I’ve been facing this issue for quite some time and to me it seems like a serious bug / flaw which I was hopping to see fixed. But maybe I am missing something.
Let me explain and let’s choose the island of Sao Miguel in the Azores as an example.
When looking for species, let’s say all mammal species, and using the iNaturalist place concept (in this case the Ilha de São Miguel, PT, AC Island) you basically find 10 species only. The most obvious miss is the endemic Azores Noctule bat.
However, when drawing a rectangular or circular area, as close to the space shape as possible, you find 32 mammal species instead. Of course, drawing a box or circle around the island, you capture a few sea mammal species, but that’s just some of missing species.
If you compare number of observations, you get 12318 when using the place vs 15943 when using a rectangular or circular area for the same place
This situation is not only specific to this island but virtually any place / location around the globe. There is always a serious “delta” of missing data for species/observations between drawing a box or choosing place.
I very much like the “Place” concept and its UI and also the inherent checklist coming out of it, however I find myself not using them or at least not trusting them when looking for the species present in a certain region or place.
There must be some bug on how a Place aggregates data from observations, as it seems to miss a lot of them.
Or maybe I am missing something.
Any thoughts anyone?
There are a number of things that explain the discrepancy.
First, the list of 10 is a checklist. Checklists are known to be a little buggy and are not recommended (and will be phased out at some point), plus they don’t include casual observations.
If you instead use Explore to list mammals on the island, and you allow for casual observations (because your list of 32 allows casual), then you get 20 instead of 10.
So what is missing between the 20 and the 32 from the drawn circle? It’s not just “a few sea mammal species”, the missing 12 species are 11 sea mammals and the bat that you mentioned. The sea mammals are easily explained by the difference in the shape between the two places, and the bat is missing because it is threatened and the locations are obscured, more about geoprivacy here.
If you want to include near shore observations and obscured observations, I would recommend using the standard (“official”) places for the municipalities on the island. You can string them all together in the URL and then bookmark it, or you can make a project that uses them.
In the specific case of the endemic Azores Noctule bat, it appears that all of the observations on/near Sao Miguel are obscured, perhaps due to its conservation status. As such, the automatic obscuring of the locations (to a random point in a large rectangle) necessarily causes the “accuracy” of all those sightings to exceed the limits of the place defined as Sao Miguel, so they won’t be collected in any list for the island. To be included in a checklist or any “Explore” results, both the point of an observation and the entirety of its accuracy circle must be encompassed in the defined polygon for the place. (I’m not sure I personally like the way this functions, but it is what it is.)
Thanks for the pointers, although they don’t fully explain the discrepancy.
Perhaps choosing mammals was not the best idea for an island, given that inside the circular area you get the sea mammals.
Here is the plants’ observations comparison: 7035 when using the place vs 9332 when using a rectangular or circular area for the same place
We are talking about a significant difference.
To me, the way the observations are allocated (or rather not allocated) to a “place” is buggy and not the checklists.
Again, I very much like browsing all species by place (in the example the plants), the UI is nice and the also checking the linked checklist for the place. I wouldn’t phase out the place-linked or auto-generated checklists, they are useful.
Hope someone will look into the root cause for this or shed more light
the first result set here includes only verifiable observations, while the second includes all observations. if you limit the second result set to only verifiable, you get 7726 observations.
here are 694 verifiable observations that fall in the circle but not in the polygon.
if you look at this set of observations, you’ll see that the others have already provided a lot of explanations for the existence of these 694 observations. the only other thing i’ll add is that there are also unobscured observations that have really big positional accuracy values which will prevent them from being found when filtering on the smaller place.
@pisum - Thanks! Good pointers! It would be then great if someone would have a set of checkboxes for the place:
- include unverifiable observations in the list of species
- include obscured observations in the list of species
Probably and hopefully with these 2 checkboxes enabled, the results will be identical, while allowing to browse species by place with the better UI.
These checkboxes could be then carried over to the checklist corresponding to the place
this already exists in the Explore page
this is already done by default. but as others have noted, this won’t work if you’re filtering on a smaller custom place where the obscured coordinates fall outside the bounding box of the place. this is just how the geoprivacy functionality works.
you can filter for only obscured observations by adding parameters manually in the URL. see https://forum.inaturalist.org/t/how-to-use-inaturalists-search-urls-wiki/63#heading--geoprivacy.
it’s not identical. as i noted before:
it’s unlikely the place checklists that currently exist are going to be changed. as i understand it, there are already plans to make a new “dynamic” version of place checklists that will be similar to the dynamic life lists for users.
Thank you for all clarifications! Much appreciated