« October 2007 | Main | December 2007 »

November 30, 2007

Controlling sex ratios

I formerly thought that when a tendency to produce the two sexes in equal numbers was advantageous to the species, it would follow from natural selection, but I now see that the whole problem is so intricate that it is safer to leave its solution for the future. -- Charles Darwin in Descent of Man
This week, I will discuss recent papers that shed light on the evolution of genes that control sex ratios in insects (fruit flies) and mammals (mice). John Dennehy recently discussed Hamilton's 1967 paper, "Extraordinary Sex Ratios." (Yes, Hamilton, as in Hamilton's r.) See also the last paragraph of this post, on the surprisingly sophisticated adjustment of offspring sex ratios by fig-pollinating moths.

As Darwin was starting to recognize, natural selection doesn't depend on whether a trait is advantageous to the species. If the effects of an allele (a particular DNA sequence) cause it to become more common, relative to alternative alleles that compete for the same space on a chromosome, then it will become more common, whatever the effects on the species as a whole. As Richard Dawkins famously pointed out, natural selection acts as if each gene were "selfish."

In particular, alleles that increase the chance of having male, rather than female, offspring will tend to become more common when males are rare, and vice versa. This is because each individual of the minority sex will generally have disproportionately high genetic representation in the next generation. If a population has one female and ten males, for example, every member of the next generation will be descended from that female, whereas the chance of being descended from any individual male is only 10%. The parents of girls born in China today are more likely to become grandparents than the parents of boys, although this could be affected by international migration.

So that's why natural selection often leads to a 50:50 sex ratio. But deviations from this ratio do occur, as Hamilton noted. This leads to two questions. What is the ultimate (evolutionary) reason for different sex ratios? And the proximate (mechanistic) question: to what extent can individuals actually control the sex ratio of their offspring, and how?

Two papers by Yun Tao and colleagues on control of sex ratio in fruit flies were just published in PLoS Biology. Because this is an open-access journal and because PLoS includes a well-written summary, by Patrick Ferree and Daniel Barbash, I will only discuss the main points.
In fruitflies, as in humans, females have two X chromosomes, whereas males have one Y chromosome and one X. Each parent contributes a copy of one of its sex chromosomes to each offspring. If males had an equal chance of contributing and X or a Y (as is approximately true in humans), then half of the offspring would be male and half would be female. But sexual equality is threatened by Dox (Distorter on X), an allele on the X chromosome that promotes its own transmission (rather than the Y chromosome) from XY fathers to their offspring. Because the offspring always get an X chromosome from their XX mother, this selfish gene leads to a preponderance of female (XX) offspring. All else being equal, it would tend to spread, leading to an all-female species and likely extinction. This could happen. This may have happened in other species. There is nothing in evolution that guarantees the survival of a species; most species that ever lived are extinct.

But in this case, Tao et al. also found another gene, which they named Nmy (not much yang), that suppresses the Y-killing activity of Dox.
The left side of the diagram is what they think happened after Dox evolved but before Nmy. The RNA or protein product of Dox kills Y chromosomes, so only female-producing sperm with the X chromosome survive. The right hand side of the diagram shows how suppression works today. Messenger RNA from Nmy binds to messenger RNA from Dox. The cell's defenses against double-stranded RNA (typically viruses) degrade it. But for RNA from Dox to bind with RNA from Nmy, they would have to be complementary. An amazing coincidence? More likely, Nmy is derived from Dox by duplication, a common source of evolutionary innovation.

Also this week, Elissa Cameron and colleagues published "Experimental alteration of litter sex ratios in a mammal" in Proceedings of the Royal Society.

It has been suggested that mothers with enough resources to produce larger-than-average offspring would have more grandchildren if they produce more sons than daughters. This assumes that big males have more opportunities to reproduce but female reproduction depends less on size, which may be true of many mammals. But how much control do mothers have over the sex ratio of their offspring?

To find out, they induced a decrease in the nutritional status of mice, using a steroid hormone that lowers glucose levels in the blood. They measured actual change in glucose during pregnancy and the sex ratio of the baby mice. Mothers whose glucose increased (almost all steroid-free) had more sons, while those whose glucose decreased (mostly steroid-treated) had more daughters. From their graph, it looks like the percentages were about 60% vs. 40%.

Of course, the steroid treatment could have had a wide variety of effects. For example, an earlier study they cite found that the same hormone reduced increases in male-biased sex ratio that were blamed on stress. But this study, and other results cited (including one on diabetes) suggest that blood glucose may be a key player in controlling sex ratio in at least some mammals.

