Reader Comments

Post a new comment on this article

How is a genotypic bias non-adaptive?

Posted by Bolton1 on 27 Apr 2013 at 05:48 GMT

This is an interesting article that may open up a whole new debate about evolvability. However, I would like to get some clarification on the subject of selection versus a non-adaptive increase in evolvability. I can certainly see that the first explanation (an unbiased drifting process) is completely non-adaptive. But the explanation that postulates a founder effect for the discovery of new niches, resulting in genotypic biases towards greater evolvability, is no less adaptive than the original selection based explanation. In fact, as far as I can see, it is one and the same. If one is going to postulate that evolvability has an adaptive value, it is surely with reference to the greater access to new niches. Selection at the level of lineages means that those with greater evolvability are likely to find larger numbers of new niches and therefore be more competitive than lineages with much less evolvability. When phrased from this perspective how is the "non-adaptive" but biased drifting process anything different to the explanation I just gave. The very use of the word bias surely implies selection and that greater evolvability is, in this case, adaptive rather than non-adaptive. In fact, there is not much that is passive or drifting about a process in which entire lineages are selected or biased over others on the basis of non-random founder effects.

I think there is a fundamental misunderstanding in this paper that is best highlighted by the final sentence of the conclusion. "In this way the story of biological evolution may be more fundamentally about an accelerating drive towards diversity than competition over limited resources." Over the course of large scale evolution one especially competitive evolutionary strategy is the rapid evolutionary drive towards a diverse range of niches. This is obviously one way to hog all the limited resources. Again, the use of the expression "accelerating drive" indicates some obviously adaptive advantage to greater evolvability, which there obviously is. The authors, in my opinion, have therefore erected a false dichotomy to push their argument across. As far as I can tell, there is no real difference between a genotypic bias towards evolvability and the greater adaptability of lineages with greater evolvability.

No competing interests declared.

RE: How is a genotypic bias non-adaptive?

jal278 replied to Bolton1 on 28 Apr 2013 at 08:43 GMT

This response is from the authors. We appreciate the sentiment behind the comment, which highlights the need for caution when applying the words selection and adaptation. Before we address the main criticism, it is first important to emphasize the equal importance of the pure passive drift model, which has no selection pressure whatsoever (as the commenter acknowledges). Even this non-selective model still separates more evolvable organisms from less evolvable. While the article overall makes a theoretical point about how evolvability can originate, at heart it serves as a cautionary note on the interpretation of nature. In that context, the pure passive drift model on its own serves to question whether an overall average increase in evolvability actually occurs in nature anyway (at least within particular evolutionary branches). That is, the apparent increase in evolvability in some branches of evolution (which is observable across niches) could indeed theoretically result without any form of selection pressure. For example, it is possible that early in evolution, when resources were for some periods effectively unlimited, such nonselective evolution might have been seen. So even if the commenter disagrees with how the authors characterize the niched models, there is no disagreement over the interesting result that a pure drifting process can nonetheless produce an increasing average evolvability across niches (as opposed to across individual organisms). Because niches are usually the focus of discussions of evolvability, this insight alone raises valid questions on the appropriateness of selective explanations in particular cases.

Of course, the main concern raised by the commenter is the possibility of a false dichotomy. However, we think there is a useful conceptual distinction between an “accelerating drive towards diversity” on the one hand and “competition over limited resources” on the other. This conceptual distinction would be difficult even to discuss if we are forced to equate both characterizations as the same thing (and as the commenter notes, our article does aim to provoke a new discussion). In particular, the “competition over limited resources” characterization of evolution offers a different kind of explanatory tool than the “drive towards diversity.” For example, if you ask how or why a particular trait evolved (as biologists often do) the explanation might be that it afforded a certain functional competitive advantage (such as faster ambulation or better sensation of the outside world), which fits with the competitive perspective of selection as a big competition among organisms. In contrast, the “drive towards diversity” explanation does not allow one to defer to any explicit functional advantage. In this way, it suggests a significantly more subtle story of origination. In a sense, it is more about an “escape from competition” (by founding a new niche, where functional performance may not even be measured on the same dimensions) than about competing at all. That is why it is helpful to make this distinction and to be able to discuss it in clear terms.

It is also important to recall that although selection is present in the fixed-capacity model, it acts only randomly within each niche. That is, there is no way for any individual to be considered objectively better than another in the model, and thus it is difficult to argue that adaptation with a niche is occurring or is even possible. Furthermore, while the commenter states that "if one is going to postulate that evolvability has an adaptive value, it is surely with reference to the greater access to new niches," that is by no means a consensus statement. In fact, several other viable theories for evolvability do not depend upon such an idea (Wagner and Altenberg 1996, Kashtan et al. 2007, Clune et al. 2013), which is one reason this new perspective can help to provoke useful discussion. In fact, the assumption that access to new niches is an “adaptive” trait faces the further problem that it does not specify the mechanism by which evolution selects for supposed greater "adaptive value," which is exactly the paradox that any viable theory for increasing evolvability must explain: Evolvability may "benefit" evolution in the long run by accelerating the rate at which forms diversify, yet selection always acts narrowly without foresight. That is, just because something might increase evolutionary potential does not mean that it provides immediate competitive advantage. Thus the results in the paper with a non-competitive niched model are interesting because they suggest that founder effects (isolated from pressure to out-adapt or out-compete other organisms) are sufficient to accelerate the trend seen in the passive drift models. The insight is that competition, often assumed to be a crucial factor for evolvability to increase, may surprisingly not be a necessary one.

