Multivariate models improve accuracy of genomic prediction for spring frost tolerance in Norway spruce

Warming spring temperatures increase the risk of frost damage to emerging Norway spruce (Picea abies) buds by advancing their spring phenology and increasing the frost event frequency. We present a field-based electrolyte leakage assay to assess basal frost tolerance in newly emerging buds. Using this assay, we estimated genetic parameters and tested the effectiveness of multivariate genomic selection (GS) models integrating frost tolerance, bud burst phenology, and height growth. Multivariate models significantly improved frost tolerance prediction accuracy, particularly when incorporating bud burst data, due to a strong genetic correlation between traits (r ≈ -0.63) and high heritability of bud burst (h² ≈ 0.60). The observed genetic correlations suggest that early-flushing genotypes exhibit higher tolerance to spring frost. Our findings underscore the importance of basal frost tolerance as a complementary trait to traditional phenological frost-avoidance strategies. Additionally, we emphasize that early-stage bud burst assessments in controlled environments can accelerate genomic predictions, overcoming the limitations imposed by long growth cycles. Integrating multi-trait genomic prediction models optimized with bud burst as an assisting trait and optimized model parameters enhances prediction accuracy of spring frost tolerance and supports the development of climate-resilient breeding strategies in Norway spruce.