Simulation Experiments on Efficiencies of Gene Introgression by Backcrossing

ferent alternatives for defining a strategy. The screened population size has been reported in simulation studies Designing a highly efficient backcross (BC) marker-assisted selecas the most important factor affecting the efficiency of tion (MAS) experiment is not a straightforward exercise, efficiency being defined here as the ratio between the resources that need to MAS (Zhang and Smith, 1992; Gimelfarb and Lande, be invested at each generation and the number of generations required 1994; Whittaker et al., 1995; Hospital et al., 1997; Frisch to achieve the selection. This paper presents results of simulations et al., 1999a). Independent of the strategies considered conducted for different strategies, using the maize genome as a model, in those papers (selection index and BC-MAS), selecto compare allelic introgression with DNA markers through BCs. tion for individuals with a desirable genotype at the Simulation results indicate that the selection response in the BC1 predetermined markers usually requires a relatively could be increased significantly when the selectable population size large population. Theoretically, if one considers an infi(Nsl ) is 50, and that a diminished return is observed when this nite population size, any BC-MAS experiment can be number 100. Selectable population size is defined as the number of achieved in three generations (one generation to cross individuals with favorable alleles at the target loci from which selection the two parental lines and produce the F1 seeds, one with markers can be carried out on the rest of the genome at nontarget loci, simulations considered the allelic introgression at one to five BC, and one self-pollination generation). With recent target loci, with different population sizes, changes in the recombinaadvances in polymerase chain reaction (PCR) based tion frequency between target loci and flanking markers, and different markers, for example, simple sequence repeats (Chin et numbers of genotypes selected at each generation. For an introgresal., 1996) and single nucleotide polymorphisms (SNPs) sion at one target locus in a partial line conversion, and using MAS (Gilles et al., 1999), a substantial improvement in the at nontarget loci only at one generation, a selection at BC3 would be capacity to efficiently screen large populations has been more efficient than a selection at BC1 or BC2, due to the increase achieved. Today, the screening of thousands of genoover generations of the ratio of the standard deviation to the mean types for a few target genes no longer poses an intractaof the donor genome contribution. With selection only for the presence ble technical problem, and can be considered in MAS of a donor allele at one locus in BC1 and BC2, and MAS at BC3, lines strategies (Ribaut and Betrán, 1999). with 5% of the donor genome can be obtained with a Nsl of 10 in BC1 and BC2, and 100 in BC3. These results are critical in the application of During the past several years, simulations to evaluate molecular markers to introgress elite alleles as part of plant improvethe efficiency of MAS as a breeding tool have been ment programs. reported by various groups. These simulations have been quite diverse; for example, MAS has been tested combining phenotypic and genotypic data in a selection I a BC scheme, the strategy is to transfer a specific index (Lande and Thompson, 1990; Knapp, 1994; Xie elite allele at a target locus from a donor line to a and Xu, 1998a), considering different breeding generarecipient line. The use of DNA markers, which permit tions (Edwards and Page, 1994), and for different breedthe genetic dissection of the progeny at each generation, ing schemes (Xie and Xu, 1998b). Efficiency of MAS increases the speed of the selection process (Tanksley has also been evaluated considering the heritability of et al., 1989). Although it is easy to plot a BC-MAS the target trait (Hospital et al., 1997; Knapp, 1998), the strategy, the design of the most appropriate and efficient genetic effect at target loci (Van Berloo and Stam, 1998), strategy is generally not a straightforward task, given the and by monitoring target genomic regions simultanenumber of parameters involved. Before any experiment, ously vs. one by one (Hospital and Charcosset, 1997). the number of target genes involved in the selection Most of the theoretical papers related to MAS present and the expected level of line conversion must be decomplex mathematical models, making it difficult to fined. Then, one must identify at each generation the directly derive a practical MAS experiment. In addition, size of the population to be screened, the number, posithe implications of using different laboratory strategies tion, and nature of molecular markers used, and the that can be considered to achieve the selection, such as number of genotypes selected. The expected level of different DNA markers, are rarely taken into account conversion is closely related to the number and distribuin those theoretical papers when comparing the effition of the DNA markers at nontarget loci and the ciency of different approaches. recombination frequencies between the target gene and The objective of this paper is not to present new flanking markers. All these parameters influence the genetic models, but rather to provide some guidelines number of the generations required to achieve a specific at both the theoretical and practical levels for identifying and successful BC-MAS experiment, while offering difthe most appropriate BC-MAS strategy based on the objectives of different types of applied breeding experiJ.-M. Ribaut and D. Hoisington, CIMMYT, Int. Maize and Wheat ments. To achieve this objective, the relationship beImprovement Center, Lisboa 27, Apdo. Postal 6-641, 06600 Mexico D.F., Mexico; and Changjian Jiang, Monsanto Life Sciences Research Abbreviations: BC, backcross; cM, centimorgan; MAS, markerCenter, 700 Chesterfield Parkway North, St. Louis, MO 63198. Reassisted selection; PCR, polymerase chain reaction; N, screened popuceived 8 Dec. 2000. *Corresponding author ([email protected]). lation size; Ni, number of individuals; Nsl, selectable population size; SNP, single nucleotide polymorphism. Published in Crop Sci. 42:557–565 (2002).