vesicles

Prev: Eco-evo Predator-Prey model (vd Laan) Next: Stochastic corrector model


 * TODO List**
 * REF Wynne Edwards
 * REF to Williams and individual-level selection
 * REF Michod

=[|Vesicles] and [|group selection]=

In the previous section we have seen that the spatial hypercycle patterns (spirals) generate a **higher-level selection** ([|Boerlijst & Hogeweg 1991]). This higher-level selection was not implemented //a priori// in the model, but emerged together with the spiral patterns, i.e. it was an **implicit group level selection**. The same was true for the wave-level selection we observed in the minimal eco-evolutionary model of replicators and parasites (Takeuchi and Hogeweg, 2009). In this section we consider a more //explicit// multi-level selection by means of **compartments** or **vesicles**, which are predefined. The question we ask then is: //Can group selection lead to stable coexistence of several species and so generate stability against parasites?// (Note that we are looking at ecological stability here)

In the models of prebiotic evolution that are based on this idea, the "vesicle first" and "evolution (replication and heritability) first" scenarios are combined (see the prebiotic evolution intro page).

[|Group selection]
Group selection has long been seen as a very controversial term (see e.g. wikipedia link above) because it was often formulated in terms of the "good of the species" ([|Wynne-Edwards]? (REF)). With the advent of [|neo-darwinism] there was a new explicit focus on the individual as the **unit of selection**, and individual selection was thought to always undermine group selection (REF). However, as described in the first paragraph we have already seen some examples in which looking at selection at the individual level only was insufficient to explain the evolutionary dynamics: the outcome was determined by selection on characteristics of higher level entities (e.g. faster moving spirals, or waves that survive long enough to produce new waves).

Recently, the debate around group selection flared up again when 3 scientists (evolutionary mathematical biologist M. Nowak, biomathematician C. Tarnita and ant-specialist and famous evolutionary biologist EO Wilson) published a paper in Nature in which they claimed that the evolution of [|eusociality] could better be explained by group selection than by [|kin selection]: the famous explanation for social behaviour introduced by Hamilton and described in his [|Hamilton's rule] (1963). Their paper was met with a large amount of heated replies, see eg Abbot et al (2011) and many of the other reactions, and until know this debate still continues. However, in groundbreaking work in the 1970s [|DS Wilson] (1975?,1979; not to be confused with EO Wilson), formulated a group selection model in which he studied the conditions in which group selection could occur given the individual as the unit of selection. The model was formulated with explicit higher-level trait groups and selection within and between trait groups. The main idea was to look at the evolution of [|altruism] and to separate [|kin] and group selection. The latter could be achieved by perfectly mixing the population once in a while to remove kin biases (i.e. individuals that are related sitting close to each other/in the same trait group), and in that sense it becomes clearer that kin selection is basically just a parameter in this model of group selection: It is an external parameter in the model, and only affects the distribution of related individuals (i.e. versus random) (see also [|Lehmann et al. 2007]). Individuals were modelled with a difference in traits (A and B) which could represent an altruistic behaviour (e.g. [|alarm calling]) or not with a fitness effect for donors (fd, for inidividuals that "give" altruism) and recipients (fr, for individuals that receive altruism). Then the fitness effect on individuals with trait value A (displaying the (altruistic) behaviour) and B (not displaying the behaviour) is:

F a = fd + (Na - 1)*fr F b = Na*fr

where fd and fr are the fitness effects of "giving" and "receiving" the behaviour, and Na is the number of individuals of type A in the trait group.

Looking at the extreme cases (paradigms) of this model one obtains the following results:
 * 1) Uniform distribution (no groups): similar to just one group, and there is no group selection. The behaviour is selected if fd > fr (positive fitness effect of displaying the behaviour is larger than the fitness effect of not displaying the behaviour).
 * 2) Totally segregated groups: altruism evolves because they have the fittest groups. This also happens if fd < 0.
 * 3) Random distribution: altruism evolves when fd > 0.
 * 4) More than random (some kin aggregation): altruism can evolve when fd < 0.

Important insights from Wilson's model are:
 * group selection works according to compartmentalization of the population
 * kin selection as a parameter of group selection
 * group selection causes a **reversal of selection** relative to individual-level selection: i.e. altruism evolves instead of non-altruism!

Next, we will consider how adding an explicit layer of groups affects a model of prebiotic evolution.

Next: Stochastic corrector model