Advancing Black Spruce (Picea Mariana) breeding through genomic selection: a comparative analysis of models using pedigree and genomic marker information.

Abstract

Long generation times in forest trees constrain the pace of genetic improvement necessary to sustain productivity under climate change. Genomic selection offers a promising approach to accelerate breeding gains in long-lived species like black spruce (Picea mariana). In this study, we evaluated genomic selection models using data from a long-term half-sib progeny trial in the Lake Nipigon West breeding zone of northern Ontario. A subset of 1194 trees from 70 families was genotyped using two platforms: a SNP array (16,217 SNPs) and a genotyping-by-sequencing approach based on RADseq (10,626 SNPs). Growth traits—including height, diameter at breast height (DBH), growth rate, and volume—were measured at multiple time points. We compared three animal models differing in their relationship matrices: pedigree-based (ABLUP), genomic-based (GBLUP), and a hybrid model integrating both pedigree and genomic information (HBLUP). The HBLUP model consistently produced the most accurate heritability estimates and the smallest prediction errors for key growth traits such as volume and DBH, likely due to its ability to incorporate both genotyped and ungenotyped individuals. Genomic models (GBLUP and HBLUP) outperformed pedigree-based models, highlighting the value of genomic information for improving selection efficiency. While early height has traditionally served as a proxy for long-term growth, its low heritability in this study suggests caution in its use as a sole selection criterion. Instead, height may be better incorporated as part of multi-trait selection indices to capture its environmental responsiveness, particularly during early testing stages. Among genotyping platforms, SNP chips consistently outperformed RADseq, indicating their preference when budget allows, though RADseq remains a cost-effective alternative that could benefit from complementary strategies such as imputation or hybrid integration. Overall, our findings support the practical integration of genomic selection into black spruce breeding programs. By aligning genotyping strategies and model choice with specific trait characteristics and breeding goals, programs can accelerate genetic gain, reduce breeding cycle time, and enhance forest adaptability under future environmental challenges.