Species that do not occur in Florida / iNaturalist suggested species problem
The following species either do not occur in Florida, they are uncommon in Florida, or I'm unsure of their status. However, these species totaled over 1600 observations, which shows there is a serious problem with iNaturalist suggesting non-FL species and users selecting them. In some cases, the CV may be trained incorrectly.
() = number of FL observations in iNaturalist
Cladonia coniocraea - Common Powderhorn (139)
Flavoparmelia caperata - Common Greenshield Lichen (was 600, currently 241 after making an effort to sort these)
Flavopunctelia flaventior - Speckled Greenshield (48)
Flavopunctelia soredica - Powder-edge Speckled Greenshield (67)
Lecanora thysanophora - Mapledust Lichen (45)
Parmelia sulcata - Shield Lichen (52)
Phlyctis argena - Whitewash Lichen (64)
Ramalina farinacea - Farinose Cartilage Lichen (119)
Usnea florida - Florida Beard Lichen (12)
Usnea hirta - Bristly Beard Lichen (was 500+, now 0)
Comentarios
@alice_herden
@chaseyb
@evanaturalist
@jaykeller
@mhodges1957
@susanhewitt
Thanks for taking this on. Lichenology is complex and seemingly in its early stages of development and/or resources are relatively scant, so identification is often a challenge. Upon reviewing some of these across users (including some of mine), it seems like in some cases the iNaturalist AI suggests IDs that are quite close but perhaps not exact, which is extremely common. For example, after hundreds or even thousands of observations submitted for specific species of Euphorbia, the iNat AI still cannot properly ID them, even when it is clear to a trained (human) observer as to the identity at a glance. It just is what it is.
I also feel the ranges of some of these may not be 100% understood, and I would caution us from specific ID based PURELY on range. For example, on your list, Lichenportal shows that Parmelia sulcata has at least one record in Gainesville, FL. Another potential example is Ramalina farinacea, which has records in far-flung places such as Hawaii, the Galapagos, Uruguay, South Africa, etc (!!), so is conceivable it could be in Florida. Usnea florida has many records in Florida. Usnea hirta has at least one accepted record in the Orlando area.
https://lichenportal.org/cnalh/collections/map/index.php
Looks like a good list to me, Steve, though likely incomplete.
We have not accepted any records of Cladonia coniocraea in Georgia, and Harris (1995, More Florida Lichens) didn't include the species in his key. There may even be no reliable records of C. ochrochlora in the state (according to Harris 1995), the name Lendemer prefers over the former. I think you can dismiss any C. coniocraea observations from Florida, but please tag me if you run across an observation of a powderhorn Cladonia with big cupped squamules at the base; it would be interesting to find a good report of C. ochrochlora there.
Usnea florida is an Old-World species that probably doesn't occur in North America, according to Esslingerand is not named for the state of Florida; the inane common name should be removed entirely because it's causing confusion for observers.
Yours,
Mal
Jay, with all respect, any iNat observation (or other report) of Usnea hirta, Usnea florida, any Parmelia, and Ramalina farinacea can be summarily dismissed for Florida. I'd lay money on those specimens having been misidentified, given my experience with what occurs in Georgia and Florida and the many museum specimens I've found misidentified. Please tag me on any observation you think provides evidence to the contrary; I promise to keep an open mind. Or write if you find a specimen at NY of any of these species mentioned, for anywhere near Florida.
The problem is that it would take a serious lichenologist, using tools such as spot tests, dissection, microscopy, and thin-layer chromatography, to identify such extreme disjunctions, and they would be unlikely to share such extraordinary data on iNat before using it for a publication.
Joanne INatobserver takes a photo of something and agrees with what is suggested; she has no idea the suggestion is misguided. Her classmates agree with her observations so that she gets credit for the class assignment, or disreputable people trying to build up numbers in the identification hierarchy agree with observation IDs even when they have no experience with or knowledge of a species. INat starts creating a false range for the species based on these naive or corrupt observations/identifications.
Mal
@jaykeller @mhodges1957 Thanks for the discussion, Jay and Malcolm!
I agree that we may not be able to fix the AI, I think it would help if at least iNat doesn't suggest that the species "occurs nearby" with hundreds of other observations. I'm working through observations now, and as Malcolm pointed out, most of the misidentified lichens are unidentifiable photos from class assignments. It has been a real chore to sort through them.
I think some of the herbarium specimens that are shown in Lichen Portal are misidentified specimens, so I don't think we can rely on what appear to be uncommon or disjunct populations based on Lichen Portal maps. I have been using Brodo et al., the Rosentreter Florida key, the Georgia and Subtropical Florida websites and other sources. I understand that some range maps in Brodo et al. aren't accurate either, so I try to keep an open mind and confirm potential ranges with other sources.
And yes - this list above is nowhere near complete, but it's some of the worst offenders. If there are other species that should be added, let me know and we can keep a running list. Thanks!
Thanks for the information
edit: it took days, but I sorted through all of the Flavoparmelia records. Most were student submissions and could not be identified. They often were photos of tree trunks taken from about 6 feet away, and a couple submissions were photos of tree trunks that had no lichens on them!
Añade un comentario