4 DISCUSSION

We tested hypotheses related to both general limitations (B,C) in rowan performance, due to harsh climate (B) or mechanical vulnerability (C), and specific (local) limitations (A,D), due to wind-induced stress (A) or low light (D). Clearly, the hypothesis C was supported, as the subalpine rowans were generally stout (with maximal H:DBH ratio around 1 and median of 0.6); moreover, the trees had long crowns (with median CL_H of 0.73), and thus slender and top-heavy trees were extremely exceptional, disregarding canopy openness. The hypothesis B stating a general size-limitation received a limited support, as many of the studied rowans had surprisingly large crowns (widths up to 7.7 m and lengths up to 11.8 m); however, given the large crowns, the trees were rather short (heights up to 14.5 m), which supports the size-limitation hypothesis (B), and also provides a further support for the mechanical vulnerability hypothesis (C). While C was better supported than B, we may already suppose that strong wind was more limiting here for rowans, than shading from spruce canopy. The hypothesis A received a considerable support, while high tree slenderness was clearly limiting to rowan fruit production under more open canopies (Fig. \ref{387654}c); nonetheless, there was little vice versa evidence that more slender trees under denser canopies produce more fruit (but see example trees a,b vs. g,h in Fig. \ref{543683}). Finally, the hypothesis D received a mixed support: better for D1 (wide-and-shallow crowns as a survival strategy under low light) and poorer for D2, while trees with high CW:CL ratio produced little fruit (see the example tree Fig. \ref{543683}e). There were several reproductive architectural strategies: increased CL metabolic cost (in the open: predominantly wide crowns, but with considerable crown lengths, and some asymmetry), increased CW metabolic costs (under more closed canopy: predominantly long crowns, but not very narrow, and with little asymmetry), or intermediate (cf. Fig. \ref{553668}). However, importantly, there were little trees with both high CW_cost and CL_cost (also further supporting the size-limitation hypothesis). Overall, trees with the fullest and the most costly crowns were frequent among the most reproductive, in all cases large relative crown length facilitated fruit production.