Obscuring on map vs lists

i think because the obscuration box is still within the bounding box of the place (different from the park boundaries)

1 Like

The project is not defined with a bounding box, it is defined with an exact mapping of the park boundaries.

Here is the project - check the map at the bottom https://www.inaturalist.org/projects/algonquin-provincial-park

here is the place the project uses https://www.inaturalist.org/places/126945

Orange (irregular) - park boundary
Red (large rectangle) - roughly drawn bounding box (see FAQ)
Blue (small rectangle) - roughly drawn obscuration area

The project does not pick up records from any kind of bounding box, it solely gets them from records within the exact boundaries of the park. Here is a record of mine from just a couple of hundred meters outside the park, and well within any bounding box you can define.

It is not in the project.


Right, the bounding box would only apply to obscured observations and those with the true centroid within the place, to my understanding.

  1. Open, accuracy circle is totally within the place and bounding box — in place
  2. Open, not totally within place, but yes bounding box — in place
  3. Obscured, not totally within place, but yes bounding box — in place
  4. Open, not totally within place and breaks bounding box — not in place
  5. Obscured, not totally within place and breaks bounding box — not in place
  6. Open, centroid is outside the boundary, but the accuracy radius touches the place — not in place (right?)

@bouteloua has it correct here. Nice diagram! So basically, the size of the place boundary (and, to some extent, where in the boundary the obscured observation was seen) determine whether or not an observation is indexed as being within that place (both on the map and on the place’s list). The idea is that you can’t use a small place within the obscuration area to narrow down the location. Certainly not perfect, but it helps.

Can you provide an example here? If it’s sensitive, you can send it to help@inaturalist.org.

I agree with all of @cmcheatle’s points here. Right now trust is predicated on the observer’s informed decision on who they want to share their obscured/private locations with, and this is done via traditional projects and user-to-user trust.

1 Like

wow, that diagram is awesome. Can we get it in the on-site Help wiki if it isn’t already?

I do like the idea of being able to set places and bioblitzes to collect all of those excluded observations and also every other obscured or uncertain observation that intersects… that would have saved us a lot of trouble in our bioblitz as not that many people were roaming outside the boundaries anyway. But i wouldn’t think that would be great as the default.

I’m not sure what you mean here.

Thank you all for a very informative and thoughtful discussion. If someone really wanted to find out where a certain observation was using the strategies discussed above, they would need to create a lot of new places that would overlap with each other. Could not the process of detecting unusual clusters of overlapping new places be automated? The circumstances under which one would want to create a large number of overlapping places in a pattern that would allow for the localization of an observation seem quite limited. A reasonable minimum size for places to show obscured observations on their lists could also be implemented to contravene a gridding approach.

I am currently running a place-based conservation project on iNat that has run into significant problems because all of the protected organisms on this imperiled land don’t show up on our project. I may resort to contacting observers who have detected anything obscured anywhere nearby to find out if it was on this property.

Not really sure I want to provide an example of that considering that with the current way decisions are being made iNat may decide to remove the observations from the AOI list was well, compounding the problem that many of us are facing.


I think you’re interpreting my words correctly, but I was thinking of really large places like Yosemite. If something was observed near the edge of Yosemite, it’s obscuration box might break Yosemite’s bounding box and thus it won’t be listed as being seen in Yosemite (example 5 in the diagram) whereas something seen right in the middle might be listed as being in Yosemite since its rectangle would fit entirely within the bounding box. But then it wouldn’t really matter in that case since the entire rectangle is already in the place. If you are thinking of a different scenario, please message me.

That’s not what’s going on in my area. I have several observations that are exactly in the middle of a roughly 400 km^2 area with only a tiny corner of their obscured box projecting past the edge of the defined project area and those observations don’t show.

If it was uncertainty based, 55% of the box (for example) many of the observation that are missing from the list of the project area would be included.

Clearly something else is going on.

You can essentially do that with the current system.

With any current project if you have a rare species that’s obscured, but the entire obscuration box is within the boundary of the project so the species shows up in the list I could make a radial array of new projects shifted by the proposed 100m and see which projects the observation remains in and disappears from. Do that a few times and I can pinpoint the specific boundaries of the obscuration box and know within which area I’m going to need to search to find the species. Considering that if I’m that dedicated to do this I probably already know a lot about the habitat requirements/preferences of the species, so once I have that obscuring box and a decent map of the area it’s not much additional work to find where it is (or is most likely to be).

This can be done with the system as it currently stands. It would take some work to do it manually.

Of course, it’s much easier to just zoom to an AOI on the Explore map, set a category filter, then scroll through the observations looking for your species of interest. I can do that currently and it’s much faster.

The point of the numerous simultaneous discussions on this topic has never been about the specific point on the map, it’s been stated by myself and others repeatedly that we are fine with the map location being obscured (within a reasonably distance so that valuable ecological information is not completely lost), the issue has been about the lack of inclusion of those species on the list of observations for any given project, as that simple presence/absence information is absolutely vital to conservation management approaches.

