Proposal: Implementing Species Groups in Asilidae

It has now been over a year and it appears that iNat staff have no intention of adding species groups to the iNaturalist taxonomy. I still believe species groups would be useful for robber flies, so a potential workaround is to use the species complex rank instead. Such taxa can be updated to proper species groups later if the iNat staff choose to add that rank in the future.

Good idea? Bad idea? Let me know your thoughts on this proposal in the comments! I will wait for feedback before making any changes.

Below are my proposed 'species group' taxa to add. What groups should or should not be included in this list? I will update the journal post accordingly.

cc @tristanmcknight @quaedfliegh @robcannings @dannywolff @bradmoon @stevecollins @giffbeaton @b_coulter @pbedell @rogue_biologist @estrada1990

Efferia [following BugGuide and/or modified from Artigas & Papavero 1997]
Complex Efferia aestuans (=Nerax group)
Complex Efferia albibarbis (=albibarbis group)
Complex Efferia anomala (=anomala group)
Complex Efferia arida (=arida group)
Complex Efferia carinata (=carinata group)
Complex Efferia pogonias (=pogonias group)
Complex Efferia staminea (=staminea group)
Complex Efferia tuberculata (=tuberculata group)

Cyrtopogon [following Wilcox & Martin 1936]
Complex Cyrtopogon anomalus (=anomalus group)
Complex Cyrtopogon aurifex (=aurifex group)
Complex Cyrtopogon callipedilus (=callipedilus group)
Complex Cyrtopogon dasylloides (=dasylloides group)
Complex Cyrtopogon dubius (=dubius group)
Complex Cyrtopogon inversus (=inversus group)
Complex Cyrtopogon leptotarsus (=leptotarsus group)
Complex Cyrtopogon lutatius (=lutatius group)
Complex Cyrtopogon marginalis (=marginalis group)
Complex Cyrtopogon maculipennis (=maculipennis group)
Complex Cyrtopogon montanus (=montanus group)
Complex Cyrtopogon nitidus (=nitidus group)
Complex Cyrtopogon perspicax (=perspicax group)
Complex Cyrtopogon princeps (=princeps group)
Complex Cyrtopogon profusus (=profusus group)
Complex Cyrtopogon pulcher (=pulcher group)
Complex Cyrtopogon rattus (=rattus group)
Complex Cyrtopogon rejectus (=rejectus group)
Complex Cyrtopogon tibialis (=tibialis group)
Complex Cyrtopogon nugator (=nugator group)
Complex Cyrtopogon vulneratus (=vulneratus group)

Stenopogon [Following Wilcox 1971 over Martin 1970]
Complex Stenopogon albibasis (=albibasis group)
Complex Stenopogon boharti (=boharti group)
Complex Stenopogon breviusculus (=breviusculus group)
Complex Stenopogon californiae (=californiae group)
Complex Stenopogon inquinatus (=inquinatus group)
Complex Stenopogon jubatus (=jubatus group)
Complex Stenopogon obscuriventris (=obscuriventris group)
Complex Stenopogon rufibarbis (=rufibarbis group)
Complex Stenopogon wilcoxi (=wilcoxi group)

Last Updated: July 23, 2021 10:20 AM PST

Posted on July 23, 2021 05:20 PM by myelaphus myelaphus


Interesting proposal, Chris. I'm out in the field this week, but I'll think about this.

Before I really wade in, a few logistical questions. Have we fixed at all how species complexes reverberate around the internet? I know it's not necessarily one website's responsibility how another uses their data, but I know I was disappointed to see "Laphria sericea complex" iNat observations interpreted as Laphria sericea s.s. over on GBIF. If that's still a problem, I'd be hestiant to make lots of species complexes.

Does the CV model work on species complexes? I just checked the L. sericea complex page and it doesn't have the tag saying it suggests that taxon. Which might just be because we haven't made enough of the observations stuck at that level Research Grade (which connects to the previous paragraph's question, since RG complexes get exported to GBIF). But if the CV doesn't suggest species complexes, I'm not sure if multiplying them would make our triage or classification jobs that much easier.

