My last research paper: Part 1: Aims and methods

January 26, 2020 • 11:00 am

My last “research” paper—one in which I collected data with my own hands by pushing flies (i.e., manipulating Drosophila under the dissecting microscope)—has finally been published in Genetics, and I was pleased to find that it is a “highlighted paper.” Well, I thought it was pretty cool, too, and a good way to end my fly-pushing career. Even better, it appears to be unpaywalled, at least for the nonce. You can access the final published version by clicking on the screenshot below, or get the pdf here (the full reference at bottom). If for some reason you can’t get the pdf, a judicious inquiry will yield you one.

To describe the whole paper, including our aims, our methods, and the results (there are a lot of them!) would take a long and tedious post. I’ve thus decided to break the post up into two bits, one giving the aims and methods and the second the results. I hope this will make it easier to digest. Here is part 1.

I wrote briefly about this paper three days ago, but can be a bit more expansive here. It stemmed from a piece of work that almost every evolutionary geneticist I know has wanted to do, but that kind of study is dicey because it’s exploratory and, more important, takes a long time to carry out. With grants lasting about three years, there’s no way this research could have been conducted and finished within a single granting period, and that means that if you start it under a funded proposal, you won’t have anything much to show when it’s time to renew your grant. As I said before, I thus wrote the NIH proposal for this study as part of my very last grant, so I didn’t need to worry about having results within three years.

But I wanted to do the work very much, and when I talk about the results at various universities, I almost always meet a researcher who says, “Hey, you know, I thought of a study like that, too.”  It’s not the result of a great idea, but it’s the result of a common question: “What happens when you hybridize two species so that their genomes are completely mixed?” What you get in this case is a “hybrid swarm”, and we wanted to know what happened to it.  There are several possibilities:

a.) It could remain as a mixed mongrel swarm with genes from both species segregating.

b.) The hybrid swarm could remain somewhat mixed, and evolve into a new species that is reproductively isolated from both parent species. This is called a “diploid hybrid species”, and they do occur in nature, but they aren’t common. (However, species that result from hybridization and then the doubling of the hybrids’ chromosome number, called “polyploid species”, are fairly common, especially in plants.)

c.) The swarm could revert back to one of the parental species, losing through selection the genes from the other species. If you produce replicate hybrid swarms, you can see if, when this happens, it always reverts to the same parental species, or whether it could revert to either of the parental species in different replicates.

We expected that the most likely scenario would be c), but we weren’t sure. The reason we thought this is that in both of the species pairs we used, hybrid males are largely sterile, and so selection would purge foreign genes causing sterility. But that doesn’t mean that all genes could be purged. And even if reversion occurred, we didn’t know how fast it would be, or if it would differentially affect the visible traits that distinguish the species, the interspecific mating behavior of the species, or the DNA.

We might even have created a new hybrid species in the lab! We just didn’t know what would happen, but we knew something would happen, and that it would be interesting.

So we made hybrid swarms. In fact, we made them between two distinct pairs of “sister species” of Drosophila (each other’s closest relatives), and for each swarm we made eight replicates. This gave us 16 chances to see what happened. And for each of those 16 replicates (each one consisting of a bottle of hybrids), we followed morphology, behavior, and DNA sequence over 20 or more generations—about a year in the lab—to see what happened. It was a huge experiment, for we measured many traits in many replicates, assayed behavior and fertility of males, and sequenced entire genomes from each replicate: over 20 million nucleotides.

I’ll give the results in one word: reversion, and consistent reversion to one parent. But more about that next time.

Below are the two pairs of species, comprising two separate hybridizations (again, each hybrid swarm was constructed 8 times independently for each pair). And each hybrid swarm was made by crossing the species together reciprocally, producing two types of hybrid females, each containing half the DNA of each parental species and with half of them having the cytoplasm and mitochondria of one parental species (the mother), and half the females having those traits from the other. Now we had the hybrid females to begin the swarm, with an equal proportion of all DNA, cytoplasm, and organelles. But since hybrid males are sterile, we needed males to get the hybrid swarm going and reproducing; so we put with those females an equal number males from both parental species (the “backcross” males generated when these pure-species males cross to hybrid females are partly fertile). There is little mating discrimination in hybrid females, so they crossed to males of both species, produced semifertile male offspring and fertile female offspring, and then we let the swarm go for 20-24 generations. Of course fertility improved over the experiment as genes producing male sterility were weeded out by selection. Hybrid swarms began with a pure 50/50 mixture of DNA, chromosomes (including the Y), and organelles + cytoplams from each species.

