Frequency distribution of the phenotypic traits
FRH and NDVI were the first data collected in the first year of planting. Both populations had a similar range of distribution for FRH and NDVI (Figure 1). B6 had higher FRH and NDVI than AP13 and VS16. For FRH and NDVI collected in fall 2018, both population distributions were skewed to the right, indicating lower population FRH and NDVI values. The reason for the shift in values was because we clipped the plants on 31st August in 2017 and on September 21st in 2018. Thus, most of the plants were already dormant or progressing toward dormancy in fall 2018, and hence the lower FRH and NDVI values. There was no B6 value in AB plot for 2018 because the plant was growing poorly in spring 2018 and eventually died in fall 2018. In the BV plot in the fall 2018 B6 value was similar to VS16. B6 had earlier emergence than AP13 but later emergence than VS16 in spring 2018. In spring 2019, B6 had earlier emergence than VS16. We think the reason for B6 year to year variation is most likely due to its intolerance to cold temperature. B6 grows throughout the winter in the greenhouse where the temperature is warmer while AP13 and VS16 go dormant (Figure 2a). Both populations had earlier emergence in spring 2019 than 2018, indicating that the environment in early spring 2019 was more conducible for growth. For the flowering date, both populations have about the same range of distribution in 2018 and 2019. B6 flowered later than AP13 and VS16, which suggests that B6 completed its growth later than the other two parents.
Analysis of variance
There was a significant genotype effect on all traits in AB, BV planting date 1, and BV planting 2, and a significant year effect for FRH, NDVI, and SE in AB; FRH, NDVI, SE, and FD for BV (Table 1). Significant genotype by year effect was observed for NDVI in AB, FRH, NDVI, SE, and FD for BV planting date 1. Because of this significant interaction, QTL mapping was done using LS means for each year and BLUP for trait values across years.
Broad-sense heritability (H2)
There were medium ranges of H2 for FRH (0.54 – 0.64) and NDVI (0.30 – 0.61), a small to medium range of H2 for SE (0.13 – 0.56), and a high range of H2 for FD (0.61 – 0.88) across AB, BV planting date 1, and BV planting date 2 (Table 2). In most cases, the highest H2 was observed for FD, followed by FRH, NDVI, and SE.
Correlation between traits
A similar trend of correlation was observed for AB and BV populations (Table 3). Biomass weight was positively correlated with FRH, NDVI, and FD, while negatively correlated with SE, which means higher biomass weight is correlated with lower dormancy level, earlier spring emergence, and later flowering. FD had the lowest correlation with other traits; it was not correlated with biomass in BV, and with SE in AB and BV.
Using 2,772 and 3,766 SNP markers for AB and BV population, respectively, for QTL mapping, we identified 16, 14, 12, and 20 QTLs for FRH, NDVI, SE, and FD, respectively, mapped in both populations and years using LS means (Supplementary table 1 - 4). For FRH QTLs, 3 were mapped in AB and 13 in BV. For NDVI QTLs, 4 were mapped in AB and 10 from BV. For SE QTLs, 7 were mapped in AB and 5 in BV. For FD QTLs, 11 were mapped in AB and 9 in BV. A higher total number of QTLs were mapped in the BV population; these could be due to the separate mapping done for two subsets of BV population. Another reason could be due to a higher genetic variance between B6 and VS16, leading to more QTLs contributing to phenotypic expression in F1 progenies. To have a meaningful comparison, QTL numbers are compared using BLUP QTLs. We discovered 9, 6, 11, and 14 significant BLUP QTLs for FRH, NDVI, SE, and FD, respectively (Supplementary table 1 – 4). A higher number of BLUP QTLs were observed in the BV population for FRH (6 QTLs) and NDVI (4), while a higher BLUP QTLs were found for SE in AB population (6 QTLs), and an equal amount of QTLs found in both populations (7 QTLs each).
Some of the QTLs mapped with BLUP values overlapped with those mapped using LS means. We identified 7, 2, 2, and 4 BLUP QTLs for FRH, NDVI, SE, and FD, that are redundant to LS means QTLs (Supplementary table 5). Adding all QTLs mapped with LS means and unique BLUP QTLs, we have a total of 18 QTLs for FRH, 18 QTLs for NDVI, 21 QTLs for SE, and 30 QTLs for FD (Figure 3 - 6). For FRH, the range of percentage of variance explained (PVE) by each QTL is 4.21 – 23.27%, for NDVI this is 4.47 – 24.06%, for SE the range is 4.35 – 32.77%, and for FD the range is 4.61 – 29.74%. FRH QTLs were mapped in LG 5N and 4K in the AB population; LG 1N, 5K, 5N, 6K, 9K, and 9N in the BV population. NDVI QTLs were found in LG 2K, 3K, 5N, and 6N in the AB population; LG 1N, 5N, 9K, and 9N in the BV population. SE QTLs were mapped in LG 1K, 1N, 2K, 5N, 7K, 9K, 9N in the AB population; LG 1N, 2N, 5K, 5N, and 9K in the BV population. FD QTLs were found in LG 3K, 4K, 5N, and 9K in the AB population; LG 1K, 1N, 2N, 3N, 5N, 6N, 7K, and 9K in the BV population.
Combining all QTLs mapped using LS means and BLUP (redundant BLUP QTLs were not included), we observed a total of 16 QTL colocalization regions within all parental maps in both populations (Figure 3 - 6, Supplementary table 6). There are three chromosomal regions with colocalization of 3 QTLs. The first region is in LG 5N of B6.AB map; the three QTLs found here are mapped for SE2019 (marker AB6919), NDVI2018 (marker AB6919), and FD2018 (marker AB8070). The NDVI and FD QTLs in this region have positive additive effects (α) while SE QTL has a negative α. This is a possible indication that the transmission of this chromosomal region can reduce the dormancy level by having a higher magnitude of plant greenness in the fall, earlier dormancy break as indicated by early spring emergence, and later flowering/maturity.
The second region with 3 colocalized QTLs is in LG 5N of the B6.BV map. The QTLs found in this region were mapped for FRH2018 (marker BV17469), NDVI2018 (marker BV17469), and SE2019 (marker BV17309). The FRH and NDVI QTLs have negative α while the SE QTL has a positive α. Since this region is associated with a higher dormancy level (positive α for FRH and NDVI) and later dormancy break (negative α), we can potentially use the opposite marker genotype to screen the plants with the opposite trait direction, i.e. lower dormancy and early emergence. A similar condition is observed in the third colocalization region with QTLs mapped for FRH2018 (marker BV7842), SE2019 (marker BV7842), and FD2018 (marker BV7842); the FRH and FD QTLs have negative α while the SE QTL has a positive α.
FRH and NDVI are two traits that shared the most number of colocalized QTLs regions with 8 regions in total. The second highest is NDVI and SE with 4 regions. NDVI has the most number of overlapped QTL regions with other traits with 14 regions in total, while FD had the least number of colocalized regions with other traits with 5 colocalized regions. In contrast, FD has the most number of the same QTL mapped across years with 3 QTLs in total.