More Recent Comments

Thursday, December 18, 2008

Speciation in Monkeyflowers

 
Within a species there may be distinctive subspecies that have different allele frequencies. The differences are maintained because there is restricted gene (allele) flow between them. The two subspecies may look very different or they may be very similar in appearance.

Genetic exchange between the subspecies is often prevented because the subspecies are geographically separated. This is the first step on the path to allopatric speciation. But genetic exchange can also be restricted by other mechanisms, for example the timing of reproduction, that occurs even if the subspecies inhabit the same environment. This could lead to sympatric speciation.

In either case, the two subspecies will become distinct species—as defined by the biological species concept—when it becomes impossible to form hybrids due to genetic incompatibility. The study of actual speciation events is a hot topic in evolution these days. One of the goals is to identify the genes responsible for preventing the formation of fertile hybrids. The other goal is to identify the mechanism by which the alleles of these genes become fixed in the subspecies. Is it by natural selection or random genetic drift? (Shuker et al. 2005)

One of the best studied examples of speciation in action is due to the work of H.D. Bradshaw and Douglas Schemske at the University of Washington in Seattle, Washington (USA) (Schemske is now at Michigan State University). They studied two species of monkeyflowers that grow near streams and rivers in the mountains and valleys of western North America.

Mimulus lewisii (top) is found primarily at higher elevations (1600 m to 3000 m) while Mimulus cardinalis (right) grows at lower elevations (sea level to 2000 m). Their ranges overlap at moderate elevations in the mountains of California but hybrids are exceedingly rare.

The species differ in a number of characteristics including leaf shape and stem height but the most obvious differences are in the flowers. Mimulus lewsii has pink flowers that are quite open. They attract bumblebees and in the wild 100% of pollinations within this subspecies are by bees. Mimulus cardinalis has red flowers with a more narrow shape. These flowers attract hummingbirds who are responsible for 98% of pollination events in M. cardinalis.

When crossed in a greenhouse, the two species produce fertile hybrids so technically they are not really species but subspecies.

Ramsey et al. (2003) have studied the barriers to gene flow in the wild. Much of it is due to ecogeographic isolation, which is a fancy way of saying that the species don't often come in contact. They grow at different elevations and each species has become adapted to that elevation so that M. lewisii, for example, does not survive well at low elevations and M. cardinalis can't take the cold and the shorter growing season at high elevations.

The fact that the two species have different pollinators is a major factor in preventing gene flow between them. Hummingbirds hardly ever visit M. lewisii and in the overlapping zones there were very few recorded instances of bees visiting flowers from both species. Thus, the opportunities for cross-pollination were effectively zero. What this means is that, "even in sympatry these species are isolated to a large degree by pollinators" (Ramsey et al. 2003).

There are other factors contributing to genetic isolation. The hybrid plants are somewhat less fit and cross-pollination results in fewer seeds than pollination within a (sub)species. The sum of all these factors means that, in the wild, the total reproductive isolation between the two species is 0.9974 to 0.9998. In other words, they don't mix! (But recall that they can readily form fertile hybrids when crossed in the greenhouse.)

In this example, a major component of the restricted gene flow is due to physical separation of the species and that separation is the result of adaptation to different environments. In that sense, the path to speciation is driven, in part, by natural selection. The species are not genetically incompatible so we're not dealing with mutations that prevent hybridization as would be the case if they were true biological species.

Attention has focused on flower color and shape since that determines whether an individual is pollinated by bumblebees or hummingbirds. It's another step toward preventing gene flow between the species. Is it due primarily to selection or drift?

Schemske and Bradshaw (1999) identified a locus, called yellow upper (YUP), that plays a large role in determining flower color in the two species. The locus affects carotenoid distribution in the petals. In M. cardinalis carotenoids are found throughout the petals and the flowers are red. Bees are not attracted to red flowers. The YUP allele in M. lewisii results in less carotenoid and the flowers are pink. These flowers attract bees.

