This is the text of a talk on natural selection I gave at the University of Nicosia on February 14th 2012.
In his studies on the history of biology, Ernst Mayr divided Darwin’s concept of evolution into five integral components, summarised in the diagram on the left. The first is that organisms change with time – the pattern we see in the fossil record. Related to this is the concept of common descent, that we can recognise common ancestors. The changes between ancestors and descendants happen gradually in processes of speciation, which are driven by natural selection.
Of course, evolutionary biology has moved on a lot since then (gradual change is now known to be very, very flexible; natural selection doesn’t necessarily lead to speciation; etc.) But the concept of natural selection is fundamental and has remained unchanged. It’s summarised by the Stephen J. Gould quote: it’s nothing more than differences in reproductive success. We’ll be looking at this in more detail over this entire post.
Diagram source: Kutschera U. 2009. Charles Darwin’s Origin of Species, directional selection, and the evolutionary sciences today. Naturwissenschaften 96, 1247-1263.
Quote source: Gould SJ. 1995. Spin Doctoring Darwin. Natural History July 1995, 8, 70-71.
The logic behind natural selection is extremely simple, and combined with the entirety of the empirical evidence backing it up, this makes natural selection one of the most elegant and powerful theories in science. The logic is summarised in the curve, which depicts a population’s members on the x-axis and fitness on the y-axis (we’ll get to the definition of fitness in a few slides). Those members of the population that have a high fitness are more likely to survive and reproduce more, and so are more likely to pass on the traits that made them so fit to their offspring; the converse applies to those members at the lower extremes.
The elegance and power comes from the fact that in nature, the parameters are always changing. Population size fluctuates, and as populations get smaller, natural selection gets weaker; this Gaussian curve-style curve might be a straight line; the environment changes all the time, so a member that’s at the top one period may be at the bottom the next period; some organisms can survive but never reproduce; etc. Natural selection can accommodate all of these, and has been shown to do so empirically.
The typical example for the logic of natural selection is the classic peppered moth story. It should be familiar to all readers from high school biology: before the Industrial Revolution in England, we know from historical collections that the typica form of the peppered moth was dominant. Then the Industrial Revolution happened, resulting in tons of air pollution, which caused the lichens on trees of the time to become black from the soot. And suddenly, the carbonaria form became more dominant. Then clean air reforms were implemented, resulting in white lichens again and the dominance of the typica form in non-polluted forests. The significance of the lichens is that they serve as the camouflage background that allows the moth to hide (try to spot the moth in the right picture).
The most common explanation for this pattern is predation pressure from visual predators, first proposed based on slightly faulty experiments by Bernard Kettlewell in the 1950s-1970s. The experiments have since been redone properly many, many times and the hypothesis has been proven (alternative hypotheses, on the other hand, have not – for example, non-visual bat predation was found to have no effect). I only mention this because there is a popular trope of dismissing the peppered moth example based on Kettlewell’s faults, while conveniently ignoring the massive amounts of research vindicating his hypothesis that have been done since then.
The results of the latest of these experiments is diagrammed in the slide. The experiments were conducted by the late Mike Majerus, who did a lot of work on melanism in evolution (I fully recommend his book Melanism: Evolution in Action). What the results show is a clear demonstration that being dark is consistently associated with higher death rate in non-polluted trees. The variation is caused by abundance of non-visual predators, but even then, there is a consistent lower survival rate of the melanic form – they are selected against.
Study: Cook LM, Grant BS, Saccheri IJ & Mallet J. in press. Selective bird predation on the peppered moth: the last experiment of Michael Majerus. Biology Letters.
This was nothing more than an introduction. From now, we’ll be going to the science, starting with the definition of an adaptation, something that’s critical to establish before talking about natural selection. Then we’ll define natural selection, and see some examples of it in action (occasionally digressing to make some points about it). This will lead to one of the main debates in evolutionary theory, about the levels where natural selection acts. Finally, to round things off, I will mention what natural selection cannot do.
The reason why we need to define adaptations before anything else is because they are the end result of natural selection. The animal pictured above is a nudibranch, a relative of the snails that has no shell. One of the main purposes of the snail shell is to provide protection from predators. But here’s a very colourful animal out in the open without any apparent defences – how did this evolve?
This particular species feeds on sea anemones. While feeding, it takes in the stinging cells, but instead of digesting them, it incorporates them into its body. When a predator wants to chomp on it, it will be stung and it will get a good dose of poison. The gaudy coloration is a memorable warning to predators, so they’ll recognise the species and won’t go near it again.
This behaviour of the nudibranch, and the associated physiological and phenotypical traits, is an adaptation to selective pressure from predators, and it apparently works, since the species is stillt here, and still without a shell.
Picture source: doug.deep
This is how we interpret it in biology today. However, before Darwin’s formalisation of natural selection in his extended abstract, The Origin of Species, another theory was the norm: the argument from design, exemplified by William Paley‘s watchmaker analogy, taken from his seminal book, Natural Theology. It’s nothing more than the intelligent design argument.
Darwin’s theory of natural selection debunked that analogy, by providing a completely naturalistic and empirically-demonstrated mechanism for producing adaptations. By doing so, Darwin allowed biology to undergo the revolutions previously undergone by geology, astronomy and physics, and disabled any need for a supernatural deity to be brought in as an explanation for complexity and beauty in biology. We will look at this in a later slide, since the ID movement still surprisingly exists.
Anyway, let’s start our exploration of what adaptations are. The first example is the snake’s skull. All lizards have a mobile skull, with certain bones being moveable to some extent. Snakes take it to the extreme though. As the picture on the left shows, snakes have a very wide gape (hence why St. Exupery could show off a picture of a hat as a picture of a snake eating an elephant); this allows them to swallow prey much larger than themselves.
This is achieved by the extreme modification of the snake’s skull, in which every single bones is moveable. In addition, the bones badly darkened in the image have also been extensively modified. But the key point is that in this adaptation, no new feature has been made. All the bones in the snake’s skull are found in any other lizard skull. Natural selection does not act to produce novelty; it can only coopt previously existing features and traits and modify them (sometimes to the point where the original derivation is unrecognisable, as is the case in many parasitic taxa). In this case, it took the lizard skull bones and modified them to provide this new functionality. This example is on a phenotypical level, but the principle applies to the genetic and developmental levels as well.
Another adaptation is that of orchids. Like most flowering plants, orchids depend on insects to pollinate them. Most flowering plants offer a reward to the insect for doing so in the form of nectar. Orchids don’t. They attract the insect by mimicking either the shape or the scent (via pheromone mimicking) of the female of the pollinator. The pollinator is deceived into thinking the flower is a female, and tries to copulate with it (high speed video recordings show that the male goes throught he entire copulation – he approaches the flower, takes his penis out, and ejaculates).
This is an adaptation for the orchid. Note, however, that this adaptation doesn’t aid the orchid’s survival, only its reproduction. This is important: reproduction is the important milestone, not survival.
Also note the unfairness of this adaptation. The orchid gains the advantage, while the insect loses a lot – it takes energy to fly to the plants, energy usually recuperated from drinking the offered nectar; the pseudocopulation also takes some energy. It’s a big loss for the insect (and he knows it; a single male will only be fooled by it up to 10 times, depending on the species).
Picture source: unknown. Pass me a link/takedown notice please!
