In defense of constructive neutral evolution - Part I

Caution: What follows is mostly an opinion piece by an undergrad. While said undergrad has done a fair amount of reading on the topic, the post is still subject to many errors. Tread carefully. [/disclaimer]

I won't go into an all-out discussion of neutral evolution here: I'm neither qualified enough nor have enough spare time at the moment. However, some issues seem to crop up multiple times, both here and on other blogs. I figured I'd try to briefly adress some of them, although do take my discussion with a grain of salt. That said, while neutral theory require a certain amount of effort to grasp properly (just like any other aspect of evolutionary biology), it is not something worth dismissing. In fact, if you consider how horribly misunderstood evolutionary biology is on the whole (even among grad students: Gregory & Ellis 2008 BioScience), the neutral elements of evolution seem to be understood by a very small fraction of biologists even.

Perhaps part of the problem is that adaptive evolution is just...flashier. It makes for fun and fairly simple stories: Eg. the peacock has a huge tail to signal to the females that it has nice genes that would compensate for the problems it causes. This "Good Genes" theory is actually taught in first year curricula, and makes very little sense upon further examination, and definitely does not survive Occam's Razor. A simpler explanation would, of course, entail something like runaway sexual selection (ie. female happened to prefer flashy tail, males with flashier tails outcompete their dimmer counterparts, tail gets longer) or that the tail may have a more important function, like scaring away predators. In any case, the adaptive approach very often results in what is mostly a story-telling exercise, and one that is very difficult to deal with experimentally. Worst of all, those stories very easily make sense upon first glance, and thus the field becomes cluttered with poorly thought out theories that sound reasonable.

This post became way too long, so I broke it up into three parts; table of contents here:
Part I
-Adaptationism vs. Neutralism
-"Population genetics ignores reality!"
-Existence of neutrality and near-neutrality
Part II
-Neutral evolution is relevant
-Evolution lacks foresight; it can neither anticipate nor respond
-Rise of complexity through non-adaptive means
-Further examples of constructive neutral evolution
Part III
-Discussion of what sparked this argument: Evolution of ciliate nuclear dimorphism

Adaptationism vs. Neutralism
Of course, none of what I said is new by any margin: the famous Gould and Lewontin 1979 Spandrels of San Marco (free access) paper does a nice job at pointing out many of the fallbacks of panadaptationism. And I don't find it much of an 'attack', as it is often described by diehard adaptationists, but perhaps that's just me. Since so many before me have pointed out the fallbacks of the 'adaptationist programme', I won't bother dwelling on it any further. Besides, this type of thing causes a great polarising effect on the community, with people being either hardcore adaptationists or hardcore neutralists. This seriously fucks up any progress on the topic, because biology hardly tolerates dichotomies. In fact, the truth in this case does not even lie 'somewhere in the middle', but in the fact that both adaptive and neutral processes work in tandem.

Let me reiterate that: Selective and neutral mechanisms work in tandem. Simultaneously. In some situations (eg. large effective population sizes, in bacteria; Lynch 2007 PNAS, Yi 2005 Bioessays), adaptive processes are more dominant; in some cases, adaptive 'forces' are negligible compared to neutral phenomena (small effective population sizes). Considering some specific structure, parts of its evolutionary history have been driven more by drift and mutational bias, interspersed with parts dominated by selective pressures. It's not like selection takes a nap for a while, and then gets back to work while drift, bias et al. chill out. There is no point to dismiss one or the other, like so many tend to do.

Curiously, I've heard numerous times that "Well, maybe selection is less important for bacteria, but it is the dominant force in vertebrates". How hilarious is it that any population geneticist will tell you the exact opposite: bacteria are under overwhelming selective pressure due to their freaking massive effective pop sizes, meaning that drift is much less effective in that situation relative to selection. Vertebrates are actually an awesome example of selection going rather easy: being a large multicellular thing with a backbone is a damn stupid way to copy your genes. Seriously!

Now, you may wonder why should anyone who's not an evolutionary biologist care about any of this? For a cell or developmental biologist, why not assume everything is there for a reason?

Because this leads to rather convoluted explanations for things. Take signalling pathways, for example. Is there any particular reason you have tens of genes required to turn on a particular behaviour in the end when you could've 'designed' it instead to use only one or two? Here we have again a problem with the adaptationist approach: you can pretty much always think of some reason why something is 'useful' or adaptive. That doesn't make it right. What if some features of these pathways originally evolved as a form of adaptively-neutral 'bloating' of the system? See Lynch 2007 PNAS, Lynch 2007 Nature Rev Genet for more on non-adaptive processes in evolution of genetic pathways. (Or just stop reading this post and go through this short list instead ;-)).

"Population genetics ignores reality!"
Now, there's some complaints that popgen kind of fails at taking reality into account. For some work in the field, that is true -- just like in any other field. Did you seriously think everyone, to the last moron, is in touch with reality in your field? If so, I'd love to hear! (no philosophers need apply, heh... although to be fair, there are some 'fringe lunatics' there who actually make sense by our standards.) That said, mathematical modeling requires simplifications to be made to get somewhere. If you have a problem with that, note all those humanities scholars who are sneering at us because we make simplifications as scientists! They wisely take the easy way out and conclude that reality is a social construct and understanding is actually impossible and thus not worthy losing sleep over...

Good mathematical biologists note their simplifications, keep track of them, and know when to simplify what, and what the limitations of their models are. Even better mathematical biologists test their models empirically, thereby producing work that is truly relevant to the rest of us. In my [admittedly still quite inexperienced] opinion, Michael Lynch belongs to that category. Seriously, I despised and dismissed the entire field of population genetics as well, and thanks to some of his papers realised it's probably not a very good idea to do that. In a field as messy as biology, any tidbit of information, even if it comes from simulations, is not only valuable, but essential to the very hope of sorting stuff out. We, that is -- all biologists -- are in no position to discard entire fields because of our petty tribalistic snobbery. Tread with caution -- yes. Dismiss without a second thought -- absolutely not.

