Abnormality bias

[mnharris]

I haven’t seen this addressed explicitly in any of the FAQs, but are observations on here supposed to reflect a general view of what is common in the ecosystem?

This is best explained through an example. I come across crows, pigeons, robins and mallard ducks quite often. I have posted all of those species. But after the first time seeing a common bird in a location, I am not going to keep taking pictures of every crow I see. On the other hand, if I see a bald eagle or a great horned owl, I am going to report it. Each and every time, even if it is a blurry or bad picture.

The problem with this is that if everyone does this, it might give a very skewed idea of species distribution. If there are as many observations of ducks as bald eagles from an area, that isn’t a very accurate census. It shows that both species are present, but doesn’t show that one is common, even comensual, while the other is quite rare.

And for some plants, it is even more of a problem. I have not taken any pictures of grass, because…well, I just can’t imagine going around taking pictures of turf everywhere I go. My botanical observations have been confined to things I find interesting. But in doing that, I am underreporting some pretty core parts of the ecosystem.

So I guess my question is, how much should we try to capture the entire ecosystem, or is taking records of the parts of it that are most of interest to us a fair thing to do?

[sarahduhon]

I think about this a lot, especially the problem that observations of nature can only happen in places where people are actually there to observe it. If you need a good representation of species and population sizes in a particular place for, say, conservation purposes, these are the reasons why iNaturalist isn’t a good tool for that, and a planned methodical survey is needed.

What iNaturalist is good for, though, is encouraging people to get involved and excited about nature. It’s also good for very general conclusions about where certain organisms are and are not. It’s also super helpful for researchers to get in touch with people who can contribute data, whether it be to look out for something, or send in a specimen, etc.

[jdmore]

I second @sarahduhon’s response. I know I experience that kind of bias in my own sampling of nature, and it will just have to be assumed to be a factor by anyone using iNat data. If one sees a Canada Goose observation in a city park in North America somewhere, for example, it’s a pretty safe bet there was more than one

iNaturalist provides a pretty general set of tools that every user can choose from to meet their own desires and needs. I know of a user in the desert southwest of North America who posts a very high density of observations of common desert plants like saguaro cacti and creosote bush. That’s great - it (presumably) meets their needs, and will likely be valuable to others also.

But yeah, most users will probably not be bucking the abnormality bias to quite that extent.

[JayAvery]

There’s no rules about it, so there’s no ‘should’ - you should do whatever you feel like doing! iNat data is never going to be a source of accurate population information, and it’s not users’ responsibility to try and make it so.

[cthawley]

There are definitely biases in the data present on iNat, but that’s really a concern for those who want to use the data for something scientific.

If you are just interested in iNat as a way to interact with nature, post whatever you find interesting and fulfilling to post!

[neontetraploid]

I feel the true strength of iNaturalist (aside from getting people excited about nature) is actually tied to this “abnormality bias.” As people become more familiar with the “normal,” they start paying more attention to the more “abnormal” things around them. New geographical records, range expansions, and even previously undescribed/undiscovered taxa; all of these are brought to the community’s attention through people uploading “unusual” observations, especially in less-studied regions. I’m sure many of you have come across at least a few observations that fall into one of these categories, and that’s where the real magic of iNaturalist happens. Sure it isn’t ideal for hard data like population size, but it is extremely useful when it comes to shedding light on the novel and unique.

[biosam]

There are certainly many biases, including where users are located geographically, what they like to photograph, their level of natural history knowledge, etc. But that’s okay! iNaturalist is first and foremost a website for connecting people to nature and sharing biodiversity info. It is not primarily intended to be a rigorous, quantitative research platform, although the data certainly can be and is used for research and monitoring. Any researcher should be aware of the limitations and biases of any data collection method. A recent study (Lehtinen et al, 2020) on color morph frequencies of the eastern gray squirrel (Sciurus carolinensis) incorporating iNaturalist observations remarks: “citizen science records are likely to be a biased sample as many participants might be more likely to submit observations perceived as novel… Thus, using citizen science data, the real frequency of color morphs in an area is not known with high confidence since the assumption that participants submit bservations of all color morphs equally is unlikely to be true. We recommend caution in making inferences based on citizen science data and especially in pooling these datasets with others”.

If you want to photograph every living thing you see in an ecosystem, in proportion to its prevalence, you can! If you only care about beetles and dandelions, or pigeons, you can just photograph those! What other people do with your observations is up to them. No researcher should blindly depend on iNaturalist to give them complete or unbiased data.

[stanvrem]

Never say never. Who knows, Artificial Intelligence 10, 20 years from now might be able to compensate for user bias and generate reasonably accurate population numbers and trends.

[sedgequeen]

I think that iNaturalist can a reasonably accurate idea of species ranges (for some species, in some areas) and about changes in ranges (as species are first recorded or people stop recording them) (for those organisms people often photo, in some areas). It’s lousy for population sizes. And it’s great for connecting people with nature. So photo what you want to photo, and let the data user beware.

I personally enjoy photographing and posting all the plant species at a given location (including grasses) but at other places or times I photo just what most interests me. Lately I’ve been taking almost daily photos of a molting warbler that’s coming to my feeder. Satisfying to me and iNaturalist is pretty tolerant.

