Greetings. I’m havin some kind of an issue. What’s happening is that, we’re working on a project to upload all observations we can insider our university. Normal observations are appearing but some observations aren’t and I know what’s the problem.
The species that aren´t showing inside the project are the ones that are in some category either by CITES or by the IUCN. I understand the reason why this is happening and in some way, I’m happy that the platform thake this kind of precautions but my question is that if there is a way to show this data as the main objective is showing how diverse our space is and maybe create some kind of conservation program.
It would really be helpful if there’s a solution to this, but like I said, I can understand why the precautions.
Thank you for your attention.
Can you please provide specifics such as URLs, examples, and screenshots? It’s very difficult to provide advice without those things.
I suspect it’s due to this: https://www.inaturalist.org/pages/help#placeindex
in case it isn’t fully clear:
if you upload an organism that is at risk, its location will be automatically obscured (if the curators have made it so for that species). that makes the location data’s accuracy very low.
if your project’s target area is relatively small, obscured organisms observed within that area may not be included because their accuracy radius is not completely included within the bounds of your target area.
if you enlarge your target area, it may help.
It’s actually a huge problem that experts can not receive the “obscured” data on the threatened species from iNat, and it even goes like that on GBIF. Because of this “displaced” data from iNat some threatened species can appear to be more widely distributed than they actually are and their conservation statuses can be wrongly assessed as not threatened, so there will be no effort to protect such species as a result. It can literally result in decline or even extinction of some species. Of course if we’re talking about species being hunted or harvested it’s a right thing to hide coordinates, but 99% species declining only because of the habitat loss and exact data on their distribution can only help to save those species and their habitats, not otherwise. There is a recent paper about this problem https://www.sciencedirect.com/science/article/abs/pii/S000632072300143X
Hi tiwane!
Of course, here I attach a screenshot of one of my observations
As you can see this one is evaluated under CITES Criteria.
And this one is an observation from a friend that’s uploading to the project too https://www.inaturalist.org/observations/172248887 and has also being obscured.
Thanks! Yes, I think this because they’re obscured and are thus not publicly indexed as being in this place.
There are a few workarounds:
- if the taxon isn’t threatened by location disclosure, you can flag the taxon and explain why you think it shouldn’t be obscured. For example, the IUCN says of Mastigodryas reticulatus:
The major threats to this species are deforestation and habitat change due to agricultural expansion, settlement and cattle ranching. There is some oil infrastructure development on the coast, but this is localized. This species occurs in a heavily-populated area with very few protected areas, and where most suitable dry forest habitat has been lost in the last 50 years (D. Cisneros-Heredia pers. comm. 2016). This species may be subject to localized persecution due to local beliefs that it is venomous (M. Yánez-Muñoz and J. Valencia pers. comm. 2014).
I’m not familiar enough with this specific situation, but you could probably argue that location disclosure might help protect this species more than harm it, if its main threats are development. To flag a taxon, go to its taxon page on the iNat website, click on Curation, then click on Flag for curation.
- You can make a “traditional” project for obscured taxa found on campus, then put that new project and your collection project under an umbrella project. That umbrella project would show all the species found there.
But there isn’t a simple solution to this, unfortunately. There are always going to be trade-offs when it comes to deciding which data should be publicly accessible.
Thank you again tiwane. I’ll talk about this with my partners in order to communicate this situation.
I certainly agree that there are tradeoffs with obscuring data. Obscured data is much less useful for certain analyses or goals, but there needs to be a balance between accessibility and protection of species. How to strike that balance is an area for reasonable debate.
I took a quick read through of this paper and it seems as though the authors are intentionally misusing iNat data and then saying that misusing the data is a problem. It’s a pretty poor straw man argument for at least two reasons:
- The authors know the limitations of the data that they are using well as they describe the obscuration process of iNat fairly accurately. They then create a similar (but seemingly different, see notw below) obscuration process and apply it to unobscured data. They then use these pseudo-obscured data points as though they were correct and say that the outputs are significantly different than those produced using the correct/unobscured data. Of course the outputs are different!
The reason this is silly (beyond the obvious) is because any data with poor precision/high uncertainty should not be used in this way. If a geospatial datapoint has low accuracy/precision, it shouldn’t be used for the type of modeling or calculations that the authors are doing. There are non-iNat datapoints with low precision too - they should also be filtered out and not used for certain analyses. As someone who has actually done the types of modeling and geospatial analyses the authors dohere, I can say that they have pretty much just created an analysis pipeline where they skipped the critical data QC checks and used inappropriate data for their analysis and then complain that the outputs are flawed. The main takeaway from this should be:
Always make sure that you understand your data and do your quality checks - as always, garbage in = garbage out.
The authors even acknowledge this in the Discussion:
“Given that iNaturalist appropriately reports increased values of the uncertainty of each record (i.e., they populate the Darwin Core field coordinateUncertaintyInMeters; Wieczorek et al., 2012), the use of data records with high associated uncertainty is ultimately the responsibility of the users of the data.” This is the case for all data! (not just that from iNat).
