What is an observation?

#1

An observation records an encounter with an individual organism at a particular time and location.
https://www.inaturalist.org/pages/help#observations1

There have been several parallel discussions about defining an observation, so here’s a place to host those.

4 Likes
Use computer vision to annotate observations?
#2

The first thing that needs to be fixed / clarified in that is ‘location’ is a meaningless, or at best ambiguous term.

A provincial park is a ‘location’, any of the individual trails with that park is a ‘location’, an accurate to 6 decimal places GPS bearing is a ‘location’.

I suppose given you must add a precise pin to any location, the closest iNat match is the last one but then that is offset by the uncertainty buffer (ie is it appropriate or not to add a buffer under the guise that what was reported was within that circle? If that is appropriate then a record with photos of different individuals must be acceptable)

EDIT - of course if ‘organism’ is taken to mean a single individual entity, then ‘location’ must be the exact spot where that encounter took place.

Of course this then goes to absurd lengths:

  • if i photograph a flock of 200 ducks, the record I submit is only ‘really’ about one of them
  • if I observe a single bird and it moves from one tree to another while I photograph, each photograph of it in a different tree needs to be a different record.
1 Like
#3

I think the definition is necessarily vague, because it comes down to common sense (which varies widely from one type of organism or situation to another) and how you use iNaturalist (which varies from user to user).

If I observe a bee for one minute visiting three different species of flowers, it seems to me not unreasonable to submit it as three observations, even though I know it is one individual at times/locations that are separated by just a few seconds/centimetres. “Bee feeding on rose” and “bee feeding on daisy” are two separate pieces of observed behaviour that are scientifically interesting, and which it could be most useful to keep separate. But if I take 10 photos of one bee over 10 minutes as it buzzes aimlessly around my garden, it may be worthwhile submitting all 10 photos (if they show different angles, for instance) but most logical to treat those 10 minutes as a single time and the whole garden as a single location.

What constitutes a different “time” is just as undefinable as what constitutes a different “location”. If I photograph the oak tree outside my house at 8am, chances are it will still be there at 9am, and again at 10am, and it benefits nobody to spam iNaturalist with hourly photos of the same tree. Monthly photos, on the other hand, could show it in different stages of leaf and flower, which is scientifically interesting information. But other organisms live life at an entirely different pace: a little toadstool might emerge, open up, release spores, and collapse over the course of a day or two, and documenting each stage of that is scientifically interesting.

This kind of information would be invaluable to people studying the flower choices of bees or variations in the date that oak trees lose their leaves from one year to the next, but of no interest whatsoever to people who use iNaturalist to compile a species list for their garden/trip/life. The latter type of person, once they have logged one feral pigeon may have no interest in ever logging another.

So I think the answer is just to try to understand the ways in which iNaturalist is useful as a scientific tool. Once you have logged a photo of a common daisy in your garden, you may not see any point in logging more. But to a person studying the number of petals on daisy flowers, the more photos of different daisies they can see, the better, and it would be most helpful if each plant is submitted as a separate observation.

7 Likes
#4

So I am going to do a little test here. Under iNaturalist rules, is this a valid observation or not ?

It contains 2 photos, a male and a female of the same species, taken at the same feeder, at the same time.

Please note I am defining ‘valid’ as appropriate to achieve research grade status, or not have a negative vote assigned on a DQA parameter (ie does not break iNat rules on what is an observation), not ‘of course it is valid because anything can be submitted into a casual grade record’.

I don’t know if it should be a 3rd option, but my personal answer would be ‘under the rules as written, this is invalid and should not be research grade, but it should be acceptable’

https://www.inaturalist.org/observations/20267669

  • Yes - this is a valid observation
  • No - this needs to be split into 2 separate observations.

0 voters

#5

I have voted Yes on your poll because regardless of the merit of this observation (for whose purpose anyway?), nothing needs to be split, voted, ID’d, qualified, etc. on iNaturalist. It’s your observation. you decide what to do with it.

2 Likes
#6

Not being a birder, I can’t tell there are two different birds in the observation. It looks to me like the first photo has one in focus and one out of focus bird (therefore, one “observed” bird,) and the second photo looks to me like the same bird.

#7

They are 2 different birds, the male and female of this species look superficially similar, but the males are darker. I added a 3rd picture where the female is in the forefront and the male is coming up behind but in poor focus.

#8

Subtle. I wouldn’t have noticed without your tip off, but since it’s your observation and you knew when you created it that it contains two different individuals, I’d say you either have to split it, or remove the photo of the male bird.

#9

I agree with that statement, but not the one about the single bird changing trees. Good thing I’m a plant person, and we don’t have that problem!

3 Likes
#10

I think you’ve missed the qualifiers above - the question seems to be is this “valid” in the sense that this is an acceptable observation to achieve Research Grade.

