Efficient genomics-based 'end-to-end' selective tree breeding framework.

Since their initiation in the 1950s, worldwide selective tree breeding programs followed the recurrent selection scheme of repeated cycles of selection, breeding (mating), and testing phases and essentially remained unchanged to accelerate this process or address environmental contingencies and concerns. Here, we introduce an "end-to-end" selective tree breeding framework that: (1) leverages strategically preselected GWAS-based sequence data capturing trait architecture information, (2) generates unprecedented resolution of genealogical relationships among tested individuals, and (3) leads to the elimination of the breeding phase through the utilization of readily available wind-pollinated (OP) families. Individuals' breeding values generated from multi-trait multi-site analysis were also used in an optimum contribution selection protocol to effectively manage genetic gain/co-ancestry trade-offs and traits' correlated response to selection. The proof-of-concept study involved a 40-year-old spruce OP testing population growing on three sites in British Columbia, Canada, clearly demonstrating our method's superiority in capturing most of the available genetic gains in a substantially reduced timeline relative to the traditional approach. The proposed framework is expected to increase the efficiency of existing selective breeding programs, accelerate the start of new programs for ecologically and environmentally important tree species, and address climate-change caused biotic and abiotic stress concerns more effectively.

Decoupling of height growth and drought or pest resistance tradeoffs is revealed through multiple common-garden experiments of lodgepole pine.

The environment could alter growth and resistance tradeoffs in plants by affecting the ratio of resource allocation to various competing traits. Yet, how and why functional tradeoffs change over time and space is poorly understood particularly in long-lived conifer species. By establishing four common-garden test sites for five lodgepole pine populations in western Canada, combined with genomic sequencing, we revealed the decoupling pattern and genetic underpinnings of tradeoffs between height growth, drought resistance based on δ13C and dendrochronology, and metrics of pest resistance based on pest suitability ratings. Height and δ13C correlation displayed a gradient change in magnitude and/or direction along warm-to-cold test sites. All cold test sites across populations showed a positive height and δ13C relationship. However, we did not observe such a clinal correlation pattern between height or δ13C and pest suitability. Further, we found that the study populations exhibiting functional tradeoffs or synergies to various degrees in test sites were driven by non-adaptive evolutionary processes rather than adaptive evolution or plasticity. Finally, we found positive genetic relationships between height and drought or pest resistance metrics and probed five loci showing potential genetic tradeoffs between northernmost and the other populations. Our findings have implications for deciphering the ecological, evolutionary, and genetic bases of the decoupling of functional tradeoffs due to environmental change.

Pest defences under weak selection exert a limited influence on the evolution of height growth and drought avoidance in marginal pine populations.

While droughts, intensified by climate change, have been affecting forests worldwide, pest epidemics are a major source of uncertainty for assessing drought impacts on forest trees. Thus far, little information has documented the adaptability and evolvability of traits related to drought and pests simultaneously. We conducted common-garden experiments to investigate how several phenotypic traits (i.e. height growth, drought avoidance based on water-use efficiency inferred from δ13C and pest resistance based on defence traits) interact in five mature lodgepole pine populations established in four progeny trials in western Canada. The relevance of interpopulation variation in climate sensitivity highlighted that seed-source warm populations had greater adaptive capability than cold populations. In test sites, warming generated taller trees with higher δ13C and increased the evolutionary potential of height growth and δ13C across populations. We found, however, no pronounced gradient in defences and their evolutionary potential along populations or test sites. Response to selection was weak in defences across test sites, but high for height growth particularly at warm test sites. Response to the selection of δ13C varied depending on its selective strength relative to height growth. We conclude that warming could promote the adaptability and evolvability of growth response and drought avoidance with a limited evolutionary influence from pest (biotic) pressures.

Generating Douglas-fir Breeding Value Estimates Using Airborne Laser Scanning Derived Height and Crown Metrics.

