Hi Nicole,
A simple example is your shapefile of Lausanne. In that shapefile there are two individual polygons. However, they really represent one "reference space" (a reference space is a specific area/point that we want to link to something, e.g. flows or stocks). However, there are other examples, like a database with trees or with biodiversity reservoirs in the city like this particular file. In these cases, we don't really want to create a unique "space" for each tree or each biodiversity reservoir, because we won't link those individual elements to flows or stocks. Rather, we want to take the entirety of that shapefile and use that information (containing details of all trees or all biodiversity layers) and put that on the map and contrast that with other infrastructure or flows). So in these cases we don't want the system to create all these individual spaces that make everything heavier and more complex, but we just want to treat the systems as a single layer without further subdivisions.
Does that make sense?