We thank the commenter for taking the time to thoughtfully argue an important point and for providing us the opportunity to clarify some insights.

No competing interests declared.

RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to jal278 on 29 Apr 2013 at 05:03 GMT

I certainly accept the pure passive drift model as a valuable and interesting contribution of this paper and so I take no issue with the first paragraph. This disagreement comes down to the scale of selection that is relevant. If the scale of selection considered is restricted to individual organisms over relatively limited time frames then there is simply no way that evolvability can be considered adaptive by anyone – the case being made by the authors is a tautology. This is because the case for evolvability being or not being adaptive mainly rests on selection for evolvability acting on broader agents, such as species. This is obviously because the rapid evolvability of a species does not provide an advantage to an individual (individuals do not evolve), and can therefore only be selected for over many generations.

This being the case, you cannot treat evolvability the same as a physiological or morphological trait. The functional benefits of most traits apply directly to individuals within species, meaning that selection acts on individuals and therefore also acts over relatively short periods of time. But for lineages evolving over long periods of time the distinction between the accelerating drive towards diversity and competition for resources is lost. By this I mean that the accelerating drive towards diversity (where it occurs) is one way for lineages to compete for a wide range of different resources over very large time scales (noting again that individuals are completely irrelevant when making a case for or against evolvability being adaptive). There is therefore a strong case for saying that differences in the rate of diversification of lineages (i.e. species level selection) will lead one lineage to dominate or replace others, in the same way that species outcompete and replace other species through higher rates of fecundity and survival. At this scale “the drive towards diversity” explanation clearly can have an explicitly functional advantage to the lineage. The ability to colonize new niche space is therefore easy to reconcile with a functional advantage when considering broader scales of selection (and a broad agent of selection is the only one that should be considered when dealing with evolvability).

To make a clear case for evolvability being something that is non-adaptive one should avoid the tautology that higher evolvability is not advantageous or adaptive for individuals within a species. Any case against evolvability being adaptive needs to pertain to the relevant scale of selection (i.e. not individuals). To make a case that there is a bias for evolvability (as the authors have done) appears to instead be making a case for evolvability being adaptive for lineages via selection at the level of species; greater evolvability resulting in a broad distribution of species across niche space (therefore reducing competition or overlapping niche-space within the same lineage), and a high proportion of species occupying novel competition-free niches.

Other than the above, I like this paper quite a bit.

No competing interests declared.

RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to Bolton1 on 29 Apr 2013 at 06:44 GMT

I should amend what I said with regard to the functional advantage of evolvability. Clearly, evolvability is not actually functional in the physiological sense and therefore it has no actual functional advantage, but it can, at least in theory, be selected for and therefore it can be considered adaptive. To argue that evolvability is non-adaptive while at the same time saying it is distinct from other traits in not being functional also seems like a tautology if a physiological or morphological function is believed necessary to qualify as adaptive. Perhaps the authors were not stating that functional advantages were needed for biological characters to be adaptive but I thought I should add this to help clarify my position.

No competing interests declared.

RE: RE: RE: How is a genotypic bias non-adaptive?

jal278 replied to Bolton1 on 01 May 2013 at 18:01 GMT

This is the authors again. It is important for us to acknowledge that while the passive drift model includes no selection pressure, the limited-capacity model does indeed produce selection pressure. So (for other readers) we are not disagreeing with the commenter about whether selection pressure plays a role in that variant, and the commenter is right to highlight that selection pressure plays a role in that case. The disagreement is on whether that particular kind of pressure should be characterized as “adaptive” in the traditional sense, which often relates to competition between individuals to increase their fitness to the current environment (Fisher 1930, Williams 1966, Brandon 1978), although there are other possible conceptions.

Perhaps it would help to explain why we avoid the term adaptive in this case to look at an example of how evolvability can be more explicitly adaptive: In 2005 Kashtan and Alon published a paper in PNAS that showed that if the environment is varied over the course of simulated evolution then the genotype becomes more evolvable in the sense that it can change faster over generations to match one of the environmental variants. (Earl and Deem made a similar point in 2004 in their PNAS article, “Evolvability is a Selectable Trait,” which focuses on environmental variation of a different type.) This kind of evolvability is more in the spirit of the traditional “competition” view of evolution, wherein the genotype is becoming more flexible to anticipate the variation that can happen in the environment based on historical precedent. In this context, wherein the environment varies in predictable ways, evolvability is adaptive. You could say that selection pressure for evolvability in this case is direct in the sense that the capability to adjust to environmental variation is selected explicitly.

