martes, marzo 11, 2008

Quick change, then stasis, is not evidence for adaptive directional selection

Gasterosteus taken from Henry (2008) see ref

In scenarios that consider positive, directional selection as the main cause of adaptation, mutation is abundant, such that there always is available genetic diversity for selection to act upon. If on the contrary, adaptive mutations are rare, no matter how intense, selection is powerless to produce adaptation. In a classic scenario of directional selection, change accumulates steadily as genetic variation is not restrictive to the action of selection. Quite different is a scenario where evolution stops each time selection is done with the existing variation, left waiting for the advent of new mutations.

It is at this point that I wish to emphasize an important conceptual distinction that fails to be made when "positive selection" is reported from the sequence of a single gene. We are talking that each allegedly beneficial mutation lies within a single gene here; it is not a bunch of mutations spread out in different genes within the standing variation of a population, ready to be accumulated by selection. Rather, "positive selection" at the gene sequence level must by force obey an "episodic" scenario. Even if selection immediately imposes each beneficial mutation in the population, the change would be episodic and restricted to the occurrence of the next mutation. It is important to make this clear since this can be easily confused with the classical scenario of positive selection: continuous, and unrestricted by genetic variation. Sequences with "positive selection" don't provide any evidence of that.

Directional selection is also not "episodic" because it is supposed not to depend on more than a single episode of environmental change; that is, you have a new selective condition, and from then on, directional selection is supposed to produce the gradual betterment over time of adaptation to that selective condition. In theory it only requires abundant genetic variation and competition; evolution occurs, despite no further ongoing changes of the environment. However, short episodes of evolution followed by stasis, rather than continuous evolution, suggest that episodes of evolution may be restricted to events of environmental change.

Even within the neodarwinian synthesis, Simpson, their most notorious paleontologist, had attempted a proposal for a saltational mode of evolution, where stasis was interrupted by short periods of rapid evolution. Wright had also provided theoretical reasons for such an evolutionary pattern. The reasons of Simpson were quite understandable: This is the pattern that can most readily be discerned from the fossil record. However these views were pushed into the background pretty much until Gould and Eldredge re-established the fact decades later under the name of "punctuated equilibrium". Cases of gradual transformation of one species into another are rare in the fossil record, and so are the cases where we could suggest directional selection is producing adaptation.

I wish to thank Gustavo Ramos for bringing to my attention a commentary in Nature (Henry 2008) regarding this classic difficulty for detecting directional selection in the fossil record. Researchers with an explicit preference for darwinian explanations are showing frustration because currently available statistical tests for detecting directional selection never deliver positive results from the fossil record. The emblematic case in point is a population of sticklebacks in an environment with low predation, where exquisite fossil documentation and temporal resolution shows a trend across time to reduce the average number of spines in their armor, from about 1.4 to 0.9 (see figure below). When the tests were negative for directional selection, some felt that observable trend means the tests MUST be wrong. What they did was to change the requirements of the tests for detecting directional selection, now retrieving positive results for the stickleback population.

Says the article: "Previous tests have tended to treat directional selection as a reasonably consistent force driving average phenotypes in a given direction. This model is obviously unrealistic in the absence of any force expected to sustain selection in a particular direction over such long time frames. Instead, adaptation should often involve the asymptotic approach of phenotypes towards a particular optimum, near which the average should then remain until the optimum is perturbed. That is, environmental change should cause initially strong directional selection that should gradually grade into stabilizing selection, a ‘hybrid’ selection model if you will. This particular process is what would be expected for heavily armoured sticklebacks colonizing a lake where predatory fishes are rare, and the hybrid model provided an excellent fit to the fossil stickleback data"

Now wait a minute. Isn't this more or less what people like Simpson, Wright and Gould had in mind? An episode of quick change followed by stasis. Henry's comment has placed all the change at the moment of environmental change, and further envisions a different, negative and stabilizing selective pressure after this episode. This is hardly the scenario by which positive selection is suppose to produce adaptation.

