I know it’s not Friday, and it’s not IDing, but I would greatly appreciate help in fixing observations incorrectly annotated as feathers (as some observations need multiple downvotes). I have gone through all observations here (both needs ID and RG): https://www.inaturalist.org/observations/identify?verifiable=true&taxon_id=3¬_in_project=found-feathers&term_id=22&term_value_id=23, but if anyone wanted to go through all observations annotated as feather I would appreciate that too. Also, if you choose to help, please check all images as some will have both birds and feathers. Thanks in advance!
Isn’t this annotation, at least for the flowering stage, something that should be delegated to the machine learning algorithm?
I mean if an average amateur can see if a plant is flowering, it should be pretty reliable for ML and could cover millions of observations without relying on these lazy humans.
1 Like
Maybe it could be delegated to the computer, but that hasn’t been done here. If we want it annotated, we humans have to do it.
(Should it be delegated to the computer – a different question. I suspect the computer would make a lot of interesting mistakes, given how odd flowers can be – consider grasses! – and how many bright things can land on plants where they could be mistaken for flowers.)
2 Likes
it’s a chicken and egg problem.
6 Likes
Respectfully, that comment is over four years old, and there are now over a millions observations annotated as flowering. Is that not enough of a training set to begin to provide some assistance/augmentation to manual human annotation?
I used to think - flowering plant - observed because it is flowering, and only annotated for fruit. Now I annotate more as I ID. And am meticulous about annotating my own obs. Phenology info is on iNat, but we need to take that closing step to make it visible.
If it is human, not automated, it is another layer of quality control - wait - that isn’t That species. Or this obs is for the brownish chameleon, not the dead Watsonia it is on.
2 Likes
i suggest that you and @chirp147 make your thoughts about computer vision + annotations known over in the other thread i noted above. the people who actually would do the required work are more likely to see your comments there.
Lynn, there’s one (swamp loosestrife) that I can’t do at all, and you’re the top. I can’t believe it’s easy! Lol i will learn jt! I have hope!
1 Like
Yeah, it makes me cringe to be at the top of a short list
1 Like
The erigeron has a second late summer bloom in the NE! I didn’t notice until this past year. Lesser, as your graph implies
I came back from a 6 months IDing break in december. I was delighted to see the small Coccinellidae IDer community had grown a bit. (Waving at @That_Bug_Guy ) I also found a backlog of more than 30000 observations to review. Well I had done the same before after a similar break. The goal now was obvious: get the backlog done before the CNC 2024. And today this:
So I made it a bit earlier.
13 Likes
Always happy to help :)
4 Likes
Patti, clearly you need go laze about in a kayak on a nice wetland this summer, when the Decodon is in bloom.
3 Likes
Yes, that seems to happen around here, too. I’ve noticed on those late summer to fall plants in bloom there is often some evidence that they have been cut down or mowed earlier in the year. Tried and true gardening advice: Dead-head your plants after flowering for a second bloom! Seems to work well on these.
Alternatively, the “second bloom” in late summer may be at least in part E. strigosus rather than E. annuus and a phenomenon driven by phenology differences between two similar species. I just talked to some of our botanists/herbarium folks at work yesterday and we all agreed that based on our own anecdotal observations E. strigosus seems to bloom later in summer compared to E. annuus in our area. Comparing the phenology graphs for the northeast, there indeed seems to be a one month difference in peak bloom with E. annuus peaking in June and E. strigosus in July. However, the phenology graphs for the Southern Appalachians suggest their bloom times overlap and they peak at the same time in June. Is that real or indicating that a lot of our local E. annnus gets misidentified as E. strigosus?
I was hoping to get that sorted but this has turned into a bit of an impossible problem due to overlapping characteristics, a history of synonymy (as well as some recent pubs treating them as varieties of each other), and lack of identifying features on iNat observations. The Flora of NA sums it up nicely by saying: “Apparent intermediates between E. annuus and E. strigosus are encountered.” There probably are a lot of misidentifications between these two making them blend together. Trying to clean this up may mean bumping hundreds to thousands of species identifications back to section for being inconclusive, and I’m hesitant to do that. I’ve already gotten some push-back on my IDs even in those cases where I can point to features that are mentioned in the keys and plant descriptions. There may be a wasp nest hiding down in that rabbit hole - not sure I want to poke any deeper. 
5 Likes
I find this very interesting because the seasonal diagram for Erigeron annuus looks quite different here in Germany, with much less of a distinct first bloom vs. second bloom and continued blooming well into October:
Because the majority of observations do not have phenology annotations, I’ve used the general seasonal diagram here, but my experience is that most observations represent plants that are still actively flowering, which often seems to continue until the first frost.
I do not have a sense of whether most observations actually represent Erigon annuus and not E. strigosus. Supposedly E. strigosus is a neophyte that is not currently widespread in Germany, but as you note, there is a history of synonymy and it is quite possible that it is being mis-ID’s by users who aren’t aware that there is more than one similar species.
Now you got me curious about how the phenology of these compares across the world. Both are native to eastern North America and introduced in various places around the globe, so I decided to compare NA vs. Europe vs. Asia. Here’s what I found. The left panels are E. annuus, right panels are E. strigosus, top row is North America, middle Europe, and bottom row is Asia.
