Hi all, just seeking people’s opinions on an issue I recently came across. One of the taxa that I work closely with has a range map on iNat that is very incorrect - it both excludes large areas where the species is present and includes large areas where it is absent (overall <50% overlap with the true range). The range map is from IUCN which is slightly concerning (luckily it is not a threatened species), but I can at least partially understand why there are some errors as the species in question is part of a relatively poorly-studied genus (although the inclusion of some areas in the range map I have no explanation for!).
My initial view was that this incorrect map should be deleted from iNat. However, another user is of the opinion that because it is an official IUCN map, it should be retained on iNat. I am interested to hear what other users think, and what a good course of action would be in this case. Personally, I cannot see any reason why an incorrect map should be retained, but I would be happy to change my mind if there is a good reason to keep it!
I generally dislike the IUCN maps as they are often inaccurate. Curious: If you look at the range map for that species on the explore page on iNat, does it show a reasonable range for just iNat observations?
You can flag the species on IUCN and discuss with a reviewer. If you’re going to do that you might go on GBIF and take a look through the ones in obviously bad areas. It shouldn’t be that hard to find ones that have faulty coordinates, and if it’s a species that is photographable, you might even find some that are clearly wrong. Finally search around for any papers discussing the species’ range.
I’ve seen inaccurate shaded range maps for some mammal species on iNat which might be based on IUCN maps or another source. Some of the errors are a matter of scale where the original was probably rather approximate/simplistic to begin with. Others are just outdated. Unfortunately inatters may use the shaded maps as a guide to assigning species IDs when the mapped range is not accurate.
The individual dots in the wrong place on the map are easy to fix when they’re based on iNaturalist’s own data – click on the dots and re-identify them. The larger areas (countries, states, counties) that are colored in must have been imported from elsewhere and I don’t know the solution for them.
I checked a Carex that has been split taxonomicly (Carex atrata). Some of the wrong places are legacies of the name changes. Interestingly the large part of what used to be considered its range in North America is no longer included in its map, except for Mexico and a county in Colorado. Also, mountains in southern Asia aren’t included in the map, again due to taxonomic changes; C. atrata seems to grow there. The map on POWO is more accurate, I think.
A species by species fix within iNaturalist is probably not practical. Can we communicate with the source of the problem? Would they fix their maps? Does iNaturalist renew its maps from this source occasionally?
If there is a published source for a better map, that would be ideal. You can create a flag and suggest using the more accurate source. We are not tied to using IUCN maps, they just happen to be convenient, so it is entirely possible to update to a more accurate map on iNat, provided someone has the time, skills and interest to create one.
On iNaturalist we have essentially three ways of understanding the distribution of a taxon: (1) the observations themselves, (2) the ‘atlas’ which gives context to ‘listed taxa’ to completely encompass the distribution of the taxon, and (3) the ‘taxon range’. While the first two are more akin to raw data (i.e. an observation is evidence that the taxon occurred at that point in space and time, while an atlas is a statement that all individuals of the taxon should be encompassed by the listed taxa), the taxon range is more of an abstraction of this data. For example, the edges of the taxon range don’t claim to be exact, the area within the range doesn’t claim to harbor a uniform density of individuals, and individuals may occur in the area outside the range.
My understanding is that we should aim to have an atlas for every taxon on iNat, as this is really just a list of all the places we expect that taxon to occur, e.g. “United States and Mexico”. Even if distribution data is very sketchy, anyone with curator rights can still create an atlas that includes the countries where iNat RG and other GBIF observations have been recorded. If we have better data, we can break this down to the state or county level.
The bar for a range map is higher, as with that we’re trying to show the actual boundaries for where the taxon typically occurs. We would exclude vagrant animals and “waif” plants and draw boundaries that include only part of a state, province or county as needed. Reliable info doesn’t exist for a lot of taxa, and so we shouldn’t create range maps for them. There was an assumption that range maps imported from sources such as IUCN would be accurate, but that’s not always true and we should try to fix those where we can.
The big obstacle here is that creating range maps requires an iNat user to have three things:
Have sufficient knowledge of the taxon to find, evaluate and synthesize research on the distribution of the taxon.
Have access to appropriate GIS software (some is free, like QGIS) and skills to use it to create a KML file.
Have curator access to permit uploading the new range map.
I went to the taxon page for the sedge. I clicked on “Map.” That’s what I looked at. From what you say, it has #1 (the observations themselves) and #2 (the “atlas,” which is whole political areas colored to indicate the plant grows there). For Carex atrata, that “atlas” is in someways inconsistent with what is known of the range of Carex atrata sensu stricto, though partly consistent with the range of C. atrata as that name was formerly used.
The POWO map is also an atlas, apparently, a map of political areas where the plant is thought to be, and it is more accurate than the map iNaturalist has.
I wasn’t referring to any maps of the 3rd type, the ‘taxon range.’
