Mission: Impossible - Identify Plantae in Africa

Ay no! The month is ending!

1 Like

I wait, each month, for this update. Use the internal links to see your taxon or location.
https://www.inaturalist.org/posts/83370-a-new-computer-vision-model-v2-6-including-1-399-new-taxa

Using data from 16 July - so the September update will include the first half of Mission Impossible, and October for the second half.
iNat CV added 74 new plant species for Africa. That is global iNat, not ‘first time observed in Africa’
Each new species added makes it easier to ID That species in future, and avoids CV offering looks good (but hasn’t been seen in Africa. Oops!!)
Everything we can nudge along to species targets future CV updates :heart_eyes:

6 Likes

Partway through this week the biggest movement has been in Unknown obs, decreasing by 1700. Obs at species level are also going down at the same pace as previously, by 700 this week so far.

With only 7 days of the challenge left, now is the time to shine!

Happy iNatting ;)

Number of observations in thousands for Africa excluding the Cape (counts including Cape in brackets).

Most recent counts shown as change relative to start of week 4
21-08-2023 24-08-2023
Not in Flora of Africa
  Plantae at Kingdom 10.4 (16.7) -0.3 (-0.6)
  Unknowns 5.4 (7.6) -1.1 (-1.7)
Flora of Africa
  Phylum to Class 44.4 (64.5) +0.6 (+1.1)
  Dicots 28.2 (39.4) +0.7 (+1.1)
  Subclass to Subtribe 52.6 (102.2) +0.3 (+0.2)
  Genus to Complex 131.7 (317.9) +0.0 (-0.1)
  Species and lower 176.3 (357.1) -0.4 (-0.7)
Captive
  Unknowns 3.5 (9.1) -0.0 (-0.0)
  Plantae 76.9 (184.5) +0.2 (+0.3)

Observations not in Flora of Africa. Black lines are including the Cape


Change in observations in Flora of Africa, relative to start of project.

1 Like

Looking good except for creeping up Phylum?
That is an impressive amount of hard work!

I expect Phylum-to-class chunk will be ticking up for a bit as Unknowns are broadly cleared out in the next couple days. A lot of distant trees are receiving a basically provisional dicots label. If some turn out to be podocarps instead, we can figure that out and correct them later- or sooner, if anyone notices big blunders while this is happening.

3 Likes

Trees to the Tree project along with annotating for flowers or fruit?
Then we might get a quantum leap for trees too!

The task switching between those tasks turns out to be too many extra clicks when done in bulk. So my own technique by this week is to bulk label as planty, then go back through for annotations in a later pass. The good news for some distant trees is that for some regions, there are enough obs of distinctive silhouettes, that some trees could later be id’ed by shape in a “who’s that Pokemon” way.

4 Likes

There are keyboard shortcuts.
I don’t use them - picked up from a previous forum post? (But no way can I find the source again!)

p + l = fLower
p + r = fRuit

and you know
x = cultivated (that is ingrained now)
r = reviewed
c = to comment
a = to agree (but I am very wary of WHICH ID iNat agrees to)

2 Likes

It’s also a mental task switching issue at this speed. Keep the faith!

3 Likes

The muscle memory does what it does on autopilot :grin:

3 Likes

Here’s my chance to bring in what I hoped to be part of the uh, mission debrief. By now, if y’all have your favorite keystrokes for easiest use for what you do the most, please help me flesh out the alphabet right in the first post (editable by others):
https://forum.inaturalist.org/t/wiki-an-unknown-identifiers-tree-of-life-draft/38591

Just freestyle in there alphabetically for now and we’ll sort it out later. I’m hoping some might share easier keystrokes than I use for some of the common categories.

Like, how many keystrokes needed to get Genus Euphorbia or Genus Oenothera to be easily clickable (in top few or choices)? If the answer is, oh many obs are just in these easier to pull up Sections, let’s list those too if you got 'em!

2 Likes

Type slowly
see what the first letter gives you
or the first 2.

For example l (ell - this is a horrible font to distinguish capital I or lower case l !) will give you Lepidoptera. Autocomplete is good once you discover the trick to catch the ID you want.

For Euphorbia
spu for spurge is quicker.

3 Likes

I’ve been going through Unknowns for the last few days.

General problems with the Unknown bucket are no different from anywhere else - beginner mistakes such as observations with multiple species, no clear indication of subject, etc. I have been adding notes and questions to the original observer, suggestions to separate the pictures into different observations, giving a generic id to cover all the pictures, etc. Many were class assignments, so I don’t expect many of those will get a response. One person with multiple observations I commented on did respond and so this was helpful in at least one case.

Otherwise, I have been thinking about strategies as I’ve worked. Here’s the strategy I now have.

Work through the chosen query, in my case the Unknown query excluding the Cape, query is from a post in this thread: https://forum.inaturalist.org/t/mission-impossible-identify-plantae-in-africa/43528/128

To start, be sure to check for placeholder text that might reveal something. If it’s text I don’t recognize, it is often a plant name in another language. Search that text on a search engine to see if it translates to an actual ID just from that. If there is placeholder text, it gets hidden when you submit an ID, so I will copy it into a comment indicating that this is is the original placeholder text, in case it’s useful for the next identifier. Anyway, it seems respectful to the observer to keep that text.