November 28, 2007

Update: kin selection in plants and bacteria?

Two papers I've discussed that seem to show the importance of kin selection in plants and bacteria merit some additional commentary.

Susan Dudley, an author of the plant paper, was just here for a seminar, at my invitation. Someone in the audience (Jeannine Cavender-Bares, but it's not her fault if I'm misinterpreting her) raised an interesting possibility. Could the results Dudley reported (differences in root growth depending on whether neighboring plants are more or less related) be the result of roots sensing the genetic diversity of roots around them , rather than relatedness per se? This could actually be a possible mechanism of predicting relatedness without a plant needing to "know" what genotype it itself is. If three neighboring plants are identical to each other, (as indicated, perhaps, by similar chemicals exuded from their roots), maybe a bunch of seeds from the same mother plant fell in the same place, and the plant detecting the chemicals is also from the same mother plant. It should be possible to test this hypothesis by seeing how a plant of genotype A responds to 3 plants of genotypes A vs. B, (same diversity but high vs. low relatedness) and also three different genotypes A,B,C (high diversity but intermediate relatedness).

Meanwhile, John Dennehy calls our attention to a post by Rosie Redfield, who comments on the "quorum-sensing" paper I reviewed recently. She notes that the Diggle paper ignores another possible reason bacteria might release and measure "quorum-sensing signals", namely, measuring how fast an excreted enzyme disappears due to diffusion or fluid flow. She published this idea a few years ago. If you're a bacterium in a fast-moving stream, there's no point in releasing enzymes, even if you have a lot of clone-mates nearby, because the enzymes will be washed away before they can do anything useful. On the other hand, maybe even a single cell could benefit from releasing enzymes if it's in a little crevice somewhere with no fluid flow. (Has anyone manipulated fluid flow around immobilized quorum-sensing bacteria?)

Redfield also points out that the "high relatedness" treatment may be unrealistic -- essentially pure cultures, so Hamilton's r = 1. Manipulating relatedness in a way that is more representative of the real world would be nice. For example, earlier I discussed an experiment that used a thicker or thinner culture medium to control ease of movement. A thicker medium would lead to greater relatedness, as dividing cells would stay near each other.

Redfield is also the author of this gem, providing an alternative view of horizontal gene transfer in bacteria: Genes for breakfast: The have-your-cake-and-eat-it-too of bacterial transformation. J Hered 84: 400-404

November 25, 2007

Conflict over parental care

My wife and I have been watching the Planet Earth series. Week after week, mother polar bears and mother snow leopards care for their young, while fathers are either absent or dangerous. But then we got to Emperor penguins. What's the difference? This week's paper Parental conflict in birds: comparative analyses of offspring development, ecology and mating opportunities tries to answer this question.

Every baby bird -- except some turkeys -- has two parents. How much care does each parent provide for their chicks, and why? If animal behavior were ordained by a god, as a guide to human behavior, then we might expect all wild species to exhibit the same exemplary behavior. Or maybe those species that have more opportunities to interact with and influence humans -- ducks, say -- would exhibit divinely inspired behavior, while those remote from human settlements -- Emperor penguins, for example -- are left to the whims of natural selection?

The authors of this week's paper didn't waste time testing such nature-as-morality-lesson hypotheses, which are left as an exercise for creationists. Instead, they explored how parental care behaviors have evolved in response to various factors. These factors include how chicks of different species depend on parental care, and also whether a bird that leaves its mate alone to care for their chicks has additional opportunities to reproduce.

They compared parental care in 193 species of birds, using "phylogenetic comparative analysis." This method attempts to subtract out correlations that are simply the result of sharing a recent common ancestor. For example, suppose that none of the bird species that can carry a coconut is migratory. Does that indicate some intrinsic trade-off between migration and coconut carrying? Not necessarily. Maybe all coconut-carrying bird species are descended from a recent ancestor that just happened to by nonmigratory. On the other hand, if both coconut carrying and migration are widely dispersed across the bird family tree, yet these two traits never occur together, then it's more reasonable to conclude that traits that give a bird the ability to carry a coconut may also reduce the ability (or perhaps the need) to migrate.

Using a large data set, they scored the contribution of each parent, from building the nest to feeding and defending chicks. They also scored the climate where each bird species lives and the degree of polygamy (from none to common) for each sex.

