Give them some space: how spatial structure affects the evolutionary transition towards mutualistic symbiosis
Mutualists construct the ecological conditions that trigger the transition from parasitism
The evolution of mutualistic symbiosis is a puzzle that has fascinated evolutionary ecologist for quite a while. Data on transitions between symbiotic bacterial ways of life has evidenced shifts from mutualism towards parasitism and vice versa (Sachs et al., 2011), so there does not seem to be a strong determinism on those transitions. From the host’s perspective, mutualistic symbiosis implies at the very least some form of immune tolerance, which can be costly (e.g. Sorci, 2013). Empirical approaches thus raise very important questions: How can symbiosis turn from parasitism into mutualism when it seemingly needs such a strong alignment of selective pressures on both the host and the symbiont? And yet why is mutualistic symbiosis so widespread and so important to the evolution of macro-organisms (Margulis, 1998)?
While much of the theoretical literature on the evolution of symbiosis and mutualism has focused on either the stability of such relationships when non-mutualists can invade the host-symbiont system (e.g. Ferrière et al., 2007) or the effect of the mode of symbiont transmission on the evolutionary dynamics of mutualism (e.g. Genkai-Kato and Yamamura, 1999), the question remains whether and under which conditions parasitic symbiosis can turn into mutualism in the first place. Earlier results suggested that spatial demographic heterogeneity between host populations could be the leading determinant of evolution towards mutualism or parasitism (Hochberg et al., 2000). Here, Ledru et al. (2022) investigate this question in an innovative way by simulating host-symbiont evolutionary dynamics in a spatially explicit context. Their hypothesis is intuitive but its plausibility is difficult to gauge without a model: Does the evolution towards mutualism depend on the ability of the host and symbiont to evolve towards close-range dispersal in order to maintain clusters of efficient host-symbiont associations, thus outcompeting non-mutualists?
I strongly recommend reading this paper as the results obtained by the authors are very clear: competition strength and the cost of dispersal both affect the likelihood of the transition from parasitism to mutualism, and once mutualism has set in, symbiont trait values clearly segregate between highly dispersive parasites and philopatric mutualists. The demonstration of the plausibility of their hypothesis is accomplished with brio and thoroughness as the authors also examine the conditions under which the transition can be reversed, the impact of the spatial range of competition and the effect of mortality. Since high dispersal cost and strong, long-range competition appear to be the main factors driving the evolutionary transition towards mutualistic symbiosis, now is the time for empiricists to start investigating this question with spatial structure in mind.
Ferrière, R., Gauduchon, M. and Bronstein, J. L. (2007) Evolution and persistence of obligate mutualists and exploiters: competition for partners and evolutionary immunization. Ecology Letters, 10, 115-126. https://doi.org/10.1111/j.1461-0248.2006.01008.x
Genkai-Kato, M. and Yamamura, N. (1999) Evolution of mutualistic symbiosis without vertical transmission. Theoretical Population Biology, 55, 309-323. https://doi.org/10.1006/tpbi.1998.1407
Hochberg, M. E., Gomulkiewicz, R., Holt, R. D. and Thompson, J. N. (2000) Weak sinks could cradle mutualistic symbioses - strong sources should harbour parasitic symbioses. Journal of Evolutionary Biology, 13, 213-222. https://doi.org/10.1046/j.1420-9101.2000.00157.x
Ledru L, Garnier J, Rohr M, Noûs C and Ibanez S (2022) Mutualists construct the ecological conditions that trigger the transition from parasitism. bioRxiv, 2021.08.18.456759, ver. 5 peer-reviewed and recommended by Peer Community in Evolutionary Biology. https://doi.org/10.1101/2021.08.18.456759
Margulis, L. (1998) Symbiotic planet: a new look at evolution, Basic Books, Amherst.
Sachs, J. L., Skophammer, R. G. and Regus, J. U. (2011) Evolutionary transitions in bacterial symbiosis. Proceedings of the National Academy of Sciences, 108, 10800-10807. https://doi.org/10.1073/pnas.1100304108
Sorci, G. (2013) Immunity, resistance and tolerance in bird–parasite interactions. Parasite Immunology, 35, 350-361. https://doi.org/10.1111/pim.12047
François Massol (2022) Give them some space: how spatial structure affects the evolutionary transition towards mutualistic symbiosis. Peer Community in Evolutionary Biology, 100143. https://doi.org/10.24072/pci.evolbiol.100143
Evaluation round #3
DOI or URL of the preprint: https://doi.org/10.1101/2021.08.18.456759
Author's Reply, 17 Jun 2022
Decision by François Massol, 24 May 2022
from Eva Kisdi's review, you will see that some clarifications are still needed regarding the mathematical developments presented in appendices. These corrections are needed for the sake of mathematical accuracy.
I am looking forward to reading your response and the revised version of the paper.
