Saturday, November 12, 2011

A New View of Evolution

There have been lots of new books about evolution in the past decade or so. I tend to divide them into three categories:
  1. The Standard View: These are books that basically support the Modern Synthesis with some small tweaks here and there. They do not advocate major shifts in the way we look at evolution. Books by Richard Dawkins (The Greatest Show on Earth: The Evidence for Evolution), Jerry Coyne (Why Evolution Is True), Sean B. Carroll (Endless Forms Most Beautiful: The New Science of Evo Devo and the Making of the Animal Kingdom, The Making of the Fittest: DNA and the Ultimate Forensic Record of Evolution), and Ken Miller (Only a Theory) fall into this category.
  2. The New View: Some books make the case for a new way of looking at evolution. I'll call it the "New View." Many of Stephen Jay Gould's books fall into this category (The Structure of Evolutionary Theory). He refers to it as extending the modern synthesis. Most of the "extension" is based on a pluralist, rather than an adaptationist approach but other modifications are important. Two recent books by Michael Lynch (Origins of Genome Architecture) and Eugene Koonin (The Logic of Chance: The Nature and Origin of Biological Evolution) fall into this category. It's a view that I share.
  3. The Radical View: Some books advocate a more-or-less complete overthrow of the Modern Synthesis, replacing it with the author's pet theory. Examples are: Marc Kirschner, and John Gerhart (The Plausibility of Life: Resolving Darwin's Dilemma), James Shapiro (Evolution: A View from the 21st Century), Lynn Margulis and Dorion Sagan (Acquiring Genomes: A Theory Of the Origin Of Species), Massimo Pigliucci and Gerd B. Müllerand (editors) (Evolution - the Extended Synthesis), many others.
You might think that these three categories represent a progression from a traditional view of evolution to a "new" view and then on to a more radical view. But the funny thing is that the "radical" views usually skip over the intermediate stage and focus on the traditional view as their main target. In fact, the authors of the "radical" books seem to be largely unaware of the "new" view.

The development of the "new" and "radical" views represent parallel tracks, not sequential tracks.

It's easy to dismiss all criticism of the traditional view based on the silliness of most of the radicals. This is a mistake. Those who advocate the "new" view have some serious points and they deserve a hearing. What does this "new" view look like? Here are some excerpts from Koonin and Lynch.

According to the Modern Synthesis, the evolution of life is a process of active adaptation of populations to changing environments. We now realize that although such adaptation is undoubtedly an essential component of the evolutionary process, it is not quantitatively dominant. Although fully aware of the oversimplification inherent in any attempts at grand definitions, I submit this:
The evolution of life is a stochastic process based on historical contingency, substantially constrained by various requirements for the maintenance of basic biological organization, and modulated by adaptation.

Eugene Koonin
Contrary to popular belief, evolution is not driven by natural selection alone. Many aspects of evolutionary change are indeed facilitated by natural selection, but all populations are influenced by nonadaptive forces of mutation, recombination, and random genetic drift. These additional forces are not simple embellishments around a primary axis of selection, but are quite the opposite—they dictate what natural selection can and cannot do. Although this basic principle has been known for a long time, it is quite remarkable that most biologists continue to interpret nearly aspect of biodiversity as an outcome of adaptive processes. This blind acceptance of natural selection as the only force relevant to evolution has led to a lot of sloppy thinking, and is probably the primary reason why evolution is viewed as a soft science by much of society.

A central point to be explained in this book is that most aspects of evolution at the genome level cannot be fully explained in adaptive terms, and moreover, that many features could not have emerged without a near-complete disengagement of the power of natural selection. This contention is supported by a wide array of comparative data, as well as by well-established principles of population genetics. However, even if such support did not exist, there is an important reason for pursuing nonadaptive (neutral) models of evolution. If one wants to confidently invoke a specific adaptive scenario to explain an observed pattern of comparative data, then an ability to reject a hypothesis based entirely on the nonadaptive forces of evolution is critical.

Michael Lynch
When it comes to the acceptance of nonadaptive evolution, we've probably reached the middle stage of a multi-step process. Today, most evolutionary biologists will say, "Of course it's true, that's what I've believed all along. What's the problem?" Then tomorrow, or the next day, they'll write or say something that sounds very much like an adaptationist.


