[Taxacom] GBIF progress

John Irish jirish at mweb.com.na
Tue Jan 6 01:16:50 CST 2009


I think GBIF is great, but it is suffering from a bad case of GIGO. It 
may be better to first fix what they have before piling on ever more 
records.

A lot of the current records are not georeferenced (e.g. Namibia: 224000 
records, only 17000, or 13%, georeferenced). But it gets worse:

A lot of the 'georeferenced' records are crap. I recently did a GBIF 
search on a rectangle in inland southern Namibia, and turned up 6840 
records. After examination, only 192 proved to be useful. The rest were 
from stated localities in other countries in different hemispheres, but 
with coordinates in Namibia; from parts of Namibia nowhere near my 
search rectangle, but with coordinates inside it; marine taxa, but my 
search rectangle was not in the sea; records identified to genus, family 
or order only, and not very useful for species-level analysis; records 
of species that are well-known and definitely do not occur in the area 
(misidentifications?); apparent nomina nuda - names that I have not been 
able to find anywhere else; fossils - not very useful for what GBIF is 
usually used for.

GBIF's strength, and weakness, is that it indiscriminately serves what 
museums offer. Fix the museum collection databases and GBIF's *useful* 
holdings will increase. Simply add more collections and the garbage will 
increase to the point where it is no longer worth the effort to use GBIF 
(at a rate of 1 useful record per 1000, as above, it may already be).

Fixing collection databases is a tall order, I know. About 2 years ago, 
I asked for copies of Namibian data holdings from a number of large 
mammal collections that are served both on MaNIS and GBIF. The 
georeferencing was as usual, but I spent time to fix it and gave all 
back to the curators involved. They thanked me very politely, but when I 
did the GBIF search mentioned above, there were some of the same 
mistakes still unchanged. As an ex-curator, I know full well that 
museums are understaffed and overworked, so this is no surprise.

So, yes! Get more data into GBIF. BUT: make sure it is properly 
georeferenced. And figure out some way to also fix the legacy data that 
is already in there. (And as an aside: have the fixing done by someone 
who lives in the country and speaks the language - georeferencing from a 
distance is what caused the problem in the first place).

John

-- 

Dr. John Irish
Gobabeb Training and Research Centre - EON Co-ordinator
P.O. Box 953, Walvis Bay, Namibia
Gobabeb Web Site: http://www.gobabebtrc.org/
Namibia Biodiversity Database Web Site: http://www.biodiversity.org.na/




More information about the Taxacom mailing list