Soybean Cultivars Resulted from More Recombination Events Than Unselected Lines in the Same Population

The selection of superior adapted cultivars has contributed to the doubling of soybean [Glycine max (L.) Merr.] yields in the USA since 1930. Genetic variation was required for this selection to be effective. The objective of this study was to evaluate the importance of homologous chromosome meiotic recombination in the creation of soybean cultivars. A set of 10 adapted high-yielding cultivars selected from the cross ‘Williams’ × ‘Essex’ was compared with a set of 156 random recombinant inbred lines (RILs) from the same population. Crossover events were identified using 143 simple sequence repeat (SSR) markers spanning all 20 soybean chromosomes. The recombination rates were standardized among chromosomes by dividing the realized crossovers by the potential crossovers. The standardized recombination rate for the entire genome was significantly greater for the 10 cultivars (0.34) than for the RILs (0.29). The cultivars had numerically higher standardized recombination rates for 17 of the 20 chromosomes, significantly higher on chromosomes defined by the molecular linkage groups C2, L, and M. The interaction of linkage groups S T E F A N I A K E T A L . , C R O P S C I E N C E 4 6 (2 0 0 6 ) 2 with the two sets of lines was nonsignificant for standardized recombination. Soybean breeding progress has been accomplished in part by creating and capitalizing on new within-chromosome allele combinations. Abbreviations: MAS, marker assisted selection; QTL, quantitative trait loci; RIL, random inbred lines; SSR, simple sequence repeat. The improvement of any species through breeding requires the creation and selection of a novel combination of alleles in progeny from the genetic variation contained within the parents. These new combinations could arise from either independent chromosome assortment or homologous chromosome recombination. Breeding progress can be made only through variation from independent assortment if the superior alleles are found on different chromosomes (unlinked). Variation in recombination frequency exists in plant breeding populations (Pfeiffer and Vogt, 1990; Tulsieram et al., 1992; Fatmi et al., 1993). Similarly, inbred progeny receive a range of alleles from one or the other of the population’s parents (Kiem et al., 1991). The inheritance of alleles in adapted soybean cultivars does not necessarily follow the ratios estimated by the coefficient of parentage for a particular cross (Kisha and Diers, 1997) because the coefficient of parentage is a probability that necessarily ignores the preferential selection by plant breeders of alleles that favorably affect a phenotype. Because the site of crossing-over is random, selection for polygenic traits will alter the number and position of crossovers found in the lines a breeder chooses to advance as opposed to those found in the entire population. Therefore, for quantitative traits, meiotic recombination has been, and will continue to be, a mechanism on which breeders must capitalize to establish novel superior linkage blocks in these regions. Demarly (1979) introduced the term “linkat” for a collection of favorable alleles that are linked and tend to be inherited intact because of the competitive advantage they give the individuals that possess them. These linkats contribute to the preferential inheritance of sets of favorable accumulated alleles that cause the actual pattern of inheritance to deviate from that predicted by the coefficient of parentage (Kisha and Diers, 1997). Using restriction fragment length polymorphism (RFLP) markers, Lorenzen et al. (1996) identified two linkage groups in four soybean cultivars that have long chromosome sections inherited intact from the same parents despite the fact that these cultivars were developed in separate breeding programs. If a collection of genes fundamental to domestication were established in linkats long ago, it would be likely that this collection would be found intact in many modern cultivars despite a lack of a common pedigree. Similarly, new favorable linkage blocks may be created by selection following hybridization. For example, Lorenzen et al. (1996) identified five cultivars that were the result of a crossover in the same location with the same parental alleles selected on either side of the crossover. Seed yield in soybean is a polygenic trait, which could be greatly affected by recombination. Whether these genes interact additively or epistatically is of great importance with respect to the efficacy of utilizing enhanced recombination in a breeding strategy. If yield genes act additively, then enhanced recombination that brings these alleles together will S T E F A N I A K E T A L . , C R O P S C I E N C E 4 6 (2 0 0 6 ) 3 be beneficial. If these genes interact epistatically, then recombination’s value is uncertain until the coupling-repulsion status of the alleles is clarified (Hanson and Hayman, 1963). Cregan et al. (1999) have combined data from mapping populations into a highly saturated linkage map of the soybean genome containing classical, SSR, and RFLP markers. This map may be used to screen parents for polymorphisms useful in evaluating crossingover in the progeny of these parents. One can then track whether linkage blocks were inherited intact or broken in the progeny of a cross. The high degree of map saturation makes it likely that some markers are linked to yield quantitative trait loci (QTL). The objectives of this study were to (i) evaluate the importance of homologous chromosome meiotic recombination in the creation of soybean cultivars by comparing standardized recombination rates between a random population and a set of adapted cultivars derived from the same cross and (ii) infer the relationship between genomic regions with high or low crossover rates and the location of previously identified QTL in the soybean genome. Materials and Methods Genetic Material The genotypes investigated in this study fell into one of two selection sets. All genotypes were from the cross ‘Williams’ (Bernard and Lindahl, 1972) × ‘Essex’ (Smith and Camper, 1973). Ten released adapted cultivars comprised the high yield selection (cultivar) set. These genotypes, ‘Pennyrile,’ ‘S4240,’ ‘RA452,’ ‘A4268,’ ‘RA481,’ ‘A3860,’ ‘A3127,’ ‘9441,’ ‘9471,’ and ‘Coker 393,’ came from various breeding programs (Gabe, 1994) and are assumed to be the product of transgressive segregation. These cultivars were probably selected as F4 or F5 derived lines, depending on the breeding strategy utilized by the particular breeding program from which the