David Dobbs mucks up evolution, part II

If you’ve been following this site, you’ll know that yesterday I wrote a long critique of David Dobbs’s recent article in Aeon magazine, an article called “Die, selfish gene, die” (the subtitle is “The selfish gene is one of the most successful science metaphors ever invented. Unfortunately, it’s wrong”). Actually, there were a fair number of comments on that piece (127 up to now!), so I was pleased that readers took the time to digest a long science post.

Sadly, I couldn’t analyze Dobbs’s entire article in one go, both because of the “TL/DR” syndrome and because I simply didn’t have time: almost all these posts are written strictly between 6 and 8 a.m., after which I start my day job. So today I’ll finish my critique by discussing the second contention of Dobbs: that conventional natural selection, in which existing genetic variation is sorted out according to the gene copies’ ability to replicate, is wrong. As he said, “Die, selfish gene, die.”

I gather that Dobbs has backpedaled on his own website, but I am not going to read whatever clarification or retraction he issued until I’ve finished writing this. As I said yesterday, I want to discuss the original piece, not some subsequent gloss on it.  I’ll then have a look at what Dobbs wrote.  But he should know by now that, regardless of whatever encomiums other science writers (or scientists) gave him, his piece was a bad job. If he were an honest man, he would say that it was indeed full of misleading statements, and apologize for it.  But we know that won’t happen!

Today’s discussion is on what Dobbs and some of the heroes of his piece (especially Dr. Mary Jane West-Eberhard) see as the truly novel and non-Darwinian refutation of the selfish gene idea: the idea of genetic accommodation.  “Genetic accommodation” has other names: it’s also been called “The Baldwin Effect” and “genetic assimilation.”  But all of these names refer to a single mechanism: instead of existing genetic variation being subject to natural selection in an existing or changing environment, the environment itself evokes phenotypic (not genetic) variation, which is then somehow fixed in the species’ genome.

This idea was first suggested in 1896 by the American psychologist James Baldwin, although little was known about genetics then.  The idea was worked out in nearly its present form in a paper by George Gaylord Simpson in 1953, and then made famous by geneticist Conrad Waddington, who demonstrated how it might work using a fruit fly experiment.  The idea has been altered and refined—in a way that makes it truly non-Darwinian—by Mary Jane West-Eberhard. Finally, Dobbs misrepresented the whole notion in his piece: not only misleading people about how it works, but by claiming that it makes hash of Dawkins’s “selfish gene” idea.

Now there are really three versions of genetic accommodation. The first of these falls firmly within the purview of neo-Darwinism, and in no way violates the “selfish gene” idea, as Dobbs mistakenly trumpets.  The second, the one that West-Eberhard has proposed, does have non-Darwinian features, but there’s simply no evidence for it.  The third, which seems to be Dobbs’s own garbled version of West-Eberhard’s idea, is non-Darwinian in two ways, and again there’s no evidence for it. Here are the three versions.

1.  A change in the environment exposes a bunch of mutations whose effects were masked in the previous environment.  This exposes a lot of new genetic variation, some of which could be adaptive in the new environment.  The genes that replicate better in the new environment (let’s just say they improve the reproduction of their carriers), sweep through the population via natural selection.

The classic example of this was Conrad Waddington’s artificial selection experiments with fruit flies. He exposed the pupae to heat shock, and this screwed up the development of the flies, so that some of the adults who hatched from the pupal cases had broken veins in their wings.  Waddington then selected the broken-veined flies and bred them to create a new generation, whose pupae were also subject to heat shock.  There were, in their offspring, a higher percentage of broken-veined flies. Waddington continued this selection for several generations. Eventually he saw that broken-winged flies were appearing even when their parents weren’t heat-shocked.  Although this looks like Lamarckian inheritance—the inheritance of acquired characteristics—what was happening was eminently neo-Darwinian. The explanation was simple: the heat shock simply exposed those individuals carrying genes that gave them a propensity for broken wing veins. These effects weren’t expressed at normal temperatures: it took a heat shock for the genes’ effects on wing veins to be seen (that shock might have done this by altering the expression of those genes).  Eventually, via continued selection, Waddington accumulated enough “broken wing vein” genes that they showed their expression without the need for a heat shock.

That, of course, and this form of genetic accommodation, is totally in harmony with Dawkins’s idea of “selfish genes”. The broken-vein genes were indeed “selfish” in the artificial selection experiment, for Waddington selected their carriers to breed.  The only novelty here is that the “selfish” genes were exposed by an environmental change.

Now Waddington’s experiment was done in the lab. Does this happen in nature? My answer is hedged: “Probably, but the evidence is thin.” I can think of only one really convincing demonstrated example: that described in a paper by Scoville and Pfrender in Proc. Nat. Acad. Sci. in 2010.  These authors looked at the small crustacean Daphnia in two lakes in which fish were introduced about 50 and 90 years ago, respectively, and contrasted them with Daphnia in fishless lakes. Normally Daphnia have a plastic response to the environment: if there is high ultraviolet light, they become pigmented to protect themselves (a crustacean sunscreen). But if put in a lake with fish, individuals can get rid of their pigmentation. This “developmental plasticity” is adaptive, because darker individuals are more easily seen by fish and eaten. Presumably natural selection favored the ability of an individual to change its color based on the environment it “perceives,” just as cats grow longer fur in colder weather and shed it when the weather gets warm.