The response to the the boundary box/project movement hypothetical proposed earlier was in reply to a somewhat off-topic though experiment presented about why XY or Z was a bad idea and potential ways of gaming the system. I was demonstrating that the reasoning employed didn’t fully hold up because you can use the current system to identify the boundary box for an observation (and yes, we are all aware that the point shown on the map is not the specific point of the observation) and use basic knowledge about the species to hone in from there on where the species in question is most likely going to be found.

As someone working directly on poaching issues and managing anti-poaching teams, my experience is that the poachers know enough about their target species of choice that even the current system doesn’t provide any additional security, and that it may actually do more harm to conservation efforts than anything else.

1 Like

right. that’s what seems to be happening in my tests.

I agree. Serious poachers have access to data and techniques (and for that matter money, insiders and corruption) that make many of these obscuration attempts worthless.
This obscuration is surely for opportunistic poaching/collecting, and must be carefully weighed up against negative effects such as allowing local enthusiasts to police and monitor populations (using iNat if needs be) for threats and activities.
A secondary factor that needs to be taken into account is that even highly desirable pets/plants are not usually under threat if they are easy to propagate/breed in captivity. In these cases, rather than obscuring data, breeding programmes may be a more useful way to tackle the problem.

We also have had claims that putting data on citizen science sites has led to their “extraction”, but to date none have been conclusively verified**. A few cases have been due to porcupines “poaching” the bulbs, and a few succulent cases were poached without being posted. But concerns do run high.
e.g. Mokala National Park wanted all observations of Rhinos on the reserve removed from iNat, even though the entire reserve comprises only two 20X20km blocks, with rhinos in both, and the reserve advertises Rhinos as one of its main attractions.
**A major exception is a Spanish succulent expedition to South Africa that extracted data from CS sites, Red List sites and other online data on a spectacular scale (and were confident enough to have all the data on site when caught), but despite all their homework, they discovered an unpatrolled area of the Knersvlakte Nature Reserve and just hunkered down and collected and shipped out everything suitable for several days before being discovered. What was most alarming (apart from their audacity and scale of operation) from their records is that their intent was to explore unknown groups of plants presumably to test their novelty value and potential market.

The real issue here is that obscuring the data also hides it from the conservation agencies that are supposed to be protecting and managing the populations. So a reserve manager / specialist scientist / anti-poaching team may not become aware of a species in their conservation area, until after the better resourced poachers/collectors have been in. It is not just a case of obscuring data but keeping it usable at the same time. Bearing in mind that the observer may not be aware of the ID, or of the obscuration, or of the data trusting procedures/limitations on the site. And that for many heavily utilized groups electronic data is not nearly as useful as local habitat knowledge, with collectors relying on the latter and conservation agencies on the former.

Obscuration has its pros and cons, and I dont think the disadvantages of obscuring data (or for that matter, the technique or areal scale of obscuration) have been explored adequately. We all intuitively understand the need to obscure data for various reasons (esp. rarity, trespass and land value), but the efficiency and consequences of this on subsequent data value are not so obvious.


it is important to note that for very small conservation areas, the majority of the centroids for threatened species will be outside of the reserve and its bounding box, and those species and populations will not be recorded from the reserve on iNaturalist.
e.g.https://www.inaturalist.org/places/harmony-flats-nature-reserve#taxon=47126 is well bioblitzed, so many of its plant species are probably (I cannot check) on iNaturalist, but only 70 of the 215 species (as per the checklist, although only 20 of these species are IUCN threatened Red List species and thus automatically obscured) are shown on iNaturalist as being in this 9 ha nature reserve.


Tony: Thanks for posting that. I agree and have noticed similar re: obscuring. I work and interact with with tons of local conservation groups including lots of informal ones - just groups of neighbors wanting to protect their land together. While there are some valid arguments for obscuring (another one is if landowners want to obscure things on their land to monitor for trespassing or access), there’s also very little data on cases where iNaturalist does cause a net increase in poaching, and unfortunately it’s very hard to impossible to document time when iNaturalist stopped poaching due to increased visibility (poachers thrive in secrecy) and awareness and community involvement. Thus in the least we need to limit obscuring to species where there’s a clear danger because when you obscure things with no conceivable danger you end up with all the downsides but no upsides.

And i can speak to my area in general: the biggest threat to wetlands in Vermont is development. We are lucky to have rules to protect them, but rules are imperfect. The best way to improve rules and resist development is awareness. iNat s one of the best ways to do that as it has become huge in our area. Even if iNat results in some people yanking pitcher plants from wetlands (they are obscured… but their locations are well known) and some slight uptick in trampling, it has increased awareness greatly and also connected a lot of us creating an informal conservation organization of its own. i have no doubt the result was greatly positive on Vermont’s wetlands… AND if obscuring were more overzealous that positive impact would have been lessened a lot. We have tiny towns we take pride in with our species lists and such, as all of Vermont is divided into small squares around each town. Things that are obscured can’t be tracked by town, much less town forest. And people here know the land and care about what is happening where. Please help us keep this wonderful resource by not increasing obscuration!


I’ve noticed that as well. I would hate to “obscure” the date, as observations have phenological information in many cases, but maybe the time of day could be obscured.

1 Like

There has been discussion of auto obscuring an entire day’s observations if any one thing that day is auto obscured (which I don’t think is a good idea at all fwiw). Considering that, I think obscuring date is a lesser “evil”. But in any event it’s a known issue.