A subsequent study by Bradshaw and Schemske (2003) established that the YUP alleles are directly responsible for much of the pollinator discrimination observed in monkeyflowers. In the second study the authors created near-isogenic lines (NIL) that differed only at the YUP locus.

The normal M. lewisii flower is pink and the petals are in an open shape (a). The normal M. cardinalis flower is red and the shape of the flower is quite different (c). The dominant YUP allele from M. lewisii prevents carotenoid deposition and when it is bred into M. cardialis the flowers are pink (d). The recessive yup allele from M. cardinalis causes more carotenoid to be deposited making the flowers orange in an M. lewisii background (b).

The plants were tested in a natural environment where the ranges of the two species overlapped and both bees and humingbirds were common. Bees preferred the pink flowers whether they were in an M. lewisii background or an M. cardialis background. Conversely, hummingbirds preferred the orange and red flowers in both backgrounds. Thus, the two species have adapted to different pollinators and a large part of this adaptation is due to flower color.

Here's where it gets tricky. Is the switch from bee pollination to hummingbird pollination driven by natural selection? In other words, when the mutation causing red flowers first arose did it confer a fitness advantage on the individuals that came to be pollinated by hummingbirds?

Here's how Bradshaw and Schemske (2003) address this question,
As ‘mutations’ at the YUP locus decrease visitation by the current pollinator guild, and simultaneously increase visitation by a new pollinator guild, are there plausible ecological circumstances in which the mutant might be favoured by natural selection? The combined rate of bumblebee and hummingbird visitation to the yellow-orange-flowered ‘mutants’ of M. lewisii is just 26% of that to the wild-type pink flowers, and the combined rate for dark-pinkflowered ‘mutants’ of M. cardinalis is 95% of the wild type. This implies that a change in the relative abundance of bumblebees and hummingbirds, compared with the pollinator assemblage present during our field experiments, would be required for the mutant to be favoured by natural selection in the common ancestor of M. lewisii and M. cardinalis. The change in relative abundance of pollinators necessary to produce equal visitation to both flower colour phenotypes can be estimated from our data. A ninefold decrease in the relative abundance of bumblebees would produce equal combined visitation rates in the wild-type pink-flowered and ‘mutant’ yellow-orange-flowered M. lewisii NILs. At the equilibrium point, 99% of visitors to wild-type M. lewisii flowers would be bumblebees, whereas 87% of visitors to ‘mutants’ would be hummingbirds. In the M. cardinalis NILs, a twofold increase in the relative abundance of bumblebees would produce equal visitation rates to pink and red flowers. At the equilibrium point, hummingbirds would be virtually the only visitor to the wild-type red M. cardinalis flowers, and remain the major visitor (89% of visits) even to the dark-pink ‘mutants.’
In order for the red flower allele to be fixed by natural selection there would have to be a significant decline in the bee population at the time the mutation arose. Presumably, this decline would have only occurred in a small part of the range leading to a subpopulation with red flowers while the main, wild-type, population (pink flowers) continued to be visited by bees.

The authors don't mention the other possibility; namely, that the red flower allele (yup) spread in a subpopulation by random genetic drift. In this scenario, there is no selective advantage to individual plants if they are pollinated by humingbirds. Clearly the evolution of pollinator discrimination by flower color will lead to restricted gene flow between the two species but it is not clear whether this epiphenomenon is due to selection for hummingbird pollination or random genetic drift.


[Photo Credits: Mimulus lewisii or Purple monkey-flower (top) is from flickr. Mimulus cardinalis or Cardinal monkeyflower (second from top) is from the Arizona-Sonore Desert Museum.


