All, or almost all of the Crataegus douglasii IDs south of the Great Lakes and east of the Rockies are wrong

This is a difficult genus, but this is a species of cool areas (Pacific Northwest, Rockies, northern Great Lakes). It seems to be a very common ID applied to a very wide variety of Crataegus in the SE. Some of the IDs are even Research Grade. Not all of the C. douglasii IDs are even Crataegus. One is Calycanthus, and one is Pyrus. The result is that the range is spectacularly distorted.

1 Like

pretty much all crataegus IDs anywhere on anything are wrong :( It seems to be just a few species with a bunch of variants described to species too. I don’t even try them unless i have to for work.

It’s probably because it’s a top autosuggestion (I just tried it on an out of range obs and that did come up as first choice).

The same thing happens with Houstonia for example, where the less-common choice comes up as the top autosuggest for a very common one that has a similar looking flower. (How I became a Houstonia “expert” for my region- I went through all prior obs to try to correct those after getting burned on my first observation. :) )

3 Likes

I wouldn’t go that far. Quite a few Crataegus are easy enough to ID to species, and the actual biology involves both distinct entities that reproduce sexually, as well as entities that are mostly apomictic, with occasional introgression impacting trait variation in both sets. If we apply a species concept to the group that’s consistent to the species concept we’ve applied to other with similar reproductive complications, like Asplenium etc, it’s necessary to describe quite a few species to cover the set of distinct biological entities.

Overzealous splitters have previously put names on inconsistent morphological variations within those entities, so there’s a bunch that aren’t valid, but heavy lumping doesn’t really fit our best knowledge to date either (Amelanchier’s in the same boat).

Regarding the OP this is a similar problem we see with many taxa where well-meaning nonexpert users find an AI suggestion plausible and just go with it. I spend time on Houstonia IDing as H. Canadensis doesn’t even come up on the autosuggester but many H. longifolia (including research grade) obs in certain areas are canadensis or can’t be distinguished from available photos. A shocking number of H. purpurea also slip through as H. longifolia and those often get multiple confirming IDs, which probably sets up a feedback loop mis-training the computer vision. Kinks to be worked out.

5 Likes

Sort of funny coincidence that we’re both on about Houstonia in our responses (you posted while I was typing) out of all the possible plants in the world!

4 Likes

So funny! My issue was the earlier ones, H. caerulea vs serpyllifolia, and then I kept finding pusilla issues while I was at it. :) I’m still a noob on the later ones.

That part of the genus is the part I don’t know or deal with. Divide and conquer!

2 Likes

Well, difficult is relative. My meaning is that a lot of people that think Carex is easy think Crataegus is difficult, and those that aren’t intimidated by taking to a key to figure out a hawthorn are still often confronted with needing both stamen counts or color as well as fruit color–inconvenience. Oh, and winter ID IS dificult relative to other larger, woody plants, most of which can be reliably IDed to species based on bark, buds, growth form, etc.

Surely, some over-split Crataegus (and Rubus), but the opposite is true of others. I know one accomplished botanist that only ever called out two species in SE WI (C. mollis and C. crus-galli). Reality is in-between. …and there are likely still real new taxa hidden in all the mess.

As for the AI, it’d be nice if it could steer away from suggesting taxa that have no vouchered specimens within 500 miles.

2 Likes

Not so. There are reasonable interpretations of Crataegus that lead to reliable and repeatible ID of consistently different taxa, but in the Midwest (my area) those tend to be resources that call out 15-30 (30 probably being a bit high) species vs. 50+ (the name slips my mind, but the Milwaukee Public Museum has the collection of a Crataegus expert from early in the 20th Century that separated the Genus into a boat load of taxa, some based only on single specimens or localities…clearly overly narrow). Some species can be reliably identified any time and leaf (e.g. Crataegus crus-galli, which some eroniously attribute to anything with thorns). Other Crataegus create fits by requiring stamen color and fruit color to run through the key. Based on characteristics like stamen counts, variation in fruit morphology, and some of the extremes in leaf morphology, there are clearly at least in the double digits of species in most of the Midwest. There are ecological differences too. C. crus-galli is almost always in uplands. Some are reliably in floodplains (either upland or wetland). Others are more reliably in wetlands.

2 Likes

I’m not saying they are ALL bad species. 15-30 is probably reasonable. I don’t see them that often but did see one in a wetland today with no reproductive material. Probably won’t id that.

2 Likes