[HELICONIUS] Heliconius heurippa -- a hybrid species

[Jim Mallet]

Dear All,

A major paper has come out about hybrid speciation, based on Heliconius, 
and I thought you might be interested.

Mavárez, J., Salazar, C., Bermingham, E., Salcedo, C., Jiggins, CD., 
Linares, M.  Speciation by hybridization in Heliconius butterflies Nature. 
441: 868-871
Available at:
http://heliconius.cap.ed.ac.uk/butterfly/publications/Mavarez_Nature2006.pdf

Needless to say, a few critics have argued to the media that Heliconius 
heurippa is not a hybrid species at all, but they haven't contacted anyone 
who might know about Heliconius to test their doubts.  By a quirk of fate, 
I was able to hear some of these criticisms which are normally kept 
relatively secret among scientific colleagues.  You can read about some of 
this disagreement from the National Public Radio website:

http://www.npr.org/templates/story/story.php?storyId=5485880

I am a reasonably independent critic, not having been an author on the 
paper, and this is what I think:

First, the background.
Emsley had included heurippa, with its red and yellow bands, within 
melpomene (1964), presuming it to be a hybrid like so many other hybrids 
between Amazonian yellow forewing band species and extra-Amazonian 
red-banded species.  Mauricio Linares has been working for many years on 
this problem.  His 1989 PhD thesis specifically advocates that heurippa is 
a hybrid between Heliconius cydno and H. melpomene.  Larry Gilbert, his 
(and also my own) PhD supervisor, also believed this, and it certainly 
seemed probable to me, given that I knew of many natural hybrids between 
the two putative parent species.

I believe the evidence for heurippa being a hybrid is adequately provided 
by the Nature paper, for the following reasons.

1) Heliconius heurippa is clearly closely related to H. melpomene and H. 
cydno, on molecular grounds and morphology.  On average, and it is true for 
mtDNA, the form seems closer in molecular genetic terms to cydno.  It 
overlaps with Heliconius melpomene, but the nearest population of cydno 
(the subspecies Heliconius cydno cordula) is a few hundred km to the 
N.  Thus, cydno is genetically distinct from melpomene, with which it 
coexists, but not as different, on average from Heliconius cydno.

2) Helionius heurippa is a colour pattern intermediate between the two 
species.  From (1), if it is not a separate species, it must be cydno.  But 
it lacks the brown hindwing underside pincer-shaped markings of that 
species.  In fact, it would be the only subspecies that lacks these 
markings.  It also has the red basal underside spots that are typical in H. 
melpomene.  It has the red forweing band of melpomene, which is pushed to 
one side, in the mode of the hybrids in Heliconius melpomene group, by a 
yellow forewing band.  The yellow band shape is typical of the adjacent 
cydno race Heliconius cydno cordula, which occurs in the NE Colombian Andes 
into Venezuela.  We know the inheritance of these colour patterns due to 
crosses performed in Mauricio Linares' lab, by Russ Naisbit et al, and by a 
variety of people including Philip Sheppard, John Turner, and others before 
them.  I think the most convincing thing about this colour pattern evidence 
is that it is the two local races Heliconius melpomene melpomene and 
Heliconius cydno cordula, rather than any other races, that could have 
produced this colour pattern combination.

3) Microsatellites show that heurippa is more divergent from melpomene and 
cydno than any other race of either species.  Thus, when you use the 
Bayesian programme STRUCTURE, three groups are identified, cydno, 
melpomene, and heurippa.

Unfortunately the microsatellite data are not provided in the Nature 
paper.  We are left in the dark as to whether heurippa differs from cydno 
and melpomene in having intermediate gene frequencies, or whether cydno is 
closer, or melpomene is closer to heurippa.  Even worse, it could be that 
cydno and melpomene are closer to each other than heurippa.  Intermediate 
microsatellite frequencies would provide good evidence of hybrid origin for 
heurippa, but because microsatellites evolve fast, it could be that 
heurippa diverged more strongly, so finding that heurippa is the most 
divergent wouldn't necessarily reject the hybrid argument.

4) There are "pure" divergent haplotype groups in two nuclear loci 
(invected and Distal-less) in H. melpomene and H. cydno, but mixtures of 
the two groups of haplotypes in the putative hybrid, heurippa.  It is 
already known that H. heurippa's mtDNA is closer to cydno than to melpomene.