Our paper states that in the limited-capacity model, there is “no direct selection pressure for evolvability,” which reflects that individuals gain access to new niches (which gives them a temporary indirect selective advantage) without ever being tested for their ability to deal with such variation (or with any particular environmental conditions whatsoever). So our distinction between adaptive vs. non-adaptive hinged on the direct vs. indirect selection for evolvability. In effect, the environment does not explicitly vary and there is therefore no way for evolution to anticipate what might be needed in the future (so it cannot “adapt” to such varying conditions). We grant to the commenter that this point is subtle and that the decision to tie the word “adapt” to this distinction may invite disagreement; as Lewontin said, "There is virtually universal disagreement among students of evolution as to the meaning of adaptation." But hopefully the commenter can see why our conception is not necessarily tautological given that there can be a kind of pressure towards evolvability that is explicitly adaptive in the varying-goals sense (that is, the argument does not imply that evolvability cannot be adaptive by definition).

However, what we think is ultimately important from both the passive drift and limited-capacity models is not the semantic issue of what is or is not called adaptive, but the fact that both of these (even the limited-capacity model) are unlike most typical arguments for the origins of evolvability, which often hinge on specific environmental challenges leading to competitive pressure (as in the example of a varying environment). The idea that limited-capacity niches alone is sufficient to accelerate evolvability is illuminating in this way, regardless of the words we choose to describe it. We hope the particular words used to make this distinction, although we think they are appropriate, do not become a distraction from this deeper point.

No competing interests declared.

RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to jal278 on 02 May 2013 at 04:05 GMT

Before I make my main point, the example the authors give by Earl and Deem cannot be an example of selection at the level of individuals. As I mentioned in my former response, selection for evolvability of any kind cannot entail direct selection of individuals, which can only be selected with regard to how well suited they are for a particular environment. So in making their point through the example of Earl and Deem I hope the authors are aware they are now contradicting a large proportion of what they have said, including the previous paragraph (although they did confess to the existence of alternative conceptions of adaptation, which they have now opted for). But I am glad they have come round to the point of view that accepts adaptation and selection occurs for lineages and species (i.e. that more evolvable genotypes require many generations to be selected for). Certainly, the point the authors are now making skillfully avoids the tautology I highlighted earlier. But the point of highlighting that tautology was to show the redundancy of their former position. Their position is now different. Somewhat fitting given that this discussion is all about adaptation. (I should point out that it is fine for somebody to change their mind about something on further reflection.)

OK, now for the main argument. The problem is that the case provided by the paper of the authors, as interesting as it is, is really no different to Earl and Deem's case except it is dealing with adaptation to spatial heterogeneity as opposed to adaptation to temporal heterogeneity. Spatial heterogeneity in resources allows organisms within the same lineage to diverge and exploit more than one resource type. This is not necessarily the case in temporal heterogeneity, in which a lineage could be forced to go from one type of resource to another without the possibility of divergent specialization. Let's say the authors ran another experiment in which there were very few different niches, let's say just two or even one possible niche. Obviously when there is only one niche there is no need for adaptation so adaptation does not occur at all. Consequently, evolvability will not increase - in fact, the concept of evolvability becomes completely redundant. But this point serves a purpose by highlighting an extreme end of a spectrum. Different levels of spatial heterogeneity should select for different amounts of evolvability (at least I would be amazed if they did not).

For example, marine environments tend to have relatively low amounts of spatial heterogeneity compared to land (coral reefs aside). Consequently, organisms that make the move to land should become more evolvable over many generations. In this case, greater evolvability represents a good example of a lineage's adaptation to a terrestrial setting. This may partly explain why we often find so much diversity and species richness on land compared to the ocean (just compare insects and crustaceans), although I admit that this diversity may just come down to the heterogeneity itself. Likewise, if you altered the numbers of niches in the authors' experiment it is very difficult to envisage an outcome in which the degree of evolvability was not to some extent correlated with the number of different niches (in this case using differences in the scale and complexity of the maze environment in which the robots ran - a very simple maize should result in a more gradual increase in evolvability and a lower final evolvability).

So whereas Earl and Deem reference the adaptation of a lineage to a temporally variable environment, the authors provide an example of adaption to a spatially variable environment. So there really is no fundamental difference here in terms of qualifying as adaptive, as the authors have tried to suggest.

No competing interests declared.

RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to Bolton1 on 02 May 2013 at 04:47 GMT

Sorry, I was referring to Kashtan and Alon regarding the temporal heterogeneity, and not Earl and Deem.

No competing interests declared.

RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to Bolton1 on 02 May 2013 at 04:50 GMT

OK, I guess all of them apply.

No competing interests declared.

RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

kenstanley001 replied to Bolton1 on 02 May 2013 at 06:29 GMT

Authors here again: Note that the citation of Earl and Deem was just an aside - the previous argument was based on the Kashtan and Alon example. In any case, the commenter's latest point hinges on the idea that selection for "spatial" heterogeneity is effectively equivalent to "temporal" heterogeneity. Note that it may be a stretch to call the diversity in the robot scenarios simply "spatial" given that it is actually based on a trajectory taken by a robot over time that leads to a final point in space. But more importantly, this argument about spatial versus temporal misses the relevant distinction, which is:

The ability to succeed in a situation that was previously encountered in the past (i.e. the "temporal" scenario) is intrinsically different from the ability to do something fundamentally novel in the future.

