Uploading garden plants as wild: The potential for false naturalization data to spread

Hi. Since using iNaturalist I’ve noticed a huge amount of garden plants being uploaded as wild. I do my best to mark them as such, but sometimes it’s simply not possible, especially when a species has thousands of observations.

I worry there is an issue when it comes to this websites potential for spreading misinformation on a species. I for one was amazed to see that the African lily (Agapanthus praecox) has such a wide wild distribution within the UK… However with further investigation I realized a majority of the observations are plants growing in gardens or in plant pots. (The species is a popular cultivated garden plant.) I tried listing some as casual, but there is so many I don’t think I can possibly get them all.

If someone was studying the global distribution of a species, such as Agapanthus praecox, they might not investigate the data further and could just use it in their scientific papers.

I worry that iNaturalist will gain a reputation for being as untrustworthy. There seems to be no real measures put in place to prevent cultivated plants reaching research grade status… People of course can list an observation as not wild, but that doesn’t remedy the problem. I’ve seen countless clearly cultivated plants reach research grade over my short time on this website.

It’s clear to me that although the data this website harvests is useful, it’s also undeniably flawed. Expecting me or the other sorry people to go through thousands of garden plants and list them as casual observations simply isn’t fair.

There should be a classification of species that includes all popular ornamental garden plants and domesticated animals. When observations including these species are uploaded within a country they are not native a marker should come up saying that it’s either a popular pet or ornamental garden plant within that area. I suggest for these species instead of having a box people tick that says it’s cultivated / captive, there should be a box you tick to say it’s wild.


Aside from every user having the ability to mark observations as “not wild” / “captive/cultivated”, once enough of a particular species have been marked so in a particular place:

  • The system will [automatically] vote that the observation is not wild/naturalized if there are at least 10 other observations of a genus or lower in the smallest county-, state-, or country-equivalent place that contains this observation and 80% or more of those observations have been marked as not wild/naturalized. (source)

so there are some measures built in, though not perfect. I sometimes create a flag and @ tag others for help with marking observations as "not wild’ when needed.

If a researcher is not investigating the source and quality of the data used in their publication, that is pretty irresponsible “science”. I don’t think that is something for individuals to be terribly worried about.

If you have a proposed new feature, you can post it to Feature Requests using the template there.


You can copy and paste the canned responses from this page when ID’ing. It helps educate the offenders so they don’t keep uploading garden plants.


Do you mean, “I do my best to mark them as NOT WILD”?


I’m a researcher, and while I agree that garden plants can really mess up the iNat range maps, I’m not too worried about that leading to bad scientific research. Good researchers don’t just randomly grab data from somewhere and use it. They examine it first to check the quality and then clean it up if necessary. Good journals have peer reviewers to check submitted papers and see what they think about the author’s data quality protocols. Even after publication, other scientists can comment on the work if they see something wrong, so hopefully there are enough checks and balances in the process to prevent bad data from getting too far.


if people are using data (from any source) that isn’t theirs in their paper without any of their own analysis of data quality, this is very very poor practice and the paper shouldn’t even make it through peer review. So yes it is important to mark things as cultivated when they are (with all the caveats for grey areas) but iNat can’t be responsible for what basically accounts to scientific negligence.


Based on my experiences talking to new iNaturalist users, I think taking a picture of a garden plant to see what it is without being aware that it should be marked cultivated is a fairly typical use case for iNaturalist, at least as far as the general public knows about it. Part of it is iNaturalist’s reputation as an app that simply tells you what stuff is. Many of those users should probably use Seek instead but their observations end up on iNaturalist, maybe even without them realizing they are contributing to a database of nature observations or that there are human beings looking at their observations to provide IDs or pulling data from this for research.

Others routinely help clean up the mess, but there is no expectation for anyone to volunteer their time doing that. As others have said, the main responsibility to curate data for research projects falls to the researchers using this data.


I’ve had the same experiences. Also, a surprising number of people have no idea what the distinction between wild and cultivated even is - very few people have home gardens anymore, and even fewer plant or tend landscaping. People seem to assume those flowers in the parking lot planter boxes just grow their by themselves, and don’t even think to wonder about how they got there.


And it gets more complicated when you have a native plant garden, and plants start reproducing on their own, i.e.: “volunteering”. Some plants in my garden I presume to be Not cultivated because:

  1. they are already widespread, e.g.: Viola sororia, Ageratina altissima
  2. I didn’t plant them.
  3. I maintain an accession list of all the plants I’ve acquired (and killed) over the years. If it’s on that list, even if it shows up where I didn’t plant it, I mark it as Cultivated.

Observations of Cultivated plants (correctly marked as such) are still of scientific value! They just won’t automagically get exported to databases such as GBIF. Phenology, for example, is used to document changes in seasonal climatic conditions. I document a lot of insect-plant interactions. Knowing the associated/host plant is important, so having an associated observation of the plant that can be verified, or corrected, by others is important.

I have notifications set for new observations in my area (Kings County, NY, aka "Brooklyn). When I notice a bunch of garden plants left as “wild”, I’ll correct them, and give the user (generally someone who recently became active) some tips for making good observations.

Now, what to do with all my honeybee observations … :-)


I do tidy up distribution maps for outliers of our South African plants. But. I have to tread carefully, because they may indeed be Wild, and invasive there.