They found a negative relationship, overall, between care provided by mothers vs. fathers. To some extent, this seems like an inevitable consequence of the way care scores were calculated. If males did 75% of the nest building, for example, then females did 25%. It would be interesting to see the relation between male and female parental care where each was measured as hours spent or calories consumed, but it would be hard to get such detailed data for so many species. Still, the percent of care provided by males ranged from near 0 to well over 50%. They did note that "total care is more closely correlated with [more] male care." What explains these differences?

In species where checks are fairly self-reliant (precocial), fathers tended to provide less care to their existing young when there were more opportunities to have chicks with someone else. So did mothers, when they had similar opportunities. But neither was true when the chicks were relatively helpless (altricial).

Climate was not a significant predictor of male vs. female care. I wonder about the most extreme climates, like Antarctica, where it may take extreme efforts by both parents to keep an egg and chick alive.

November 19, 2007

Biological evolution vs. word games

Each generation tends to resemble the previous one, so evolution of whales from land animals, for example, took many generations. One limitation on the power of natural selection is that each generation must be viable. Some creationist suggested that the problem is analogous to "evolving" a sentence one letter at a time to make a substantially different sentence, while requiring that each intermediate step be a valid sentence. The Mosquito Eater has solved this challenge. Cool!

But we no longer need to rely on imperfect analogies to biological evolution. Molecular tools now make it possible to explore multistep evolution experimentally, as I discussed in an early post.

November 16, 2007

Communication doesn't automatically prevent cheating

There are enough examples of ‘‘cheating’’ in bacteria ... that mindless obedience to such [quorum-sensing] chemical signals cannot be assumed. Mindlessness can be assumed, but not obedience. -- Denison et al. (2003) Ecology 84:838-845
Millions of cooperating cells can do things far beyond the ability of an individual cell. This is most obvious in multicellular organisms, whose cells cooperate because they are all genetically identical, or nearly so. Genetically diverse populations of cells could often benefit from cooperating, but do they? For example, the mixed bacteria populations associated with plant roots might benefit from keeping the plant healthy, so that it can continue to feed them with its root exudates. But for this to happen, they need some method of coordinating their plant-benefiting activities. Furthermore, cells whose genes lead to this form of cooperation must, on average, survive and reproduce more than "cheaters" who don't invest in cooperative activities. Otherwise, cooperative traits will disappear.

Quorum sensing, an exchange of chemical signals among bacteria, can solve the coordination problem. But this week's paper Cooperation and conflict in quorum-sensing bacterial populations shows that quorum sensing doesn't automatically solve the problem of cheaters. The paper is by Stephen Diggle, Ashleigh Griffin, Genevieve Campbell, and Stuart West and published in Nature.

There are two key elements to quorum sensing: 1) release of signal molecules, which build up when many signal-producing bacteria are nearby, and 2) responding to high concentrations of signal molecules by doing something that is only beneficial if many cells do it at once. For example, the pathogen Pseudomonas aeruginosa typically only produces protein-attacking enzymes when they can produce enough to be effective in digesting the lung tissue of whoever they're infecting. Quorum sensing is used to determine whether enough cells are present to do this. Mutants exist that fail to produce the signal ("signal-negative") or fail to respond to the signal ("signal-blind"). Are there any conditions under which either mutant would tend to become more common?

To find out, the authors started with genetically uniform cultures of normal (quorum-sensing "wild-type") and mutant bacteria. The only way to get energy in their "quorum-sensing medium" was by breaking down a protein, using the same quorum-sensing-controlled protease enzymes that these pathogens normally use to digest their host's proteins. Under these conditions, a pure culture of either mutant grew slower than the wild-type bacteria. Adding signal molecules increased the growth of signal-negative but not signal-blind cells. So if everyone cheats -- can you still call it cheating then? -- everyone loses.

On the other hand, when they mixed a few cheating mutants into a population of wild-type bacteria, the signal-blind cheaters increased from 1% of the population at the beginning to 45% after 48 hours. These cheats got to use the protein digested by the normal strain, without paying the metabolic cost of producing and releasing the protein-degrading enzyme. Similarly, the signal-negative mutant saved the cost of producing signal, and increased from 3% to 66% of the population.

Cheating mutants must arise all the time by random mutation, so why are they fairly rare in nature, rather than 45-66% of the population? Kin selection, maybe. If bacteria usually interact with others of the same genotype (high Hamilton's "relatedness"), then cheaters mostly cheat other cheaters and wild-type quorum-sensing cooperators get to cooperate with each other.