Progeny test trials in British Columbia are essential in assessing the genetic performance via the prediction of breeding values (BVs) for target phenotypes of parent trees and their offspring. Accurate and timely collection of phenotypic data is critical for estimating BVs with confidence. Airborne Laser Scanning (ALS) data have been used to measure tree height and structure across a wide range of species, ages and environments globally. Here, we analyzed a Coastal Douglas-fir [Pseudotsuga menziesii var. menziesii (Mirb.)] progeny test trial located in British Columbia, Canada, using individual tree high-density Airborne Laser Scanning (ALS) metrics and traditional ground-based phenotypic observations. Narrow-sense heritability, genetic correlations, and BVs were estimated using pedigree-based single and multi-trait linear models for 43 traits. Comparisons of genetic parameter estimates between ALS metrics and traditional ground-based measures and single- and multi-trait models were conducted based on the accuracy and precision of the estimates. BVs were estimated for two ALS models (ALSCAN and ALSACC) representing two model-building approaches and compared to a baseline model using field-measured traits. The ALSCAN model used metrics reflecting aspects of vertical distribution of biomass within trees, while ALSACC represented the most statistically accurate model. We report that the accuracy of both the ALSCAN (0.8239) and ALSACC (0.8254) model-derived BVs for mature tree height is a suitable proxy for ground-based mature tree height BVs (0.8316). Given the cost efficiency of ALS, forest geneticists should explore this technology as a viable tool to increase breeding programs' overall efficiency and cost savings.

Multiple-trait analyses improved the accuracy of genomic prediction and the power of genome-wide association of productivity and climate change-adaptive traits in lodgepole pine.

BACKGROUND: Genomic prediction (GP) and genome-wide association (GWA) analyses are currently being employed to accelerate breeding cycles and to identify alleles or genomic regions of complex traits in forest trees species. Here, 1490 interior lodgepole pine (Pinus contorta Dougl. ex. Loud. var. latifolia Engelm) trees from four open-pollinated progeny trials were genotyped with 25,099 SNPs, and phenotyped for 15 growth, wood quality, pest resistance, drought tolerance, and defense chemical (monoterpenes) traits. The main objectives of this study were to: (1) identify genetic markers associated with these traits and determine their genetic architecture, and to compare the marker detected by single- (ST) and multiple-trait (MT) GWA models; (2) evaluate and compare the accuracy and control of bias of the genomic predictions for these traits underlying different ST and MT parametric and non-parametric GP methods. GWA, ST and MT analyses were compared using a linear transformation of genomic breeding values from the respective genomic best linear unbiased prediction (GBLUP) model. GP, ST and MT parametric and non-parametric (Reproducing Kernel Hilbert Spaces, RKHS) models were compared in terms of prediction accuracy (PA) and control of bias. RESULTS: MT-GWA analyses identified more significant associations than ST. Some SNPs showed potential pleiotropic effects. Averaging across traits, PA from the studied ST-GP models did not differ significantly from each other, with generally a slight superiority of the RKHS method. MT-GP models showed significantly higher PA (and lower bias) than the ST models, being generally the PA (bias) of the RKHS approach significantly higher (lower) than the GBLUP. CONCLUSIONS: The power of GWA and the accuracy of GP were improved when MT models were used in this lodgepole pine population. Given the number of GP and GWA models fitted and the traits assessed across four progeny trials, this work has produced the most comprehensive empirical genomic study across any lodgepole pine population to date.

Integrating genomic information and productivity and climate-adaptability traits into a regional white spruce breeding program.

Tree improvement programs often focus on improving productivity-related traits; however, under present climate change scenarios, climate change-related (adaptive) traits should also be incorporated into such programs. Therefore, quantifying the genetic variation and correlations among productivity and adaptability traits, and the importance of genotype by environment interactions, including defense compounds involved in biotic and abiotic resistance, is essential for selecting parents for the production of resilient and sustainable forests. Here, we estimated quantitative genetic parameters for 15 growth, wood quality, drought resilience, and monoterpene traits for Picea glauca (Moench) Voss (white spruce). We sampled 1,540 trees from three open-pollinated progeny trials, genotyped with 467,224 SNP markers using genotyping-by-sequencing (GBS). We used the pedigree and SNP information to calculate, respectively, the average numerator and genomic relationship matrices, and univariate and multivariate individual-tree models to obtain estimates of (co)variance components. With few site-specific exceptions, all traits examined were under genetic control. Overall, higher heritability estimates were derived from the genomic- than their counterpart pedigree-based relationship matrix. Selection for height, generally, improved diameter and water use efficiency, but decreased wood density, microfibril angle, and drought resistance. Genome-based correlations between traits reaffirmed the pedigree-based correlations for most trait pairs. High and positive genetic correlations between sites were observed (average 0.68), except for those pairs involving the highest elevation, warmer, and moister site, specifically for growth and microfibril angle. These results illustrate the advantage of using genomic information jointly with productivity and adaptability traits, and defense compounds to enhance tree breeding selection for changing climate.