A few other random questions / notes:
-- One that immediately comes to mind is a Machimus notatus / virginicus group.
-- In Wilcox 1971, Stenopogon inquinatus was a "complex" of one species (not counting unspecified Palearctic taxa that would need someone to dig into the genus to find), so I'm not sure that group would be needed.
-- Similar concerns about some of the Cyrtopogon groups. Do we need groups for singletons like C. anomolus, C. nitidus, C. nugator, or C. tibialis? But on that note, I'm not sure species groups would actually be that effective for reducing the work of IDing Cyrtopogon species on a photo platform like iNat. If you can see the fly well enough to take it to those levels, you can often just take it all the way, and if you can't, you'll usually need rather bigger groups. Especially for females. I think this dynamic would also apply for several other small-bodied asilid taxa with documented subgroupings, like Lasiopogon.
-- Implemented at this large scale, how do we balance fine-tuning taxonomic accuracy with being robust and externally documented? Every version of triaging Efferia has made slightly different groups. Would we get to "fix" various misclassified species (like Eric has pointed out?) or would they be tied to a published outline? Martin suggested some other groups that never got used (like a tagax "subgroup" of the albibarbis group), and Papavero's subgenera collapsed some of the Wilcox groups. But waiting for a good enough phylogeny that would resolve some of these issues to get published might leave this genus with an insurmountable backlog, though. I'll think about it.

Thanks for bringing this up! I look forward to reading more responses and mulling this over. Now-- to catch a fly!

Posted by tristanmcknight almost 3 years ago

I feel a bit like a simpleton here as an amateur (citizen-scientist yeah!). As far as I know we don't officially work with such complexes in Europe. But I think we should with platforms like iNat, it would make my work a lot easier. Users want clear answers based on pictures (often of bad quality). Some complexes might be only temporarily when taxonomic problems are resolved. I would be helped with the following complexes but I am afraid to have overlooked some. The following are the first that come to mind, (only for Europe!!!) but there will be more:

Complex Stenopogon junceus (cervinus, elongatus, junceus)
Complex Dasypogon diadema (bacescui, diadema, gerardi, iberus, irinelae, kugleri, magisi, melanopterus, octonotatus, regenstreifi, tsacasi)
Complex Machimus chrysitis (chrysitis, chrysitis nigrofemoratus, fortis, subdolus)
Complex Machimus setibarbus (caliginosus, cyanopus, intermedius (reintroduced by the germans without publication), setibarbus)
Complex Machimus concinnus (concinnus, nevadensis)
Complex Choerades fimbriata (castellanii, fimbriata, fulva)
Complex Choerades marginata (dioctriaeformis, femorata, marginata, mouchai)

I will follow this discussion... tomorrow we will be studying some Choerades problems which might help in the future

Posted by quaedfliegh almost 3 years ago

I agree that being able to lump together more species that can't be easily separated in photos would be beneficial. I think Complex Laphria canis is a perfect example of this feature working as intended. I've only seen someone take the impetus to advance their ID past the Complex level a few times and personally I think that's because most users curiosity is satisfied when it shows up. The AI is trained pretty well to recognize the Complex and it doesn't bother to suggest the more specific species which means we have to spend less time going in and correcting observations and explaining why. If we can identify it further from the photo that's great but at least we aren't muddying the waters by allowing users to choose whatever they feel like.

To offer an example where I feel something needs to be done, Diogmites neoternatus seems to act as a catch for a large number of observations and we've largely left it alone which perpetuates its status as such. Only ~10% of D. neoternatus observations are research grade, correct me if I'm wrong but this says to me that we don't have a high degree of confidence in our abilities to identify them from photos and a complex could be beneficial.

I am also curious about what Tristan said regarding L. sericea complex IDs being exported to GBIF as something more specific. If that's the case with all Complex IDs I do feel that's an issue and maybe rather than a complex we should just be taking a harder stance on pushing certain ambiguous observations back to Genus?