To test this hypothesis, the authors manipulated "relatedness" and let quorum-sensing evolve over six rounds of growth. For high "relatedness" they started each round with a single colony (Hamilton's r=1), while the low-relatedness treatment got the mix of genotypes that evolved in the previous round. Starting in round 1 with a 50:50 mix of quorum-sensing wild-type and cheaters, the frequency of quorum-sensing cells went to nearly 100% when relatedness was set to 100% at the beginning of each cycle, while it fell to about 35% when relatedness was low, as shown in this graph from their paper.
What levels of relatedness are found in actual P. aeruginosa infections? An infection starting from a single cell might be expected to have high relatedness, due to very high genetic similarity among cells reproducing by cell division. But relatedness is measured relative to the population with which individuals compete. In a Grafen diagram, an isolated population has B close to A, but P close to B, for a low value of Hamilton's r. So the authors suggest that P. aeruginosa relatedness may actually be fairly low in long-term lung infections, as in patients with cystic fibrosis. Quorum-sensing cheaters have been found in these patients. When the bacteria cycle back and forth between cooperating (or cheating) within a host, and then competing for hosts, quorum sensing should be favored.

Additional commentary and pictures at Not Exactly Rocket Science.

Also this week:

Ice-age survival of Atlantic cod: agreement between palaeoecology models and genetics

Self-Organization, Embodiment, and Biologically Inspired Robotics

Transgenerational Plasticity Is Adaptive in the Wild

November 9, 2007

When hybrids are best

This week's paper is Facultative Mate Choice Drives Adaptive Hybridization by Karin Pfennig of the University of North Carolina, Chapel Hill (where I went to kindergarten), published in Science.
Spadefoot Toad (Wikipedia)

In contrast with the “hybrid vigor? sometimes seen with crosses between different genotypes within a species -- corn, for example -- hybrids resulting from mating between related species are usually less likely to survive and reproduce. For example, when two different species of spadefoot toad mate, their daughters usually produce fewer eggs. But apparently there are situations where genes from the other species are beneficial enough to outweigh problems due to genetic incompatibilities.

When Spea bombifrons females mate with Spea multiplicata males, their tadpoles mature faster. This is beneficial in shallow ponds that don’t last long, but not in deeper ponds, which will last long enough for nonhybrid tadpoles to mature. Therefore, the smart thing for S. bombifrons females to do would be to mate with their own species in deeper ponds, but with S. multiplicata in shallower ponds. But how smart is the average toad?

To find out, Dr. Pfennig looked at the response of S. bombifrons females to recorded calls of S. bombifrons males Download file (their own species) versus S. multiplicata males Download file, to see how they respond. (Mark Bee, in my department, uses similar methods.) In deep pools, they moved towards the call of their own species about 2/3 of the time, but in shallow pools they were slightly more likely to approach the call of an S. multiplicata male. There was some evidence of individual preference also, with each female choosing the same species call 76% of the time. Toads in worse physical condition were slightly more likely to switch species, consistent with the observation that tadpoles from healthier mothers mature faster, so are less at risk in rapidly drying shallow ponds.

What controls these preferences? There are at least two aspects to consider. First, the ability to distinguish the calls of different species is needed for any kind of preference. This ability apparently evolves only when both species are present, as S. bombifrons females from areas where S. multiplicata is not found often chose the other species, even in deep-water trials. This was a little surprising to me after I got to hear how different the calls sound -- thanks to Karin Pfennig for sending them -- go back and download the audio files if you skipped them. But female preferences can apparently evolve fast enough to have changed since they migrated to (or from) areas where S. multiplicata is found.

This seems to contrast with data presented in a recent seminar here. Michael Ryan showed that, in the tungara frogs he studies, female preferences have been inherited from their fairly distant ancestors, while the “tuning? of male calls to match those preferences came later. (Hungry bats that home in those calls make mating a dangerous game.)

This is another example of the sophistication of reproductive strategies in species with very small brains. For example, female wasps that pollinate figs adjust the sex ratio of their offspring, depending on how many other females are laying eggs in the same fruit. When only one wasp is present, she lays mostly female eggs; all her sons will die in the fruit after mating with their sisters, and only daughters will carry her genes on. But if other wasps are present, her sons can carry on her genes by mating with another wasp’s daughters, so she lays more male eggs. Recently, it was found that this sex ratio adjustment is even better than we thought. When two different “cryptic species? are present, we can’t tell them apart (except by their DNA), but apparently the wasps can, and adjust the sex ratio of their eggs accordingly (PNAS 100:5867). The wasps die after laying their eggs, so we could be sure this behavior is evolved rather than learned, even if their brains were bigger.

