Wednesday, September 6, 2017

Descent with Modification: The Evolution of a Conceptual Figure

Almost every student talk – and many by postdocs and profs – starts with a conceptual diagram linking ideas and concepts. The simplest possible version of such a diagram for eco-evolutionary dynamics is two boxes, one for ECOLOGY and one for EVOLUTION, with arrows linking the two. This figure still leads many talks as a way of emphasizing the fact that researchers have long focused on the arrow from ECOLOGY to EVOLUTION but not the reverse, which has recently become the primary new emphasis of research in Eco-Evolutionary Dynamics.

Starting from this simplest framework, I developed an expanded – but still simple – version that sought to make several key points explicit. First, phenotypes are the true nexus of eco-evolutionary dynamics because it is PHENOTYPES (not genotypes) that are under direct selection and it is PHENOTYPES (not genotypes) that have ecological effects. Thus, arrows from ecology to evolution and back must flow through phenotypes. Second, the direct effects of phenotypes on one ecological level (e.g., populations) can cascade to indirect influence other ecological levels (e.g., ecosystems).

I tried to unearth out when I first came up with this figure and was unable to be definitive. However, it was first published in Bailey et al. (2009), a New Phytologist “Forum” describing a symposium on eco-evolutionary dynamics at the ESA meeting in Albuquerque, New Mexico, in August of 2009. I had used the figure in my talk at that symposium. Another early appearance was an F1000 Biology Report that I did with Eric Palkovacs in 2010. Since those uses, the figure has become widely used in various publications and talks (and my book) as a simple intuitive way of conceptualizing eco-evolutionary dynamics.

In short, the basic figure has stood the test of time, largely without revision for almost 10 years. However, almost every time I show it, I end up having interesting discussions with someone about how it could perhaps be improved. In nearly all such cases, I have convinced myself that no change was necessary but, twice now, I have been compelled to admit that it could be better. The first time came during a “the genomic of eco-evolutionary change” symposium and workshop in Monte Verita, Switzerland, in 2016. Victoria Stork pointed out during her talk that my use of “genes” at the top of the figure ignored other, potentially important, genomic changes that influenced phenotypes and could therefore have ecological effects. Epigenetic changes, such as DNA methylation, are perhaps the most obvious example. At this point, my book was basically done and at the printer but I was able to add “genomes” to “genes” at the last minute. My intent in doing so was be inclusive of people studying epigenetics.

The second time I have felt compelled to make a change came just last week at a “significance of sexual selection for population fitness” workshop, again in Switzerland – but this time in scenic Fafleralp and organized by Claus Wedekind. I gave a talk on the first night of the meeting and, afterward, several people – most notably Jacek Radwan – argued for the addition of an arrow directly from population dynamics to genes (bypassing phenotypes). The most obvious reason for such a pathway would be that small population sizes can lead to genomic changes through inbreeding or drift, which could then have phenotypic effects with ecological consequences. I had heard this argument before and had not been convinced, because I felt that the effects still had to flow through phenotypes. This time, however, the argument was clearer to me because I had been specifically talking (and therefore thinking) about the genetic effects of population size.

Specifically, I now agree that small population sizes can directly influence genomes without having to pass through phenotypes, with my apologies to those who made this point previously and I dismissed it. However, I still think that effects of genomes BACK to ecology must flow through phenotypes. That is, inbreeding and drift will change genomes but they will have ecological effects only if those genomic changes modify organismal fitness (leading to a phenotypic effect on population dynamics that could cascade to indirectly influence communities or ecosystems) or traits (a potential direct effect on communities and ecosystems).

Here’s hoping no more major changes are needed. Or, wait, maybe not. Keep ‘em coming – just realize it might take me years to agree! 


Fafleralp: our workshop was held in the building in the lower left.

Fafleralp: looking back where I had hiked on my first day.

Fafleralp: looking back where I had hiked on my second day.

More photos from Fafleralp.


  1. In my opinion the biggest missing feature in this classic diagram is the likely major role of environmental context in mediating essentially all of the arrows. I don't think it's sufficient to just say that the environment and ecology are just the same thing, because some environment features are likely shaped by ecological feedbacks, but other represent potentially weak or strong extrinsic factors like temperature or cultural eutrophication. In Tuckett et al. 2017 (Copeland) we created a figure depicting the role of environmental context as a 'lens' of sorts that modifies feedbacks (and is shaped by some feedbacks) using a variant of the simplest schematic you show here, but it would be nice to somehow capture the potentially major role of environmental context in some version of the Bailey et al diagram.

  2. I think you are right Mike. I tried to tweet this out to Andrew, but it wasn't the best format. Van Nuland et al. 2016 put a figure like this together (I keep trying to post it here, but can't). The point of the figure is really redefining the basic definition of a phenotype (in a sense). For example, If a phenotype has a genetic and environmental component then P=GxE. We know that E can be partitioned into abiotic and biotic components. Biotic components have a genetic basis so the definition expands to P=GxG(indirect genetic effects within a population or Interspecific IGE among species)xE(abiotic). This is effectively the definition of the geographic mosaic theory of coevolution and explicitly acknowledges the role of "environment". Still only coevolutionary feedback and no Niche Construction. GxG interactions are the foundation of the extended phenotype and niche construction. Based on the laws of thermodynamics (and basic ecology) as species interact energy flows across trophic levels and nutrients are cycled. These are ecosystem processes and are geneless products (abiotic environment) that emerge from genetically based species interactions and represent niche construction/extended phenotype. If these ecosystem processes feedback to effect the fitness of the interacting species (and models in Schweitzer et al. 2016 show they can), then we should be thinking about phenotypes=GxG(interacting individuals)xE(abiotic)xE(niche construction). The effects of E(niche construction) are likely to vary in space and time though and may contribute more in some generations than others and so should be modified with a time component E(NC)t1, t2 etc. This accounts for feedbacks from a variety of levels and abiotic gradients (Fig 1 of Van Nuland et al. 2016). Everyone is trying to move towards the price equation and we talked about this in Genung et al. 2011. It is amazing powerful and general, but this is a much more accessible perspective for most folks than going after multilevel selection. It does not take multilevel selection to see niche construction evolve on the landscape. My two cents. JB