Genome‐wide locus–allele comparison reveals differential evolution dynamics from annual wild to landrace and released cultivar soybeans

Previous studies on population evolution relied primarily on allele frequency analysis using molecular markers or genome sequence segments, like selective sweeps. With the sequencing technique developed, we suggest the genome-wide locus-allele comparison to detect the genomic structure variation among populations. Its key point lies in taking SNP linkage disequilibrium block as uniform genomic marker for genome-wide gene and inter-gene regions to meet the requirement of multiple alleles in natural populations. A sample composed of 750 annual wild accessions (WAs), landraces (LRs), and released cultivars (RCs) of soybean from southern, northern, and northeastern China eco-regions (SC, NC, and NEC, respectively) were analyzed for their evolution dynamics involving four evolutionary processes (WA→LR→RC, WA_SC→WA_NC→WA_NEC, LR_SC→LR_NC→LR_NEC, and LR_SC→RC_SC/LR_NC→RC_NC/LR_NEC→RC_NEC). Our major finding was the discovery of allele and locus zero/one variation between/among ancestor-filial populations involving a large part of the whole population alleles and loci, 25.10% and 18.62% in domestication and modern breeding stages, respectively, which was not detected by selective sweeps. The essence of population evolution is the allele zero/one changes based on ordinary allele frequency changes, which causes the locus zero/one changes. The allele/locus zero/one variation happened more often when their frequency was at 0.0-0.3 and 0.8-0.99 in the previous stage generation, respectively. The WA and LR geographic evolution are different processes due to different combination of allele/locus zero/one changes by natural versus artificial selection pressures. Compared to per-year allele exclusion, the rate of per-year allele emergence is relatively stable in domestication and modern breeding (2.75E-5 vs. 1.34E-5 and 1.42E-3 vs. 1.10E-5), respectively.