InsectNet forum

According to the Guide to the Butterflies of the Palearctic Region(part 1 second edition, 2024), Graphium mullah is divided into three subspecies: ssp. mullah, ssp. kooichii, and ssp. chungianus. However, based on the phylogenetic tree constructed using mitochondrial genome analysis in the paper Complete Mitochondrial Genomes of Two Insular Races of Pazala Swordtails from Taiwan, China (Lepidoptera: Papilionidae: Graphium), the relationship between Graphium mullah mullah and Graphium eurous asakurae appears to be closer than that between ssp. mullah and ssp. chungianus.

Did I misinterpret the tree in this paper, or could there be any other possible explanations for this?

All the best,
Wu Ming Hsuan

Wu Ｍing Hsuan wrote: Sat Dec 07, 2024 5:53 am Did I misinterpret the tree in this paper, or could there be any other possible explanations for this?

No, you didn't misinterpret the tree, and yes there is an explanation.

Unfortunately the eurous group (both mullah and eurous belong here) is very poorly resolved using mitochondrial DNA alone so a tree with very few specimens (one of each) produces anomalous topography. This is why the third part of the subgenus Pazala analysis has been delayed due to the time it is taking to run full genomic analysis on the whole group.

That is why the Bozano 2nd edition did not update this species group. Bozano asked me to help with a second edition but I asked him to wait until our current analysis is finished (it has taken way longer than anticipated because genomics is needed - these species and subspecies separated relatively recently compared to the mandarinus group species which were very well delineated by mtDNA alone [genomics confirms the findings in this group by the way]) but he went ahead and published a 2nd edition without my involvement and sadly followed the 'peer-pressure' results published in Huang (2023) - many of Huang's friends didn't like the idea that there could be several species within what looked to them like a single species. A number of Huang's illustrated mandarinus group specimens were misidentified (his so-called 'hybrids' were actually 1st generation G. confucius, and several other specimens were misidentified) and he made assumptions about the Vietnamese taxa, and as a result his findings on Pazala in that paper were flawed.

Adam.

Thank you Adam. You truly know the best.

While reading this book, I noticed that morphologically, ssp. kooichi seems to be more similar to Taiwan's ssp. chungianus. This made me curious about the phylogenetic relationship of this species.

Ming

So it appears from that tree that chungianus is a distinct, undescribed species. Though granted, if I understand Adam they only analyzed one specimen.

Chuck wrote: Tue Dec 10, 2024 1:43 am So it appears from that tree that chungianus is a distinct, undescribed species. Though granted, if I understand Adam they only analyzed one specimen.

Thank you Chuck. That's why I brought up this question. Haha..
Considering the extremely limited distribution of the "subspecies" in Taiwan, its ecological divergence in host plants compared to the nominate subspecies, and its morphological differences from the nearest nominate subspecies, the phylogenetic relationship of this taxon is intriguing.

Ｍing

To demonstrate the ambiguity in COI: a soon-to-be-published paper has both COI 3’ and 5’ tree diagrams for P rutulus and eurymedon. One of the trees highly suggests that one is a ssp of the other; the other tree slightly suggests it might be the other way around.

Further, only some specimens of said ssp fall under the other species; but other specimens group oddly, kinda all over the place, including one level up. I believe these inferences to not reflect the real situation, however it has happened for other taxa that “one species” has been split, some elevated, etc.

The lesson is that COI trees for 3’ and 5’ don’t always agree, and any tree built with few, or ONE, specimen are highly suspect to the point of being useless at species level and below. That’s where SNP and other analyses shed more light.

To be clear COI is useful, and more so with more specimens. If a series of specimens all group together you’re probably on to something. But when they don’t differentiate well with COI- as is the case with the group in the original question- it means other tools are required, tools which are more costly and time consuming.

I have been away for a few days, but I asked Shao-Ji Hu about the tree, and he replied:

The topology is due to missing taxa in the eurous-mullah group. When we fill in with more data, the topology will change. The tree in any paper of Mitochondrial DNA Part B is required to validate the placement of the sequenced specimen ONLY, not intended for any phylogenetic inference.

Adam.

Chuck wrote: Tue Dec 10, 2024 3:09 pm To be clear COI is useful, and more so with more specimens. If a series of specimens all group together you’re probably on to something. But when they don’t differentiate well with COI- as is the case with the group in the original question- it means other tools are required, tools which are more costly and time consuming.

Absolutely agree, and this is the same problem we have found with some other Papilionidae such as P. machaon and G. sarpedon.

Adam.

There is yet another explanation: provisional misidentification of specimens.

The ID you see for a specimen in a tree must be assigned by a human. If that ID is wholly wrong it will show up there anyway, based on MtDNA. The tree is automated based on COI, not names. If someone IDs a Graphium mullah as T Rex, then it will look like T Rex groups with those butterflies.

For example, in the Tiger Swallowtail trees you will see soon there is an outlier- my nemesis. Amongst all the glaucus is one Papilio alexiares ssp. garcia. This is simply because some moron misidentified a glaucus as the Mexican Papilio alexiares ssp. garcia, and it was uploaded to BOLD with that ID. And, since most studies that use BOLD COI also use that specimens ID, it shows up in the tree that way.

When there are two or more easily misidentified taxa, and the tree is generated, it will show two or more distinct groups, with the named specimens all jumbled up in both or all groups. It just means that many were misidentified.

When I assign an identifier to a specimen it includes location and date. Not name. In a tree you can see what groups together and AH HA moments can arise if the lab test ID has this data. As in, all the AZ P rutulus group together, aside from the main rutulus. When the specimenID is “Jim vacation” it’s useless, you have to research every damned specimen. When a provisional ID is just wrong it will be misleading in the tree. A bit of common sense goes far.

Chuck wrote: Thu Dec 12, 2024 12:17 am There is yet another explanation: provisional misidentification of specimens.

I agree this can happen, and I know of instances in published papers where specimens have been misidentified.

However, in this case there is absolutely no way any specimens were misidentified. My colleague, Dr. Hu, is an expert on Pazala and there are only two species in Taiwan, easy to separate.

Adam.

Chuck wrote: Thu Dec 12, 2024 12:17 am There is yet another explanation: provisional misidentification of specimens.

It happens more than often indeed. Having sampled hundreds of Delias legs myself and put them in little vials... I can tell you that if someone talks to you or phones you in the middle of the process... you don't know where you are and you hesitate to start all over again.

In a study I took part in, the source specimens from the Museums were misidentified... by the Museums. As a result, conclusions of synonymy or differentiation were erroneous. If the scientists put a label on or took photos of the sampled specimens, we can confirm the initial error. Otherwise, it's impossible to know... and even then, it has to be verified.

This is why any DNA analysis must be carried out on SEVERAL specimens per taxon. But this is obviously not possible for rare species in collections. If a DNA study concludes something really strange... the first reflex must be: “can we see photos of the sampled specimens”.