It gets even murkier with fungi, where “the individual” is more often than not invisible, and many observers are unaware of it. A blanket rule of what is a “proper observation” doesn’t appear useful.
I think @upupa-epops has the right idea: an observation is of a single species at a particular location and time. If it’s a single individual, then there’s no difference. If there are multiple species present, then the observation has to be of one of them, so one photo with two species would have to be duplicated to get IDs for both, and an observation with two photos which don’t show the same species must be split before it can be ID’d to species. If there are multiple individuals of the same species, then there’s no particular reason for the observation to be invalid. Indeed, having multiple individuals of the same species can aid ID, as in the example, where males and females look different. I know of several genuses where the females of a couple of species are nearly indistinguishable, and it’s very valuable to have a photo of a nearby male to help ID her.
Pushing the boundaries of this a bit, there are a couple of cases where I photographed too many individuals of a species I know well within a few minutes and a few meters of each other, so rather than post a dozen observations, I just merged them into one. I think these are valid observations. The only reason I’d split them would be if someone came along and pointed out that I’d mistakenly included an individual of the wrong species.
In the egg-case case (ha!), I don’t have a clue how many individuals are in each photograph. Presumably ~100 each.
Here is my take at the moment ;)
Possibly widening the breadth a little to encompass things such as scat, fur, feathers, other remains, tracks, and etc:
An observation is the evidence of the presence of life preferably recorded in the form of an image or sound with the specification of time and place.
A stimulus resulted in that record. That record can just be that unto itself or can have a purpose or give a result on its own or in combination with other records.
The stimulus is multifactorial and can be impulsive, planned, automated, reactive, academic, non-academic, and etc.
The purpose and/or result is/are also multifactorial with additional evidence of behavior/interaction, habitat, growth, individual, difference, scale, morphology, rerproduction, range, persistance, posture, gregariousness, group dynamic, visibility, blend/camouflage , diet, checklist, and etc.
Because of the multifactorial nature, observations are difficult to define from one person to the next and from one moment to the next.
What about photos with two species in it:
This observations shows Bombus affinis on Eutrochium purpureum, currently the observation is a research grade observation of Bombus affinis, and the ID for the plant is in the observation fields section. However, if I want to download the research grade observations of Eutrochium purpureum, this will not be one of them.
Is this a problem? should there be two observations, one with the plant ID and one with the insect ID? Does it matter that the plant ID can’t become RG or be integrated into the research dataset?
Yep, the user can use the duplicate feature to quickly create a second observation for the plant, if they’re interested.
@cmcheatle I think that reveals different ideas about what observations are meant to show and how they are used. Are they just to indicate presence/absence or are they to indicate abundance as well. If it’s merely presence/absence then a single observation even if there are hundreds of individuals is adequate, but if observations are meant to indicate abundance then multiple independent observations would be the way to go.
Both methods have their uses, strengths, and weaknesses.
In practice I take iNat observations to be of the presence/absence type because there is no consistency in how observations are made… some make independent observations for every photo of the same individual, others make on observation of a single individual when there is a large number present, etc.
I would agree that a single observation that includes 3 or so different life stages should not be a single observation though. Observations, at their core, are about individual organisms and discrete moments in time and specific locations.
I dont dispute the fact inaturalist can and should be used for abundance data. I wish it were in fact. What I dispute is that the optimal way to record having seen 200 of the same warbler species in the same park is to record 200 separate records.
I agree that it’s impractical to have 200 of essentially the same observations.
Perhaps a feature request could be an “abundance” tag to go with the “life stage” & “sex” tags.
It has been rejected multiple times by the site. They will not implement that feature.
I think observations are a way for people to “share what they are seeing”, a place to converse and learn about them, and an opportunity for science to capitalise with a data point.
When we start talking about whether we have to count everything we see, just imagine that you have been invited on a bush walk, and you are standing next to the person that invited you, and they point out something that catches their interest. Are you going to suggest they should count every single one? Or that they should point out every single one? Please DO CONSIDER THE OBSERVER in this equation… not every observer is a field worker out to collect every single data point they can…
If abundance data is wanted, then you can use iNat to be a sort of pre-scout into an area, and use it to design and implement a field operation that gathers the data you need. If I am looking to do a specific study on katipo, then I can use iNat to locate potential sites that are more likely to yield the results I am looking for. I most certainly wouldn’t use raw iNat data for such a study!
