Missouri Botanical Garden Open Conference Systems, TDWG 2013 ANNUAL CONFERENCE

Font Size: 
Miss the forest for the trees: bringing together multiple classifications
Heimo Rainer, Wolfgang Koller

Building: Grand Hotel Mediterraneo
Room: America del Nord (Theatre I)
Date: 2013-11-01 11:15 AM – 11:30 AM
Last modified: 2013-10-09

Abstract


A topic in active discussion in the Biodiversity Informatics Community is the question of taxonomic opinions and their computational implementation. Is it feasible to have multiple classifications for one and the same taxon or shall it be more desirable to have only one consolidated taxonomic tree.

One question that immediately arises: is there any comprehensive taxonomic treatment available at all? For some taxa only regional treatments are published, for others monographs are available. Working in a large botanical institution with a worldwide coverage of plants, algae, lichens and fungi quickly reveals the patchy publication record. Monographs are outdated concepts and very time consuming to create and thus comparatively rare. On the other hand regional treatments are quite amply produced because of their local, regional, or national importance, e.g. Flora of China, Flora of North America. Astonishingly for Europe, no modern treatment of all higher plants exists. How can this be overcome? We suggest as a basic idea, to link sources of taxonomic and floristic treatments on the basis of scientific names and all their synonyms, be they taxonomic or nomenclatural, but always keeping track of the provenance, citing the original work as well as the first place of publication and if possible also the type collections as a proof of the concept.

Work on the Flora of Austria as well as participation in the Global Plants Initiative led us to implement an integrated system for data capture along these lines. As part of an open data environment access to the classifications is provided through a JSON (JavaScript Object Notation) service architecture.

Storing data-trees within a relational database is a common task. However storing a large number with a high node count and overlapping paths leads to a complex data model. The actual challenge is maintaining an acceptable performance level. We have developed strategies for pre-rendering those trees to guarantee faster response times. In addition it allows us to provide classification structures that will be static and are therefore citable.