Ivory poaching imposing selection on elephants to evolve shorter tusks

October 24, 2021 • 9:30 am

Here we have a case of selection by humans—killing elephants that have tusks because ivory is so valuable—increasing the frequency of tuskless African elephants in Mozambique over a 28-year period. (As we’ll see, only the proportion of tuskless females increased.)  We have similar examples from other species, as in the reduction of horn size in bighorn sheep hunted for their horns as trophies, and the reduction in the size of some fish due to commercial fisherman going after the big ones.

Is this artificial or natural selection? Well, you could say it’s artificial selection because humans are doing the choosing, but after all human are part of nature. And this selection was not conducted to arrive at a given end. Dachshunds were selected to look like hot dogs to root out badgers in their burrows, but the reduction of tusk size in elephant, or horns in sheep, was not a deliberate target of selection, but a byproduct of greed. So I would hesitate to characterize this as artificial selection, since it’s not like breeders choosing a given characteristic to effect a desired change. In fact, the evolutionary change that occurred is the opposite of what the “selectors” wanted.

You can find the article in Science by clicking on the screenshot below, or get the pdf here.  There’s a two page shorter take that’s an easier read, “Of war, tusks, and genes,” here.

The phenomenon: a civil war in Mozambique from 1977 to 1992, which increased the frequency of tuskless female elephants from 18.5% to 50.9%, nearly a threefold increase. Why? A model showed that such a change (which occurs among generations, so it’s not just selective killing within a generation) must have been due to natural selection rather than genetic drift. The killing was motivated by a desire to get money to fund the conflict.  A female without tusks had five times the chance of surviving as a tusked female. That imposed strong selection in favor of tuskless females.

Usually, tuskless elephants are at a disadvantage, for tusks are multi-use features, employed for defense, digging holes for water, male-male competition, and stripping bark from trees to get food. But the natural selection to keep tusks in females was weaker than the “artificial selection” by humans against tusks.

Here’s a photo of a tuskless vs. a tusked female:

Photo by Finbarr O’Reilly for The New York Times

And the only kind of male that we see: ones with big tusks (tusk size varies, of course, as they continue to grow as the elephant lives). Tusks are homologous with our incisor teeth.

The authors first tried to determine the genetic basis of having versus lacking tusks. It turns out that, by and large, tusklessness behaves not as a complex trait caused by changes in many genes of small effect, but as a single dominant mutation on the X chromosome (like us, elephant males are XY and females are XX). Further, the dominant mutation causing tusklessness is lethal in males, killing them before birth. (This is probably not because the tuskless gene form is itself lethal, but is closely linked to a gene that is a recessive lethal.)

So here are the “genotypes” of the elephants. I’ve used “x” as the gene form on the X chromosome that produces tusks, and “X” as the alternative dominant allele that makes you tuskless.

Males: All have tusks and are thus xY. (Males have only one X chromosome and also a Y.) The XY genotype is lethal, so we never see males carrying the tuskless gene form (XY). Ergo, there are no tuskless males.

Females: We see two types:

Tuskless: Xx. These females will lose half their male offspring because when mated to an xy male (the only viable type), they produce half xY males, which are tuskers, and half XY males, which are lethal. Thus a population of tuskless females will produce a sex ratio in their offspring skewed towards females, which is what is observed.

We never see XX tuskless females because they’d have to inherit one “X” from from their fathers, but that XY genotype is lethal.

With tusks: xx.

There are a few complications, as other genes are involved (for example tusked mothers, who are xx, produce only 91% of tusked daughters when you’d expect the xx by xY cross to produce 100% xx (tusked) daughters. So things are not quite so simple, but in general a single gene seems largely responsible for the tuskless condition. (You might expect this, because if many genes were involved you simply wouldn’t get females lacking tusks: you’d get females with slightly smaller tusks, who would still be killed for their ivory. It would thus take many generations instead of a couple to raise the frequency of tuskless females.)

I won’t go into the gory genetic details, but the authors sequenced entire genomes from tusked and tuskless males and females and looked for signs of natural selection on some genes, comparing the tusked versus tuskless females. (One sign of rapid selection for tusklessness, for the cognoscenti, is the presence of DNA bases recurrent and common near the gene causing tusklessness.)

