I am making a program for simulating natural selection. Now, I have discovered an interesting difference between positive selection and negative selection.

Positive selection occurs when the population size is limited by resources. Take bird territories as an example. Assume that there are not enough suitable territories for all the birds of a generation. The fates of the individual birds are interdependent in the following way: Every time a bird finds a territory and successfully defends it, the chances for the remaining birds are reduced. In other words, the chances for each bird of finding a territory depends on how successful other conspecifics have already been.

Now assume that there are two different phenotypes with different fitnesses. A rare mutant has higher chances of finding and defending a territory than the wild type.

The probability distribution of the mutant after selection is described by the little-known Wallenius' noncentral hypergeometric distribution (see Wikipedia: https://en.wikipedia.org/wiki/Wallenius'_noncentral_hypergeometric_distribution )

Now consider negative selection. This occurs when the population size is limited by predation. Assume that the birds are hunted by predators and that the predators have a limited appetite or limited time and energy for hunting, so that the number of surviving birds is fixed. The fates of the individual birds are also interdependent under negative selection: Every time one bird is caught by a predator, the chances for the remaining birds are increased. Assume that the mutant birds have a higher chance of escaping predators. The Wallenius distribution now models the number of birds killed rather than the number of birds surviving. The distribution of surviving birds under negative selection can be called the complementary Wallenius distribution.

The difference between positive and negative selection only applies when the fates of the birds are interdependent. Under positive selection, the birds are selected for survival, one by one. Under negative selection, the birds are selected for elimination, one by one.

I have simulated these distributions and discovered that positive selection has the strongest effect when the overall survival rate is high, while negative selection has the strongest effect when the overall survival rate is low.

The difference is shown in the following graph. The graph shows the expected number of surviving mutants after positive versus negative selection. The initial population has 10 mutants and 990 non-mutants. The fitness ratio is 5.

positive and negative selection as function of survival rate

The green line indicates positive selection and the red line indicates negative selection. The grey line indicates the situation where the fates of the individuals are independent. We can observe that both positive and negative selection have stronger effects than when fates are independent.

Independent fates means that the chance of survival of each individual is independent of what happens to the other individuals. Independent selection can happen in the following cases: (1) Unrestrained growth. (2) Feeding, when food is plentiful, but difficult to find. (3) Predation, when predators have unlimited appetite, but limited success in hunting the species in question. (4) Diseases: The ability of one animal to survive a disease has no negative effect on the other animals.

The total number of survivors is not fixed under independent selection, but determined post factum.

The difference between the effects of positive and negative selection can be explained most easily by the number of individual selection events. A selection event under positive selection is a bird finding or winning a territory. The number of such events is higher when the survival rate is high. A selection event under negative selection is a bird being caught by a predator. The number of such events is higher when the death rate is high and the survival rate is low.

Now, here comes my question: I find it hard to believe that I could be the first one to discover the statistical difference between positive and negative selection, but I cannot find any literature describing such a difference. Can you please direct me to any literature describing the difference between positive and negative selection in terms of the mathematical/statistical effects.

  • 1
    $\begingroup$ I don't think that you are using the same definition of positive selection as other people. Positive selection is usually more considered to be the outcome of innovation that allows competitive advantage; the rate of new innovations is positively related to the summed population mutation rate. There are many descriptions of positive/negative selection behavior in terms of gene frequencies, but it seems that you are assuming perfect heritability of different phenotypes and seeing what happens rather than doing pop genetics. $\endgroup$ Oct 26, 2023 at 19:19
  • $\begingroup$ Thank you for your comment Maximilian. You are right that the terms positive and negative selection are often used in a different meaning than here. The concepts I am adding here are not described anywhere in the literature, as far as I can find. For lack of better terms, I am calling them positive and negative viability selection. I am open for other ideas if you have a good suggestion. My models are described in this preprint: researchgate.net/publication/… $\endgroup$
    – A Fog
    Oct 28, 2023 at 6:03
  • $\begingroup$ I skimmed this question and the preprint again and I think that the main stumbling block here has to do with vocabulary. For example your "positive viability selection" is very close to "negative frequency dependent selection", though I suppose you can make some definitional arguments for a difference. Similarly "positive frequency dependent selection" looks a lot like your "negative viability selection". I am not an evolutionary ecologist though. I added the ecology tag to your question since I think that's more appropriate. $\endgroup$ Oct 29, 2023 at 20:32

1 Answer 1


I have a hard time to understand exactly what and how you are modelling your scenarios, so it is difficult to answer the question. For one; are you modelling survival rates of individuals under different scenarios, or are you modelling actual evolution across time (e.g trait or allele frequencies across time)?

However, I think (as also suggested in the comment by @maximilian) what wording and language plays a role here. For instance, your framing of territorial competition (and the predator-prey scanario) are very similar to what is used in evolutionary game theory. You should look into theory in that area for inspiration and similarities.

You should also look at density dependent selection since theory in that area also deals with similar issue that you mention. For instance “K-selection” in crowded environments with fixed resources or territories. One recent paper you might find interesting is Bartram & Masel. 2019. Density-dependent selection and the limits of relative fitness.

I’m not sure that you will find what you are looking for here, but it might provide a couple of starting points

  • $\begingroup$ Thank you for the reference. You ask if I am modeling survival rates. I am modeling the statistical distribution of individual selection events as a function of phenotype. How many birds of each phenotype will find a territory? How many rabbits of each phenotype will avoid predators? These are statistical distributions. The statistical distributions get complicated when the fates of the animals are interdependent. The survival rates of birds of one phenotype depends on how many birds there are of a different competing phenotype. That's what I want to model. $\endgroup$
    – A Fog
    Oct 31, 2023 at 6:22

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .