Genomic selection for tar content in Nicotiana tabacum: genetic architecture analysis and model evaluation

TL;DR

This study explores the genetic factors influencing tar content in tobacco and identifies the best model for predicting low-tar varieties.

Contribution

The study introduces rrBLUP as a highly accurate and efficient model for predicting tar content in tobacco.

Findings

Tobacco tar content is influenced by many small-effect genetic loci, especially on chromosome 17.

The rrBLUP model achieved the highest prediction accuracy (0.84) for tar content in tobacco.

rrBLUP's performance was confirmed in an independent test group with a strong correlation (Pearson r = 0.888).

Abstract

Despite its economic importance, reducing tobacco tar content remains challenging due to its complex genetic basis.

Here, we evaluated 436 diverse tobacco accessions to characterize the genetic architecture of tar content and develop an optimized genomic selection strategy. Based on these findings, sixteen genomic prediction models were assessed using five-fold cross-validation.

Genome-wide association analysis detected no major-effect loci, and regional heritability mapping revealed localized enrichment of small-effect variants, particularly on chromosome 17, indicating a predominantly polygenic architecture. rrBLUP achieved the highest prediction accuracy (0.84) with superior computational efficiency, followed closely by GBM (0.83). The robustness of rrBLUP was further confirmed in an independent panel of 36 accessions (Pearson r = 0.888).

Together, our results demonstrate that tobacco tar content is governed by dispersed small-effect loci with regional aggregation and establish rrBLUP as a robust and practical model for genome-wide prediction, providing methodological guidance for low-tar tobacco breeding.

Figures