The set of the relative traits had less co-linearities (Fig. \ref{154230}b), and there were several strong trade-offs within the detected trait dimensions (Fig. \ref{553668}b). The first two, co-dominant PCArel axes were: crown proportions (PCrel1, 38.5% variance explained ) and tree slenderness (PCrel2, 27.8% variance explained). Crown transparency (PCrel3, 13.5% variance explained) clearly formed a separate axis. The first three PCArel axes jointly explained 79.7% of the total inertia. Trees with high crown density index (DBH_CA) had shorter and less costly crowns, while trees with slender stems (high H:DBH ratio) had narrower crowns (low CW_H, and low CW_CL). The variation in PCArel was mainly driven by trees with high crown density index and with wide-and-short crowns, but also by the stouter trees (Fig. \ref{553668}b).