A library of nine aromadendrane-type sesquiterpenes (1-9), including eight new natural products (1-5 and 7-9), was isolated from Landolphia dulcis var. barteri along with a previously described cadinane derivative (10) and a new muurolane derivative (11). The structures of all compounds were established by means of NMR methods including COSY, NOESY, HSQC, and HMBC experiments, supported by HRMS and optical rotation data. Virtual characterization of the aromadendrane library (1-9) was performed using chemoinformatics tools. 3D molecular fields were calculated with the GRID program using low-energy structures obtained with the MMFF force field. VolSurf descriptors were calculated from the GRID maps and subsequently analyzed by multivariate statistics. The analysis disclosed the presence of a common motif for possible interactions of the aromadendranes with a putative target receptor. At the same time, a considerable chemical diversity within the library was disclosed, despite a close biosynthetic relationship of its members. The results can be interpreted in terms of evolutionary optimization of structures of secondary metabolites for interaction with macromolecular targets and are of interest in terms of assessment of potential "drug-likeness" of natural products.