Reviewed by Eva Kisdi, 14 May 2022
Evaluation round #2
DOI or URL of the preprint: https://www.biorxiv.org/content/10.1101/2021.08.18.456759
Author's Reply, 17 Jun 2022
Decision by François Massol, 31 Mar 2022
given my reading of your ms and the four reviews, I ask you for a revision of the manuscript.
As you will see, one review (by Eva Kisdi) is pointing at important issues to be solved in Appendix A2. Three of the four reviews also list some points that need clarifications.
Personally, I had the same reading as E. Kisdi regarding Appendix A2 -- it seems that the "probability of getting a symbiont" have been lost and replaced by fecundities instead. I don't think doing the calculation with the right quantities is going to be much more difficult, so I encourage you to reiterate the exercise of obtaining some mathematical insights in this Appendix.
Regarding the mutation distribution, I am afraid I was misunderstood (and thus I deeply apologize for being too obscure in my description of the beta distribution). What I was thinking of as a perfect distribution for mutation was not to add a beta-distributed noise, but rather to draw the value of the new trait from a beta distribution centered on the value of the parent trait. In this way, there cannot be any problem of overshooting 0 or 1 (beta distributions are constrained between 0 and 1). The technical point is just to use the parameteriation of the beta that uses mean and precision, rather than the usual (a,b) parameters that is the natural parameterization of the distribution, such that mean = a/(a+b). With such a distribution of mutations, the mean of mutant traits is always the value of the parent's trait. The variance will vary depending on the localization of the parent trait value, but it will be less brutal than the doubly truncated exponential.
Aside from these issues, I was very pleased to see the arrival of the new appendices which are very useful to gauge the generality of your study.
I hope you will find all reviews useful for revising your paper. I am very eager to see the next version and I'm quite hopeful it will lead to swift recommendation.
Reviewed by anonymous reviewer, 22 Feb 2022
Reviewed by anonymous reviewer, 09 Mar 2022
Reviewed by Eva Kisdi, 15 Mar 2022
Reviewed by anonymous reviewer, 04 Mar 2022
Evaluation round #1
DOI or URL of the preprint: https://doi.org/10.1101/2021.08.18.456759
Author's Reply, 17 Jun 2022
Decision by François Massol, 30 Sep 2021
four referees have reviewed your ms. Based on their remarks and my own reading, I would like you to revise your paper before I can recommend it.
Overall, I was very pleased to read your manuscript as it is very rich in terms of both biology and theoretical results. The way you present the goals of the study as well as the background / state-of-the-art is helpful to understand where you are going. Thank you for the excitement!
One of my main concerns comes from the fact that you plunge right into the simulation model without trying to give some rough predictions based on some analytical approximation of the model. Considering global dispersal, some computations can be done, e.g. finding the proportion of cells occupied by hosts at "equilibrium" (I found 1-sqrt((1/f)*(m/(1-m))) because the way you formulate P_I and the fact that a cell must be unoccupied to be colonized make the density-dependence squared). Such an equation is interesting because it yields the occupancy of the grid when parasites are absent (64%) or present (20%), which set the limits for what you expect when evolution comes into play. You can also compute population growth rate using Euler-Lotka equation -- this would help you justify why the host population is viable in the absence of mutualistic symbionts, etc. I do appreciate the importance of a simulation model in the present case, but it is very difficult to be surprised or not surprised at the results if one has absolutely no clue as to what would be predicted by a simpler, approximated analytical model.
As remarked by at least one or two referees, the mutation model looks strange. As all the traits you want to have evolve are between 0 and 1, why don't you use beta distribution (or logit Gaussian) for mutation? The issue with truncated exponential is that the closer you get to the trait boundaries, the less likely mutation swill be accepted. So you would get an acceleration of mutations at mid values, and this could be problematic. The beta distribution looks like the cleanest contender for a mutation distribution in [0,1] and you can always specify it so that it looks hump-shaped rather than u-shaped.
The reproduction process also looked strange to me since you basically use the integer part of f as a given (i.e. you fix the number of offspring) and then us e the decimal part as a stochastic bonus. Why didn't you use a more classic approach such as drawing from Poisson distributions? Effectively, the variance in offspring number is very low in your simulations, which counteracts the possibility of extinction of the whole population. This might be worth discussing as extinction might sometimes happen before the transition to mutualism.
The introduction probably lacks a little bit of definitions/context:
* as noted by one of the referees, I guess you should use "joint evolution" instead of "coevolution" when you refer to the evolution of different traits in the same species
* the definition of symbiosis you use is not completely consensual -- some people still understand symbiosis as mutualistic symbiosis. For this reason, I encourage you to write your definition early in the introduction.
* since you use a parallel with altruism, I guess you also need to explicit what is altruism and how you distinguish mutualism from altruism
As you tackle the evolution of symbiotic entities, some background on symbiosis as an evolutionary force would help, e.g. Margulis & Sagan (2002). The notions of interactors and replicators of Hull (1980) are also interesting in your context, as well as the more recent idea of the holobiont (e.g. Bordenstein & Theis 2015; Doolittle & Booth 2017). I saw you referred to Sachs et al. (2011), maybe you could give the transition counts they put in their Fig. 3 to illustrate the importance of mutualistic transitions?