This next adaptation expands on the idea that survival isn’t a prerequisite for natural selection. The organism concerned is the redback spider, a highly-poisonous endemic of Australia. These spiders have a pecular mating ritual: after copulation, the male will jump straight into the female’s mouthparts and get willingly cannibalised. At first glance, this doesn’t seem like an adaptive behaviour for the male at all. But when Andrade (1996) tested the patternity of offspring from cannibalised versus non-cannibalised males, she found something incredible: cannibalised males fertilised almost all the eggs of the mated female, whereas non-cannibalised males had a much lower success rate.
What this tells us is that this behaviour is adaptive, even though the male doesn’t survive. He gets to pass on his genes, and that’s all that matters.
Picture source: PacificKlaus
Study: Andrade MCB. 1996. Sexual Selection for Male Sacrifice in the Australian Redback Spider. Science 271, 70-72.
This final adaptation showcases a more unique trait that some adaptations might have. The pictured ants are weaver ants in the process of beginning to build their nest. Two groups of ants stands on each end of a leaf and grab the edge with their mandibles. Across the leaf, other ants form a line, all holding each other with their mandibles, and they pull to bring the 2 edges next to each other. Meanwhile, standing nearby off the leaf are other ants with larvae held in their mandibles. When the leaf edges are close, the larvae shoot silk out of their abdomen. The silk acts as glue and stitches the leaves together. This is repeated with many leaves to form the nest.
Such elaborate behavioural patterns can only be adaptations. How they’re selected for is an area of debate, and we’ll touch on it later; what is clear is that some adaptations, such a this one, are beneficial not only at an individual level, but also at the level of a tribe or group. Also, the fact that the larvae of Oecophylla can shoot silk is a cooption of the ancestral feature of using silk to build the pupation coccoon.
Picture source: wildsingapore
The examples you have seen so far are all proven adaptations. There was a time in biology when it was common to see any behaviour or any structure as being an adaptation, a viewpoint called adaptationism.
Adaptationism is now mostly dead, because it’s an extremely flawed perception of biology, as shown by Gould & Lewontin (1979), a classic paper that is rightly obligatory reading for all biologists (although see here for some good critique!).
Adaptationism fails because of one critical fact: natural selection is not the entirety of evolution. Evolution is a framework comprising many different processes; for adaptationism to be valid, then natural selection must be one that is always acting the strongest. And this simply goes against all the evidence – natural selection certainly is powerful, but it’s not omnipotent, and is overshadowed by genetic drift under the right circumstances.
The downfall of adaptationism made several things very clear and set the biologist’s research agenda straight. We can no longer assume that natural selection is the null hypothesis; natural selection is something we must prove have acted, and only then can we say that a trait is adaptive.
And this burden of proof is very important, because there are many ways in which a seemingly adaptive trait could have arisen without natural selection. For example, a trait might be a side-effect of selection acting elsewhere. An example fo this is with blood. Our blood is red because it contains iron. Natural selection acted on maintaining the iron, since the haemoglobin molecule is built around it, and without iron, the oxygen-carrying capacity of the blood is diminished. The resultant red colour of the blood is not an adaptation; it’s merely a side-effect.
A trait might be phylogenetically ancient and only present due to developmental constraints. That tetrapods have four limbs isn’t an adaptation in itself; it’s a consequence of tetrapods having evolved from four-limbed fishes.
The way a trait is currently used may not be the way it was used when it first evolved. Traits we identify as predaptations or exaptations are examples of this. For example, many catfish species have very well-developed non-visual sensory systems, adaptations to their nocturnal behaviour. That cave-dwelling cave fish retain these isn’t an adaptation to cave-dwelling; they simply took the existing systems and used them in their new situation.
The final large problem with adaptationism is that in order for every trait to be adaptative, we have to view every trait uniquely. This is simply impossible. Morphology is the result of massive networks influencing each other during development, and development is the result of massive gene netowrks influencing each other (and also being influenced by environmental conditions). Having a single gene:trait linkage is extremely rare; pleiotropy (genes having cascading, knock-on effects) is the rule.
So basically, what this means is that to declare a trait an adaptation, proof is needed. You can’t just make up stories, even intuitive-sounding ones, and call your traits adaptations.
Here is an example of how to prove an adaptation. The diagram above shows the sequence of some segments of chromosome 7, sequenced from European-Americans and African-Americans. I’ll break down each row from top to bottom. A shows 4 genes: EPHB6, TRPV6, TRPV5, and KEL. AA and EA SNPs indicate the presence of single nucleotide polymorphisms (AA = African Americans, EA = European Americans). B-E show various metrics, with red being the value for European Americans, black for African Americans. B show PD, the frequency of the derived SNPs. C shows Tajima’s D, which can be taken as an index of non-neutrality (i.e. it shows the strength of selection). D shows nucleotide diversity. E shows FST, a measure of the differences in allele frequency; the red line shows the average FST and the green line shows the significance threshold.
Study: Akey JM, Eberle MA, Rieder MJ, Carlson CS, Shriver MD, Nickerson DA & Kruglyak L. 2004. Population History and Natural Selection Shape Patterns of Genetic Variation in 132 Genes. PLoS Biology 2, e286.
What’s of interest to us for this example is TRPV6. Notice that European Americans have a high number of derived SNPs in TRPV6, with a low index of non-neutrality (i.e. a strong index for selection), and low nucleotide diversity. These are tell-tale signs of a selective sweep, when natural selection acts strongly on the genome to promote only certain mutations (in this case, the derived SNPs) and getting rid of others (hence the low nucleotide diversity).
It’s interesting that this took place in the European-American version of TRPV6. TRPV6 is a gene that codes for a protein involved in calcium absorption. We know from archaeology and anthropology that European populations adapted milk as a staple part of their diet. This is a genomic adaptation to that diet.
This is how you show an adaptation: you prove that natural selection has acted, and then you muster up evidence from relevant areas of biology to support the hypothesis.
Now that we’ve got adaptations – the results of natural selection – out of the way, it’s time to get to natural selection itself. As I said before, natural selection is not synonymous with evolution, it’s just one of the mechanisms acting during evolution.
There are many slightly-varying definitions of natural selection; we will be using the one from Futuyma (2009):
Natural selection is any consistent difference in fitness among phenotypically different classes of biological entities.
Lots of clauses there, so we’ll break it down into its constituent parts in the next slides.
Source: Futuyma DJ. 2009. Evolution, 2nd ed.
First and foremost is the concept of fitness, a concept first formalised by the great Ronald A. Fisher in his landmark 1930 book, The Genetical Theory of Natural Selection. Fitness is nothing more than a synonym for reproductive success, and is split into three components: probability of survival to reproductive age, and average number of offspring produced by male input (e.g. sperm) and by female input (e.g. egg).
Notice the language used: probability and average. These are terms that come from statistics. You can’t talk about the fitness of an individual except by comparing it to a baseline calculated from the general population, because you can’t do statistics when n = 1.
The larger implication of this is that we refer to natural selection only in population terms – it’s all about differences in allele frequencies.
The next part of the definition is the part about different classes of biological entities. This is nothing more than an allusion to the levels of selection debate, i.e. at what level natural selection acts. We’ll touch on this subject again later on, but suffice it to say for now that natural selection can act on any conceivable level, at least in theory.