The two species pairs we hybridized:

1.) Drosophila simulans/D. mauritiana. D. simulans is a worldwide human commensal, while D. mauritiana is found only on the Indian Ocean island of Mauritius, where it’s endemic. DNA sequencing puts their divergence between 250,000 and 500,000 years ago. While they look much alike, the males differ in a number of secondary sexual traits, implying fairly rapid sexual selection (females cannot be told apart by inspection). Here are some differences in male genital clasper shape, the number of bristles on the anal plate, the number of bristles on the genital clasper, and the number of “teeth” in the sex comb, a structure on the male foreleg that he uses to hold on to females during copulation. First, the pictures:

And here are the average differences between the species for several traits (standard errors aren’t shown):

Thus, as the experiment proceeded we could measure these traits, as well as others (width of forehead, wing area, and so on), to see if the trait values remained intermediate in the hybrids or if the population began reverting to the trait values characteristic of one species or the other.

Further, the species show substantial mating discrimination: although D. simulans females cross readily to D. mauritiana males, the reverse cross almost never happens: mauritiana females just reject the persistently courting simulans males. This gives us a way to see if the mating behavior of hybrids reverts to one species or the other. This can be judged by crossing males from the hybrid swarm at various times to D. mauritiana females. If they’re rejected, then they are becoming simulans-like in the male traits that females reject, but if they’re accepted, then those traits are reverting to mauritiana-like ones.

Further: when copulations do occur between simulans males and mauritiana females, they’re short and abnormal: the females don’t like the males climbing on their backs and try to kick them off, so a “mating” involves a male being dragged around the observation vial by his genitals. It looks painful! This kind of bizarre mating is abnormally short, and often results in no fertilization. simulans X simulans matings take about 30 minutes, mauritiana X mauritiana matings take about 13 minutes, and the abnormal hybrid mating lasts about eight minutes.

This gives us another way to measure mating traits: use a male from the hybrid swarm and put him with a mauritiana female. If the mating is abnormal and lasts about 8 minutes, he’s got mating traits that have reverted to those of D. simulans. If the mating is normal and lasts about 13 minutes, the males have reverted to D. mauritiana.

Finally, because male hybrids between the two pure species are sterile, we can see whether the males and females in the hybrid swarm have reverted to one species or the other for sterility genes. Just cross a female from the swarm to D. mauritiana males and, separately, to D. simulans males. If she produces fertile hybrids with the former and not the latter, she’s reverted to D. mauritiana. If the opposite result obtains, she’s reverted to D. simulans. You can, of course, do this for the males too, seeing if they have the sterility relations of one pure species or the other.

Finally, there is the DNA. These species differ in a number of sites on the DNA, and thus one can identify a bit of genome from one species or the other simply by sequencing it. Does it have a mauritiana-like sequence or a simulans-like sequence? To do this, we sequenced entire genomes from all hybrid swarms at generation 20, seeing if their genomes were still greatly mixed, or had moved toward one species or another. (We knew the complete DNA sequences of the two parental species.) This was the laborious part of the experiment given the need to first sequence the pure species, find the differences between them, and then compare the sequences of the hybrids.

Thus, we used a combination of morphological, behavioral, fertility, and DNA traits to track what happened to this hybrid swarm. We did the same thing for the other pair of species:

2.) Drosophila yakuba/D. santomea. I spent the last decade of my lab work on this pair of species, as D. santomea (discovered by my late colleague Daniel Lachaise) is endemic to the island of São Tomé, about 250 km due west of Gabon. But that island also harbors its sister species, D. yakuba, and the two diverged about a million years ago. D. yakuba is widespread in mainland Africa, found in savannas, grassland, and open forest. The formation of D. santomea probably did not occur on the island in the presence of its sister species, but rather after their common ancestor invaded the island a million years ago. After that ancestor evolved through transformation of the invader into D. santomea on the island, D. yakuba re-invaded, probably in the last 10,000 years or so.

The two species coexist on the island now, but yakuba lives at lower altitudes and santomea is restricted to altitudes above 1000 m. There is a hybrid zone between them where they meet, so we can also see what happens in nature when the two species hybridize.

In both cases, then, we produced replicates between a widespread “mainland” species (we call it “dominant” in the paper), and an island endemic.

Like simulans/mauritiana, yakuba/santomea also differ in morphological traits, though not as profoundly. The most striking difference is in pigmentation. Like the 7 other species in the D. melanogaster subgroup, D. yakuba has black pigmentation on both males and females, which is especially pronounced in males at the tip of the abdomen. D. santomea, on the other hand, is unique in almost completely lacking black pigment, as you see below in photos of two males:

This gives us a way to measure how the hybrid swarm is evolving, as we developed a way to quantify blackness on a scale from 0 to 120. (All flies were scored blind.)

Yakuba/santomea also differ in two morphological traits on average: number of hypandrial bristles (on the male genitals) and number of teeth in the sex combs, though that difference isn’t as pronounced as between simulans and mauritiana. (I discovered the sex-comb-tooth difference.) So we can measure these three traits over time to see how the yakuba/santomea hybrid swarm was evolving.

The sterility relationships between this pair are also similar to those of the other pair: hybrid males are sterile. So we can also take males and females of the hybrid swarm, cross them to members of the pure species, and see whether the swarm is adopting the fertility genes of one species or the other.