  1. " Many aspects of evolutionary change are indeed facilitated by natural selection, but all populations are influenced by nonadaptive forces of mutation, recombination, and random genetic drift."

    What makes this a 'new' view? This has been known for a very long time. I guess people argue about the percentage of contribution.

    I am just starting to read Shapiro's book, knowing that I need to keep an open, but skeptical mind. This is not so easy for a layman like me. It can sometimes feel like drinking from a fire hose.

  2. Hi Larry,

    I confess that I'm not sure I understand how the New View differs from the Standard View. It seems that the New View simply says that more mutations may occur before selection "kicks in" than the Standard View allows. But I'm probably just confused about all this.

  3. I'll agree w/ Prof. Moran here, most presentations of evolution are focused on adaptations, and I recall that there was a helluva debate over whether Dawkins or Gould was right, people didn't just say 'Well yeah Stevie, that's not controversial'. People fought it, people said it was wrong or at best minor, or that multiple levels/pluralism was a step on a slippery slope to vitalism and such.

  4. I suspect that the difference is that scientists of the traditional view often sound like theologians in how often they ascribe natural selection to the cause of a biological phenomenon. A theologian would say "purpose", and that seems to be how some scientists view natural selection.

  5. I just want to let you know, that PZ have moved Pharyngula to - you should update your link.

    Kind regards

  6. What I find interesting is how these three views aren't set in stone. You place Sean Carroll in the Standard View, and I totally agree, but I'm not sure if Carroll himself would agree. He uses "new synthesis" language in regards to development (new view), but still pushes a selectionist account of evolution (standard view).* (So is the belief that development needs to be integrated into the Modern Synthesis count as standard, new, or radical?)

    And then you have Michael Lynch, who in the last chapter of his book, says that "nothing in evolution makes sense except in the light of evolution" (standard view) but seeks to restore(?) the role of non-adaptive evolution in evolutionary theory (non-standard). He also criticizes Carroll and other evo-devo biologists for not creating a populational framework for evo-devo (also standard).

    * Although I have come to realize that Carroll is not representative of the evo-devo viewpoint, but it seems like he thinks he is.

  7. Niels
    Pharyngula hasn't "moved" to FreethoughtBlogs, it is duplicated there (with some extras).
    It's still going strong at ScienceBlogs.
    (I sure Prof. Moran knows this.)

  8. ""nothing in evolution makes sense except in the light of evolution" (standard view)"

    You of course meant "nothing in biology", but I'm not here to nitpick, I'm confused as to how this is the standard view in your eyes?

    I see that famous quote as compatible with all "views" on evolution (well, except the creationist's of course).

  9. If researchers were a computer program selection would be the default setting. All of the other buttons are available, but in general they are not switched on until needed.

    This sort of vague bias is enough to be off putting for people interested in exploring different questions, but modern Evolutionary theory is pretty widely regarded as pluralistic for the past couple of decades. In that context a true revolution is highly unlikely, but in pushing a significantly new idea one is often backed into the revolutionary rhetoric corner.

    I'll counter balance that by stating that there is a lot to be said for fighting over the default settings. Those very general initial directions can have a profound impact on the course that investigations take, and I do think that biologists did get a little bit to selection happy in the post Synthesis period.

  10. Does Gould really belong with Koonin and Lynch?
    Lynch and Koonin are very much updated standard. Gould is way off taking a track on his own. Lynch and Koonin are not indebted to Gould but to Sewall Wright - who after all was part of the standard.
    As to the Radicals, some better knowledge of physiology and embryology is very profitable to evolutionary biology. However, part of the radicalism of the Radicals is caused by a severe misunderstanding of standard evolutionary biology, equating it with what is taught in a first year introductory course in population genetics.

  11. heleen says,

    Lynch and Koonin are very much updated standard. Gould is way off taking a track on his own. Lynch and Koonin are not indebted to Gould but to Sewall Wright - who after all was part of the standard.

    I don't lump them together because they all deal with the same subject matter. Koonin and Lynch focus on molecular evolution while Gould's main focus is macroevolution.

