Alicia,
EDG is organized around asset collections. There are different pre-built asset collection types. And you can define your own.
Taxonomy is one of the pre-built asset collection type. It is used to govern assets of a type skos:Concept. Each time you create a taxonomy, SKOS will be included into it. Ontology is also an asset collection type. It is used to govern assets of type owl:Class (or rdfs:Class), properties, etc. It could also include instances of classes - although best practice is to separate them.
Many EDG capabilities are common across all asset collection types, but some are specific to a type of collection. For example, certain imports may be specific. EDG offers import of MultiTess and this importer is specific to taxonomies as it converts data in the source files into SKOS concepts. Similarly, a DDL import is specific to the data asset collections as it converts information in the DDL file into resources that are instances of edg:Database, etc. classes.
Another feature specific to taxonomies is that there is a specialized editor application that is optimized for creating and working with SKOS concepts. With this, my first recommendation for you would be to use taxonomies, but the final recommendation may change after I see your datasets. Taxonomies and ontologies can and often do work together. For example, if you have defined any custom (not SKOS) classes and properties, you may put them into an ontology and have only instances in the taxonomy.
Regards,
Irene Polikoff