Predicting the number, death, and self-pruning of branches in Scots pine

Branchiness of Scots pine (Pinus sylvestris L.) was studied in 19 thinning experiments in southern and central Finland. Data were collected from 229 trees growing on sites of different fertility, stand densities, ages, and canopy positions. The data were used to construct models for (i) the number of new branches on the stem apex, (ii) the probability of a branch being alive, and (iii) the proportion of the actual number of dead branches out of the predicted initial number of branches on the whorls below the crown base. Generalized variance component models were used to separate the stand-, plot-, and tree-level variation in the dependent variables. The number of new branches was closely connected to the height increment of the current year, slightly modified by the height/diameter at breast height ratio (h/DBH). The probability of a branch being alive was connected to its age, tree age, h/DBH, and its relative diameter within the branch whorl. Self-pruning of branches was faster on large trees with long crowns than on thin trees with short crowns. The indices describing the stand density and spatial arrangement of the neighbouring trees around the sample tree reduced the residuals only slightly.