GWAS Mediated Elucidation of Heterosis for Metric Traits in Cotton (Gossypium hirsutum L.) Across Multiple Environments

Abstract

For about a century, plant breeding has widely exploited the heterosis phenomenon-often considered as hybrid vigor-to increase agricultural productivity. The ensuing F₁ hybrids can substantially outperform their progenitors due to heterozygous combinations that mitigate deleterious mutations occurring in each genome. However, only fragmented knowledge is available concerning the underlying genes and processes that foster heterosis. Although cotton is among the highly valued crops, its improvement programs that involve the exploitation of heterosis are still limited in terms of significant accomplishments to make it broadly applicable in different agro-ecological zones. Here, F₁ hybrids were derived from mating a diverse Upland Cotton germplasm with commercially valuable cultivars in the Line × Tester fashion and evaluated across multiple environments for 10 measurable traits. These traits were dissected into five different heterosis types and specific combining ability (SCA). Subsequent genome-wide predictions along-with association analyses uncovered a set of 298 highly significant key single nucleotide polymorphisms (SNPs)/Quantitative Trait Nucleotides (QTNs) and 271 heterotic Quantitative Trait Nucleotides (hQTNs) related to agronomic and fiber quality traits. The integration of a genome wide association study with RNA-sequence analysis yielded 275 candidate genes in the vicinity of key SNPs/QTNs. Fiber micronaire (MIC) and lint percentage (LP) had the maximum number of associated genes, i.e., each with 45 related to QTNs/hQTNs. A total of 54 putative candidate genes were identified in association with HETEROSIS of quoted traits. The novel players in the heterosis mechanism highlighted in this study may prove to be scientifically and biologically important for cotton biologists, and for those breeders engaged in cotton fiber and yield improvement programs.

Keywords: F1 hybrid; GWAS; hQTNs; heterosis; multiple environments; upland cotton.

FIGURE 1

Violin plots based on phenotypic variation of ten agronomic and fiber quality traits of F₁ hybrids (Yaxis) from four sets (SetA, SetC, SetD, and SetE) across multiple environments for two years; 2012 and 2013 (Xaxis). Legends on top right in different colors are representing ten evaluated phenotypic traits.

FIGURE 2

Distribution of SCA and five heterosis types (HB, MP, HI, K3, and K4) of agronomic and fiber quality traits among four F₁ hybrid sets (SetA, SetC, SetD, and SetE) across multiple environments for years 2012 and 2013. Legends on the top right in different colors are depicting ten evaluated phenotypic traits.

FIGURE 3

Population structure of 284 female parents in the association panel (A) Principal component analysis (PCA) of female lines (B) Population structure of 284 female parents (K = 3).

FIGURE 4

Single nucleotide polymorphism (SNP) distributions on 26 chromosomes of (A) parents, (B) F₁_A, (C) F₁_C, (D) F₁_D, and (E) F₁_E. A_t1∼A_t13 and D_t1∼D_t13 in vertical axis are the serial number of 26 chromosomes; the horizontal axis shows chromosome length (Mb); = 0 ∼>702 depicts SNP density (the number of SNPs per window).

FIGURE 5

Phenogram displaying the 2847 significant (–log (p) ≥ 4) associations among phenotypic traits and 1348 significant SNPs residing on 26 chromosomes of upland cotton.

FIGURE 6

Summary of significant association signals and significant SNPs. (A) Representation of significant associations among 10 phenotypic traits, four F₁ sets, five heterosis types, SCA and significant SNPs (B) details of significant SNPs commonly associated across four different sets of F₁ hybrids (C) number of significant SNPs/hQTNs associated with heterosis types, SCA and F₁ sets.

FIGURE 7

Depicted here are results from the multivariate analysis of pleiotropy. For each associated SNP, the method returns the best-fitting solution of which phenotypes were associated with that SNP. All SNPs with one or more associated phenotypes are shown here. For example, every SNP associated with FE was found to be pleiotropic for other phenotypes. The total number of pleiotropic as well as unique associated SNPs for each trait from these analyses were 181 (FE), 176 (BW), 147 (LP), 146 (MIC), 141 (FUI), 134 (FS), 113 (FL), 111 (BN), 107 (PH), and 92 (FU).

FIGURE 8

Detailed view of significant SNPs related to five types of heterosis, F₁, SCA and studied traits on 26 chromosomes with their physical positions (bp).

FIGURE 9

Heat map for expression patterns of the 275 genes nearby significant keys SNPs/QTNs associated with studied agronomic and fiber quality traits. Shaded portion is representing expression >1 while white portion is representing <1.

FIGURE 10

(A) Summary of GWAS results for Fiber micronaire (MIC) including Manhattan plots, QQ plots, violin plots displaying differences for MIC among two haplotypes of SNP/hQTN D09_43629201 in trait phenotype and five heterosis types. (B) Regional Manhattan plots showing presence of hQTN D09_43629201 in trait phenotype and five types of heterosis further narrowing down to genomic location of gene hqMICD09_43629201_C on chromosome D09, sun plot displaying the p-values of variables harboring hqMICD09_43629201_C and gene Gh_D09G1604, Expression levels of representative gene associated with MIC during different growth stages and Haplotype region (55 kb) surrounding the peak on chromosome D09 associated with MIC.