By far the commonest level is individual selection, simply because it’s the individual that survives and reproduces. All other cases are exceptions. We have group selection, where natural selection acts on the level of a group; its power is under considerable debate. On the other hand, we have gene-level selection, in which natural selection acts at the level of the gene; this is relatively rare, since it often leads to extinction. A combination of sorts between group- and gene-selection is kin selection, where natural selection acts on a group of genetically-related individuals. On the largest scale, we have species selection, one that can only be detected using palaeontology or other historical data.
But one can talk about any level of selection, as long as one can prove it. For example, cancer can rightly be seen as natural selection acting on the cellular (somatic vs. cancer cell) or tissue (tumour vs. regular) level. There is some talk of ecological selection, an extreme of which is the Gaia hypothesis.
The next bit of the definition is the part about phenotypically different classes of biological entities. This basically means that we have to have differences between our genes (in the case of genic level-selection), our phenotypes (in the case of individual selection), our groups (in the case of group selection), etc.
Despite this, we always talk about natural selection in terms of genes and alleles. The reason is simple: they’re the most practical and objective level at which we can measure selection, and all our tools are built around them. This is why we take a “gene’s eye view”, even in cases of individual or higher level selection.
Finally, no matter what level natural selection is acting on, the response to selection is always phenotypical – there has to be a tangible, visible change to the structure of the gene, to the character in the individual, to the trait of the group, for natural selection to act on.
So basically, here is a diagram summarising natural selection. You have a population with variety in it, variety that can be genetically inherited. A selective pressure comes along, and only those variants that are most fit manage to reproduce. Repeat ad infinitum (or for 4+ billion years).
Diagram source: Gregory TR. 2009. Understanding Natural Selection: Essential Concepts and Common Misconceptions. Evolution: Education and Outreach 2, 156-175.
Going into the entire history of natural selection would be overkill, but I do want to just go through the landmark events of natural selection theory.
Before Darwin, there was only one comprehensive evolutionary theory, that of Lamarck. It viewed organisms as being moulded by their environment, with an implicit trend towards progression and perfection. The way it worked was that organs that were used more would get extended; those used less would shrink. The mechanism was some “neural fluid” inside the organs.
It was a nice try, and he did gather up some evidence for it (the mechanism was way off though), but it was ignored by 19th century biology, due to both sociopolitical reasons and due to lack of scientific merit. The view that dominated before and after Lamarck was that of essentialism – species were static, unchanging from when they were created (or somehow came about).
The other landmark was the publication of Malthus‘s Principle of Population essays. Though it predated Darwin’s work, it was both a sign of the zeitgeist of the time when competition was viewed as an integral part of life (see my Darwin post, search for “As an aside”), and when Darwin did get around to reading it “for amusement”, it made all the pieces of the puzzle click in his head and gave him “a theory by which to work”.
The development of Darwin’s thoughts can be traced fairly precisely because of his meticulous notekeeping and extensive correspondence. He first began having transmutationist thoughts while on the Beagle, and when he returned to England, he opened up the first of four Notebooks on the Transmutation of Species in 1837. You can look at them all here, look for Notebooks B-E.
Here he outlined his thoughts as he had them – we have the first iconic “phylogenetic tree” he drew in these notebooks, for example. As trivia, that’s not actually a tree, it’s based on the structure of corals.
Anyway, from the end of the last notebook, he did most miscellaneous biological work and also began compiling as much evidence for transmutation as he could. The rest of the story is well-known: he received a letter from Wallace outlining a very similar theory of natural selection; they presented them together; Darwin rushed to release his ideas to the public, in the form of an extended abstract that became one of the foundational books of biology. Darwin’s actual book on natural selection is incomplete and has been reconstructed from whatever was found of his notes, and is pretty spectacular.
In any case, I think it’s interesting to do a comparison of natural selection as we understand it today versus Lamarckism as it would be understood today. Lamarckism nowadays isn’t ignored because of historical factors, but because it was disproven experimentally at the start of the age of bacterial genetics.
The only difference between natural selection and Lamarckism is the notion of directionality: natural selection follows no direction, while Lamarckism is implicit in a strive towards progression.
We can see this at the genetic level, when looking at molecular evolution. Before the 1960s, it was thought that mutations could only come in two flavours: good or bad. Most mutations were bad and disruptive, some mutations were good. This is called the selectionist view (you can see it as an analogue of adaptationism), because natural selection was always equally strong. This is obviously wrong, and it was taken down by Motoo Kimura (1968) where he outlined the neutral theory of molecular evolution. In this view, mutations come in three flavours: bad, good, and neutral. Neutral mutations simply have no effect, and selection thus doesn’t act on them.
Neutral theory wins out because that’s what the data shows. It was expanded by Kimura’s student, Tomoko Ohta, in Ohta (1973), where she proposed that the neutral mutations can also be split into two types: true neutral ones, and nearly neutral ones that have a very slight, almost negligible, effect on fitness.
Neutral Theory: Kimura M. 1968. Evolutionary Rate at the Molecular Level. Nature 217, 624-626.
Nearly Neutral Theory: Ohta T. 1973. Slightly Deleterious Mutant Substitutions in Evolution. Nature 246, 96-98.
Diagram Source: Bromham L & Penny D. 2003. The modern molecular clock. Nature Reviews Genetics 4, 216-224.
That said, there is a kind of notion of progression in natural selection, but it’s of a different nature than that proposed by Lamarckism. R.A. Fisher first proposed it in The Genetical Theory of Natural Selection as the fundamental theorem of natural selection, and says that natural selection increases the mean fitness of a population. In other words, it merely states that deleterious mutations will tend to be more deleted by natural selection.
The history of the FTNS is one of acceptance, then rejection, then resurging acceptance. I’m one of those who accept it, perhaps not as R.A. Fisher first forumlated it, but as an explanation for certain patterns of natural selection (the ones you learn of in high school), as seen in the next slide.
These patterns are the typical disruptive, stabilising, and directional selection, where natural selection will change the general phenotype of a population by selecting for specific phenotypes in response to the environment.
Disruptive selection can happen when multiple niche polymorphism develops in a population. An example could be when the same species develops two phenotypes adapted for eating different foods, as seen in the black-bellied seedcracker, a bird which has a form with wide bills and another with narrows bills. Wide-billed birds can eat hard seed, narrow-billed birds can eat soft seeds; those in the middle aren’t efficient at eating either hard or soft, so natural selection has led to only selection of the extremes.
Study: Smith TB. 1993. Disruptive selection and the genetic basis of bill size polymorphism in the African finch Pyrenestes. Nature 363, 618-620.
Stabilising selection is when the currently dominant phenotype gets reinforced. An example of this happening is with established tolerance to extreme conditions. Say you have a soil with heavy metal contamination. The plants that are adapted to such conditions will be fine, but any deviation from their adaptations will be deleterious.
Directional selection is probably the most relevant to applied evolution: it’s when the complete opposite phenotype gets selected for, and it’s what happens in antibiotic or pesticide resistance. Before pesticide is applied, the resistance-confering genes are neutral. When pesticide is applied, they are suddenly selected for, and only the resistance gene-carrying phenotype survives.
So, to go back to the Lamarckism-natural selection comparison, a final slide with a direct comparison, using the typical example of the giraffe.
Under Lamarckian evolution, a short-necked giraffe would not be able to reach taller branches with more food. So the neck will lengthen, and the offpsring are born with this longer neck.
