Appalachian snakeroot identification resources

Not too long ago (earlier this year?), iNaturalist started recognizing Appalachian snakeroot (Ageratina roanensis) as a species rather than a variety of A. altissima. Prior to this, many of these plants were identified to Research Grade under A. altissima with only few being narrowed down to variety. I know I have contributed to this as I rarely ID to variety or subspecies, and as far as I can tell nobody has revisited these observations yet to see if they need updating. Since there aren't enough observations identified to A. roanensis yet (77) to qualify for inclusion in the Computer Vision training set, iNat CV consistently suggests A. altissima for all of them, further adding to the pool of potentially misidentified plants. The purpose of this post is to summarize a couple of resources that might help with sorting them out.

Distribution and Habitat:
A. roanensis appears to be a Southern Appalachian endemic found in the mountainous areas of Alabama, Georgia, Kentucky, North Carolina, South Carolina, Tennessee, Virginia, and West Virginia, and primarily at elevations above ~4,000 ft. Its typical habitat is moist to mesic northern hardwood forests and spruce-fir forests.

Distinguishing Characters:

  • Leaves: The leaf blades tend to be more deltoid (triangular) in shape compared to more ovate in A. altissima, and the leaf base is said to be generally subcordate or truncate vs. generally broadly cuneate. However, I see a lot of overlap and I'm not confident I could tell them apart for sure just based on leaf shape.
  • Inflorescences: These are said to be arranged in more dense corymbs in A. roanensis vs. more open corymbs in A. altissima. A. roanensis has more florets per head (18-34) compared to A. altissima (9-20). This can be hard to see on the typical iNat photographs - I find it easiest to count if the individual flowers haven't opened up yet. Again, there appears to be quite a bit of overlap with Weakley suggesting A. r. can go as low as 15 florets and A. a. can show up to 25 florets. Another difference is the length of the phyllaries with 5-7 mm in A. roanensis vs. 3-5 mm in A. altissima.

If anyone knows of other features that can be used to distinguish them, and in particular how to tell them apart on the typical iNat-quality photographs, please feel free to add those tips in the comments below!


Posted on November 08, 2022 12:49 PM by annkatrinrose annkatrinrose


Thank you for this guide! I am not sure I could tell them apart at the moment, despite being aware of the two taxa for a while and your summary of the characteristics, but I will keep my eyes out. I like the focus on getting the 100 observations together to be included in the computer vision model!

Posted by mycoweise 11 months ago

Same here - I've been trying to figure out how to distinguish them for a while now and I still feel like I can't tell for sure. There are very few observations where I would feel confident to claim it to be A. roanensis. At least the range and elevation info may indicate areas where that option can be ruled out.

Posted by annkatrinrose 11 months ago

Thanks for the info! I wasn't aware of Ageratina roanensis, so this is very helpful. If I were to look through old A. altissima observations to try to help to reclassify any errant IDs, and have any questions should I just tag you?

On another note, I was under the impression that the AI criteria was changed in the last year so that now it only requires 100 photos, and not 100 observations for the AI to begin training on a species. Do you know anything about that?

Posted by dsmorris 11 months ago

Here is an iNat blog post from April 2022, that talks about the AI changes:

The part that I'm remembering comes from this line in the post:
"[AI training] includes more taxa right on the borderline of inclusion (taxa with at least 100 photos but fewer than 100 observations will now make it into the export, but didn’t previously)"

Am I misunderstanding the new AI training standards? Maybe AI is trained on these photos, but it is not going to offer them as suggestions yet? Or would there be another important reason that Ageratina roanensis is not coming up as an AI suggestion yet?

Posted by dsmorris 11 months ago

The way I understand it, the AI training set may only take pictures from observations that have a community ID. Captive/cultivated observations can make it in there as long as they have a community ID. That would mean it only considers pictures from RG but not Needs ID. I may be wrong though - it seems trying to understand how CV works and what is included is a moving target, just like trying to learn all these plants species that taxonomists keep splitting right under our noses.

Posted by annkatrinrose 11 months ago

Oh, that makes a lot of sense! And it does look like around 50% of those 77 observations you mention are research grade at this point, and only maybe a 1/4 of those have multiple photos, so it's probably no where near enough for the computer vision to work with. And I totally have the same feeling as you about the taxonomy! An ID I made recently for what I thought was Fritillary butterfly, turned out to be a Gulf Fritillary, which is in fact no longer considered a Fritillary butterfly--which was super confusing and took me a while to understand. I feel like I am having experiences like that every few weeks. Something has been switched and I had no idea!

Posted by dsmorris 11 months ago

Add a Comment

Sign In or Sign Up to add comments