Saturday, September 20, 2014

A Tale of Two Morphs

Back in the summer of 2010, I was working on what became a chapter of my Ph.D. in Austria, as part of the Young Scientists’ Summer Program (YSSP) at the International Institute for Applied Systems Analysis.  You can read more about that amazing experience, and the research project that it led to, in my post on Evolutionary branching in complex landscapes.  While I was there, Andrew Hendry, my Ph.D. advisor, sent me an email saying “Hey, there’s a researcher in Zurich who has an interesting modeling idea, perhaps you might want to talk with her.”  And thus was a collaboration born that ultimately led to another thesis chapter – serendipity!  At the end of the YSSP, I hopped on a train from Vienna to Zurich (a truly lovely journey of about 8 hours), spent a couple of days with Elena Conti, Barbara Keller, and Jurriaan de Vos at the Institute for Systematic Botany, and together we sketched out the idea for what became a paper we just got published in PLoS ONE.  And that’s what I want to write about today.

A view from the train from Vienna to Zurich.  Photo: Ben Haller.  (For lots more photos from my time in Austria, including side trips to Prague, Budapest, Munich, and Abisko, you can check out, my photography website.)

It’s a difficult project to write about, because it is deeply intertwined with the precise workings of a floral syndrome called heterostyly that is itself not easy to explain.  Heterostyly attracted the attention of Darwin himself, who wrote at some length about it in his treatise The Different Forms of Flowers On Plants of the Same Species, and it has proved fascinating to evolutionary biologists ever since.  In essence, heterostyly typically has two components: a floral polymorphism, and an intra-morph sexual incompatibility system.  In the simplest form, called distyly, there are two floral morphs, and these two morphs can cross with each other but (mostly, to varying degrees in different species) cannot cross with themselves. The two morphs are called pins and thrums, and they are characterized by different heights of the reproductive organs within the corolla tube of the flower, as drawn by Darwin:

A pin, on the left, with a high style and low anthers, and a thrum, on the right, with a low style and high anthers.  Drawing: Charles Darwin.

The key point to notice is that the positions of the reproductive organs in the two morphs are reciprocal: the high style in the pin is at about the same height as the high anthers in the thrum, and the low style in the thrum is at about the same height as the low anthers in the pin.  Although the precise way in which heterostyly functions to increase the fitness of flowers is a bit subtle, having to do with reduction of sexual interference between the male and female functions of the flower (see Barrett 2002 and Barrett & Shore 2008), the important point here is that studies indicate that this reciprocal positioning tends to lead to reciprocal pollination: pollen tends to move from pins to thrums, and from thrums to pins, rather than between flowers of the same morphology, because pollinators tend to transport pollen positionally.  Pollinators do this because they approach flowers in a somewhat consistent manner, and the same parts of their bodies tend to contact the same parts of the flower in each flower they visit.  So if a particular part of a bee’s glossa (like a tongue) brushes against the anthers of a thrum, and picks up pollen, then if the bee next visits another thrum, that pollen is relatively unlikely to be delivered to the second flower’s stigma; instead, that particular part of the glossa will brush against the anthers again, and will just pick up more pollen from the second flower.  When the bee eventually does visit a pin, that particular part of the glossa will again brush against the same spot in the flower – but now, there is a stigma there, because this flower is a pin, and so the pollen stuck to the glossa will be delivered.

I don’t mean to make this sound too precise; no doubt it is a very messy business, and pollen ends up all over the place.  What I am describing is only a tendency: the probability that pollen will be transported between different morphs is somewhat higher than the probability that pollen will be transported between identical morphs.  Or to put it in a more particular way, for reasons that will become clear: if a pollen grain starts at a particular height h in the corolla tube of a flower (the height of the anthers), what in the probability distribution for the height h′ at which that pollen grain is likely to be offered to the next flower by the pollinator?  What is the precision of pollen transfer, for a given species of pollinator?  The short answer is that nobody really knows the answer to this; it is extremely difficult to measure this empirically.  However, heterostyly has independently evolved more than 20 times in the angiosperms, so it appears to be quite beneficial; that suggests that the precision of pollen transfer might be high enough to warrant some careful thought about its potential consequences.

Pollination: a messy business!  (This bee is on a sunflower, a non-heterostylous species.)  Photo: Ben Haller.

The key idea that we wanted to pursue in this project is that this precise pollen transfer might lead to reproductive isolation between different populations of a heterostylous species, reducing gene flow between the species and thus promoting speciation.  How would this work?  Consider the diagram below:

A: Morphs with reciprocally matching reproductive organ positions cross well when pollen transfer is precise, because the pollen picked up at the anthers of one morph is delivered to the correct height to be received by the stigma of the reciprocal morph.  B: Morphs with mismatched reproductive organ positions do not cross well when pollen transfer is precise, because the pollen gets delivered at the wrong height (i.e., it is stuck to a part of the pollinator’s body that does not contact the stigma in the destination flower).  Diagram is from our PLoS ONE paper, Figure 1.

So if two populations of heterostylous plants evolved different reproductive-organ heights, that would decrease the gene flow between them to some extent – an extent that would depend on the precision of pollen transfer.  That decreased gene flow would allow them to diverge in other respects as well, since they would be (somewhat) released from the homogenizing effects of gene flow.  In the end, the divergence afforded by precise pollen transfer in the context of heterostyly might be the first step down the road to ecological speciation.

This post has rambled on long enough, and I haven't even gotten to what we actually did!  So, to make a long story short: we designed an individual-based model that tracked the movement of every pollen grain between every pair of flowers, we ran the model using various types of simulated pollinators, and we observed the degree of ecological divergence that two populations of flowers could reach when subjected to divergent selective pressures in their respective environments.  The pollinators occasionally flew from one patch to the other, transporting pollen with them and thus producing gene flow that constrained that adaptive divergence; but if the flowers evolved mismatched reproductive-organ positions, they could decrease that gene flow, and thus be free to become better-adapted to their local environment.  We varied the frequency of inter-patch pollinator movement, the strength of divergent selection, and the precision of pollen transfer, and we observed the extent of adaptive divergence attained between the two populations (relative to a set of control runs without precise pollen transfer).

What did we observe?  For details on that, you will have to look at our paper!  Suffice to say that it was a complex tale: in one scenario we observed increased divergence due to decreased gene flow, but in a different scenario we observed decreased divergence due to increased, asymmetric gene flow – a result that quite surprised us!  Furthermore, in one scenario the reproductive-organ traits acted as magic traits, influencing both the ecological fitness and the reproductive isolation of the populations, while in the other scenario they did not.  There’s discussion of “magic environments” and “magic modifiers” and all sorts of wonderful magical things; the upshot of all of it is that, as I and coauthors have written about before, the term “magic trait” is perhaps a bit of a misnomer.  The “magic” really happens in the interaction between an ordinary trait that happens to have an effect on assortative mating, and an environment that happens to produce divergent selection on that trait.  The effect size of the magic trait on divergence and speciation, furthermore, is again not entirely a function of the trait itself, but is instead affected by other genetic and environmental factors that influence the strength of the effect of the trait on non-random mating and on local adaptation.  In this paper, we call these external factors “magic modifiers”.  The precision of pollen transfer proves, in our model, to be such a magic modifier.

What does that mean in empirical terms?  What would be the actual magic modifier?  One possibility that we point to is the shape of the corolla tube.  Flowers with wide-open corolla tubes place relatively little constraint on pollinators, and so the precision of pollen transfer might be quite low between such flowers.  Flowers with long, narrow corolla tubes tend to constrain how the pollinator interacts with the flower, increasing the precision of pollen transfer because the same parts of the pollinator will contact the same parts of the flower in each visit.  Evolving a longer, narrower corolla tube might therefore be a way of evolving a higher precision of pollen transfer, increasing the degree of “magicness” afforded by the reproductive-organ heights of the flower, and thus allowing greater divergence and speciation.  This ventures well into the realm of speculation, but it is interesting to note that the ancestrally heterostylous clade Primula (including nested genera), with ~550 species, typically possesses long, narrow corolla tubes with the sexual organs concealed inside, whereas its sister clade, the genus Soldanella, is non-heterostylous, is typified by open, dish-shaped corollas, and has only 25 species.  So perhaps heterostyly does not always provide a magic-trait mechanism that promotes diversification; but perhaps Primula is an example of a case where a magic modifier, the corolla shape, evolved to allow it to do so.

So if you ask “are the floral reproductive-organ heights in heterostyly magic traits?”, the answer appears, according to our model, to be “it depends on the environment”.  This suggests that if we want to understand why magic traits appear to be common in nature, and how they influence the process of speciation, we will need to broaden our perspective from thinking about the evolution of traits to thinking about the evolution of ecological interactions, and in particular, the evolution of magic environments and magic modifiers.  It’s an eco-evolutionary problem!


B.C. Haller, J.M. de Vos, B. Keller, A.P. Hendry, E. Conti. (2014). A tale of two morphs: Modeling pollen transfer, magic traits, and reproductive isolation in parapatry. PLoS ONE 9(9), e106512. DOI: 10.1371/journal.pone.0106512

Carnival of Evolution #75

Whoops!  The delay in the August Carnival of Evolution (#74) got me confused, so I forgot to post about the September Carnival, which came out right on the heels of the August one.  Well, maybe this gave you all some breathing room, anyway.

So, the September Carnival, #75, has been out for a couple of weeks now.  Our contribution to it was my post about interactions between theoretical and empirical research in ecology and evolution.  I conducted a survey of scientists in ecology and evolution, and got more than 600 responses.  The survey results were interesting, and it’s a subject that ought to be relevant to everybody!

There’s lots of other cool stuff in this Carnival too, from a discussion of a new (possible) case of sympatric speciation, to a takedown of a new claim for intelligent design from Behe, to Carl Zimmer on new findings in the evolution of land-living tetrapods from fish!  Check it out!

ALSO: Our very own Felipe Pérez-Jvostov will be hosting the next Carnival, #76, here at eco-evo-evo-eco in just a little over a week!  So if you’ve got a blog post about something evolution-ish, and you want it in the Carnival (because why not?), submit it to Felipe!  His email: felipe [dot] perezjvostov {at} mail (dot) mcgill <dot> ca, with appropriate symbolic replacements.  :->

Monday, September 15, 2014

The ideal experiment...

[ This blog post is by Sinéad Collins; I am just putting it up.  –B. ]

In a previous blog post, Andrew outlined the “ideal approach” for investigating whether plasticity facilitates evolution, and to my delight, he proposed experimental evolution. Not only that, but he proposed the experiment my group got published this week in Proceedings B. I was pleased to be accused of doing anything ideal, much less an ideal experiment, even if Andrew doesn’t seem to remember me presenting this very data at the American Genetics Association Meeting.

When Elisa Schaum (the PhD student behind all this work) and I planned this experiment 3 years ago, our main concern was that it was, if anything, too obvious: in order to test whether plasticity facilitates evolution, take plastic and non-plastic populations, put them in new environments, and watch them evolve. Not so easy if you study elephants, but completely doable if you study microalgae (like we do).

To conduct our “ideal experiment” (do I like the sound of that too much?), we used plastic and non-plastic isolates of the small but mighty marine picoplankton Ostreococcus. Ostreococcus is exciting for many reasons, among them that it is the smallest known free-living eukaryote and yet manages to house a huge virus. However,  we chose it mostly because it is distributed over most of the world’s oceans, and we supposed that Ostreococcus from different locations would differ in how plastic they were in their response to CO2 enrichment (we were right, and we published this in Nature Climate Change). We used 16 different isolates of Ostreococcus from different locations.  We found that isolates from environments with more variable and less predictable CO2 levels showed the largest plastic response to changes in CO2, meaning that we had plastic and non-plastic (and intermediately plastic) genotypes of Ostreococcus.

Two TEM images of Ostreococcus. Photos: C.E. Schaum.

