How to deal with observations of fish on fish markets?


I see a lot of observations of fish on fish markets. The problem with these observations is we do not know where it was caught. For all we know, the animal was caught a hundred miles away and so setting the location to the fish market it ended up on leads to false dots on the map.
Therefore I generally mark these observations as ‘Location Inaccurate’, which sets them to casual grade. Is this the correct way to go about this? I mean, sometimes people have more information and know the fish was caught nearby, in which case the observation does provide valuable information.

1 Like


For the most recent identify trip (nightmare) I took to Malaysia (
where I barely even made a dent…I was marking those as captive/cultivated. Likely many of those were wild caught but I didn’t know where like you’ve said, and I also realized there was a good chance some of what I was looking at was farmed. Because that was a major rush job I loosened my high standards and just marked them that way figuring it got them out of needs ID rather quickly. Going forward though, I’m also interested in guidance for future piles of fish on ice. Additionally, an identifier at a market isn’t the person who caught or farmed them, somewhere, making it technically not really theirs to observe as wild. I’m confused!

1 Like


Some judgement is required on the nature of the fish market and the fishery - certainly large city fish markets are likely to have fish flown in from around the world but local fish markets, especially in developing nations, are likely to be stocked with fish caught locally by artisanal fishers. Visiting local markets in, say, Indonesia or the nations of the South Pacific is a very useful way to get a quick overview of what species occur locally.



I would either mark them as not wild or not the correct location. I can see the point about local small fish markets but unless you know how far their boat went from shore you can’t set the uncertainty buffer correctly anyway. I think marking not wild/captive makes a bit more sense than mapped wrong because the organism WAS found where observed, it just was brought there by a person. Also there is the potential that someday the ‘not wild’ observations will still be able to get research grade or train the algorithm whereas ‘mapped wrong’ locations are basically useless for most things



This seems like it definitely should be tagged as captive/cultivated. The location is correct, the species is at that location. It is just that it was taken to that location by people.



I still argue that there should be no hard, fast rule and that it should be judged record by record, based on the evidence and using common sense. If more info is required to make a decision then ask the observer. Here’s an observation that is important because it’s one of very few (perhaps the only) record of the species in this part of Indonesia.
The fishers, who undertake overnight trips in small vessels (without refrigeration), indicated that it had been caught locally and there was no reason to doubt them. In my role as a museum collection manager, I would readily accession this specimen into the collection with the data supplied (and, incidentally, many new species have been described and named from specimens purchased in Indonesian fish markets). Even with the uncertainty, it’s way too valuable a record to lose to casual status based on a recommendation that it be marked down because it’s in a fish market. This one has been included in research of the genus Mola, which I think says enough.



well the issue of casual status is an issue of how the site works, and isn’t a reason to improperly not mark things as captive. I really think that fish should be marked captive unless you are absolutely sure the people fishing stayed in that circle, which seems pretty unlikely. If you don’t like captive things being banned from research grade, go vote on the feature request to reverse that (or if there isn’t one, make one), and email admin. We’ve got a few people breaking the protocol by doing stuff like that to make things research grade when it isn’t. I understand the frustration and largely agree, but forcing things as ‘wild’ when they aren’t isn’t a good solution



That is the region as indicated by the fishers, and there’s no reason to doubt them, but with a pelagic species like Mola even tens of kilometers out is not at all a big deal. That the specimen was wild is undisputed - caught and killed certainly, but not captive which I believe is primarily aimed at ruling out pets, domesticated or not, zoo and farm animals etc. that really do screw up distributions. Do we now mark every fish taken on a line as captive or any lizard being held for a photo?

1 Like


To me, a \captive\ organism is one that without human intervention or pending intended human action like releasing it (or force majeure like a hurricane destroying its enclosure) can not remove itself from the captive state. A bird netted and held for banding, a snake scooped up into hand to photograph etc dont meet that criteria.

1 Like


I think in the case given by rfoster, it’s fine to keep as Research Grade if the circle of accuracy is set to include the possible radius within which the fish was caught. The basis for this should ideally be mentioned in the Description field. For example, that local fishermen said they’d caught it during an overnight trip less than a mile offshore, or whatever. In many cases in fish markets, that information is not available (unless people ask), and in those cases Captive/Cultivated would be more appropriate.



Marking “Captive” doesn’t seem appropriate for fish in fish markets.

A good analogy is plant/insect specimens which are allowed on iNat as long as the date and location are correct for the time the animal was collected.

