Strengths and Weaknesses of iNaturalist Data

I will mention that data does need a bit of curation and most of the dots on extreme edges of ranges need to be checked carefully.

I will be mentioning that iNaturalist’s primary purpose is to get people to connect with nature. But I wanted to showcase that iNaturalist’s data is useful, and that people shouldn’t consider contributions to iNaturalist a waste of time scientifically. iNaturalist can not answer questions that the third breeding bird atlas for Ontario I am participating in next year. I believe breeding bird atlases do a great job in determining the population status of bird species, and is one of the best citizen scientist projects to participate in. That said, a breeding bird atlas needs a ton of volunteer effort to be successful (hundreds of thousands of hours) and ignores the other 99% of nature.

3 Likes

I spend quite a bit of time looking at range expansions of species I am familiar with. It’s true most are false alarms but I just disagree with them. Occasionally they are real though! I think the data is very valuable when one checks the suspect observations themselves.

5 Likes

(reposted for clarification)

I use “opt out” to manage my own observations. I have some desire to keep my records consistent, and having everything changing is too tricky to keep up with, so this allows me to go through them one-by-one and “vet” the changes before agreeing to the updated ID.

I’ve been given a hard time by people, who accuse me of being arrogant and rejecting the “greater cause and promise of data quality”. Because, yes, I’m just one person and I can’t claim to have expertise everywhere. I’m always than open to correcting and adjusting my ID if rational evidence is provided. If I’m not adhering to community ID even after my “review”, then there’s always a good reason for it. Personally I do believe that data wise, opt-out is arguably negative on average. But in my case I’m very careful and I’m certain it actually provides a small benefit. There are a few cases where letting the community ID override would have kept incorrect IDs (from non-experts, mostly) at the forefront. Or cases where people commit to species without adequate evidence (e.g. “guessing” a species without reasoning).

But at no point am I letting “incorrect IDs” stay around, in my case.

5 Likes

@trh_blue @melodi_96 @rayray Observer location bias will skew the spatial sampling distribution at a point in time but over time the spatial sampling distribution will converge on the actual distribution. Twelve years is nowhere like long enough to get complete coverage for an abundant, widely distributed species, especially in parts of the world with limited participation in iNat. I would argue that Marina’s mallard map is actually pretty good, given the relatively short duration of the effort and the issues with collecting records in remote parts of the range. Does anybody believe that the blank areas on that map will not continue to fill in as time moves on and observation totals mount?

As others have noted, the issues with iNat are largely the same issues encountered with mapping presence/absence generally with slightly different details.

1 Like

That’s quite what I said, we’re making it better, though Mallard have over 100k observations that is pretty much enough to not have such big blank areas, as not many observers needed to cover them in sparse dots, it’s totally a problem of iNat (or similar systems, of course) not being a common thing in those areas, plus Mallard is the most observed species, others usually have worse maps.

1 Like

I’m not trying to be a jerk here, and I’m sure these ideas have already been argued at length for years on the forums, but I am curious since you are one of the more prominent and conscientious opt-outers I’ve run into. (I do appreciate that you’ve always been responsive whenever I’ve identified robber flies, and I admire your thirst to keep on learning!)

But… Do you have a statement in your will to help your executors continue curating with your signature careful vetting after you’ve passed? Or to opt your account back in to community consensus? Because if neither of those is set to happen, then in a hundred years half of these observations will probably be broken, just due to ever-improving taxonomy. Maybe iNat will be long forgotten by then (after all, how many websites have lasted even the 25 years since the modern internet took off), but that’s the kind of multi-generational perspective that museums and taxonomists have to plan for. iNat sometimes dreams of being useful for that world (e.g., pushing records to GBIF) but tension will remain so long as we try to serve two masters.

I figure this is one reason why some experts are loathe to spend time on here curating a resource that is ephemeral by design. I mostly gave up on BugGuide a few years ago when I realized that they were letting the eternal mirage of “BugGuide 2.0” keep them from making any real structural fixes and because iNat had the momentum and power backing to take BugGuide’s place. But I still worry sometimes whether my energy would be better spent curating physical bugs on pins that have a longer shelf life.

Anyway, carry on… I’ll go back to my hole.

