Add interactions to species pages

FWIW, yes, semantics (what things mean) and ontology (what things exist and how they’re related) are important, and there has been a lot of work done on this, e.g., sometimes including/inspired by iNat discussions. GLoBI does try to stick to established ontologies, but one problem with the obs field approach is that iNat users aren’t necessarily aware of the implied semantics, e.g. we all need constant reminders of what the difference between “eating” and “preys on” might be. Something we will need to keep in mind if we ever get to this feature.

Also, my personal opinion is that iNat should only focus on interactions that can be verified in a photo, so a photo of a birch in a hemlock forest doesn’t prove an interaction. The co-occurrence is interesting, but that could be derived from existence of an observation of a birch and an obs of a hemlock at the same time and place. Same thing with pollination: a bee covered with pollen on a flower doesn’t demonstrate that pollination is happening, though it is highly suggestive.


Makes sense. I won’t add any more of those. However, if you can see the birch tree and a bunch of hemlock growing all around it in the photo, would you include that, or no? (Nowadays i’d just do the same thing by mapping both in Hemlock Forest natural community anyway, so i don’t mind not doing it! Sorry if it caused any confusion)

1 Like

If you’re asking would I enter that as an interaction, I don’t know. Gut reaction is “no” b/c the photo only demonstrates co-occurrence and not interaction, but it depends on the biology. If biologists have found that that kind of growth pattern always implies a symbiotic interaction, then sure, a photo like that demonstrates the interaction.


sounsd like broadly, since it doesn’t provide any concrete value to others and it isn’t necessary for me to get the info I want out of observations, it just doesn’t make sense for me to do it. Which is good because i haven’t been lately anyhow. I’ll continue to use the fields in the more traditionally obvious interactions like an insect visibly eating a plant, etc

1 Like

Hey, this is super coo!! I hadn’t been aware of that test=interactions thing. It’s certainly quicker than mucking about with URL strings in searches. I’m pleased to hear that you’ll keep it around.

I’ve been waiting for a long time for a proper way to connect together two observations with an interaction. :grinning:

I’m less keen on the suggestion that an interaction needs to be verified in a photo every time a user sees it and wants to record it. I can imagine casual and research grade interactions just like we do with taxon identifications. If someone sees a falcon take a pigeon but doesn’t manage to photograph it, I’d still like them to be able to add a pigeon and a falcon observation and connect them with an interaction (if they want to).

That’s a different issue from people jumping to the conclusion that it must have been pollination if a bee was on a flower, or that two plants must have been competing if they were growing next to each other. That will be trickier to manage.


How can we get this higher up the “desired” list of features?
Both the New Zealanders and southern Africans have projects dealing with this.
Ours is visible at

Basically, we record only the active interaction (i.e. “a eats b”, not “b is eaten by a” - the latter just being the reciprocal of the first), although user pressure has resulted in us adding a passive field for the reciprocal observation, given that observations fields link only one way, so that these observations do not display their hosts) as:

Visiting flowers:
Attached to:
Associated with:
& the passive

Note that in each case the field value is the url of the interacting observation. Unfortunately we cannot use this is a query to summarize the interactions.
We can ask
“What flowers does the Cape Sugarbird Visit?” -
but we will only see the bird, and not the flowers, even though all the urls to the flowers are in the field - see:

In over 5 years of using this “set” of interactions, we have never had a request to add additional interactions (e.g. Eating = preys on = killing to eat - i.e. “killing for fun” has not cropped up), although it would be nice to have a hierarchical dictionary of interactions (e.g. visiting a flower > pollinating a flower (> for nectar, pollen, oil, gum)/robbing a flower/, etc.

Just a request for - is that it is searchable/filterable and displayed in explore/observations - species view in descending order (and perhaps also taxonomic order?).

Oh: these may tickle you?
(Note that one chilli bush has over 500 interactions! -


@tonyrebelo Impressive stuff. I like the way you’re doing this with observation URLs rather than taxon lookup fields (I wish I’d done it this way when setting up the NZ interactions). I’d bet Ken-ichi and team could build an elegant interface onto something like this.

This gets one of my votes.


linking to individual observations is great. It might mean you sometimes would need to create a casual observation to link to but that would tell you that one of the organizations in question couldn’t be verified with media - which is fine!

1 Like

Wow, that is an impressive series of observations from the chilli bush! And what a terrifying beetle larva!

I thought I’d mention how I’m currently recording interactions as there seem to be different ways of doing it. I use the observation field “Associated observation” to add the URL of the other species - I try to make observations of both species when I can. Then I use one of the following observation fields to add the taxon, which has to be manually updated as and when the other taxon gets more finely identified:

Interaction->Herbivore of:
Interaction->Visited flower of:
Interaction->Preyed on:
Interaction->Ate seed of:

For example:

There seems like a definite need to standardise how this is done, and it would be fantastic if the interaction updated itself via the link whenever either taxon got more finely identified.

