Last week, I posted some thoughts on Pierrick Bourrat’s new paper in Philosophy and Theory in Biology, focusing on his criticism of Rick Michod’s ‘export of fitness’ framework. This week, I’ll take a look at the second of Bourrat’s criticisms, regarding the transition from MLS1 to MLS2, as first defined by Damuth & Heisler, during a transition in individuality.
MLS1 and MLS2 refer to two different versions of MultiLevel Selection. As Bourrat describes it (and this is pretty much in line with other authors), fitness in MLS1 is defined in terms of the number of particles (or lower-level units, or cells) produced, while in MLS2 the fitnesses of the particles and collectives (or cells and multicellular organisms) are measured in different units. Cell-level fitness (for example) is defined in terms of the number of daughter cells, organism-level fitness is based on the number of daughter organisms. (As with last week’s post, I’ll generally stick to cells and organisms, though the principles apply equally to any two adjacent levels.
Okasha’s 2006 book, Evolution and the Levels of Selection, describes a transition in individuality as a transition from MLS1 to MLS2 and claims that these two frameworks represent “ontologically distinct causal processes of selection.” “Ontologically distinct” is, as far as I can tell, philosopher-speak for “really” different as opposed to merely convenient to treat as different. We can group or categorize things in different ways, and those categories fall on a spectrum from fundamental to arbitrary. For example, it might be argued that “fruits” (seed-bearing parts of flowering plants) and “meats” (flesh of animals) are more ontologically distinct categories than “vegetables” and “spices,” which are defined more by how we use them than by any biological criteria. When Okasha says that MLS1 and MLS2 are ontologically distinct, he’s saying that they’re actually different processes, not just different ways of looking at the same process. The (not universally accepted) claim that kin selection and group selection are mathematically equivalent “accounting methods” is a claim of ontological non-distinctness. [This is my interpretation; I’d be happy for someone with more of a philosophy background to set me straight]
Bourrat’s second objection to the Michod/Okasha framework revolves around the distinction between MLS1 and MLS2:
According to Okasha, once an ETI is completed, the particle and collective fitnesses become incommensurable: one could not, even in principle, measure the fitness of the collective in terms of fitness of the particles. Where does such incommensurability come from? To this question there no clear answer is given, and it is hard to see how there could be one, even in principle…I will show that the claim that there is incommensurability between the particle and collective fitnesses in cases in which there is evolution by natural selection is unwarranted.
To show that particle and collective fitnesses are commensurable, even in MLS2, Bourrat uses an example from Okasha’s book relating to species-level selection, which he (Bourrat) calls “the paradigmatic case of MLS2.” As a test case, he uses sexual vs. asexual reproduction. The idea here is that, because of the two-fold cost of males, asexual reproduction is advantageous at the individual level, but because it allows faster adaptation, sexual reproduction is selected at the species level. For Okasha, this is an example in which individual- and species-level fitness must be incommensurable, since they predict different outcomes. If individual-level selection is driving one direction and species-level selection is driving the opposite direction, they must be different processes, not merely different ways of quantifying the same process.
On the contrary, Bourrat claims that the apparent opposition between the two levels of selection is an artifact of measuring them over different time periods:
At the organism level, fitness is usually measured as the reproductive output after one organism’s generation. At the species level, fitness is measured as the rate of extinction or speciation over much longer periods of time, sometimes many millions of years…the difference observed could be either due to two processes of selection pushing in opposite directions or to two measures of one and the same process of selection over different periods of time, pushing in one direction over the short term and in the other over the long term.
In the test case of sexual vs. asexual reproduction, Bourrat says that, yes, in the short term, when the environment is stable, the asexual organisms have a selective advantage over the sexual. But measured over a time scale that encompasses environmental fluctuations, not only the sexual species but the sexual individuals leave more descendants, that is, they have greater (long-term) fitness. The apparent opposition between the two levels of selection results from measuring individual fitness over one or a few generations and species fitness over thousands or millions.
Bourrat goes on to say that this conclusion applies not only to species-level selection but to all MLS2 models:
There is no logical barrier to extending this argument to all the other cases for which MLS2 has been the framework of choice. In each case, if fitness could be determined over the same period of time, or more precisely in the same environment, at each level, what seem to be ontologically different levels of selection could in principle be unified under one and the same process.
Bourrat concedes that MLS2 is often a better framework, in practice, for dealing with collective-level fitness, but he treats this as a matter of tractability rather than a fundamental distinction. Collective-level fitness can always be expressed in terms of particle level fitness, he claims, as long as both are measured on the same time scale.
Finally, Bourrat gives an example using cells and multicellular organisms. In this example, the apparent opposition between cell- and organism-level fitness evaporates if both are measured on the time scale of organismal generations:
…this artefact quickly disappears if we measure the fitness of the cells and multicellular organisms over the same period of time. Then we can see that both selection at the cell and at the multicellular-organism level go in the same direction.
I’m not convinced that Bourrat is wrong that collective-level fitness and particle-level fitness, measured on the same time scale, will always predict the same direction of adaptive change. Certainly this is the case in his toy example. I’m not convinced that he’s right, either. A single counterexample in which particle-level fitness and collective-level fitness predict different outcomes over the same time period would show that he’s wrong, at least in his claim of universality.
Specifically, such an example should show that an allele that increases organismal fitness can nevertheless decrease in frequency among cells. When I first started thinking about this, I thought it would be easy to come up with one. Each attempt ended up the same, though: colony-level and (average) cell-level fitness were not only of the same sign, they were, in every case I tried, identical. As long as cell number and proportion of somatic cells are fixed, there is no way for cell-level and colony-level fitness to change independently. So the obvious candidates for a counterexample are mutations that change cell number or proportion of soma (or both). A mutation that causes colonies to have half as many cells will generate more, smaller colonies, increasing colony fitness but decreasing cell fitness, right? Nope, both colony-level and cell-level fitness are cut in half (I’m assuming here equal generation times and one offspring per reproductive cell, as in the volvocine algae). What about a mutation that causes previously undifferentiated colonies to produce 50% somatic cells? Let’s say the colonies have four cells, so the undifferentiated colonies produce four four-celled daughters and the differentiated colonies produce two four-celled daughters. Colony level fitness of the differentiated colony is half that of the undifferentiated colony, as is average cell-level fitness (eight daughter cells from the four cells in the differentiated colony; sixteen daughter cells from the four cells in the undifferentiated colony).
Maybe I’m thinking about this too simplistically, but I would like to see an example in this kind of gene-centric terms: an allele causes a certain change in phenotype that causes its frequency to increase in the population of colonies but decrease in the population of cells (or vice-versa). Even change in the same direction but at different rates would be sufficient to show incommensurability.
I’m still not totally convinced that collective- and particle-level fitness are always commensurable, but this paper has undermined my conviction that they are (sometimes) not. One valid counterexample would show that Bourrat’s claim of universality is wrong. Even then, I think his time scale perspective will be valuable in showing that perceived incommensurability often (if not always) evaporates when collective- and particl- level fitnesses are considered on the same time scale.