The former ability is naturally called "adaptation": an individual or a group is becoming better suited to a set of historical conditions through repeated exposure. The latter is much less clearly labeled adaptation because it involves the idea of becoming better at doing something one has never done before. If tomorrow you discover for the first time to your surprise that you can stand on your head brilliantly, it would be an unusual choice of words to say, "I adapted to standing on my head," because that would imply you had been working on it over time and became better as a result. It would be similarly strange to say, “I adapted to doing things that I didn’t know I could do.”

No competing interests declared.

RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to kenstanley001 on 02 May 2013 at 10:05 GMT

The authors appear to have misunderstood my argument about spatial heterogeneity. I simply mean that there is variation in space that is obviously relevant to niche divergence, and that more than one niche exists at the same time (as a result of spatial variation). The evolvability they model could be thought of as an adaptation to this situation as opposed to the cases of evolvability being an adaptation to temporal variation they mention with regard to previous work, whereby a single resource and niche is changing from one form to another without any actual divergence. So selection for evolvability can arise through divergence (lineages branching into multiple niches at the same time) or non-divergence (multiple niches in succession but only one niche at a time due to the absence of branching). The difference here is not so fundamental that the authors are able to justify the characterization of one as adaptive and the other as something else. As divergent niches can also arise without spatial variation it might be better to contrast the two in terms of divergent versus non-divergent or linear adaptation. The spatial example is easier for me (and I suspect others) to visualize (it is the first thing that came to mind), but it does not really get the concept across as precisely as it should so I apologize for the confusion.

In reference to the last couple of paragraphs it is clear there is some confusion on the part of the authors. They are referring to the novel characters or niches and that the shift to any particular niche is not adaptive. I take it they mean the adaptations to the particular niches in the experiment. Shifts to new niches and the origination of novel characters often occur through selection and adaptation. This cannot be a point that the authors now wish to seriously contend, surely (that novelty is never adaptive). But the issue is not adaptation to the particular niche in the experiments. The question we were debating is one of whether, specifically, your example of evolvability is or is not adaptive, an entirely different issue.

Message to authors: There is nothing wrong with graciously accepting you made a fairly fundamental error. You would genuinely get my respect. I don't want this to be painful for anybody and I think it is becoming awkward and painful for all of us now.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

kenstanley001 replied to Bolton1 on 03 May 2013 at 04:39 GMT

Authors again: Note that because there is more than one way to increase evolvability to adapt to varying environments, nothing in our perspective precludes divergent adaptation in this sense either: evolvability can adapt to varying environments divergently or non-divergently. Nevertheless, the limited-capacity scenario in the article does not match either case, so the commenter is alleging the wrong dichotomy.

Of course no one disagrees that “shifts to new niches and the origination of novel characters often occur through selection and adaptation.” The question is simply whether the word “always” should replace “often” in the sentence (which the commenter would need to do to maintain a consistent objection). However, as the commenter notes, the real question here is whether the example of evolvability in the limited-capacity models should be called adaptive. No further arguments were offered on this question beyond divergence vs. non-divergence.

It is important for the commenter to recognize we are talking about a definition here – the definition of the word “adaptive.” To quote Lewontin again, "There is virtually universal disagreement among students of evolution as to the meaning of adaptation." Thus the suggestion of “graciously accepting you made a fairly fundamental error” is a bit grandiose. It is entirely healthy to be discussing how we should use the word “adaptive.” In fact, the tautology appears now to be in the commenter's court: It seems to be that the commenter is ultimately asserting that selection is always adaptive by definition (notice “always” again). Of course, in that case, by definition any increase in evolvability due to selection would be adaptive, so there would be nothing here to discuss.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to kenstanley001 on 03 May 2013 at 12:55 GMT

I think the authors will find, on a perusal of the literature on what qualifies as adaptive, clear directional selection for an attribute (in this case evolvability) almost invariably qualifies as characterizing that attribute as adaptive. I can think of no cases (there may be some strange exceptions I am unaware of) where this is not thought of as adaptive. However, I am not saying that selection is always adaptive. Stabilizing selection is obviously an example of non-adaptive selection. Therefore the claim, made by the authors, that I am pushing a tautology is entirely invalid. The issue of determining if something is adaptive is often with regard to whether it is possible to find evidence of direct selection for something. As this is often difficult to do it is seldom clear if an attribute is adaptive. This is where contention normally lies. Lewontin was referring to evolutionary "spandrels" and the way that many attributes should not be seen as adaptive because there is no direct selection for them. To use Lowontin's words to bolster their own case is somewhat disingenuous when I think none of the details of Lewontin's arguments could be used to support the claim of the authors.

I think that the case of evolvability highlighted by the authors is fairly unambiguous in how it should be interpreted. To argue this is just down to definition is not the point. The authors have practically distorted the definition of the word "adaptive" beyond all recognition in their desperation to show they are right. The authors got this paper into PLOS ONE under the guise of showing evolvability increases without being adaptive. But rather than own up to this error, over the course of this debate the authors have gone to many lengths to pretend that adaptation does not apply when I think almost every evolutionary biologist would argue that adaptation does apply here. Please someone weigh in here if I am wrong.