For each panel, I included the phenology graph with the gray curve for “no annotation” (on the left) and rescaled without “no annotation” (on the right). Red is flowering, orange budding, blue fruiting, and green for no evidence of flowering (e.g. leafy rosettes). The dotted lines marks June, because that seems to be average peak bloom time for NA. The numbers in each panel are total number of observations for the species in that location and which percentage of observations has been annotated. As you noted, the majority of observations do not have phenology annotations. However, for the most part the gray “no annotation” curves peak in the same month as the flowering annotations, which makes sense as that is probably when observers take notice of these plants.
Some take-aways from this:
Marked with star #1 are early spring “bumps” in the curves for both species in NA. From what I’ve seen while ID’ing in my area, this may represent misidentified E. pulchellus and E. philadelphicus, especially for E. strigosus. Both are spring bloomers and absent (or only very few observations) in Europe and Asia, so it is not surprising that this “spring bump” is missing in those locations. This might be an easy target for someone looking to clean up some of this data, as the spring bloomers are relatively easy to tell apart from the others by their clasping leaves.
Marked with star #2 are the fall “bumps” in the curves for E. annuus. Interestingly, this bump is much more pronounced in Europe and Asia than NA, and it seems to be smoothed out in E. strigosus. As I mentioned before, it went away almost completely after I went through the Erigerons for my area and disagreed on anything that was an aster (spreading involucral bracts) instead of a fleabane. The bump for Europe is huge though and I’m not sure how reasonable it is to assume that all of those are misidentifications as well. It would be interesting to see if and how this might change with more annotations and IDs.
Some other random observations:
It seems Europe has more observations of the introduced E. annuus than NA where it is native. Is that for real? Or a CV suggestion artifact? We usually complain about all the invasives in the US, but this looks like a case where it might be turned around.
I’ve come across some German articles/theses claiming that there is no E. strigosus over there but three subspecies of E. annuus instead. Some are treating E. strigosus as a synonym of E. annuus. Polyploidy and apomixis are involved. How much of that is going on in the US, and how does that affect the IDs being made by local identifiers?
Both POWO and Wikipedia claim that E. strigosus has been introduced to China, while E. annuus according to POWO is NOT present in China. However, based on iNat observations, almost all of the Erigerons in China are identified as E. annuus with only a few questionable observations ID’d as E. strigosus. Are both POWO and Wikipedia wrong on this? Or are they all misidentified on iNaturalist? Or are identifiers using different keys? Looking at iNat observation numbers, it seems questionable that there is any real E. strigosus in Asia at all.
This rabbit hole goes far deeper than I thought…
3 Likes
Anecdotally, my experience in Germany is that the weedy Erigerons bloom fairly consistently well into the fall. I did key out the ones in an empty lot I’ve been monitoring, so I am reasonably certain that they are indeed E. annuus and not E. strigosus (at least, as defined in the key I use…). The last couple of years we have had a very dry first half of the summer, a wetter late summer, and an unusually warm fall, which has disrupted some of the usual phenology patterns and resulted in second late summer/fall blooms of plants that are normally finished by that time, but I doubt this would be enough to skew the charts for all of Europe over a longer period of time.
Looking at this at a continental level means that each of the data sets include a wide variety of climate zones. This seems to be less significant than I would have expected, but a more fine-grained breakdown could potentially offer some insights.
It is very common in ruderal urban spaces here (i.e., areas where observers are likely to spend a lot of time). It also seems that E. annuus is mostly limited to the eastern half of North America, which is geographically a somewhat smaller area than its range in Europe. The German Wikipedia page does suggest that it is an invasive of some concern (“In Mitteleuropa zählt es zu den wichtigsten invasiven Neophyten.”) though apparently it is not a recent arrival (18th century).
True, so here’s a south-north breakdown for NA at least. Same type of figure as above, but split between Southern Appalachians (where I went through and annotated all flowering/budding/fruiting plants; the ones not annotated here are basically no evidence of flowering), northeast US, and Canada.
A couple more things to note:
The “spring bulge” (1) for E. annuus probably includes a bunch of leafy rosettes. This species is a winter annual, its seeds germinate in fall and the plants overwinter to resume growth in spring. Their rosettes are presumably easier to ID to species than E. strigosus. I only annotated those where I was sure of the ID that had been suggested.
The “fall bulge” (2) is present in both NE US and Canada, but much less so where I’ve been through the data for the Southern Apps. E. strigosus shows a bit of a bump there, too.
Phenologies differ north to south. In the south, bloom time for both species appears to peak in June. In the north they have separated with E. annuus peaking in June and E. strigosus peaking in July. Interestingly, this matches the observations for Europe. In Canada they both seem to be shifted into July. The place map for NA also includes Alaska and Greenland but neither of those places had any observations for these two species, so Canada probably represents the northernmost occurrences.
I wonder if the second “bulge” in fall is due to some seeds germinating in spring rather than fall further north, resulting in a mix of biennial (early bloomers) and annual (late bloomers) individuals? Again, maybe a target for some clean-up to remove misidentifications and add some annotations to see what’s really going on here.
4 Likes
Even though I know @annkatrinrose put a good bit of time into this analysis, I shudder to think of how much more time it would have taken before iNaturalist.
4 Likes
After helping to fix an Argiope problem in Asia I kind of got stuck in Taiwan now … soooo many needs ID-arachnida that are pretty easy for me to push to family, genus or even species… already half-way done for that broad level, will dive into the family-needs-ID next… will keep me busy for a while :-D
As we had this discussion last years several times on the forum about how much leading-improving-supporting IDs tell you about an IDer… I find it interesting how different this scale looks at the moment for me when comparing to last year. Supports me in my idea that those percentages tell you not much about an IDer actually…
5 Likes