How exhausting plant taxonomy curation has become

[choess]

OK. Having vented a bit, I am a long-time taxonomic curator in a particularly fraught vascular plant area (ferns).

For vascular plants, there are two databases that are intended to cover all the vascular plants in the world. One of these is Plants of the World Online (POWO), the backbone of which is the World Checklist of Vascular Plants (WCVP). It is maintained by RBG Kew, and principally by Rafaël Govaerts. The other is World Flora Online (WFO), which is a consortium aggregating data from a variety of sources (including WCVP). (Tropicos, hosted by Missouri Botanical Garden, is wide-ranging but not universally comprehensive.)

WFO is actually the older of the two databases (2012) and was initiated to meet Global Strategy for Plant Conservation (GSPC) Target 1, but its development stalled out for many years–I assume there were funding problems in the consortium or something but I have no knowledge of this. In the interim, POWO was launched in 2017. In the following year, some Kew-adjacent people published the Global List of Accepted Vascular Plant names (GLOVAP), which was intended to place all extant vascular plants into genera that were (believed to be) monophyletic, which subsequently became the framework for POWO and allowed the creation of a world flora by the target year of 2020.

Unfortunately, GLOVAP was prepared in some haste and was not necessarily well-received. In the areas I’m more familiar with, the authors tended to deal with non-monophyletic genera by very sweeping lumping of genera, in ways that are not accepted outside of POWO. While there has been some improvement over time there are some places in POWO where their classification is very much out of sync with the rest of the botanical community. There’s been no systematic attempt by the GLOVAP authors to maintain or revise their classification, so as new species are discovered, they either become unplaced in POWO or even more alarmingly, POWO declares that the accepted name is one that has not yet been published (“ined.”), i.e., https://powo.science.kew.org/taxon/urn:lsid:ipni.org:names:77217357-1

Also, updates to WCVP are essentially all funneled through Rafaël Govaerts. He’s extremely active and helpful (I’m not sure he sleeps) but for lack of time he’s more or less limited to looking at the most recent literature and seeing if it looks credible.

World Flora Online attempts to solve this problem by parceling out different parts of the vascular plant tree to “Taxonomic Expert Networks” (TENs). When these are well-populated and active, you get well-curated species lists developed by expert consensus, similar to what workers in, say, tetrapods are probably familiar with. In areas like that, “pick one reference and stick to it” does work pretty well. However, my impression is that in some areas of WFO, the TENs have been slow to revise and there are lots of unplaced names and uncertainty, and in others, no TEN is yet available and WFO falls back on POWO/WCVP. https://about.worldfloraonline.org/world-flora-online-curation-red-list-2

So why is this all so controversial? Our vascular plants are nominally aligned with POWO, but because our taxonomy well predates that, there are some areas where iNat taxonomy is arguably “better” than POWO, because it incorporates some specialist judgment that hasn’t really entered POWO, or because it avoids some weird GLOVAP circumscription that won’t change in POWO because of internal politics. Essentially, a lot of the discrepancies between iNat and POWO represent real errors or points of contention that we are surfacing–iNat is playing a role in improving the backbone taxonomy, not just copying it.

From conversations with staff, I know they are concerned with the continued growth of taxonomic flags and deviations and their scalability–they would like to process to be more regular and mechanical like the annual bird updates. But premature alignment to an imperfect backbone also carries a lot of risks with it. In particular, if you lump a taxon that is later recognized as distinct, it’s at best going to require tedious back-end work by staff to reverse the lumping (if people complain immediately after the swap) or it’s going to require laborious manual re-ID to retrieve it. Even one-to-one changes of name create gratuitous database load and leave a trail of re-IDs which I would hope to avoid.

The changes to plant curation over the past year have, AIUI, largely been attempting to grapple with this problem. It’s a difficult one to solve because it reflects the overall levels of uncertainty in plant taxonomy; it’s not just an iNaturalist problem.

[Davidia]

There are a lot of factual errors in your post, which I will not go into but the reason there are so many changes are twofold

1. Before iNat used POWO, we used TPL (The Plant List). So the vast majority of names in iNat come from TPL, both are largely based on WCVP (which started in 1988, POWO is merely a new website), so basically iNat uses taxonomy from 2011.

2. From the IPNI statistics page you can see that last year 9603 new names were added, so that means that POWO/WFO/iNaturalist has to make 9603 changes. To put that in perspective, that is the same as all tetrapods every 3 years. And this does not include new synonymy or resurrections. So basically WFO (and iNat plants) needs a community the size of the tetrapod community added every 3 years, of course that does not happen and therefore many names remain unplaced or changes not made.