    I lump them together because they advocate a change in our way of thinking about evolution but it's the kind of change that's been in the air for decades.

    However, part of the radicalism of the Radicals is caused by a severe misunderstanding of standard evolutionary biology, equating it with what is taught in a first year introductory course in population genetics.

    It would be wonderful if introductory evolution courses were actually taught correctly from a population genetics perspective. That would go a long way towards eliminating the "standard" view of evolution.

    As Michael Lynch says, "Nothing in evolution makes sense except in the light of population genetics." How many instructors in introductory university evolution courses believe that? How many even know what it means?

    Very few, I'll wager. And even fewer high school teachers.

  12. anonymous says,

    If researchers were a computer program selection would be the default setting.

    And if pigs could fly we'd all have to wear wide-brimmed hats.

    All of the other buttons are available, but in general they are not switched on until needed.

    That's completely wrong. You cannot stop evolution by random genetic drift. It's going on all the time. There is no "off" button for mutation and drift.

    On the other hand, selection is relatively rare in biology, but it's the only thing that counts when you're writing "evolutionary" algorithms.

    It's almost impossible for a computer scientist to really understand evolution. Evolution is not an algorithm. That's the mistake Daniel Dennett made when he wrote Dawin's Dangerous Idea.

    1. Laurence, a biochemist myself I appreciate your emphasis on non-adaptive processes. Computer algorithms are not only limited in how they model evolution, they are actually biased towards the programmers preferred outcome of natural selection, e.g. walking robots. Chance in evolution is underrated, and I personally believe that evolution operates on a 'good enough' mechanism.

  13. My computer metaphor was not targeted at representing evolutionary theory. It was merely intended to provide an outline of how I have observed researchers approach potential evolutionary questions, they tend to look for selection as an explanation first. If that does not seem to be doing the job most are perfectly willing to use the other explanatory tools available in modern theory.

  14. Dr Moran quotes Michael Lynch: "Nothing in evolution makes sense except in the light of population genetics."

    It is too bad that population genetics is so wrong about some of its most basic concepts. We can't base evolution on population genetics until geneticists clean up their act. As I've mentioned here before, they are wrong about the factors controlling population divergence, for example.

  15. I'm not sure that this terminology is entirely useful or accurate. I subscribe to what you outline as the "New View" and yet, if asked, I would describe (and, indeed, consider) it to be the "standard view". It was what I was learnt at University in the '90s, and what I teach my own students now.

    The importance of neutral evolution and chance has a long history and I see their recognition as major, even dominant (depending on the focus), forces in evolution as something that has come naturally from the older position. I think it is a continuum, not a divide. You have staunch "old school" Adaptionists at one end, extreme Neutralists at the other - both pretty rare, I would wager - and everyone else is in the middle, exploring how much adaptation and how much neutral evolution and chance play a role in different situations. Indeed, selection and neutrality are themselves a continuum. Making a false division just encourages people to falsely believe that they have to choose. Demonstrably both are important.

    Personally, I prefer to split views/positions into "standard" (including "new") and "radical". Radical views are sometimes right but usually more by luck than judgement - if enough evidence has accrued to really support a position, it is unlikely to be radical. (I suspect my own division opens a whole new can of semantic worms, though.)

  16. Lou, could you please elaborate on the wrongness of PopGen? Is it about the diversity/similarity metrics?

  17. Lou Jost

    It is too bad that population genetics is so wrong about some of its most basic concepts.

    Perhaps as a weak mathematician, I have always felt rather unconvinced by many pop-genetic arguments. The panmictic population that underlies many models seems something of a fiction, as does the assumption of constancy - even stochastic constancy - in selective differentials.

    I have been roundly roasted for saying so in the past, but I think the 'cost of sex' is a pop-genetic fallacy. And Lynch's use of pop genetics to explain eukaryote-prokaryote differences in junk DNA seems to suffer from similar deficiencies. For an asexual species, one of the principal 'motors of mixing' - mate seeking - has been turned off, and with it much of the homogenisation assumed by pop-genetic models. It's not the effective 'population' size that bothers a prokaryote, it is its immediate neighbours, whatever their species.