5 Likes

This is one of the central aspects why I’d also support the removal of iNat’s global opt-out function while still allowing it on a case-by-case basis. Not the topic of this thread, though, so maybe better started as a new one.

Passing away is the most exteme case in that respect, but there are and will be loads of users who at some point simply stop using iNat without further thoughts concerning the afterlife of their data.

3 Likes

5 posts were split to a new topic: How to Note Extra Information / Inter-species Relationships?

Totally agree. With a bit of double-checking, you can fix all the false alarms rather easily. As long as you take all the observations with a grain of salt, iNaturalist is really useful for tracking range expansions.

2 Likes

As far as “Bugguide 2.0” INat has a big advantage in usability than Bugguide, which is more old school and often tedious forum uploads one by one. I still think INat has a long time to go to catch Bugguide in the small and cryptic insect category, where there is sizable species-level identification by many experts. And because there is no community vetting, the specimens are readily searchable and not hidden in the vast dustbin of “genus” or “family” where very few experts will have the time or patience to plow through one by one (especially if they show no details). So I think Bugguide will always have a place, BUT I think these rare, tiny, or difficult to determine species SHOULD be added to iNaturalist by the experts so we have a full species inventory in place for “Identify” tabs to at least inform people of this species’ presence. If people knew there were several similar species of Dung Beetle X, maybe they would stop labeling it as the most common species and help the future curation of the site. I know this is a big project given the thousands of species of tiny moths/beetles/flies/etc. but I think even adding singleton species bookmarks (dead museum specimens) is useful to give that species a presence on iNat finally.

1 Like

One thing I’ve noticed on some of the iNat species pages is that the shaded range maps – the ones that show the supposed geographic distribution based on other sources outside of iNat – are often not very accurate. Not all species maps on iNat have such shading but, for example, many mammal pages do and they are often inaccurate. Since I’m familiar with North American mammals, it’s probably the fault of the sources being used as many field guides have rather poor or out-of-date range maps. Unfortunately, iNatters might use those shaded maps as a guide to what their photo records should be.

1 Like

This is a rapidly developing program, with burgeoning numbers of new users. Even though I joined with expert guidance during a bioblitz, I have found the learning curve to be much steeper than I expected for a non-expert. A photo or two might easily differentiate a regional species, but when you throw in the whole world of possibilities and varying taxonomic systems, it’s easy to miss a subtle point if you’re adding a number of observations at once and trying to move fast. For example, Alnus incana in north-central Minnesota is almost always Alnus rugosa or Alnus incana var. rugosa, if you wish.
The benefit of this, of course, is a deeper knowledge of plant taxonomy for an amateur naturalist like me, and an understanding of what photos are needed for a definitive ID. It has totally changed my outlook on this.

1 Like

Long story short, no I do not. To be blatantly honest, it isn’t something I’ve thought much over (and without considering the museum curation mindset, I imagine few people do). So, I can see why this is a potential problem, and frankly, I have no answer that I can give here, because opt-out is flawed in that respect.

But for me to feel the most fulfillment and enjoyment of using iNaturalist as my chosen archive and database, opt-out is still the way to go for me. I have that many observations and species that there are just too many cases for me to comfortably “let go”. I still have a large number of obs where someone joined and IDed a random species that isn’t even from that continent, and using opt-in would diminish the value of those observations. Selectively opting-out is just too time-consuming and requires a lot of extra management, with the amount of observations I have.

Honestly it’s not so much a thirst to keep learning, although knowing the features to photograph is the crux of the hobby, so that I can provide better data in the long run. It’s more a thirst for having the knowledge at least out there, on how this case was distinguished. BugGuide suffers from this a lot where an expert just moves something, and then 3 years later no one knows why or how that ID verdict was reached. This is especially pertinent for “citizen scientists” or really even professional entomologists without the full expertise that one or two specialists have, and still want to help contribute to that dataset. Sometimes a search for literature answers, but it’s not always available or it is ambiguous. At least when I inquire, and an expert responds, that opinion can be expressed and referred to later.

5 Likes

The iNaturalist stengths are, in my opinion, LOTS of data points and the points are verifiable. (A researcher can check the records and decide if they are correct.)

Weaknesses are uneven sampling (geographically, taxonomically, in every possible way) and the identifications must be considered tentative until checked.

2 Likes