So TPL was static from data harvested in 2011 and iNat started using POWO in 2017, that means iNat ran behind in some 57,600 new names.

As to GLOVAP, for vascular plants I don’t think POWO accepts many more then most others. For ferns of course there are 2(or more) schools of thought, but I believe for ferns iNat follows World Plants in any case, though there has been a lot of pushback on this site from implementing that. Turning every clade into a genus is not for everyone….

[choess]

There are a lot of factual errors in your post

I would appreciate you pointing them out. I obviously have opinions, but I am trying to present a veracious narrative here. Some things I cannot know but I have tried to be clear when I am speculating.

I’m not sure quite sure which changes you’re referring to when you say “so many changes”. Many, many taxon changes have been committed since iNat began switching to POWO in 2018. Our taxonomy has definitely not been static since then. Annual novelties in IPNI create work for POWO and WFO, but not necessarily for us. Staff has been explicit about not adding names to our taxonomy unless they are necessary to ID observations, so most of that does not affect us.

I don’t think the volume of novelties are directly germane to the site policy changes on vascular plant curation that have been made over the past year or so. Rather, some of the plant taxa that have an unknown relationship were left there because curators looked at the literature, saw it was not clear whether to deviate or match POWO, and left them alone. So the backlog continues to grow, which staff do not think is sustainable. More recently there have been some high-energy attempts to resolve taxa lacking taxon relationships without background research. Sometimes this cleans up things that should have been synonymized but often it is discovered that POWO ought to recognize that taxon, but only after it has been lumped and lost in a bigger pool of observations. That’s generated a lot of drama reaching from our flagging system to staff email. The policy changes have been an effort to strike a balance between these forces.

Ferns are a bit of an odd case, which I can speak to because I did the bulk of curatorial work on them until a few years ago. When taxon frameworks and POWO alignment rolled out a few years ago, ferns were not ready in POWO, and with the encouragment of staff, I started aligning genera with the 2016 publication of the Pteridophyte Phylogeny Group, PPG I. I did this in a fairly conservative fashion: there were some changes I didn’t make, like Osmunda claytoniana to Claytosmunda claytoniana because it would involve many observations and I wasn’t sure the splitting of Osmunda would endure, and several genera where the New Zealand Flora and PPG I circumscriptions differed, the former having been used as a basis for our taxonomy early on. So ferns are in a weird space where they’re under the POWO taxonomic framework but there are a large number of deviations, generally one-to-one (species circumscription is the same but genera are different).

I am trying to get us switched to WFO for ferns and lycophytes, where PPG is now the TEN, but have not heard back from staff as of late (World Plants is in pretty close sync with this as well). I think wanting fewer genera is a defensible desire (and PPG II has done some lumping, particularly in Microsoroideae) but if new names are neither published in those genera nor recombined into them, then the system is not being maintained and is not very useful.

[zygy]

I still stand by my recommendation. If there’s no single universal reference for plants that is usable, choose different references for each group. It sounds like you’re already doing this for ferns (switching to the WFO). The main goal should be having an easy to manage taxonomy, not having a cutting edge or “scientifically correct” taxonomy (as there is no such thing). Most research museums use taxonomies that are 50+ years old for their catalogs. Even Wikipedia uses 20 year old taxonomies. The standard for mammal taxonomy on Wikipedia is Mammal Species of the World, 3rd edition (2005). I used to be frustrated by the fact that iNaturalist adheres strictly to the World Spider Catalog as it sometimes meant waiting months or years for taxonomic changes to be reflected on iNaturalist. And sometimes I didn’t agree personally with the taxonomy it provided. But after seeing how much chaos and argument there is for other groups, I’m now very happy to wait. And for those people not willing to wait there are observation fields like “Provisional Species Name” that you can use in the meantime. iNaturalist curators shouldn’t be having debates about taxonomy. They should be having debates about what taxonomic catalogs to use and then stick to those catalogs, even though they are all slow and imperfect. If the catalogs are truly unusable, you should as a community figure out ways to improve or replace those catalogs (outside of iNaturalist). That’s what the bird folks did. As the iNaturalist staff have reiterated many times, iNaturalist is not intended to be a taxonomic reference. Sometimes the solution to an intractable problem is accepting imperfection. Now with all that said, I have to admit that I have no actual experience with plant taxonomy curation on iNaturalist so I’m just an armchair commentator throwing out (probably useless) opinions.

[upupa-epops]

@zygy

The main goal should be having an easy to manage taxonomy, not having a cutting edge or “scientifically correct” taxonomy (as there is no such thing).