Under natural selection and evolution as we understand it today, you have a population of giraffes in which all sorts of neck sizes coexist. Those with the taller neck can reach the higher branches with more food, and so are more likely to have the enrgy to survive to reproductive age. On average then, they will reproduce more than those with the shorter neck, passing on the developmental program that leads to a longer neck (note: this assumes that it is developmental, not environmental; just go with it for the sake of the example!) and so long-necked giraffes will become more common in the population.
The point of this exercise is to show that natural selection itself has no creativity or drive, even taking into account the FTNS – it’s merely a matter of statistical difference. Natural can explain why there are long-necked giraffes, but it can’t explain why the giraffes have long necks – it doesn’t care about that.
I do want to expand on one issue that’s an active area in current research (and one that I’m involved in in my own research): the link between the genome and the phenotype, and how natural selection plays into it.
The genome is a giant set of raw materials. Development takes those raw materials and puts them into modules and combines them to form the phenotype; these modules are typically preserved and recycled. The phenotype then can inform the ecology of the organism, and the ecology can also modify the phenotype.
Natural selection acts on all these levels. A screw-up in development can lead to a fatally-deformed phenotype. A phenotype can be unfit. The behaviour of an animal can be deleterious. The unfit organisms don’t contribute to the gene pool, and so the evolutionary response plays out on the genome, leading to conservation of good developmental patterns, good phenotypes, and good ecologies.
An example of the above scheme can be seen in the Himalayan rabbit, a breed of rabbit which differs from the regular rabbit with one important mutation in one of the enzymes that produces melanin. The mutation causes melanin to be produced only in cold temperatures.
So what will happen is that melanin will only be produced at the coldest parts of the body – the extremities, explaining the coat colour of the animal. This will then provide an ecological advantage by allowing for camouflage in the snow with the white colour, leading to overall positive selection for this mutation, and the evolutionary response is the maintenance of the mutation.
This particular example is one of developmental plasticity, so it’s not quite ideal, but it’s a cool story nonentheless.
Now that we’ve looked at the general principles of natural selection, it’s time to look at it in action, with examples showing more of these principles.
The first example we’ll look at is a classic of evolutionary biology which came out of Sol Spiegelman‘s lab in the 1960s using the Qß phage, a virus that infects E. coli bacteria, using the bacterium’s resources (replicases and nucleotides) to replicate itself. The study here is that by Mills et al. (1967).
What they did was set up a culture of Qß in Petri dishes containing replicases and nucleotides, so they can replicate themselves without having to infect E. coli. The only limitation that the Qß face is one of resources, so this sets up a selective pressure towards faster replication. After a set amount of time, a sample Qß from the dish was transferred to brand new dish, establishing a new population from the gene pool of the most successful variants in the previous dish. This was repeated for 75 transfers.
The trend is extremely clear. With each successive transfer, the Qß phage population exhibited a faster rate of growth. In addition, as early as the 5th transfer, the Qß had lost its ability to infect E. coli (see inset diagram). Finally, by the 75th transfer, the Qß had shrunk in size from the original 3300 nucleotides to 550 nucleotides. This latter point highlights an important philosophical distinction.
Study: Mills DR, Peterson RL & Spiegelman S. 1967. An extracellular Darwinian experiment with a self-duplicating nucleic acid molecule. PNAS 58, 217-224.
This is the issue of selection of versus selection for, which Elliott Sober, a prominent philosopher of biology, did a lot of work on; I recommend his book The Nature of Selection for more info. To explain it, we’ll use the above experiment’s results: by the end of it, there was faster replication speed and smaller size.
Because the experiment was set up specifically, we know that there was selection for high replication speed, and the selection of small size was a side-effect of this – a smaller virus is obviously going to replicate faster.
In other words, the effect of selection for speed is selection of small size; there was no selection for small size. On the surface, it may seem like merely a semantic distinction, but it has implications for how we explain the origin of adaptations – we need to know the conditions under which selection happened to find out the real trait that was selected for.
Qß in replicase culture has proven to be an excellent model system for experimental evolution. In this next experiment, Saffhill et al. (1970) added a stumbling block for the Qß to go through: a replication inhibitor, ethidium bromide (EtBr). The diagram on the left shows the results. At every (x), more EtBr was added to the serial transfer solution; (a) was the first addition at 4 µg/ml, and it increased until (i), where 50 µg/ml were added.
The results are clear: when EtBr is first added, replication is severely affected and the growth rate crashes. But by the next transfer, resistance to the EtBr evolves, since it’s very strongly selected for. But it is clear that resistance will only get you so far, and by (I) replication is almost impossible.
Study: Saffhill R, Schneider-Bernloehr H & Orgel LE. 1970. In vitro selection of bacteriophage Qβ ribonucleic acid variants resistant to ethidium bromide. Journal of Molecular Biology 51, 531-539.
Kramer et al. (1974) investigated the genetic basis for the resistance, and found that three mutations are responsible, termed mutations α, ß, and γ. The diagram on the right shows them on the complementary strands.
Study: Kramer FR, Mills DR, Cole PE, Nishihara T & Spiegelman S. 1974. Evolution in vitro: Sequence and phenotype of a mutant RNA resistant to ethidium bromide. Journal of Molecular Biology 89, 719-728.
What Kramer et al. (1974) also saw was that these mutations always come in the same pattern: ß is first, then comes γ, then α. To explain this, we have to invoke the abstract concept of the adaptive landscape. Imagine a field with several high hills. The field represents the entire spectrum of mutational possibilities. The hills (i.e. the y-axis, or z-axis if you’re in 3D-mode) represent adaptive value or fitness; the higher the hill, the more adaptive the mutation.
When one of the viruses acquired mutation ß, it’s basically the same as hammering in a safety hook in mountain-climbing: it’s started climbing one particular hill. From there, the adaptive possibilities gradually get less (these are peaks, not plateaus); and so mutation γ, then α, pop up successively, as that’s the only way up this particular mountain.
For someone not familiar with this widespread concept of adaptive landscapes, and who also has a beef with evolutionary biology for whatever reason (stupidity and/or ignorance is the commonest), such a specific sequence stinks of design. They’re obviously following some predetermined path laid out for them by a supernatural force, and chance could obviously not have played any role. I chose a random schmuck to illustrate this viewpoint from my “Idiots” bookmark folder, Robert Herrmann, a professor of mathematics.
This is pure, unadulterated bullshit. The appearance of design is merely an artefact of a zoomed-out perspective, seeing only the “final product”. In fact, what happens in this case when climbing up the adaptive hill isn’t that every individual virus will gain mutation ß. Most will die off, because mutations are random – it’s a giant process of trial and error, because copying genetic material is always imperfect (there is no such thing as a clone).
These mutations, some of which may be mutation ß but will also include thousands of other mutations, lead to differences in the virus’s performance and survival. Those that survive to reproduction end up acting as templates for the offspring, which will also add on new mutations. When you see this whole population on the adaptive landscape, most will be on the flat land, some will be vlimbing hills, and some will have fallen into ditches.
In other words, all these templates reproduce. Those on the ground reproduce at a rate of 0.2 (number made up); those in the hills have higher rates depending on how high up they are; and those in the ditch reproduce at rate 0, i.e. not at all. The latter will not be passing on any of their mutations.