Then, we set up the evolution experiment. We let all of the genotypes evolve in 4 different environments. First, we used a control environment where CO2 levels were normal and stable. Second, we used a fluctuating environment, where mean CO2 levels were the same as the control, but they fluctuated around this mean every few generations – we hypothesized that this environment would select for plasticity, but not for adaptation to high CO2. Third, we used a stable high CO2 environment, where we could look at how the initial plasticity of the genotypes affected evolution in a new environment even if there was no further need for plasticity. Finally, we used a fluctuating high CO2 environment, where mean CO2 levels were high, but also fluctuated every few generations, to look at how plasticity affected evolution in a new environment when there was also selection for plasticity.  Then, we let everything evolve for a few hundred generations. We are now up to 1000 generations in the lab, but the paper was written before we reached this point of insanity.

Aaaannnnnd… plasticity facilitates evolution. Genotypes that were more plastic evolved more in high CO2 environments. Not only that, but populations in fluctuating high CO2 environments evolved more than populations in stable high CO2 environments. And to make matters even more exciting, populations evolved in fluctuating environments were more plastic than populations evolved in stable environments, no matter what the level of CO2. So, even when plasticity itself is selected for, populations evolving in response to an environmental change still evolve faster than populations dealing with that same environmental change who don’t have to bother with selection for plasticity. I may have done a happy dance when I saw that data.

Dr. Collins expressing her love for Osteococcus, post-results. Photo: Jane Charlesworth. [We tried to obtain a video of the good-data happy dance, but it was not available at press time. – The Management]

Of course, things are never that simple. The evolutionary response of Ostreoccocus to high CO2 can only be described as weird. I think this is because CO2 is food for many photosynthetic organisms, including Ostreococcus. So, when CO2 levels increase, Ostreococcus cells divide faster. This means that working with high CO2 here is at odds with the usual way of doing an evolution experiment with microbes, where researchers generally starve, poison, overheat, or do some other horrible thing to decrease microbial fitness substantially at the beginning of the experiment. However, we discovered that we were (eventually, and inadvertently) also guilty of torturing our microbes, as it turns out that a higher growth rate is all well and fine for a few generations for Ostreococcus, but after a while, dividing so quickly takes a toll, and the cells become less able to survive the slings and arrows of outrageous fortune (heat), have leaky mitochondria, and are bad at competing against other Ostreococcus. So, the evolutionary response to high CO2 in Ostreococcus – the response that results in cells that have normally-functioning mitochondria, can handle a bit of heat, and can overgrow other genotypes – is to grow more slowly. Basically, evolution reverses the plastic response to high CO2. Even though cells grow faster in the short term in high COenvironments, they slow back down again if given enough time to evolve. Most theory for evolutionary biology isn’t tested in enriched environments, so it took us a while (and quite a few cups of hot chocolate) to figure that out.

So yes, I would say that the experiment was ideal. It had everything: tiny protagonists (Ostreococcus), clear results (plasticity facilitates evolution!), weird and surprising twists in the clear data (evolving slower growth than your own ancestor!), and a happy dance (possibly two).

Sinéad Collins and Elisa Schaum
Institute of Evolutionary Biology, University of Edinburgh

Proceedings B paper:
Nature Climate Change paper:

Tuesday, September 9, 2014

MYScience: A newer faster cheaper easier BETTER open access journal

  • Tired of being rejected from PLoS ONE, PeerJ, and Arxiv?
  • Don’t want to reformat for the next journal?
  • Annoyed by referee and editor comments?
  • Irked by suggestions that your paper is better suited for some other journal?
  • Bothered by arguments to reduce text and transfer things to supplementary appendices?
  • Wearied by having to select only a few relevant references?
  • Irritated at having to submit your data to an online database?
  • Chuffed to say what you want how you want as long and as often as you want?
  • Want your paper accepted and published instantly?
  • Hope to publish the same study twice – or more?
If the answer to any of these questions is yes (or no), then we would like to encourage you to submit your paper to our new open access journal: MYScience.

