Oh boy! Another taxon swap to learn!

Since this change a year ago, there are still thousands of misidentified observations listed as Neltuma glandulosa from Western US states that should be Neltuma odorata. In my opinion, it was inappropriate to digitally automate the transition from Prosopis glandulosa to Neltuma glandulosa because the treatment split P. glandulosa varieties out into multiple species. The N. glandulosa concept is not a direct 1:1 equivalent of the P. glandulosa concept. Observations of P. glandulosa var. torreyana that were correctly identified as P. glandulosa, but not identified to the varietal level were incorrectly lumped into N. glandulosa.

This swap needed a regional modification at minimum. According to POWO, the authors of the paper, and other sources like Jepson; N. glandulosa does not occur in California, Nevada, Utah, or Arizona; but this split transferred over 1,000 observations in these states into N. glandulosa. Now, that number is still rapidly growing because the iNat's computer vision model is automatically suggesting N. glandulosa to people in those states telling them that there are "Research Grade" observations nearby. There are not; or shouldn't have ever been. A good start to fixing this would be dumping every observation of N. glandulosa from those states into N. odorata. Observations within the areas where N. glandulosa and N. odorata overlap (New Mexico, Texas, and Mexico) still need to be evaluated on a case by case basis.

To solve the computer vision suggestion issue, it would be nice if every observation impacted by this swap could be removed from Research Grade until one or two individuals manually confirm IDs, at least in the areas of geographic overlap. This automated taxon swap should not have left those observations of P. glandulosa/N. glandulosa as "Research Grade" when there was a known issue with the synonymy identified during discussions prior to the swap. This would probably be good practice for most taxon swaps; but it is especially important in cases like these when significant changes are made to generic, specific, and varietal concepts at the same time without precise synonymy. The predictable errors here are the direct result of the taxon swap, not of actual peer review. Having them listed as "Research Grade" as if two identifiers actually agreed on applying this particular name is not the case.

This swap has no doubt confused countless people. It is spreading botanical misinformation across the west for an important dominant species. It's greatly reduced the quality and accuracy of the iNat computer vision model for these taxa. I think something should be done, but given the huge number of observations to go through (~28k N. glandulosa), it seems like the task might take more time to complete than waiting for someone to redo the Mimosoideae treatment and change all the names again.

@kevinfaccenda @aguilita @salvadorcabrera21 @joshua_tx @nathantaylor @stevejones @txlorax @sambiology @aidencampos @davidia @alexbinck @loarie

Might we benefit if the AI included subspecies and varieties, rather than just dealing with species level IDs?

Actually, as I've been going through more observations since my comment, I realize it isn't as simple as that. Due to the popularity of Prosopis glandulosa var. glandulosa (and it's cultivars) as a horticultural species, there are valid or semi-valid observations of the new N. glandulosa concept within those ranges. Some of these should be listed as cultivated, but others are escaping and naturalizing around populated areas. It seems the NV and UT observations have been pretty well separated (thanks in large part to @escalanteslim). CA and AZ seem to have many issues still, but it's probably still best to do those manually as it's commonly planted around Phoenix and Los Angeles.

Presumably, N. glandulosa var. glandulosa (or N. glandulosa s.s.) is the eastern form more likely to take on a tree-like form with var. torreyana (N. odorata) the western form more likely to form shrubs. There are leaf characteristics that trend in different directions also. However, it's all intermixed in New Mexico and West and Central Texas (Oklahoma plants form shrubby trees too). To quote Powell and Worthington (2018), "Forms seemingly intermediate between var. torreyana and var. glandulosa are abundant in the Trans-Pecos." For this concept to work, we have to hypothesize a huge hybrid zone across these states. If you want some proof of that, try to ID across the full range of West Texas using more than one key characteristic (habit, leaves, and spines per node are the three main characters). If someone can make sense of that mess and teach me how to, I'll freely admit to making a mountain out of a molehill, but otherwise, I remain highly skeptical of this change.

This is the primary source POWO is referencing: https://phytokeys.pensoft.net/article/75379/download/pdf/

It would probably be best to email the authors about this since they provide no justification in their text.

Yeah, unfortunately, I probably should email the authors at this point. I really want to know if they considered the Trans-Pecos material when they made their decision. However, unless someone decides to challenge or change their decision in print, it doesn't really help us (I'm very tempted to do so, but it's more work than I have time for at the moment). I looked into it and Neltuma glandulosa var. torreyana isn't even available as a combination. At the end of the day, it looks like using Neltuma forces us to use N. odorata unless someone publishes the new combination.

If this is the case, the complex idea seems like the only way to preserve the appropriate level of continuity between glandulosa and odorata/torreyana. We would probably have to change everything in N. glandulosa to complex N. glandulosa where the two taxa overlap which might be undoing a lot of work.