We suggest using maternal haploid plants for increasing the efficiency of recurrent selection in maize. Maternal haploids exhibit a fair degree of viability. Partial fertility of their ears enables the selected haploids to be involved in the breeding process. Selection of haploid plants carrying useful genes is much easier as compared with diploid plants. Expressed in haploids are all the alleles: both dominant and recessive. Dominance and overdominance are completely lacking. This facilitates selection for genes with additive and epistatic effects.
Segregation of maternal haploids differs significantly from that of diploids. In haploids, it corresponds to the segregation of gametes. The frequency of useful gene combinations is much higher in haploids than in diploids. Natural selection is an important factor in improving diverse material in breeding programs involving haploids. Haploid plants carrying harmful genes either die at various stages of embryo and seedling formation or are rather stunted and sterile.
In our work maternal haploid plants were used to improve two synthetic populations of maize: SA and SP. These are populations of dent maize carrying 50% each of germplasm of North-American inbred lines. Maternal haploids were produced in a space-protected nursery. The synthetic population and the line inducing maternal haploids were sown in alternate rows, two rows of the former per row of the latter. Before flowering, tassels were removed from the plants of the synthetic population. Pollination with pollen from the inducer line occurred spontaneously, by wind. The resulting seeds were selected for the expression of gene R-nj. The putative haploids were those kernels in which the R-nj gene was expressed on the endosperm but was lacking in the embryo.
The resulting seeds with haploid embryos were sown in the field using an ordinary seeder. Each year, no less than 2000 haploid plants obtained from the improved synthetic population were grown. The haploid plants produced from the initial synthetic population were pollinated with a mixture of pollen harvested from diploid plants of the same synthetic population. Both the largest and medium-sized haploid plants were pollinated. No less than 200 to 300 haploid plants were involved in pollination. Selection was for ear size in haploids. 20 to 30 of the largest ears showing no sign of any disease were selected. The seeds harvested from selected ears were the product of the first cycle of selection (C1). As was shown before, they contain a normal diploid embryo.
The next year the bulk of the C1 seeds was sown in a space-protected nursery to produce haploids. The remaining C1 seeds were stored in a refrigerator. The next season the haploids produced from C1 and C1 seeds stored in a refrigerator were sown in the experimental field. At flowering, C1 haploids were pollinated with a mixture of pollen harvested from diploid C1 plants. Upon selection of ears, the resulting seeds were the product of the second cycle of selection (C2). Thus, a two-step scheme of recurrent selection was employed: the first step involved producing haploids, the second growing the haploids, pollinating them with a mixture of pollen from diploid plants of the same cycle of selection, and selecting haploids exhibiting traits valuable for breeding. The above procedure for improving synthetic populations has been termed haploid sib recurrent selection. A total of two cycles of the haploid sib recurrent selection was carried out for each synthetic population.
In 1998, diploids of the initial synthetic populations and of the first and second cycles of selection, C0, C1 and C2, were planted in the field. The plot area was 30 square meters. The planting density was 60,000 plants per hectare. 180 plants were planted for each cycle of selection of each synthetic population. The primary objective in 1998 was to propagate the seeds of both synthetics and their C1 and C2 cycles. Concurrently with seed multiplication, various plant and ear traits were being measured. For seed multiplication, half the plants in each treatment were used. The other half were open pollinated. Productivity and seeds per plant were only measured in plants whose ears had been open pollinated. The other traits were measured on all the plants in each treatment. The results allow some rather interesting conclusions to be drawn.
Application of recurrent selection resulted in a significant increase in productivity in both synthetic populations. The initial synthetic population SP exhibited a yield of 53.9 g of seeds per plant, with the mean yield of the second cycle of selection (SP C2) being 65.7 g/plant (Table 1). The gain per cycle of selection amounted to 11.0%. This is a fairly large gain per cycle of selection. However, the gain was even larger in the SA synthetic population. In the initial synthetic population SA, the yield was 56.0 g/plant whereas in plants from the second cycle of selection (SA C2) it was 74.6 g/plant. The gain per cycle of selection was as high as 16.7%.
Table 1. Ear trait values for the initial synthetic populations SP and
SA and two cycles of haploid sib recurrent selection.
|Ear trait||Population||Mean||Gain per cycle, %||Coefficient of variation, %|
|Yield, g/plant||SP C0||53.9±3.5||58.6|
|Ear length, cm||SP C0||12.8±0.3||23.8|
|Ear diameter, cm||SP C0||3.77±0.05||12.2|
|Seeds per plant, No.||SP C0||232.4±12.8||49.5|
|Weight of 1000 kernels, g||SP C0||228.2±5.1||19.9|
*,**,*** Differences from initial synthetic population are significant at 5%, 1% and 0.1%
Selection of haploid plants for ear size resulted in larger ears in diploid plants of synthetic populations. Thus, for example, the ear length and diameter in synthetics of the second cycle of selection were significantly larger than those in the initial synthetics. The number of seeds per plant was significantly increased (Table 1).
Interestingly, selection of haploid plants for larger ear size resulted in larger plant size of diploids (Table 2). For instance, the mean plant height was 144.8 cm in the original synthetic population SP, and 171.2 cm in plants from the second cycle of selection (SP C2). The gain per cycle of selection was 9.3%. In the SA synthetic population, the gain per cycle of selection for plant height was 12.1%. Besides plant height, some other traits showed an increase, such as ear height, leaf length and, to a somewhat lesser degree, the number of leaves per plant.
Table 2. Plant trait values for the initial synthetic populations SP
and SA and two cycles of haploid sib recurrent selection.
|Plant traits||Population||Mean||Gain per cycle, %||Coefficient of variation, %|
|Plant height, cm||SP C0||144.8±2.0||17.2|
|Ear height, cm||SP C0||42.9±0.9||25.8|
|Leaf length, cm||SP C0||62.2±0.7||14.6|
|Leaf number||SP C0||15.9±0.1||8.7|
*,**,*** Differences from initial synthetic population are significant at 5%, 1% and 0.1%
We suggest that the high efficiency of haploid sib recurrent selection was due to two factors: (1) natural selection of haploid plants; this eliminated harmful genes from the synthetic populations, and (2) artificial selection of haploids; haploid plants made it possible to select useful genes with additive and epistatic effects. Genes with additive and epistatic effects form a basis for the improvement of populations (Hallauer, Quantitative genetics in maize breeding, 1986). A combination of the above factors resulted in a large gain per cycle of selection.
An important factor in improving synthetic populations is maintaining population genetic variability during selection. The coefficient of variation can provide a general idea about the way the population genetic variability changes. In the synthetic population SA, the coefficient of variation for yield showed little change. If anything, it exhibited some tendency to increase. The genetic variation presumably did not decrease during the first two cycles of selection, thus making it possible to expect a large gain per selection cycle for subsequent cycles as well. In the synthetic population SP, the phenotypic variation for yield decreased during the second cycle of selection, C2.
We expect the gain per cycle of selection to be considerably decreased after 4 to 5 cycles of selection because of reduced genetic variability. Therefore, new germplasm is planned to be introduced into synthetic populations starting with selection cycle 4. Work is underway to select donors to be used for improving synthetic populations SA and SP. Our experience shows that a useful means in fulfilling this task can again be provided by haploid plants which are a good indicator of the presence of useful genes in the material studied.
The above suggests that maternal haploid plants can be a useful tool
for the corn breeder whose work is aimed at improving synthetic populations
to the MNL 73 On-Line Index
Return to the Maize Newsletter Index
Return to the Maize Genome Database Page