The researchers found one X-linked gene form with strong signs of selection called AMELX, which in other mammals codes for a protein that leads to the mineralization of enamel and regulates other tooth-associated genes. Another gene not on the sex chromosome, MEP1a, also is associated with tusklessness, but not as strongly. This gene, too, is known to be associated with tooth formation in other mammals. Here’s the diagram from the paper of which parts of the tusk are controlled by which gene. You can see that AMELX is expressed only in the “tusky” part of the tusk:

(From paper): Putative functional effects of candidate loci on tusk morphology.A cross section of an African elephant tusk shows the anatomical position of (a) enamel, (b) cementum, (c) dentin (ivory), (d) periodontium, and (e) root of the tusk. Dark blue circles indicate regions known or proposed to be affected by candidate gene AMELX. Light blue circles are proposed to be affected by candidate gene MEP1a. Neither gene is known to affect the formation of the dental pulp (black interior of cross section).

The upshot: Human-imposed (“anthropogenic”) selection that causes evolution in the wild has been demonstrated before, so this phenomenon is not new. What is new is that the genes involved in an anthropogenic evolutionary change—the increase in frequency of the tuskless allele, which is evolution—have been identified for the first time, and we know the kind of selection that’s caused the evolution. What is also unusual (I know of no other case) is that selection for tusklessness is in opposite directions (“antagonistic selection”) in the two sexes so long as tuskless females survive better. As the authors note:

Physical linkage between AMELX and proximate male-lethal loci on the X chromosome, such as HCCS, may underpin the proposed X-linked dominant, male-lethal inheritance of tusklessness in the Gorongosa population. If our interpretation is correct, this study represents a rare example of human-mediated selection favoring a female-specific trait despite its previously unknown deleterious effect in males (sexually antagonistic selection). Given the timeframe of selection, speed of evolutionary response, and known presence of the selected phenotype before the selective event, the selection of standing genetic variation at these loci is the most plausible explanation for the rapid rise of tusklessness during this 15-year period of conflict.

What of the future? Even though the conflict is over, poachers continue to kill tuskers for their ivory in much of Africa. What will happen? We expect the frequency of the dominant tuskless allele to increase. That itself will not lead to extinction of the population because tuskless males are simply not produced: all tuskless females will remain Xx and produce half the normal number of males. Tusked females will still be produced as Xx females crossed to xY males will produce both Xx (tuskless) and xx (tusked) females.  But the reduction in the number of males produced by anthropogenic selection, coupled with continual poaching of both males and females with tusks may drive the population size so low, with an unequal sex ratio, that it could become severely endangered.

Since tusks are good for elephants, the solution is not only to ban the trade in ivory, which has been done in part, but some countries continue to trade in elephant ivory. Further, we must stop the poachers cold, as there’s still a market for both legal and illegal ivory, and prices are high. That’s easier said than done given the area that must be monitored. Note, though, that in 2017, Donald Trump lifted the ban on ivory imports from Zimbabwe, which had been put in place by his predecessor. And the elephant is the Republican symbol!

h/t: Pat, Matt, and several other readers.


S. C. Campbell-Staton et al.. 2021. Ivory poaching and the rapid evolution of tusklessness in elephantsScience 374, 483-487.

Heavy human harvesting of a valuable medicinal plant leads to evolution of new leaf and flower colors

November 22, 2020 • 10:30 am

If humans harvest an animal or plant, especially if they harvest it heavily, the species often evolves to make itself less “harvestable”.  For example, commerical fisheries that take the larger fish in the sea have led to the evolution of individuals that mature earlier at a smaller size, for it is the small reproducing fish who don’t get caught. Elephants harvested for their ivory have, in some populations, evolved smaller tusks or even tusklessness, for it’s the tuskless elephants who leave more offspring. (The condition for all such evolution, of course, is that the evolved conditions have at least a partial genetic basis.)

Finally, there’s a similar phenomenon called “Vavilovian mimicry”—named after the great Russian geneticist and botanist Nikolai Vavilov, who was imprisoned by the Soviets and died in the gulag because he dared to embrace Western genetics and science against the teachings of the charlatan Lysenko.

In Vavilovian mimicry, weeds are selected among agricultural crops with which they grow to get themselves in the next generation of the crop. Farmers have mechanical ways to sort out the weed seeds during harvesting, and this imposes selection on the weeds to produce seeds of the same size and shape as the crop; it’s those mutant weed seeds that get replanted the next year.