This is a bit of a tangent but Chris and I talked about this briefly so I figured I'd throw it out there and see if anyone else had thoughts. I am taking issue with a lot of these Leptogaster and Machimus sadyates/notatus observations that have been popping up since the AI got updated. I don't know how accurate the Machimus ones are, but I'm also relatively confident neither do the observers. Same idea with the Leptogaster observations. It seems like the status quo right now is to leave them alone if we can't verify but doesn't that just make the problem worse?

Overall I think in lieu of some sort of species group option, a complex sounds like it fits our needs as long as we aren't producing lower quality data because of it.

Posted by rogue_biologist almost 3 years ago

After thinking of this some more, I think I'd more favor making complexes for targeted groups of species that are too difficult to tell apart from photos and not as a proxy for us to overlay taxonomic species groups in general. I think complexes will be most useful for triaging morphospecies out of large speciose genera and will have more power if we aren't worrying about making them comprehensively monophyletic clades. They seem to be working fine in this style for our eastern North American Laphria. I also think it'd be good for us to periodically revisit our complexes so we can evaluate whether they're helping or whether we're picking up on ways to split the taxa more effectively.

For example, I think a Complex Efferia aestuans would be most effective if it's just restricted to aestuans + kondratieffi and maybe apicalis, and not all the "Nerax". I can hope to eventually learn how to recognize something like E. belfragei, but I apparently have no hope of separating E. kondratieffi from E. aestuans without dissection.

Similarly, I'd rather see a Complex Efferia tagax (tagax, neosimilis, incognita, sonorensis, armata) and a Complex Efferia zonata (zonata, peralta, vertebrata, fisheri) that aren't collapsed into a massive Complex Efferia albibarbis. I think those are where I run into dead ends when I'm IDing this group from field pictures.

As for other Efferia groups, at first I felt resistant over a carinata catch-all since there are some taxa in that group that are relatively easy and others that are hard. But maybe it'd make sense to make a carinata group for all them so we can more readily find and sort these and then maybe in future years we can propose more focused complexes? (Or talk the admins into a new level...) I guess the same might as well go for the other Efferia groups.

Some others I think I'd recommend include:

-- Like I said before, a Complex Machimus notatus (notatus, virginicus). This would be especially useful for females in Northeastern North America. Other Machimus groups would also help but might need some more refining. For instance, perhaps a paropus complex (paropus, snowii, erythocnemius) might make sense, as might one for novascotiae + autumnalis + frosti. I'm actually not too bothered by recent M. sadyates since I try to curate that one periodically and it's time for that species to start showing up (though some of them are indeed wrong and the observers certainly aren't making an informed decision). I have been wondering what to do with the new CV model's insistence on recommending Tolmerus for Nearctic observations, though-- it seems there's a few every day now that I have to boot.

-- Complex Diogmites angustipennis (angustipennis, bilobatus) comes immediately comes to mind, since it accounts for a lot of ~western USA observations and they're indistinguishable without a microscope. I also agree that D. neoternatus has become a catch-all that needs cleaning, but I don't think making a neoternatus group is the solution since it's apparently the only species in its taxonomic group in the USA. So I guess we'll need to make some other species complexes for lookalike taxa to break them out of neoternatus. Possible, but I'd need more time to study those taxa.

-- Complex Stichopogon trifasciatus (trifasciatus, coquilletti, venturiensis). This would let us pull down trifasciatus-like things in Southern California where the photo is too far away to judge setation or from Central America where we're not sure it's the same species.

-- Promachus is a mess (what else is new?...) I think I'd recommend at least three groups to help with our Nearctic taxa: a Complex Promachus bastardii (bastardii, quadratus, texanus, dimidiatus), Complex Promachus aldrichii (aldrichii, nigrialbus), and Complex Promachus giganteus (giganteus, magnus, minisculus).

-- Complex Laphria vultur (vultur, asackeni).

-- Complex Proctacanthus rufus (hinei, rufus). I might be able to get Riley Nelson to suggest a few others for that genus-- he's started working on it again with a new masters student.

