Fig. 1.
Fig. 1.

STRUCTURE and principal component analysis. (a) STRUCTURE analysis of the Middle American diversity panel (MDP) from K = 2 to K = 7. Line K = 2 shows the split between the two major races, and from K = 3 to K = 7, each race subdivides into market classes. (b) Three-dimensional principal component (PC) analysis of the MDP. The first dimension separates the two major races. The color coding of bean market classes is as follow: great northern (purple); black (blue); navy (yellow); pink (green); pinto (light blue); small red (orange); small white (white); and black mottled, carioca, flor de mayo, and tan (black).

 


Fig. 2.Fig. 2.
Fig. 2.

Linkage disequilibrium heatmaps of 11 chromosomes in (a) race DJ and (b) race MA after controlling for population relatedness. The green lines denote the boundaries of the pericentromeric region. A subset of single-nucleotide polymorphisms with minor allele frequency ≥ 5% were used.

 


Fig. 3.
Fig. 3.

Genome-wide linkage disequilibrium heatmap in races (a) Durango–Jalisco and (b) Mesoamerican. Data above the diagonal represent the null model, and data below the diagonal image represents the model that accounts for individual relatedness. A subset of single-nucleotide polymorphisms spaced every 50 kb was used, and only pairwise r2 > 0.6 are shown. The gray rectangular areas show the pericentromeric regions. Gray dashed lines define the chromosome boundaries.

 


Fig. 4.
Fig. 4.

Density plot of phenotypic distribution of six agronomic traits. Color coding is provided in the label section. CO, Colorado; MI, Michigan; NE, Nebraska; ND, North Dakota; Combined, across all the locations; DJ, Durango/Jalisco; MA, Mesoamerican.

 


Fig. 5.
Fig. 5.

Correlation between traits and locations based on adjusted means. DF, days to flower; DM, days to maturity; GH, growth habit; LG, lodging; CH, canopy height; SW, seed weight; CO, Colorado; MI, Michigan; ND, North Dakota; NE, Nebraska. Cells with correlation values not significant at P-value < 0.01 are left blank.

 


Fig. 6.
Fig. 6.

Manhattan plots of the best models for flowering time in the Middle American diversity panel (MDP) and Mesoamerican (MA), Durango-Jalisco (DJ) races across all the locations. The two major peaks on Pv01 (8 and 15 Mb) vary between races. The best model is indicated in parenthesis. The green lines are the cutoff values to call a significant peak. Single-nucleotide polymorphisms that pass the 0.01 percentile are highlighted in red, and those falling between 0.01 and 0.1 percentile are highlighted in blue. The quantile–quantile plot shows the goodness of fit of the best model, and the histogram shows the trait distribution for adjusted means across all the locations in that population.

 


Fig. 7.
Fig. 7.

Manhattan plots of the best models for plant architecture related traits (LG, lodging; CH, canopy height; GH, growth habit). The peak on Pv07 (46 Mb) is in common among all three traits when the determinate genotypes are excluded from the population for growth habit analysis. Growth habit encompasses major peaks from both lodging and canopy height. Growth habit using the entire Middle American diversity panel (MDP) shows a strong signal at the end of Pv01, which captures the determinacy characteristic of genotypes with type I growth habit and masks the peak on Pv07. This peak is shared with a flowering time peak in Michigan and is close to flowering time candidate genes such as fin locus (TFL-1). The best model is indicated in parenthesis. The green lines are the cutoff values to call a peak significant. Single-nucleotide polymorphisms that pass the 0.01 percentile are highlighted in red and those falling between 0.01 and 0.1 percentile are highlighted in blue. The quantile–quantile plot shows the goodness of fit of the best model, and the histogram shows the trait distribution for adjusted means across all locations in that population.

 


Fig. 8.
Fig. 8.

Manhattan plot and linkage disequilibrium (LD) heatmap over 3.7-Mb region (45.03 – 48.37 Mb) on Pv07 for lodging. This region includes the three significant peaks (Pv07 [45.18 Mb], Pv07 [46.11 Mb], and Pv07 [48.63 Mb]) and their associated candidate gene homologs in Arabidopsis. The marker at 46.11 Mb, which resides in Phvul.007G221700, is colored in blue in the Manhattan plot and the rest of the markers in this region are color coded based on their pairwise r2 value with this marker. Single-nucleotide polymorphisms (SNPs) with r2 < 0.25 are white colored, those with 0.25 ≤ r2 0.5 are orange, and SNPs with r2 > 0.5 are colored in red. The r2 values in Manhattan plot and LD heatmap are corrected for population structure and relatedness based on the best model using all the SNPs with minor allele frequency ≥ 5%.

 


Fig. 9.
Fig. 9.

A simplified schematic illustration of the Arabidopsis flowering time pathways. Only pathways and genes related to our candidate genes are shown. The common bean orthologs are denoted with a star.