My rule of thumb for this sort of thing is that if it’s reasonable to assume the same individual is present throughout all of the photos, then it’s fine if other individuals are present in some or all photos. For example, Photo #1 is just Individual A, Photo #2 is Individual A and Individual B, and Photo #3 is Individual A and Individual C, then I think it’s fine for that to be a Research Grade observation for Individual A.

In @cmcheatle’s example, you could say this is fine because the male is technically in all of the photos (albeit blurred in the background), but I think the point they were trying to illustrate was the situation where Photo #1 is Individual A, Photo #2 is Individual B, and Photo #3 is both.

#11

In Mirko’s defense, I added the qualifiers after he replied, as I was concerned about how he may have interpreted valid.

1 Like
#12

Fair enough. If the purpose of the case is to call the rules -as written- into question, then perhaps the observation is not valid in that sense.

It certainly is valid in the sense of how I see people submit observations in practice. The observation would be accepted as RG without question by anybody not actively evaluating the rules.

I stand by my statement that there is no objective need to single out this observation for any purpose other than your own. Cases that could be considered edge, like this one, are certainly a much smaller problem that confirmation bias and the rampant uncritical acceptance of the AI. Those are not the topic of this thread, of course.

1 Like
#13

This thread was actually spun out of this one https://forum.inaturalist.org/t/use-computer-vision-to-annotate-observations/3331/12 which was about having the AI do more work in terms of not just identiying species, but annotation level characteristics of the photos, and what do you when observations have more than 1 photo or individual. That is just a different form of tacit acceptance of the AI.

1 Like
#14

I think in practice I generally make observations as if they are defined as “an encounter with a species at a particular time and location”, because I will include in one observation photos which I know are of multiple different individuals of the same species in one area (similar to @cmcheatle’s Yellow Warbler example).
I guess I could submit each photo separately, but it seemed a bit silly, sort of as an extension of the 200 ducks example.

3 Likes
#15

wait, by this logic i can’t add a plant with both male and female flowers even if they are the same individual. That doesn’t make any sense. Most photos have multiple organisms visible anyway and maybe i am just misunderstanding something but i don’t see how the bird could be disqualified from RG because there’s another blurry bird in the background

#16

I think it is because there are multiple photos. Photo 1 has bird A (with bird B blurry in the background), implying that bird A is the focus of the observation. Photo 2 has bird B with no bird A present. Now what? Which “individual organism” is the observation for?

#17

It seems to be ambivalent. According to the rule stated in the opening post of this thread, it is invalid because it contains multiple individuals. However, there is another rule which states that “iNat observations record one taxon at one place and time”. So, since your observation only records one species, it is also valid.

The next question is: what exactly is an individual/taxon? Are composite organisms like lichen two species (and therefore two individuals), or one? How many individuals should be counted for modular/colonial organisms? There is obviously no clear-cut answer to these sorts of questions, so iNat’s “rules” seem to be unenforceable for the general case.

#18

ah so the equivalent would be more than i take a photo of one trillium plant, walk 2 feet down the trail, take another photo of a different plant because it shows some other diagnostic feature or something and add them both to the same observation. I guess i see that as technically kind of fuzzy or in a grey area but certainly not worth worrying about. I’ve done it.

3 Likes
#19

I do it too. But there is always the possibility, despite our belief that they are the same species, that they might turn out to be different species!

For what it’s worth, I think an observation is the recording of someone having noticed (encountered) another living thing. It is a place (in the observation view) to find out what that other living thing is, and to partake in discussions about it. Just as I could be walking through the bush with a friend, and point something out and ask if she knows what it is, and even though she doesn’t know the name, she might tell me that her Mother used to use it as a tonic for upset stomachs… so too in iNat we are walking through the bush with hundreds upon thousands of people, many of whom CAN tell us the name, and with the name we can learn even more about that thing. And we meet people that have a scientific interest in those type of living things, and sometimes our observations (or answers given to questions) help in forming a clearer scientific understanding of these things.

to me, the scientific data is just a bonus, a wonderful side effect derived from empowering people to record their encounters with the natural world. The gold for me is in being able to see the world through other peoples observations. Being introduced to the fauna and flora, of being able to learn about what they do and how they “fit” into the complex web of nature. When someone posts an observation, I see a window into the world they live in, of the things they value and deem worthy of being noticed by others.

4 Likes
#20

If you had specified the sex as male it would be invalid as it’s not in all three photos. The sex was left open so it is an observation of the female bird. I have photos with insects and plants together. Should the insect be out of focus in one of the photos it doesn’t make it invalid as an insect observation IMO. Insect-plant is more different, IMO, than male-female.

I also have seen pictures with several different waterfowl. As long as the user can get close enough so that confusion doesn’t abound(Aves would be a bad start point), they are fine. The web app actually has a duplicate option for cases where multiple possible subjects are present in a photo.