- The authors make a big deal of the fact that a high proportion of data for some threatened species on GBIF is from iNat where it is obscured, and that, based on argument 1 above, the assessment of these species can be riddled with errors. Though again, this is only the case if these data are used inappropriately. However, the authors (and any other researchers) are free to (and indeed should) not use any data that they don’t think is valid for use in their analysis. If they just screen out all iNat data, they are no worse than they would otherwise be in a situation where iNat never existed. iNat data has not had an adverse effect, except in the situation where it is misused which is the fault of researchers using it, not the data itself.
NB: the methods sections isn’t written with enough detail to actually assess (which is a problem in itself…), but it looks like their methodology doesn’t actually replicate iNat’s obscuration process. The procedure described on page 3 of the article describes creating a separate bounding box for each observation, but this isn’t how iNat obscuration works (boxes are set for a given area). While I don’t think this would produce hugely different results, it should lead to obscuration with more noise than the actual iNat process.
Well, it’s kind of strange to argue that a paper published in the top journal of it’s field is a rubbish. The main fact here is that “obscured” displaced coordinates from iNat being transferred to GBIF. Here is an example of a very rare mammal with very limited distribution in very limited habitats being recorded obviously outside its distribution as a result of “obscured” observation from iNat:
https://www.gbif.org/occurrence/3113410502
Some people are using datasets from GBIF to assess conservation statuses of some species. Authors of the article has shown on their artificial examples that data from GBIF is basically unusable as a whole now, because of the observations from iNat. If someone wants to use it then each record should be checked individually, which loses whole point, you can’t do it with thousands of records. The most unclear thing for me here is why GBIF is still uploading such data.
It’s not strange at all. When I was working on my PhD, a paper was published in Conservation Biology (one of the top journals in the eponymous field) that stated the group I was working on at the time was entirely extinct. And it was written by Paul Cox, a prominent conservation biologist.
Also, GBIF and similar databases are well known to be riddled with many kind of errors. This is the GIGO problem (“garbage in, garbage out”), where aggregators like this obscure the flaws in the sources during the aggregation process.
I should add, obscuring coordinates is not just about protecting commercially exploited/exploitable species as the paper says. It’s also about protecting direct and indirect habitat destruction from well-intentioned people who simply want to see rare species. There have been many cases where people damage the area around an endangered plant by walking up to it, eventually leading to it dying. For this reason I also obscure records of common taxa observed on the same day.
Yes, and this is something that iNat and similar platforms/databases will always be reevaluating. There won’t ever be a perfect balance struck that will please everyone or work in all cases, unfortunately. But we’ve tried to mitigate the issue - which is a real one - and provide workarounds. They’re not perfect, but they exist.
Currently if you’re interested in obtaining true locations of obscured/private iNat observations, you can:
-
flag the taxon and make a case for why it should not be automatically obscured.
-
reach out to observers directly and ask for coordinates.
-
reach out to the observers and ask them to “trust” you on iNat.
-
reach out to the observers and ask them to join and trust your project.
-
reach out to your region of interest’s iNat Network partner organization, if there is one. They get exports of most auto-obscured coordinates for observations made in their region.
not to mention their statement that “According to iNaturalist (https://www.inaturalist.org/pages/help#geoprivacy), the original, unmodified data for threatened species can be obtained only by writing to the authors of each of the records” [emphasis mine] is just blatantly false. They provide a link to the help page and still get it wrong; if you follow that link, there are multiple ways beyond ‘ask everyone’ explicitly listed there:
In all these cases, the following applies:
No, it is not true for
This requires asking a single person
@igor117 @jasonhernandez74 Do you have suggestions for how to improve the way iNaturalist handles this? The issue of obscuration and access to hidden coordinates is a known and admitted one, as is the possibility of causing harm to species/habitats from openly publishing location data. Ideas for better and more scalable solutions would be appreciated.
As an example of the opposing view (that locations of rare and endangered species should not be made easily available) there are also high-profile publications, for example an article in the journal Science titled “Do Not Publish”. There’s a nice article in Yale Environment 360 that references this and other work to discuss the difficult trade-offs involved in making location data available, or not, for particular species.
On the first place, you should simply give unobscured data to the affiliated scientists on the request, after basic verification.
On the second place, there are many large taxonomical groups, not just singular species, where obscuring coordinates gives nothing. It’s the case for most of the invertebrates. Even if invertebrate species is being commercially used on some occasions it doesn’t means that numerous poachers will go on exact spot to find this exact species, it’s not how it happening. Species should be expensive and there should be some significant market for someone to bother that much. Also the populations of invertebrates are mostly very numerous and few collected specimens without significant damage to their habitat would mean nothing for their conservation. So among invertebrates it’s a problem mostly for some butterflies and very few other exceptions. It would make sense to revise it first on the level of groups and to make it possible to flag and obscure/unobscure coordinates for all species within large taxonomical groups all together.