  18. Geoxus, yes, it is because geneticists have based the analysis of population divergence on a measure, Fst, that has virtually nothing to do with divergence but rather with fixation. So they think Nm controls divergence at neutral loci. They (and I suppose Larry as well) think that a set of demes evolve as a unit if Nm>>1, and diverge due to mutation/drift if Nm <<1.

    You can see this is false if you run individual-based simulations. If you do that, you'll see that divergence is controlled by m(n-1)/u rather than Nm.

    That changes a lot of things about pop gen. For one thing, it means there is no genome-wide divergence at neutral loci, since u (mutation rate) varies from locus to locus.

    It also changes our understanding of the conditions that favor speciation.

  19. Allan, I think some of these suspicions are justified. Geneticists recognize that models like the finite island model are almost never correct for any given system, but many results are based on that model (out of necessity, since more realistic models often don't give analytical solutions).

    In fact I think a more robust population genetics is possible. In the case of differentiation, the interpretation of Gst or Fst is very sensitive to some of the assumptions of the underlying model, but interpretation of my D is model-independent. It can also give good estimates of some model parameters even if deme sizes are not constant over time. I bet there is more that could be done along these lines.

  20. Thanks, Lou. I have a very superficial (and probably naive) understanding of PopGen, so if I wanted to go deeper into it, what introductory readings would you recommend to "get it right"?

  21. Geoxus, what is your background? Are you more familiar with ecological diversity and differentiation measures? If so, it is easier to understand the mathematics of diversity and differentiation from that angle, and then move to the pop gen stuff.

    Here are some starting references about these issues in pop gen and ecology, with some comments about them:

    Hedrick P (2005) A standardized genetic differentiation measure. Evolution, 59, 1633-1638.

    This paper by Hedrick is an influential paper which noted that the "differentiation" measure Fst is confounded with within-group diversity when this diversity is high. The paper proposed an ad hoc solution, which is pretty close to the true solution.

    Gregorius, H-R., Degen, B., & Konig, A. Problems in the analysis of genetic differentiation among populations -a case study in Quercus robar, Silvae Genetica 56, 190-199 (2007), see its Appendix

    Gregorius, H.-R. Linking diversity and differentiation. Diversity 2, 370-394 (2010).

    These two papers noticed something that Hedrick missed: Fst and its relatives (Gst, for example) also fail at the low end of the diversity spectrum.

    Jost L (2006) Entropy and diversity. Oikos, 113, 363-375

    Jost L (2007) Partitioning diversity into independent alpha and beta components. Ecology, 88, 2427-2439.

    These two articles treat the mathematics of diversity and differentiation in ecology. These lay the mathematical groundwork which I applied to the pop gen problem in the following paper:

    Jost, L. Gst and its relatives do not measure differentiation. Mol. Ecol. 17, 4015-4026 (2008).

    This paper has stirred up a lot of reaction, both favorable and unfavorable. I think the criticism demonstrates a fundamental misunderstanding of my paper. They criticize my measures because it does not behave like the classical ones, but you can see from simulations that my measure makes the correct predictions and the classical approach makes dramatically incorrect measures. I've posted simulations in the Comment section of Nolan Kane's post in the Molecular Ecologist:

    Here are a few of the critical and articles reacting to mine, and one response of mine:

    Heller R, Siegismund H (2008) Relationship between three measures of genetic
    differentiation Gst, Dest, and Gst’: How wrong have we been? Molecular Ecology, 18, 2080-2083

    Ryman, N., and Leimar, O. Gst is still a useful measure of genetic differentiation- a comment on Jost’s D. Mol. Ecol. 18, 2084-2087 (2009).

    Jost, L. D vs. Gst: Response to Heller and Siegismund (2009) and Ryman and Leimar (2009). Mol. Ecol. 18, 2088-2091 (2009).

    Meirmans, P.G., & Hedrick, P.W. Assessing population structure: Fst and related measures. Mol. Ecol. Resources 11, 5-18 (2011).

    Whitlock, M. G’st and D do not replace Fst. Mol. Ecol. 20, 1083-1091.

    Ellison, Aaron M. 2010. Partitioning diversity. Ecology 91:1962–1963.

