More GBIF questions (was: ITIS)

[Roderic Page]

>I'm surprised that Rod Page's "definition of a classification is a tree" got
>by, or maybe I missed something. Check the dictionary.
>


A classification, in the sense of a list of accepted names arranged
in a Linnaean hierarchy can be represented by a tree, that is, graph
in which all nodes (representing names) are connected, and each node
has a single parent (for example, the parent of a species is its
genus, the parent of a genus is a subfamily, and so on). This is how
databases such as ITIS and NCBI model their classifications. This is
what I meant by the  "definition of a classification is a tree".

My point was that, even though ITIS models their classification as a
tree of accepted names, in fact it is not because it is broken. For
example, the fly tribe Stomatosematidi (tsn 124358) is accepted by
ITIS, but its parent taxon is listed as the subfamily Lasiopteridi
(tsn 123426), which ITIS treats as a synonym of the subfamily
Cecidomyiinae (tsn 123421). The parent of Stomatosematidi is not an
accepted name, which means that the tribe Stomatosematidi is not
connected to the rest of the ITIS animal classification, and so we no
longer have a tree.

We've found a number of these errors. The reason I brought this up is
that the notion of a classification as a tree is crucial to using
classifications to answer questions such as "list all insect
species." To do this, we take a classification of animals (a tree),
locate "Insecta", then visit every species that is a "descendant" of
the node labelled "Insecta". Now, if the tree is broken, we are not
going to be able to list all the insect species (for example, we will
never visit members of the tribe Stomatosematidi).

Regards

Rod



--
--------------------------------------------------------
Dr Roderic D. M. Page
Professor in Taxonomy
DEEB, IBLS
Graham Kerr Building
University of Glasgow
Glasgow G12 8QP
United Kingdom


Phone: +44 141 330 4778
Fax:   +44 141 330 2792
email: r.page at bio.gla.ac.uk
http://taxonomy.zoology.gla.ac.uk/rod/rod.html