These authors found, though, that in the permanent-fish lakes, where predation was high, the absence of color became “canalized”: that is, individual Daphnia were no longer able to make as much pigmentation when exposed to UV light. What presumably happened is that, when fish were constantly around, they selected for those individuals that didn’t have the ability to change color at all (a mistake like that would get you spotted and eaten, and there’s also a metabolic cost to maintaining flexibility)—those individuals who, though natural selection, simply lost their ability to alter their color in response to intense sunlight.  That, of course, involves “selfish gene-ery”: the preferential replication of those gene variants that have lost the ability to effect color change. You don’t need to become pigmented in lakes where it never pays to lose your pigment, and you pay a metabolic (and presumably reproductive) cost to keep that flexibility.

But that’s the only example of genetic assimilation I find convincing.  It has undoubtedly happened in other cases, but we can’t know for sure because we weren’t there to observe  the evolutionary history. That’s important because, to make a convincing case for genetic accommodation, you need to know whether the genetic variation was there in the first place, and simply selected in the new environment (normal natural selection), or was actually revealed by the new environment.

There are other possible cases, but these are purely speculative.  For example, if an early Tiktaalik-like fish were to walk temporarily on the land, and gain an advantage by so doing (e.g., get more stuff to eat or the ability to scuttle from a drying pond to a more permanent one), those individuals who were able to develop more muscular “fin-legs” during their terrestrial sojourn, due to exertion, would be favored. If that difference in ability to develop musculature had a genetic basis, then eventually such sojourns could select for a better ability to walk on land. Perhaps that is how terrestriality evolved. But of course there’s an alternative scenario: those individuals who had stronger fin-legs to begin with would have an advantage walking on land. That’s not genetic accommodation, but simple natural selection. But neither example violates the notion of the selfish gene.

In contrast, we know of many cases in which existing genetic variation, not revealed by an environmental change, responded to selection: antibiotic resistance in bacteria (the mutations are there before the antibiotic is given), the evolutionary change in color of the peppered moth Biston betularia from white to black during industrialization in England (the black moths were there in low frequency before the polluting began), and the increase in size of finch beaks in the Galapagos in the seventies when there was a drought, killing plants with small seeds so that only finches with the larger beaks were able to eat (finches with bigger beaks were there before the drought).  So the existing evidence supports conventional selection, not genetic accommodation. My guess is that the conventional mechanism would be far commoner, simply because there are few environmental changes that expose new genetic variation that could adaptively respond to that change. (Ancestral brown bears moving north would not get whiter coats due to developmental plasticity.) But regardless, this doesn’t, contra Dobbs, suggest that the “selfish gene” notion is wrong. I say that so often because Dobbs says the opposite so often.

2. A change in the environment alters the phenotype in a directional and adaptive way, and genes are not initially involved. Later, that phenotypic change becomes read into the DNA by selection of genes that promote the adaptive phenotype.  This is, I think, Mary Jane West-Eberhard’s view of how genetic accommodation works. This seems less probable to me than version #1, simply because I can’t envision an environmental change preferentially invoking an organismal change that’s useful in that environment. Instead, like the effects of heat on variants of the heat-shock protein Hsp 90, an environmental change (in this case the heat shock) simply exposes a bunch of genetic variants with random effects on the organism, nearly all of which are maladaptive (deformities and the like).  Heat-shock doesn’t expose mainly those variants that are resistant to heat.  It’s is “non-Darwinian” to envision an organism being able to respond adaptively to an environmental change UNLESS it’s prevously evolved adaptive plasticity, as in the Daphnia case above. But in that case the situation reverts to scenario #1, the exposure of a bunch of genetic variation, only some of which is adaptive, followed by selection among the adaptive variants.

I know of no case of this type of genetic accommodation, and none are really given in West-Eberhard’s 800-page book, Developmental Plasticity and Evolution, which I’ve read.

Despite this, Dobbs not only touts the views given in that book, but quotes West-Eberhard as saying that, in light of her ideas, Dawkins’s selfish gene idea “risks ending up on the wrong side of history.”  Pardon me, but that’s a rather presumptuous thing to say about one’s own theories.  She also says, “the gene does not lead; it follows”, which is not really accurate in her situation if the plastic response of the organism to the new environment evolved through natural selection. In that case the gene originally led, and then follows after the plastic response is expressed in a new environment. But here again no violation of the “selfish gene” paradigm is seen: the environmental changes eventually become fixed in the genome via the selection of genes that promote a response.

3. Dobbs’s Lamarckian view of genetic accommodation.  Dobbs’s characterization of how genetic accommodation works is deeply muddled, for it invokes a Lamarckian process in which a developmental change gets read into the genome, but not by selection among pre-existing variants. It also involves an environmentally invoked phenotype being inherited across generations—that is, the inheritance of acquired characters.

