If you spend a lot of time testing the early generations, cultivar release may be delayed