Production carries 12 species of Salix, a genus with 350-500

[lambda2]

Production carries 12 species of Salix. The genus has 350–500 accepted worldwide.

Found while dry-running the TRY import against production: 786 of 10,815 species in a 50,000-measurement sample failed to resolve, and the list is dominated by willows — Salix viminalis, caprea, cinerea, alpina, appendiculata, aurita, acutifolia, retusa, daphnoides, eleagnos, reticulata. Every one of the most common European willows.

They are not resolution failures. Checked directly: none of them exists as a species or as a synonym.

Salix viminalis | species=- | synonyme=-
Salix caprea    | species=- | synonyme=-
Salix cinerea   | species=- | synonyme=-

It is an outlier, not the general coverage gap

That was my first assumption and it is wrong. Neighbouring genera are fine:

Genus	Species in production
Carex	2,915
Quercus	856
Festuca	729
Rosa	546
Acer	363
Pinus	292
Salix	12

The twelve are all Salix alba variants plus S. arbutifolia: Salix alba, Salix alba f. veniaminii, Salix alba subsp. alba, Salix alba subsp. micans, Salix arbutifolia…

Not a duplicate-genus artefact either — there is exactly one Genus row named Salix (id 16547, slug salix), and all twelve hang off it. 54 synonyms carry a Salix name.

Why it matters beyond willows

The TRY import resolves by name. Any trait data for a species we do not carry is dropped with a report and never ingested — so this gap silently costs data on every import, not just this one. In the 50k sample alone, 786 species' worth of measurements went nowhere.

It also means the 7.3% unresolved rate on that dry run is not a quality signal about TRY or about the resolver. It is a measurement of our own taxonomy.

What this needs

• Find out what happened to Salix specifically — a failed WCVP pass for that genus, a purge that took more than it should, an ingest that errored mid-genus. The synonym purge (25,385 rows) is the obvious suspect to rule in or out first.
• Then ask whether other genera are in the same state. Sampling seven genera found one bad; a full pass comparing per-genus counts against WCVP would say how many more there are, and that is the honest version of Measure the taxonomic coverage gap, and settle whether hybrids are excluded on purpose #359's "measure the coverage gap".

Related: #359 (taxonomic coverage gap, and whether hybrids are excluded on purpose — the same dry run also failed to resolve Iris x germanica, Amelanchier x lamarckii and Helianthus x laetiflorus, which is that issue's hybrid question showing up in practice).