Import of iNaturalist observations

12 posts / 0 new
Last post
Barry Walter
Import of iNaturalist observations

I recently discovered that iRecord has started importing observations from iNaturalist. I have been using both platforms, so there is a lot of potential for duplication (a verifier recently informed me of one). My initial reaction to this is to wonder whether I should stop using iRecord in order to prevent any further duplication. What exactly are the plans going forward? Can I now assume that all of my research grade observations on iNaturalist will eventually find their way onto iRecord?

When it comes to entering records, the iNaturalist site is vastly superior to iRecord. I can automate the entire process (including adding photos), so it saves me a lot of time and effort. On the other hand, I greatly appreciate the feedback I get from the iRecord verification process, so I don't want to cut myself off from it entirely.

I found a thread on the iRecord issue tracker which gives some insights into what might be implemented. However, it would be very helpful if someone could explain what to expect from a recorder's point of view. Ideally, I would like to use iNaturalist exclusively for uploading all my records, but still get some sort of feedback from iRecord. Is this a realistic prospect?

Matt Smith
iRecord is a UK based system

iRecord is a UK based system with records verified by specialists / recording scheme experts, and with links to the NBN and Local Record Centres.  Both LRCs and National Recording Schmes can access iRecord data.  iNaturalist is a US based system where IDs are crowdsourced.  Data in iNats goes no further so it is essentially "lost" to the UK biological recording community, which is one of the reasons this link was developed.  I don't think there are any plans to have any feedback on records or verification going "the other way" and updating iNaturalist from comments or decistions made in iRecord.  I would recommend using iRecord as your primary recording platform adds iNats adds nothing "extra" to the record.

Barry Walter
Re: Import of iNaturalist observations

I have been using both platforms for quite some time, so there is no question that iNaturalist is a superior system from a recorder's point of view. This is no surprise really, since they have far more resources at their disposal than iRecord has, and have therefore developed a much more sophisticated interface with many more features. This doesn't mean that iRecord is a bad paltform - it just means it will always have far less to offer than iNaturalist (again, mostly from a recorder's point of view).

For now, I will think I may just stop using iRecord. I will certainly miss the feedback from the verifiers, but it's not a deal breaker. A much more important issue is the question of exactly which records will be imported. From what I can tell, the current thinking seems to be that only research grade observations will be imported. I can see why this is a sensible approach (as least to start with), but I think special consideration should be given to well-established iRecord users who also have an iNaturalist account. For them, I think there should be a way to explictly link the two accounts so that all their records could be imported from iNaturalist, regardless of their current research status. To put it another way: it shouldn't matter which specific portal I use to upload my records - so long as iRecord sees me as a trusted user, it should just treat them both the same. Ideally, iRecord would also continue to use their existing notification system to inform me of any changes that were made to such imported records, so that I could manually update my iNaturalist observations where necessary.

Thanks for giving your views

Thanks for giving your views on this.  We are likely to explore further interactions with iNaturalist - you are certainly correct that it has considerable funding behind it.  We would be interested in hearing your perspective on the advantages of iNaturalist (from a recorder's point of view), so this can inform any improvements we might make.  David

Work load

There are now many duplicate records from the iNaturalist imports.  I'm not a verifier, but and verifiying must be an onerous task already, and I should think this is just another added burden for them. I really value verification of records. Also these duplicates are annoying when searching.

I can't remember when I first signed up to iRecord but wasn't there an agreement not to duplicate records on other systems? I use iSpot, but mostly for queries.

