Among them leaf_type=palm seems most promising to me as it is not relying on botanical knowledge like being ale to decide whether something is in Arecaceae taxon.
It also does not put data that is known to be wrong (like say taxon:en=palm tree would be)
Also, I expect that there is at least one palm tree outside Arecaceae family - while “looks like palm tree leaves” seems more human-readable and decidable without expert knowledge.
The leaf_type of a palm tree should be broadleaved.
The Wikipage of leaf_type actually shows a palm tree as one of the example pictures for the value broadleaved.
To mark a tree as palm tree, other tags should be used.
Why is taxon:en=Palm wrong? You are thinking about species:en= ? taxon= can be any level. That’s its advantage, together with taxon:family:en=Palm .
On top of the possibility of considering them as =broadleaved , there are many “leaf type”. Palm can be pinnate (feather), or palmate (fan), aside from simple or compound leaf in other trees. And there are many other aspects Glossary of leaf morphology - Wikipedia
For others’ reference, one of the listed method, trees=palm_trees is a hack from =orchard , although trees= was once suggested as a replacement by someone Key🪵 Difference between revisions - OpenStreetMap Wiki (taxon:en=Palm has been listed in Key🪵 Difference between revisions - OpenStreetMap Wiki some time ago)
It’s unfortunate wood= was deprecated entirely without fully specifying replacements, and reserved it for undetermined generic use. trees=*_trees is redundantly verbose, and may seem awkward in =tree singular. (ironically fits perfectly on landcover=trees compared to =wood )
Personally I’m a fan of taxon=Arecaceae for palm trees.
I do think that tree mapping in OSM have a big margin of improvement at the moment. For example, we have many trees tagged with species=* missing leaf_type=* and leaf_cycle=*. I created a MapRoulette and a Wiki page for this but making this automated would be a big help. I proposed a new category for trees in the NSI. I see they are considering using Wikidata for fixing the localizated names problem. This would help in this situation since a user looking for “Palm” would get the Arecaceae preset suggested. @tordans made a similar proposal for iD.
IMO, the question here is if we want to be scientifically correct, or use the (misguided) common knowledge that palm leaves are neither needles nor broadleaves. Botanically, palms have leaves that are classified as broad leaves. This is not my opinion, it is a matter of fact.
I am not aware of palms that are outside Arecaceae family. Again as a matter of fact, Palmae Juss. is a synonym for Arecaceae Bercht. & J.Presl.
I can agree with you if you tell me that for most people there are plants that look like palms but aren’t palms. Since the judge of “it looks like a palm” is the observer, the judgment is subjective, and I am quite sure that no botanist will make the mistake to look at any Cycadaceae sp. or Strelitziaceae sp. and tell you it looks like a palm tree.
Back to my opinion, I suggest we use a tag like taxon:family=Arecaceae to mark palm trees. There are already 35k elements in our database marked this way.
But +30k were added with an automated edit (“discussed on the OSM World Discord”) in feb 2021 and 98% of its values are =Arecaceae. This tag is basically a boolean is_plam_tree=* at its current state. I think we should evaluate tags not only by their uses but also by their natural growth.
From a botanical point of view, using “taxon=Arecaceae” is a bit forcing the term taxon to mean family. I’m not going to put here the whole explanation, I don’t think this would be the right place. Any species, genus, family, etc are taxa. If you can’t identify a plant at rank species, you attempt identification at the lowest possible rank.
For palms, a very specialistic field in botany, it is quite common not being able to say much more than “it is a palm”. This is also why, as you say, 98% of the taxon:family tags refer to Arecaceae.
I’ve used it. Seems fine to me and is far more memorable than having to know the Latin name. There are also university pages with a whole bunch of information about them that don’t mention “broad” once.
Tree tagging in general is a bit of a pain. I’m long enough out of school that I can’t remember which word goes with which level of the family tree (they are seemingly randomly assigned) so even knowing which tag to stick :en to requires Googling every time I try to do this.
Most of the time I don’t bother with more than natural=tree because even if I know that e.g. “that’s some sort of citrus” it will take more effort to figure out how to translate that into a dead language than I can generally be bothered with. It’s a shame because if there was an easy way to tag what I do know I probably would do that more often.
if the point is enabling mappers to add a tag to plants that look like palm trees, to say just that, why don’t we simply add a tag “is_palm_like=yes”? Then we don’t need any botanical knowledge, and we’re not putting questionable and potentially wrong information in the database.
Yep, that’s why I use taxon, which is as you say a “wildcard tag” for taxonomic ranks. If a new user checks the natural=tree wiki and CTRL-F for “family”, two things will pop up:
A documented tag:
taxon=* - scientific name describing any taxonomic level e.g. order, family, genus, species, sub-species or cultivar.
A paragraph for palms that bounces you to a talk page in which 20 different tags for the same thing are listed.
So I guess they would choose the first option, and so do I. I’m not against using a dedicated tag for families, but doesn’t seem like there’s a consensus right now.
Also, I don’t understand why they decided to use taxon:family= for the +30k automated edit on the OSM World Discord. Species is a taxonomic rank, and we use species=, not taxon:species=, same for genus=* instead of taxon:genus=. So why taxon:family= was choose instead of family=* ? I don’t have Discord so I can’t check it. I think it would make sense to have family=, genus= and species= OR taxon:family=, taxon:genus= and taxon:species= since the trees tagging situation is already a mess as it is (again, 20 different tags for palms only, and we are not even talking about cultivars ecc.).
I would also be happy with more tagging consistency. We would ruffle too many feathers and be in conflict with the KISS principle if we went for taxon:everything tags, and I find plain taxon=* too unspecific, so my vote is on family=Arecaceae.
This is just guaranteeing that barely anyone uses it as (as a first order approximation) no one knows the Latin name.
Edited to add: Even when you think you know the Latin name you might not know it. As a non-tree example one of the most culturally significant marine species where I grew up has had it’s scientific name changed twice since I went to school.
There is a broad choice when it comes to tagging tree specifics, species:en for instance is already used 210k times species:en | Keys | OpenStreetMap Taginfo
species:de even 412k
“species”, which ideally is tagged in latin, which also seems to be the actually the case, has more than 1.6 million: species | Keys | OpenStreetMap Taginfo
On the contrary, I settled for taxon:*= because I can agree the plain form is unclear. While the majority of unique family | Keys | OpenStreetMap Taginfo is for taxonomy, ignoring palms, family=yes has been used on US =toilets reasonably. taxon:*= is more scalable. There are a few taxon:subfamily= , many taxon:cultivar= (mostly imported, over cultivar= ), and ~9k taxon:species:varietas | Keys | OpenStreetMap Taginfo (imported). They can easily be interpreted as part of taxonomy, unlike family= and subfamily= alone.