That’s exactly why I said this:
However, a pull-down number field that is something along the lines of, “1, @5, @10, Many, Enter Approximate Number” with 1 being the default for any non-selected option could have value.
Realistically people wouldn’t use it , just like they usually don’t use the “life stage” or “sex” options.
yes, it has been discussed before about adding “count” annotations, but I think annotations have to be discrete (ie set values) rather than scalars.
[0 | 1 | some | heaps] would be doable!
I take it you are using the @ as an indicator of “approximately”? I have always used the tilde (~) character for that as I think of it as the wavy line above the equals part of the mathematical symbol for approximation, although I don’t think I’ve ever seen anyone else use it that way. Or I use +/-, depending on situation.
I do think this is an excellent point for data users: inaturalist effort is not standardized across observer type, which is something rarely accounted for in analyses
Haha, yeah, I was using @ as ‘approximately’. I grew up with that as the standard shorthand, but it’s been co-opted and now is often used to mean ‘at’ in an online sense.
Be nice if my keyboard had a key for ≈ … I could hot-key it, but that just seems like a lot of hassle for little return.
I use it like that too. You are not alone.
I’ve just finished reading (hopefully) the entirety of this “What is a valid observation?” thread. I can’t say I yet understand where @cmcheatle comes down on the question except that Chris points to the technical definition found in one corner of the iNaturalist guidelines (“one individual…”) which seems to give rise to the dilemma. I tend towards the “one taxon…” variation of the definition for some common sense biological and practical reasons.
My suggestion: Encompass into the definition of an “observation” some simple allowance (“fudge room”) to better fit organism behavior and our observational tactics. I’d suggest a rewrite such as:
“One ‘observation’ provides the documentation of one or more individuals of one taxon at one location at one time. The key elements are that the observation should refer to a single taxon, the location should be as precise as practical and, if of multiple individuals, they should reflect a common biological collective unit such as a herd, flock, school, or other set of individuals in close proximity due to the taxon’s expected behavior (e.g. moths at a moth sheet).”
Now I’m usually pretty verbose in my writing so some editor could tweak this but it seems to capture the various conundrums offered up by Chris and others.
– There should be nothing in the definition of an observation which precludes a researcher from uploading separate individuals as separate observations–whether by design, by caution, or by accident. That type of effort may be part of a research design/protocol which of course will be perfectly valid. This can be another caveat to the definition–and it may need several caveats.
– For sexually dimorphic animals or dioecious plants, I’d suggest that the two sexes may be uploaded together or separately at the observer’s discretion, again as long as they are in reasonably close biological proximity (an interacting pair, members of a flock, etc.).
– For fungi and lichens, we shouldn’t get “hung up” on what an individual is. More often than not, what will be documented is the “sporulating body” (mushroom) or a whatever a blob of lichens is called. That’s why my suggested definition is inclusive and biological based.
– Documenting different parts from different individual plants occurring together seems so practical and logical that it shouldn’t raise an eyebrow…unless the plants are miles apart, etc.
– AI will just have to “deal with” multiple individuals in an image or multiple images of different individuals. I would humbly suggest that should be part of the learning curve for iNat’s computer vision.
– There will always be cases where multiple species are included in one image or in a set of images–typically by accident or ID errors. Those are to be expected; they become the grist for identifiers and further clarification by comment, etc., such as, “Your third image shows a female Farkbleberry Warbler rather than a male Twinkleberry Warbler…”
I could recite a large array of examples (e.g. from my experience with plants, birds, and moths) to illustrate special cases but I’ll shut up for the moment. A set of examples (with links to characteristic examples) will be of immense value for a settled definition.
Where I come down on it is pretty simple - iNat rules as they are written say each observation must be for one and only one individual. Not for a taxa, not for a number of them you saw in close proximity, but one individual.
And that this definition is silly and should be changed.
So who on staff draws the black bean of rewriting that tiny portion of the user guidelines and suffering the slings and arrows which will follow? ;-)
Before someone tells me iNaturalist is not used to do abundance data, please understand, in no way should collecting of abundance data be mandatory for recording iNat observations, however, it should be standardized for those who wish to.
I’ll give you my own example. I made this as one observation because cardinals usually mate for life. It seems logical to me that a mated pair should be one observation. https://www.inaturalist.org/observations/28575721
Granted, I am not interested in iNat for it’s research data ability. But my layman thought seems like the fact this is a mated pair should be important to any data.