Table S3. Environmental data for all collections.
 
Table S4. Minor allele frequency (MAF) for all collections genotyped at all 13 markers. The table includes average MAF for all collections and by lineage. Fixed alleles are outlined in red and averaged MAF for markers 8-12 are highlighted gray.
 
Figure S1. Bayesian Information Criterion (BIC) values averaged over 25 iterations and divided by the standard deviation for K values 1-10.
 
Figure S2. Minor allele frequency (MAF) for 13 candidate markers with coastal collection averages represented by the orange line and inland collection averages represented by the gray line.
 
Figure S3. Comparison of average genotype frequencies (all populations) for different sets of candidate markers. The six sets of markers include the following: a single marker (9 as in Micheletti et al. 2018a); three markers from the same greb1L haplotype block (2,3,6); four markers available for all collections (2,3,6,9); five markers from the intergenic haplotype block (8-12); six markers from the greb1L haplotype block (2-7); and 11 markers excluding one from each of the distal ends of the candidate genomic region. Error bars represent standard error.
 
Figure S4. Maps of haplotype and genotype proportions for all collection locations. Pie chart size corresponds to population size. Populations with ≥ 100 individuals were reduced to 100 with the same genotype proportions to keep the charts on the map as visible as possible. See Table S1 for collection names and exact genotype proportions. The first map (a) incorporates three candidate markers (2, 3, and 6) on chromosome 28 associated with maturation timing in O. mykiss. These 3 markers were evaluated to include as many populations as possible, while excluding marker 9 due to a greater association with haplotype block 2. The haplotypes representative of the heterozygote genotype are depicted as a gradient corresponding to the number of markers that match either fixed genotype. There are 5 unique haplotypes for markers 2, 3, and 6. The percentage of individuals with each haplotype is reported in the table. The completely blue haplotype matches the mature genotype and is the most frequent, while the completely red haplotype matches the premature genotype and is the third most frequent. The haplotypes with a mixture of blue and red represent the different possible heterozygote genotypes. The second map (b) incorporates only candidate marker 9 (Omy_RAD47080-54).
 
Figure S5. RDA of all collections in Columbia River basin to model the degree to which the variation in environmental variables explains the variation in allele frequencies for candidate markers for all collections in the greb1L haplotype block (2,3,6). The populations are represented by text and colored black or red in accordance with their lineage determined by DAPC in adegenet. The arrows spatially denote a significant influence of environmental variables and the length of the arrow indicates the extent of the effect. Coastal populations (a) and inland populations (b) were analyzed separately. Environmental variables retained in RDA of coastal populations were mean temperature coldest quarter and precipitation of the wettest month. Environmental variables retained in RDA of inland lineage populations were August water temperature over a 20-year average and minimum temperature of the warmest month.