Supporting Information Figs S1–S5, Tables S2–S4, S8 & S9 and Methods S1. Methods S1 Estimating genetic gains from selection for RCN, XG, and TDM. First generation gains were predicted for RCN, XG, or TDM individually and for a linear combination of all three traits in CCLONES. The across-site additive genetic trait values were estimated by adding the population mean to the across-site additive genetic deviations for each individual (i.e. the vector ‘a’ in the across-site model). Percent change in the population mean from selective breeding was predicted by comparing the mean genetic values of all individuals that were sampled for resin canal traits in CCLONES to the mean of individuals in the 95th percentile. The Box-Cox transformed oleoresin dry mass (TDM) data were transformed back into the original units (g) to estimate gains in oleoresin dry mass (Westbrook et al., 2013). Westbrook JW, Resende MFR, Munoz P, Walker AR, Wegrzyn JL, Nelson CD, Neale DB, Kirst M, Huber DA, Gezan SA et al. 2013. Association genetics of oleoresin flow in loblolly pine: discovering genes and predicting phenotype for improved resistance to bark beetles and bioenergy potential. New Phytologist 199: 89–100. Table S2 Genetic distances between mapped genes in CCLONES and BC1 LG 1 2 3 4 5 6 7 8 9 10 11 12 Average Total All All ALL CCLONES CCLONES CCLONES BC1 BC1 BC1 N genes avg. dist. max dist. N genes avg. dist. max dist. N genes avg. dist. max dist. 259 0.575 3.672 107 1.308 6.346 53 2.275 9.911 309 0.500 6.826 118 1.252 23.732 55 2.634 32.538 245 0.642 8.061 110 1.142 7.625 65 1.843 12.917 267 0.571 3.610 100 1.367 6.944 56 2.422 16.615 344 0.413 4.006 135 1.057 9.763 75 1.831 11.853 292 0.434 5.211 115 1.055 7.269 52 2.044 8.549 296 0.510 2.686 125 1.140 5.572 54 2.152 22.866 284 0.665 16.769 118 1.270 8.036 64 1.757 16.370 273 0.550 9.323 118 1.183 7.533 66 1.907 14.404 287 0.611 8.000 112 1.314 7.585 59 2.028 15.897 240 0.507 2.676 99 1.118 11.202 58 1.467 10.363 256 0.465 3.568 116 1.001 8.016 64 1.827 10.879 279 0.533 6.201 114 1.180 9.169 60 1.998 15.263 3352 1373 721 N, number of mapped genes; avg. dist., mean distance between adjacent genes; max dist., maximum distance between adjacent genes; LG, linkage group. Columns with ‘All’ represent the total set of mapped genes, while ‘CCLONES’ and ‘BC1’ columns represent the mapped genes that are polymorphic within these populations. Table S3 Identity-by-descent (IBD) proportions for pairs of individuals in CCLONES that were sampled for resin canal traits IBD proportion 0 0.0625 0.125 0.1875 0.25 0.3125 0.375 0.5 N relations 99,868 6,118 8168 812 16,276 616 240 3,362 % relations 73.7% 4.5% 6.0% 0.6% 12% 0.5% 0.2% 2.5% Relationship type Unrelated Half first cousin First cousin Half sib Full sib Identity-by-descent proportions between all pairs of 520 individuals that were sampled resin canal traits were obtained from the pedigree additive genetic relationship matrix. N relations, number of pairs of individuals within a IBD proportion class; % relations, percent of pair wise relationships. Table S4 Additional genetic correlations between RCN and TDM RCN site/yr A6+7 C6+7 N6+7 N6 P6+7 TDM site/yr A7 C7 N7 N7 P7 rg ± SE LLfull LLnull P 0.1825 ± 0.0991 0.1781 ± 0.1407 0.2606 ± 0.1378 0.1229 ± 0.1531 0.1256 ± 0.1473 3104.52 -2213.54 -2217.37 -1673.46 -2483.51 3102.99 -2214.30 -2219.00 -1673.76 -2483.81 0.08 0.48 0.07 0.43 0.43 A, across sites; C, Cuthbert; N, Nassau; P, Palatka; 6, age six; 7, age seven; 6+7, ages six and seven; LLfull, log-likelihood of full model where rg is estimated; LLnull, log-likelihood of the null model where rg is fixed to zero; P, P-value of likelihood ratio test. Table S8 Number of significant xylem growth marker–trait associations shared