Range maps for vascular plants

It definitely comes from a much more painstaking research process, though it still has its occasional accuracy and documentation issues, like any such broad compilation.

1 Like

‘widely accepted ranges’ are often missing a lot of stuff, and for reasonably identifiable plants in areas with a fair bit of iNat use, the iNat range maps are actually better than the published ranges in a lot of cases. So i agree that we really need a curated mix of the above. Some of the iNat range maps have extraneous points thanks to cultivated species or the ID algorithm, but most are pretty easily cleaned up.

I use or at least think about range maps a ton: what part of a plant’s range we are in tells us a lot about what it indicates when found in a wetland or other ecosystem. Then there’s cliamte change and geographically limited tree pests, etc etc etc. The de facto range maps are a lot of the reason why i am here.

1 Like

The iNat map for Lomatium hooveri accessed through the taxon is actually okay in the sense that it doesn’t have a dot for the observation in Placer County. But it still highlights Placer County, so I don’t know whether that is what is causing it to appear on the Placer County checklist. If that could be fixed somehow, I wouldn’t have such a problem with it.

So just investigating, and looking at the Lomatium hooveri checklist entry for Placer County, it shows the list entry was:

Added on April 23, 2018
Updated on April 23, 2018
Source: Observation: Hoover’s lomatium on April 21, 2018 at 01:59 PM PDT by cloudya

and

First iNat observation on April 21, 2018 in California, US by cloudya, added April 22, 2018

Both Occurrence Status and Establishment Means are currently set to Unknown (the initial defaults).

From this I infer that the checklist entry for Placer County was added automatically by the system based on the location of the random obscuration point for this observation. I think this is the issue that needs to be addressed.

I didn’t want to try any edits to fix this, both because it’s still an active observation, and because I didn’t want to change anything that staff might want to investigate here. But if I was going to try fixing it, I could either set Occurrence Status to “Absent” and see what happens, or just delete the checklist entry. If I did the latter, though, I wonder if the system would just find and add it again?

1 Like

I suppose if we did pursue populating atlases and/or checklists via GBIF data globally, we could carve out geographic/taxonomic spaces where a better source is available and would be used instead, like BONAP for North America north of Mexico (assuming we could even get access to the data). NatureServe for rare species might be another potential overlay.

But we would have to keep in mind that even projects like BONAP may be behind iNaturalist and/or GBIF data for some things. Big compilation projects like that always have inherent lag times.

So whether we used GBIF as a single source, or overlayed additional sources, any auto-population effort would probably end up doing the easy 80-90% of the work for us, and we would have the harder 10-20% of the work left in the form of manual corrective curation. Realistically that would have to be on-demand, as particular errors surfaced that needed to be addressed. Just the nature of beast, I’m afraid.

1 Like

Thanks for investigating this. Toward the end of this other thread https://forum.inaturalist.org/t/geoprivacy-obscuring-and-auto-obscure-discussion/457, we were coming to the conclusion that it is showing up in Placer County because the accuracy circle is 110 km, before the obscuring. So I was thinking the huge accuracy circle is what is causing it to end up on the Placer County checklist. I also think the checklist entry was added automatically by the system, but based on the original location and the large accuracy circle as well as the random obscuration point. I don’t understand why, given that the point is gone, Placer County is still highlighted on the taxon map, or how that interrelates with the checklist. If the large accuracy circle is the reason, then maybe checklists could be made to not include those observations with very large accuracy circles and the maps could be made not to highlight the area they fell in?

Yeah, it could have been the large (in)accuracy circle too, hard to tell just by looking.

But to your question, it looks like the system does the easy addition to a place checklist when it detects a new observation, but then doesn’t do the harder work of reassessing that checklist if that observation changes or goes away later (there could be other corroborating observations by that time). Problems like that get left to the curators to resolve as they surface (though any user can edit place checklist entries).

I think the suggestion to require a minimum accuracy threshold before an observation can automatically affect place checklists is a good one. How to pick a number?

1 Like

Good question. I am having a hard time figuring out how an observation with a very large accuracy circle can really be considered to have a location.

2 Likes

I would actually be surprised if the large uncertainty circle is the cause of it being on that checklist. I don’t know California geography that well, but I would expect a 110km circle to touch more than 1 county, yet only this single one has it added to the checklist. Without being able too see the real location, it is impossible to judge, but I would bet the mapped centroid is in that county.

2 Likes

A quick calculation shows that the distance between the obscuration rectangle and the main cluster of observations is ~90km, so a 110km circle would cover all observations of this species. The checklist and GBIF are obviously using the random “obscured” point in the obscuration rectangle. The obscuration rectangle is actually far more precise (and less accurate) than the original observation. Obscuration rectangles are 0.2 deg x 0.2 deg, so ~22km x 22km, but narrower in the east-west direction the further you get from the equator (~18km on this observation). The rectangle contains the center of the unobscured observation’s accuracy circle, but is not centered on it (obscuration rectangles form a fixed grid aligned with latitude/longitude lines). The point displayed (and apparently exported to GBIF) is a random point in the rectangle which was chosen once and saved, so the same point is shown to everyone.

