Is anyone aware of a way to connect a feature layer from ArcOnline to iNaturalist? We use Survey123 to collect field observations that contain information and photos of species, and we’d like to import all of them to iNaturalist.
As I’m sure many of you are aware the iNaturalist API is very extensive and could be useful in this case. ArcOnline also has a pretty good API, but I’m not skilled enough to connect the two. Has anyone tried this before? It seems a waste to have thousands of potential observations sitting in a database locked down from a community like iNaturalist.
I don’t know enough about the API to help you myself, but users with more experience check on the Forum periodically.
As for a mass transfer, if the sample is sufficiently large, iNat staff may be able to give you a hand.
To what end do you want to import the data, though? iNat isn’t a data aggregator, it is meant primarily to be a platform to engage users with nature.
Also, there are issues of ownership / copyright when it comes to importing images you did not create. And in general it is not recommended to upload observation data from something you did not personally experience.
Thank you for your thoughtful response, I’d be happy to discuss the purpose of the data transfer. I welcome any feedback or thoughts regarding this topic and whether the community deems this an appropriate use of iNaturalist.
The purpose of the transfer is to help identify wildlife observations along Montana highways. Montana suffers from a lot of roadkill along major highway-wildlife corridors, both in the sense of wildlife and human health. What a team and I have been doing is cycling major segments of highways and recording wildlife (alive and dead). We’ve been using Survey123 as there were a lot of other parameters to collect, as well as other benefits of using the Esri platform for data analysis and sharing.
The issue we are trying to solve now is the lack of identification down to species of these observations. We aren’t exactly qualified naturalists with the necessary skills to identify all of these species accurately. The hope in transferring into iNaturalist would be to involve the broader naturalist community to help us identify these observations.
As I mentioned above I welcome any feedback regarding the use of iNaturalist for this purpose.
I’m not sure I’ve never tried. You should be able to test it pretty easily, but I have a feeling it won’t work. It does seem like something iNat should have an application for. Hopefully you aren’t working with very large datasets (like thousands of observations) because adding them separately would be tedious.
That’s an interesting case. I think that @astra_the_dragon did a good job pointing out some things to think about. Copyright would be a big one, but if this isn’t a huge group and everyone were to agree, that might be ok. There is some precedent for group accounts on iNat (like for classrooms) that are administered by one person in charge.
Otherwise, this certainly seems like a dataset that would benefit from community input and does represent individual interactions of people with organisms.
One other thought would be that iNat doesn’t want a bunch of observations to just get uploaded and then not be maintained/responded to. It sounds like your team is interested in the ID process, but it might be a little tricky to figure out who might maintain the account, respond to comments, etc. if there are a bunch of different people involved.
I’ve no idea on the API side, but it sounds like an intriguing project so hopefully you can figure it out!
are you intending to replace your current system with iNaturalist, or is the idea to keep your current system going and duplicate the observations in iNat?
if the latter, this seems problematic because it seems like the primary purpose of connecting your system to iNaturalist in that case would be to leverage the power of the identification community and AI in iNaturalist, and that seems to go against the spirit of what iNaturalist is about.
Thanks for your thoughts pisum. This touches on the point that trh_blue brought up in the first response. The current plan is to keep our system of data collection to utilize the robust GIS processes that are available in the Esri platform. The reason for bringing the observations into iNaturalist is to leverage the power of the identification community. The AI is less of a factor here, as many of the observations are roadkill that the AI is likely not trained to identify.
On the flip side these are still valid species observations with photos, location, and physical descriptions. In many cases we’re able to get closer to species as well for greater detail. I personally have gotten more rattlesnake photos with very detailed photos of their patterns (much closer than I would’ve have gotten otherwise).
To respond to cthawley’s inquiry about record maintenance above, I would be the main point of contact for curating these observations. I would certainly encourage my team to help out, but I am the lead on this aspect of the project if we were to take it on.
I haven’t been an overly active iNaturalist user so I am curious to hear more about the spirit of what iNaturalist is about. I feel like I have a sense for it, but I think based on the fact that we’re having this conversation we have different opinions about what that spirit is. I would like to hear from other users, as well as you pisum, what your interpretations are of the spirit of iNaturalist and if it fits with this potential use case. Thanks all for your thoughts thus far! This has already been a fantastic discussion, and I’m excited to see where it leads.
Hi christawhite, yes I don’t think the csv upload will work unfortunately. And yes we have 3,371 observations that we would like to add now. That was just this year, and we’ll be repeating for the next 2 years with more people as we add more to our team. So we’re projecting another 10,000 to 15,000 observations by the end of 2022. I would very much like to avoid a manual process :)
i’m not personally aware of anyone having made this kind of connection between AGOL and iNaturalist before, but it doesn’t look particularly complicated as far as working with APIs goes.
as trh_blue noted, the primary purpose of iNaturalist is to engage people in nature and to build a community of people interested in nature (see https://www.inaturalist.org/pages/what+is+it). so leveraging iNaturalist primarily for identification is not really in line with that core mission, at least from my perspective.
there are cases where folks have made systems that collect data from their users and ultimately feed that data over to iNaturalist in the name of the organizations running those systems (which some folks refer to as “group accounts”). these are probably not ideal from the perspective of iNaturalist because it fragments the iNaturalist community, which is something that iNaturalist explicitly asks you not to do. ideally, any individual contributor to iNaturalist should be able to interact with any other individual contributor, which is hindered when there are group accounts. the group accounts also generally don’t properly attribute and license the observations within iNaturalist.
as far as I can tell, the iNat staff don’t often crack down on these because i’m guessing most of these cases are helping engage people in nature more than they are causing problems. but i can think of at least one case that became problematic. coincidentally, in that particular case, as far as i could tell, one of the the primary reasons they connected to iNaturalist was to leverage the identification community here.
Thanks pisum, that all makes sense. The geodatabase is indeed the only method for extracting photos from a feature layer (outside of an API) which I think will be the crux of making this work.
My hope in getting these observations into iNaturalist is to engage my entire team in the identification process. My guess is a lot of them haven’t used iNaturalist before and would be new users contributing to the community. I discovered iNaturalist from a similar project.
I think I’m going to go ahead and try it. But from what it sounds like, no one else has done this before. I guess it’s time to learn how to build APIs. Thanks everyone for your input!