-- Perhaps a Complex Atomosia puella? I see that Eric Fisher has suggested separating puella from melanopogon might be more tricky than I thought from Barnes 2008:

[apparently there is a character limit. To be continued...]

Posted by tristanmcknight almost 3 years ago

[cont. from above]
-- For Lasiopogon, some complexes might prove useful, especially for females and especially for the European taxa. Though I'm not sure it's worth the extra step (I doubt there are enough observations of any of these to get picked it up by the CV). But here's what I might recommend if we do this, listing the currently valid species.
-- Complex Lasiopogon montanus (bellardii, montanus)
-- Complex Lasiopogon apenninus (apenninus, fourcatensis, grajus)
-- Complex Lasiopogon cinctus (cinctus, intermedius, macquarti, novus, peusi)
-- Complex Lasiopogon aldrichii (aldrichii, pacificus, yukonensis)
-- Complex Lasiopogon opaculus (appalachensis, marshalli, opaculus, piestolophus, schizopygus, slossonae)
-- Complex Lasiopogon actius (actius, dimicki, puyallupi, ripicola, willametti)
-- Complex Lasiopogon drabicolum (anaphlecter, californicus, drabicolum, esau, nelsoni, odontotus, sierra). If you want to make it monophyletic you should also add arenicola and asilomar, though they're easily distinguishable from the rest.
-- Complex Lasiopogon bivittatus (bivittatus, canningsi, tumulicola)

-- I want to mull over the Cyrtopogon groups some more.

-- Some complexes would also surely help with Ommatius and Leptogaster, but I don't know them well enough to make good proposals yet.

Ok, that's what I've come up with so far. Thanks to all for your contributions so far!

EDIT: I meant to ask about Holcocephala. Would it help to have a species complex for H. abdominalis / fusca? I wonder if by always bumping those up to genus but leaving H. calva as the main possible species-level tip the CV won't realize that they are different.

Posted by tristanmcknight almost 3 years ago

Thanks for mentioning this to me. Sorry it took me a while to get to it. Most of the things I would have mentioned, and a lot more, have already covered above. I'm not an expert with robber flies but am slowly coming to know a decent proportion of local species with some confidence. Most of what I can say is centered on species in my region (east Texas across Louisiana and maybe as far east as the Florida panhandle). So treat my comments as coming from a biologist with some but limited knowledge of asilid systematics and taxonomy.

Overall, I like the idea of using more complexes for asilids and have wondered vaguely about suggesting somethign like it to curators at some point, at least for some species in my area. It seems worthwhile for narrowing down an ID to the lowest level for which we have confidence, which I think would be an improvement over leaving so many IDs at genus level, especially for large genera. But the utility and value of it would depend heavily on how well supported and stable the species groups are, which is well beyond my knowledge right now.

If species groups used in complexes are not very well supported yet (I'd guess usually because so few people are studying them), then complexes could be used for species that are just generally similar or poorly described/differentiated in the literature, such as Cerotainia albipilosa/macrocera, Holcocephala abdominalis/fusca, Holopogon guttulus/phaeonotus, and many species Diogmites, Promachus, and Proctacanthus, but that risks using complexes as somewhat arbitrary and potentially large categories that could end up being used a lot!

More narrowly, species reported in the literature to be impossible to tell apart except by details requiring microscopy or dissection would be good to put in complexes. For example, Diogmites angustipennis & bilobatus in this area. I suspect this is what the complexes currently in the iNat taxonomy reflect, and a few more such complexes could be helpful even if not as many as for the species groups listed above or generally similar/related species.

Overall, I like the idea and would use complexes that apply to the species I observed. But deciding how broadly or narrowly to set them up and how many to use seem challenging and largely beyond my expertise at this point.

Posted by bradmoon almost 3 years ago

The consensus seems to be that species complexes (closely related species that are often indistinguishable from photos alone) would be more useful on iNat than species groups (a more broad grouping of related/similar species). I agree and so will table plans for species groups for now. I have an existing journal post for species complexes here, so I suggest we move our conversation there.

