Yeah, there's a lot missing from open data projects to make it easy.

I think the "correct" way forward is as follows:

1. On wikidata, add "subclass of (P279) Fruit Tree (Q904653)" to all the taxa which are, well, fruit trees. You can use something like QuickStatements to help with the process;
2. Then, extract the list of taxons with this property from there, including both the wikidata IDs and labels in English (which seem to always be the latin name). This can easily be done with SPARQL;
3. Then do something similar to https://overpass-turbo.eu/s/269o but for all the taxa we've extracted, and turn it into a standalone app (maybe use https://openrecyclemap.org/ (https://github.com/meta-systems/openrecyclemap) as an "inspiration" - shouldn't be too difficult to adapt it here);
4. Finally, add all the fruit trees you want to OpenStreetMap, using existing tags like species or genus etc.

This is a lot of work but if you're into it (and know both the subject matter and javascript) doable within a month or two of after-hours work.

If you decide to go for it, feel free to ask for help here and if I know something from the top of my head I will try to help you.

PS

Actually, I just realised that everything is mostly here already on the Wikidata side. What I actually wanted is something like this: https://w.wiki/EVgg

This does produce a list of all fruits and all species/gena from which you collect those fruits, sorted by "popularity" (sitelinks). It's like half the interface for our supposed map already!