5) Mating behaviour.  In Heliconius, the males usually approach females 
visally, and mate after a short courtship interaction.  Butterflies are 
highly visual creatures, and it is perhaps not surprising that 
intermediate, heurippa-like patterns in hybrids between cydno and melpomene 
are preferred by heurippa males.  This (together with previously published 
evidence) suggests that reproductive isolation between Heliconius is partly 
generated by colour pattern choice.  Probably, once the new heurippa 
pattern is fixed, the behaviour of males must coevolve with the new colour 
pattern environment to maintain maximum sexual success in males.  In any 
case, heurippa is partially - strong isolated from both the parent species 
by mate choice differences, which depend strongly on colour pattern.

6) Hybridization is regular in the wild between the two parent species 
wherever they co-occur.

7) Linares and Mavarez cite a population in San Cristobal Venezuela where 
H. cydno cordula and H. melpomene melpomene coexist, a few hundred km N of 
the heurippa site, and where colour pattern intermediates are common.  In 
this unusual population, containing the same subspecies of the putative 
parent species, even the mtDNA is flowing between the species (Mavarez 
pers. comm.).  It is simple to imagine that such a polymorphic situation 
might end up with fixation and male choice adaptation to the new colour 
pattern.

Now the critiques of the Mavárez et al. paper, and some comments.
1) Jerry Coyne at Science Now 
http://sciencenow.sciencemag.org/cgi/content/full/2006/614/2

"...others remain skeptical. "Maybe this is a hybrid species, but I'm not 
convinced from the genetic data that it is," says evolutionary geneticist 
Jerry Coyne of the University of Chicago in Illinois, who says much more 
sequencing would need to be done to prove H. heurippa is a hybrid. Coyne 
believes hybridization can't be common in nature because current animal 
family trees would reveal such mixes."
We don't know what his criticism was, but we get the flavour that he 
doesn't believe this sort of thing is likely, and therefore tends to ask 
for stronger proof.  He doesn't understand some of the strong 
Heliconius-specific evidence, perhaps, and maybe the Nature paper doesn't 
try to head off criticisms enough.
2) Anonymous commentator to Current Biology (this and the following are 
paraphrased to avoid any unnecessary aggression).

No evidence for hybrid origin.  Author's cite only two genes supporting 
this [i.e. invected and Distal-less].
Apart from the approx 20 microsatellites, it is true that only two 
sequenced genes were analysed.  However, as they were sampled from from the 
local populations (this was not mentioned in the paper), the hybridization 
hypothesis seem rather convincing in view of the haplotpe mixing.
Color pattern similarities might well be due to evolutionary convergence.
This is certainly sometimes the case.  However, it is an extraordinary 
coincidence that this parallel evolution has led to the duplication of 
patterns expected from transfer between LOCAL populations, rather than 
unusual parallel evolution of genes that occur in populations not in contact.
"Evidence" cited in an earlier paper is based on asymmetrical 
incompatibilities, and is wrong.  Michael Turelli pointed this out and had 
the story KILLED on National Public Radio.
This refers to the paper: Salazar, C. A. et al. Hybrid incompatibility is 
consistent with a hybrid origin of Heliconius heurippa Hewitson from its 
close relatives, Heliconius cydno Doubleday and Heliconius melpomene 
Linnaeus. J. Evol. Biol. 18, 247–-256 (2005).  However, I still don't 
understand the criticism.
There are no phylogenies showing reticulation.
Species-level phylogenies usually show a consensus of binary branching 
trees, and therefore almost never show reticulation.  However, genealogies 
at the two above-mentioned loci do show reticulation.
The evidence that the butterflies show wing-color related assortative 
mating is not there. No statistics are given in the paper.
I find this mystifying, and believe that the critic has completely 
misunderstood the paper and the statistical analysis.
It is not clear how, according to their scenario, "drift and 
frequency-dependent selection" would increase the frequency of a NEW color 
pattern in a species that is ENTIRELY SYMPATRIC with one of its supposed 
ancestral species.  Strong selection for mimicry should homogenize any 
variation.
It seems likely that the critic is one of the authors of a well known 
diatribe against the shifting balance, which involves genetic drift, and 
also believes a priori that sympatric specition is unlikely.  The critic is 
not interested in the evidence.  It seems to me that the non-mimetic 
hybridization seen between the parent species at San Cristobal, Venezuela, 
disproves this argument.
The main results are either wrong or buried somewhere else.."
The critic seems not to have carefully read the paper, or checked out the 
supplementary data for evidence against his unshakeable belief that the 
paper is totally misguided.
Well, I hope you find this edifying!  All the best, Jim

James Mallet
http://www.ucl.ac.uk/taxome/jim/
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.mailinglists.ucl.ac.uk/pipermail/heliconius/attachments/20060622/35594fcc/attachment.html