The OpenScience Laboratory The Open University

iSpot Forum

Has anyone seen / used [email protected] Identify?

eg

Those labelled up ‘Plant Net’ use online picture matching ID software

Keys have the problem that the plant might not be at the right stage of development, or you have to know the answer before you can interpret the key - as often as not I run into an impass with a key. There are multikey apps which might be more of a happy medium.

(I’ve been thinking of the possibility of a crowdsourced Critical Illustrated Flora of the British Isles aimed at epub and pod publication, where by relaxing the constraints of paper one can incorporate multikeys, illustrations of field marks/key features, variation, etc. And a sister project - Woody Plants Hardy in the British Isles.)

1 Like

I’m glad to know someone else suffers from this. I take photos and try to observe what I think might be important then I get home and find, often early in the keying process, I’ve not noticed what turns out to be a critical feature. If you don’t know the critical features, you can’t use the key; if you do know them, you probably don’t need the key!

I don’t carry books with me much of the time - after waterproofs, food, water, camera, tablet (for recording), powerbank, maps (for if the tablet battery runs out) I’ve got enough weight on my back without Stace never mind 5 volumes of Sell & Murrell - but I’ve run into an impass with a key in the field before, never mind working from photographs when I get home.

lavateraguy
Herringbone
I much admire the way you phrased THE question in the comment here https://www.ispotnature.org/communities/uk-and-ireland/view/observation/790940/bioblitz-s295-2019

one of my friends is pondering whether there might be interest in/a market for a field guide that basically tells you want you need to look at/measure/sample in the field in order to get to a reliable ID. Not sure its a best-seller, but perhaps useful?

I always try to have either a basic ID guide (Stace field key or in unfamiliar territory I rather like Streeter) or better still a more experienced botanist (self-propelling and doesn’t take up rucksack space!)

Sounds like there is a gap in the market for something like this, but perhaps nationally and with multi-keys and illustrations?
http://webidguides.com/
Mike has been developing this for a few years now - it attempts to bring you back if you’ve gone off in the wrong direction. I am sure he would happily expand it to the wider UK/build in other features if some sort of funding were available

Had a brief glance at webguides you mention, trouble was that I was thinking of a species (of tree) that was not included so it was not very useful. On wikipedia they tend to list most of the species even if only a relatively small number then get a description, this can help if you think you have got to the right group but the detailed descriptions don’t match - at least the other species are listed even if they don’t have a description.

The other point I wanted to make about having a guide that tells you what to record is that ‘what to record’ depends on many factors e.g. if you are just trying to tell between a particular known set of 3 sedges or want to distinguish all the trees in an arboretum. This becomes very clear if you are trying to write keys and have to set the boundary or add in a few extra species.

I was thinking that a revised Plant Crib, aimed at a less advanced level, would be handy - a book specifically aimed at the identification of difficult to identify plants. (I have on a tablet the Plant Crib PDFs from the BSBI web site, and the Kindle sample of Stace (which covers Ferns and Conifers); I also have Kindle editions of Sterry on Wild Flowers and Sterry on Trees, but the Kindle editions of those aren’t very useable.)

The East Anglia wesite mentioned http://webidguides.com/_templates/intro_template.html
has an Introduction with this rider:
“Always remember rule 1: Never identify a plant from a single feature!”
One might add “Always remember Rule 2: Never identify a plant from a single website”.
I like the East Anglia website (still under production so there are 404 pages) because it is trying to help beginners and at the same time acknowledges that identification is not always easy. The latter statement being, perhaps, the reason why those of us in this forum find it at times frustrating, always absorbing.

http://webidguides.com is lovely. There are some key links broken -
about us
references
glossary
communities
I had no way of reporting them.
I liked it from hereon in http://webidguides.com/_templates/section_flowers.html

Following up my plan from 12 days ago… i posted one photo from this in plantnet

And it offered F. excelsior at 4.76/5.
Clicking on that species name there were lots of photos which the poster could compare with his post.

So I’m wondering if, having checked it out ourselves to see if it’s a likely, we might consider directing people to plantnet. Would that be encouraging?

I think anything that helps folk to an ID is useful and, as I’ve suggested before, the pattern recognition / AI route is very much up-and-coming. From those points of view I think, yes, we should be pointing people towards such resources. But, at least for the moment, that advice must come with some warnings: not to put too much trust in the machine; to understand the n/5 probability given with each suggestion and to recognise the relatively limited scope of the machine. iSpot, for all its faults, is still better in many circumstances.

Have you seen the link where the “probability” is explained. (Since I think that I’ve seen numbers above 5 I’ve assumed that it’s not strictly speaking a probability.)

What would be the point in going to my Math teacher and saying “I can’t answer this, please do it for me”?
name withheld

Much the same as posting something on iSpot and saying “I don’t know what this is; please tell me.” In each case. very little point unless you use the answer to learn how to do it yourself next time or can go on to use the answer to do somthing further.

Keys! Grrrrr!
I sat down this morning to try to identify some insects from genera I’m not familiar with (wide scope there!) I skipped over a fair number of photos where the quality was probably not good enough for ID at this stage but eventually found a photo that looked possible for ID. I think the fly is probably one of the Muscidae so I looked up a key. The first entry was
1 Vein R1 with small bristles on the apical third of the
dorsal surface …
Time to give up and have a coffee.

Your keys comment opens up a whole field, some of which I touch on in the https://www.open.edu/openlearn/science-maths-technology/citizen-science-and-global-biodiversity/content-section-overview?active-tab=description-tab free course.

It is particularly interesting in times of DNA and AI where the DNA might ‘definitively’ tell what the species are but at the same time allow you to look for new/better morphological characters that can be used in keys (because you are sure those are two separate species and not just variation within one).

Possibly something similar might apply with AI but it is much less clear there - the AI might really distinguish between two real species but it won’t be able to tell you how it has distinghished them i.e. what characteristics it has used. The DNA uses changes in DNA molecules but the AI has something in its algorithm.

It’s relatively recently that botanists have started using the fine details of leaf venation as identifying characters. (They’re less subject to variation and convergence that overall leaf shape.) Previously neobotanists focused on flowers and fruits, and palaeobotanists got identifications badly wrong - see where the various species of fossil “Ficus” have ended up. AIs might well end up being trained to look at the fine details of leaf venation, and do better at leaf identification than the average human field botanist.