Posted by myelaphus almost 3 years ago

I like the complexes you listed for Phases 2 and 3, but I have some comments:

Complex Diogmites angustipennis (angustipennis, bilobatus, texanus)
My understanding from Barnes' paper is that angustipennis & bilobatus are indistinguishable except by dissection of reproductive parts. However, I don't understand why you listed texanus in that complex. Hyaline wings that at most are lighlty infuscate along veins characterize angustipennis & bilobatus, whereas dark smoky wings characterize texanus. At least in older literature, I think I read that missouriensis is similar to texanus except that the wings are only lightly infuscate (vs very dark in texanus). I could be missing some literature, but if not then maybe make a angustipennis & bilobatus complex that doesn't include texanus.

Complex Efferia aestuans (aestuans, kondratieffi)
Efferia femorata is reported to look essentially just like a very large aestuans. The main features reported to distinguish femorata are large size (seems true), a long stump vein (seems true by somewhat variable in my limited experience), and tree-trunk perching behavior (the only place I've found femorata so far, whereas aestuans can be found in much more varied microhabitats. Would it be good to include femorata in this complex? I'm on the fence with this one because femorata does seem distinct in the ways noted above, although otherwise specimens look like large aestuans.

Complex Proctacanthus rufus (hinei, rufus)
I find these two reasonably easy to distinguish if one has good views of them. But I suppose there is utility in this complex because they don't always allow good views or close photos, so it may not be possible to distinguish these species in a lot of observations.

Another possible species complex for Phase 3:
Proctacanthus longus, milbertii (although it seems that milbertii flies much later into the fall around here--it, or whatever it is, is the only Proctacanthus flying here in October)

And a minor editorial question to consider as you setup the complexes:
Is there a system for which species name to use in the main name of a complex? If there isn't a specific rationale, such as the species with taxnomic priority or the wider range and therefore more likely encountered, then why not name complexes alphabetically based on the species included? For example, use "Complex Cerotainia albipilosa (albipilosa, macrocera)" rather than "Complex Cerotainia macrocera (albipilosa, macrocera)". There are several complexes in your list that aren't based on alphabetical order. I wouldn't recognize taxonomic priority or broader range at my level of expertise, and I don't have any problem with any one of those rationales or another one I haven't thought of or noticed yet. I just tend to notice details like that from my writing and editorial experience (and so I'm good for proof-reading things and often willing to do that if it would be helpful at any point).

Posted by bradmoon almost 3 years ago

I've updated the D. angustipennis complex.

Regarding naming of complexes, the curator guidelines state: "If a 'principal species name' is not established in the literature, use the earliest published species name for the name of the complex." As far as I know, most if not all of these species complexes proposed in Phase 2 are not named or established in the literature, so the iNat complex name would need to be the earliest published. Since complexes are not governed by the Code, I don't think this is particularly necessary and I would rather they be named after the most widespread or well-known species. I'm open to debate on what these complexes should be called.

Posted by myelaphus almost 3 years ago

I agree that complexes are a great idea, especially if it helps with iNat's computer vision. I hesitate to recommend complexes for poor photos when species can be identified from stellar photos. In those cases, leaving the ID at genus seems the more appropriate route.

With respect to complex naming, it sounds like the curator guidelines are to use the earliest species name for the complex even if the complex is not established in the literature.

Posted by stevecollins almost 3 years ago

A technical question...since these "taxa" aren't defined in the literature, where will their definitions be published? How will people know what organisms to place in each "taxa"?

Posted by pfau_tarleton almost 2 years ago

@pfau_tarleton All of the groups listed in the first post have been published in the literature. But as species groups, not as 'complexes'. I don't think that any asilid pub has actually defined/delimited a species complex. If I were to create a species complex that only exists on iNaturalist, then by necessity the species included by me in that 'taxon' will serve as the basis for the definition of that 'taxon'.

Posted by myelaphus almost 2 years ago

Add a Comment

Sign In or Sign Up to add comments

Gracias al apoyo de:

¿Quiere apoyarnos? Pregúntenos cómo escribiendo a