Using iNat for invasive species monitoring?


See my project ( aimed at getting pampas grass (Cortaderia selloana) listed as an Oregon noxious weed (alongside jubata grass that is already so classified). Washington and California already recognize the invasiveness of pampas grass, but we need to prevent the Oregon coast from becoming the eyesore that the northern California coast has become. A listing means that nurseries can no longer sell and landscapers can no longer plant a listed plant. While 3/4 of my observations are cultivated plants, these are the source of the invasive naturalized ones. Unlike the California invasion, the situation on the Oregon coast is not yet hopeless.Through EDRR (early detection and rapid response) perhaps we can stop the invasion. BUT the first step is to recognize the threat–that’s the purpose of this project.



@kueda I am one of the people running Vespawatch, a citizen science project directed at beekeepers, naturalists and the public on surveillance for invasive Asian hornet which is part of the surveillance for Union List invasive species regulated by the EU IAS Regulation (1143/2014) and is run by my institute (INBO). The idea of our site, where you can report observations, was a seemless early warning and rapid reponse supported by IT infrastructure. Reports can be done through our site, or through iNat and we sync back and forth. Once a nest is validated, an email is sent our to the reponsible company (mostly fire departments), who then take action on the nest. They then report on the management action through the site (this last part requires registration on our site and is still under development).

  • I think it is fair to say that linking our website to iNat was very efficient
  • I would say the approach is generally succesful in terms of following the invasion and in supporting management on the species, although we did not quantify engagement or success indicators or other yet
  • it requires dedicated human resources for stimulating volunteers to report, press releases etc.
  • there is another platform widely used in Belgium ( run by the major conservation NGO (and where my institute also supports a dedicated surveillance portal for IAS financially). We did receive a deal of criticism and non-collaboration because we chose iNaturalist (the reason for our choice was the open software which we could easily build upon, the dataflow to gbif, the rather non-naturalist target audience, the possibility of reporting without having to register - I know this is perhaps not iNaturalist’s preferred choice and the swift open-to-all peer validation) and we are still not allowed to show their records on the invasion map on our site. I could have included them as partners, but then using iNat would not have been an option. We did not yet compare hornet reports on both websites, but I think our survey yields more so clearly there is a project effect besides a recording platform effect.

We did have to deal with the criticism of getting many pictures of dead insects, and our account was suspended (and reinstated) twice by iNat curators because some people do not get that they need a picture of the actual thing they observed and thought they could just support their observation with pics from the web, which leads to copyright infringements. You can read more in our answer to some questions raised on this thread

1 Like

The EU IAS Regulation has also boosted the emergence of a few other surveys/record collections on iNaturalist, for instance: (invasive plants in Lithuania)

I know that Spain/Catalonia also have some projects related to invasive species.(e.g. seems to be built on iNat infrastructure and

One element that is perhaps interesting to mention in your talk, is that data from iNaturalist are effectively used in official regulatory processes on invasive species. Data are harvested from GBIF by the European Alien Species Information Network (EASIN) which is the official reporting tool of the EU IAS Regulation where member states have to notify the European Commission on first incursions of Union List IAS and which also serves the EU Regulation’s 6 years reporting by member states (on distribution, actions taken etc.). Hence the importance of the link to gbif.

Thanks Carrie. The FreshData platform is no longer being supported so the National Capital Region PRISM will be relying on iNat for our data needs. It is taking a lot of work to add 18,000 taxa one at a time to the list of taxa to exclude but when that is done it should be a great tool.


Awesome, this is all really helpful, everyone. Some follow-ups:

@murray_dawson, since you created, would you care to comment on how data from iNat actually gets used? Are there any specific stories you can relate of particular observations triggering a management response? If you have ones regarding California natives that are invasive in New Zealand, like Pinus ponderosa / contorta / radiata / muricata, that would be an added bonus for my Californian audience.

Chuck, in this case, you were both the reporter and the manager, correct? So this is sort of a case of iNat providing a way to document this novel occurrence, but not one of iNat helping managers learn about occurrences they might not have otherwise? I guess you’re also implying iNat is helping you monitor the population subsequently, although lack of observations doesn’t really imply extirpation. Documentation of occurrences isn’t something invasive plant people usually tell me about, so that’s interesting, especially considering your corresponding record on EDDMapS isn’t public.

@vespawatch, regarding not being able to use data from iNat and simultaneously, is that because of some policy of’s that they don’t want their data mixed with those from other sources, or is there some legal limitation there?

1 Like

Ken-ichi, you wrote: " especially considering your corresponding record on EDDMapS isn’t public."

I was surprised to see that record as not public. I didn’t intend to hide the record; I’m not sure how public accessibility for data is set up on EEDMapS. I’ll check into that.

The Oregon iMapInvasives Project has been collecting data through iNat, first through the traditional projects and now through a collection project. The contact/admin for the project, Lindsey Wise (@wisel), might be willing to answer your questions.

1 Like

Ken-ichi, I can affirm that Chuck’s addition of the observation to iNaturalist has alerted at least some of us in the vicinity to be alert for an infestation we might not have thought was a concern. I am just a volunteer, and I am not plugged in to EDDMapS, but I have come to trust the information found in iNaturalist more than I trust any database I have found that is purported to document observations of invasive species.

I know, for example, that at least 243 of the 584 observations of Ligustrum japonicum in EDDMapS are of no value. Yesterday, I reviewed the database that is the source of those observations. I found that 90 percent are not L. japonicum at all:

  • The lion’s share are other Ligustrum species—186 L. lucidum, 16 L. sinense, and 8 L. quihoui.
  • Five cannot be identified to species.
  • Two are in some other genus.
  • Two show no evidence of the organism.