[arjonker]

I agree with most people here. There is a bias. In the Netherlands they have tried to quantify the observations submitted, but that 1 rare bird is entered a thousand times and the common birds only a few times. I think these platforms are good for range data. Does a species occur there or not (yet). It is also good for getting an ID of a species. It will never be possible to fit all living and dead individuals of all the species in the whole world into 1 database. I think trying to do that will make iNat to succumb. I think it is better to photograph the variation within a species and giving an estimate of the number in the comments/description space. Or give an estimate of the area it occurs in if you wish. Pictures of different parts an different stages are very useful. Just the ‘same’ picture over and over again is not.

[mamestraconfigurata]

I also struggle with that. In Winnipeg, in the winter, and where I roam daily, there are the ‘usual suspects’. Birds (mainly) that are hardy enough to spend the winter here. I don’t bother much with crows. If I saw an extreme reduction, I would try to document that somewhere. I do take way to many photos of chickadees, nuthatches and woodpeckers (Downy and Hairy), just to document their presence. Crows tend to bugger off when it gets really nasty, but those four stay in place. To be truthful, I like them, and if I noticed a drop, I would say something to the City, or somewhere.
Such observational biases can throw up some interesting questions, though. I didn’t bother with Blue Jays or American Robins initially, but around my place they have seemed to decrease in numbers. They are still around the area, but robins especially seem to be more scarce than they were 5 years ago. I don’t have hard data for that, but the fact that I now try to capture photos of robins and jays is a change.
For what it’s worth!

[simono]

I have to practice abnormality bias out of necessity. I live in a remote and very rugged area whose biodiversity has not been well-surveyed. It is simply not possible for me to regularly document common species as otherwise I would not have the energy, battery power, fuel etc to make observations of insects that have not been recorded in decades or centuries.

[xpda]

I think the observations here represent what people see and what people like and find interesting. That probably couldn’t be changed even if it was attempted. To get a general view of what is common in the ecosystem, you would need to consider the “attractiveness” of subjects, human population densities, organism densities, and how attractive the area is for outdoor observations, when organisms are likely to be visible and identifiable, and other factors. iNat provides a ton of valuable data that has never been available before, and makes it available to everybody. But it is raw data, as it should be.

[robert_taylor]

You might not be able to estimate population sizes but you should be able to pick up general trends in population growth or decline of a species/family/order.

Observations as a relative proportion of all observations (of all taxa) made on the site helps smooth out the effect of the overall growth of the site. You can click on this option on the species pages - https://www.inaturalist.org/taxa/5305-Haliaeetus-leucocephalus.

As an example I looked at the 74 011 Lepidopteran observations in Africa up till the end of 2019 - just eyeballing it - there might be a bit of a decline in recent years. I assume that prior to 2014 there might have been too little data to tell anything meaningful.

[mamestraconfigurata]

It’s interesting - I took a look at the iNat graph for American Crow (https://www.inaturalist.org/taxa/8021-Corvus-brachyrhynchos) and the Black Capped Chickadee (https://www.inaturalist.org/taxa/144815-Poecile-atricapillus). Both curves reflect basically what I have intuitively inferred - higher numbers in the winter, lower in the summer, then increasing as fall and winter begin. My guess is that in the winter, there are more of the birds flying around, so they get photographed more. In the spring and summer they raise their chicks, so are not as visible (plus there are more birds passing through to pay attention to).

[karyncook79]

I definitely post anything uncommon - but I also upload lots of repeat observations of the same species. I do this because (and I may be completely wrong here) I thought it might one day prove helpful by including it in the data, and uploading enough observations that it would be obvious that it is common and abundant, should a common species suddenly become less common/uncommon due to some environmental event, etc. I also do this for the common invasive species in my area for a similar reason - so that it’s plain that I’m finding it everywhere.

[botanicaltreasures]

I see much more than I observe. Yesterday, I saw butterflies and birds but most of them were moving too fast for me to capture with my iPhone. Also some plants I find almost impossible to photograph in situ because of busy backgrounds. I’ve almost consider whether it would make sense to carry a small collapsible pocket sized background with me. Outdoor plant & critter photography carries it’s own challenges.

[dianastuder]

that is why my pictures often have fingers in - to force the camera to focus on The Flower, not whatever. My hiking group also commandeered my hat for a cryptic greenish brownish creamish small orchid.

[karen5lund]

I was thinking of this discussion when I went for a walk this morning, so I photographed a robin and a dandelion. Looking for squirrel and pigeon next time (whenever that is). No intention of trying to get a representative distribution of species–certainly not by number; that would be incredibly boring–but I will make some effort to notice the common local species.

I was also thinking about what constitute normal or abnormal species. (I do some of my best thinking while walking, so this is a difficult time for me, as it is for many.) What is normal to me is not normal to someone I met virtually on Mastodon–the open source social media site similar to Twitter–who posts iNaturalist observations in Singapore. I’m in New York City. Are my observations as “exotic” to her as hers seem to me?

And it’s not only location but time. When I was much younger we didn’t see white-tailed deer and wild turkeys in our outer boroughs neighborhoods; now they’re so common as to be considered a nuisance. Likewise, lichen on trees was just about non-existent, but thanks to less air pollution I have some lichen photos waiting to be uploaded. The above changes have happened within the past 20 years or so.

So I really have no idea what is “abnormal” among my observations. It depends on where or when the viewer is.

[karen5lund]

Is it actually possible to photograph only one Canada Goose? As I’m going through my old photos to upload I’m finding lots of them, but I can’t recall a photo of just one.