Bradshaw, H.D. Jr. and Schemske, D.W. (1999) Allele substitution at a flower colour locus produces a pollinator shift in monkeyflowers. Nature 426:176-178. [doi:10.1038/nature02106] [PDF]

Ramsey, J., Bradshaw, H.D. Jr., Schemske, D.W. (2003) Components of Reproductive Isolation between the Monkeyflowers Mimulus lewisii and M. cardinali (Phrymaceae). Evolution 57:1520-1534. [PDF]

Schemske, D.W. and Bradshaw, H.D. Jr. (2003) Pollinator preference and the evolution of floral traits in monkeyflowers (Mimulus). Proc. Natl. Acad. Sci. (USA) 96:11910-11915. PDF]

Shuker, D.M., Underwood, K., King, T.M., and Butlin, R.K. (2005) Patterns of male sterility in a grasshopper hybrid zone imply accumulation of hybrid incompatibilities without selection. Proc. Biol. Sci. 272:2491-2497. [DOI: 10.1098/rspb.2005.3242]

10 comments :

Anonymous said...

Happy Monkey!

Anonymous said...

I agree. Note that the authors assume that pollinator abundance has been a limiting factor. Without competition for pollinators, the trait is neutral and genetic drift is a plausible mechanism. Further specialisation to hummingbird pollination in M. cardinalis may then have evolved as a reinforcement of reproductive isolation, after secondary contact.

Anonymous said...

Are you suggesting that the “red” allele, which attracts humming birds, got fixed by neutral drift in the absence of a change in selection pressure (i.e., no change in relative abundance of bees and hummingbirds)? It seems to me that the red allele currently imposes a substantial fitness penalty compared to the wild type. If it’s even a fraction of the 74% decline in visits they measured, it’s not going to drift to fixation. That’s an allele that surely is destined for elimination!

Although you point out that fixation by selection would require a change in the bee-to-hummingbird ratio, you don’t point out that this would also be required for drift to work. The problem for both explanations is that, relative to the wild type pink species, the playing field is tilted against the red allele. For drift to work it needs to be nearly level, and remain that way for a long time. Selection would require a bigger change so it tilts towards red, but it wouldn’t have to persist very long.

I donno. Seems to me that this is not really a good candidate for proposing drift as an explanation. The trait does appear to be important to reproduction, seems to be sensitive to an environmental feature (bee/hummingbird ratio) that reasonably could vary, and your hypothesis has even fewer degrees of freedom than the selection one.

Larry Moran said...

Divalent asks,

Are you suggesting that the “red” allele, which attracts humming birds, got fixed by neutral drift in the absence of a change in selection pressure (i.e., no change in relative abundance of bees and hummingbirds)?

Let's try and keep our terminology straight so we don't get confused.

I'm suggesting that the "red" allele might be selectively neutral with respect to survival of the individuals that carry it relative to those who carry the "pink" allele.

It's that's true, then the allele may increase in a population due to random genetic drift. (It's best to avoid terms like "neutral drift.")

It seems to me that the red allele currently imposes a substantial fitness penalty compared to the wild type.

Why do you say that? If a bunch of "red" flowering plants arose (by combining recessive mutations) among the "pink" plants then why would their survival be at risk? There are plenty of hummingbirds around so the chances of them setting seed are no worse than the pink plants.

We then have a situation where both red and pink plants are competing for the same ground but neither has a fitness advantage. That's what "neutral" means.

The trait does appear to be important to reproduction, seems to be sensitive to an environmental feature (bee/hummingbird ratio) that reasonably could vary, and your hypothesis has even fewer degrees of freedom than the selection one.

That's not the point. The point is that the mere existence of these two different types of flowers is not prima facie evidence of natural selection. You need to postulate additional factors in order to make the case.

In this particular example, there would have had to have been a declining bee population in some part of the range. That change favored the red flowers but elsewhere in the range the pink flowers survived because there was no change in the bee population. Over a period of thousands of years the two subspecies diverged and one had red flowers while the other had pink flowers.

We don't know how long the absence of bees persisted but we do know that, today, there's no disadvantage in being pink.

I'm not saying that this selection scenario is impossible. I'm merely pointing out that you have to specify it if you want to make the case. It's not sufficient to just point to red and pink flowers and say, voila! - selection.

That's what Orr does in his Scientific American article.

I think you're as confused about this as he is.

Anonymous said...

You stated “I'm suggesting that the "red" allele might be selectively neutral with respect to survival of the individuals that carry it relative to those who carry the "pink" allele.”