A cool and famous example is how the common vetch (Vicia sativa), a weed, has evolved in crop areas so that its seeds come to closely resemble that of the edible lentil (Lens culinaris), a crop that the weed infests.  Because lentil seeds, which are what’s eaten, are tasty but vetch seeds are bitter, farmers have used mechanical and visual sorting to discard the wild vetch seeds. Over time, the vetch seeds have undergone what’s called “unnatural selection” (for Vavlovian mimicry) to have the same size, color, and shape (flattened) as the lentil seeds. Here’s a diagram showing the cultivated lentils (A) along with the wild vetch seeds growing on their own (B), and the seeds of the same vetch, but which have grown in lentil fields. Look at the big evolutionary change in the vetch seeds!:

Today we have another example of plants mimicking other things—in this case the environment—to hide themselves from being harvested.  Fritillaria delavay, is a perennial alpine Asian plant that grows from a bulb, living about five years. The bulbs, particularly the small ones, are very prized in Chinese medicine, especially for treating tuberculosis, fetching up to nearly $500 per kilogram. (Since they’re small, it takes about 3,500 bulbs to make a kilogram.) They are picked visually, with harvesters looking for the bright green leaves and flowers of the plant that stand out against their rocky background.

Since harvesting is heavy, you can guess how the plant evolved. That evolution is documented in this new paper in Current Biology (click on screenshot below, or go here to get the pdf, both of which are free).  If you want a journalistic summary, there’s one in the Times and another in the Guardian.

In short, the plant has undergone evolution of both leaf and flower color to make it more inconspicuous and thus harder to find and harvest (harvesting, since it takes the bulb, kills the plant). You’re more likely to reproduce if you’re not seen, and in harvested areas those plants with mutations making them match the background better are those that survive. Herbivores apparently aren’t involved in this system, as nothing has been observed to eat the plant, which is full of alkaloids and toxic.

Here are pictures F. dlavayi in an unharvested area (left) and one in an area heavily harvested (right). You can guess which is which. Note the difference in the color of both leaves and flowers. In fact, the green color can evolve to either reddish, brownish, or grayish colors depending on the color of the background.


In the paper, the authors collected plants from eight populations in southwest China, and found significant divergence of color among the populations using a special “vision model” to measure the colors and luminescence seen by humans. Here’s a plot of the variation among the eight populations (each dot has a color that is related to the plant color, with each color representing a single population):

(From paper): Plant Color Variation of Fritillaria delavayi among Populations. (A) Color divergence from eight populations in human CIE L∗a∗b∗ color space

Are the plants camouflaged in their local area, and is the degree of the camouflage correlated with how heavily the plants are harvested? The authors derived a measure of how camouflaged a plant was by comparing leaf and flower color with the color of the soil or rock background (also measured using the human-vision algorithm). Collection intensity was assessed by questioning the locals and deriving an estimate of intensity = [amount of bulbs collected]/[relative abundance of the plant in the area]. The higher this fraction, the heavier the collection effort (i.e., the proportion of the population that gets taken by collectors).

As you see from the plot below, the higher the collection intensity in a population (position to the right), the better the mimicry (lower values on the Y axis). The relationship is highly statistically significant (p < 0.001). Clearly, the prediction that the color evolved in response to human harvesting is supported.

Finally, the authors looked at an ancillary relationship: that between the difficulty of digging up bulbs (some are hidden under dirt and rock piles) and the degree of camouflage of the population. The relationship they found is shown below. One predicts that the easier it is to dig up a bulb, the more camouflaged the population would be, for easier digging makes for heavier harvesting and thus stronger “unnatural selection”. The relationship below affirms the prediction, though they left out one population where collection is easy but the plant is green—yet collection isn’t heavy in this population. (This sounds like post-facto discarding of data, but could be kosher.)

Whether each dot is statistically independent of the others, which seems to be the assumption when doing the nonparametric correlations, is dubious, since plants in a given area are related to one another, and each plant didn’t evolve its color independently—the population as a whole evolved its color as a gene pool.

Leaving that possible quibble aside, the authors finally did a computer experiment on target slides showing plants matching their background to various degrees. They found, as expected, that the locals took longer to detect a plant when it matched the background, confirming that your chance of escaping “predation” is likely higher when you’re better camouflaged.

Here’s one more photo from the paper showing the cryptic nature of the plant in brown and gray backgrounds (C and D), and how readily the bright green plants stands out against a scree background (A and B; this is clearly a low-harvest area).

