Lewontin’s Paradox of Variation
There’s nothing like a good paradox to keep a scientific field energised. In the field of evolutionary genetics, interest has recently been rekindled in an enduring riddle known as “Lewontin’s Paradox”. Neutral models of molecular evolution predict that the amount of genetic variation within populations should scale positively with the number of individuals in the population. Yet in the 1970s, Richard Lewontin observed that while population size varies by orders of magnitude across species (think oceanic zooplankton versus snow leopards), variation in their genetic diversity is much paltrier by comparison and indeed shows no obvious relation to census population size. While several potential explanations have been posited, these have defied rigorous empirical testing until the recent genomics era. Nowadays, entire genomes (or large chunks thereof) of any species can be sequenced at increasingly lower costs and the resulting avalanche of data has opened the door to a range of “Big Questions” in comparative genomics. This week in our journal club, we discussed a 2015 Nature paper by "Romiguier et al" that addresses a simple question, with a not-so-simple answer: how and why do genetic diversity levels vary between species?
In what must have been a mammoth undertaking, the authors sequenced the transcriptomes (the protein-coding regions of the genome transcribed into messenger RNA) of 76 non-model animal species spread across eight major metazoan phyla. Genetic polymorphism within each species was quantified as synonymous nucleotide diversity (πs): a measure of genetic differences among individuals at silent sites where nucleotide changes do not affect the phenotype. For each of these species, data were also collated from open source repositories on candidate explanatory variables that fell into two classes: life-history traits describing the species’ biology and ecology, versus “historical and contingent factors” captured by variables related to geographic range.
The results revealed no detectable influence of geographic range (or invasive status, interestingly), but beautifully clear effects of all life-history traits examined, which together explained a remarkable 73% of the variance in πs. In short, K-selected species with long lifespans, low fecundity and high parental investment seem to harbour less genetic diversity than r-selected species that live for shorter and trade offspring quality for quantity. Amniotes (turtles, mammals, birds) exhibited low polymorphism relative to other taxa examined, but non-amniotes that varied in parental investment strategies also varied in their genetic diversity, e.g. brooding marine species like seahorses and urchins were relatively genetically depauperate, whereas highly fecund, broadcast-spawning, sessile species like mussels and sea squirts were oozing genetic diversity. The results were also robust to phylogenetic non-independence and the same patterns were found for non-synonymous genomic sites that are presumably under selection.
Like all interesting science, the paper raises as many questions as it answers, and there was unanimous agreement that it was an excellent paper worthy of publication in Nature. I chose it as a great example of Big Theory being confronted with Big Data to great effect (with R-squared values that most of us can only ever dream of encountering!!), where complex genomic data is interpreted in an ecologically relevant light. The authors comprehensively ruled out the possibility of spurious explanations for their results related to methodology, while confounding factors such as interspecific variation in mutation rate could also be excluded (K-strategists should if anything have higher mutation rates). A few downsides (certainly not deal-breakers) were noted though, most obviously the rather opaque and poorly verbalised explanation for why species biology is so important. Several of us also felt that the conservation implications they drew were stretching their data a little (given that molecular genetic diversity is a “poor predictor” of adaptive potential in the face of environmental change) and they perhaps would have been better returning to the big theoretical questions in their concluding paragraphs. Personally I would have liked to have seen some clearer hypotheses and predictions laid out a priori regarding the expected effects (if any) of life-history on genetic diversity.
Two quite sophisticated explanations for their results were actually developed as mathematical models in the Supplementary materials, but their essence was clearly challenging to distil down to a few digestible sentences in the main text. One scenario is that population size of K-strategists “bounces back” more slowly following episodic bottlenecks (e.g. driven by periods of climatic change) compared to r-strategists, who spend less time at low population sizes following ecological disturbances. The other relates to the fact that r-strategies are thought to evolve in unpredictable, highly variable environments, whereas K-strategies evolve in more stable, competitive circumstances. The population dynamics of r-strategists are therefore more strongly buffeted by environmental fluctuations, which increases extinction risk. Only those with naturally high equilibrium population sizes might persist in the long-run, whereas even small populations of K-selected species might be viable, as they fluctuate less around their carrying capacity.
Both mechanisms would lower the long-term effective population size, Ne (which is highly sensitive to periods of low census population size) more in K-selected relative to r-selected species; Ne, in turn, is the key parameter determining contemporary nucleotide diversity. In that sense, the r-K continuum is really just a proxy here for a systematic gradient in Ne across species. Other explanations may be possible, however, such as stronger fluctuating selection in r-selected species that maintains genetic diversity at both non-silent and linked silent sites, while balancing selection may be weaker or less effective in smaller populations of K-strategists.
So has this paper solved Lewontin’s Paradox? Certainly it has provided a partial answer as to why current census size (or contemporary population dynamics) may not be a good guide to genetic diversity: historical bottlenecks and past demographic fluctuations (which play out differently depending on species’ life history) are what determine long term Ne and hence contemporary polymorphism levels. But other resolutions to the paradox that invoke selective sweeps have recently been shown to play a role (“Corbett-Detig et al. 2015”), so multiple explanations may apply. The riddle is not dead yet!!
Photo credit: https://www.3d-puzzlewelt.com/en/magic-cube-wild-animals-out-of-the-blue-4029811309870-en