This is something I keep running into and it’s not clear to me if there’s a formal priority when these conflict. There seem to be many cases where old resolutions have been found to be non-monophyletic, but taxonomists are still working towards a new formal revision. At higher levels the old version is maintained for stability, but at lower levels monophyly seems to take priority a lot more. Although I think the Curator Guide only explicitly requires monophyly for complexes.

@upupa-epops

@trscavo

Does iNat allow infrageneric subdivision based on morphology alone?

Personally I’d prefer using whatever taxonomy there is that covers an entire genus even if it’s old and morphology-based, but a lot of curators will disagree. If the last revision of the entire genus was in 1849 and it’s a well known genus, it’s possible that phylogenetics has messed up the old morphology-based taxonomy, and most curators will want to follow the phylogenetics. That’s been the main hindrance for some genera (e.g. Cladonia, Potentilla) where the phylogenetic work has been fairly recent and hasn’t reached a formalized resolution yet.

[Davidia]

@zygy It is not that POWO is unusable (ferns have never been used from POWO) and if you look at WFO, they have not found “Taxonomic Expert Networks” for most plant families, so splitting plants is a non-starter. Overall it works well, iNatters find differences, they discuss among themselves and collaborate with POWO to get an agreement or not, in which case a deviation is implemented.

The problem is 1. notification; Most interested parties only find out a change was proposed after it was implemented as there is no easy way to monitor them or be notified unless you spend a lot of time and this then leads to intense discussions and this is something iNat wants to do something about by having the 5 vote trial that is currently going on.

2. there are still a lot of deviations that are made without conversations with POWO. Often no information is given at all, which I think should be avoided at all times. iNat could implement a policy that if a vascular plant name taxonomy is changed, it should say something like “complies with POWO” or “does not comply with POWO” in which case another published source must be given.
3. I think we should try and keep some consistency which will always lead to unavoidable changes like not accepting species in genera that are not accepted, so a name change in California may result in name changes in Japan and Mexico. Same for species, we cannot have a situation where a species native to Spain must retain its “old” name because that is the name widely used in the Southern USA where it is introduced.

[Davidia]

@choess since you insist:

“For vascular plants, there are two databases”

This paper says 4: https://doi.org/10.1111/nph.18961 and I can think of Catalogue of life as at least one other current one.

“WFO is actually the older of the two databases (2012)”

First WFO (like POWO) is merely a website that presents other people’s databases. WFO as a website dates from 2019 so that is later than POWO (the WFO institution was set up in 2012). Also the POWO taxonomic data from the WCVP database is from 1988 ( https://doi.org/10.1038/s41597-021-00997-6 ) much earlier than WFO which is a mix of some 60 databases(including WCVP), World Plants dates from 1994.

“GLOVAP…subsequently became the framework for POWO”, If I look at the genus Pyrus, I cannot find any of the 850 new combinations in GLOVAP accepted in POWO, so definitely not a framework.

“….the framework for POWO and allowed the creation of a world flora by the target year of 2020.” That makes no sense, the “framework” as mentioned above was not a framework and POWO is not a “world Flora” as the about pages say, it aims to ”make freely available electronic data created by different projects but that no longer have an online presence or where data was never made available externally” People see POWO and WFO somehow as rivals, competitors, alternatives while in reality they are completely different; POWO merely wants to present legacy data as originally published, mostly associated with Kew, while WFO wants to make an up-to-date flora of the world where POWO is one partner among many that provide data. The only overlap is that they both hang their data on a names backbone which are currently not identical, the WFO latest release says “Updates to the default classification used for the 303 non-TEN families are provided by RBG Kew’s WCVP. Since the December 2024 release the following 17 non-TEN families have been updated to follow WCVP version 12 (May 2024)” https://about.worldfloraonline.org/june-2025-release So they must be collaborating.

I could go on but you get the gist and only get this from what they published, which may not be the whole truth.

As to your other points: “Annual novelties in IPNI create work for POWO and WFO, but not necessarily for us. “ I think about a third are new combinations so that will entail taxonomic decisions.

“I don’t think the volume of novelties are directly germane to the site policy changes on vascular plant curation” Someone on here gave a link to the proposed changes, when I looked, I think there were some 15K and certainly a large part were spellings, https://www.inaturalist.org/taxon_changes/166733 new combinations https://www.inaturalist.org/taxon_changes/166717 old synonymy that does not look controversial https://www.inaturalist.org/taxon_changes/166680 changes published very recently https://www.inaturalist.org/taxon_changes/166715 (publ. 3 Nov. posted 10 Nov.) these are some of the examples from the past weeks, none of which seem controversial in the way you describe but they do make up the largest part.

