Ecological Epistasis

Glad to hear my passing remark during the BEACON 2013 Congress sparked some debate (see David Baltrus’ blog post). The example that I gave was of plant growth when inoculated with two different microbes. In that particular case, the effect on leaf number was additive–however, one of my research questions is how often multi-strain community effects can be predicted from single-strain experiments. I believe that the term “ecological epistasis” is in fact an appropriate analogy in thinking about this problem, and will attempt to convince you of this.

First off, some clarification about what “ecological epistasis” is not. Ecological epistasis is not GxE (see Wade 2007 for a discussion of GxE versus GxG, as mentioned by Baltrus in his post). Genotype by environment interactions occur when the traits and/or fitness measured for a given genotype vary across environments; this has been termed phenotypic plasticity or equivalently a “reaction norm”.  Bradshaw (1965) writes:

“We are becoming increasingly aware that the individual cannot be considered out of the context of its environment. The way in which is reacts to different environments is as much part of its characteristics as its appearance and qualities in a single environment.”

He goes on to provide extensive evidence of plant traits that respond to environmental conditions such as light, water, and nutrient levels. To the extent that these reaction norms and interaction norms have a genetic basis, their shapes can evolve. Via and Lande (1984) model the evolutionary trajectories of phenotypic plasticity and show that correlated phenotypes across selective environments can constrain the response to selection in a given environment. Phenotypic plasticity can also occur in response to the biotic environment, such as population density (examples in Bradshaw 1965) or the presence of other species, such as the helmet and tail spike that Daphnia produce in reponse to fish predators. Agrawal (2001) discusses evidence that phenotypic plasticity can occur reciprocally by interacting species, which can be termed an “interaction norm”.

My original motivation for thinking about ecological epistasis came from reading a paper that described the term “ecological pleiotropy” (Strauss & Irwin 2001). (In my memory, this paper also used “ecological epistasis”, but re-reading it shows that this must have just been something I was thinking about at the time–good scholarship always involves re-reading papers you think you know!) Via and Lande (1984) parallel the evolution of phenotypic plasticity to pleiotropy, in which genetic correlations between traits under contrasting selection in a single environment can influence evolutionary trajectories (Lande & Arnold 1983). Just as gene can have pleiotropic effects on multiple traits and thus respond to multiple selection pressures, Strauss and Irwin (2004) propose that a single trait can show “ecological pleiotropy” and be involved in multiple ecological interactions. They give the example of floral traits that increase attractiveness: these may be selected for by enhanced pollination, but selected against when herbivores use these cues. Despite my fondness for it, this term has not gained traction within the evolutionary ecology literature. A paper titled “Niche separation in space and time between two sympatric sister species—a case of ecological pleiotropy” shows that habitat preferences are correlated with emergence timing and protandry in two Lepidoptera species (Friberg et al. 2008), which isn’t really in keeping with the definition proposed by Strauss and Irwin (2004).

The field of community genetics grapples with three levels of biological complexity: genes, species, and communities. Strauss et al. (2005) write, echoing Bradshaw (1965) above,

“How species evolve depends on the communities in which they are embedded. The idea that species exhibit traits shaped by collections of other species that co-occur with them received growing attention from the late 1970s through to the present. D. S. Wilson (1976) stated that ‘… every effect of a species on a community will loop back to influence the species itself, either positively or negatively.’ “

Ecological pleiotropy is thus the analogy of genetic pleiotropy–genes that have effects on multiple traits within a species–to communities, with traits that have effects on multiple interactions. I believe that “ecological epistasis” is a useful analogy in the same sense.

As an aside, analogy := “a comparison between two things, typically on the basis of their structure and for the purpose of explanation or clarification.” Lorenz (1974) writes about the usefulness of analogy in scientific reasoning about adaptation, cautioning that logical errors can be made when analogy (similarity of form) is mistaken for homology (similarity by descent from a common ancestor). He won the Nobel Prize (Physiology or Medicine) in 1973 for his work on individual and social behavior.

Click for open access paper from his 1973 Nobel prize lecture!

Baltrus writes that “you can define epistasis in the quantitative genetics sense (multiple loci interacting in a non-additive way)”, and it is this sense in which I intend ecological epistasis. Epistasis can occur not only among genetic loci in the same genome, but also across genomes. This is widely appreciated in cytonuclear interactions, such as a study in Drosophila that found segregating epistatic variation between cytoplasmic genetic elements (likely mitochondrial) and nuclear genes (Dowling et al. 2007). Mitochondria are an advanced case of endosymbiosis, and similar genotype by genotype (GxG) effects occur in Wolbachia interactions (potentially leading to speciation; Brucker & Borderstein 2012). Genotype by genotype effects are widespread in the horizontally-transmitted rhizobia-legume symbiosis, such as a paper titled “Intergenomic epistasis and coevolutionary constraint in plants and rhizobia” (Heath 2010). In this example, plant traits depend non-additively on the genotypes of both hosts and symbionts. These genetic interactions can additionally depend on the abiotic environment, i.e., GxGxE (Heath et al. 2010). Baltrus writes: “dynamics that apply to interactions between genes residing in different genomes may be inherently different than those in linked together and vertically inherited”. This point is well-explored in Austin, Trivers, and Burt’s 2009 book “Genes in conflict: the biology of selfish genetic elements” and in Michael Wade’s 2007 review.

