Identifications taking longer?

General
62

All of these would have definitely helped to encourage me to submit more IDs earlier.

Something I think would be beneficial on the species pages (the inat formatted/hosted ones) better notes about cases where it’s difficult (if not impossible) to distinguish species from photographic evidence. I’ve come to not be a fan of these details pages pulling from wikipedia.

Through work, I’ve been learning more about local cottontails, particularly the Appalachian cottontail. My supervisor is a bit of a cottontail expert and has informed me how difficult it is to distinguish between the Appalachian cottontail and the eastern cottontail, which are sympatric in my area. In our project, we get LOTS of photos of cottontails and we are stopping at the genus level with our identifications.

Given the difficulty described to me by a cottontail specialist and the note provided here from the South Carolina DNR:

Swamp Rabbit (I don’t know why the link gets parsed as Swamp Rabbit for the title, but it’s a fact sheet about Appalachian cottontails)

These differences between the Eastern and Appalachian cottontails are so subtle that identification usually requires handling of the animal.

And also this fact sheet from the NC Wildlife Resources:

Appalachian Cottontail Rabbit | NC Wildlife

A more reliable method of distinguishing between Appalachian and eastern cottontails is to compare skulls. When examining the Appalachian cottontail’s skull from above, the suture line where the nasal (nose) bones attach to the skull forms an irregular, jagged line. That line is smooth and regular in the eastern cottontail. However, genetic analysis provides the most accurate determination of species.

there are FAR TOO MANY at research grade on inat. Especially considering how many observers (and subsequent identifiers) appear to be new and not specialists in the species. I guess that is also evidence that there aren’t enough folks going through those observations and giving that info. I did a few yesterday that were lower-quality photos (shots from a distance, for example).

I also noticed yesterday when looking at the old observations in my area that don’t have many (or any) identifications was how many are salamanders with reasonably good quality photos. I know that some require DNA sequencing to distinguish species. I know that some have such heavily restricted range (say, a single valley) that an obscured location eliminates an important element in their ID that I’m not surprised by the lack of identifications of salamanders.

I guess these are examples of needing to use this checkbox more liberally.

Would there be value to auto-marking this box for some cases, such that folks would have to UN check it in order to supply a finer ID? And upon un-checking the box, they’re given a popup explaining that distinguishing such-and-such species is extremely difficult or impossible from field marks, and only to uncheck this box if you have additional info about identifying characteristics (and encourage supplying said info either as photographs or as a note in the description). Then after reading that note, they are asked if they still wish to uncheck the box, and they have to click “yes” or “no”.

I could see doing something like this would require a good bit of programmatic work. But maybe trying a test case that’s really obvious where you need DNA sequencing or dissection would help to see if doing so would be worthwhile. Instead of something like my cottontail example where it’s more fuzzy and it’s possible to ID the species in the field, but instead very difficult.

Doing something like this might help slow the addition of overly-ambitious IDs on these species and maybe give identifiers the ability to start catching up on the backlog?

4 Likes
63

First, salamanders (Order Caudata) are not especially speciose with only a few hundred species (just over 800 at last check) compared to other clades (nearly 8000 Anura [frog] species, nearly 12,000 Squamata [lizard and snake] species).

Second, (and in part related to the first) there just are not a lot of salamander biologists and only a couple I’ve seen active on iNat. Much more research goes into those other clades, so they kinda get overlooked.

Third, salamanders are secretive living under things for the most part (logs, rocks, in burrows) and mostly active at night. What’s more, their peak of activity is on rainy nights when it’s not the most pleasant to iNat meaning the bar for entry for amateur enthusiasts is probably higher than other, easily learned taxa.

Fourth, we have learned in the molecular systematics era that salamanders are incredibly micro-endemic sometimes occupying single mountain tops meaning without really specific location data two closely-related species can be hard to distinguish.

Just give it time though. Keep IDing those to Caudata or even family if you know it and some salamander enthusiast will emerge to help clear that backlog…those observations aren’t going anywhere.

4 Likes
64

An average identifier can identify a page of 30 observations in about 3 minutes. Experts might be able to drop it to 30 seconds for common species.(I can do a full page of 20 supporting and 10 leading in about 45 seconds for local plants.)

Unfortunately, there are more than 500,000 observations uploaded every week(this obviously decreases in the N. Hemisphere winter). I don’t have time to identify for more than 1 hour a day.

1000 identifications per person per week is great, but are 500 people really going to do that plus get through the old Needs ID obs?

I don’t think so.

2 Likes
65

I do find that checkbox quite useful.

I wondered if perhaps there were more useful tricks for hunting anomalies hiding in his linked searches. Always glad to find new tools!

