Supporting Information Functional multi-locus QTL mapping of temporal trends in Scots pine wood traits

Marker sorting and mapping was performed with all the available genotype data simultaneously with the aim of constructing one pure AFLP dataset (A-set) for the larger subset of individuals and one mixed SNP-AFLP dataset (S+A-set) intended for the smaller subset. Because the studied full-sib family was generated by two non-inbred and highly heterozygotic parents, a two-way pseudo-testcross mapping strategy was employed (e.g. Grattapaglia and Sederoff, 1994; Grattapaglia et al., 1995). Markers for which genotyping scoring success was inadequate (< 80%) and poorly genotyped individuals (< 70%) were excluded from further study. By checking parental heterozygosity and genotypic frequencies of the offspring, an additional selection was performed where codominant SNP markers were accepted given that genotypes segregated according to 1:1 proportions (one of the parents being heterozygote) or to 1:2:1 proportions (both parents heterozygotic). For dominant AFLP however, only a 1:1 band presence/absence segregation proportion was accepted (only one parent heterozygote). Selection was performed based on submitting genotype proportions to χ-tests using 0.01 as a threshold for the p-value. Among the selected markers, recombination fractions were calculated for all pairs of markers in order to estimate marker linkage and phase. For most marker pairs this could be done by division, but for pairs of 1:2:1-segregating SNPs, recombination frequencies were estimated via the maximum likelihood EM-algorithm (Dempster et al., 1977). Linkage between markers was declared based on a log-likelihood ratio of 23.03 or higher (equivalent to LOD>5). Even though the inclusion of 1:2:1-segregating markers made the construction of a consensus map possible, it was nonetheless reasonable to assume that different sets of QTL segregated within each of the unrelated and heterozygotic parents and it was thus more meaningful to separate the marker linkage mapping into maternal and paternal sections. The 1:1-segregating markers were assigned to linkage groups based on the parent for which they exhibited heterozygosity as well as on linkage. The 1:2:1-segregating markers were instead duplicated into two identical 1:1segregating markers each assigned to male and female linkage groups respectively. Heterozygote genotypes of 1:2:1-segregating markers were henceforth treated as unknown (replaced by missing