I will only tackle the few, can’t imagine how many Ag prae there are!


I have a healthy clump of goldenrod that was all volunteers, once they flower I’ll probably throw them up on inat to get an ID - and technically theyre not casual since they were self seeded.

I probably wont upload them after that though unless i notice something new, since i am tending the patch


btw - things to document to get a good species-level ID for goldenrods:

  1. view of the whole plant
  2. view of the main stem at mid-height (just below the flowering section) showing any hairs present there
  3. views of the top and bottom of a leaf coming off the stem at mid-height. Wrap the upper surface of the leaf around your finger to get an easy in-focus shot of the hairs (if any) on the leaf underside.
  4. side-view of a flowering branch/stem
  5. length of the longest phyllary

(jargon explanation: Phyllaries are the green parts analogous to sepals that surround each “flower”. The things that look like yellow “flowers” in goldenrods are technically bundles (‘capitula’ or ‘heads’) of two types of tiny flowers (‘florets’). Some of those tiny flowers look at first glance like petals (‘rays’) and some look like the middle of a normal flower (‘discs’). The “flowers” are bundled up with greenish, scale-like structures analogous to sepals (‘phyllaries’ or ‘involucral bracts’). The absolute length and shape of the phyllaries is often very important for an ID.)


Wonderful info, thank you!

1 Like

As others have said above, any scientist should know not to do this, and most do know. That said, as a scientist who has often done the work to figure out how to properly use data collected by others, any dataset is more valuable if those who collected it made attempts to clean it up. The people collecting the data will be able to spot problems with it that may not be apparent to, or fixable by, even the careful data analyst.

On a different note, I often, with the full iNat definition of captive at hand, and making an effort to use it properly, simply don’t know if a plant is wild or captive. Did someone plant it here? I can’t tell. Is this the individual that was planted, or its offspring? I don’t know. In these frequent cases, I do not mark the plant as captive, which makes it wild by default. That is just inherent to the structure of the data, and so we will never know if a plant is definitely wild or just of unknown status.


Agapanthus in the UK is actually the perfect example of this - it’s a casual in much of the south and outright naturalised in Cornwall and Scilly. I know of several plants which seemed to be in natural scenarios (I doubt anyone’s planting them in pavement cracks or on a forest path where the only thing near is a car workshop). Given the right conditions they can persist for a good while though outside those two areas I don’t think they’re considered truly naturalised.

iNaturalist does automatically vote things captive if that species gets voted captive a lot. Half the time I upload a picture of English Lavender (a decently common garden escape in pavements and other “near cultivation” places, but hugely underrecorded on this site) in the pavement or in woodland near gardens I have to vote it up because so many of the observations of it here are actually garden plants.


I know for example there are clusters of Narcissus sp. all throughout Cuyahoga Valley National Park that are definitely not native and are definitely old remnants of homesteads that used to be in the area but are, obviously, no longer than. But the plants remain.

Obviously, these were cultivated plants initially, but over the years the patches have grown and they have spread - so clearly some of those plants have naturalized, but how do you tell which ones exactly? There’s really no way.


I’ve gone on sprees about that. I pick a map region, filter for a species for which people do that a lot (such as Bougainvilleas), and go assembly-line style marking them all cultivated. Sometimes I find that someone else had already done so, but the observer countervoted.

…and, every now and then, I see a Bougainvillea that does appear to be an escape. I deal with those on a case-by-case basis as I see evidence.


Very true, it can be difficult to tell the difference between cultivated and wild unless it is your own garden or you know the history of the site. Typical ornamental plants or crops in pots or mulched garden beds or showing signs of pruning or other care are very likely cultivated. But even plants in pots can be “wild” if they are the result of reseeding on their own. For example, I routinely come across “potted” stuff like bittercress, wood sorrel, or burnweed in my area - all common garden and nursery weeds that I doubt anyone would plant on purpose.

Same in the Smokies and around some areas here. Old graveyards can have this, too. The houses/grave markers are long gone, but some of the plants planted around them still grow in these areas. And with long-lived plants like trees and shrubs it can be almost impossible to figure out whether something was planted a hundred years ago or not if there is no longer an indication of human structures nearby.


As a researcher that uses iNaturalist data, I think the mission of getting people interested in wildlife and research is very important; potentially more important than the mission of generating an ecological database. It is true that research grade items aren’t usable in raw form, but I think that’s the case for any leading ecology database. They always have errors, flaws and missing metadata. OBIS for example is not a public-facing database, but no one would expect to be able to use its raw data for a study without pre-treatment and cleaning.
Databases are very messy things, but we wouldn’t have the breadth and depth of iNaturalist’s coverage if not for its inclusivity and accessibility. Right now we are openly advertising for people to submit observations of dead fish resulting from the current harmful algal bloom developing in SF Bay. We are doing this with full cognizance that there will be misidentifications, duplicates, incorrectly tagged locations, etc. But the data is still valuable and we wouldn’t be able to cover hundreds of miles of coastline fast enough ourselves!


in Australia, Agapanthus is increasingly emerging as a nasty, naturalised weed, and I’ve been seeing it pop up and proliferate more and more in bushland over the past few years. For example, a bushland reserve near my house abuts a golf course where Agapanthus is planted prolifically. Surprise surprise, the Agapanthus has invaded the reserve and is starting to naturalise in there.