Copulation latency (the time from when a male meets a female to when they copulate) and copulation duration also differ among the species, giving us another behavioral trait to measure by crossing individuals from the hybrid swarm to individuals from the pure species.

Finally, the species of this pair, like the other, differ in DNA sequence in a species-specific way, so we could sequence the DNA of yakuba/santomea hybrid swarm members at generation 20 to see how it had evolved. Was it still a big mixture of the parental species’ genomes, or was it reverting to the sequence of one species or the other?

In the next post (I hope in a few days), I’ll describe the results, concentrating on the mauritiana/simulans swarm because the results are pretty much the same in both pairs of species, and there’s no need to give all the results when they’re in the paper.

I’ve tried to be brief here, but as you can see it still takes space to describe what we did. (And it took two hours to write this post!) Remember, you can read the paper, and inquire if you can’t get the pdf.


Matute, D. R., A. A. Comeault, E. Earley, A. Serrato-Capuchina, D. Peede, A. Monroy-Eklund, W. Huang, C. D. Jones, T. F. C. Mackay, and J. A. Coyne. 2020. Rapid and predictable evolution of admixed populations between two Drosophila species pairs. Genetics 214:211-230.

20 thoughts on “My last research paper: Part 1: Aims and methods

    1. Well, it did between Neandertals and “modern humans”, and the hybrids had half the genome of each group (I consider them subspecies). But due to either selection or backcrossing, the modern human genome is what largely remains. This hybridization could have been responsible in part for the demise of the Neandertals.

      1. From what I recall, though individual modern humans may have only 0% – 3% or so Neanderthal DNA 40% – 70% (of the Neanderthal genome is still extant scattered among modern human populations.

        Do the results from your study give any insight into the question of how extensive interbreeding between modern humans and Neanderthal was? What I mean is given what we know about Neanderthal genes among current human populations and trying to work backwards from that to try and determine the degree of “inter-mixing” between the two groups, it seems like the results of your study might give insight into how to reconstruct that history.

        Perhaps I am speculating way beyond what’s merited, but your results seem as if they lend support to the hypothesis that Neanderthal could have been absorbed by modern humans rather than being killed off.

  1. Very interesting! I will follow this as closely as time allows.
    I did not know that male fertility would return over time as the genes causing male sterility were removed. This seems a valuable resource for further study since those genes could be the ones that lie behind Haldane’s rule, which as I understand it has been something of a mystery.

    1. We understand Haldane’s rule pretty well know; it’s due to both the dominance of genes affecting sterility (more expressed in hybrids of the heterogametic sex), and to the faster evolution of genes on the X chromosome. I can refer you to review papers that summarize our knowledge.

  2. If I understand correctly, the two different ways you’ve described for measuring mating traits both involve the same procedure: put a hybrid male with a mauritiana female and see what happens. If she rejects him prior to copulation, we infer that his courtship behavior has reverted to simulans. If she rejects him after copulation has begun, we infer that his copulatatory behavior and/or genital morphology have reverted to simulans.

    Or is there some methodological difference between the two cases that I’ve failed to grasp?

    1. Well, there are two measures here: whether copulation occurs and, if it does, how long it lasts. Those result in two different statistics: percentage mating and duration of copulation. They may well involve different traits, including pheromonal differences for the former (and wing-waving that stimulates mating) and, I suspect, genital morphology in the latter. We’re not claiming that everything we measure is independent (copulation duration may be influenced by genital morphology), all we argue is that for some of these traits there are different ways to measure “reversion.”

      I’m not sure I answered your question.

    1. I know; that’s what my colleagues keep telling me. I’m pretty sure I won’t work with flies with my own hands any more, as I’m giving up my lab, but I’m not at all sure I won’t publish scientific papers that need to be refereed.

  3. Fascinating. Flies reproduction takes about 7 days, right? So, about 52 cycles per year. That’s a lot of work for the flies and the experimenters. I can imagine back muscles can get very sore. Congrats on the project.

  4. Sex Comb?

    Sounds like something used by guys in the 80’s hair metal bands.

    Sorry, I’ve nothing intelligent to add. This is a bit over my head and makes me wish I could retroactively give my high school self a dope-slap for not taking all the biology courses that were on offer. I can remember the racks of flies in the corner of the room and I would stare at them with curiosity, but I was already suffering from depression, anxiety, and imposter syndrome and thought I was too stupid to take those higher level electives.

    So, if there are any high school-aged readers here, DON’T LIMIT YOURSELF! TAKE THE RISK! The worst thing that could happen is not that you’ll do poorly and be embarrassed, but that you’ll end up a miserable middle-aged putz like me with nothing but the cold, dead dreams of woulda-shoulda-coulda to keep you company.

  5. Lovely — will wait for Results write-up, but interesting if there’s a pattern for speciation in such repeated insular invasions [literal here, but generalizable for other types of geographic separation].

  6. Fascinating indeed. Sounds like a poor feature of the whole grant-awarding system that you weren’t able to do it until now, since as you say it does sound like just the sort of thing that anybody working in this area would want to do.

Leave a Reply