Xuanjun Fang¹ , Weiren Wu²

1 Hainan Provincial Key Laboratory of Crop Molecular Breeding, Hainan Institute of Tropical Agricultural Resources (HITAR), Sanya, 572025, Hainan, China
2 College of Agriculture, Fujian Agriculture and Forestry University, Fuzhou, 350002, Fujian, China
Author    Correspondence author
Molecular Plant Breeding, 2026, Vol. 17, No. 1   
Received: 15 Apr., 2026    Accepted: 25 Apr., 2026    Published: 30 Apr., 2026

This article was first published in Fenzi Zhiwu Yuzhong (Molecular Plant Breeding) in Chinese (24(9): 2817-2829) in Chinese, and here was authorized to translate and publish the paper in English under the terms of Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

The genetic dissection of complex quantitative traits has long been constrained by polygenic architecture, small effect sizes, gene–environment interactions, and limited statistical power. In response to these challenges, statistical genetics has undergone a paradigm evolution from linkage analysis and candidate gene strategies to genome-wide association studies (GWAS). From a statistical genetic perspective, this study systematically reviews this methodological transition, with a focus on the changes in underlying assumptions, data structures, and statistical models, as well as their internal logical connections. Starting from classical assumptions such as Mendelian segregation and relatively stable recombination rates, we examine how factors including segregation distortion, genetic background heterogeneity, and genotyping errors may affect recombination rate estimation and likelihood distributions, thereby leading to systematic biases in linkage statistics and LOD curves. We further compare likelihood-based LOD statistics in linkage analysis with p-value–based significance measures derived from test statistics in GWAS, highlighting their differences in statistical foundations and significance assessment. It is emphasized that genome-wide significance thresholds generally require empirical calibration, such as permutation testing. With the development of high-density SNP data, large population samples, and mixed linear models, GWAS enables higher-resolution mapping by weakening locus-specific prior assumptions and explicitly modeling population structure and relatedness. Through a systematic comparison of different approaches and their respective limitations, this study argues that GWAS does not simply replace traditional methods, but represents a paradigm shift and extension in terms of statistical assumptions, model structures, and analytical scales. In the context of plant molecular breeding, the integration of GWAS with approaches such as eQTL analysis and genomic selection is expected to enhance the robustness of genetic inference and provide stronger statistical support for breeding decisions.

Statistical genetics; Paradigm evolution; Linkage analysis; Candidate gene strategies; Genome-wide association studies (GWAS); Mixed linear models