So dead fish should I think be marked “location inaccurate” if you aren’t confident that the accuracy circle encompasses where it was caught (and people uploading these observations should be encouraged to use very large accuracy circles such that observations can legitimately reach research grade)



my understanding is ‘captive’ is just a simplified word to mean a human put it there. If you capture the wildest tiger in India, and bring it in a cage to Cleveland, it should be captive regardless of the fact that the animal is far from domesticated and would eat you. If a fish is brought onshore, it should be marked captive unless you mark the location as where it was caught. Conversely a bird in a mist net that was documented where it was found should be marked as wild even though it’s in your hand or a net because it came to the mist net on its own. it sounds like this observation is OK, but i am vary wary of this sort of thing because we’ve already seen cases where policy was influenced by things like fish (10KM uncertainty OK) and imposed on things like plants where you lose the ability to do any spatial ecology after 100m or so. if you pick a flower and move it 200 meters, it should be marked captive if the point on the map is not where it was collected.

I also don’t think we should necessarily be encouraging people to map things with huge uncertainty circles to get research grade. Unless you can somehow limit that to animals. And admin has in the past been reluctant to set up different protocol or displays for different taxa.

Ultimately we may need admin feedback on this one though. @tiwane ? I know i am fussy about geoprivacy but this stuff makes me really nervous.

1 Like


if you pick a flower and move it 200 meters, it should be marked captive if the point on the map is not where it was collected.

I think “location is not accurate” would be the correct choice here, not captive/cultivated. As the very same observation could be research grade if the point showed where it was originally picked.



what if you dig it up with roots attached still? And plant it later? I guess this is getting into a difficult grey area… i do see your point here!



there are clearly multiple issues, or at least variables, here. My example of the recent nature challenge in Penang show another variable that isn’t being addressed (perhaps with good reason). The example of a fisherperson accurately marking a location (or responding to inquiries) in an observation showing dead fish for sale isn’t the same as a single new (city challenge) user going to a fish market, uploading 5-10 observations of dead, iced fish which are either incorrectly identified or left unknown and do not give the community the sense that these are dedicated responsive committed users who will be happy to provide you with details about catch and location accuracy. In these cases, unless we can access knowledge about the specific catch, location inaccuracy seems a bad fit. I am not in a position to visit Penang and research fishing or fish farming practices so, in this case, from a practical functional level, I think if I don’t even know if something was wild caught, and the observer hasn’t demonstrated good practices in their observations, then it seems captive/cultivated is more appropriate. I guess I’m confused because these examples may seem similar on the surface but don’t really jive in practice.

Then we have packaged vegetables submitted as wild organisms…but I won’t say more on that because I think you can catch my drift.



I would say the observations of vegetables are less related to the fish market example because those were not submitted as wild on purpose. I think those students a) likely did not know about the wild and captive/cultivated differences on iNat and b) were motivated by an extrinsic reward so likely didn’t care.



it’s more than just students from this project who do that…plenty of “what’s this salad?” without that actual question. I’m really interested in guidance on the rest of what I wrote there though. The veggies comments were just an aside, although indirectly related, to demonstrate the broad spectrum of concerns with “market” observations and the frustration of those trying to help and feeling in need of guidance.



If a bird was caught alive and placed in a cage in a market nearby in Asia, would it be more appropriate to mark it “not wild” or “location inaccurate” (assuming the location it was caught is unknown)?
What if instead it was killed and is being sold for body parts in a market in Africa?

1 Like


Unless you specifically know the date and time of capture (for example slightly different a museum specimen), and intend to record that, this is a captive animal. It is there because a human brought it there, and it can’t remove itself from the situation without human aid.

1 Like


I don’t disagree with this at all - it’s exactly what I was referring to. If the locality is put in as Cleveland it’s clearly an observation of a captive animal. However if the same photo of the tiger, even if taken in Cleveland, is accompanied, instead, by the capture point in India and date of capture then I would have no qualms about passing it as wild if the data can be reasonably verified - same individual organism and a point and time that it existed naturally.

Because an observation, plant or animal, is not suitable for spatial ecology doesn’t mean it’s not extremely valuable to a researcher in another field of biology. When it comes to data use it’s a case of caveat emptor - apply your filters to the data when you harvest but don’t impose strictures on the data itself that may preclude potentially useful observations from being found by researchers. If a plant observation has an uncertainty of 10km you would filter it out but someone doing a broad-scale distribution study may well choose to retain it. Nonetheless, it should be a case by case assessment. The Penang fish market is large and receives fish from all over the world so, in this case, it seems very reasonable to mark these observations as unreliable in some way (neither of the available options really hits the mark, imo). The circumstances of different observations can be so nuanced, as @mira_l_b has noted, that the context of each does need to be considered before making a decision.