Also this week:

Avian-like breathing mechanics in maniraptoran dinosaurs

Character-based DNA barcoding allows discrimination of genera, species and populations in Odonata

Evidence for adaptive design in human gaze preference

Facultative mimicry: cues for colour change and colour accuracy in a coral reef fish

The island rule: made to be broken?

Convergent dental adaptations in pseudo-tribosphenic and tribosphenic mammals

A Cretaceous Hoofed Mammal from India

November 2, 2007

When did social learning evolve?

Two papers this week may shed some light on human evolution. We aren't descended from modern monkeys or lemurs, but we can often learn something about our ancestors by studying our distant cousins.


If several cousins share an allele (version of a gene) that is rare in the general population, they probably inherited it from one of their shared grandparents. It's not absolutely impossible that the allele arose by independent mutations in their parents, instead, but it's much less likely. So, one of the reasons we study the genes and behaviors of related species is to figure out how far back in the human family tree a particular trait evolved. Of course, it helps to know how closely related various species are, that is, how recently a pair of species shared a common ancestor.

Molecular and Genomic Data Identify the Closest Living Relative of Primates was published in Science by Jan Janecka and others. They concluded that the closest living relative of the primates, a group which includes lemurs and monkeys as well as humans and other apes, is the colugo, the so-called "flying (actually, gliding) lemur." (I've heard of flying nuns, but primates?) They estimate that our last common ancestor lived in the Cretaceous, prior to the meteor impact that killed the dinosaurs 65 million years ago. Most primates can't glide, so this trait presumably evolved in the line that led to colugos from their common ancestor with primates.

Social diffusion of novel foraging methods in brown capuchin monkeys (Cebus apella) by Marietta Dindo, Bernard Thierry, and Andrew Whiten, was published in Proceedings of the Royal Society. They showed that, under experimentally-controlled conditions, a capuchin monkey that had been taught a particular technique for getting food could be imitated by another monkey, who was imitated by another monkey, and so on. Those taught an alternative method transmitted that method. This is the same species used in a study of symbolic token use I discussed earlier.

I will let others comment on whether the experimental setup was realistic, in terms of extrapolation from these experiments to social learning in the wild. If so, then it seems possible that the common ancestor of humans and capuchins could learn by imitating others. We would need to look at other related species before drawing that conclusion, however; octopuses can learn by imitating other octopuses, yet I doubt that our distant common ancestor could do so.

November 1, 2007

Group selection is dead; get over it!

It is the sort of thing that people like, and want, to believe. Thus, though better theories supplant it in scientific usage, we may be certain that the 'hypothesis' will persist for a while as an element of folk-science. Eventually, that remnant, too, may vanish in light of discordant facts, and the essential imagery of this once-scientific hypothesis will recede to a revered position in the popular environmental ethic, where it doubtless will do much good.

PZ Meyers suggests that David Sloan Wilson has "made a persuasive case for group selection (but then [he admits], I'm already partial to the idea anyway). " This partiality is surprisingly common, despite the lack of empirical evidence that group selection (as opposed to kin selection) is of more than negligible importance as an evolutionary force in natural ecosystems.

Theory, too, suggests that group selection can be ignored. We've known since the 1970's that the conditions under which group selection can counter individual selection almost never occur, namely: ("deme sizes of less than 25... and migration, could not be too much greater than five percent per generation" Levin BR, Kilmer WL. 1974. Interdemic selection and the evolution of altruism: A computer simulation study. Evolution 28: 527-545).

It's not that group selection can't work. For example, Muir (Poultry Sci 75: 447) increased egg production by selecting for the collective egg production of groups of 4 hens (4<25, no migration!), with reduced aggression a side effect of human-imposed group selection. But in nature, a bird whose selfish behavior leads to decimation of its flock just switches to another flock. In nature, there's no human in charge to impose selection at the group level, limit group size to <25, and prevent migration between groups.

Kin selection theory generates lots of experimentally verified predictions every year; group selection none (under nonartificial conditions). But people want to believe in it because it makes them feel good. Sound familiar? Group selectionism is a religion. (The quotation above is from Goodman's 1975 paper "The theory of diversity-stability relationships in ecology", but seemed equally applicable to group selection.)

I know it's risky to disagree with the awesome PZ, but his flame-throwing tank is no match for the 6.7-meter walls of Fort Denison.