As to the question of how to get this higher up the list of desired features - it’s all based on how many votes it gets!


I think adding a way for users to link different observations is important. This would facilitate documentation of predator-prey interactions, mutualisms, habitat associations, pollination events, etc. Unless there’s a way to associate multiple iNat entries, it may be difficult to properly document these interesting natural history observations. One simple method to achieve this would be to allow multiple taxa to be identified in a single observation.


Hi @cypseloides ! I hope you don’t mind… but I moved your post. there is a thread about this very thing! In terms of adding interactions to observations. There is also discussion here about linking the observations more formally though that does not exist yet. Great thought, that’s why it keeps cropping up!

1 Like

Oh great, thanks! I scanned through the other topics, just to see if it’d been suggested before, but I guess I missed it.


All, I think a good first step would be to allow multiple taxa to be identified within a single photo. As iNat operates now, if a bee is observed gathering pollen from a flower, the same photo will have to be uploaded as two different observations, once for the bee and once for the flower. Any interactions that will be added at later date with more sophisticated methods will likely be lost for the bee/flower observation, since there’s no easy way to search for multi-taxa observations. Yes, observers can add tags, but as has been mentioned above, the tags system is a real mess, and there’s no way to add tags when submitting observations from the app. Allowing multiple, independently verifiable taxa per photo would at least be a good first step, and would put in place a framework for later, more complex interaction identifications.


i would like that feature- i imagine it being like tagging people in a facebook picture where you can put multiple people in one photo. You can currently copy an observation and the copy uses the same photo, but there’s no indication of that which i am aware of.

1 Like

Might a better solution perhaps be to - instead of allowing multiple taxa to be identified on the same photo - to “explode” the observation to multiple observations with the tagged species being made a cropped image #1 and the entire image #2.
Each observation can then be individually identified.

Of course, you can already do this on iNat by just hitting the duplicate button, but it would be handy to have a cropping or marking tool, for those too lazy to do these edits on their own computers.

@cypseloides - your solution is not a solution. You now have multiple identifications, but no way of recording their interaction or type of interaction. I am afraid you will have to resort to observation fields (not tags!!!), of which there are several workable alternatives - but only if one has alternative observations that can both be independently identified and verified.

But I for one will be asking for interactions to be upgraded from an inadequate set of observation fields to a more robust module that will allow one select one or both partners (at various taxonomic ranks) as well as an interaction, in a “filter box”. Questions such as: What aphids species eat English Oak? or: what daisies are visited as flowers by hummingbirds? or What interactions are recorded between cattle and insects? or even: What interacts with Dewsticks?

The sticking point at this stage is the interaction types. We need a hierarchical dictionary for these, that will satisfy the nitpickers, but still be useful to the casual layman. And which, like the identifications, can be community verified and improved.

I suspect that we will need a working website version, before we can have an App version. Sorry.


i don’t think so. The observer can already do that. The problem is there is not a clear indication that copied observations are linked (though creating one might be one angle to solve this). It’s also neat to see the organisms together. I think an in-website cropping or selected zooming option is another neat feature but off topic for this one.


Selected zooming would be great! Cropping might not be a good idea, because so many photos are already low resolution – cropping would make them even lower. @charlie, I’m not sure if this is what you mean, but maybe have a clickable box around each organism in an observation, which when clicked would bring the part of the photo bound within the box out.

@tonyrebelo, my solution wasn’t intended as a solution, but instead as a stop-gap between now, when there’s no way to link observations at all, and some future point, when interactions can be easily documented. If multiple species could be tagged in a single photo, then it’d be easier to search for those observations, rather than going through 1000, 5000, 50,000 or more observations one by one, searching for interactions.

It would also have the side benefit of reducing the number of duplicate photos, which just take up space with unnecessary redundancy. For example, this observation: is completely redundant with this observation:; if I could have just tagged each fish, then the load on the database for these two observations would have been halved.


It depends how you do it. See the interactions project
e.g. all observations of Dewsticks with interactions:

click on any one: e.g.
and you can immediately see in the observation fields that it has interactions (or links) with
Passive Partner to: (Interaction):;

The idea of an interactions module is to provide a table that links these - in this case something like:
18719545 “associated with” 18719540
18719544 “associated with” 18719540
18719543 “associated with” 18719540
18719543 “eating” 18734954
18719545 “eating” 18720598

These links already exist as observation fields and can just be clicked. But unfortunately at present, cannot be simultaneously displayed and filtered.

1 Like

There is already a thread for this: Note that duplicating does not duplicate the photograph: the one instance is shared between observations.


1 Like

Yes, I have long wanted ways to properly connect observations, it would be good if it was a more general mechanism than just dedicated to interactions. Eg to connect observations of the same entity over time, or by different observers.

I don’t think it was mentioned above, but there is also a discussion of being able to label elements within the photos at


1 Like