Plant Color Variation of Fritillaria delavayi among Populations

There are no new principles demonstrated in this paper, but the results are still fascinating, and show a mixture of artificial and natural selection that’s called “unnatural selection.” That is, the color isn’t a deliberate product of the breeder, like the grotesquely long bodies and minuscule limbs of wiener dogs, but is an inadvertent result of “artificial” selection. (I’m not even sure I’d call this artificial selection, for humans are part of nature and are gathering something they need.) And, like natural selection, all this process requires is differential reproduction of individuals that have different genetic variants.

If you want to read more about “unnatural selection” and how it’s affected many species, click on the screenshot below.

h/t: Ben, Matthew, Florian


Niu, Y., M. Stevens, and H. Sun. 2020. Commercial Harvesting Has Driven the Evolution of Camouflage in an Alpine Plant. Current Biology. Online. https://doi.org/10.1016/j.cub.2020.10.078

Discovery Institute makes hay of Dawkins tweet, and a geneticist mistakenly says that artificial selection won’t work in humans

February 20, 2020 • 9:45 am

Unless you’ve been in Ulan Bator (and actually, some people in Mongolia do read WEIT), you surely know about Dawkins’s latest twitter kerfuffle, in which he said, correctly, that human eugenics would “work”. That is true in the sense he meant it: artificial selection practiced on human traits would yield a change in mean trait values, for most traits have appreciable “heritability.” People misinterpreted that—most of them deliberately, I think—to excoriate Dawkins as favoring eugenics, something that’s clearly untrue, especially in light of his subsequent clarifications. It’s not clear to me why these people won’t admit that they mischaracterized Dawkins’s tweets. But of course people get stuck in their ideology and are loath to admit error.

[UPDATE: Dawkins reiterates what he meant on a comment on yesterday’s post: here.]

One commenter noted that most of the pushback seemed to come from the Left or from the woke. That may be true, for many of them hate Dawkins for being “the wrong kind of atheist”: seen (wrongly) as shrill and uncaring about oppression. But I have to note that the Right has been making hay about Dawkins’s tweet as well—especially the religious, who tend to be on the Right. That goes double for creationists, including those who are wasting their lives at the Discovery Institute.

And so, on the DI “Evolution News” website, David Klinghoffer, an Orthodox Jewish ID creationist, has evinced some glee about Richard’s tweet. To see it, click on the screenshot below (it goes to a Wayback Machine link that I’ve archived so that the site itself doesn’t get clicks):

Klinghoffer takes two approaches to denigrating Dawkins.

The first is to cite geneticist Dave Curtis’s recent Twitter thread arguing that eugenics wouldn’t work. (Curtis is an Honorary Professor in the Division of Biosciences at University College London.) Curtis’s tweets have cited widely to show that Dawkins was wrong, but, sadly, I think Curtis himself is wrong. Dead wrong. I’ll give a few of his tweets and briefly explain why.

Here’s the first one, and the assertion that “eugenics simply would not work” is not at all supported by human data, as I document below.

I’ve indented Curtis’s subsequent tweets (only the ones I see as relevant). My own comments are flush left.

I work on human genetics and am honorary professor at the UCL Genetics Institute. I’m the editor-in chief of a journal which used to be called Annals of Eugenics. I just wanted to say that we now know from the latest research that eugenics simply would not work.

This is not true at all. The up-to-date data we have suggests strongly that artificial selection on human traits would “work” in the sense of changing mean trait values in the direction you select. Moreover, it would work in this way for nearly all human traits (see paper below). By saying eugenics would “work”, of course, I am, along with Dawkins, not at all saying it should be practiced. While a limited form of selection in humans is acceptable—for example, preventing a couple who are carriers of a recessive genetic defect or disease from producing an offspring with that condition—the kind of wholesale and directed selective breeding of humans suggested by the word “eugenics” is immoral, and I don’t favor it at all.

On to more tweets. In this one, Curtis flaunts his expertise, but that doesn’t make the data showing him wrong any less convincing:

I have published hundreds of scientific papers on human genetics including on intellectual disability, mental illness and the predictive ability of genetic. You can view the list here: scholar.google.co.uk/citations?hl=e

On to his objections:

Animals are bred in controlled environments and have short generational times with large numbers of offspring. In these circumstances selective breeding can produce desired changes in a small number of specific traits such as milk yield or racing performance.

There are a number of different kinds of reason why eugenics would not work. One is that humans have long generational times and small numbers of offspring. This would make any selective breeding process extremely slow.

Well, “controlled environments” doesn’t mean that selection wouldn’t work, any more than saying selection wouldn’t work in nature because the environment in nature is variable. Artificial selection in animals is successful even in variable environments: I could, for example, select for more bristles on Drosophila flies, even while changing the type of food they get every generation and letting them experience variation in room temperature. We’d still get an increase in bristle number over time. If you think otherwise, I’d bet you a lot of money that you’re wrong.