And this is how natural selection acts. It’s just the result of accumulated differences and selection of the most advantageous combnations, over many, many generations. While the mutational basis is mostly chance (a mutation is a random event, although there is a slight predictable basis to it, but only probabilistically – you will never be able to foresee what mutation will happen where, only likelihoods), the selection of mutations is the very antithesis of chance.
The next example we will look at is another classic of evolutionary biology, from one of the architects of the Modern Synthesis, Theodosius Dobzhansky. The organism involved is Drosophila pseudoobscura, a species characterised by having chromosomal inversions according to geographical location.
Inversions are sequences within chromosomes that can be flipped upside down (inverted), leading to new combinations. Obviously, they will have an effect on fitness, and so natural selection can be conceived as acting on them. What Dobzhansky did was keep flies with two different polymorphisms (ST and AR) in cages together in a constant lab environment, and measured the frequencies of the polymorphisms as they changed every generation.
Study: Dobzhansky T. 1948. Genetics of Natural Populations. XVIII. Experiments on Chromosomes of Drosophila Pseudoobscura from Different Geographic Regions. Genetics 33, 588-602.
His results show that not any polymorphism will be preferred. The two curves refer to two founder populations, one with a low frequency of ST and one with a high frequency of ST. By the end of the experiment, both populations had converged on very similar levels of ST in the population (and conversely similar levels of AR).
These results illustrate another principle of natural selection: as the FTNS predicts, natural selection will act to raise the fitness of the population. In many cases, such as this one, fitness is increased by having genetic diversity, and natural selection will aim to preserve and enhance that.
The next experiment is a textbook study of natural selection in action, coming from John Endler‘s lab, and serves to demonstrate the role of trade-offs in selection. The study organism is the male guppy. Male guppies have large, conspicuous spots on their body used to attract females – the more spots, the more attractive.
Endler (1980) took a founder population of guppies and put them in a tank by themselves with no predators, and let them reproduce for 6 months. He then took two batches from that population, putting one in a tank with a weak predator, and the other batch with a strong predator (their natural enemy Crenicichla, typoed in the slide). His aim was to count the evolution of the number of spots on the male guppy; he left the initial population with no predators as a control.
The results show that in the populations with weak to no predation, the number of spots increased to an average of 13; those with no predation reached to 14-15 spots, while those with weak predation were more limited in range. Those under strong predation experience the complete opposite, and their number of spots sunk below that of the initial wild population.
Guppy picture: tartaruga33
The explanation for this pattern is simple. The number of spots make the male easier to spot by the females, but the same is true for predators. In other words, there has to be a trade-off between promiscuity and camouflage. In the control’s case, there is no trade-off, since predation isn’t a risk – males with more spots will reproduce more than those with less spots, and so the number of spots will increase in the population.
In the case of weak predation, the reproductive success outweighs the predation rate: those with more spots can still manage to reproduce because their predators are stupid. But in the case of strong predation, those with more spots get spotted more easily, and get eaten. Hence those with less spots reproduce more often and the number of spots decreases in the population.
These trade-offs happen all the time in biology and natural selection. Everything comes at a price, be it promiscuity, energy consumption, time, etc. Natural selection isn’t a conscious process, it merely refers to those organisms that happened to have the correct combination of trade-offs.
The above study is also an example of the most common special case of natural selection: sexual selection. This is when a selective pressure comes from the mate, most often the need to attract a mate. A famous example is the male peacock’s tail: the bigger and flashier it is, the more attractive it is for a female. The guppies in the above study are another example where promiscuity is the attractant. In many animals, combative ability can also be rewarded; see mantis shrimp as an example. Elk antlers are another example of a character selected for by sexually-motivated combat. Of course, there is also the potential for sneakiness to be rewarded: in some dung beetles, while the big-horned males are duking it out, small, hornless males will sneak up to the female, copulate and fertilise all the eggs. These all fall under the category of sexual selection, and all cases of sexual selection involve trade-offs between the need to attract mates and the need to survive (by not attracting predators, or by having enough energy to mate after building enormous horns).
Most of the time, it’s females who choose males, but sometimes it’s males who choose.
Peacock picture: Henry McLin
The next two slides will showcase two recurrent phenomena of natural selection. The first is Batesian mimicry, a type of mimicry that occurs when an unrelated and non-toxic species mimics in appearance a toxic species, in a bid to fool predators into not eating them.
It was first discovered by Henry Walter Bates in South America, where he observed it among species of heliconiid butterflies (Bates, 1862); look at the bottom right pair as an example – those are two different species that look extremely similar; one is toxic, the other isn’t. A predator would not take the chance.
When Charles Darwin first saw Bates’s publication, it immediately struck him as an excellent example of natural selection and he mentioned it in the next edition of the Origin (4th).
Mimics can belong to the same genus or family, but can also be completely unrelated – there are spider-mimicking ants, ant-mimicking spiders, snake-mimicking moths, wasp-mimicking beetles, bee-mimicking flies, etc. They all evolve by natural selection, with the more successful mimics being selected for and passing on their good mimicry.
In some cases, as with the Amazonian heliconiids, mimicry rings can form, encompassing a large geographical area. In these, you get a sort of musical chairs happening, with two geographically adjacent species resembling each other and going through a chain all around the large geographical area (e.-g. the Amazon). If you look at the northernmost species and the southernmost species, they don’t look at all like each other; but you can link them through mimics.
First discovery: Bates HW. 1862. Contributions to an insect fauna of the Amazon valley (Lepidoptera: Heliconidae). Transactions of the Linnean Society of London 23, 495-556.
Ring mimicries are spectacular, but rare (or hard to investigate/find). What is very clear is that mimics must inhabit the same area as the models, so that the predators can be fooled, as seen in the study above.
However, that study is not a case of Batesian mimicry, but of Müllerian mimicry. This is when toxic species mimic each other; this evolves by exploiting the mutual advantage gained. It was first discovered by Fritz Müller, a German biologist who emigrated to Brazil, again in heliconiids.
The example above involves a hemipteran genus, Dysdercus, in South America. There are several phenotypes present in Dysdercus, the above shows phenotype A, and its geographical and systematic distributions (black branches are those that have it). As you can see from the tree, it evolved convergently in several unrelated taxa; this is how powerful the mimicry is. And as can be seen from the left map at least, the mimic inhabits the same area as the model.
Study: Zrzavý J & Nedvěd O. 1999. Evolution of mimicry in the New World Dysdercus (Hemiptera: Pyrrhocoridae). Journal of Evolutionary Biology 12, 956-969.
The above examples involve natural selection modifying the appearance of the organism to evade predators, in those cases by mimicking other organisms. There are also the cases when natural selection causes the organism to mimic its environment, by evolving camouflage. The example I will be showing is a pretty spectacular one, and one I tpredict will be in the next generation of high-school textbooks.
The setting is the White Sands National Monument in New Mexico, USA. As the name and the picture say, it’s an area full of white sand. Regular, yellow sand is derived from quartz crystals or eroded shells/microfossils. In White sands, it’s derived from gypsum. Anyway, the areas surrounding White Sands is regular shrubland.
Picture Source: Almond Butterscotch
The picture above summarises the entire case study. In A, you see three species of lizard. The ones at the bottom live in the shrubland, and are appropriately coloured to blend in. The top pictures show the same species of lizard, but the variety that lives in the White Sands. As you can see, they all convergently evolved to become white, as camouflage.