Improving lodgepole pine genomic evaluation using spatial correlation structure and SNP selection with single-step GBLUP.

Modeling environmental spatial heterogeneity can improve the efficiency of forest tree genomic evaluation. Furthermore, genotyping costs can be lowered by reducing the number of markers needed. We investigated the impact on variance components, breeding value accuracy, and bias of two phenotypic data adjustments (experimental design and autoregressive spatial models), and a relationship matrix calculated from a subset of markers selected for their ability to infer ancestry. Using a multiple-trait multiple-site single-step Genomic Best Linear Unbiased Prediction (ssGBLUP) approach, four scenarios (2 phenotype adjustments × 2 marker sets) were applied to diameter at breast height (DBH), height (HT), and resistance to western gall rust (WGR) in four open-pollinated progeny trials of lodgepole pine, with 1490 (out of 11,188) trees genotyped with 25,099 SNPs. As a control, we fitted the conventional ABLUP model using pedigree information. The highest heritability estimates were achieved for the ABLUP followed closely by the ssGBLUP with the full marker set and using the spatial phenotype adjustments. The highest predictive ability was obtained by using a reduced marker subset (8000 SNPs) when either the spatial (DBH: 0.429, and WGR: 0.513) or design (HT: 0.467) phenotype corrections were used. No significant difference was detected in prediction bias among the six fitted models, and all values were close to 1 (0.918-1.014). Results demonstrated that selecting informative markers, such as those capturing ancestry, can improve the predictive ability. The use of spatial correlation structure increased traits' heritability and reduced prediction bias, while increases in predictive ability were trait-dependent.

Quercus species divergence is driven by natural selection on evolutionarily less integrated traits.

Functional traits are organismal attributes that can respond to environmental cues, thereby providing important ecological functions. In addition, an organism's potential for adaptation is defined by the patterns of covariation among groups of functionally related traits. Whether an organism is evolutionarily constrained or has the potential for adaptation is based on the phenotypic integration or modularity of these traits. Here, we revisited leaf morphology in two European sympatric white oaks (Quercus petraea (Matt.) Liebl. and Quercus robur L.), sampling 2098 individuals, across much of their geographical distribution ranges. At the phenotypic level, leaf morphology traditionally encompasses discriminant attributes among different oak species. Here, we estimated in situ heritability, genetic correlation, and integration across such attributes. Also, we performed Selection Response Decomposition to test these traits for potential differences in oak species' evolutionary responses. Based on the uncovered functional units of traits (modules) in our study, the morphological module "leaf size gradient" was highlighted among functionally integrated traits. Equally, this module was defined in both oaks as being under "global regulation" in vegetative bud establishment and development. Lamina basal shape and intercalary veins' number were not, or, less integrated within the initially defined leaf functional unit, suggesting more than one module within the leaf traits' ensemble. Since these traits generally show the greatest species discriminatory power, they potentially underwent effective differential response to selection among oaks. Indeed, the selection of these traits could have driven the ecological preferences between the two sympatric oaks growing under different microclimates.

Linkage disequilibrium vs. pedigree: Genomic selection prediction accuracy in conifer species.