    A more detailed explanation of the relation between diversity and differentiation is:
    Jost, L., Chao, A. and Chazdon, R. L. (2010). Compositional similarity and beta diversity. In Biological Diversity: Frontiers in Measurement and Assessment (eds A. Magurran & B. McGill) Oxford: Oxford University Press.

    That is probably more than you wanted to know!!!!

  22. @ Geoxus

    Joe Felsenstein has a .pdf for you personal use on his website.

  23. The Felsenstein book is a excellent intro text, and a prescient one. If you look at it carefully, his measure rho (page 234) turns out to be 1- my differentiation measure D. It had been used even earlier by Spaeth, Latter, and Maynard-Smith, and a simple transformation of it has bee used by Nei to measure genetic distance. Because of historical accidents and the absence of a fundamental proof connecting it to differentiation, it has been overshadowed in the literature by Wright's Fst and Gst.

    I should mention that Ayala also used this measure in the 70's, and showed empirically (by applying it to fruit flies of different taxonomic categories) that it tracks the biological species concept. The usual measure of differentiation, Fst, does not. SO it is surprising (even shameful) that Fst and its relatives still dominate the literature as the preferred measure of differentiation.

    In the list of articles in my previous post, I forgot to mention probably my simplest intro to the mathematical issues surrounding diversity and differentiation measures:
    Jost, L. (2009) Mismeasuring biological diversity: Response to Hoffmann and Hoffmann (2008). Ecological Economics, 68, 925–928.
    Read this first.

  24. Lou, TOJ, thanks for your answers!

    That's more than what I was looking for and I suspect some of it will be over my head, but it makes a very handy list. I've been using Joe's book for reference, so it's great to know it is OK on this.

    As for my background, I'm more into phylogenetics, but I've done some palaeoecological work with differentiation and diversity measures (Jaccard, Simpson, &c), so I'll take your advice and start from there.

  25. If you need any of the papers, let me know. You do have to be careful of Felsenstein's treatment. He sometimes interprets Fst as a measure of genetic divergence or differentiation between populations (e.g p 213, 214), and this is wrong. He also gives a more nuanced interpretation, which is fine. Fst is not monotonic with respect to any reasonable model-independent definition of divergence, and can be near zero (supposedly indicating virtually no divergence) even when the populations don't share any alleles at all, because Fst is confounded with within-group diversity.

    Joe sometimes participates in these Sandwalk comments. I hope he joins this thread; I would be very interested to hear his comments about all this, which is still being hotly debated in the literature.

  26. Geoxus, since you mentioned the Simpson index, it will be helpful to remember that the Gini-Simpson index of ecologists (1- sum of squared species relative abundances) is the heterozygosity or "gene diversity" of genetics. Gst is derived from that, mistakenly using additive partitioning on a non-additive measure. The criticisms I have made in the ecology literature regarding the additive partitioning of the Gini-Simpson index can therefore be translated directly into pop gen.

  27. Thanks Lou, I'll drop you an email if I can't find some of the papers. I also think I should get myself into a proper pop gen course some time. My education was lacking in that respect and I probably won't be able to understand the mathematical nuances just by myself.

  28. Geoxus, that's probably a good idea. Just be alert, since the textbooks will teach the old ways. Be skeptical about their interpretation of Gst or Fst (it is always fun to look at artificial datasets to see how wrong the classical interpretation can be---even more fun to watch your teacher puzzle over it!!!). Also watch out how they use the coalescent, which often gets abused. And remember heterozygosity always needs to be converted to effective number of alleles before it can be interpreted as a diversity.

    And of course, don't buy their explanation of drift/mutation/migration equilibrium. Nm does not determine whether a set of populations drifts apart or drifts as a single cohesive unit. That is determined by m/[u(n-1)].

    Note a typo in my earlier post of that formula, by the way.

    Wish Larry would post on this, as he is a big fan of drift.

  29. @Allan Miller
    I have been roundly roasted for saying so in the past, but I think the 'cost of sex' is a pop-genetic fallacy.
    My experience too. ‘The cost of sex’ due to basic genetics does not exist. I’ve constructed life histories with different relations between sexuals and asexuals, already in 1980. One can make the balance go to sexuals depending upon the life history and ecology.