2 Likes

i don’t even see why we want these observations showing up anywhere. I understand for some cases of marine fish or albatrosses or such they have some conservation value, but for plants they certainly do more harm than good to tracking biodiversity and should just be blocked from any map displays or places imho

1 Like

@cmcheatle @JeremyHussell I agree, it had to have been that obscuration point that got picked up both by the iNat place checklist (and thus the taxon map) and by GBIF.

If I am understanding correctly, the point is only gone from the iNaturalist range maps because of a decision to hide sufficiently inaccurate observation points.

But the underlying point still exists as part of a research grade observation, and until that is addressed, it’s unlikely that it will disappear from either GBIF or the Placer County checklist.

The submitter is still an active user, so hopefully they would respond to questions. There are at least 4 distinct scenarios I can think of but am not qualified to judge

  • it is mapped incorrectly
  • it is identified incorrectly
  • it is mapped and identified correctly but is planted
  • it is actually valid

OK, I left a comment on the original observation, pointing to this discussion and asking the observer to help us out.

But it remains a larger use case that we need to consider for constructing range maps in general.

1 Like

cloudya responded in her observation and fixed it https://www.inaturalist.org/observations/11352005

I wonder if it would be possible to alert observers during upload about their large accuracy circles and/or “first records” in a place, since she was unaware of both issues on her observation? It’s great that she was still around to fix it, but no one noticed for about a year, so if it could be automated maybe they would be caught sooner and before the observer becomes inactive.

Another thing that might help is if we had the ability to reverse the order of species in a “place” search–I had to ask how to see the least-seen species in a place, given the limited number displayed, but by finally getting to the least-seen in Placer County I found cloudya’s observation. It would be helpful if the least-seen were easier to see because that is where the not-supposed-to-be-there species are going to be.

2 Likes

I have just had a request from a researcher looking at alien and invasive plants.
They have natural range maps for pine species which could be added as range maps or atlasses (what is the difference?)
They want to extract the data of these species outside of their natural range - whether planted or going wild.
The question is simply: would it be possible to filter and download observations that occur outside of the area of natural range maps/atlasses? Or are such filters only possible on place names?

(this covers it, but is not very promising: https://forum.inaturalist.org/t/how-to-find-out-of-range-observations-wiki/4271)
Using places is possible: https://www.inaturalist.org/observations?locale=en&not_in_place=14&place_id=any&taxon_id=53421&verifiable=any
But often places (even many) are a very poor second option to range maps/atlasses.
Is there a “not in range” option or some other way of easily accessing observations extralimital and invasive/planted?

1 Like

The range map is a polygon, while atlases are essentially a visualization of checklists at the country, state, and county levels. Some parts of the tree (like mammals) have gotten range maps from outside sources; most plants don’t have maps (see conversation above).

Here’s a mammal with both a range map (the pink polygon) and an atlas (the green political boundaries).

Depending on whether your target plants have good atlases and already have out of atlas observations, you may be able to get somewhere with those. If your species already has out of atlas observations, you can follow the link on the atlas page to view out of atlas observations – this just constructs a query for not in all the atlas places.

For example, this is the first plant I saw in the list of atlases with out of atlas observations. Clicking on View sends you to an Observations query that only has the out of atlas observations – they can be downloaded in the normal fashion by clicking Filters -> Download.

If you want to use range maps instead of atlases, I would imagine you can upload those polygons as new place boundaries and then search not in that place for out of range map observations.

2 Likes

That’s a great explanation by @jwidness.

As far as I understand it, @tonyrebelo’s researcher friend (or someone with curator rights and willing to assist) could add range maps for these species of interest. I’m assuming the maps are accurate and follow iNat’s guidance. Those range maps would then display precise known boundaries in relevant iNat map views but wouldn’t by themselves provide the ability to query based on in/out of range. Here’s a tutorial on adding ranges.

In addition (or separately) someone with curator rights could create an atlas for each of these species. That would allow a search to reveal observations outside of the atlas area, but would be less precise because an atlas is a combination of whole countries / states / provinces / counties / etc. Here’s a guide to creating Atlases.

Another consideration here is “natural range” vs. “introduced range”. I believe iNat ranges do allow for the concept of native and non-native ranges. This is how @bouteloua explains the native/non-native treatment by atlases (from the guide linked above):

Displaying establishment means and observed/unobserved on atlas presence places

On taxon maps throughout the site, establishment means (native/non-native) is indicated by a dotted line and color indicates whether there are research-grade observations associated with a listed taxon (green) or not (orange). Currently, atlas maps are not displaying whether there are observations associated with atlas presence places or not. They are showing establishment means as follows: if any relevant listed taxa have native (or endemic) establishment means the atlas presence place displays as green, if all relevant listed taxa are introduced they display as red. There is currently not good atlas UI for setting establishment means.

Calflora now offers two range maps for all 10,000+ plant species that grow wild in CA: one based on watershed and one based on climate model. To find a range map for a specific species, search for that species at www.calflora.org then click on Plant Range to the right of the map. Here for example is the range for Verbena lasiostachys (Common verbena): https://www.calflora.org/entry/dgrid.html?crn=8233. Here’s more about Plant Ranges on Calflora: https://conta.cc/3sqgpiY

2 Likes

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.