Timothy, more of a mystery than you might think

I've been checking identifications of Timothy Grass (Phleum pratense) since mid-2019. The level of misidentification on iNaturalist is . . . interesting. My list is below. If the plant is not a grass, its family name is added in parentheses. Sorry about providing only scientific names.

Note: These misidentification have been corrected. Of course, there are no doubt others I haven't found (yet).

Plants identified as Phleum pratense. The first four species listed are very commonly misidentified as Phleum pratense on iNaturalist, and I've seen the others at least once. (Updated 2022)

Alopecurus pratensis
Alopecurus arundinaceus
Plantago lanceolata (Plantaginaceae)
Phalaris aquatica
Agastache sp. (Lamiaceae)
Agropyron cristatum
Agrostis sp.
Alopecurus aequalis
Alopecurus brachystachyus
Alopecurus geniculatus
Ambrosia sp., probaby A. artemsiifolia (Asteraceae)
Anemone cylindrica (Ranunculaceae)
Anthoxanthum odoratum
Apera interrupta
Holcus lanatus, immature
Baptisia sp., immature (Fabaceae)
Betula sp., catkin (Betulaceae)
Calmagrostis arundinacea
Carex acutiformis (Cyperaceae)
Carex barbarae (Cyperaceae)
Carex heteroneura (Cyperaceae)
Carex kelloggii (Cyperaceae)
Carex nebrascensis (Cyperaceae)
Carex obnupta (Cyperaceae)
Carex pendula (Cyperaceae)
Carex sect. Racemosae (Cyperaceae)
Carex sp. (Cyperaceae)
Celosia spicata (Amaranthaceae)
Chamaelirium luteum (Melanthiaceae)
Cynosurus cristatus
Dalea candida (Fabaceae)
Dalea purpurea (Fabaceae)
Dactylis glomerata, immature
Eleusine indica
Elymus repens
Elymus sp.
Gastridium phleoides
Hilaria mutica
Holcus lanatus
Hordeum brachyantherum
Hordeum pusillum
Hordeum sp.
Hyptidinae, maybe Hyptis mutabilis (Lamiaceae)
Hypochaeris radicata, in bud (Asteraceae)
Itea virginiana (Iteaceae, formerly Saxifragaceae)
Koeleria sp.
Lagurus ovatus
Lepidoptera (a moth caterpillar on a grass stem)
Liatris spp. (Asteraceae)
Luzula spp. (Juncaceae)
Melica transsilvanica
Muhlenbergia glomerata
Muhlenbergia ringens
Muhlenbergia sp.
Pennisetum glaucum (now in Cenchrus)
Pennisetum setacea (now in Cenchrus)
Phalaris angusta
Phalaris arundinacea
Phalaris caroliniana
Phalaris coerulescens
Phleum alpinum
Phleum arenarium
Pleum hirsutum
Phleum sp.
Plantago coronopus (Plantaginaceae)
Plantago media (Plantaginaceae)
Plantago patagonica (Plantaginaceae)
Poa arachnoidea
Polypogon monspeliensis
Populus tremula, catkin (Salicaceae)
Rostraria sp. (probably)
Salix sp., catkin (Salicaceae)
Secale cereale
Setaria faberi
Setaria pumila
Setaria viridis
Trifolium angustifolium (Fabaceae)
Triticum aestivum (club wheat, T. a. compactum)
Triticum aestivum (an awned wheat)
Turritis glabra (Brassicaceae)
Verbascum thapsus (Scrophulariaceae)
a blond middle school boy with freckles (likely correct, in one sense)

These have been misidentied as Phleum alpinum (in the broad sense):
Alopecurus arundinaceus
Betonica, maybe B. hirsuta (Lamiaceae)
Carex aterrima ssp. medwedewii (Cyperaceae)
Carex breweri (Cyperaceae)
Carex scopulorum scopulorum (Cyperaceae)
Carex spectabilis (Cyperaceae)
Carex section Ovales (Cyperaceae)
Cynosurus echinatus
Phyteuma nigrum (Campanulaceae)
Psilathera ovata
Trifolium pratense in fruit (Fabaceae)

In addition, a Typha minima observation was misidentied as Phleum (no species), corrected from the initial identification of Tenodera angustipennis, the Narrow-winged Mantis. Sometimes one almost wants to weep.