Dobbs’s scenario begins by describing selection for predators that ambush their prey by leaping on them (e.g., a serval).  Then the environment changes: there’s a forest fire. This changes the situation: now to catch prey you must run them down over open ground.  But instead of proposing the simple theory that predators with longer legs, better endurance, and so on, have a selective advantage—conventional natural selection—Dobbs postulates a combination of non-genetic adaptive change of phenotype based on changed gene expression, followed by Lamarckian inheritance of that change, and then, finally, the occurrence of some mutations that favor that phenotype. (Why doesn’t he just cut out the middleman and assume that mutations were there at the outset? Also, why would fire just happen to change the expression of your genes to make you a better open-ground predator? Dobbs does not explain.)

Read his explanation and see if it makes sense to you.  The first part does, because it evokes conventional natural selection:

For example, suppose you’re a predator. You live with others of your ilk in dense forest. Your kind hunts by stealth: you hide among trees, then jump out and snag your meat. You needn’t be fast, just quick and sneaky.

But comes the muddled non-Darwinian part (my emphasis):

Then a big event — maybe a forest fire, or a plague that kills all your normal prey — forces you into a new environment. This new place is more open, which nixes your jump-and-grab tactic, but it contains plump, juicy animals, the slowest of which you can outrun if you sprint hard. You start running down these critters. As you do, certain genes ramp up expression to build more muscle and fire the muscles more quickly. You get faster. You’re becoming a different animal. You mate with another fast hunter, and your kids, hunting with you from early on, soon run faster than you ever did. Via gene expression, they develop leaner torsos and more muscular, powerful legs. By the time your grandchildren show up, they seem almost like different animals: stronger legs, leaner torsos, and they run way faster than you ever did. And all this has happened without taking on any new genes.

Then a mutation occurs in one grandkid. This mutation happens to create stronger, faster muscle fibres. This grandchild of yours can naturally and easily run faster than her fastest siblings and cousins. She flies. Her children inherit the gene, and because their speed wows their mating prospects, they mate early and often, and bear lots of kids. Through the generations, this sprinter’s gene thus spreads through the population.

Now the thing is complete. Your descendants have a new gene that helps secure the adaptive trait you originally developed through gene expression alone. But the new gene didn’t create the new trait. It just made it easier to keep a trait that a change in the environment made valuable.

Note the bold part, which invokes adaptive change in gene expression that was not due to a pre-evolved plasticity (or perhaps it was, in which case the genes were leading!), and, especially, the Lamarckian notion that “your kids run faster than you do” because the changes in gene expression are inherited. But how does that work? It’s Lamarckian, an adaptive and purely epigenetic change in phenotype that can be passed on for at least two generations (“grandchildren).

Now I suppose this is possible, but it is unparsimonious, evoking as it does a concatenation of unlikely circumstances: an “adaptive” plasticity that might not have rested on genes, the Lamarckian inheritance of an adaptive but nongenetic change for several generations, and then a new mutation that is somehow connected with the plastic response.  Why the unnecessarily multiplication of entities? Had Dobbs started this scenario with the mutation in paragraph 2, he wouldn’t need to suggest this convoluted scenario.

And it’s not that the data demand that we invoke this complicated scenario. It’s simply concocted out of thin air, with no evidence supporting it (and no known examples), and then waved around like a flag, with the flag-bearer crying “Horray! Hooray! The selfish gene is dead!”

Parsimony, my dear Dobbs, parsimony. We need not concoct such convoluted scenarios unless we have evidence showing that the present theory is inadequate. But we don’t have that evidence: there is no “black body radiation” phenomenon in evolutionary biology.

Dobbs’s piece goes on to further tout this idea and drag in some scientists (many of whom I’ve criticized before) who like the idea, but I’ve done my job now and must rest (or rather, work on my book).

Let me add one thing, though. I’m constantly puzzled these days by how often people argue that the neo-Darwinian synthesis is wrong, and that we need a new paradigm. Genetic assimilation, epigenetics, horizontal gene transfer—all of these buzzwords are evoked as reasons to jettison our “conventional” view of evolution. But always, when you look at the data, the evidence that these phenomena will overturn neo-Darwinism is nonexistent.

I’ve already written a lot on the epigenetics hype, and have shown that there’s no evidence that a single adaptation in nature involves the fixation in the DNA of an epigenetic alteration of the genome that isn’t initially inherited.  Yet people keep banging on about epigenetics.

I’m not sure why the hype continues, but perhaps it has to do with the fact that the main paradigm of evolution—the neo-Darwinian synthesis—is largely consolidated, and is correct. Sure, there are surprises to come, and interesting new phenomena, but there’s no “quantum mechanics” of evolution on the horizon.  Some theories don’t need to be overthrown because they’re generally right. Perhaps people don’t like working in a field where there’s no new “paradigm” to forge, and Kuhn has ruined us all!

The “neo-Darwinism is dead” trend may have to do with ambition, or perhaps with boredom. I don’t know. What I do know is that the many recent challenges to neo-Darwinism have all failed to hold water, but people keep pouring liquid into that sieve.