BACKGROUND: The presupposition of genomic selection (GS) is that predictive accuracies should be based on population-wide linkage disequilibrium (LD). However, in species with large, highly complex genomes the limitation of marker density may preclude the ability to resolve LD accurately enough for GS. Here we investigate such an effect in two conifer species with ~ 20 Gbp genomes, Douglas-fir (Pseudotsuga menziesii Mirb. (Franco)) and Interior spruce (Picea glauca (Moench) Voss x Picea engelmannii Parry ex Engelm.). Random sampling of markers was performed to obtain SNP sets with totals in the range of 200-50,000, this was replicated 10 times. Ridge Regression Best Linear Unbiased Predictor (RR-BLUP) was deployed as the GS method to test these SNP sets, and 10-fold cross-validation was performed on 1,321 Douglas-fir trees, representing 37 full-sib F1 families and on 1,126 Interior spruce trees, representing 25 open-pollinated (half-sib) families. Both trials are located on 3 sites in British Columbia, Canada. RESULTS: As marker number increased, so did GS predictive accuracy for both conifer species. However, a plateau in the gain of accuracy became apparent around 10,000-15,000 markers for both Douglas-fir and Interior spruce. Despite random marker selection, little variation in predictive accuracy was observed across replications. On average, Douglas-fir prediction accuracies were higher than those of Interior spruce, reflecting the difference between full- and half-sib families for Douglas-fir and Interior spruce populations, respectively, as well as their respective effective population size. CONCLUSIONS: Although possibly advantageous within an advanced breeding population, reducing marker density cannot be recommended for carrying out GS in conifers. Significant LD between markers and putative causal variants was not detected using 50,000 SNPS, and GS was enabled only through the tracking of relatedness in the populations studied. Dramatically increasing marker density would enable said markers to better track LD with causal variants in these large, genetically diverse genomes; as well as providing a model that could be used across populations, breeding programs, and traits.

SNP variable selection by generalized graph domination.

BACKGROUND: High-throughput sequencing technology has revolutionized both medical and biological research by generating exceedingly large numbers of genetic variants. The resulting datasets share a number of common characteristics that might lead to poor generalization capacity. Concerns include noise accumulated due to the large number of predictors, sparse information regarding the p≫n problem, and overfitting and model mis-identification resulting from spurious collinearity. Additionally, complex correlation patterns are present among variables. As a consequence, reliable variable selection techniques play a pivotal role in predictive analysis, generalization capability, and robustness in clustering, as well as interpretability of the derived models. METHODS AND FINDINGS: K-dominating set, a parameterized graph-theoretic generalization model, was used to model SNP (single nucleotide polymorphism) data as a similarity network and searched for representative SNP variables. In particular, each SNP was represented as a vertex in the graph, (dis)similarity measures such as correlation coefficients or pairwise linkage disequilibrium were estimated to describe the relationship between each pair of SNPs; a pair of vertices are adjacent, i.e. joined by an edge, if the pairwise similarity measure exceeds a user-specified threshold. A minimum k-dominating set in the SNP graph was then made as the smallest subset such that every SNP that is excluded from the subset has at least k neighbors in the selected ones. The strength of k-dominating set selection in identifying independent variables, and in culling representative variables that are highly correlated with others, was demonstrated by a simulated dataset. The advantages of k-dominating set variable selection were also illustrated in two applications: pedigree reconstruction using SNP profiles of 1,372 Douglas-fir trees, and species delineation for 226 grasshopper mouse samples. A C++ source code that implements SNP-SELECT and uses Gurobi optimization solver for the k-dominating set variable selection is available (https://github.com/transgenomicsosu/SNP-SELECT).

Genomic selection of juvenile height across a single-generational gap in Douglas-fir.

Here, we perform cross-generational GS analysis on coastal Douglas-fir (Pseudotsuga menziesii), reflecting trans-generational selective breeding application. A total of 1321 trees, representing 37 full-sib F1 families from 3 environments in British Columbia, Canada, were used as the training population for (1) EBVs (estimated breeding values) of juvenile height (HTJ) in the F1 generation predicting genomic EBVs of HTJ of 136 individuals in the F2 generation, (2) deregressed EBVs of F1 HTJ predicting deregressed genomic EBVs of F2 HTJ, (3) F1 mature height (HT35) predicting HTJ EBVs in F2, and (4) deregressed F1 HT35 predicting genomic deregressed HTJ EBVs in F2. Ridge regression best linear unbiased predictor (RR-BLUP), generalized ridge regression (GRR), and Bayes-B GS methods were used and compared to pedigree-based (ABLUP) predictions. GS accuracies for scenarios 1 (0.92, 0.91, and 0.91) and 3 (0.57, 0.56, and 0.58) were similar to their ABLUP counterparts (0.92 and 0.60, respectively) (using RR-BLUP, GRR, and Bayes-B). Results using deregressed values fell dramatically for both scenarios 2 and 4 which approached zero in many cases. Cross-generational GS validation of juvenile height in Douglas-fir produced predictive accuracies almost as high as that of ABLUP. Without capturing LD, GS cannot surpass the prediction of ABLUP. Here we tracked pedigree relatedness between training and validation sets. More markers or improved distribution of markers are required to capture LD in Douglas-fir. This is essential for accurate forward selection among siblings as markers that track pedigree are of little use for forward selection of individuals within controlled pollinated families.

Quantitative Genetics and Genomics Converge to Accelerate Forest Tree Breeding.

Forest tree breeding has been successful at delivering genetically improved material for multiple traits based on recurrent cycles of selection, mating, and testing. However, long breeding cycles, late flowering, variable juvenile-mature correlations, emerging pests and diseases, climate, and market changes, all pose formidable challenges. Genetic dissection approaches such as quantitative trait mapping and association genetics have been fruitless to effectively drive operational marker-assisted selection (MAS) in forest trees, largely because of the complex multifactorial inheritance of most, if not all traits of interest. The convergence of high-throughput genomics and quantitative genetics has established two new paradigms that are changing contemporary tree breeding dogmas. Genomic selection (GS) uses large number of genome-wide markers to predict complex phenotypes. It has the potential to accelerate breeding cycles, increase selection intensity and improve the accuracy of breeding values. Realized genomic relationships matrices, on the other hand, provide innovations in genetic parameters' estimation and breeding approaches by tracking the variation arising from random Mendelian segregation in pedigrees. In light of a recent flow of promising experimental results, here we briefly review the main concepts, analytical tools and remaining challenges that currently underlie the application of genomics data to tree breeding. With easy and cost-effective genotyping, we are now at the brink of extensive adoption of GS in tree breeding. Areas for future GS research include optimizing strategies for updating prediction models, adding validated functional genomics data to improve prediction accuracy, and integrating genomic and multi-environment data for forecasting the performance of genetic material in untested sites or under changing climate scenarios. The buildup of phenotypic and genome-wide data across large-scale breeding populations and advances in computational prediction of discrete genomic features should also provide opportunities to enhance the application of genomics to tree breeding.

Temporal quantification of mating system parameters in a coastal Douglas-fir seed orchard under manipulated pollination environment.

Seed orchards main function is delivering breeding programs' gains in the form of genetically improved seedlings. They are unique experimental populations, perfectly suited for studying various pollination environments (natural or otherwise), affecting their mating system parameters. Here, under different pollination environment (natural and intrusive (pollen augmentation and/or bloom-delay)), the mating system of a second generation, wind-pollinated, coastal Douglas-fir (Pseudotsuga menziesii var. menziesii (Mirb.) Franco) seed orchard was evaluated over four years. Using DNA microsatellite markers and bulk seed samples, we conducted pedigree reconstruction to assign each seed's male and female parents, followed by determining the extent of pollen contamination (external gene flow), selfing rate, and, parental gametic contribution for each year. Overall, external pollen contamination rates ranged between 10 and 28%, selfing rate varied between 12 and 17%, and 80% of the seed crops were produced by 37-64% of the orchard's parents. Pollination environment and seed crop size substantially influenced the observed results, particularly for small crops as pollen contamination was high in natural (28%) vs. intrusive pollination (10%). Generally, irrespective of the crop size, seed produced under natural pollination had higher pollen contamination, confirming the role of pollination environment manipulation in improving seed crops' genetic quality.

Multienvironment genomic variance decomposition analysis of open-pollinated Interior spruce (Picea glauca x engelmannii).

The advantages of open-pollinated (OP) family testing over controlled crossing (i.e., structured pedigree) are the potential to screen and rank a large number of parents and offspring with minimal cost and efforts; however, the method produces inflated genetic parameters as the actual sibling relatedness within OP families rarely meets the half-sib relatedness assumption. Here, we demonstrate the unsurpassed utility of OP testing after shifting the analytical mode from pedigree- (ABLUP) to genomic-based (GBLUP) relationship using phenotypic tree height (HT) and wood density (WD) and genotypic (30k SNPs) data for 1126 38-year-old Interior spruce (Picea glauca (Moench) Voss x P. engelmannii Parry ex Engelm.) trees, representing 25 OP families, growing on three sites in Interior British Columbia, Canada. The use of the genomic realized relationship permitted genetic variance decomposition to additive, dominance, and epistatic genetic variances, and their interactions with the environment, producing more accurate narrow-sense heritability and breeding value estimates as compared to the pedigree-based counterpart. The impact of retaining (random folding) vs. removing (family folding) genetic similarity between the training and validation populations on the predictive accuracy of genomic selection was illustrated and highlighted the former caveats and latter advantages. Moreover, GBLUP models allowed breeding value prediction for individuals from families that were not included in the developed models, which was not possible with the ABLUP. Response to selection differences between the ABLUP and GBLUP models indicated the presence of systematic genetic gain overestimation of 35 and 63% for HT and WD, respectively, mainly caused by the inflated estimates of additive genetic variance and individuals' breeding values given by the ABLUP models. Extending the OP genomic-based models from single to multisite made the analysis applicable to existing OP testing programs.

Genomic prediction accuracies in space and time for height and wood density of Douglas-fir using exome capture as the genotyping platform.

BACKGROUND: Genomic selection (GS) can offer unprecedented gains, in terms of cost efficiency and generation turnover, to forest tree selective breeding; especially for late expressing and low heritability traits. Here, we used: 1) exome capture as a genotyping platform for 1372 Douglas-fir trees representing 37 full-sib families growing on three sites in British Columbia, Canada and 2) height growth and wood density (EBVs), and deregressed estimated breeding values (DEBVs) as phenotypes. Representing models with (EBVs) and without (DEBVs) pedigree structure. Ridge regression best linear unbiased predictor (RR-BLUP) and generalized ridge regression (GRR) were used to assess their predictive accuracies over space (within site, cross-sites, multi-site, and multi-site to single site) and time (age-age/ trait-trait). RESULTS: The RR-BLUP and GRR models produced similar predictive accuracies across the studied traits. Within-site GS prediction accuracies with models trained on EBVs were high (RR-BLUP: 0.79-0.91 and GRR: 0.80-0.91), and were generally similar to the multi-site (RR-BLUP: 0.83-0.91, GRR: 0.83-0.91) and multi-site to single-site predictive accuracies (RR-BLUP: 0.79-0.92, GRR: 0.79-0.92). Cross-site predictions were surprisingly high, with predictive accuracies within a similar range (RR-BLUP: 0.79-0.92, GRR: 0.78-0.91). Height at 12 years was deemed the earliest acceptable age at which accurate predictions can be made concerning future height (age-age) and wood density (trait-trait). Using DEBVs reduced the accuracies of all cross-validation procedures dramatically, indicating that the models were tracking pedigree (family means), rather than marker-QTL LD. CONCLUSIONS: While GS models' prediction accuracies were high, the main driving force was the pedigree tracking rather than LD. It is likely that many more markers are needed to increase the chance of capturing the LD between causal genes and markers.

Single-Step BLUP with Varying Genotyping Effort in Open-Pollinated Picea glauca.

Maximization of genetic gain in forest tree breeding programs is contingent on the accuracy of the predicted breeding values and precision of the estimated genetic parameters. We investigated the effect of the combined use of contemporary pedigree information and genomic relatedness estimates on the accuracy of predicted breeding values and precision of estimated genetic parameters, as well as rankings of selection candidates, using single-step genomic evaluation (HBLUP). In this study, two traits with diverse heritabilities [tree height (HT) and wood density (WD)] were assessed at various levels of family genotyping efforts (0, 25, 50, 75, and 100%) from a population of white spruce (Picea glauca) consisting of 1694 trees from 214 open-pollinated families, representing 43 provenances in Québec, Canada. The results revealed that HBLUP bivariate analysis is effective in reducing the known bias in heritability estimates of open-pollinated populations, as it exposes hidden relatedness, potential pedigree errors, and inbreeding. The addition of genomic information in the analysis considerably improved the accuracy in breeding value estimates by accounting for both Mendelian sampling and historical coancestry that were not captured by the contemporary pedigree alone. Increasing family genotyping efforts were associated with continuous improvement in model fit, precision of genetic parameters, and breeding value accuracy. Yet, improvements were observed even at minimal genotyping effort, indicating that even modest genotyping effort is effective in improving genetic evaluation. The combined utilization of both pedigree and genomic information may be a cost-effective approach to increase the accuracy of breeding values in forest tree breeding programs where shallow pedigrees and large testing populations are the norm.