Rather, this looks more like fairly typical change in conditions of negative selection, like the peppermoth: population averages change dramatically at the beggining, when the change in the environment is new; as time goes on, the population settles on the new average and stops changing. This is not "creative" selection. It's a shift in the populational frequencies of fishes with either 3, 2, 1 or 0 spines (more than 3 spines are never observed). No new morphologies are produced or even lost, despite the fact the average number of spines clearly moves into a new range.

Can we then confidently say that adaptive directional selection is responsible for what was on average the rapid loss of about half a spine, with stasis therafter? In my opinion, this pattern is suggestive of a shift in conditions of negative selection, specifically, a relaxation of negative selection. When transferred to an environment with less predators, since selective pressure is released, forms with 1 or 0 spines rapidly become more frequent and drag the average down to a new low. Before the environmental change there was little intergenerational oscillation in the average of spines, suggesting strong negative selection. After the episode of rapid change, and as a result of more relaxed negative selection, the averages became lower, AND there is also much greater intergenerational oscilation in average spine number (indeed, much LESS "stabilized" than before the environment changed!) . We go from fairly straight lines to a more wildly zig-zagging pattern. This to me suggests the relaxation of selection; quite according indeed with the shift to a lower presence of predators.
Hardly a compelling case of adaptive directional selection, no matter what the new tests may say.


Bell MA, Travis MP, and Blouw MD. 2006 Inferring natural selection in a fossil threespine stickleback. Paleobiology 32(4): 562–577

Henry AP 2008 Darwin in the fossils. Nature 451: 779-780

5 comentarios:

Roberto E. Yury Yáñez dijo...

Del mismo modo cuando en un trabajo reciente Hunt (2007)observa que al ajustar linajes de fósiles a tres modelos evolutivos cambio direccional, caminata al azar y estasis solo el 5% se ajustaba al primer modelo y el restante 95% se ajustaba equitativamente al resto (caminata al azar o estasis) sostiene que reafirma el equilibrio puntuado "demostrando" que el cambio direccional es de muy corta duración para ser registrado en el registro fósil.

Al final todo depende de como se quieran "leer" los datos.

PNAS 104 (47): 18404-18408

Sanders dijo...

la resolución temporal no es problema en el mencionado caso del lago de los sticklebacks...

Sanders dijo...

I must say I have becom completely convinced that this is a relaxation of selection, not directional selection. It only looks directional because there is an upper limit (constraint) to the possible number of spines (no more than three)

When predation is less intense, forms with less spines become more frequent, lowering the average in general, but also allowing for much greater oscillation (which shows no sign of narrowing out, quite plainly against the notion of "stabilizing selection" proposed for this period)

And that's it.

Chico dijo...

In my opinion what makes clear it is a case of negative selection is that the morphological diversification happened independently of any increase in competition. New phenotypes appeared when more phenotype became ecologically viable. And it’s what is observed in the fossil record. The episodes of bigger diversification occurred during periods of abundant resources ("adaptive radiations"). Competition generates stasis, not change.
Sander, I think it is a very important discussion and should be enlarged. There is so much confusion and misinterpretation about what is directional selection!

Sanders dijo...

At the level of the trait of spine number, no new condition evolved. Only their relative frequencies changed.
No new adaptation, and thus not at all the kind that is supposed to be responsible for shaping new adaptations, call it directional, positive, etc.)

However, at individual rather than at trait level, a new phenotype did arise, as the result of the combination of all the lower range conditions of all three armor-related traits: a new lineage of very low-armored fish.

I was supecting this could be the case; as forms with lower armor become more frequent, the probability of previously unseen combinations of low values increases.

So indeed in this case relaxation of selection allowed the establishment of a new very low-armored individual phenotype (despite no actual new traits showing up).

Another interesting thing is the pelvic score trait, which started changing 3500 years after spine number. One of their explanations is that it was selected for separately and later than for spines. The other explanation is that this required the Pitx1 mutation, which is recessive and would not have been frequent enough to offer any phenotypes to selection. I assume in this model that after some while drift must have finally surpassed the "threshold" frequency...