Strengths and Weaknesses of iNaturalist Data

I am doing an article about iNaturalist for the Woodland Observer, a newsletter of a naturalist club in North Bay, Ontario. I have a short list of the strengths and weaknesses of iNaturalist data. Anything missing?

Strengths:

  • Documenting aberrants
  • Documenting visual difference between members of the same species. These can include variation over ranges, potential subspecies, phases etc
  • Recording species new to science
  • Documenting ranges of species, range extensions and maybe range contractions. (Note: GBIF is usually better, although in some cases Needs ID data can supplement GBIF)
  • Documenting phenology
  • Can expand the value of specimens
  • Early detection of newly arrived invasive species

Weaknesses (what iNaturalist’s data is usually not good at)

  • Figuring out population size of a species
  • Figuring out population trends in a species
3 Likes

I would see the phenology data as a weakness. While the flowering phenology graphs are interesting they are biased towards flowering you only need to look at the seasonality graph to get an idea of a species flowering time, not many images in the system that document the flower bud developement or fruit development over time, at least not for the species I have been targeting.

You have to be careful with the range expansions. As has been discussed in other threads, some people uncritically agree with the computer vision, and then another person uncritically agrees with the first one; as a result, there are taxa at research grade on the wrong continent. If the intention is to get your readers to use the site, I recommend that you explain this so that hopefully they will not do this themselves.

9 Likes

Yes, a weakness I would include is that identifications are not always right. Like any community science project, IDs come from a wide spectrum of experience and expertise, and sometime people exceed their abilities in offering or agreeing with IDs. Identifications should be vetted by the data user and not taken on faith.

That said, this is not a weakness peculiar to iNaturalist. Even long established scientific specimen repositories have some rate of identification error, and in my experience iNaturalist does fairly well given its different and diverse data sources.

8 Likes

One of the biggest (perhaps the biggest) strengths of iNat data is the sheer volume. With so many people plugging away at observations it becomes possible to do things that would not otherwise be possible, provided, of course, that the researchers are willing to curate the data.

I think it’s important to be clear that the data are not iNat’s actual objective. It’s about expanding public understanding of the natural world. The data are a bonus, not a primary objective.

13 Likes

One of the most notable traits of INat data is its heavy dependence on user density. Data on widespread mass taxa actually reflect observer density better than distribution of species. For the most taxa, Moscow is the biodiversity hotspot of Russia, while the fact is the opposite :)
The same stuff works with specific taxa also. By selecting “Insect” observation, you can virtually count registered macro photographers on the map. Underwater observations are almost limited to popular dive sites, which is especially evident in the case of soft-bottom fauna: almost all observation of them come from Lembeh, Ambon, and Tulamben(popular “muck diving” dive sites), while these species likely live everywhere between, but on larger depths, boring sands or in just in places with no diving infrastructure…

9 Likes

Weakness: record frequencies of certain species. From my field of expertise (lichens): flashy species get more attention,so the equaly frequent and even recognizable but less eye-catching species will be recorded less. This impacts distribution maps, too. Areas where identification resources or experts are available, will have higher species numbers,even though in reality the localities are not so rich in biodiversity. Strength: huge number of observers, among which some may upload something really outstanding.

6 Likes

Add “documenting escapees” to the list, one of the most important functions of iNat.

One thing iNat does poorly is tell you what species are NOT present. Often in conservation it is just as important to know where a species isn’t found, and iNat doesn’t document this (eBird does, one of the strengths of that site).

1 Like

To represent iNaturalist’s data honestly and completely, it’s important to realize that many of the problems of iNaturalist data also are problems for scientific collections in museums. Incorrect IDs, incorrect locations, etc (as well as the two that you mentioned originally). For species that can be identified by a photograph, it’s actually easier to correct IDs of iNat observations than it is for scientific collections.

11 Likes

Agreed. Unlike a specimen sitting in a museum collection, the photo records on iNat are there for all to see and evaluate. Wrongly IDed specimens in a museum can languish for years or decades – while the data associated with the specimen (e.g., on Arctos) potentially gets used in publications – until a specialist comes along and examines/corrects them. And there might be only one person doing the ID. Some museums are starting to make available photos of their specimens, but for some taxa that still is not helpful. Which is also a shortcoming with iNat records – some taxa just aren’t identifiable based on photos.

5 Likes

Unless the observer has opted out of community ID, one of many examples of how iNat is fundamentally a nature social media network at its heart and not a scientific database. I often work with insect specimens in museums that were collected generations ago and they may have been misidentified or georeferenced improperly, but at least an expert can eventually fix or flag that when they visit the collection. There is no plan for correcting such data here after the users who observed it drift away or die short of demoting the observations into “casual” oblivion. User-controlled private locations are another example.

1 Like

Actually it can do this very well provided either good data is available to populate a checklist, or you have more advanced knowledge of how to access the site.

There is a difference between ‘a specific observer did not note this species on this outing’ which is what ebird can do well vs. the species is not found somewhere.