The authors initially asserted that adaptation did not apply here because adaption should only apply at the level of individual selection (not lineage or species based). That is a fairly narrow definition. But if they are going to make that case, it is a clear tautology to talk about evolvability being non-adaptive. When I pointed this out they broadened their definition accordingly and then tried to make a case that it was non-adaptive by saying it was different to adaptation to an environment that varies over time. I then pointed out that their computer simulated environment could be considered an example of how evolvability is similarly adaptive to spatially varying and complex environments. If temporally varying environments lead the authors to suggest that evolvability is adaptive then why not complex and spatially variable environments as well?

They have now tried to suggest this does not apply here without giving a reason. The limited capacity scenario is somehow suppose to not pertain to the argument I made. I cannot see this. The authors have generated a computer simulated environment in which robots reproduce. As the robots reproduce they evolve in different directions, allowing them to move to different niches. As an ancestral robot can have many different descendent lineages (some of which move into one niche and some of which will move into another) I fail to see how they can make a case that divergence does not apply here. I know that robots cannot adapt to individual niches (or at least that there is no bias for the individual robots), but evolvability is adaptive for lineages and not robots. In the Kashtan and Alon case (referred to by the authors as an example of evolvability being adaptive) the change of niche (in this case niche only meaning the ecological requirements) is only occurring over time for lineages so that there is selection for lineages that shift niche more rapidly. In the computer simulated model the selection is for lineages that move out into multiple niches more rapidly rather than those that make a complete shift of niche. That is the only distinction. Both involve selection of lineages.

So if the authors are happy to accept that the pace at which a lineage can relocate niche space can be adaptive then why is the pace at which a lineage is able to spread out into niche space not also adaptive.

The authors have now clearly lost their case. They have not addressed the issue I raised and they have shifted definitions in order to avoid conceding a fundamental error in their claim. They must realize by now that all of their squirming has only made their case worse.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

kenstanley001 replied to Bolton1 on 04 May 2013 at 00:27 GMT

Authors again: The commenter wishes to end the discussion just as it begins to address the heart of the matter. It turned out helpful to push the commenter to reveal what in the commenter’s view makes selection adaptive. As the commenter helpfully states:

“The issue of determining if something is adaptive is often with regard to whether it is possible to find evidence of direct selection for something.”

While the commenter is now suggesting an internal inconsistency to our position, at least we agree now on the litmus test – and in fact it turns out exactly aligned with what we said in the article: there is “no direct selection pressure for evolvability” (quoting the article) in the limited-capacity models.

As the commenter now also states, “… it is seldom clear if an attribute is adaptive.” (note the instrumental word “clear”)

So we now know what we are actually disagreeing about. It is often the case in discussions of subtle and complex phenomena that it takes a little while for the participants to understand each other's positions, and there is no harm in that. The commenter is in effect asserting that selection for evolvabilty in the limited-capacity models is clearly “direct” while our view is that it is indirect.

In fact this tension aligns well with the quote from Lewontin because that is exactly the issue in contention here – again as the commenter says, “Lewontin was referring to evolutionary "spandrels" and the way that many attributes should not be seen as adaptive because there is no direct selection for them.”

So what we need to address here is whether there is (1) clear (2) direct (3) selection for evolvability in the limited capacity model. We both agree there is selection, so there is no disagreement there, so the issue is whether it is clearly direct. However, there are a number of reasonable arguments for taking the position that it is not clearly direct. These are again consistent with what we have been saying:

1) It is not “clear” in general that limiting the capacity of niches is directly selecting for evolvability. It is conceivable instead that it could lead to stagnation because niches are unable in this scenario to evolve internally. Or it could lead to stagnation because rewarding diversity alone early in evolution might lead to unstable genomes that produce nonfunctional behaviors. Or, even if it does not lead to stagnation, it could conceivably lead to increasing diversity without causing evolvability to increase. After all, if you limit the capacity of niches and wait long enough, of course more niches will eventually be filled, but that does not guarantee that anyone in those niches is necessarily more evolvable. So the clarity (i.e. is it “clear”) of the selection pressure is certainly in contention.

2) Selection is not “direct” in limited-capacity scenarios because the fact that a particular genetic tendency leads to more niches in the past does not ensure that the same genetic tendency will lead to more niches in the future. In other words, the proclivities that allow a species to diverge in one case are not necessarily correlated to those that lead to divergence in another. The fact that we observe consistent acceleration of evolvability in a limited-capacity model like the robots is an a posteriori observation of those particular models that reveals that such a correlation happened to exist to a sufficient degree in those models, but there is no a priori reason to expect in any given domain that such a correlation would always exist. That is what separates the “varying environments” examples from our limited-capacity scenario: in varying environments, you would *a priori* expect a correlation between (a) the genetic proclivities that allow a species to switch from environment A to B in the past and (b) the proclivities that would allow its descendants to switch from environment A to B in the future.

