Q&A: Yannick Wurm wrangles RADseq to learn why some fire ants bow to more than one queen

DSC_9777.jpg

Fire ants, Solenopsis invicta. Photo by Flickr user HankPlank.

Yannick Wurm grew up in Redwood City, California, and his initial plan was to design interfaces for Apple. But he went to university at the Institut National des Sciences Appliquées in Lyon—where, after two years of general engineering courses, the new Bioinformatics and Modeling department looked like “fun.” Looking for opportunities in research, he found an opening with a “red fire ant genomics” project on Google … and eight years later, he’s still studying exactly that.

Yannick got in touch to let us know that a cool new result in that project had just been published in Nature, and he was kind enough to answer some questions about the discovery.—Jeremy

Wang, J., Wurm, Y., Nipitwattanaphon, M., Riba-grognuz, O., Huang, Y., Shoemaker, D., et al. 2013. A Y-like social chromosome causes alternative colony organiziation in fire ants. Nature, doi: 10.1038/nature11832.

Okay, so let’s start off with the key natural history: some fire ant colonies have more than one queen? How does that work? What features of worker behavior or other phenotypes are associated with supporting more than one queen?

First, you’ve got to remember that there are more than 20,000 species of ant, and their lifestyles are very diverse. For example in some species each colony contains only a few dozen individuals living in a dead twig or an acorn, but Oecophylla weaver ants make large networks of nests up in the trees by sticking leaves together, and Eciton army ant colonies of hundreds of thousands of workers lack permanent home base and instead are regularly on the move.

But there can also be variation within species. In particular this has been extensively studied in the red Solenopsis invicta fire ant: some colonies have up to hundreds of wingless queens, but other colonies contain strictly one single wingless queen. And this is stable: any additional queen you try to add to a single-queen colony is executed by the workers.

It turns out that several additional traits vary alongside queen number: when there are multiple queens, each queen is less fertile than if you had a single queen there is a broader worker size distribution in single-queen than multiple-queen colonies, single-queen colonies are more territorial than multiple-queen colonies, and the dispersal strategies differ: a young queen destined to form a single queen colony will disperse far and wide and is able to independently found a new colony. Instead a young queen destined to a multiple queen colony will join an existing nest… only once that colony becomes too large, part of the colony (workers carrying queens and brood) may leave on foot to form a new nest nearby (this process is called “budding”).

What did you already know about the genetics underlying this “social polymorphism”?

Ken Ross at the University of Athens, Georgia, USA was the first to look at the genetics of this species in the 1990s. Using allozyme markers—proteins for which the two alleles are distinguishable after starch gel electrophoresis—Ken realized that something interesting was going on. More than 15 years of work involving Ken and his collaborators including Laurent Keller, Dewayne Shoemaker, Mike Goodisman, Michael Krieger and Dietrich Gotzek led to quite deep understanding of this amazing system.

In short, one of the fourteen allozyme markers, Gp-9, is associated to social form: colonies containing only homozygous Gp-9BB workers accept only a single Gp-9BB queen, whereas colonies containing both Gp-9BB and Gp-9Bb workers will invariably accept several queens, but only Gp-9Bb queens. This is probably the only example of such a fundamental aspect of social organization being under the control of a Mendelian element. Furthermore, because the presence of workers carrying Gp-9b leads to the execution of queens that lack Gp-9b, Gp-9 has been described as being a green beard gene.

But until our study, the extent to which other genes may be linked to Gp-9 was unclear.

And so it turns out that these Mendelian variants underlying the social polymorphism are actually quite a large portion of one fire ant chromosome. What methods did you use to determine this?

Yes, we found that during meiosis in Gp-9BB queens recombination occurs normally across all sixteen fire ant chromosomes, but in Gp-9Bb queens it is completely absent over more than 13 megabases surrounding Gp-9. This large region contains more than 600 genes, at least some of which must affect social behavior.

The next gen sequencing tools made all this possible. Our work was in fact easier than in many other non-model organisms because male ants (like other hymenopterans) are haploid: an unfertilized egg develops into a male. Having only a single version of each geniomic region makes genotyping and sequence analysis much easier than if you’re working with a diploid organism. We performed two approaches in parallel. To be able to perform sequence comparisons we sequenced, assembled and annotated de novo the genome of a Gp-9B male and independently of a Gp-9b male. In parallel, we performed RADseq assays on families of related individuals. For example we isolated a Gp-9Bb queen, collected up to 100 of her haploid male offspring, determined which were Gp-9B and which were Gp-9b and performed RADseq on all of them. We did this for several such families of both Gp-9BB and Gp-9Bb offspring. And then spent many many many months analyzing the data. The data analysis aspect has clearly become the biggest bottleneck.

queenPupaWorkers096-cropped

Two fire ant workers attend a much larger queen. Photo courtesy Yannick Wurm and Romain Libbrecht.

