1 INTRODUCTION
Trees realize multi-objective optimization to fulfill growth and reproduction \cite{Farnsworth_1995}. These are governed by several trade-offs among light and hydraulic efficiencies, storage capacity, and mechanical support functions \cite{Pratt_2016}. In resource-limited environments, or in otherwise stressful conditions, these trade-offs become acute and crucial for the overall tree performance \cite{Rasmussen_2004,Dobbertin_2005,Hacket_Pain_2018}. Whole tree architecture, at any time of its development, is the effect of the past interactions between the internal growth processes and exogenous constraints exerted by the environment \cite{Barth_l_my_2007}. Thanks to the rapid progress in close-range remote sensing methods to quantify tree architecture \cite{Barbeito_2017}, many recent studies addressed the problem of encoding and extracting the information about tree functional fitness from the structural information \cite{Verbeeck_2019,Nunes_2023,McNeil_2023}. Much focus was on the economically relevant, abundant large-statured trees, either mature or saplings, growing in good conditions \cite{McNeil_2023}. The small-statured trees or shrubs attained much less attention \cite{Charles_Dominique_2012}, although these may provide large fruit crops \cite{Greene_1994}, and while often being both pioneer and stress-tolerant \cite{Brzeziecki_1994}, form stable elements of disturbance-prone ecosystems, "already there" at gap formation \cite{_ywiec_2007}. In this paper we analyze a unique data from detailed tree measurements of a fleshy-fruited phenotypically plastic tree species, growing near the upper distribution limit, to test the idea that similar tree architectures may result in contrasting performances, represented by long-term fruit yield, depending on the ecological context and canopy openness gradient.