The problem of multilevel selection

In the chapter about selection of my book, I wrote a draft on what would be the unit of selection in the selective process, namely, what biological entity has its frequency increased because of selective processes. This is a rather controversial topic in evolutionary biology, and I quickly decided to drop it, focusing only on the basic explanation of fundamental concepts related to the selection itself, because, from the beginning, I planned to write a fairly brief and short book. But now, here in the blog, I can try to write a line or two about the unit of selection. This post is a first superficial exploration on this subject.

Regarding what is conventionally called Darwinism, or in other words in classic evolutionary biology, the unit of selection is the organism. That is, selection is a process in which different individuals, with distinct attributes and properties, have a difference in their reproductive rates. Considering the organism as an individual (i.e., that which cannot be divided), we should analyze the organism as a whole, competing with other organisms of the population in a resource-constrained environment, and having a higher or lower reproductive rate (that we call fitness) compared to these other organisms. This is the orthodox view in evolutionary biology.

However, selection process, even if it occurs at the level of organisms, have the effect of significantly increasing the frequency of some genes, some allelic variants. According to Stephen Gould, this change in gene frequencies must be considered as a consequence of the selection process at the level of individuals, but many biologists now consider the genes, and not the organisms, as the “unit of selection”. Although this approach has been widely popularized by Richard Dawkins, to the point that many people immediately associate him with the concept of gene as the unit of selection, it was initially postulated by George Williams in the 60’s.

Particularly, despite all the dangers of this abominable thing called reductionism, I do not see why not admit, in certain situations, that selection occurs at the level of genes. The occurrence of selection at one level does not necessarily exclude the occurrence of selection at other levels. Stephen Gould, who is usually seen (incorrectly) as an opponent of this idea, makes it clear in his book “The structure of evolutionary theory” that he admits the possibility of different levels of the biological hierarchy functioning as units of selection. For those who are still a little confused, these levels are, from smallest to largest: genes, cells, tissues, individuals, social groups, populations, species, higher taxa etc.

I don’t mean to accuse anyone of reductionism, but I think there’s some radicalism on the part of Dawkins and his fellows, when they affirm categorically that genes are the only valid unit of selection. Well, the last Dawkins’ text I read where he reiterates this point of view is from 1994 (“Selection chooses only replicators such as DNA molecules”) and, so, a bit old. Therefore, I don’t know if he ever admits other levels for selection or if he’s still considering the genes as the only valid unit of selection.

However, we should now move to the other side of the hierarchy: selection at levels higher than the individual.

The problem arises with a fairly common example in biology: suppose a particular prey, amid other of its kind. At the sight of a lurking predator, the prey emits an alarm signal, warning his colleagues about the approaching predator. The problem here, to explain the evolution of this type of behaviour, which we define as altruistic, is that the individual who initiates the alarm clearly draws the predator’s attention to itself.

The first attempts to explain the evolution of this and other types of altruistic behaviour gave rise to what was initially called group selection. The individual who gives the alarm, despite increasing the chances that he himself is captured, reduces the overall number of catches. Thus, he acts for the good of the species, and the group as a whole is benefited. This can sound as a reasonable explanation, but I am strongly opposed to this interpretation. What was originally called group selection in the 60’s is imbued with an absurd teleological view: it’s like saying “organisms reproduce to perpetuate the species”, a misconception that we hear constantly, and that constitutes a rather serious error (I treated this issue of “reproduction serves to perpetuate the species” in the chapter on gene pool of my book).

So, how should we interpret these altruistic behaviours? One of the most popular alternatives is called kin selection, proposed by William Hamilton, among others. What matters here is not exactly my altruism, but what individuals I can save (or benefit) with it. When, for example, my selfless action saves my father, my brothers, some cousins ​​and uncles, who were with me in the herd, for example, even if the predator captures and kills me, several copies of my genes are preserved. Note that, here, the genetic relationship is very important: when asked if he would give his life to save a brother from drowning, John Haldane reportedly said “no, but I would to save two brothers or eight cousins”. For those not familiar with the genetics of kinship, notice that you share with a sibling, on average, 1/2 of your genes, while with a cousin you share 1/8 of your genes. Thus, sacrificing yourself to save 3 brothers or 15 cousins would be interesting, since the number of your gene copies will increase in population.

Note, however, that kin selection, explained this way, favours the view of genes as units of selection: an altruistic behaviour is evolutionarily selected because it would increase the frequency of the genes involved. This is a coherent explanation, but there’s another possibility: selection at levels higher than the individual level can be explained otherwise, without bringing up again the fallacies of group selection.

Before we go any further, a brief digression: a few supposedly altruistic behaviours may be not that altruistic. A classic example is the stotting behaviour of gazelles, shortly before fleeing from a feline, or even during the chase. This behaviour does not serve to warn its peers about the presence of the predator. On the contrary, this signal is directed exactly to the feline! It may be that the gazelle wants to warn the feline “I’ve seen you”, or show the feline that it is in good physical condition, so the chasing will be hard or even useless. Whatever the case, felines more often abandon the chase of stotting gazelles than of non-stotting gazelles.

A springbok stotting (photo: Yathin Krishnappa).

A young springbok stotting (photo: Yathin Krishnappa).

Returning to our subject: I believe there is a new and adequate explanation for selection at levels above the individual level. It is a simple matter of self-regulation from an evolutionary standpoint. An interesting example, published in Scientific American, in the case of bacteria Pseudomonas fluorescens. There are variants which produce a polymer that allows the colony to float, and which have a greater energy expenditure producing this polymer; other opportunistic bacteria, taking advantage of their fellows, do not produce the polymer. Selection at the level of individuals, or even at the level of genes, will cause the proportion of opportunistic bacteria in colony to rise. However, if this proportion is very large, the colony sinks, and all of them die.

What we can do in this case is to define the entire colony as a unit of selection. There are a number of colonies, some colonies where opportunistic bacteria are more “aggressive”, and other colonies where opportunistic bacteria are more restrained. Without any teleology needed, selection process can be explained as usual: treating the colony as the unit of selection, certain colonies originate more colonies than others (higher reproductive rate), and thus propagate their genes and their individual bacterial behaviours.

Currently, these proposals for selection at various hierarchical levels, including levels higher than the organism, is what has been called multilevel selection. For these evolutionary biologists, kin selection, for example, would be just one of several cases of multilevel selection. Selection at the organisms’ level, or even at genes’ level, is not discarded. On the contrary, for most of the proponents of multilevel selection, as Edward Wilson and David Wilson, “adaptations at a given level require a process of natural selection at the same level and tend to be undermined by selection at lower levels.”

To summarize this whole issue very simply, I think that an excellent explanation can be found in “Evolution: an introduction”, by Stephen Stearns. According to him, wherever there is variation, reproduction and heredity, there is selection. Therefore, selection can occur at different levels of the biological hierarchy, not just at the level of organisms.


2 thoughts on “The problem of multilevel selection

  1. Great article. I’ve also written a few articles on multilevel selection, and can send you links if you’re interested in reading them. I also had a few questions on the topic I’d like to ask you. Do you have an email address I can contact you at?

    • Hi,
      thanks for your comment. I’ve just seen your blog, great content.
      This little post of mine about multilevel selection is quite old, I’ve recently written a lot more on this subject. The thing is I am in the painstaking process of translating all my posts from my original portuguese blog, in chronological order.
      You can find my email in my “about” page. It is

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s