But that is contradicted by the results you described above! I’ll concede I may be misunderstanding the situation here, but instead I think you are. Here’s how I viewed it:

The “red” allele on the pink subspecies imposes a fitness penalty, as shown by the result above (going from image ‘a’ to image ‘b’ leading to a 74% decline in pollinator visits). Naturally, this is based on *today’s* conditions of the bee/hummingbird ratio acting on the background of the particular secondary characteristics of the wild type pink flower, so there is uncertainty of what the situation was back then. Nonetheless, it does show that petal color has an effect on fitness, and that the effect of the red allele can be negative (and that, in the pink subspecies, it is negative today).

So the problem is, how would a red allele have any chance of getting fixed if it is currently negative on the wild type pink subspecies? “Random genetic drift” (RGD) only works when the fitness field is approximately level (neutral), e.g., at most a 1-2% fitness advantage for any allele; anything bigger and all models of prevalence show the more favorable allele will eliminate the others in short order. This is an important thing to keep in mind, and it bears repeating: RGD only works when the alleles are nearly equivalent (and remain that way for the time needed to fix one or the other).

One can easily envision that the pollinator selection pressure would vary from year to year, and not just due to relative changes in the population of the two, but also due to fluctuations in the overall pollinator populations. (That is, even with a constant bees/hummingbirds ratio, in times when both populations are high, even the least fit flowers get pollinated, but if the overall population is low, the selection pressure will get accentuated.)

Thus, for RGD to fix the red allele, the selection pressure will have had to change to one that is almost nearly neutral, and 1) to have remained neutral for the many years required to eliminate the wild type in at least one population, *and* 2) to remain neutral during all of the time when there is any significant gene flow between the two populations (i.e., during the time when other genetic changes build the genetic barrier that exists today). [the second requirement is removed if there is geographical isolation.]

So, I just think an RGD hypothesis is wildly implausible given what is known: a) the red allele is phenotypically visible to a known selective process, b) that selection pressure likely is variable from year to year (and so unlikely to remain neutral), and c) many many years of a selective neutrality would be required. You base your RGD “just so story” on the assumption that the red allele is neutral relative to the pink allele and that it remained neutral for a long time, in the face of strong circumstantial evidence that it not neutral now, and that it is likely to vary with changes in the ecological milieu. That does not strike me as reasonable.

Rei Akira said...

@ Divalent: Well articulated! *applauds* Couldn't have said it better myself.

Nevertheless, here is my (much more than) two cents worth.

@Moran: When Gould and Lewontin argued for biologists to think outside the adaptationist programme, I don't think they were advocating jumping blindly and ridiculously onboard the non-adaptationist programme either.

"If a bunch of "red" flowering plants arose (by combining recessive mutations) among the "pink" plants then why would their survival be at risk? There are plenty of hummingbirds around so the chances of them setting seed are no worse than the pink plants.

We then have a situation where both red and pink plants are competing for the same ground but neither has a fitness advantage. That's what "neutral" means."

Lets start from the basics. What we see in nature is evolution from bee pollination to hummingbird pollination (to moth pollination, though that isn't important here). By doing a phylogenetic analysis, its obvious that M. lewisii and M. cardinalis are sister taxa, and their most closely related species is M. gutattus (also bee pollinated). M. lewisii is probably most indicative of what the most recent common ancestor of lewisii and cardinalis looked like, because it was also bee pollinated (assuming that this model of bee-->hummingbird pollination holds true across all flowering species, which to all our knowledge it does.).

Next, keep in mind that the recessive mutation at the YUP locus in the lewisii background causes yellow flowers. The change from pink to red only occurs with additional anthocyanins and carotenoids. What we are most likely looking at is the closest probable map of how a speciation event could have begun that led to these two species. As Divalent pointed out (and, for that matter, as Bradshaw and Schemske not only clearly state, but that you were kind enough to restate), this yup NIL in the lewisii bg led to a huge decline in bee visitation, but attracted some hummingbirds. Thus some shift in the pollinator landscape would have been required.