Because your initial state is one of parasitism, I found it bizarre that there is no mention of Red Queen dynamics in your introduction or discussion. For instance, would you expect the same speed of transitions if both hosts and symbionts had alleles that should match for the parasite to infect the host (or reciprocally for the host to be defended against the parasite)? It would probably be better if you could discuss a little what you think would happen in your model if either parasites and hosts were differentially locally adapted (as in the matching allele paradigm) or if parasites and hosts sequentially gained adaptations to counter the effect of the other (as in the gene-for-gene paradigm). In Sasaki et al. (2002), the effect of migration in a spatially explicit Red Queen dynamics model is studied and the authors conclude that migration is a "cheap alternative" to sex in Red Queen dynamics -- since your model includes dispersal but not sex, you might want to use this study as a benchmark for discussion.
Since you introduce and compare some of your results with those known in the evolution of altruism, I guess you could make the distinction between results due to kin vs. group selection. Group selection has had a bad press for some time, but modern models are able to incorporate both group and kin selections (e.g. Simon et al. 2013). In your model, the existence of clusters of mutualists is due to philopatry and increased fecundity, which leads to increased competition. However, if dispersal becomes local also for symbionts, these might suffer from a lack of percolation of cells in the grid (if host-occupied cells are not adjacent, a philopatric symbiont might not be able to "jump" over gaps in the host population). For a model dealing with the ecological considerations associated with local vs. global dispersal, see Huth et al. (2015)
At the end of the introduction, I was left with a big question: in your model, we don't know what exactly is understood by mutualistic effort. Since mutualism can take many forms, it would be useful if you could specify whether you think your model fits exchanges of resources more than shared immunity or anti-predator behaviours or other types of mutualisms involving symbionts.
Regarding the simulations, as noted by all referees, it would be nice to have the code somewhere accessible. Also, if you could give more details regarding simulations (duration, number of replicates per run, all tested parameter values, size of the grid, torus or absorbing boundaries or reflecting boundaries?), this would be invaluable.
In terms of statistics to present your simulation outputs, you use an assortment index which I did not understand -- can you write down its formula please? Also, why don't you give simpler things like Moran index of hosts and symbionts, join count statistics, spatial autocorrelation statistics, etc.? Computing the population growth rate (using Euler-Lotka or something similar) and plotting that in relation to e.g. competition strength might also help.
As some of the referees remarked, you can still improve the clarity of the text. Some sentences (especially at the end of the introduction and in the discussion) should be rewritten. Regarding presentation and formatting, my biggest remark would be to check thoroughly the names of authors in the references and when cited in the text: it looks like your bibliography is not very fond of authors with two surnames without hyphens, like the poor John MAYNARD SMITH who appears sometimes as Smith, sometimes as Maynard... Some other authors might have received the same treatment (I only saw Minus VAN BAALEN who lost his VAN, but others might have been injured too), so please pay attention and revise your .bib accordingly.
Another presentation/clarity item stems from the mathematical notations: some symbols (like c) are used to note very different quantities. My suggestion is to use other letters -- use a different letter for each different quantities. Also, as remarked by referees, please try to stay consistent in the way you write parameters (superscript vs. subscript, etc.).
It is quite frustrating not to have a more thorough explanation of the model in the main text -- can you please move some of the content from appendix A to the model & methods? Also, if you could add all parameters in the associated table, so that readers can refer to the table and find the interpretation of each parameter easily, that would be really nice.
A final remark: please pay attention to give all parameter values under plots if their values are not given by default in the summary table. For instance c_s is never given.
Bordenstein, S. R. & Theis, K. R. (2015) Host biology in light of the microbiome: Ten principles of holobionts and hologenomes. PLoS Biology, 13, e1002226.
Doolittle, W. F. & Booth, A. (2017) It’s the song, not the singer: an exploration of holobiosis and evolutionary theory. Biology & Philosophy, 32, 5-24.
Huth, G., Haegeman, B., Pitard, E. & Munoz, F. (2015) Long-distance rescue and slow extinction dynamics govern multiscale metapopulations. American Naturalist, 186, 460-469.
Margulis, L. & Sagan, D. (2002) Acquiring Genomes: A Theory Of The Origin Of Species, Basic Books, New York.
Sasaki, A., Hamilton, W. D. & Ubeda, F. (2002) Clone mixtures and a pacemaker: new facets of Red-Queen theory and ecology. Proceedings of the Royal Society of London. Series B: Biological Sciences, 269, 761-772.
Simon, B., Fletcher, J. A. & Doebeli, M. (2013) Towards a general theory of group selection. Evolution, 67, 1561-1572.