Observations of absence vs presence

#21

Point well taken. Which means @jon_sullivan (quoted above) and others who have used the “sought but not found” field have some decisions to make…

#22

As Jon pointed out, the problem was well understood and an alternative method was found. No one should still be using “sought but not found”, and I don’t recall seeing it being used.

@tony_wills @jon_sullivan This does raise an interesting dilema with the Great Kereru Count, which has historically been loading absence records ID’d to taxon and a count field of 0 (some years it has been “kereru count” :frowning_face:)

#23

Our Great Kereru Count is an outreach that has members of the public logging observations of Kereru, which are a very easy to identify bird, and for the most part pretty hard to miss, if you are looking for them! It is not a complete count, anymore than it is possible to state categorically that there are none! But as far as the question “how many did you see while you were proactively looking out for these birds” goes, 0 is a useful piece of data!

2 Likes
#24

there’s no doubt that absense or ‘fail to find’ data is very valuable! It’s just that iNat isn’t well set up to record it and i don’t think it’s high on the devs development list (though i could be wrong)

4 Likes
#25

Hi everyone. I’m late to this party but @tony_wills has already nicely summarised my current solution to this. That is, to add an observation with no ID, so it doesn’t show up on the map. And, set to “No” both of “Based on the evidence, can the Community ID still be confirmed or improved?” and “Evidence of organism” (when there’s a photo), so it doesn’t show up on the Identify page. And, the observation field “Taxon sought but not found” is set to what was being looked for (eg Papilionoidea for butterflies). It’s clunky as anything but does work.

Why do this at all? I use if in class exercises at university where I send my ecology classes out to look for some taxa, eg. butterflies, in a set period of time (for butterflies, we use 20 minute surveys on sunny warm days). iNat works fabulously, especially with all the computer vision and community help with the IDs. I also like that it gives the students familiarity uploading observations on iNat so in the future they’ll use it when they find new and interesting things. The only problem is when they find no butterflies at all, which is where we need our clunky “sought but not found” solution. If there wasn’t some way to enter these observations, we wouldn’t be able to use iNat for these surveys.

I’d love to see a more elegant solution to this on iNat. I also understand the developers’ reluctance to feature creep in this direction.

8 Likes
#26

this was posted in another thread dealing with absence / presence:

2 Likes
#27

@htrudell – it sounds like Trips (https://www.inaturalist.org/trips) might address your use case, though i’m not sure how well-supported that functionality is.

3 Likes
#28

With plants absence is often observed, and documented very well with a photo

split this topic #29

A post was merged into an existing topic: Trips Feature on iNat

Trips Feature on iNat
#30

Seconding (/thirding?) @jon_sullivan - not having a standard way of recording absence at a fixed sample point is a large hole in functionality.

1 Like
#31

Guess I just have a semantic quibble with the term absence. To me, absence means that a taxon is not present at a location. What a photo can document is lack of detection. To document absence takes a lot more search effort, since one is attempting to “prove a negative.”

1 Like
#32

absolutely… if you want to prove there is no needle in a haystack, you literally have to examine every straw of hay. To rummage about randomly through it (or even worse, to just visually scan the surface) is only going to give you “no needle detected”, not absence…

2 Likes
#33

Fair point on the semantics.

The obvious scenario in my mind isn’t plant based - it’s a trap line, where you have a very defined set of target species and you’re inspecting at regular intervals. ‘No possum’ is just as valid an observation in this scenario as ‘possum’. The data is in that detail, and meta information about the method used - the photo is just an additional level of confidence for the community that the trap was inspected, if you upload one.

2 Likes
#34

these situations don’t neccesarily need to have photos and RG status observations. As the creator of the data, you have confidence in using it, so “evidence” per se is not needed. What the photo will give you, however, is gps and date/time, and fields can record the presence/absence, and any other data such as non-target species caught could be documented with photo evidence to aid in identification where applicable.

#35

A little off topic, but suppose there was a new undetected/absence type of “observation”. in that world would you record a raccoon track in a presence type of observation or in an undetected/absence kind of observation?

#36

That will work very well to record the absence (ok, lack of detection) of weeds . In the case of large invasions suppressing other vegetation, the absence of detection is ample info. If we cant see the weed in the image, the adjacent vegetation can develop.
Of course it doesnt record the absence of seeds, roots etc so a repeat observation is needed to document whether the weed became present again.
The absence record in between is vital to assessing both the effectiveness or otherwise of the first intervention or lack of, and the means and rate of reinvasion.
So thanks jon, i will do that. I am not a scholar and dont know what people mean by the maps, so to date I have been afraid of posting absence observations

#37

My dilemma is that if “bird” isn’t found when searching for glass collisions, we can’t prove it didn’t happen - or to what species.

#38

Observation fields like https://www.inaturalist.org/observation_fields/10102 are quite problematic because they misrepresent observations which by definition record the presence of an organism in space and time as an absence. Since many of the ‘subsystems’ in iNat are interpreting observations as presences doing this can have lots of bad downstream repercussions on distributions, image recognition models etc.

Please use https://www.inaturalist.org/trips if you want to record absence data on iNaturalist

2 Likes