Existence of neutrality and near-neutrality
To touch on the existence of 'true' neutrality, let me show you a few diagrams. The first one outlines the history of selectionist and neutral theories:

Story of neutralism and selectionism. (Bernardi 2007 PNAS, OA) Similar diagram and accompanying story can also be found in Ohta 2002 PNAS

You can also see Tomoko Ohta's big review here: Ohta 1992 Annu Rev Ecol Syst (free access).

The major breakthrough here is not even so much the Neutral Theory (which was more of a wake-up call to the stagnating ultra-selectionists), but the Nearly Neutral Theory. Accepting near-neutrality enables one to deal with truly messy and ambiguous situations that don't fit so neatly into the advantageous-deleterious dichotomy. Furthermore, it allows a full spectrum, encorporating the fact that some deleterious mutations are worse than others, and vice versa for the advantageous ones. The 'strict-ness' of selection is further affected by additional factors, such as effective population size.

In short, selection acts probabilistically, not absolutely:

The probability of selection-dependent fixation of an allele vs. product of effective population size(N_e) and selection coefficient(s) of the allele. Note that holding s constant and increasing N_e shifts the selection-dependent fixation probability away from 1 (neutral). Thus, selection is probabilistic; there is also a time dependency here: the longer you run a given situation, the greater the chance that even the slightest deviation from neutrality will be acted on by selection. Given either infinite time or infinite population size, we would see a sharp cut off between advantageous and deleterious alleles. Neutrality would disappear. However, given that no biological system persists for infinite time or has an infinite population size, what one gets is a zone of effectively neutral (Nearly Neutral) mutations; again, the width of this 'zone' is inversely proportional to N_e Disclaimer: I am not well-versed in popgen by any means, please correct me where I'm wrong! (Yi 2006 BioEssays)

The further a mutation deviates in either direction from the true neutral (here assumed to be the same as previous state of the allele, in the context of mutations), the more likely selection is to do something about it, wither selecting against the allele in question, or selecting against those who lack the allele in question (aka "positive selection"). Crudely put, the larger the effective population size, the steeper the selection curve (ie slightly deleterious now becomes quite deleterious, etc).

This actually means that according to pop size alone, selection is far stricter in bacteria than it is in vertebrates. This makes sense: bacteria exhibit a lot less 'design stupidity', if you will, than we do. Again, creating a multicellular organism with a ridiculously long and expensive generation span and a pathetically low reproduction rate is about the stupidest way to perpetrate a handful of genes. But it is tolerated -- in large part because there is a niche for it beyond the reach of more efficient organisms, and to some extent because the effective population size there is so small as to allow the system to drift towards foolish complexity.

Want concrete examples? Genome complexity is a really nice one; I won't discuss details here, but see Yi 2006, Lynch 2007, Stoltzfus 1999, and more Lynch if interested.

To make take an analogy from ecology, a similar argument can be made for the existence of commensalism. Yes, strictly speaking, if you measure all effects and interactions to the finest detail, there are no commensal relationships -- even the slighest extra drag produced by free-riding fish on the manta ray (provided they don't do anything beneficial to it, too lazy to research) would be harmful thereby rendering the relationship parasitic. However, just as in the case of selection, there lies a fuzzy line between parasitism and mutualism, in many cases fuzzy enough to contain 'nearly-parasitic' and 'nearly-commensal' interactions. In that case, the relationship would not be harmful or beneficial enough to really matter within a finite (and turbulent) timeframe. Thus, it is probably more useful to use 'commensal' (well, nearly-commensal) than have to always scratch your head over whether the net total of a relationship is mutualistic or parasitic.

Now that we are hopefully at least considering the possibility of neutral mutations, why should anyone care? Aren't pure selectionists just a strawman anyway? Besides, aren't neutral mutations incapable of really doing anything useful or noticeable? And what the hell is constructive neutral evolution anyway? Stay tuned for this and more in Part II!

References and further reading:
Bernardi, G. (2007). The neoselectionist theory of genome evolution Proceedings of the National Academy of Sciences, 104 (20), 8385-8390 DOI: 10.1073/pnas.0701652104

Gregory, T., & Ellis, C. (2009). Conceptions of Evolution among Science Graduate Students BioScience, 59 (9), 792-799 DOI: 10.1525/bio.2009.59.9.11

Lynch, M. (2007). Colloquium Papers: The frailty of adaptive hypotheses for the origins of organismal complexity Proceedings of the National Academy of Sciences, 104 (suppl_1), 8597-8604 DOI: 10.1073/pnas.0702207104

Lynch, M. (2007). The evolution of genetic networks by non-adaptive processes Nature Reviews Genetics, 8 (10), 803-813 DOI: 10.1038/nrg2192

Ohta, T. (1992). The Nearly Neutral Theory of Molecular Evolution Annual Review of Ecology and Systematics, 23 (1), 263-286 DOI: 10.1146/annurev.ecolsys.23.1.263

Ohta, T. (2002). Inaugural Article: Near-neutrality in evolution of genes and gene regulation Proceedings of the National Academy of Sciences, 99 (25), 16134-16137 DOI: 10.1073/pnas.252626899

Stoltzfus A (1999). On the possibility of constructive neutral evolution. Journal of molecular evolution, 49 (2), 169-81 PMID: 10441669

Yi, S. (2006). Non-adaptive evolution of genome complexity BioEssays, 28 (10), 979-982 DOI: 10.1002/bies.20478