How well did RADseq work for this analysis? If you started over from scratch, would you use a different method?

There were indeed some initial hiccups with getting the Illumina sequencer to produce useable data from the RAD libraries. And processing the data wasn’t trivial because RAD analysis methods were still in their infancy when we began this. But overall we were very happy with the data and results. The most frustrating aspect in dealing with the data was that the coverage was so heterogeneous: some individuals gave almost no data, and some RAD sites had deep coverage for only a portion of sites. So you had to almost randomly choose which individuals and sites to throw out and which to keep.

We actually had some other sources of data that also show recombination is absent across this large region: RNA and genomic sequence information from pools of Gp-9B brothers and from pools of Gp-9b brothers obtained from single Gp-9Bb queens. Within each pool normally recombining regions show both possible maternal alleles, while only one of the two maternal alleles is found in the non-recombining part of the social chromosome.

Starting from scratch today, I would go for one of the many RADseq variants that reduce the number of markers. Or—if the budget allows—would simply sequence the whole genome of every individual.

So how do you think these chromosomal variants stopped recombining?

There must have been normally segregating genes that affected how likely a colony was to accept multiple queens, and others that affected fitness in a social-form specific manner. You would expect selection for reduced recombination between alleles that are advantageous in a multiple-queen setting and alleles that make you more likely to accept multiple queens, right? Based on current sequence comparisons we know that around 400,000 years ago recombination ceased between the B and b variants of this chromosome. This must have been due to some kind of large structural rearrangement that made it impossible for the two variants to pair up correctly during meiosis. We can see two such rearrangements today—where segments of the b variant are in a different order than on the B variant of the chromosome.

What kind of natural selection do you think might be responsible for creating and/or maintaining the two social forms? Is there an advantage to one form over the other?

Yes that’s a very interesting question. The relative success of the two social forms actually depends on the habitat. Because young queens destined to establish single-queen colonies disperse far and are able to found new colonies independently, they are very good at invading new territories. But in saturated habitats—with intense competition between colonies—multiple-queen colonies are able to produce larger numbers of workers and thus can outcompete single-queen colonies. The two social forms are thus under balancing selection.

But could one of these social forms go to fixation?

I guess if all multiple-queen Solenopsis colonies died, the b variant of the social chromosome would be lost, thus you’d have fixation of single-queen-ness. But based on what we know today it would be impossible for multiple-queen-ness to go to fixation: Indeed, bb individuals are inviable thus B cannot be eliminated from multiple-queen colonies. And because this genetic variation is retained, young BB queens – destined to independently found their own colonies – can emerge from multiple-queen nests.

Given that social structure is a property of a whole ant colony, rather than individual ants per se, would you say that it’s appropriate to talk about the evolution of this social structure polymorphism in terms of selection on whole colonies? And does that mean we’re talking about group selection, or that, for fire ants (and maybe other eusocial insects), the “individual,” from the perspective of natural selection, is the whole colony?

Social structure is indeed a colony-level phenotype. But selection occurs at many different levels. For example, workers in many social insect species occasionally misbehave—selfishly laying their own eggs instead of contributing to the colony. Furthermore the fire ant is unique because the b variant of the social chromosome is in fact a selfish genetic element—a “green beard gene” whereby bearers of the b help b by executing queens that lack b.

Thus it would be incorrect to think in terms of a single level of selection. In fact, selection can act on any heritable entity—and this is formalized with the general kin selection or inclusive fitness framework. It is unfortunate that recent discussions of group selection—which is simply a specific case of kin selection—may have seeded some confusion.

Many thanks for answering/putting up with my questions. I have just one more: How many fire ant bites have you had?

All too many!! But actually it’s the sting that hurts: the worker bites you with her mandibles, and then brings over her stinger to inject venom.

RedditDiggMendeleyPocketShare

About Jeremy Yoder

Jeremy Yoder is a postdoctoral associate in the Department of Plant Biology at the University of Minnesota. He also blogs at Denim and Tweed and Nothing in Biology Makes Sense!, and tweets under the handle @jbyoder.
This entry was posted in bioinformatics, interview, next generation sequencing and tagged , , . Bookmark the permalink.