Back to: Projects

Distribution of Afrotropical Kite Swallowtails

Results

On this page we present the findings of our investigation. First we give our answer to one of our initial questions:

Then we describe the results of our cladistic analyses of the relationships and classification, discussing:

before answering more of our initial questions:

We then discuss:

  • what else our analyses show
  • and give a summary of our results in which we also indicate some possibilities for future research.

How many species are there?

We have identified 39 species of Afrotropical Graphium,one of which was new to science. There may be more species as yet undiscovered both in the wild, or overlooked in public or private collections. We list the species and subspecies we know in the annotated checklist.

The 'new' species is from the Central African Republic. We named it Graphium abri in honour of the African Butterfly Research Institute, Nairobi, Kenya, and its founder, Steve Collins, who lent us the specimens. Sadly, we have seen only two specimens, both male, and know of no others.

Graphium abri upperside Graphium abri underside
Graphium abri holotype -
upperside
Graphium abri holotype -
underside
© African Butterfly Research Institute, Nairobi, Kenya

We also found it necessary to revise the status of several taxa, raising some from subspecies to full species, downgrading others from species status to that of subspecies of other species, and transferring other subspecies from one species to another.

Relationships and classification

General considerations

Our data yielded two similar, but not identical, patterns depending on whether the two unordered multistate characters were included (see analysis).

When we excluded (or deactivated) the multistate characters we obtained 1,656 equally parsimonious trees. When all those branching points or nodes not supported in all the trees were collapsed, the 'strict consensus tree' shown below left was produced.

cladogram of Graphium with multistate characters inactive cladogram of Graphium with multistate characters active
L = 176; CI = 33; RI = 68
multistate characters inactive
L = 214; CI = 37; RI = 66
multistate characters active
Taxa shown in green are swordtailed species; those in red represent oriental members of the genus Graphium.
Clicking on either diagram will open it in a new window. Readers may want to keep these open when reading what follows.

With the multistate characters included (active), just 756 equally parsimonious trees were generated. The strict consensus tree is shown above right.

As can be seen, a considerable amount of structure was retained in each case. There is also a significant amount in common between the two.
In both cases, the basal part is of the tree is pectinate with branches for Iphiclides, Lamproptera and two branches for Graphium (Pathysa). Note that the position of Protographium was fixed in advance as the most basal clade in the analysis.
At the other end of the tree, both analyses give a 'crown group' including most of the non-swordtailed species (apart from the angolanus group), together with Graphium (Paranticopsis).

There are also major differences.
With the multistate characters inactive, the angolanus group appears — in polytomy with a non-swordtailed G. (Graphium) — as sister to the other non-swordtailed species (+ Paranticopsis). With the multistates active, it is part of a much larger and more mixed polytomy.
Both diagrams show a close relationship between G. antheus, G. evombar and Graphium (Pathysa). With the multistates inactive, this trio forms part of a polytomy with G. kirbyi and G. junodi basal to the remaining Afrotropical (and some oriental) species. With the multistates active, the trio is further associated with G. kirbyi and G. policenes and this group is higher up the tree.
The tree with the multistate characters inactive recovers more structure higher up the tree, and less basally. By contrast, the tree with the multistates active gives more structure to the basal part of the tree, but is less structured towards the crown.

Are the Afrotropical Graphium a natural group?

We have not been able to confirm that the Afrotropical Graphium form a natural, monophyletic group excluding all other species of the genus.

In all our analyses, members of the oriental subgenera G. (Graphium), G. (Pazala), and G. (Paranticopsis) all emerged among the Afrotropical species. G. (Pazala) was consistently placed as sister to a pairing of G. antheus and G. evombar. G. (Paranticopsis) was always placed among a terminal group including many mimetic species.
Of the oriental subgenera, only G. (Pathysa) fell outside the Afrotropical species, and the two examples of the subgenus we examined did not hold together as a clade, but appeared paraphyletic.
G. (Paranticopsis) and G. (Pathysa) were formerly 'lumped' in a single subgenus. Our findings at least support the separation of these two subgenera.

The monophyly of the genus as a whole was well supported.

Are the swordtailed species a natural group?

Again, our analyses do not support a monophyletic clade for the swordtailed Afrotropical Graphium.

A glance at the two cladograms shows the swordtailed species to be paraphyletic, even if the oriental species are ignored.
In the left hand diagram, an unresolved grouping of G. antheus and G. evombar plus G. kirbyi and G. junodi are shown to be sister to another unresolved grouping of swordtails plus the remaining species.
In the right hand figure, there is more resolution, but in this G. junodi, G. polistratus and G. colonna are placed successively as sister species to the rest.
With the oriental species taken into account, the situation is even more complex.
Further work is needed to confirm or refute these putative relationships

Are the non-swordtailed species a natural group?

The answer to this question is more problematic.

If we ignore the oriental species, the left hand figure does show a clade of non-swordtailed species, with the G. angolanus group sister to the remainder. With the multistates active, the G. angolanus group forms part of an unresolved grouping including swordtailed species. The remaining non-swordtailed species do hold together.
Once again, however, a member of the oriental subgenus, G. (Paranticopsis) is also included. It may be that the single character, 'swordtail absent', caused G. (Paranticopsis) to be placed here.

What else do our cladograms show?

Our analyses have shown other features, corroborating some previously assumed groupings, but not supporting others:

  • G. antheus and G. evombar appear to be closely related sister species, which is not surprising considering the similarity of their genitalia and their dissimilarity from other species. More surprising is that G. (Pazala) consistently emerges as the apparent sister of these two.
  • Though we have made some changes within it, a clade including G. policenes and three other species is well supported.
  • G. tynderaeus and G. latreillianus hold together well.
  • The group of probable Amauris mimics, including G. adamastor, G. ucalegon, G. abri and 11 other species, is supported, but in our analyses G. philonoe is always included in the polytomy. Though at first sight this is surprising, there are sufficient similarities in their genitalia to suggest a relationship. We wonder whether G. philonoe might be the sister to the adamastor group. Extra characters, or refining the definitions of existing ones, could test this hypothesis.
  • An angolanus clade of six species including G. ridleyanus is well supported.
  • Most earlier accounts have included G. leonidas in a group with G. levassori and G. cyrnus. Despite similarities in their genitalia, our analyses did not recover this group. Instead, the three species emerged as part of a large polytomy with G. (Paranticopsis), the tynderaeus/latreillianus pair, and the adamastor group/philonoe 'crown group'.

Summary

Our findings are somewhat of a mixed bag. We do feel that we have contributed to the understanding of the group not only by corroborating some previous hypotheses, but also by calling into question some others and by refuting yet others. All in all, our reconstruction is much less neat than previous efforts.
There is clearly a need for further work.
Some of this we hope to do ourselves. We intend to extend our study to the oriental subgenera to unravel the relationships between them and the Afrotropical species.

That still leaves considerable scope for others. A particular project might be to investigate the internal and external relationships of the adamastor group. More material of these variable and reclusive butterflies would be needed.
Other sources of data would no doubt yield benefits. Due to lack of material we have been unable to make use of the early stages.
DNA analyses would surely be enormously useful. But that requires both the resources of a well-equipped laboratory and specimens collected recently enough to allow good extractions to be made of the appropriate genes.

Taxonomy, like any science, is an ongoing process. We regard our results as provisional. Although we have not managed to produce a final and definitive classification of the Afrotropical kite swallowtails, we believe that we have provided a useful platform for future work.