MYScience has a number of critical advantages:
  • We accept all papers regardless of importance, content, scientific accuracy, or statistical rigor.
  • We don’t care if you have published your paper previously, in whole or in part. If it was good once, it will be even better twice. (The exception: we do not consider papers already published in our sister journal: The Journal of Universal Rejection.) You might want to take particular advantage of this benefit if you previously published in a journal that was NOT open access. This way you can publish in that other journal and yet still have it be open access in our journal. Moreover, you can just submit to us the version published elsewhere, thus avoiding having to reformat it. In these cases, please do add (handwritten is fine) a note at the top of the paper that states: “modified from [INSERT ORIGINAL CITATION HERE]”. 
  • We promise to not provide any suggestions for revision – of any sort. It’s your paper, not ours – and it will look like it.
  • We won’t waste your time in obtaining “expert” review. In fact, we won’t even waste our time obtaining any reviews. You know what you want to see in print. Who are we to quibble?
  • We encourage the use of ALL relevant references (and any non-relevant references that you think should be cited more often). You can cite each paper as often as you like and we encourage self-citation of your earlier (or forthcoming or planned or envisioned or imagined) work.
  • Add as many tables and figures as you like. In fact, we encourage taking those cluttered multi-panel figures that other journals encourage and splitting them up into as many individual figures as possible.
  • Don’t worry about supplementary materials – just put it in the main text. If it wasn’t interesting or important, you wouldn't have done it, so let us see it – all of it. If you have already formatted with supplementary materials, just past the supplementary stuff after the main text and we will publish them together.
  • If you feel compelled to do so, send us your data, raw or otherwise, in a file of any format you like and we will post it as is alongside your paper.
  • We will accept your paper instantly. In fact, consider your paper accepted immediately after submission even if you don’t hear back from us. We are likely accepting other papers (or tweeting) at the moment and will get to yours soon.
  • We will publish your paper as soon as we see it. Format it exactly how you want it to look – we won’t make any changes of any sort and we will post it on our website ( instantly and in the exactly form in which it was submitted, including the original file format. In fact, we won’t even open the file before publishing. That would just slow us down.

As this is an open access journal, we expect the authors to contribute to the cost of publishing. Our fee is $10 per submitted (and thus published) page, which is considerably cheaper than any other open access journal. We know this matters to you.

We look forward to receiving your paper so you can get on with your life.


Wednesday, September 3, 2014

The Ghost of Plasticity Past, or Why Plasticity Inferences are too Plastic

Several months ago, I attended a meeting of the American Genetics Association organized by Robin Waples in Seattle, Washington. The theme for the meeting was “Evolution and Plasticity: Adaptive Responses by Species to Human-Mediated Changes to their Ecosystems.” The meeting was a particularly clear example to the current excitement about the role of plasticity (including epigenetics) in the evolutionary process – an enthusiasm that really crystalized around Mary Jane West-Eberhard’s “Developmental Plasticity and Evolution”. Much of the excitement centers around the idea that plasticity (a single genotype has different phenotypes in different environments) can promote evolutionary change. For instance, plasticity can provide the raw material necessary for evolutionary change by revealing otherwise cryptic variation. In addition, plasticity might generate immediate adaptive phenotypic shifts that allow individuals and populations to persist in new environments, thus allowing and promoting subsequent adaptation (without the initial adaptive plastic shift, the population would have gone extinct). This idea is not new, having originated with James Baldwin in 1896, but it has certainly become vastly more popular of late.

These (and other) ideas for how plasticity might promote evolution are certainly interesting, but how does one go about testing them? One of the things I found simultaneously most interesting and frustrating about the Seattle meeting was that seemingly any pattern in the data was being interpreted as evidence that plasticity promotes evolution. Perhaps most commonly, a finding that plastic responses to a particular environmental difference were in the same direction as evolved differences between populations evolving under the same environmental difference was interpreted as evidence that plasticity promoted evolution. For instance, guppies might shoal more in the laboratory when exposed to predator cues and guppy populations evolving with predators might shoal more than those evolving without predators (independent of proximate predator cues in the laboratory). On the exact opposite hand, however, evidence that plastic responses to a particular environmental condition were in the opposite direction as evolved differences was also interpreted (by other presenters) as evidence that plasticity promoted evolution. In this case, the idea is that the plasticity is likely maladaptive and so requires evolutionary compensation.