However, I am not arguing that GxG is “ecological epistasis”–“intergenomic epistasis” (Heath 2010; Wade 2007) captures this concept well. Instead, I would define “ecological epistasis” as the case where the effects of an interaction between one pair of species depends non-additively on the interaction with another pair of species. By way of analogy with genetic epistasis (two loci interact non-additively to influence an organism’s phenotype), ecological epistasis occurs when the presence of two species in a community has non-additive effects of some property of that community. A strictly ecological example was given by my graduate student Prateek Shetty. Imagine an allelopathic plant that inhibits the growth of other plants that might otherwise occur in a habitat. If this allelopathy only occurs when a particular species of ant is present, then the presence of either plant or ant alone has no effect on community-level diversity. However, when both are present, community diversity might decrease. I would call this non-additive effect “ecological epistasis”. A major question in community ecology is the extent to which you can predict the community properties from the properties of the member species. This is directly parallel to the genetic problem of mapping genotype to phenotype within an organism. In host-microbiome interactions, as I was referencing in my talk, the community property of interest the health of the host. If two strains of bacteria are each pathogenic when inoculated alone, but interfere with one another to produce no effect on host health when inoculated together, this ecological epistasis could have major implications for our understanding and potential treatment of disease. On a more positive note, ecological epistasis between beneficial microbes such as rhizobia and mycorrhizae could enhance crop yields beyond what might be predicted by studying bipartite interactions. For example, inoculating with arbuscular mycorrhizal fungi enhanced the amount of nitrogen fixed by rhizobial bacteria and the subsequent growth of soybean plants (Asimi et al. 1980). My collaborator Michelle Afkhami has a forthcoming paper reviewing these “multiple mutualist effects”.

To close, not all analogies in science are bad. Engineers are attempting to mimic birds in designing more fuel-efficient airplanes:

It’s a bird! It’s a plane!

Some may not be as useful… trying to enhance torpedoes by adding a sleek layer of dolphin fat seems unlikely to be a great improvement on current design. In the BEACON talk that followed mine, Chris Adami remarked that a lot of his research comes about by seeing something in biology that “looks like” something in physics, and then exploring it to see if the mathematical tools of physics can be applied to the biological phenomenon. I believe that “ecological epistasis” may be such a useful analogy in the integration of community ecology and genomics.

Acknowledgement: Thanks to Ian Dworkin for encouraging me to write this response.


Agrawal, A. A. (2001) Phenotypic plasticity in the interactions and evolution of species. Science 294.5541: 321-326.

Asimi, S., Gianinazzi-Pearson, V., & Gianinazzi, S. (1980). Influence of increasing soil phosphorus levels on interactions between vesicular-arbuscular mycorrhizae and Rhizobium in soybeans. Canadian Journal of Botany, 58(20), 2200-2205.

Austin, B., Trivers, R., & Burt, A. (2009). Genes in conflict: the biology of selfish genetic elements. Harvard University Press.

Bradshaw, A. D. (1965). Evolutionary significance of phenotypic plasticity in plants. Advances in genetics, 13(1), 115-155.

Brucker, R. M., & Bordenstein, S. R. (2012). Speciation by symbiosis. Trends in ecology & evolution, 27(8), 443-451.

Dowling, D. K., Friberg, U., Hailer, F., & Arnqvist, G. (2007). Intergenomic epistasis for fitness: within-population interactions between cytoplasmic and nuclear genes in Drosophila melanogaster. Genetics, 175(1), 235-244.

Friberg, M., Bergman, M., Kullberg, J., Wahlberg, N., & Wiklund, C. (2008). Niche separation in space and time between two sympatric sister species—a case of ecological pleiotropy. Evolutionary Ecology, 22(1), 1-18.

Heath, K. D. (2010). Intergenomic epistasis and coevolutionary constraint in plants and rhizobia. Evolution, 64(5), 1446-1458.

Heath, K. D., Stock, A. J., & Stinchcombe, J. R. (2010). Mutualism variation in the nodulation response to nitrate. Journal of evolutionary biology, 23(11), 2494-2500.

Lande, R., & Arnold, S. J. (1983). The measurement of selection on correlated characters. Evolution, 1210-1226.

Lorenz, K. (1974). Analogy as a source of knowledge. Science, 185(4147), 229-234.

Strauss, S. Y., & Irwin, R. E. (2004). Ecological and evolutionary consequences of multispecies plant-animal interactions. Annual Review of Ecology, Evolution, and Systematics, 435-466.

Strauss, S. Y., Sahli, H., & Conner, J. K. (2005). Toward a more trait‐centered approach to diffuse (co) evolution. New Phytologist, 165(1), 81-90.

Via, S., & Lande, R. (1985). Genotype-environment interaction and the evolution of phenotypic plasticity. Evolution, 505-522.

Wade, M. J. (2007). The co-evolutionary genetics of ecological communities. Nature Reviews Genetics, 8(3), 185-195.