So there was a recent blog post celebrating many taxa having their ranges mapped, which is great. So I checked some arthropods and the results were strange.
I could bring up many more examples. The question I have is, why? Is it some kind of error, or is this intentional? If it is intentional, what is the basis for it? Is it counted if someone saw a cockroach on their ship? And could this be corrected by simply eliminating all occurrences in ocean grid cells in their range maps, if necessary?
EDIT: I checked the cellar spider, drain fly, house fly, house centipede and bedbug, and it’s the same thing for all of them. I think this affects all cosmopolitan synanthropes. Still doesn’t answer why this is the case though
If they disguise the location close to the shore, the pin may land out in the water. This is what I commonly see on lakes and such. The other most reasonable explanation is boats/planes
I know there are grid cells that overlap the land and sea, but that’s not the same thing as vast swaths of ocean being included in the range of fully terrestrial animals.
For the zoomed in pic ( Hadrurus arizonensis), the geomodel hexagons do largely seem to contain coastline (land) - it’s just that they also contain ocean as well (though there might be a few that are all ocean or have real small islands. That’s just a limit of the resolution of the geomodel probably.
For things like Culex pipiens and Periplaneta americana, this looks like a case of the geomodel not performing very well. It’s not the most sophisticated modeling approach, so things like this might happen. Certainly something that could be improved I would think.
That may be the case. The map for the Red Nail Gall Mite (Eriophyes tiliae) is pretty good, though there does also seem to be a few pure oceanic cells highlighted as well for some reason. I think there should be a filter to automatically delete all occurrences in purely oceanic grid cells for purely terrestrial animals, and vice versa should it be an issue. This of course wouldn’t apply to species that frequent both marine and non-marine environments.
In my excitement to show this new feature to some friends, I quickly came across a coastal dune plant (sea oats - Uniola panicula) with a geomodel that extends well into the Atlantic and covers almost half of the Gulf of Mexico. I don’t have time to figure out how to include a photo but I’ll try to do so later.
But, it doesn’t seem to be just arthopods. I do assume these models will improve over time with new updates, I’m looking forward to it.
It uses AI analysis of observations taking into account elevation, so it probably sees observations at sea level and treats the ocean nearby or in between the seal level observations as similar habitat, since it is also sea level (being the sea and all…)
I don’t think this matters too much—I can’t think of a case where this would affect something. And anyway, what if someone does find an arthropod on a local boat? Would likely be accurate
I think there must be something going on here other than it simply being ferries/boats. In Panthera onca’s map, it shows the blue on several islands and cuba- places where Jaguars do not and have never occurred. It also shows a massive extension into California, and while Jaguars were present there historically, there has not been any documented in the state in over 200 years, so something else must be going on.
Are official range maps ever shrunk to show the (assumed) current range, or do they only ever expand when things are found in new areas? I wonder whether the issue here could be just that old records are retained indefinitely, rather than perhaps using a rolling window.
On a related note, evidence of presence is easy - if you find it, it’s there - but how can one be certain of absence? Is five years without a sighting enough? 50? 500? If you want to change maps to remove areas, I assume you’d need an official cutoff time.
What if you want to make a species richness map by overlaying all the range maps? A range map showing the house centipede living across vast swaths of the ocean is really dumb and uninformative.
I looked up the Jaguar map under Taxon. Those California and Cuba dots aren’t on the map. My guess is that they’re Casual – records from zoos, probably. So that inaccuracy is a result of the question you’re asking, not any error in the dataset.
I was typing a very similar response when I saw your message I completely agree.
Honestly, I don’t even think this really needs to be improved. It provides the basic idea of where their range is. Sure it could, but the reason I think it might not be worth it is:
In order to incorporate that the code involved would likely be very complicated. For example, I don’t know much about species that frequent both marine and non-marine environments but if they were in the same genus/family of non-marine arthropods adding in the code to ignore ocean cells from them would cause a noticeable problem. Maybe allowing curators to change the grid cell mistakes would fix this problem? Regardless, I don’t think any simple fix would solve the problem fully and if you can’t solve the problem fully it wouldn’t be worth it.
Simple. Lets use hemiptera as an example. The vast majority of them do not occur naturally in marine environments. So simply code all hemiptera species so that occurrences in the ocean are flagged and removed. But then there are marine water striders in the genus Halobates, so code them specifically to differ from all the other hemiptera by having their marine occurrences not flagged.
Sorry to double post, but I noticed something interesting. There seems to be at least a few instances where the Geomodel maps genera rather than just species. I found two examples, Demodex and Manticora