The current inferential scheme for whether or not plasticity promotes evolution
If either of two opposite patterns is interpreted as evidence for the same thing (i.e., plasticity promotes evolution, albeit in different ways), how do we proceed with rigorous hypothesis testing. Surely we also need a set of results that would be interpreted as evidence that plasticity does NOT promote evolution. That is, the hypothesis must also be clearly falsifiable through some particular outcome in the data. The lack of such a criterion reminds me of the classic criticism by Joseph Connell that competition was invoked when species have similar diets (because they obviously then use the same resources) and also when species have different diets (because past competition had presumably reduced diet overlap – the ghost of competition past).

In the past, when I pointed out to some ecologists that competition seemed of little importance as a mechanism determining a particular species' distribution, they often gave the following answer. The reason, they said, for my inability to find evidence for competition was because it had already been eliminated by past coevolutionary divergence between those species. However, for the reasons discussed in this paper, and until some strong evidence is obtained from field experiments along the lines suggested above, I will no longer be persuaded by such invoking of "the Ghost of Competition Past". (Connell 1980 – Oikos)

Although the Seattle meeting ended several months ago, I was just prompted to write a post about the topic owing to the recent publication in Nature of another study about the role of plasticity in evolution: “Developmental Plasticity and the Origin of Tetrapods.” Em Standen, Trina Du, and Hans Larsson wanted to test whether plasticity would promote major evolutionary transitions, especially the transition from swimming in water to walking on land. To test this possibility, they invoked a classic hypothesis test: if an ancestral species shows adaptive plasticity in response to an environmental shift, and that plasticity mirrors the evolutionary changes that accompanied the environmental shift, then plasticity likely promoted the transition. Specifically, they reasoned that rearing an air-breathing fish out of the water (by misting them regularly) would cause developmental changes in morphology that would lead those fish to be better walkers out of water. If so, the same thing may have taken place in the water-to-land transition. This is a cool idea – although, of course, plasticity in the ancestor is merely consistent with the hypothesis rather than proof of it.

The difficulty in any such study is that the actual ancestor (here the ancestor of land-dwelling tetrapods) is not accessible for study – and so Standen and colleagues needed an analog species. Coecocanths obviously wouldn’t work and lungfish are just too pathetic on land. So the authors chose to use another basal tetrapod – the Polypterus or “bichir.” Performing the experiment, they found that Polypterus raised out of water were indeed better at walking out of water than those raised in the water (see the cool video below) and showed developmental morphological changes that were similar to those in the fossil record associated with the move out of water.

This result is fascinating and I was privileged to see its implementation as the work was done in the lab of my close colleague Hans Larsson. In fact, I was able to talk to him about it multiple times on the train on the way to work. In addition to simply be jealous that I hadn’t thought of it first, I came to crystallize a particular criticism of the work. Specifically, I contend (if only as Devil’s Advocate) that the results could just as easily be interpreted as showing that developmental plasticity does NOT promote evolution. The reason is that Polypertus has never made the transition to land despite millions of years of opportunity to do so. Thus, all this wonderful plasticity did NOT accomplish the task it is inferred to assist.

Given all these flexibility in ad hoc interpretation, it seems to me that the field needs a critique and a careful (a la Connell) outline of the various patterns that might be observed in an experiment and what inferences they would and would not allow. Until such an endeavor is undertaken and adopted, inferences about plasticity are simply too plastic.

I wrote the above on the train a few seats away from Hans and, while then walking to work, we discussed what the optimal experiments would look like. We think that the ideal approach would be experimental evolution: have replicate plastic and non-plastic genotypes/populations, expose them to new conditions, and track their subsequent evolutionary trajectories. If the most plastic genotypes show the fastest and most dramatic evolution, then plasticity promotes evolution. If the least plastic genotypes show the fastest and most dramatic evolution, plasticity constrains evolution. If plastic and non-plastic genotypes don’t differ, plasticity does not influence evolution. Until then, I will no longer be persuaded by such invoking of "the Ghost of Plasticity Past".