Resolving population structure and genetic differentiation associated with RAD-SNP loci under selection in jute (Corchorus olitorius L.)

We performed individual-based restriction site-associated DNA (RAD) sequencing to discover single-nucleotide polymorphisms (SNPs) across a diverse set of 225 fibre-type lines in jute (<i>Corchorus olitorius </i>L., Malvaceae s. l.) <i></i>and identified a set of 1115 polymorphic RAD-SNP markers (SNP containing RADseq loci), each supporting a single SNP with &gt;0.05 minor allele frequency (MAF). Based on multilocus RAD-SNP genotypes of 221 lines (four lines were excluded due to &gt;50 % missing genotypes) with a call rate of &gt;0.95, we examined the geographic patterns of genetic diversity across nine predefined populations, viz. AFR1 (Kenya and Sudan), AFR2 (Tanzania), CI (central India), EI (east India), NI (north India), SI (south India), NPPK (Nepal and Pakistan), ESEA (China, Myanmar, Indonesia and Thailand) and RoW (Australia, Brazil, Germany and Russia) and determined their genetic relatedness by assessing the <i>F</i>_ST, AMOVA, <i>N</i>_mand PCA at the population level. Using five assignment tests with different statistical bases (<i>k</i>-means clustering, STRUCTURE, sNMF, DAPC and frequency-based assignment test), the most exhaustive yet to our knowledge, we inferred how these geographic populations are structured. We further applied subpopulation-based <i>F</i>_ST- (LOSITAN and BayeScan) and <i>G</i>_ST-outlier (HacDivSel) tests and an individual-based global approach (PCAdapt) based on principal component analysis (PCA) to detect putative RAD-SNP loci under selection. Instead of BLAST alone, we employed a serial approach based on BLAST, Blast2GO mapping, protein domain annotation (DoMosaics) and REViGO semantic analysis to identify candidate genes and retrieve the overrepresented gene ontology (GO) terms associated with the outlier RAD-SNP loci putatively involved in local adaptation. <br>