There many ways of dividing up and categorizing Natural Selection. For example, there are the trichotomies of Natural Selection, Sexual Selection and Artificial Selection, and Modes of Selection (Stabilizing, Directional, and Disruptive) trichotomy.
We sense that these are good because they are “threes” and “three” is a magic number. Here, I’m focusing on the Mode Trichotomy, and asking that we consider that there are not three, but four modes of Natural Selection. This will cause tremors throughout the Evolutionary Theory community because Four is not a magic number, but so be it.
In Stabilizing Selection the extremes of a trait are selected against and the mean value of the trait remains the same. Mutations constantly introduced into the population tht produce traits out at the extremes are selected against. In Directional Selection the values of a trait at one end of the distribution are selected against and/or values at the other end are selected for, so that the distribution of values, and it’s mean, move in one direction. In Disruptive Selection the average values are selected against so that the distribution of the trait becomes bimodal.
That was pretty simple, but operationalizing these definitions, displaying them graphically, and thinking about how they work in shaping the overall pattern of evolution reveal important details that are often sidelined or not discussed. And, we have to consider the fourth mode: Null Selection: This is where there is no selection on the trait at any particular value. As mutations (or allelic novelty of any source) are introduced into the population what might have been a nice bell curve representing the trait’s values spreads and flattens.
One might argue that since “Null Selection” is not really selection, that it should not be a mode. I agree, but I still want it on the list of modes of selection. Why? Because without a concept of null selection, the lack of change in trait values is often incorrectly interpreted as “nothing is happening here.” But in fact, something fairly major and impressive is happening. Stabilizing selection is the process of ongoing introduction of variation and ongoing reduction in variation. It balances out because the more introduction of variation there is, the stronger selection becomes. A trait that remains the same for eons is a trait experiencing a dynamic evolutionary processes. Having no concept of Null Selection does not allow this thought to develop, or if it is mentioned, it may not stick as well.
Below I provide graphics depicting the modes. (They are available for non commercial use. For commercial use, that’s $1,000 Euros each. Oh, and click on the graphic to get a larger version.) I’ve made the graphics very simple but they are also meant to be very precise in selected details. as described below in the text.
Stabilizing Selection
As stated above, stabilizing selection occurs when the “central” value of a trait is not selected against or favored by selection but extreme values are selected against. The graphic shows “selection against” only, and this is depicted as rather menacing looking arrows pointing down at the upper and lower reaches of this “bell curve” shaped distribution. Note that the “after selection” graph shows that the extreme values from before selection are gone, the total range of variation is lower, and the mean is unchanged.
Directional Selection
Here the nasty looking Force of Selection Arrow is only affecting traits near one end of the distribution. The entire distribution squishes to the right. Note that the upper end of the distribution does not move up … in other words, directional selection does not simply move the bell curve along in one direction. The total range of variation reduces and the mean moves, in this case, to the right.
Disruptive Selection
In this case the central or average value is being selected against and/or the extremes selected for. My favorite example of this, and one often given in the textbooks, is the selection for gamete size. Fitness may be enhanced with a gamete with a certain amount of nutrition stored for use in a growing zygote (seed or embryo). Or, fitness may be enhanced by a small lightweight and mobile gamete (a pollen spore or a sperm). You can’t have both, and the compromise is less than ideal. [see this on Anisogamy] This example also forces us to realize that fitness needs to be considered in relation to the morph … the individual as it exists with a certain gender, developmental age, etc. Monty Python and the Catholic Church notwithstanding, a sperm is an individual with it’s own little genome and it’s own little Darwinian problems. So is an egg or a spore. They don’t have a lot of personality but they do have a fitness function.
Those are the usual three forms of selection, and the one I want to add is “Null Selection.” Is this the same as “relaxed selection” you may ask? If you want it to be that’s OK. Neither have definitions that are both formal and accepted. They are probably the same.
In null selection there are no Arrows of Selection happening to the bell curve, but there is still the constant introduction of mutations, so over time the distribution goes wacky and essentially becomes random.
For this to be really clearly conceptualized, we can go back to Stabilizing Selection and redraw the diagram like this:
Here, the mutations are seen constantly bothering the bell curve from below, and selection is working in an uneven way (more against the extremes) in the opposite direction. In cases where people have actually measured a trait over time, one sees this dynamic process. This is the equilibrium in punctuated equilibrium.
Have you read the breakthrough novel of the year? When you are done with that, try:
or Paperback
*Please note:
Links to books and other items on this page and elsewhere on Greg Ladens' blog may send you to Amazon, where I am a registered affiliate. As an Amazon Associate I earn from qualifying purchases, which helps to fund this site.