Moreover, in IUCN Red List it’s clearly indicated for each species if its use is a threat and this data can be checked in the advanced search. Among 42,108 species listed with threatened categories there are 2056 species threatened by “Hunting & trapping terrestrial animals”, 2157 by “Gathering terrestrial plants” and 2020 by “Fishing & harvesting aquatic resources”. Ok, also 2326 by “Recreational activities” (obviously very few of those because of the interest to these exact species, but let it be). So why to obscure distribution for other 34,000 threatened species (80%) if there are no any related threats to them?
This is something I think we’d like to do in the future but don’t have the staff capacity at the moment to do it at scale. Some sort of data manager position would be cool.
I’d love to get data on this, I’ve heard many differing opinions from various stakeholders and people experienced with wildlife trafficking.
That’s an interesting idea. FWIW here’s how we automatically imported IUCN statuses a few years ago.
I think this would be cool (and a cool job!), but I also think that there are multiple potential challenges.
- There are multiple reasons that data is obscured - one is via automated obscuration based on taxon status, the other is observations “manually” obscured by users. This approach could work for observations which are auto-obscured, but I think it would be highly problematic for user-initiated obscuration.
Most users want to maintain control of their data, at least to some extent, and obscuration is one way that they do this. Data-sharing/privacy concerns are already a reason some folks do not use iNat. If users knew that their data was accessible by scientists on request, it might have a chilling effect, leading some users not to use iNaturalist or reduce their usage even if they did continue or start using it.
A change to this model which included allowing scientists access to manually obscured data would also, I think, require an opt-in setting for users who have previously uploaded data that they’ve obscured. If users weren’t given a choice in this, I suspect a good chunk might delete their accounts or similar, which would be a major loss.
-
As a practicing scientist, I do not trust all other scientists with data. There have been repeated past instances where publishing geospatial data, even to describe a new species, has lead to the poaching of that species and conservation threats. Here’s one instance, in pop sci format: https://www.smithsonianmag.com/smart-news/poachers-are-using-scientific-papers-to-guide-them-to-their-next-victims-180948082/
And notice I am not really blaming the scientists/authors here - the potential misuse of geospatial data is a challenge even for scientists who want to use/publish data responsibly. There are other scientists who are careless or ignorant of conservation issues surrounding species that they work with, and even a couple instances where iNat data was used carelessly or inappropriately could reduce trust in and usage of iNat. I will also note that scientists are often not experts on whether the species they are working with are actually threatened by collecting or not - natural resource managers and government folks often have more expertise in this (even than IUCN reports, many of which are old/less correct than we might hope…), and they should ideally be consulted. One other potential issue here is whether scientists from some other country (like Global North) are making decisions that benefit them/their science without appropriately consulting people in the country where these organisms are found and are generating the data - there’s the potential for some pretty colonial science to occur here, I think. -
Providing access to obscured coordinates on demand would require creating guidelines for who can/cannot receive access, how they prove it, and also a data-sharing agreement that scientists would need to agree to I would think - not trivial.
One other potential complication is that the system of asking for access to obscured data from iNat couldn’t be done through GBIF (at least not in any existing functionality I know of). Using iNat data via GBIF is preferable by far to accessing it directly from iNat. So the pathway to make requests to iNat is not one that is simple to implement and would take a lot of time and resources. I think this challenge could be overcome with time and resources, but this would be a tradeoff with other areas iNat could grow.
Given these challenges, I think that using the existing mechanisms in place to access/create less obscured data is likely the quickest/most efficient way to see less obscured data. This includes:
Working with iNat portals can pretty much allow scientists access to auto-obscured data in a country for which a portal exists already (and this would have done for all three species in the Biological Conservation paper cited above btw, at least if the scientists’ request was approved).
Removing unneeded autoobscuration on iNat is not too difficult. If scientists have good reason to think that a species does not benefit from obscuration (or obscuration in a part of its range), they can provide support and ask for this via a flag - it usually happens pretty quickly in the cases I’ve seen. I think the idea to start with using IUCN assessments to see what species could be targeted for deobscuration from @igor117 is a good one, but I would still want someone familiar with the species to make this determination and not do it automatically. One additional reason for this in addition to the ones above is that threats are not constant across species’ ranges - a species may be abundant in one place where limited collecting/disturbance will not hurt populations, but rare in another where it is more vulnerable. Additionally, there are situations where poaching/hunting varies with culture, so obscuration might be necessary in some areas but not in others.
Lastly, I would point out that I think the precautionary principle is called for when making deobscuration decisions. I believe that less obscured data is a good thing for science. However, once the data has been deobscured, whether on iNat itself or via publication by a scientist, it is out there - there’s no putting the cat back in the bag, so we want to be very sure that the decision to deobscure is the correct one.