Table 1.

Summary of multi-environment QTL detected in this study using two-stage QTL analysis

Trait	Na	Thresh.b	ICMc	cMd	LODe	Expl. Varf	Modelg	Act.h	Dir.i	Effectj	SEk
Stem length	121	4.0	2	19 (19.8)l	4.4 (4.6)l	0.14 (0.15)l	oQ2Q3o × ooQ7Q8	CD	—	****	****
Stem prickles	121	4.4	3	22 (27.4)	6.5 (7.8)	0.21 (0.25)	ooQQ × Qooo	SD	+	2.75	0.67
			4	74 (74.5)	5.8 (6.1)	0.19 (0.20)	oooQ × oQoo	A	−	1.92	0.34
			6	14 (14.4)	4.4 (4.1)	0.14 (0.13)	QQoQ × oQoo	A	−	1.61	0.32
Petiole prickles	121	4.5	4	66 (64.7)	6.7 (7.0)	0.21 (0.22)	QQoo × Qooo	A	+	0.78	0.12
Num. petals	120	5.2	2	88 (92.7)	6.5 (7.1)	0.21 (0.23)	QooQ × oooQ	A	+	8.62	1.46
			3	27 (33.7)	6.1 (6.6)	0.20 (0.21)	oQQo × oooQ	DD	−	6.71	1.01
Chlorophyll	121	3.0	4	50 (50.3)	3.3 (3.5)	0.11 (0.11)	oQQo × QQoo	SD	+	2.95	0.95

^aN number of offspring for which genotype and phenotype data were available for each trait

^bThresh. experiment-wide LOD significance threshold, determined by permutation test with N = 1000 and α = 0.05

^cICM chromosomal linkage group, using the integrated consensus map (ICM) numbering of Spiller et al. (2011) and Bourke et al. (2017)

^dcM centiMorgan position of QTL peak

^eLOD logarithm of odds at QTL peak

^fExpl. Var. fraction of phenotypic variance explained by the QTL model at the peak position ($R_{\text{adj}}^2$ of the linear model)

^gModel QTL model that best fit the data at the QTL position (i.e. minimised the BIC). “Q” signifies a predicted QTL allele with an estimated effect, whereas “o” denotes an allele with neutral effect (i.e. the remaining, grouped alleles at a QTL whose effects are not estimated). In cases where the most likely model was multi-allelic, subscripts (Q₁ etc.) are used to denote the parental origin of the allele. Note that in the case of dominance, the Q (or QQ) alleles are taken to be dominant

^hAct. mode of action of QTL model, A (bi-allelic) additive, CD (multi-allelic) codominant, SD (bi-allelic) simplex dominant, DD (bi-allelic) duplex dominant

ⁱDir. Direction of the allele effect, either positive (+), i.e. increasing trait value, or negative (−). In the case of multiple QTL alleles, the direction of each QTL allele Q_n is given

^jEffect QTL allele effect, estimated using the slope of a (weighted) regression of the genotype means from the 36 genotype classes against the QTL allele count (coded as 0/1 in the case of both simplex or duplex dominance to signify the absence or presence of the dominance-causing alleles)

^kSE standard error of the estimated slope of the regression line

^lNumbers shown are results from initial QTL scan, with those in parentheses the results following re-saturation of QTL region with additional markers

****Q₂ = 1.28 ± 1.76, Q₃ = 2.35 ± 1.73, Q₇ = 2.78 ± 1.73, Q₈ = 5.87 ± 1.47