My last research paper. Part 3: Significance

In the past week I’ve written two posts about what I think will be the last “research paper” I write, i.e., a paper in which I actually did work at the bench (pushing flies). I’ve covered the paper at some length because I think the experiment is cool, because the results were conclusive, and because it’s an experiment that many evolutionary geneticists have wanted to do in their careers, but couldn’t because it would take too long to get results within a single grant-funding cycle. It’s the kind of experiment that you do out of pure curiosity: to find out what happens.

Here’s the paper again; clicking the screenshot takes you to it (if it doesn’t, ask for a pdf):

Part 1 was a description of the study: its aims and methods, which corresponds roughly to the “Introduction” and “Methods and Materials” section of the published paper (and of most published science papers). In that post I described how we made “hybrid swarms” of two pairs of sister species, each pair comprising one widespread Drosophila species living on a continent and an endemic sister species restricted to an oceanic island (D. simulans/D. mauritiana and D. yakuba/D. santomea). Each swarm began with 50% of the DNA, organelles, and cytoplasm from the two parental species, and we made 8 replicate swarms for each pair.

The object was to simply find out what happened when we “mogrelized” two species into a gemisch and then let the population go, without any experimenter-imposed selection over 20 generations. Would the mixture evolve back to one of the parental species? If so, which one, and was that reversion repeated in all eight replicates? Or would we simply get a mixture that persisted over time, without much happening? Alternatively, could we even get a new species—a “diploid hybrid species” that was reproductively incompatible with its two parental species? These questions bear on the compatibility of two species’ genes in a single mixture: are the genes of a species “coadapted“, that is, do they work well together but can’t tolerate the presence of genes from another species? Or is there broad tolerance for genes from a close relative?

In Part 2, posted yesterday, I described what happened (the “results” section of the paper). In short, the results were conclusive and replicable: all the swarms in both replicates reverted back to a parental species—in both cases the “dominant” mainland species (D. simulans in one case and D. yakuba in the other).  This reversion was seen in several morphological traits that distinguish males from each of the parental species, in the mating behavior of the flies, in their reproductive relationships (fertility) when crossed to the two parental species, and in the sequences of the DNA itself. Since all the trait and behavioral differences are surely based on differences in DNA, all the results are mutually supportive.  Nevertheless, foreign DNA was not completely eliminated from the swarms after 20 generations, but that’s only a year in the lab and it might have been further weeded out by selection over longer stretches of time.

So what does this all mean? I’ll use subheadings here.

Selection caused each of the hybrid swarms to revert to a constitution pretty much that of one parental species. In no case did we get anything like a new hybrid species: in every case the swarm reverted to looking and acting like the mainland-species parent, and its DNA sequence was on average about 93-94% that of the dominant species, with the rest of the genome showing some DNA from the minor species.

This means that some form of selection eliminated the island-species’ genes from the mix over time. We don’t know what kind of selection that is (see below), but surely a lot of it had to do with incompatibilities between the species that lead to hybrid sterility and mate discrimination. These will be eliminated no matter what environment you test the flies in, as such selection is largely independent of the environment. It just involves getting rid of genes that produce malfunctioning hybrids. And the elimination of those genes will also eliminate any DNA linked to those genes, whether it be neutral or also bad. This is the “hitchhiking effect” caused by the fact that genes sit next to each other on chromosomes, and if you eliminate one section quickly it will eliminate the adjacent genes from the population as well. What surprised me is how quick the reversion was, and how repeatable it was: not just that all the replicates reverted to the same parent, but because the sections of DNA that “allowed” foreign genes tended to be the same among replicates.

This leads to the next question:

Why did the populations always revert to the mainland species?  There are several possibilities here, and the short answer is that we just don’t know. Here are all the alternatives (several of these could, of course, operate together).

a. The island species, comprising smaller populations, might be more inbred than the mainland species. If that’s the case, the island species might have had high frequencies of deleterious genes, as such genes tend to reach higher frequencies in smaller populations. (This is why small, inbred groups of humans, like the Old Order Amish and the Dunkers, show a high frequency of genetic disorders.) In such a case the “healthier” genes from the less inbred mainland species would replace these deleterious genes. I consider this unlikely because the island species are still present, at least today, in appreciable numbers, which would make them less prone to “inbreeding depression”. Further, that depression would have to have affected nearly every area of the island species’ genomes.