Same thing with the observer’s comment text. It could be a name for the plant that they entered correctly as text, but did not get it into an actual official ID, leaving it at Unknown even though they had an ID.

(1) Identify to dicot, monocot, conifer, fern, moss, fungi, insect - high level areas that don’t require detailed knowledge of the flora and fauna of the region, but can be inferred from the apparent structure of the organism. Looking for parallel leaf veins (monocots) vs branched leaf veins (dicots). Is there wood? Then at least Tracheophyta. Conifers are usually recognizable as such. Shrubs with bare woody branches are dicots. Some families or genera that are obvious from visual appearance I can identity: palms, Poales (grasses, sedges, rushes), Fabaceae (legumes and friends), aloes, citrus, sometimes Asteraceae, etc. A few specific ones, mostly weeds, I was able to recognize, and a few houseplants and other things.

(2) For angiosperms/dicots/monocots : go to the annotation tab and specify Flowering, Fruiting, etc.

(3) Click the View link to go to the observation view. If it’s an obvious tree in the southern part of Africa, add to the “Trees of Southern Africa” project.

(4) View the Computer Vision suggestions by clicking on the ID part of the observation tab and waiting for the suggestions to load. This seems to be the best way of getting relevant suggestions. Suggestions tab in the ID view doesn’t seem to give anything relevant unless it’s much more narrowed down. I view the top one or two suggestions, maybe more if it seems promising. If a reasonable match exists, I’ll choose that and note that the ID is from CV and needs confirmation.

I’ve noticed a few of you in there, so I’m not feeling quite as lost as a solo adventurer deep into the heart of Africa. I have learned some interesting things. I have seen several species of wild Asparagus. I note some of the dry areas show very thorny plants with tiny leaves, I guess they must be subject to intense grazing pressure. It’s an interesting experience.

I think a possible better approach could be to familiarize oneself with plants in one area and then create a custom query for that area, and work from there. I haven’t tried this, but I think getting to know the plants (maybe using the Explore tab and viewing identified observations for that area) would be important to make progress beyond dicot or monocot.

5 Likes

I started to wonder if there’s an opportunity to handle some special plant forms (other than trees) by putting them into a project. The one that comes to mind is succulents. There are a few different families of succulents in Africa, but no common taxon that you can use to signify “succulent” even though it’s fairly obvious from the pictures.

Crassulaceae covers many plants I came across, but there are several more. Some are monocots, like aloes.

There are people that specialize in succulents, some people collect them, etc., so they could likely take the ID to species in many cases once they were put into the project.

1 Like

I would love to learn more about the relevant families. I can add to a project on the web version, but not on the Android app, so that slows down the process a bit.

If you would like me to @ you until the end of the month or until you say “enough” we can trial that. But I am deep in trees at the moment.

1 Like

I’ve moved several grass observations out of “Needs ID” at the species level! Unfortunately, to the genus or family level. I’m sure no South African observer following this Forum post needs to be told this, but I wish I could communicate to many South African observers that not every straggly grass near the coast is Ehrharta erecta! Some of the observations with that label were Poa annua, Holcus lanatus, Eragrostis sp., or what looks like Ehrharta longiflora. (For non-grass people, those are relatively different-looking grasses.)

One problem: though I can be sure many of the observations were NOT Ehrharta erecta, I don’t know the alternatives well enough to say with confidence that an observation IS E. erecta. If you can do that, there’s some low-hanging fruit for you. You’ll also find more E. erecta observations you can move to Poaceae.

1 Like

Tony has added regional links, so that project (despite the name) is open to any trees in Africa.

If the future ID of an obs interests you, you can fave or follow (without adding an ID) then learn as the discussion unfolds. Year by year I can ID more.

Do I also add to the “Mission Impossible” project, or does that happen automatically?

1 Like

For right now the project is just manually added stuff that may be of interest for whatever reason! Reasons I’ve put stuff in there include mysteries that I hope to solve along the way; good discussion, of various topics but especially of field marks; good or interesting field marks; cool or surprising finds. Also the “life’s little victories” ones where the learning process started at Tracheophyta but then got markedly better by a later week.

Join the project to add your own finds or toughies or whatever! We’ll sort things out later ahead of the next bioblitz.

Sometimes I alternately hit the Favorite button instead, so that I know to go add it to the project later. Going outside the modal at scale is a whole different workflow. ;)

2 Likes

I have set myself a challenge to populate the map. For some reason I haven’t come across an interesting obs for Egypt. I wanted papyrus, but the obs I found all looked cultivated? Will pick a volunteer …

If you compare our Mission Impossible map to the one for Flora of Africa I expect the coverage to be similar. But it isn’t. The Sahara and the rainforest heart don’t have as many obs on ours.