66

I’m generally opposed to anything that involves gauging species like this on a case-by-case basis, because ultimately, that will fall on taxa experts to decide what is cryptic and in a lot of groups that ARE that cryptid, taxa experts are already overburdened and potentially rare. I am skeptical that using an algorithm to determine what is cryptic would be innacurate and create frustration for users.

Plus auto-checking the box could lead to a lot of frustration from new users - I WANT better observations, but not at the expense of the new user experience. We need better onboarding, for sure, so people understand it better - but people need to be allowed to make mistakes so they can learn, and being auto DQAd means that observation would never be seen and they may never know why if they don’t realize what the DQA fields are.

3 Likes
67

We need to bridge the gap between

  • I’m new and I want to learn to help ID

and

  • I’m a taxon specialist and I need to mentor ‘newbies’ to help resolve this particular slice of the backlog

There is willingness and enthusiasm on both sides of the divide.

6 Likes
68

I’ve noticed a slowdown in IDs this year as well. Maybe it’s just the summer doldrums for IDers or maybe it’s the glut of new observations coming in each day. For bird observations, confirming IDs that once took 30 seconds (!) now take almost a day. Not that I’m complaining, of course, it’s just interesting.

For most of the IDs I get on my submissions, it’s from individuals who follow me or, less often, individuals I’ve reached out to (tagged) for assistance. I suppose we all want instant gratification by having someone look at our observations sooner rather than later but if that’s important to you, you may need to put in some effort to get others to see your records. Sitting passively and hoping someone notices isn’t the answer.

4 Likes
69

I wish there were some level of identification need between “needs ID” and “as good as it can be”. The difficult-but-not-impossible IDs could be concentrated over time in their own category, so broad-level identifiers can focus on new and unreviewed observations that might be more easily identified, while the tough nearly-cryptic stuff is taken out of their default feeds. A taxon expert with limited time would also have a way to go straight to all the observations at any level that someone has designated as “tough”, and focus on just those. Obviously what someone designates as “tough” is completely subjective, so this would have no bearing on whether the observation is research grade or casual or anything like that. It would just be a way to divide “needs ID” into the “this is obviously a blue jay and just needs any birdwatcher to click agree” category and the “20 bug people have looked at this little brown beetle and no one knows where to put it” category.

In fact, a thought I’ve had several times when IDing- it would be neat to be able to sort observations by “number of times reviewed”. In other words, show observations that the most users have clicked “reviewed” on, to go straight to the ones that are the most mysterious or confusing. Like personally, if I’m IDing birds, my time is best spent confirming obvious things and leaving the “which dowitcher is this” questions to the pros. But when I’m IDing moths, I want to skip over the obvious stuff and see the smallest, plainest brown micros that don’t get much attention otherwise. Having to click through the Empidonax photos or click confirm on 20 Luna Moths doesn’t seem like a good use of my time, and there has to be some metric to at least roughly split up these categories of “needs ID” material.

12 Likes
70

I asked for that. And it was turned down.

I would continue to look thru all of my chosen batch - but would love to start by knowing - MANY identifiers have already ‘despaired’ of this obs. iNat has the info, but we cannot see it. It would be similar to here in the forum. The stats at the bottom of the thread. 702 views.

As an observer, instead of nobody has bothered with my obs :sob: I would know x people have at least looked at it.

9 Likes
71

People mark observations as reviewed for all sorts of reasons that may not have anything to do with whether the observation is IDable.

Possibly being able to sort by how many IDs an observation has received would help to surface ones that nobody has nobody has looked at vs. the ones that probably can’t be refined any further because half a dozen specialists have already added an ID (e.g., for taxa that can’t usually be ID’d to species from photos). It wouldn’t be completely foolproof, of course – sometimes observations end up with a lot of IDs not because they can’t be refined, but because they need additional confirmations because of initial wrong IDs.

4 Likes
72

It is interesting to see that everyone is seeing this. I know in my area I have noticed a significant change recently in the amount of unknowns that aren’t unknown and needs IDs. In General I’m a “generalist IDer” not knowing high level information on much except for some local things but even for that, the identify tab is getting longer, not shorter.

4 Likes
73

I am personally not noticing this trend, which is probably due to the fact that I am observing in a pretty well covered region (knowledge and IDers) for the first time in a long time and I din’t have that much attention to my observations for a while.

However, I fully agree that IDing could be made easier which would probably remove some hesitancy against IDing.

The notification issue that already came up is for sure a mayor annoyance for me. I just lost a couple of 100 notifications today because of shitty internet at the moment and I really do not feel like searching around through the dashboard, which is very inefficient and time consuming.

9 Likes
74

I have also found some maybe-not-obvious-but-doable-for-me. Sometimes, it’s just that the right identifier wasn’t there at the right time.