cultivars were developed. Selection at this level of inbreeding was typical in U.S. soybean breeding programs during this time frame. Seeds of cultivars were obtained from the soybean germplasm collection or from the breeding program that developed the cultivar. An unselected set of 156 lines from this same cross was compared with the cultivar set. The random lines were created by single seed descent and were advanced to the F6:8 generation (Hyten, 2002). These lines will be subsequently referred to as the random RIL set. The RIL population was previously used in a QTL study that involved agronomic trait testing in five environments (Hyten et al., 2004). Three other similarly sized populations of unselected RILs were utilized from the crosses ‘Peking’ × ‘Hamilton’ (Nickell et al., 1990), ‘Pershing’ (Anand and Shannon, 1985) × ‘Hamilton,’ and ‘Peking’ × ‘Essex’ to validate our analysis of the SSR markers on linkage group C2. SSR Analysis About 5 g of leaf material was collected from plants from each random line and each selected cultivar. This leaf material was stored frozen until desiccated in a lyophilizer. The DNA from the 10 cultivars was extracted following the CTAB procedure of Kiem et al. (1988). The DNA from the 156 RILs was extracted with the Qiagen DNAeasy mini prep kit (Qiagen, Valencia, CA). The DNA was quantified with a fluorometer and diluted to 10 S T E F A N I A K E T A L . , C R O P S C I E N C E 4 6 (2 0 0 6 ) 4 ng/mL. A total of 10 mL of this DNA suspension was run on a 1% (w/v) agarose gel to check for quality and to verify concentration. Williams and Essex were screened with 568 SSR markers to identify polymorphisms. The SSR primer sequences were obtained from the Soybase (1995) web site. From this screening, 277 markers were found to be polymorphic between Williams and Essex, and 143 of the 277 were used to obtain crossover data (fig. 1). These markers covered all 20 soybean linkage groups (Cregan et al., 1999), with a minimum of four and a mean of 7.15 markers per linkage group. Figure 1. Schematic representation of the relative positions of the 143 SSR markers on soybean’s 20 molecular linkage groups (Cregan et al. 1999) used to identify regions of crossing-over in the progeny of Williams × Essex. S T E F A N I A K E T A L . , C R O P S C I E N C E 4 6 (2 0 0 6 ) 5 The polymerase chain reaction (PCR) was conducted as defined on Soybase (1995). Amplified PCR fragments were separated by either metaphor agarose or polyacrylamide gel electrophoresis, depending on the size of the polymorphism between Williams and Essex. Polymorphisms greater than 10 base pairs were run for 4 h at constant 70 V on 3% (w/v) metaphor agarose gels stained with ethidium bromide. Polymorphisms that were less than 10 base pairs were run on 6% (w/v) nondenaturing polyacrylamide gels at 200 V for 4 h and stained with ethidium bromide. Genetic Map Marker order was initially defined on the basis of the public soybean composite map (Cregan et al., 1999). Because some of the markers used in this study are placed on the composite map with distances less than 5 cM between them, the RIL population was mapped by the Mapmaker program (Lander et al., 1987; Lincoln et al., 1992) to verify the map order. This was accomplished by first defining 20 linkage groups and then using the “assign” command to place markers onto soybean’s 20 chromosomes. In this analysis, all markers were assigned to their defined linkage group except those found on linkage group A2, which Mapmaker designated as unlinked. Next, the order of the markers on the individual linkage groups was determined using the “compare” command. The four markers unassigned by the “assign” command were assumed to be on linkage group A2 and analyzed as such by Mapmaker and agreed with the published map for 14 marker order calculated linkage groups. The remaining six linkage groups (A1, A2, B2, C1, C2, and D1a) were ordered differently compared with the published map. On each linkage group, the order of two markers in close proximity (< 5.0 cM) of each other was inverted. For each of these linkage groups one of the two markers was dropped from the data set. Recombination was detected by first listing the markers for each chromosome in the order found on the genetic map. Next, crossing over was counted by following each line’s marker scores along the chromosome and noting where a line’s score changes from one parent’s allele to the other. All crossovers for each line and linkage group were then simply summed and standardized by putting this sum in the numerator of a fraction with the potential crossovers (number of markers per chromosome – 1) in the denominator. The analyzed variable was standardized crossovers. Crossovers were standardized to permit analysis among linkage groups in which different numbers of markers were available and to adjust individual lines in which the allele designation at a locus may have been unassigned. Data Analysis The expected 1:1 segregation ratio of the inheritance of parental alleles in the RILs was tested by the chi-square test. The Yates correction factor for a chi-square test with one degree of freedom was not used because the expected number in each class (n = 78) was relatively large (Bailey, 1961). Because of the large number of loci analyzed, deviations from this ratio at a single locus were considered significant at p < 0.01. Deviations from expected on an entire chromosome were considered significant at p < 0.05. Segregation ratios were not analyzed for the cultivars. The number of cultivars was small for a chi-square test, and S T E F A N I A K E T A L . , C R O P S C I E N C E 4 6 (2 0 0 6 ) 6 selection was expected to favor positive alleles negating the 1:1 segregation ratio expectation. The data were analyzed for all 20 linkage groups together (sources of variation: selection sets, linkage groups, selection sets × linkage groups, all factors fixed) as well as all 20 linkage groups individually (source of variation: selection sets) by PROC GLM of SAS (SAS Institute, Cary, NC, release 8.1). Standardized crossovers were compared between the cultivars and all the 156 RILS. Ten sets of 10 RILs, drawn from the larger set of 156 RILs, were also created to compare standardized crossovers in equal sample sizes between the cultivars and the random lines using the analysis above. The locations of regions where comparatively few or many crossovers were detected in the RILs compared with the cultivars were related to QTL data from previous studies. These comparisons on individual linkage groups, in specific regions, did not include replication and hence were not analyzed statistically.