With wrong ids there’s a sister problem of not marked captive/cultivated organisms, thousands of them.

5 Likes

At least people can bounce the ID back to genus/family/etc. or force it into “casual” status which researchers avoid. So there is a way to skirt around this annoying feature.

I still remain perplexed by the “opt out of community ID” being a feature as it has no positive result. The only useful data are correctly IDed specimens and I think everyone using this site wants to know the correct ID for an organism. Who would even want incorrect ID’s, or go through the effort of using the site and having an inaccurate list? I know some people are confident in their own ability, but if they are so sure they should have full confidence that everyone else (including experts) will agree with their ID. Does anyone have a situation where someone would opt out of comminity ID? I’ve never had an instance where people mass ID the wrong name and refuse to retract it.

But back on topic, the use of only two agreements to get “research grade” is a problem as I’ve encountered many observations IDed to species and a friend of that person agrees and so it goes to “RG” and no one else is able to catch the error unless they vet that entire species for errors. I would think that there should be a tiered system requiring more votes.

A really good thing are the developing range maps which should become the working model for all field guides and making accurate maps. If a species has a large number of obs (1000+?) you could consider that range map to show a good part of the actual range of that species. For species like moths the range maps are far better than any other resource as they show the modern distribution very accurately. For some species just spotting outliers allows you to find incorrect ID’s just based on range maps.

1 Like

Not directly related to data, more to postings. But as a frequent poster who does it for phenological reasons, hoping to be part of many more observers to give the data validity, I have been contacted by researchers to monitor specific species and in one case requested to get a speciman for a herbarium. Although my photo conditions restrict only commercial use of my photographs, I have been contacted by educational and public institutions to use some of my photos or authors who wish to include them in their articles. I was able to inform the invasive plants coordinator of a distict in SW Alberta where there is a provincial Invasive Weed Council about checking on posts within their area of jurisdiction to determine when particular species on their watch list might be occurring in their area. Just to test it, I posted spotted knapweed along a lake. Two weeks later, I visited that same lake – the knapweed patch was no longer there. (Since you don’t make evaluations on invasiveness, I’m not sure it would meet your positive standard). It would seem that for those people who contacted me, having this source which seems to have helped them, would make the site a plus.

On my part, with no coursework in science at the collegial level, the information under each species often does not give adequate ecosystem information, nor when it comes to species which undergo metamorphic changes, only scant note of each life stage without describing the conditions nor the variations in changing status. I’ve found, for instance, alligatorweed flea beetle larvae in hollow stems, but more frequently in envelopes of top-of the-plant leaves sealed with silk. One independent source notes that this latter situation occurs on alligatorweeds which grow on land near water, but tubular larval housing more common on those in water. My observations are almost all on water or next to water. Yes, there are citations, but having more depth in the text would be much more helpful and would help in learning much more about the observed species. I would hope that those familiar with this beetle in its home territory might be better able to contribute to furthering information (in the US it was introduced in the 80’s as a biocontrol agent.) Wouldn’t that be a real collaboration internationally?

1 Like

Not sure I would share that assumption, because of the sampling bias noted earlier:

2 Likes

I, on the other hand, am certain that the range maps for even widespread species do have large areas missing from their iNat range maps. Israel had about 40k observations when I started observing there, and I was the first observer in Israel of a number of plants which have thousands of worldwide observations. eg Euphorbia maculata. Not to say that physical collections are necessarily better about this… I was also the first to bring a specimen of the species to the main herbarium in Jerusalem – though a later thorough search revealed one other, mis-identified as E. prostrata, from 2007.

3 Likes

I guess with E. maculata it was a question of time, it’s alien in Europe and such plants can avoid herbariums for some time. I agree that many common species have far from real areas on iNat maps, taking a classic Mallard and half of its range have no observations at all. And what to say about insects? Many common species have no observations and many others have no speciealists to prove them. Thinking about professionals keeping their colelctions away from iNat, they’re cruel, but it’s in our hands to change.

2 Likes

I don’t think I’ve ever used it, but my impression is that it’s for disagreements about what the correct ID should be. One example would be if you’re confident that a specimen is a certain species, but for reasons that can’t be shown through photo or audio. Identifiers could legitimately push the observation back to genus based on the evidence provided, so if you want to keep it as a record for the species you’d have to negotiate with them and/or opt out of community ID.

A couple of the top observers for species on the website “opt out”, and for them it’s because they want to manage and control their own observations, especially for dealing with either overconfident identifiers (e.g. subspecies by range) or oversensitive identifiers (e.g. wanting to keep it at genus when there’s sufficient evidence for species).

3 Likes

I will change my mind on this. Phenology on iNaturalist is suggestive and better used by other naturalists to get a general idea of when things occur. In some cases phenology gleaned by iNaturalist could be useful but it would depend on the species, how commonly it is observed in an area, how identifiable it is in all its stages and how much available data exists outside of iNaturalist. A scientist could use iNaturalist as a starting point on the life stages of a species and launch a study to generate more data to fine tune the timings of life events.

1 Like