The “long generation time” of humans isn’t a barrier to getting a result with artificial selection. It only means that, in terms of years (not generations), getting a response would be slower. But not infinitesimally slow!

For example, if you have a trait like height, which appears to show a heritability of about 0.8 (80%), then if you breed only from a group of humans whose average height is 5 inches above the population mean, in the next generation (ca. 20 years later), the response to selection—the average height of the selected group’s offspring when mature—would be 5 X 0.8, or four inches above the mean. That is, you would have raised the height of the population by four inches. That’s a big change in one generation: you’ve gone 80% of the way to your goal. It all depends on the heritability of the trait (which is usually appreciable) and how strongly one selects.

The “breeder’s equation” for this kind of calculation is simply response to selection = heritability of the trait in that population X the strength of selection practiced. And in fact this experiment is performed in miniature every day: tall couples produce tall offspring, short ones produce short offspring. That is really a form of artificial selection performed because couples tend to mate assortatively by height. Each couple mating gives us an idea of how much variation in that trait is genetic.

Finally, the low number of offspring doesn’t matter so long as you can keep the population going after selection. Note that offspring number doesn’t figure in the breeder’s equation.

As for the statement “selective breeding can produce desired changes in a small number of specific traits such as milk yield or racing performance”, that’s extremely misleading. As I’ve said before, I’m aware of only two artificial-selection experiments, out of hundreds practiced on genetically variable populations, that failed to yield a response, and both of those experiments were mine. (I was selecting on “directional asymmetry”, a trait with very low heritability.) “Small number of specific traits” is the misleading bit here. Better that he said, “selective breeding can produce desired changes in almost any trait.” After all, remember what Darwin said in The Origin, a conclusion based on breeders’ results before evolution and genetics were accepted or even understood:

 “Breeders habitually speak of an animal’s organization as something quite plastic, which they can model almost as they please.”

That, of course, means that animal traits have substantial heritability, for it is that heritability that make animals (and plants!) quite plastic. And so it is with humans, for we have evidence that natural selection has altered several human traits in the past 10,000 years or so, and in populations that are relatively small.

Another tweet by Curtis:

Another reason is that humans are exposed to very different environments, so most of trait variation is not due to genetic factors but to differences in environment. One consequence is that it makes it hard to identify subjects who have desirable genetic characteristics.

Here Curtis is again being misleading. What we do know from studies of heritability in our species is summarized in the article below (click on screenshot). The article shows that virtually all human traits have appreciable heritability (“selectability”), with none having zero heritability. And is it really hard to identify humans who are taller than others, or have higher IQs or better teeth? Yes, there is often substantial environmental variability contributing to the trait variation (diet and education in the cases I’ve cited), but this doesn’t mean that selection wouldn’t work.

Here’s the paper’s summary, showing that most traits have a substantial heritability (49% means that about half of the variation among individuals in a population is due to “additive” genetic factors). Further analysis suggests that (shared) environmental influences aren’t overwhelmingly important in these twin studies. Other studies of both identical and fraternal twins reared together and apart also show a substantial heritability and lower environmental effects than expected (see here, and here, for example). Further studies using not twins but identity by descent (e.g., here) also confirm heritabilities derived from the twin studies.

And from the paper’s discussion (my emphasis):

We have conducted a meta-analysis of virtually all twin studies published in the past 50 years, on a wide range of traits and reporting on more than 14 million twin pairs across 39 different countries. Our results provide compelling evidence that all human traits are heritable: not one trait had a weighted heritability estimate of zero.

That means that virtually all human traits would change when subject to artificial selection.

More of Curtis’s tweets with my responses.

We can now measure genetic potential directly from genetic markers and what we know from this is that these genetic predictors perform extremely badly. We can also tell that there are many important, very rare genetic variants which we will never be able to identify. 9/n]

Individual genetic markers are largely irrelevant here; what is important in judging whether selection would “work” is the heritability of the trait in the population, which reflects variation at all relevant genes, not just one or a few genetic markers. Again, using one or a few genetic markers is not the way to change traits. The way to change them is to select for or against certain trait values.

“We should bear in mind that harsh selection pressures have been acting on humans up to the present and that there may be very little scope for overall improvement. In any event, we can confidently say that selective breeding to improve desirable traits is not practicable.