The remarkableness of this convergent evolution doesn’t stop there. Rosenblum et al. (2010) also studied the genetic basis of the white colour, and what they found was that while the mutation in each species was different, they all occurred in the same gene, and had effects in the same part of the resultant protein.
The protein in question is called MC1R, and is diagrammed in B. It’s a hormone receptor that spans the cell membrane, and is integral for melanin production. Each red dot is the responsible mutation for the respective species. Notice how they’re all in the transmembrane segment of the protein. This is an amazing case of convergence not only at the phenotypical elvel, but also at the molecular level. This is why I think this will soon be a typical textbook example.
As an aside, humans also have MC1R; mutations in it are what cause ginger hair (and presumably the associated loss of soul).
Study: Rosenblum EB, Römpler H, Schöneberg T & Hoekstra HE. 2010. Molecular and functional basis of phenotypic convergence in white lizards at White Sands. PNAS 107, 2113-2117.
The previous example of camouflage is the usual case where the appearance is set regardless of environmental conditions – the White Sands lizard morphs will be white even if they’re born and raised in the shrublands. There are cases, however, where appearance can vary depending on the environment, as a result of phenotypic plasticity. Even further, some of these environmental morphs can be so advantageous that natural selection will select for their occurence, leading to adaptive phenotypic plasticity.
An example of this is the Junonia (Precis) octavia nymphalid butterfly of Africa. It has two adult morphs, a blue one that emerges in the dry season and an orange one in the wet season. The molecular basis for the colour difference is ecdysteroid production, a hormone regulating the wing colour patterns. Studies have shown that there is only one factor that causes the appearance of each form: temperature (McLeod, 1968), corresponding to the temperature of each season.
The way this evolves is that there is some advantage to being orange in the wet season, and an advantage to being blue in the dry season. The initial mutations that caused this difference have to do with ecdysteroid production according to temperature, and the advantages gained caused these mutations to be selected for. It’s not as if the butterfly decides it will be orange now; it’s a pathway that was cemented by selection during the evolutionary history of the species.
Similar pathways can be found in survival instincts. For example, a deciduous tree drops its leaves when winter comes. This isn’t a “decision” taken by the tree. It’s merely a response to lower temperatures (or some other abiotic factor) that triggers biochemical processes that lead to the dropping of the leaves. The advantage is that those trees that drop their leaves have a much better chance of survival, and thus they pass on the mutations that lead to the leaf-dropping-at-low-temperatures to their offspring.
J. octavia study: McLeod L. 1968. Controlled environment experiments with Precis octavia Cram. (Nymphalidae). Journal of Research on the Lepidoptera 7, 1-18.
Much more famous cases of adaptive phenotypic plasticity are found in all the eusocial insects – those insects that have castes specialised for different functions, e.g. ants, termites and bees. The slide shows several examples of ants, one wasp and one termite, with striking differences between the castes.
These individuals share the same genome. These differences are not caused by a mutation. Instead, they’re caused by differences during the larval and pupation stages – they are given different types of food, raised in different temperatures, etc., and these are what determine the adult’s phenotype. These stereotypical phenotypes and the pathways leading to them were selected for by natural selection, back in the early evolutionary history of the animals (we have fossil evidence that castes have existed since the earliest ants and eusocial bees).
Diagram source: Whitman DW & Agrawal AA. 2009. What is phenotypic plasticity and why is it important? In: Whitman DW (ed.). Phenotypic Plasticity in Insects.
The last example of natural selection is one from the fossil record. I choose this on purpose because despite the “palaeobiological revolution”, many evolutionary biologists still consistently ignore palaeontology (in my experience, I don’t mean to stereotype), so I like to point out whenever possible that palaeontology can offer insights into every evolutionary phenomenon.
The example I chose is a classic one, the Steinheim Basin snails. I already wrote an extensive summary in two posts, so refer to them for the background info. Basically, they come from a crater lake that persisted for hundreds of thousands of years. Such lakes are referred to as ancient lakes and are highly sought-out by evolutionary biologists, because the taxa in them have remained isolated for a very long time, and evolution went by undisturbed; they’re natural test tubes. A modern example is Lake Tanganyika, where cichlid fish have undergone an amazing radiation with much specialisation (one species has become adapted to eating only the eyeballs of other fish while they’re alive).
The Steinheim Lake has a similar story, except with snails. It was first discovered and described by Franz Hilgendorf, who recognised that the Gyraulus snail species in the oldest sediments had undergone a progressive radiation, and drew a phylogenetic tree to support his thesis. It had a tumultuous history, with most of the 20th entury workers saying that there was no radiation, just anagenetic change due to environmental conditions, but this is now considered obsolete – refer to the posts for more info.
The changes observed in the snails are a tickening of the shell and an increase in the sculpturing. We know from modern snails that these are adaptations to predator pressure – they make the shell harder to break. Through mapping of the formation, we know that the radiative bursts (when the new snail species formed) are correlated with water level changes. Related to these is presence of predators, especially a type of fish found there with teeth specialised for breaking snail shells.
As the water level rose (and associated predator abundance rose), the snail shells became thicker and more sculpted. When the lake started evaporating, the snails reverted to become smaller (see 7, 8, 9 in the tree). This is exactly what we would expect if the snails were responding to predator pressure – as the pressure lets up, the adaptations start disappearing (it’s the same pattern observed in cave species: pressure for vision goes away, eyes disappear).
Another thing that palaeontology contributed to evolutionary theory and natural selection is the notion of species selection. This is still under debate among neobiologists, especially those who only consider population genetics. I don’t know why; I attribute it to lack of the temporal perspective gained when studying palaeontology (the alternative is that they have no respect for palaeontology, but I don’t want to stereotype).
Species selection was proposed in order to explain a peculiar pattern found in the fossil record, and diagrammed on the right. On the y-axis, we have time, and on the x-axis, we have disparity. The pattern is left behind by the increased success of some clades over other clades, leading to those successful clades surviving longer and thus having more opportunity to speciate, while the non-successful clades go extinct.
In other words, one can think of it as natural selection on the individual level, but with death and reproduction replaced by extinction and origination.
Original paper: Stanley SM. 1975. A theory of evolution above the species level. PNAS 72, 646-650.
Diagram source: Stanley SM. 1979. Macroevolution: Pattern and Process.
In fact, that is more than an analogy, it is how species selection is envisaged as working. Species selection considers individual selection at its base. Individual selection, as we said at the beginning, is the most common type of selection, relying on the survival and reproduction of individuals.
In species selection, the emergent properties of a population are where natural selection acts out. These are properties of the population as a whole, such as geographic range, dispersal ability, population size, generation time, etc. I chose three examples that showcase the validity of species selection.
On the left is a study on molluscs, where the geographic range of fossil and Recent molluscs. The plot shows the essential result: as the range grows larger, the speciation rate drops. This is a pattern observed only on the species level, and other studies have shown it in other clades.
Study: Jablonski D & Roy K. 2003. Geographical range and speciation in fossil and living molluscs. Proc. R. Soc. B 270, 401-406.