Barry Walter
Re: Work load

Yes, the terms and conditions asks that you do not "submit records to this site that have already been submitted to ... another online wildlife recording system". But since it never defines "online wildlife recording system", it is unclear whether or not this applies to sites like iSpot or iNaturalist. Certainly they both can be used in that way, but that is not their stated priimary purpose. Despite this, I understand that iRecord has previously imported records from iSpot, even though it is primarily intended for identification purposes only. And as for iNaturalist, they describe themselves this way:

iNaturalist is an online social network of people sharing biodiversity information to help each other learn about nature.

and, more to the point:

It's NOT a science project. The data generated by the iNat community could be used in science and conservation, and we actively try to distribute the data in venues where scientists and land managers can find it, but we do not have any scientific agenda of our own aside from helping to map where and when species occur.

All of iNaturalist's research-grade observations are available as a GBIF dataset. If iRecord has now chosen to make use of this, it is entirely their responsibility to deal with any potential for duplication (just as it was when they took data from iSpot). Until very recently, the only reason I have continued to use iRecord is precisely because my iNaturalist observations wouldn't otherwise be passed on to the various recording schemes in the UK. But I now have a difficult decision to make. I have quite a lot of iNaturalist observations that are unlikely to become research-grade any time soon because there is currently no one with the expertise/inclination to confirm my identifications. If I don't submit these to iRecord, they will probably never find their way to the relevant UK recording schemes. I am not going to stop using iNaturalist just to avoid this problem (which was not of my own making). So, for the time being, I suppose I will just have to risk creating the occasional duplicate on iRecord. I have already suggested a fairly simple way for these issues to be avoided - but if iRecord chooses not to implement it (or something similar), that's just too bad, I suppose.



Matt Smith
Duplicate records are not

Duplicate records are not really a problem.  It is fairly easy to filter these out and to end up with only 1 "good" record on a system, although the duplicates should be retained in there somewhere, just not mapped or referred to.

As for iNats "research grade" records, there are only a limited number of experts out there, and a limited number of hours in the day one can spend verifying records.  When iSpot first came along I was set up with an "Expert" level rating to help get the system off the ground.  However, now iRecord is here, I spend most of my time Verifying records on here, I really don't have "spare" time to do the same on iSpot or iNaturalist and more, though I do pop in occasionally.

Barry Walter
Re: Duplicate records are not really a problem

Although I don't like the idea of creating duplicates, it's good to get some positive confirmation that these won't cause too much of a problem on iRecord. So I think I am just going to stop worrying about this and carry on as normal.

The main problem I see with the verification system on iRecord is that the status of a record is much more static than it is on iNaturalist. Once an identification has been determined, it's unlikely that it will ever be looked at again unless the recorder edits the original record. The community identification system on iNaturalist means that there are usually many more pairs of eyes looking at an observation, and it is quite likely to be reviewed multiple times both before and after it has reached research-grade (or whatever you prefer to call it). But a much more important aspect of community identification is that it actively promotes the development of expertise. Anyone can suggest an identification, and they will usually get feedback on it quite quickly (often with explanatory comments). A lot can be learned by everyone from this simple process (even the experts). In the long run, a self-training and self-correcting system like this will become much more powerful than a static verification system, because its pool of records, recorders and identifiers is constantly improving.

The biggest problem that iRecord has is that it does not really have a coherent community that encourages knowledge sharing. The recorders are largely isolated both from each other and the verifiers. Even worse, due to the vice-county stucture, they are also isolated from other parts of the country (and the rest of the world). This introduces an inherent bias to the whole system, because some vice-counties are much better served with verifiers than others. Getting positve feedback from your contributions therefore becomes a postcode lottery. I happen to live on the border between two vice-counties. One is quite active, but the other is a virtual dead-zone. There are many posts on this forum from people complaining about this situation; very often, frustration gets the better of them and they leave and never come back.

Overall, I get a very strong impression from iRecord that it cares much more about the data it receives than the people that provide it. With iNaturalist, it's the exact opposite.

Import of iNaturalist observations

Thanks to Barry Walter for his comments.