Certainly the commenter may disagree with these points and insist e.g. that it is nevertheless entirely “clear” from the commenter’s perspective that the selection is directly for evolvability in the limited-capacity case despite these obfuscating factors, or that even without an obvious correlation between past divergence and future divergence we can still say evolution is “direct.” But at least we can see that the clarity and directness in this case is reasonably contended.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to kenstanley001 on 04 May 2013 at 03:17 GMT

Indirect selection means selection via another attribute. The authors' lineages are being directly selected. Those with greater evolvability are at an advantage. There is no indirect mechanism by which they could be selected for within the parameters of the simulated environment (note that individuals are not under selection). I would advise the author who is making these more recent comments to study evolutionary biology more seriously than they have.

Indirect selection is not selection at another hierarchy than individuals. This appears to be the case that is being made. Indirect selection is selection of one attribute via another attribute that is being directly selected. In other words, something has to be directly selected for the other thing to be indirectly selected. In this case it is the lineages that are being directly selected due to greater niche spread.

Because this is a clear case of direct selection for evolvability, it is also a clear case that in the specific case of the computer program (I would not dare assume this stuff applies to reality), evolvability is adaptive. Simple as that. There is nothing more to this debate.

I would request that the author refrain from the gibberish that makes up points 1 and 2. They are fooling nobody. He seems to think he can make a case here by spilling out whatever nonsense comes to mind. This has been all too common lately.

I wish the authors had come clean when I first challenged them. They obviously meant that there is still selection for evolvability when you take adaptive competition out of the mix at the level of individual organisms (competition for each niche was neutral). This is not an uninteresting result. One tends to think of evolvability being useful because it allows individuals to out-evolve one another. In this case this is clearly not happening. The authors should have faced up to the fact that there is still selection at the species or lineage level and that, therefore, evolvability is still adaptive. By trying to disingenuously wriggle out of this error they detracted from this interesting result and made themselves look foolish.

I hope this will be a valuable lesson for them.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to Bolton1 on 04 May 2013 at 03:26 GMT

Quick correction. I said "One tends to think of evolvability being useful because it allows individuals to out-evolve one another". I meant to say "we tend to think evolvability is adaptive because it allows lineages to out-evolve one another [individuals cannot evolve, as I said many posts ago] with respect to a evolving in particular directions or towards particular niches".

There you go, I messed up. It really isn't that difficult a thing to admit to is it?

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

kenstanley001 replied to Bolton1 on 04 May 2013 at 16:03 GMT

Note that we (the authors) speak together and agree upon our responses before posting them.

Although we are arguing here a semantic point about the word “adaptive” at the level of evolvability in the limited capacity model, we wholeheartedly agree with the commenter that a more interesting point is indeed with respect to the lack of competition among individuals within niches. Thank you for emphasizing this point. That is an excellent reminder for the reader of one of the most important aspects of the results. We can see the commenter is intellectually honest and genuinely seeking truth.

Nevertheless, while we continue to comfortably disagree with commenter on the use of the word “adaptive” in this case and are tempted to argue further, unfortunately the commenter’s frustration has led to an increasing proportion of argumentation consisting of ad hominem attacks in place of reasoned arguments: Condescension like “gibberish,” suggestions to “study evolutionary biology more seriously” or “come clean,” and accusations of “disingenuousness, “spilling nonsense” or “looking foolish” are all unnecessary for the commenter to make a strong point, and such judgments could instead be safely left to the reader if the commenter is confident in the commenter’s own arguments. Note that we have refrained from any such condescension on our own part throughout this discussion and have not once tried to argue that our perspective is correct simply by fiat of being more knowledgeable or authoritative (which is a temptation for anyone with credentials). In this context, despite the opportunity to further strengthen our position and how much we have genuinely enjoyed this discussion, we are hesitant to continue to respond when the response to us is now predominantly condescension in place of reasoned arguments.

However, we bear no ill will towards the commenter and would be happy to talk further if the commenter can resist such expressions of frustration. We understand that any heated discussion can become frustrating so we do not take it personally. We are simply worried that the signal-to-noise ratio is reducing to a point where time invested in responding becomes less justified. The substance of the discussion has been illuminating and useful in any case and we are sure that if we ever meet the commenter in person someday it would be a pleasure.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to kenstanley001 on 05 May 2013 at 01:35 GMT

I apologize to the authors for taking a more aggressive approach than necessary. As the authors correctly pointed out, I am a tad frustrated by this dialogue. It appears to be going nowhere.

I think the remarks where I show my annoyance make up but a tiny proportion of my points, so I am not sure if it is correct for the authors to say that my responses are predominantly condescending.

I do hope - forgive me if I am being overly suspicious here - that this sensitivity to my slightly abrasive approach is not being used as a way of evading the points that I have made. I cannot help but feel that the authors are being deliberately evasive and that they have picked up on their error but will not admit it.

On the issue of adaptive being a semantic point, it plays a role in the title of the paper. By stating that evolvability is not adaptive the authors are challenging the established view on the reasons for differences in evolvability. It is therefore important. The authors have claimed that evolvability is adaptive with regard to anageneis, but then claimed it is not adaptive in relation to cladogenesis. They have failed to give a reason why they should differentiate the two. I see the former as adapting to environments that vary over time and the latter as adapting to complex and highly structured environments.