As an identifier, I’ll sometimes comment on how long something went untouched before I made an attempt. That way, when someone disagrees the next day…

3 Likes
75

That’s a great idea. I may implement that as I slowly slog through old events with backlogs.

1 Like
76

Those, actually, are some legitimately good ideas. Bravo. I see that @DianaStuder says the second idea was already turned down, but it may be worth making a feature request for the first.

EDIT: maybe an ‘expert opinion necessary’ DQA or something. IDK people may end up abusing it, unfortunately, but it would be a way to filter things stuck at higher taxa levels that actually could be refined down with expert help from stuff that is a blurry picture of a leaf taken out of a car window at 60 MPH

…though to be honest, sometimes just IDING that thing as something WILDLY incorrect will get expert eyes on it… (see: all the Macrolepiota procera IDs I went and corrected earlier that are somehow STILL being suggested by the CV as being present in the US when they aren’t and the proper species have been described and noted by the CV ;_;)

6 Likes
77

I get what you’re saying. I agree that relying too heavily on an algorithm would turn people away. I think taxon experts should be involved somehow. Maybe helping to craft a popup when someone selects a species saying “Many species in this genus/family are cryptic and cannot be identified to species without dissection/DNA sequencing/skull measurements/microscopic examination. Are you certain that this is Xxx yyy?” or whatever is relevant for that group. And the circumstances when it should be displayed. Whether there should be a geographical component to it, or whether it should pop up any time someone selects one of the relevant species.

For my rabbit example, you could use the statements from the factsheets I linked and craft a popup phrased something like “In areas where both Appalachian cottontails and eastern cottontails occur, they can be difficult to distinguish visually. The external marks tend to be very subtle and need close examination to distinguish. Are you certain you have enough information to distinguish between these species?” Maybe give them a couple “No” options. One to erase the ID and start over, and the other “No” option that puts the genus instead. You’ll still get people clicking “Yes” when they shouldn’t, but I guess this takes a more educational approach to maybe reduce that burden. It would still require significant effort on the website side to craft these messages in cases where they’d be relevant.

I also agree that it would be useful instead of dumping these into “Casual”, sorting them into a subsection of “Needs ID” in case a particular expert comes along who has such a refined eye for the species that they can tell. It would help these experts sift through the “hard” ones to see if they could help more.

3 Likes
78

One way of making it clear how hard it is to ID each species that would be completely automated and harder to abuse would be to give each a “Percent of IDs disputed” or something similar. If every ID of a species is disputed (100%) that is a really hard thing to ID correctly. If IDs of that species are never disputed (0%) anyone might be able to ID it correctly.

This could be calculated as the ratio of number of observations ID as this that also have conflicting IDs over the total number of observations ever IDed as this. Potential identifiers could even search for things with initial IDs with less than x% disputed.

7 Likes
79

For the record, if they get marked as ‘can’t be identified further’ at genus level, they go to RG, not casual - so they’re still there and useful, and could easily be reviewed by any expert who might be able to take them further, but out of the way of people who don’t have a clue and just want to reduce the number of Needs ID.

4 Likes
80

Just to clarify, they become RG if they have a community ID (not observation ID) at family or below; if the ID is higher they become casual. I think the logic is that really broad categories are not likely to be relevant for research purposes. My impression is that people are generally reluctant to use the DQA for more broadly ID’d observations because of the stigma that is attached to casual. Perhaps it would be preferable for such cases to have a more neutral label that would indicate the ID has been confirmed but the observation is no longer needs ID.

I agree about the usefulness of the DQA for managing workflows for difficult-to-ID taxa, but this statement could be interpreted a couple of ways, so I want to note:

As someone who has been working on a genus that is IDable to species from photos approximately half the time, I would prefer that people don’t use this DQA unless they are knowledgeable enough to assess whether the observation can likely be further refined or not (i.e., they know what sorts of details/features would need to be visible for ID, even if they are not skilled enough to ID it themselves) – it is not necessarily helpful to click the DQA just because several people have provided a genus ID without suggesting a species, as there are various reasons why people add genus-level IDs and it does not always mean the observation cannot be ID’d further.

Skilled IDers may be using the DQA button as part of their own workflow (if they are cleaning up a neglected taxon, they may want to be able to distinguish between which ones really cannot realistically be ID’d further and which ones still need to be looked at). In addition, because of the way the DQA works, it means that if someone subsequently adds a finer ID, the observation will be displayed as RG with the species-level ID, even though this ID has not been confirmed. So both the use of the button and interacting with observations where this button has been used requires a certain amount of care and vigilance.

4 Likes
81

You could get together (digitally speaking) with other moth experts and create a traditional project, to which experts could add observations of “tough” species. Then, instead of 1000 moth observations to sort through every day, you might only have 10.

(And thank you for your IDs on my moth observations, by the way, I really appreciate it.)

2 Likes