b. Mainland species are more ecologically generalized than island species, as the latter live in a restricted environment while the former roam over many diverse habitats. (D. santomea, for example, is restricted to the high-altitude mist/rain forest of Saõ Tomé while D. yakuba can live in open forests, grasslands, and savannas). This means that mainland species may have genomes that comprise “jack-of-all trades” genes, and thus would be more likely to replace the island species’ genes in a foreign laboratory environment. Several ecologists have proposed that island endemics are often more narrowly adapted than are their mainland relatives.

c. The mainland species was more fit in the particular environmental conditions we used (standard cornmeal/agar/yeast medium, rearing at 24°C, 12 hour light-dark cycles, and high humidity), but in other conditions the island species’ genes could have been more fit. For example, D. santomea prefers cooler temperatures than D. yakuba, and perhaps if we left the swarms at, say 18° C, the swarm would have reverted to D. santomea. This remains to be studied.

d. The genomes of the island species may simply contain more genes that cause hybrid incompatibilities than do the mainland species. This would lead to the more rapid elimination of “island” genes, and thus reversion to the mainland species. There is a bit of evidence for this in one of these pairs, but not for the other.

Why were there parallel regions of the genome that retained “foreign” genes from the island species? As I mentioned last time, some regions of the genome, like the middle of the right arm of the second chromosome in the D. santomea/D. yakuba swarm, and the tip of the left arm of the third chromosome in the D. mauritiana/D. simulans swarm, more readily retained genes from the island species, though not at high frequencies. We don’t know why this is, but it indicates a parallelism in the selective forces acting on the different replicates. The retained regions could have contained “neutral” genes from the island species, or island-species genes might even have been subject to positive selection in these locations, driving them towards high frequencies. Since the island-species genes didn’t really attain high frequencies (we saw no cases of regions that had two copies of island-species genes), I suspect these are just regions that don’t have a lot of genes that affect the fitness of the hybrids.

Further, the X chromosome had the lowest retention of foreign-species genes. This is in line with our previous observations that X chromosomes have more genes causing hybrid problems, for because they are present in only one copy in males, and can also express both dominant and recessive genes that cause hybrid problems.

Are there parallels in the amount and nature of introgression (gene admixture) between what we saw in the lab and what occurs in nature? We can answer this question in only one pair: D. yakuba/D. santomea. That’s because these species co-occur in a narrow “hybrid zone” at about 1000 m high on São Tomé, and hybrids are formed naturally. We can thus see how much foreign genome goes from one species to the other in nature.

The answer to that is: very little. There are only a few regions of each species that contain small amounts of genes from the other species, suggesting that even in nature foreign genes are not tolerated well. But the regions of the genome where we see some gene “pollution” in nature aren’t the same as the ones that show it in our lab experiment. (In nature, low amounts of gene exchange have occurred pretty evenly across the genome, while we find them concentrated in particular areas in the lab.) This could reflect either the fact that selection wasn’t finished in our swarm, or that the nature of selection in the wild differs from that in the lab, so that different foreign genes would be tolerated.

A few caveats. Our study does not show that the formation of a new species after hybridization is impossible: after all, we used only four species of Drosophila (hybrid species are unknown in that genus), and new hybrid species are known in other groups, like butterflies and sunflowers, though for diploid species they aren’t common. (Some people say they are common, but the evidence for that is very weak.)

Further, we used only two strains in each hybridization: one from each species. Though the strains weren’t inbred, it’s possible that if we started with other genetic material from these species, we’d get different results. We were constrained to use only a few strains because we had to get their DNA sequences to be able to determine which DNA in the swarms came from which species. That would have been much harder if we used more genetically heterogenous starting material. We also had to eliminate chromosome inversions in these strains whose presence would have impeded gene exchange, so were constrained to use these “homosequential” strains.

And, as I mentioned above, even with these strains the results may have differed had we reared them under different lab conditions—say using different food or different temperature. Such studies should be done, but I doubt they will be given the difficulty of doing our own research using four species, one strain of each, and under a single laboratory condition.

But the consistency of the results in our study shows that, at least under our experimental conditions, not much foreign DNA is tolerated in even closely related species.

Here are the nine species in the D. melanogaster subgroup (a male from each). D. santomea (e) is in the middle, notable for its derived lack of pigmentation. And so endeth this series of posts.