Implementation of the Realized Genomic Relationship Matrix to Open-Pollinated White Spruce Family Testing for Disentangling Additive from Nonadditive Genetic Effects.

The open-pollinated (OP) family testing combines the simplest known progeny evaluation and quantitative genetics analyses as candidates' offspring are assumed to represent independent half-sib families. The accuracy of genetic parameter estimates is often questioned as the assumption of "half-sibling" in OP families may often be violated. We compared the pedigree- vs. marker-based genetic models by analysing 22-yr height and 30-yr wood density for 214 white spruce [Picea glauca (Moench) Voss] OP families represented by 1694 individuals growing on one site in Quebec, Canada. Assuming half-sibling, the pedigree-based model was limited to estimating the additive genetic variances which, in turn, were grossly overestimated as they were confounded by very minor dominance and major additive-by-additive epistatic genetic variances. In contrast, the implemented genomic pairwise realized relationship models allowed the disentanglement of additive from all nonadditive factors through genetic variance decomposition. The marker-based models produced more realistic narrow-sense heritability estimates and, for the first time, allowed estimating the dominance and epistatic genetic variances from OP testing. In addition, the genomic models showed better prediction accuracies compared to pedigree models and were able to predict individual breeding values for new individuals from untested families, which was not possible using the pedigree-based model. Clearly, the use of marker-based relationship approach is effective in estimating the quantitative genetic parameters of complex traits even under simple and shallow pedigree structure.

Prediction accuracies for growth and wood attributes of interior spruce in space using genotyping-by-sequencing.

BACKGROUND: Genomic selection (GS) in forestry can substantially reduce the length of breeding cycle and increase gain per unit time through early selection and greater selection intensity, particularly for traits of low heritability and late expression. Affordable next-generation sequencing technologies made it possible to genotype large numbers of trees at a reasonable cost. RESULTS: Genotyping-by-sequencing was used to genotype 1,126 Interior spruce trees representing 25 open-pollinated families planted over three sites in British Columbia, Canada. Four imputation algorithms were compared (mean value (MI), singular value decomposition (SVD), expectation maximization (EM), and a newly derived, family-based k-nearest neighbor (kNN-Fam)). Trees were phenotyped for several yield and wood attributes. Single- and multi-site GS prediction models were developed using the Ridge Regression Best Linear Unbiased Predictor (RR-BLUP) and the Generalized Ridge Regression (GRR) to test different assumption about trait architecture. Finally, using PCA, multi-trait GS prediction models were developed. The EM and kNN-Fam imputation methods were superior for 30 and 60% missing data, respectively. The RR-BLUP GS prediction model produced better accuracies than the GRR indicating that the genetic architecture for these traits is complex. GS prediction accuracies for multi-site were high and better than those of single-sites while multi-site predictability produced the lowest accuracies reflecting type-b genetic correlations and deemed unreliable. The incorporation of genomic information in quantitative genetics analyses produced more realistic heritability estimates as half-sib pedigree tended to inflate the additive genetic variance and subsequently both heritability and gain estimates. Principle component scores as representatives of multi-trait GS prediction models produced surprising results where negatively correlated traits could be concurrently selected for using PCA2 and PCA3. CONCLUSIONS: The application of GS to open-pollinated family testing, the simplest form of tree improvement evaluation methods, was proven to be effective. Prediction accuracies obtained for all traits greatly support the integration of GS in tree breeding. While the within-site GS prediction accuracies were high, the results clearly indicate that single-site GS models ability to predict other sites are unreliable supporting the utilization of multi-site approach. Principle component scores provided an opportunity for the concurrent selection of traits with different phenotypic optima.