In the middle is another emergent property: population size. The study uses data on birds in the UK, an the plot shows the number of nesting pairs (a surrogate for population size) on the x-axis, and extinction risk on the y-axis. The trend is very clear – the larger the population, the greater the extinction risk (as predicted by theory). A lot of other correlations are found in the paper (e.g. body size, migration), and all are indicative of such emergent properties having a large effect on extinction rate.
Study: Pimm SL, Jones HL & Diamond J. 1988. On the risk of extinction. The American Naturalist 132, 757-785.
The last study on the right is also the most obvious. It shows the expansion of insectan ecology through time, through their mouthparts. As time went on, insects became more and more specialised ecologically, as reflected by their mouthpart types (= feeding style). On the bottom graph, we see the corresponding taxonomic richness (at the family level) through time, and the pattern is obvious: with more ecological specialisation came more families (on average, it’s not a perfect fit due to historical contingency and the incompleteness of the fossil record). In other words, the rate of origination increased with ecoogical specialisation, another emergent property.
Study: Labandeira CC. 1997. Insect Mouthparts: Ascertaining the Paleobiology of Insect Feeding Strategies. Annual Review of Ecology and Systematics 28, 153-193.
Species selection is the highest level of selection we will be looking at. Below it lies another, much more controversial, level: group selection. I will admit to some bias, as I don’t find much wrong with group selection per se; we will get to its problems later. What I want to show here is that group selection is possible, and this is an important point that is missed by its most ardent detractors.
The experiment I will describe here comes from Michael Wade’s lab, and is a very important one in evolutionary biology for the above reason. The experimental set-up is simple (although I imagine quite laborious to conduct). The organism is Tribolium castaneum, the red flour beetle, a model organism that’s very easy to keep in the lab – all it needs is a test tube with flour to sustain a colony.
The slide explains what was done. 48 initial populations were set up and allowed to breed. After one generation (37 days), the experimental groups were set up. The control was simply taking 16 adults at random and starting the new population (each generation started with 16 adults). There were 2 experimental groups. For one, he initiated the new populations from 16 beetles belonging to the most successful populations (where success was measured by number of offspring). For the other, he did the opposite, initiating the new populations from the least successful groups.
In other words, he was artificially selecting for group size – a group-level, not individual-level, trait. In the control, there was no such selection. This went on for 9 generations.
Picture source: Eric Day
Study: Wade MJ. 1977. An Experimental Study of Group Selection. Evolution 31, 134-153.
The results are shown above. All populations experienced a decrease in population size; the control and low population size groups experienced it most, with those populations selected for high population size experiencing it least.
The explanation lies in the behaviour of the organisms. The lower the population size – as when only 16 adults are selected – leads to cannibalistic tendencies. This leads to the females not laying as many eggs – they won’t waste the energy to go through the egg formation, copulation, and laying if the eggs will be eaten anyway. So no matter what you’re selecting for, as long as the founding population has such a low number of individuals, you can expect a decrease in population size as the females don’t lay so many eggs.
But the pattern in each experimental group is different. Those with high population selection experienced occasional extreme spikes in population size, contrary to what we would expect. This is the action of group selection: natural selection at the individual level opposes the large population size because of the cannibalism; but when hgih population sizes are selected fro artificially, group selection will counteract individual selection. In the opposite group, the individual selection was reinforced by the group selection’s effect.
What this shows is that group selection is a viable, empirically-proven method of natural selection. However, this is only in the lab; whether it occurs in nature is still a matter of debate, and one in which we will not go into here.
Now we come to kin selection, a level of selection that acts on groups of genetically-related individuals. William D. Hamilton, one of the greats of evolutionary biology, formalised it as Hamilton’s Rule: kin selection occurs when rb > c. r is the relationship coefficient (how related the individuals are to each other); b is the benefit (i.e. success); c is the cost (disadvantage). We will come back to this when we look at the evolution of eusociality and altruism in the next slides.
Kin selection has been proven by many lab studies. The one I’m highlighting here comes from David Pfennig’s lab. The organism involved is the tiger salamander larva. These larvae are polymorphic, coming in two morphs, one of which is larger and cannibalistic.
What Pfennig et al. (1999) tested for was whether the cannibalistic larvae disriminated against eating their own kin. To do this, they gave them a buffet of food: related salamander larvae, unrelated salamander larvae, diseased salamander larvae, and other species. The results are shown in the bar graphs.
On the left, we see the benefit: those who discriminated against eating their kin had more surviving kin than those who didn’t disriminate. Obviously. This discrimination came at a minimal, negligible cost, as seen in the red graph, where cost was measured as body length.
So what is the effect of this on evolution? Simple. Discriminators, assuming their discriminatory ability is encoded for genetically, preserve the discrimination genes, since their kin also have them. On the other hand, non-discriminators keep reducing the chance of their non-discrimination genes getting passed on. Therefore, the discrimnatory genes increase in frequency in the population. This is how kin selection is measured – on the genetic level (remember the “gene’s eye view” we mentioned at the beginning).
Picture source: utahmatz
Study: Pfennig DW, Collins JP & Ziemba RE. 1999. A test of alternative hypotheses for kin recognition in cannibalistic tiger salamanders. Behavioral Ecology 10, 436-443.
Now let’s look at a classically contentious case where the level of selection has been debated: the evolution of eusociality. Eusociality is what termites, bees, ants, etc. have with only one reproductive caste, and other colony members being sterile.
There have been two main camps to explain the evolution of such a twisted system, where individual selection doesn’t apply. On the one side are the group selectionists. As already mentioned, group selection has a long history. The “original” group selection, termed naïve group selection is now thoroughly outdated. It was espoused by Wynne-Edwards, especially in his book Animal Dispersion. It viewed groups as static, whole, undivisable entities competing against each other. The group selection that’s alive now doesn’t resemble this at all, with competition occurring both within the group and between the groups.
On the other side are the gene-centered kin selectionists, exemplified here by George C. Williams, one of the legendary figures of evolutionary biology. His 1966 book Adaptation and Natural Selection is a must-read – it’s an extremely thought-provoking book, and one that will force you into becoming a proper evolutionary biologist (you will not agree with everything in it – that’s what’s so great).
Anyway, in Adaptation, Williams conclusively took down group selection as it was viewed back then. I don’t want to go into this, except to tell you the most simple argument: reproduction. Reproduction is the most important enabler of natural selection. Individuals reproduce much more often than groups. Therefore, any effect of group selection in anture will be far overwritten by the effect of individual selection. I highlighted that because it’s still the biggest stumbling block for group selection.
In Adaptation, Williams also tackled the eusociality problem, since up until then, it was dominated by the group selectionists.
Williams explained eusociality entirely in terms of gene-level kin selection – a highly-reductionist view, but also a powerfully explanatory one. The key thing to explain is why all workers are sterile – essentially, they’ve put a stop to their own possibility at reproducing. How can such a thing evolve?
Key to the argument is the sex determination system of the eusocial insects: they’re haplodiploid, meaning males arise from unfertilised, haploid eggs, and females from fertilised, diploid eggs. The resultant relationship coefficients are shown on the slide. All worker bees, ants, termites are sisters, and they’re all daughters of the queen. Combined with the haplodiploidy, that means that they’re all genetically very close – r is high.
Eusocial insect colonies are machines that run on altruism. Forget any pop-sci definition of altruism you might have read; altruism has a very specific definition: it’s an action that’s beneficial to the recipient and costly to the actor.