Based on the locations of the observations that either are L. japonicum or cannot be identified to genus, none of those plants are wild. In other words, these 243 observations document not even one L. japonicum that has escaped cultivation, let alone invaded anything, but they are used to establish the notion that it is highly invasive. (The other databases behind EDDMapS aren’t necessarily better. Some document only that a plant is known to exist in a county—without indicating whether it’s someone’s hedge or a 20-acre monoculture infesting a preserve. Others document the size and location of an infestation, but lack any photos of the invasive species. It’s the classic case of having lots of data, but little information.)

So far as I can tell, the databases that feed into EDDMapS are based on an expert verification model. Many people may submit observations, but the observations are not added to the database until a designated expert verifies them. There is no built-in mechanism to flag errors or question decisions made by the designated expert. Consequently, the designated experts have no opportunity to learn of, let alone from, their errors. That sets us up for many systematic errors, leading to a situation in which 90 percent of the reported observations of a species thought to be invasive aren’t that species and the other 10 percent aren’t outside of cultivation.

By contrast, iNaturalist allows for immediate publication of the observation. Review and verification starts right away, too—and continues until the community settles on the best identification. In this crowdsourced peer review, mistakes get fixed. Recently volunteers cleaned up data on L. japonicum and L. lucidum in North America.

If you refer to sources that use these expert verification databases, you will learn that L. japonicum is spread by birds, outcompetes other plants, and forms monocultures in fields and forests throughout the South and at least in riparian areas farther west.

If you refer to iNaturalist, you will learn that L. japonicum has hardly ever escaped cultivation—but that infestations of L. lucidum, which is known to establish monocultures, have often been misidentified as L. japonicum. (From this information, a graduate student might be inspired to look for reasons that L. japonicum seems to be well behaved while other members of its genus are so highly invasive.)

iNaturalist is correct. But the databases built by expert verification are the ones that are used to develop policies—policies that determine what plants can be sold in nurseries, as well as policies that determine what tips homeowners are given about the plants they could choose for their own landscapes.

It sure would be nice for the policies to be developed from data that has been demonstrated to be correct. Or would that make too much sense? :-)


Just wondering how your project is going! Can’t really answer your questions but wanted to reach out. I’ve been pulled into a project to use iNaturalist to map and detect invasive shot hole borers ( in socal. We have a traditional project - Santa Monica Mountains Bad Beetle Watch. We are using it a bit differently. There are three “layers.” 1st, observers are asked to focus on 6 tree species (common wildland and urban species that are easy to see beetle infestations), then within these, observers are detecting a presence-or absence- of the invasive shot hole borer. All observations require data fields, such as number of holes detected. 0, of course is the absence. Like others, we are relying on the photographs for the QAQC.
We are still trying to figure out how to get through QAQC quickly besides 1) looking at every single borer hole, 2) having trusted observers that we know they know their stuff.
Any ideas would be greatly appreciated about how you are doing your invasive monitoring!


I don’t really think there’s a way to do that aside from human labor and/or trust. Even if you have some sophisticated machine learning approach to validation, you still have to trust your model, and prove to yourself that you can trust it. Same thing with a crowdsourcing approach.

Honestly, given the fact that you are interested in attributes beyond the core iNat observation fields and that you’re interested in absence, I don’t think iNat is a good fit for your project. I would consider setting something up with Their platform and tools are much better-suited to tasks like counting, and you can take different approaches to QA/QC like requiring that 3 different counters counted the same number of holes in a given image before you accept it as usable data.

To all, thanks for all your feedback. My talk went well, though it was, uh, a bit shorter than anticipated so I wasn’t able to cite many examples, but your feedback here allowed me to say things with a little bit more real-world rigor than I would have had otherwise. One interesting outcome was that I chatted with the folks from EDDMapS and learned that they already have some kind of process for ingesting iNat observations, and an API that allows us to see which records have been ingested, so we’ve started showing badges on iNat pages when we know EDDMapS is using that record, just like we do with GBIF. Here’s an example by @gcwarbler: Interestingly, the record Chuck mentioned before doesn’t show the link, I assume b/c it’s private on EDDMapS and isn’t available through their public API.


Someone took Spotted Lanternfly data and created a nice spread map. Not sure if this is a success or a failure. It’s a great example of using iNat to track a recent invader.

1 Like

Late to the party but I’ll share a bit of my experience. Many of us in the iMapInvasives network are making use of iNaturalist for a source of invasive species observations, and encouraging folks to use iNat to report things they think might be invasive. I wrote a web article about one of our early detection successes for this, at

The iNat invasive sightings are also really good at filling in places that managers aren’t always working/looking (or lacking funding to do). For the more detailed info on management and follow-up, we point people back to iMap for the additional fields/features there. But iNat is a great source of obs.


Surprised it wasn’t mentioned - Euromediterranean Alien Species ( Probably you should ask @blue_celery how successful it is.

1 Like

Thank you for mentioning Euromediterranean Alien Species.
As regards I would say that it has been “lights and shadows” so far. I have noticed that there is a massive addition of observations from Russia and it is really good to know that there is an interest in tracking the distribution of alien species.
It would be great if there would be 2-3 users for each country of the project that would survey what is observed in their area and add the wild aliens to the project.

1 Like

I know there are several of us regulatory entomologists from Canada monitoring what we can of iNat. More eyes on the ground is really useful. As mentioned earlier, the box tree moth was first picked up on iNat and based on that observation inspectors were sent out to conduct a site survey.

Unfortunately, I don’t have enough time to look at everything that comes through and some days I don’t get on to iNat at all. I have set some alerts for individual species, which helps me out, but that’s not terribly effective in the larger scheme of things.


This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.