Here Curtis is saying something not supported by the data. We know that there is still substantial genetic variation in humans from the heritability studies above, which directly contradict Curtis’s claim that “there is little scope for [change].” The average heritability at present is nearly 50% among all traits, which means that there is huge scope for “overall improvement” (I prefer “change”, as I don’t know what would constitute “improvement” in humans.) Of course it’s not “practical” to perform such broad-scale selection as a form of eugenics because of moral considerations, but that’s separate from whether that kind of selection would change the mean of a population.

With a recessive disease it may be possible to eliminate cases of the disease from the population using a combination of carrier testing, prenatal screening and selective termination. However this is not eugenics because the variants are still present in the population.

Of course that kind of selective breeding (termination of genetically afflicted embryos) is eugenics! Some variants would remain in the population, but their frequency would be reduced. That is a response to selection! And that involves “terminating” (a euphemism) genetically defective embryos. Just because calling selective elimination of embryos “not eugenics” doesn’t make it not eugenics.

Again, using one or a few genetic markers is not the way to change traits. The way to change them is to select for or against certain trait values.
TLDR: People who support eugenics initiatives are evil racists. Also, modern genetic research shows that eugenics would not work. 19/end

The first part of the statement is true in a qualified sense. However, “eugenics” practiced as “elimination of cases of disease from the population,” as Curtis mentions above, are certainly “eugenics initiatives”, and most of us support such practices. That doesn’t make us evil racists. What does make us evil racists is selective breeding practices on entire races or populations with an eye to differentiating races.

The second part of the statement—that artificial selection on human populations wouldn’t work—is just wrong, and dead wrong. I’m surprised that a man of Curtis’s expertise would make a statement like that. His Twitter thread should not be used as evidence against Dawkins’s claim for the efficacy of artificial selection in humans.

Klinghoffer’s second approach is to cite Behe’s claim that while artificial selection may create some success changing species, it can’t effect big changes. It can create breeds of dogs from wolves, for instance, but can’t change a wolf into a puma. This is an old creationist trope. Quotes from Klinghoffer and Behe are indented:
In an email, a geneticist friend notes the irony. Darwinian evolution is a massive extrapolation from selective breeding in animals. Of course animal breeding “works,” up to a point. Darwin in the Origin of Species cited the efforts of pigeon fanciers. In a New York Times book review, Dawkins once taunted Michael Behe with the successes of dog-breeding. But there are limits. Dogs can’t be bred to become cats, nor pigeons into bats. There appear to be set limits.

There appear to be set limits. Why? Behe has noted the problem that dog-breeding, canine eugenics, is accomplished largely by breaking genes:

Popularizers of evolution said if we can breed dogs that are so different from each other and only do it in the past few hundred years, how much better could nature do? But again, we didn’t know what was going on in the biology of these dogs. In the past 10 years, the entire genomes of many different dog breeds have been sequenced. And again, it turns out if you want a Chihuahua, you can break one of the genes involved in growth. If you want French poodles with curly hair, you break a gene involved in hair growth. If you want a dog with a short muzzle, you break a gene involved in facial shape development.

I and others have already refuted Behe’s claim (see also here and here) that selection for new features is ineffective because it involves broken genes that eventually stop selection in its tracks. And besides, the kind of changes that racist eugenicists proposed in the past (and again, I don’t favor them) are small-scale changes of the type involved in other forms of artificial selection. They didn’t propose turning humans into pumas!

Klinghoffer goes on:

Dave Curtis’s well informed observation is that even given the success of animal breeding, the analogy with humans is mistaken. But that leaves evolution…where? The extrapolation from dogs or pigeons to macroevolution fails because building genuine biological novelties, not just a Chihuahua as distinct from a poodle, requires more than merely breaking stuff, aka devolution, as Behe has shown in his book Darwin Devolves. If the many wonders of the animal world could not have proceeded from Darwinian blind shuffling alone, then human evolution, which can’t even stand on the shaky ground of human eugenics, all the more cannot have done so. 

As I’ve showed, Curtis has no data supporting him, and considerable data contradicting his claims. But in the end, this whole kerfuffle has nothing to do with macroevolution: it’s about microevolution. Even noting that, I’ll argue—but not here—that the IDers’ supposed “unbreakable barrier” between microevolution and macroevolution is totally bogus.

Dawkins makes a tweet

February 16, 2020 • 2:30 pm

UPDATE: Crikey, there’s a whole Twitter “event” devoted to Dawkins’s tweet. People can’t wait to jump all over him.


