*2.5. QTL Identified from 11 Traits*

Using the best linear unbiased prediction (BLUP) values of phenotyping data collected from six to eight year/location environments with both generalized linear model (GLM) and mixed linear model (MLM), we identified a total of 33 QTL for 11 traits, one for YLD, eight for OIL, five for PLH, four for PAL, three each for IOD, LIO, and LIN, two each for DTM and STE, and one each for protein content (PRO) and OLE (Table 3, Figure 1, Figures S1 and S2). Thirty-one of the 33 QTL were detected using GLM and 13 with MLM (Tables S4 and S5). Of these latter 13, two QTL (QTL 18 for IOD and QTL 31 for LIN) were detected only by MLM, while the remaining 11 were identified by both MLM and GLM (Table S4).

Out of 33 QTL identified, 12, 6, 3 and 27 were from EV, SU, BM and merged population, respectively. Only six QTL were detected exclusively from two individual populations, including four (QTL 2 and 6 for PLH, QTL 8 for DTM and QTL 17 for OIL) from EV and two (QTL 3 and 4 for PLH) from BM. Eighteen were identified exclusively from the merged population. Ten QTL were detected simultaneously from the merged population and one or more individual populations (Tables S4 and S5).

QTL for YLD (QTL 1) was identified only in two environments (2010/Morden and 2012/Saskatoon) (Figure S2) but not in other environments or using BLUP estimates over the six year/location environments. We also performed GWAS for all other traits with phenotypic data from individual environments and obtained similar results with the QTL identified using BLUP values over multiple environments (Table S6).