Although I agree with much of what he says I would add a word of caution about his belief that the dynamic nature of iNaturalist will continue to improve the expertise of observers through the continued feedback. The downside of this is that if an identification is incorrect then others can learn wrongly from it and become experts in confident misidentification and then lead others in the same way. Does it happen? Yes indeed. There are a number of records of Arion ater, the Large Black Slug. They have images of a large black, or mostly black, slug. Unfortunatley the other three (or more) species in the sub-genus Arion (arion) can also have the same appearance. All that it needs is for two people to agree it is Arion ater and the record becomes research grade and the recorder will continue to submit dubious records. The comments given by the iNaturalist 'experts' generally agree it is Arion ater but do not suggest submitting images showing the foot fringe or sole nor whether the characteristic rocking behaviour of Arion ater has been observed. The nomenclature is also not up-to-date and I have the feeling that the experts' confirmation of the species is based on outdated (i.e. 40 year old) field keys.

At first I redetermined the records as Arion (arion) agg. which was safe (but not very useful) and gave what I hoped was helpful, encouraging feedback to the recorder. I have since discovered that iNaturalist does not pass iRecord feedback to the recorders. I have, therefore, just rejected all Arion ater records from iNaturalist unless there is convincing evidence that it is that species. The species group is common just about everywhere so the aggregate records add no real value to our national data collection.

I should say that not all species are as awkward as this one, and there have been some excellent slug records from iNaturalist,  but that the verification system is a two-edged sword and records need to be treated with understanding.

For what it's worth I won't

For what it's worth I won't be accepting any records that come from iNaturalist - some users seem to be using it to get ID's rather than submitting records, which it perfectly valid use of that website. The problem is they therefore are not concerned with accurately recording where they saw it. As a result I can't trust any location data as I have no idea if a record has really come from  the location attached to it. I can spot the one's that clearly are wrong (The chap with a string of BAP species from a diverse range of habitars all tagged with his back garden springs to mind) but the plausible is simply impossible to verify. Yes they photographed a Red Admiral - but was it where the system says it was, or did they see it on holiday? Is this user a concientious naturalist who took care to get the location right or just some guy who wants to know what the pretty butterfly is called? I have no idea. At least if they are submitted via iRecord the user intended to submit it.

Barry Walter
Re: Import of iNaturalist observations

Thanks for your feedback, Chris.

I understand your concerns, but the most effective way to address them is via active participation. Community identification only has a downside when people fail to share their knowledge. So if you see a long-standing problem and are confident you have a good solution, then fix the affected observations and add some comments explaining your reasoning. There is no point complaining about misidentifications if you aren't prepared to contribute constructive feedback yourself. Most people will quite happily engage with you once you start a dialogue - especially if they disagree with you!

You also seem to have your conception of the relationship between iNaturalist and iRecord backwards. Feedback must go from iRecord to iNaturalist, since the latter is just a passive source of data. In the short term, the simplest way to do this is for verifiers to set up iNaturalist accounts and add the necessary corrective identifications and comments themselves. In the l;onger term, an interface could be added to the iRecord verfication system to do most of this automatically. This should not be a very thing difficult to do, since iNaturalist already provides public APIs for precisely this kind of purpose.


Barry Walter
Re: Import of iNaturalist observations

Thanks for your comments, harasseddad - but don't you think you might be throwing out the baby with the bath water here? My immediate reactions to your concerns are very similar to the ones I gave in my reply to Chris above. The situation cannot improve if you don't provide constructive feedback. Every iNaturalist observation has a Data Quality Assessment attached to it, which covers location accuracy (amongst many other things). This allows any community member to flag problems like the one you mention, which will in turn prevent an observation qualifying as "research grade" (as long as enough people agree).

I cannot stress enough that iNaturalist is above all a participatory community. In the long-term, I think iRecord will have to engage much more actively if it wants to take full advantage of what iNaturalist has to offer. It is clear that iRecord has a great deal of expertise that could both help iNaturalist improve their data quality and also help their users to become better recorders and/or identifiers. But that cannot happen without active participation.

Log in or register to post comments