However, perhaps I have made my argument too much about adaptation. After all, over the course of this discussion the authors have gone to great pains to differentiate selection from adaptation in order that they can show that their results are consistent with their highly unusual and rapidly shifting definition of the term adaptation. I think they originally thought of selection and adaptation as synonymous, at least from the point of view of the paper. This is because they say "in this way, the emergence of a complex evolvable genotypic code and biological development may have been bootstrapped from far simpler reproductive processes by similar non-adaptive mechanisms. In other words, there may be no selective benefit for development or a complex genetic system, which may do no more than potentiate greater phenotypic possibilities".

Although they might push the case that selective benefit is a more all encompassing term and therefore not one and the same as adaptive, they appear to rest their case for the absence of adaptation being strongly linked to the absence of selective benefit. Why do this when you have such a restricted definition of what is required to be adaptive? If they are so different, perhaps preceding the second sentence with "in other words" is a bad idea. So I don't think they originally thought the two were that different. It just became convenient to distinguish them after being challenged.

But they accepted that selection pressure was going on in the limited-capacity model (without really acknowledging their error). Because I pushed the point of adaptation, and not selection pressure, they did their best to maintain that their interpretation was correct and that the disagreement was over generally recognized differences of opinion in relation to precise meanings of terms.

So when I challenged them on the use of the term "non-adaptive" they very conveniently changed the meaning of the term adaptive from one that closely matched their paper - pertaining to individuals and not lineages - to one that was quite different - also applying to lineages but under the circumstance of direct selection only. I should point out their definition of indirect selection was completely wrong. They stated "in the limited-capacity model, there is 'no direct selection pressure for evolvability,' which reflects that individuals gain access to new niches (which gives them a temporary indirect selective advantage) without ever being tested for their ability to deal with such variation (or with any particular environmental conditions whatsoever)." This is not indirect selection. Although the individuals may have the privilege of being part of a more evolvable lineage, to say that this means that evolvability is therefore being indirectly selected in individuals is wrong. Sure, evolvability physically exists in their DNA, but indirect selection has to pertain to some phenotype that can be indirectly selected via another phenotype. The authors mistakenly argued that the indirectness refers to the beneficial agent (the individual), i.e. that the individual is indirectly benefiting and that the individual is not able to express the phenotype. This is an entirely novel way of defining indirect selection. I suggest they come up with another term for this phenomenon as I am comfortable with the definition of indirect selection as it currently stands.

Their arguments now reveal they have either attempted to wriggle out of admission to their error or that they are just so misinformed and obstinate that they have actually bought into their entirely novel definitions of well established terms. Sorry if this charge is seen as abrasive and condescending. I do not wish to personally attack or hurt the authors. But they must surely accept that they were making flawed claims about their own data and then trying to avoid this thorny issue in a somewhat disingenuous way. I was full of sympathy early on, but that is now very difficult given the overall nature of their response. Again, I am sorry if I have hurt anybody. As the authors correctly point out, I am honest and I am sorry if I value this attribute so greatly that it leads me to anger when I observe its absence in others.

That said, I also bear no ill will towards the authors, and I do wish them the best of luck in the future if they wish to end this discussion here.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

kenstanley001 replied to Bolton1 on 06 May 2013 at 02:01 GMT

We (the authors) believe the commenter is starting to make an effort to avoid argumentation based on condescension, which gives us some hope that we can reengage in an enlightening discussion, but the commenter needs to recognize that the commenter continues to interleave technical argumentation with insinuations about our personal motivations (e.g. now suggesting we are "deliberately evasive," trying to "wriggle," lacking "honesty," and "obstinate"), which is a similarly unprofessional argumentation tactic. The commenter should understand that combining a technical argument with implications of intentional evasiveness and obstinacy on our part is a bullying debate tactic that makes it difficult for us to focus on the actual issue being discussed because the commenter is at the same time making it into an argument about our personal motivations. Making the argument personal forces us either to ignore the personal side of the argument and thereby seem to be conceding that there is something unseemly about our motivations, or to engage the commenter in an unprofessional and distracting debate about whose motivations are most pure.

Furthermore, continual entreaties to "graciously accept" the commenter's perspective have been a similarly manipulative and condescending tactic that creates the perception that if we do not agree right away with the commenter then there is something ungracious about our attitude, while at the same time the commenter remains conveniently free to attack us with no expectation of any reciprocal such “gracious” concession on the commenter’s part because we have not resorted to the same kind of bullying tactics.

The commenter may not understand that it is easy in any complicated argument to attribute "wriggling" or "obstinacy" to either party - in fact our perception is that the commenter has shifted positions during this discussion. However, we nevertheless have not succumbed to the temptation to maintain a side narrative about the commenter's obstinacy or personal motivations and instead have made an honest attempt to drive towards the commenter's core assumptions so that we can see where the fundamental disagreement actually lies (which becomes difficult when the commenter attempts to avoid engagement with pronouncements such as, “There is nothing more to this debate," that "It appears to be going nowhere," or calls reasoned arguments "gibberish" instead of responding to them with counterarguments). In any case, it is the presence of the personal side narrative and the fact that the commenter is so willing to publicly malign our personal motivations that still makes it difficult to continue the discussion with confidence that it will stay focused on scientific questions. If the commenter is able to recognize the importance of keeping personal imputations out of a permanent discussion thread attached to a publication in a respected scientific journal it may be possible to continue.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to kenstanley001 on 06 May 2013 at 19:04 GMT