When Matthew sent me this new tweet from Richard Dawkins this morning, I thought “Oh no! I know what he means, but there are a gazillion people out there ready to misinterpret it as an endorsement of eugenics.” And Matthew said, “Yeah, and everyone’s going to jump on that word ‘work’.”

Yep, what we predicted happened.

I didn’t read Richard’s followup (below) until a few minutes ago, and I don’t know whom he was arguing with, but his tweet was clearly intended to show that many human traits are heritable—that is, they would respond to artificial selection, which is what eugenics is. (In “negative” eugenics, you cull or prevent from breeding individuals with undesired traits; in “positive eugenics”, the other side of the same coin, you breed from individuals with desired traits.) As I’ve written before, nearly every artificial selection experiment conducted, on a gazillion different species, has been successful: the mean value of the trait has changed in the direction the experimenter wanted. (Two of the few unsuccessful experiments have been mine: attempts to select for directional asymmetry in flies.)

Artificial selection will work if a trait has any positive heritability, that is, if any proportion of the total variation in a trait among individuals in a population is due to genetic variation—what we specifically call “additive genetic variance” in the trade. And virtually all morphological or behavioral traits have some positive heritability.

Look at domestic dog breeds, for instance. All of them descend from the wolf, yet all the huge variety of their traits: the variation in their size, their shape, their color, and even their behavior (retrievers, border collies, etc.) have come from selecting on traits that have a positive heritability. As Darwin said in The Origin, “Breeders habitually speak of an animal’s organization as something quite plastic, which they can model almost as they please.”

That happens to be true. And it would be true of humans as well if we were able to select on them. Have a look at this paper, for instance (click on screenshot, and if you can’t get it, make a judicious inquiry or use the legal Unpaywall app on Chrome):

From the discussion:

We have conducted a meta-analysis of virtually all twin studies published in the past 50 years, on a wide range of traits and reporting on more than 14 million twin pairs across 39 different countries. Our results provide compelling evidence that all human traits are heritable: not one trait had a weighted heritability estimate of zero.

Not one! They take into account “shared environment” causes of correlations as well. And behavioral traits are among these. The heritability of IQ, as I recall, is about 50%, so if we wanted to improve the IQ of humans, we’d just let the smartest ones breed, and lo, we’d get a fairly substantial change in a few generations.

Should we do that? Hell, no!!  Nobody wants to go back to the era of eugenics, when “feeble-minded” people were sterilized in America and all kinds of mentally disabled people (as well, of course, as non-Aryans like the Jews, who were deemed to have “bad genes”) were killed by the Nazis. But we still practice a mild form of negative eugenics today, in the form of genetic counseling and selective abortion, to prevent couples from having children with deformities or genetic disease. That’s not to improve the population’s genes but to allow couples to have healthy babies. We do no large scale eugenics, positive or negative, to improve the population, and that’s the way it should stay.

In general, no biologist that I know wants to return to the bad old days of wholesale eugenics, which involved not only killing or sterilizing people but demonizing whole groups for their genetic endowment. So I understood what Richard was trying to say.

Should Richard have issued that tweet? Again: Hell no!! Richard knows (or surely must, just as Matthew and I knew) that there are many people out there ready to misinterpret what he says and would use it to imply that Dawkins favors eugenics—that he’s a latter-day Nazi. I see that it’s already happened.

Here’s someone else who didn’t get the tweet at all:

Crikey, can’t Dr. Blommaert read?

Further, a discussion about artificial selection in our own species or others should surely be more extensive and nuanced, not conducted in the medium of Twitter where you have only a few words to say what you want.

I see that Richard has already gone his usual route of trying to explain what he meant:

So what is Richard guilty of? Unwise tweeting! He’s not a neo-Nazi, and, knowing him, I know for sure that he’s not in favor of eugenics. But he should have learned by now to stay away from Twitter, at least on issues like this one.

Now I don’t know who he was responding to in the initial tweet, but it looks as if somebody somewhere said that eugenics wouldn’t work in humans. If they said that, they’re wrong, and Richard is right. But it doesn’t look as if his tweet was responding to anybody in particular (or at least I haven’t been arsed to investigate), and so he should have kept his thoughts to himself.  But it is, I suppose, useful to emphasize that we, like all animals, contain a reservoir of genetic variation for almost all traits, and that means that we could respond to artificial selection. But that should immediately be followed, for those eager to demonize you, by the statement that of course we shouldn’t artificially select on members of our own species. What’s possible is of course not identical to what we should do.