On the other hand, these names have been applied to what was actually Phleum pratense:
Acorus calamus (Acoraceae)
Agastache sp. (Lamiaceae)
Alopecurus pratensis
Ammophila breviligulata
Carex sp.
Koeleria macrantha
Phalaris canariensis
Phalaris aquatica
Toxicoscordion venenosa, in fruit (Melaniaceae, formerly Liliaceae)

Although I'm sure humans often misidentified Meadow Foxtail (Alopecurus pratensis) without computer assistance, I think many of these names originated as suggestions by the CV (computer vision program), which seemed to consider anything with a more or less cylindrical inflorescence made up of many little bits to be Timothy. Given the state of identifications for Timothy, this was understandable; the computer is trained on iNaturalist photos and if many of them are misidentified, errors result. I hope that recent corrections help the latest CV version to recognize Timothy more precisely. Of course, even if CV improves, we humans have to choose from among the CV suggestions and are also entirely capable of misidentifying Timothy independently.

Posted on August 09, 2021 06:15 PM by sedgequeen sedgequeen



Posted by bit over 2 years ago

It is impressive, in a negative way, isn't it!

Posted by sedgequeen over 2 years ago

have you tried to see what the app thinks plantago and some of the others are? In otherwords, I wonder how many of the bad ids are App suggestions.

Posted by bit over 2 years ago

Bit: Thinking back on this project, I think the computer contributed to many, but not all, of the identifications.

Posted by sedgequeen over 2 years ago

The APP can be helpful, but in ignarant hands it can be disastrous. That is why we are still needed and why people still need to learn something about id in shcool and not just rely on APPS.

Posted by bit over 2 years ago

I agree.

Posted by sedgequeen over 2 years ago

Yes, an impressive list! That said, not surprising to me, after seeing all of the computer vision, and human errors, in my years of spending a lot of time on iNaturalist.

Posted by stewartwechsler over 2 years ago

At least all these errors were corrected! (I often find that although "Research Grade" observations aren't all correct, they're correct a higher percentage of the time than "Needs ID" records.

Posted by sedgequeen over 2 years ago

I once estimated that "research grade" observations were correct 20% more of the time. That said, I might have been my more skeptical estimate. Could be something more like they are right at least 20% more of the time.

Posted by stewartwechsler over 2 years ago

I think it depends on the taxon. Some grasses are rarely ID'd wrong simply because nobody but serious botanists even know they exist. (A friend went through one species recently and found very few errors -- and not a whole lot of observations, either.) Some thing are so easy, they're usually right to start with. And then there are things like Timothy, a name people have heard of, with an appearance that the computer vision can over simplify, that people get wrong often.

Posted by sedgequeen over 2 years ago

I only started working on the grasses after I felt I had nearly mastered the majority of the other vascular plants I was seeing, or knew to grow in my area, and felt I needed to move on to the next group. I was also given a 5 Volume Hitchcock at about that time, and I made an effort to learn all of the local grasses both by floral characteristics, and to recognize them without flowers. I was up half the night with my Vol. I trying to memorize all of the distinguishing features. I never made that same effort with the sedges, but was also more frustrated with the sedges when I tried. The mosses came after the grasses. but they came more because I was finding multiple vascular plants using moss substrates as important habitat, rather than feeling I had mastered most of my grasses. I then started working on the fungi after discovering a Seattle population of the mycoheterotrophic Cephalanthera / Eburophyton austineae - Phantom orchid, that was missing from a Seattle area plant list I was working from and was wondering what their fungal partners they had. (There was one 1935 Seattle area record for that orchid, but the Seattle checklist had just missed recording it.) I also knew that the fungi provided critical habitat for a lot of our vascular plants, but after studying them for years, I still don't know many of which fungi partner with which plants.

Posted by stewartwechsler over 2 years ago

If I had a dollar for every Alopecurus misidentified as Phleum....

Posted by lysandra over 1 year ago

Oh, yes!

Posted by sedgequeen over 1 year ago

Wow, I'm impressed by the list but also by the thought that you recognized all those species when they were proposed as Timothy. I frequently end up saying "I don't know what this is, but it isn't Smilax" and let it go at that (though I do have a 'most often mis-identified' list). I hope you are finding that the CV is giving fewer bum steers. I do see improvements in my taxa. Human identifiers matter!

Posted by janetwright about 2 months ago

@janetwright -- Misidentifying Alopecurus as Phleum is a problem in the field here, so I headed into iNaturalist expecting it to be a problem. Then I kept finding other misidentifications and I started a list. After a while, finding new errors became a goal. I have to admit that some of the odder ones were identified only because I asked other people what they thought.

Posted by sedgequeen about 2 months ago

There was no way the CV could correctly ID Phleum pratense when I started, due to the errors in photos it was trained on. I think it's getting better now.

Posted by sedgequeen about 2 months ago

Add a Comment

Sign In or Sign Up to add comments