Honestly, I can’t stand the textbook variety of selective models to explain selection, I think they’re too simplistic. At the same time, I realize that it takes far more time explaining the myriad of possible types of variations and all the compounding factors, then giving specific examples of these types followed by explaining how exactly the population is changing.
1) Disruptive selection can be highly favorable for homozygous alleles, but more favorable to one than the other. (“Directional and disruptive”)
2) “Null” or “relaxed” or “unselected” alleles will undoubtedly have differential rates of mutation (or origin)
3) For traits with multiple genes influencing them, (most traits) this becomes even more complex; directional selection may favor a certain point along the continuum not at the either extreme, but away from average of the previous generation.
I like having these cut and dry simple models, but I sometimes wonder if we’re not oversimplifying the education of this.
If average individuals are the most heterozygotic in the population, then stabilizing selection assures an initial increase in genetic variation in each generation. I suspect that in most populations individuals around average are the most abundant group in the population. If so, this suggests that stabilizing selection is the most common form of selection. Good for Professor Bumpus.
Is it me or does darwin’s beard make him look more like bin laden? They have the same goals you know – destroy Christianity. Scary isn’t it?
No, ED, it’s you. Would that it were only you.
If Darwin looks like anybody it’s William Jennings Bryan.
Your explanation is inconsistent with respect to including mutations; you started out without them, and then included them.
“Note that the upper end of the distribution does not move up … in other words, directional selection does not simply move the bell curve along in one direction.”
“In null selection there are no Arrows of Selection happening to the bell curve, but there is still the constant introduction of mutations, so over time the distribution goes wacky and essentially becomes random.”
To avoid confusion, I would recommend including three pictures for each mode: bell curve before, bell curve after selection, and bell curve after multiple generations of selection and mutation.
qbsmd: I have done it in the manner you suggest, and it does work very nicely. Or, really, one animation for each mode.
In fact, these pictures often fail in a classroom unless the instructor gives a LOT of support as to how these graphs are constructed and what they mean.
I have used this post almost exactly as it is as a supplement to pre-existing course material that talks bout modes. It works pretty well in that case as well.
I take your “null selection” to mean that we have a Hardy-Weinberg situation except with respect to neutral mutations. To take the simplest case: We have a gene which exists in a single form and is neutral, ie. p=1. A mutation, similarly neutral, occurs, so we now have a small q and p is less than 1. Call the genes A and A1. A is going to have a rate of mutation to A1. Similarly A1 is going to have a rate of back mutation to A. Eventually, equilibrium will be reached, where the number of A to A1 mutation equals the number of A1 to A mutations during a period of time. p and q will reflect the ratio of A to A1 at equilibrium and will not change into the future unless we change a paramater. The “null curve” is not a random curve, but is rather determined relative mutation rates.
Jim, mutations don’t exist in equillibrium like the dissociation constant of a weak acid. Hardy-Weinberg equilibrium is a result of allelic frequencies in a population not subject to pressure, not mutation from one to the other. Differential rates of mutation do exist, but not from A to A1 and A1 to A, but from A to A1, A to A2, A to A3… and A1 to A1A, A1A to A1B, A1B… and A9Z9Z to A9Z9Z1…
There are some “back mutations,” but these are mostly compensatory mutations rather than reverting exactly to the parent genetic sequence.
Jared, Good points. As I understand it, the Hardy-Weinberg condition vis mutation is either (1) no mutation, or (2) mutation in equilibrium. I was using the simplest case to make the point that mutation frequencies are not a matter of magic, but rather a matter of mathmatically describable interactions. I will also say that my example was about as complex as my mathematics can deal with.
@2: A bit of thought should indicate that stabilizing selection should be the most common form of selection (well, of the classical three) – directional and disruptive selection are inherently self-limiting, as they’ll drive the population into regimes where they don’t affect a trait anymore, or where some other force comes up on the other end and turns it into stabilizing selection.
Thanks for that blog-entry – I think it’s about time that the fact that a particular trait might be spread so that no selection pressure is exerted on any value of it gets the attention it deserves. In addition, we might even consider situations where a trait is uniformly distributed so that concerning this trait, there is null selection. Or we might consider traits with have a flat fitness landscape (presuming we can – as some suggest, work with a fitness-concept for traits) and thus null selection.
Evolution denier for life? It’s so sad to me, and immeasurably more scary than Darwin’s position, that a Christian(?) would think that a humanly derived theory could destroy Christianity! Is the healing/saving message of the Christ so weak and conquerable? Where is your faith?
im shleep