Bad optics, my friend Richard. But those of you ready to use that first tweet to demonize the man—please lay off. He may be a hamhanded tweeter, but he’s no Nazi.


Artificial selection in action: more elephants are being born without tusks

June 1, 2017 • 9:30 am

What do you expect if hunters or poachers selectively kill elephants with big tusks—either for trophies or their ivory? This is actually a form of artificial selection, and it will have the expected results: elephants with smaller tusks will be more likely to survive and reproduce, and if there’s genetic variation for tusk size or presence, which there almost certainly is (there’s genetic variation for nearly every trait, accounting for phenomena like the ability of humans to change the gray wolf into Chihuahuas, greyhounds, sheepdogs, and so on), the “tuskiness” of elephants will change over time. Tusks will get smaller, or even disappear.

You can also predict that if tusks are more important for one sex than the other, that the natural “counterselection” against tusk reduction will be stronger in that sex, so that the reduction in size or presence over time will be slower and, ultimately, might stabilize at a larger size than in the sex having tusks less important for survival.

This is precisely what an article by Robby Berman at The Big Think reports. Berman notes that in non-poached populations of African elephants (Loxondota spp.), 2-6% of female elephants are born without tusks. I’m actually surprised that the percentage is that high given that tusks are used by both sexes to deter predators, dig water holes, clear obstacles, and strip bark from trees. But in poached populations that percentage can nearly reach 100.

But tusks are more important in males since they’re intimately connected with reproduction: males joust for mates using them, and a tuskless male is a childless male. Thus one would expect that, given equal intensity of poaching, males would still wind up with larger tusks than females—unless poachers kill every animal with tusks, which, by eliminating males, would drive the species extinct.

According to Berman, selection is indeed working this way, and because it’s strong—a large percentage of elephants are killed for their ivory—we’d expect the change to be rapid. As he writes:

In areas where there is poaching, however, the story’s very different, and the quest for elephant ivory is changing the types of offspring now being produced. In Gorongosa National Park in Mozambique, half of the older females have tusks. The situation has improved since poaching was brought under control there 20 years ago, but a third of the younger elephants are tuskless nonetheless, a meaningful increase over the historical norm.

In Zambia’s South Luangwa National Park and the Lupande Game Management Area, tuskelessness increased [JAC: read the reference] from 10·5% in 1969 to 38·2% in 1989 The numbers have improved slightly since then there as well, but only due to more tusked females migrating from nearby areas.

How strong is the selection? The Independent reports some populations have almost no females with tusks:

An increasing number of African elephants are now born tuskless because poachers have consistently targetted animals with the best ivory over decades, fundamentally altering the gene pool.

In some areas 98 per cent of female elephants now have no tusks, researchers have said, compared to between two and six per cent born tuskless on average in the past.

Almost a third of Africa’s elephants have been illegally slaughtered by poachers in the past ten years to meet demand for ivory in Asia, where there is still a booming trade in the material, particularly in China. [JAC: this trade will shut down at the end of this year by government decree.]

. . . The most striking example is in the Addo Elephant National Park in South Africa, where 98 per cent of female elephants have no ivory. Big game hunters there had killed all but 11 elephants by the time the park was created in 1931. Four of the eight surviving females were tuskless.

In 2008, scientists found that even among elephants that remained tusked, the tusks were smaller than in elephants’ a century before – roughly half their previous size.

What will happen? Given the strength of selection on tusks (ivory goes for $730 per kg on China’s black market, a 2/3 reduction since the ivory trade started to be banned), both the number of elephants and the size of their tusks will decrease. They will remain larger in males since there’s an additional penalty—a strong one—for being tuskless in that sex. One might then expect females to select for mating with those males having smaller tusks, counteracting this trend, but since females may not have a preference with whom they mate (males win in competitions), that kind of selection might not occur.

This is all speculation, but what’s not speculative is that the selective poaching of elephants with tusks is having the expected (but unwanted) evolutionary effect.

Here’s a tuskless male, thus a luckless male:

A female with small tusks:

My first thought was to anesthetize elephants and remove their tusks to foil the poachers, but that can’t be done for several reasons, most important that the tusks are alive and contain nerves and blood vessels (they are in fact incisor teeth of the upper jaw), not to mention the difficulty of doing that to a lot of elephants.

Read more about this in the article “Going tuskless” at the African Wildlife Foundation.

h/t: Steve