What we see in the natural system is a range in the Sierra Nevada Mtns where the two species live in sympatry but do not hybridize (thus they are reproductively isolated, thus they are ACTUAL species, not subspecies). But they can be hybridized in the lab BECAUSE the basis of their reproductive isolation is due entirely to pollinators. No genetic hindrances, merely pollination preference. And Bradshaw and Schemske are clearly isolating individual phenotypic differences (i.e. color) that can be traced to mutations in a single gene, and they're showing in the field with intelligently designed experiments that these phenotypic differences lead to huge pollinator discrimination. Since pollinators are the cause for their reproductive isolation, this demonstrates where natural selection is acting. Thus something as simple as a change in a single gene can possibly lead to - or at least catalyze - speciation events.

Naturally, we cannot go back in time and watch this unfold. But Divalent stated the whole "genetic drift is random with respect to fitness" thing pretty clearly, or else you'd have replied, wouldn't you? :)

In fact, I think the textbook for my Introductory Biology course, Biological Science (3rd ed, Scott Freeman), states that pretty clearly as well...

Larry Moran said...

Divalent says,

This is an important thing to keep in mind, and it bears repeating: RGD only works when the alleles are nearly equivalent (and remain that way for the time needed to fix one or the other).

That's completely wrong. Random genetic drift works on all alleles; beneficial, detrimental, and neutral. You need to re-read your textbooks. Beneficial alleles can be eliminated in populations due to random genetic drift and, conversely, detrimental alleles can be fixed.

Once you get a small cluster of plants with red flowers (by accident) they will not be at a selective disadvantage relative to the plants with pink flowers because they can be easily pollinated. However, it's not clear to me that they will have a selective advantage either.

If the red flowers were so great why didn't they take over the entire population?

Why can't you accept that the two subspecies are just DIFFERENT? Why does everything have to be explained by selective advantage?

Larry Moran said...

Rei Akira says,

... as Bradshaw and Schemske not only clearly state, but that you were kind enough to restate), this yup NIL in the lewisii bg led to a huge decline in bee visitation, but attracted some hummingbirds. Thus some shift in the pollinator landscape would have been required.

Exactly. In order for the adaptationist explanation to make sense you have to postulate a drastic shift in the relative frequencies of bees and hummingbirds just at the time the flower mutations arose.

While this is theoretically possible, it does require an additional step in order to salvage the adaptationist explanation. This additional step is usually not mentioned.

Why bother? Why not consider the possibility that the evolution of the two subspecies is just one of the millions of chance events that have occurred during the history of life? It could have happened even when the numbers of bees and hummingbirds didn't change. That seems to be a more parsimonious explanation, don't you think? (Not that parsimony makes it correct.)

In fact, I think the textbook for my Introductory Biology course, Biological Science (3rd ed, Scott Freeman), states that pretty clearly as well...

This is not relevant. One of the points I am making is that evolution is not taught correctly in university courses and that introductory biology textbooks are partly to blame.

loujost said...

I love drift as much as anybody, but I think this case is almost certainly driven primarily by natural selection.

On tropical mountains like the one I live on, there are very many species pairs (in both monocots and dicots) in which the lower-elevation member of the pair has pink or purple flowers and is pollinated by bees, and the higher-elevation member has orange or red flowers and is pollinated by hummingbirds. This is a very common pattern, and it makes sense because at chilly high elevations, cold-susceptible bees are less common than warm-blooded hummers. The reverse is true at lower elevations. Bees are smaller and more energy-efficient than hummers and can reach much higher densities than hummers if temperatures permit. So there are clear reproductive advantages for the orange mutant at high elevations.

In the Andes, which are quite young (a few million years) it is also common that these species pairs are cross-fertile, and isolated only by their pollination mechanism. Nevertheless this is a perfectly good isolating mechanism, and in my group (orchids) every taxonomist counts such pairs as perfectly good biological species.

loujost said...

I see my last sentence might be misunderstood. Taxonomists regard EACH member of such a pair to be good, distinct species.