Sterilisation is an enormous cost, not outweighed by the relationship coefficient between the worker and the queen. Workers not laying eggs (which they are physiologically able to do in some cases) is an altruistic act; when they do lay eggs, the eggs are cannibalised/killed by police.
This is the currently most accepted model for the evolution of eusociality. Altruistic behaviour enforced by kin selection that selected for genes that enabled altruism. Of course, it’s more complex when you take into account the intricacies of insect colonies, but we won’t go into all the details.
Finally, the last level of selection is also one of the most misunderstood among the public, thanks to a book that inexplicably became popular in the public: gene-level selection, a.k.a. selection for selfish genetic elements. If you want a good book about this, check out the one advertised in the slide: Genes in Conflict: The Biology of Selfish Genetic Elements.
Contrary to popular perception, selfish genes aren’t ubiquitous. It’s actually a rare phenomenon, with a specific definition. A selfish genetic element is one that will favour its own reproduction regardless of the effect on individual fitness. Very often, the effect is negative. An example is the t haplotype in the mouse.
When a male mouse is homozygous for t (tt), then it dies. When a female is homozygous, then nothing happens. A population where t is common will eventually get a very twisted sex ratio, with a lot of females and few males, leading to a big bottleneck (or even to extinction).
As I said, these are very rare cases, for good reason – they’re too deleterious to persist.
Study: Lyon MF. 1991. The genetic basis of transmission-ratio distortion and male sterility due to the t complex. The American Naturalist 137, 349-358.
The misconception in the public about selfish genes arises from the gene’s eye view of things, which we already saw is essential but also merely a consequence of our tools.
Samir Okasha, a prominent philosopher of biology who does a lot of work on the levels of selection issue (his book Evolution and the Levels of Selection is a must-read for the interested), outlined a way to resolve such problems, and a way to visualise levels of selection in general. What is clear is that even in cases of individual selection, there may be other levels of selection at play – multilevel selection. Kin selection and group selection are reconcilable; gene selection and kins election can play off of each other; etc. So what Okasha suggests is that we view everything in terms of particles and collectives, with natural selection acting on both. A particle can be an individual, with the collective being a group or a population (species selection). A particle can be a gene and the collective the individual.
In all cases though, we have to take the gene’s eye view in order to remain consistent – our view of natural selection is, after all, nothing more than statistical changes in gene frequencies.
But this is where the misconception with selfish genes comes from too. Just because we consider everything in terms of genes, it doesn’t mean that that’s where evolution is acting. It’s merely a practical view; it’s not an explanation. I cannot stress this enough.
Finally, we come to the last section of the talk, where we will look at the limits of natural selection. There are two main limits: genetic and developmental.
An example of a genetic constraint can be found in sickle-cell anaemia. This is a serious disease caused by a mutation leading to malformed red blood cells. These are a double-edged sword: they don’t deliver oxygen properly (hence the anaemia), but they do provide foolproof resistance to malaria, since the virus can’t attach to the cells and multiply.
There are three genetic possibilities. Homozygous dominants are regular: they have no anaemia, but also no malaria resistance. Heterozygotes have the best of both worlds: no anaemia, and resistance. Homozygous recessives are resitant to malaria… but they also have the disease, so they cancel each other out.
Obviously, the heterozygous condition would be the overall most fit, especially in a malaria-ridden environment. But no matter how strong the selective pressure is, it will never become fixed in a population, by its genetic nature as a heterozygous condition: there will always be the statistical homozygotes in every batch of offspring. This is how natural selection can be limited – the FTNS is not all-powerful.
Developmental constraints arise out of side-effects and developmental canalisation. We’ll look at a popular example here, with two wonderful human male specimens as our models, Ron Jeremy, prominent actor, and David Hasselhoff, archetypal male.
The constraint here is the male nipple. It’s only there because of developmental canalisation – the nipple is a default feature. In males it’s useless, but in females it’s vital. But because it gets specified later in development, natural selection won’t be able to get rid of it only in males. Natural selection cannot override developmental constraints; it can, however, work around them.
We’ve reached the end. By now, you should all know what natural selection is. What should have been made clear, but will now be spelled out, is what natural selection is not.
Natural selection is not a necessity. As mentioned in the adaptationism explanation, natural selection is merely one of the mechanisms of evolution, but it doesn’t have to be acting. For example, the polar bear’s white fur is an advantage, but it’s something that can also evolve without the action of natural selection – after all, it doesn’t help it survive or reproduce (a polar bear has no predators to hide from and doesn’t need to ambush hunt from the snow).
Natural selection is not a force for perfection. As we saw with sexual selection and the guppy experiment, trade-offs are always involved – the selection of one trait is always going to have a negative effect elsewhere, and will only be retained if the benefit outweighs the cost. For example, many grasses, when growing under good conditions, will produce large seeds. That largeness is traded-off with number of seeds, though – the optimal would be many large seeds, but that will not happen.
In that same vein, natural selection is not a force for universal progress. The picture shows a crocoduck, the product of the fervent imagination of creationists. Such an animal doesn’t exist, but I show it because it would be awesome for a duck to have a crocodile’s head – it would certainly prove very advantageous for catching prey. But natural selection doesn’t care. A crocoduck will never evolve because of the foresight I just made up (not to mention the phylogenetic impossibility…).
Finally, natural selection is not a force for harmony. This is something we hear from the insipid brand of nature-lovers that want to just go back to nature and live among the trees and the animals, because nature really is a mother to us all. Bullshit. If nature was my mother, I’d complain to the child protection services. As an example of the benevolent love of Mother Nature and her Natural Selection toy, look at Cordyceps, the insect-loving fungus. It loves them so much that it grows inside them, feeding on their internal organs – not important ones first, then gradually to the more important ones – before invading the brain and taking control of the insect to direct it to a place where the fungus’s spores will spread most efficiently. The insect gets dragged there, in its last throes of life since its got no more organs and tissues, and finally dies. I’m overwhelmed by such a showing of love. For more on such parasites, see this post.
So, to summarise everything. Natural selection merely acts on heritable variation in fitness. No trait evolves alone – natural selection will have knock-on effects on other traits, and all adaptations have a negative side, often through a trade-off.
Natural selection happens gradually – adaptations don’t occur overnight, but through the successive selection of superior variants or deletion of inferior variants (a largely semantic difference) in a population. These variants are all modifications of previous traits – natural selection doesn’t create anything new.
Perfection is impossible due to the trade-offs, and the thought of it is really just the result of human wishful thinking.
And, most importantly, while selection can be viewed as a force or a mechanism (these are philosophical issues we haven’t touched), at its most correct, it is merely what we refer to as a statistical phenomenon – a consistent statistically-significant change in allele frequencies.
I would like to acknowledge the Human Biology Society at the University of Nicosia, and especially Lucia, for getting the room (a UNESCO amphitheater, very prestigious!) where I babbled for two hours; one reviewer of the talk who gave very incisive commentary (the original version of this was math- and model-heavy and required multiple transitions from presentation to R and back).
And last but not least, I would like to thank Douglas J. Futuyma, a prominent evolutionary biologist. He doesn’t know me, nor do I know him (although I’d love to have the chance to meet him), but his textbook Evolution (2nd ed.) formed the basis of this talk after the scathing review. If we were to go to court, he would have a definite case for accusing me of plagiarism.