between field sites and ages in the P. taeda CCLONES population Avg. # shared A6+7 5.36 A6 3.91 26 A7 3 3.09 25 C6+7 7 2 3.09 38 C6 6 1 12 2.18 24 C7 1 3 8 2 1.82 17 N6+7 5 6 0 0 1 4.18 45 N6 4 1 0 0 0 12 2.91 27 N7 3 7 0 0 1 13 4 3.27 28 P6+7 0 0 0 0 0 0 1 1 1.18 10 P6 1 1 0 0 0 0 3 2 9 2.36 32 P7 0 2 0 0 0 1 1 0 2 5 1.18 31 2.88 Avg. # shared among all analyses A, across sites; C, Cuthbert; N, Nassau; P, Palatka; 6, age six; 7, age seven; 6+7, sum of ages six and seven. A6+7 38 13 8 5 3 4 8 6 5 0 5 2 A6 A7 C6+7 C6 C7 N6+7 N6 N7 P6+7 P6 P7 Table S9 Tests for whether the number unidirectional associations shared between RCN, XG, and TDM exceeded random expectation Trait combination RCN & XG RCN & TDM XG & TDM RCN, XG, & TDM Observed N Max N random P unidirectional unidirectional associations associations 49 17 0 19 18 0 16 17 0.00002 7 4 0 Random samples of SNPs equal in size to the observed number of loci associated with resin canal number (RCN, 251), xylem growth (XG, 220), and transformed oleoresin dry mass (TDM, 231) were selected 50,000 times from the set of 4854 SNPs used for association analysis. A positive or negative effect was randomly assigned to each SNP and the number SNPs that had unidirectional effect on combinations of two or three traits were tallied. The probability observing the number of unidirectional associations for all trait combinations by chance was calculated as one minus the quantile of the observed number unidirectional associations within the random distributions of shared associations. Fig. S1 Example CT image (resolution 10 m pixel-1) of wood cores used for resin canal number (numbers) and xylem growth (brackets) phenotyping. Three 2 mm 15 mm wood cores were sampled from each tree and phenotypes were averaged prior to quantitative genetic analysis. Fig. S2 Phenotypic distributions of resin canal traits, xylem growth, and oleoresin dry mass from the Pinus taeda CCLONES population at three field sites. RCN, resin canal number (ages six and seven); XG, xylem growth (ages six and seven); RCD, resin canal density (average of ages six and seven); RCLWP, resin canal latewood proportion (average of ages six and seven); MRCA, mean resin canal area (average of ages six and seven); RDM, oleoresin dry mass (age seven). Fig. S3 Decay of linkage disequilibrium (R2) with genetic map distance in BC1 and CCLONES. In BC1, pairwise R2 was estimated among 721 mapped genes containing SNPs segregating in 345 individuals. In CCLONES, pairwise R2 was estimated among 1373 mapped genes containing SNPs segregating in 520 individuals. Estimates of R2 were obtained with the R package ‘LDcorSV’. Kernal regression of R2 vs genetic distance was performed in R with the ‘ksmooth’ function using a bandwidth of 10 cM. Fig. S4 All pair wise genetic and phenotypic correlations between resin canal number (RCN), transformed oleoresin drymass (TDM), xylem growth (XG), resin canal density (RCD), and resin canal latewood proportion (RCLWP) within and across sites in the P. taeda CCLONES population. Significant deviation from zero correlation was assessed with likelihood ratio tests: ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. Fig. S5 Contrasts of resin canal number and xylem growth family means at age five among parental half sibs and full sibs in the BC1 population. Pt1, half sib progeny of the P. taeda grandparent from the first generation of selective breeding; Pe1, half sib progeny from the P. elliottii grandparent from first generation selections, Pe3 – half sib progeny of the P. elliottii parent from third generation selections, and BC1– pseudo-backcross ({Pt1 Pe1} Pe3) full sibs. Family means were contrasted with Tukey tests. Different letters indicate significant differences at P < 0.05.