[zygy]

Thanks for the explanation! I think I understand the problems better now. I agree that it would be very nice indeed to have some method of monitoring flags and proposed changes for an entire group.

[sjl197]

Interesting reading. I sit back with popcorn, but as i watch the world burn, i do continue to wonder how there’s a single online regularly-updated taxonomic database of >50,000 species of spiders since the dawn of the internet, with pdfs for every published work openly available and each taxonomic act detailed, yet, for vascular plants at only 7x the scale that’s somehow impossible.

[Post edit, following from choess with his “Having vented a bit” above. I now just add an appendix in the cold light of the next day, onto my own exasperation from the previous night. I’m really saddened that for you plant folk, it looks like the current implementation on the iNat end requires steps that have increasingly made any action even more cumbersome than needs, on top of delays with adoption of updates in external sources, plus conflict across those possible informative source! Working out what are the solutions and ways forward at the iNat end are of course hard and often unclear, with differing opinions. Anything challenging is by its very nature challenging, but given frustrations above round on several aspects, are we not hearing back anything from iNat central? All i see is Tiwane responding about a sideline issue. Hello? Do they want informed contributions from us, and constructive comments about possible ways forward as abundant in the comments above, or do i pass staff some popcorn and we together watch the plant taxonomy scheme burn?]

[choess]

Thanks for the detailed response! I’ll try to give a condensed summary:

1. You are right, I had forgotten the Lepzig Catalogue. I did not include World Plants in this context (although I carry on a great deal of correspondence with Michael Hassler, and had the pleasure of meeting him this summer) because he has not set up URLs with unique identifiers for each of his species, and does not intend to do so. Therefore it cannot be used as a taxon framework. (I see in his “News” that much of the Catalogue of Life plant data draws on him, though.)

2. I think the discussion of what is a database, website, list, framework, etc. obscures the point I was trying to make, perhaps not very well. TPL was a coalition project. WFO is also a coalition project, intended to replace it, but it was very slow getting off the ground. So POWO, which was intended as Kew’s contribution to the WFO coalition, wound up occupying that ecological niche of “global checklist of plant species”. I’m grateful they stepped up to do so, but of course that means the checklist has more strongly reflected Kew’s particular sensibilities.

3. I’m not sure what is meant by stating that POWO “wants to present legacy data as originally published”. POWO does update parts of its taxonomy as literature is published (I distinctly remember the circumscription of Ranunculus having fluctuated). It is hard to track down changes in what POWO has recognized over the years; the fact that they have abandoned GLOVAP Pyrus sensu latissimo now doesn’t mean that they never recognized it. (There is a complaint about POWO adopting lots of GLOVAP combinations in 2020 at https://forum.inaturalist.org/t/how-do-we-deal-with-the-thousands-of-highly-contested-glovap-new-combinations/10277 but they were accepting Cotoneaster hence rejecting Pyrus s. latis. by 2021.)

4. Even new combinations only entail taxonomic decisions if the original species was already in our taxonomic tree. But yes, most taxonomic changes to vascular plants here were, and are, uncontroversial. The recent agita has come about because formerly there was more use of taxon framework deviations or just leaving taxa unlinked to the framework when people did not want to follow the POWO treatment. Site policy shifted to discourage that, trying to reduce deviations and “relationship unknown” taxa. But that led to more taxon changes that followed POWO but were not popular, were contrary to recent literature or common floras, etc. Further policy changes like waiting period for swaps, voting, etc., are attempting to facilitate cleanup while avoiding taxon changes that will be rejected by the community.

[choess]

Thanks for your sympathy. Vascular plant taxonomy is in an awkward spot, where the different lists agree well enough that fungus-like anarchy does not make sense, but there are enough discrepancies that converging absolutely on one list creates real problems (not just “I think this subspecies should be a full species because I like it”). e.g., https://www.inaturalist.org/flags/790678 where POWO has synonymized a US Federally-listed species for some reason that I can’t follow based on literature.

While I am a little depressed waiting for a reply on experimenting with a fern & lycophyte switch to WFO, when I and another user raised some concerns about the way things were trending in plant swaps, Scott & Tony set up a Zoom meeting and gave very generously of their time to hear things out. It was very informative and helped me understand why staff were making some of the decisions they were. I think the biggest issue for them is just bandwidth–for the size of the site, iNat does not have a huge number of people keeping it up and running and I am sure they are constantly battered by a stream of urgent! things. The silence is probably because they’re swamped, not because they don’t care.