Stand characteristic models for genetically improved and unimproved silver birch plantations in southern and central Finland

Abstract

The genetic gain from improved seedlings in silver birch (Betula pendula) plantations in Finland is receiving increasing attention, yet quantitative knowledge of their growth characteristics remains limited. This study developed stand characteristic models for silver birch plantations to predict growth differences between genetically improved and unimproved stands. Response variables were basal area–weighted mean height (HW, m), mean diameter (DW, cm), and stand basal area (G, m² ha−¹). Data comprised 2,183 plot observations from thinning, spacing, and breeding trials, and commercial forests across Finland (1,954 unimproved, 229 improved). Modelling employed a multivariate mixed-effects approach using seemingly unrelated regression with a random site intercept. Predictor variables included site index, stand age, number of trees, a dummy for former agricultural land, and thinning-related variables. The genetic improvement dummy variable was statistically significant in the DW and G models (p < 0.0001), reflecting superior stand yield. HW values in improved stands were higher than in unimproved stands, based on greater dominant height and site index for the improved stands as indicated by simulation results. Model fit statistics and simulations demonstrated logical predictive behaviour and reliability. The models can be used in forest decision-support systems for managing silver birch plantations, particularly those established with genetically improved material.