iDigBio type data, DwC-A, ~1,000,000 TB records. Very similar to the GBIF type records. The full resource for this will probably be a couple hundred thousand taxa
Discoverlife maps, connector, ~600,000 maps connector has run, set to force harvest
NHM London type data , DwC-A, 300k TB records, but 2.4M rows in the measurementOrFact file, counting rows for metadata
German wikipedia, DwC-A, large, good sample of weird text objects with images embedded, etc. hold off, we might do this via wikidata instead
Flora of Zimbabwe, "currently fine, 10657 taxa, 17932 images, 13996 articles, xml resource file" The names has a problem where some names are just "x" (the times symbol, actually).
Pending Resources