I am sorry that the authors are so readily upset by what can only be an unambiguous reading of their comments. If they wish to avoid accusations of wriggling and evasion, I suggest that rather than blame the accuser, that instead they stop wriggling. This latest ploy to make this all about my abrasive line of enquiry (which is really not all that abrasive) only serves to highlight just how far they are willing to go to avoid confronting the fundamentally flawed arguments that they have made. It is shrewd and devious. It is largely designed to distract the reader from the very embarrassing part of this discussion. For I do now think that their new strategy is to cover up their errors by diluting this commentary with the subject of how they are being mistreated and how I am being unprofessional. This really is quite pathetic. Instead of attacking me for my abrasive line of enquiry and dissection of their arguments, I suggest that they try forming a case for their work. They do not need to convince me that they are right, only the reader. If my line of attack is misguided and too abrasive, the readers will judge me for that. So they can stop worrying about how bad a human being I am (it has no effect on the credibility of their arguments) and start actually making a case for themselves.

Of course I could feel justified in feeling attacked myself. They have accused me of being unprofessional and of using bullying tactics. Is this really true? Anyone who begins reading this commentary now (or who started a long way in), please go back over the entire piece and see if this is justified. But I would much rather be having a discussion about the issues (something the authors now seem incapable of) than worry about trivial aspects of personal conduct.

They have now accused me of shifting positions. But they have very conveniently refrained from stating their reason why. They state that this would be derogatory and bullying. No, please tell me. If I have shifted position I would love to know why. I am tough enough to handle it. Stating that I have shifted position and then not following through with an explanation of what they mean is absolutely cowardly. But this type of tactic does in no way surprise me now. It is pretty much in line with the rest of their commentary.

If I was angry in the past, I am livid now. To the authors: Make a case. Don't pretend you're being bullied. You're not. I am being frank. You, as ever, are being manipulative and devious. To the readers: this is not a personal attack. My interpretation of their comments are based on their comments alone (I do not know the authors and have no wish to make them look bad). The authors have themselves to blame for the charge I am making that they are being evasive and devious.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to Bolton1 on 06 May 2013 at 20:08 GMT

What I meant was, If I have shifted position I would love to know how [not why]. As I said, the failure to make this accusation without divulging their reasons is far more manipulative than pointing it out with reasons to back it up. And if they can't handle it when this plain fact is pointed out to them, I suggest they follow through with their threat to disengage from this debate. I very strongly suspect that will not happen. It was almost certainly an empty threat that was designed to manipulate myself and the readers. It actually almost worked for a moment.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to kenstanley001 on 06 May 2013 at 20:29 GMT

One more thing. Lets forget this third person nonsense as now is no longer the time for it. I am desperate to stay focused on the scientific questions. But if you wriggle and shift your position I can only be forced to call you out. The failure to point out the inconsistencies of your arguments and how they fail to align with previously made arguments would be doing the scientific questions a disfavor would it not? Likewise, when I am inconsistent I would much rather you be honest. The lack of honesty will only make it difficult for all of us to discover each other's core assumptions. Trying to hide behind a veil of sensitivity does nothing for this issue. It only makes you look as though you have something to hide from, which I now believe you do.

If you'll go back to the central question, so will I. When you shift position I will call you out and then you can tell me why I am wrong and make me look like the fool. But you have not been able to do that so far. Please get off your high horse and get back to the issue.

No competing interests declared.

RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: RE: How is a genotypic bias non-adaptive?

Bolton1 replied to Bolton1 on 06 May 2013 at 22:24 GMT

To give you a helping hand here, you need to explain why your limited capacity model (which I accept does not allow adaptation for individual niches) does not show evolvability is adaptive for lineages through the provision of faster niche spread (one could consider niche spread a divergent form of evolvability), i.e., why you don't think that rapidly evolving lineages are therefore better adapted to more complex and spatially variable environments, biomes, etc. (in which there are many potential niches). It is especially important that you clarify why you think this given that you believe that evolvability IS adaptive via faster niche relocation (this time a non-divergent form of evolvability), i.e. rapidly evolving lineages being better adapted to temporally varying environments, biomes, etc.

Please don't address this question from the point of view of the model being or not being real. I know it is a computer model and not real. You need to address why you think that your simulation shows that evolvability is non-adaptive.

That is the only question you need to address right now.

No competing interests declared.

Comment from PLOS ONE Staff

PLOS_ONE_Group replied to Bolton1 on 09 May 2013 at 23:35 GMT

PLOS ONE staff have made the decision to close this comment thread. PLOS values post-publication discussion of our published articles, but we have comment guidelines that were not followed by the commenter. Please see In particular, “Language that is insulting, inflammatory, or obscene will not be tolerated” and “Discussions should be confined to the demonstrable content of articles and should avoid speculation about the motivations or prejudices of authors”. We have agreed at the request of the authors to leave the existing comments in place, but any comments made in reply to this thread will be removed.

Competing interests declared: PLOS ONE Staff