By mapping entities one can map abstract "classes" and not only properties of classes. For specimens for example instead of only mapping the catalog number and a taxon name, one wold also map the entity/idea of an specimen:

Simple mapping:

 <Specimen>        
   <CatalogNumber>  - abcd:/DataSets/.../Unit/UnitId
   <Name>           - abcd:/DataSets/.../Unit/Identifications/.../FullScientificNameString

Entity mapping:

 <Specimen>         - abcd:/DataSets/.../Unit
   <CatalogNumber>  - abcd:/DataSets/.../Unit/UnitId
   <Name>           - abcd:/DataSets/.../Unit/Identifications/.../FullScientificNameString

This would help the wrapper software to decide which RepNode corresponds to which conceptual schema entity. Thus it would probably solve the EmptyRepeatableRegionProblem