Information integration and retrieval have been important problems for many information systems --- it is hard to combine new information with any other piece of information we may already possess, and to make them both available for application queries. Many ontology-based applications are still cautious to integrate and retrieve information from natural language (NL) documents, but rather from structured or semi-structured sources. In this paper, we choose the botanical domain to investigate how to use ontologies to facilitate integrating and querying information on parallel leaf shape descriptions from NL documents. Our approach takes advantages of ontologies to precisely represent the semantics in shape descriptions, to integrates parallel ones according to their semantic distances, and to answer shape-related species identification queries. From this highly specialised domain, we learn a set of more general methodological rules, which could be useful in other domains.