There’s the increasingly out-of-date range map editing tutorial, which hopefully lowers the skill barrier. If there were some demonstrated interest in updating it, I could do that. (Or someone else could, if they were so inclined.)
That all makes sense. Actually, there are very few plants with iNat range maps, probably because a lot of these came from IUCN, which mostly created them for animals. I believe the original question from @matthew_connors was about range maps rather than atlases, but we should probably try to improve both.
If you’re a curator and you’re familiar with an organism, it would be great if you could review the associated taxon atlas and adjust it to include the countries/states/counties where the organism occurs and exclude the ones where it does not. If you’re not a curator and you see something wrong, add a flag on the taxon.
Any curator working on a taxon split should definitely review and fix the atlases for all related taxa (including sister species) before committing the split. Taxon atlases are the way iNat avoids unnecessarily bumping observations and IDs to higher levels.
Changing taxon ranges requires a more secure knowledge of distribution. If it’s clear that the current range is wrong and a better one can be drawn, then it should be fixed. Curators can do this themselves (by following the guidance linked earlier); other users should raise a flag.
Thanks for the responses everyone, great to have some discussion. A few points of clarification:
I’m talking about Range Maps (#3 in @rupertclayton’s excellent explanation above) rather than atlases or just the observations themselves.
I have manually reviewed every observation of the taxon on iNaturalist and to the best of my knowledge they are all correct - the species is luckily quite easy to identify from most photos.
I am a curator and am happy to remove the incorrect range myself, however I have no better/more correct range map to add in place. The distribution of the species is still a bit fuzzy but it is known well enough that the current iNat range is certainly incorrect.
I had not considered adding an Atlas but I think that is a good thing to do and I will add one shortly, as it would not be too difficult at least in terms of broader geographical areas.
I guess the crux of my question is: Should demonstrably incorrect range maps be kept on iNat for any reason?
Depending on what basis it is known that the range map is incorrect (absence of records is not necessarily evidence of absence of the species) it would seem justifiable to download the current range map, edit it to remove the parts that are clearly wrong, and add areas that are clearly part of the distribution, and use the edited version as the new range map. Even if it’s not perfect, it’s an improvement. Remember that some of the IUCN range maps are basically just redrawn from blobs drawn on maps by the creators of field guides. Some are very good and are based on a lot of data and knowledge, while others are less so.
If they are demonstrably incorrect, best to update them and state clearly the evidence and sources for doing so, that this information appears in the Taxon history. The main reason incorrect maps will persist is lack of curator time and experience to update them, or lack of sufficient evidence/reliable sources by which they can be corrected. Another possibility is that some maps may show historical range, while species may have expanded or contracted since. In such cases, again, the best course would seem to be update the range maps where feasible.
Thanks @deboas, that’s helpful information. I think my main issue here is that nobody is in a position to produce a corrected range map. I am very confident that the current range map is incorrect for a number of reasons - there are more than a thousand observations on iNat alone and at least several hundred museum specimens, but I am still in the process of revising and catalogueing everything so I am not particularly confident on the exact boundaries. If it wasn’t clear, I’m currently working on the taxon in question!
If it was only a small range I would be happy to ‘estimate’ a range map, but the species occurs over an enormous area - at least 3 million square kilometres - and any attempted ‘blobby’ map that I could make could easily be out by more than 500km in some areas. It would be better than what is there currently, true, but I think it is of limited use until more work is done. On top of that, I’m not actually sure how to make a range map in the first place.
So my reasonable options are: Keep a range map that is demonstrably very incorrect, or remove it entirely. In my opinion it does more harm than good to retain an incorrect map even if there is no corrected map to replace it with.
In my opinion, the maps drawn by man or computers inherently have some inaccuracies. It is a useful tool but does not exactly represents the actual range of some species. It makes work easier for identifiers to isolate species.
I’ve often come across grasshopper species having an entire country as its range. In reality, the species only exist in restricted areas in that country. There may be spurious records of grasshoppers in iNat on another continent, such as asian species in America. This is due to the iNat’s CV recommendations is going that direction. It keep thinking the species is there, but it is usually wrong ID maybe due to the species looking very similar. I guess have to weed out the spurious records in individual country, which usually is 1 or 2 records. Range of species can span several countries. It may be dynamic, fluctuating with weather or being transported to new location by man. or its range was previously underestimated. So I think maps are not that accurate even when technology is very advance. We may have to make assumptions which we may not be able to verify all the time, like we will assume pictures are correctly identified.
That makes sense. I think you’d be justified in removing the inaccurate range map, and updating the Atlas (political regions within which the species occurs) to serve as the best current guide to the geographic distribution of the species.
Update: the iNat help site says (my emphasis): “If a taxon range does not match what we mean by a species due to taxon splits or other taxonomic mismatches, they should be removed or altered. However, all range maps are imperfect abstractions and it is not worth replacing an IUCN range for small errors in the distribution (“range should extend east to border”) unrelated to taxonomic mismatches. Use your discretion to decide if a range is sufficiently misleading to warrant removal.”
