Thursday, December 1, 2011

What should be the basic unit of community ecology, 2011.

Why intraspecific variation matters in community ecology Bolnick et al. 2011, Trends in Ecology and Evolution.

Intraspecific variation in gastropod
shell morphology (Goodrich 1934).
 There has been a long debate in community ecology on the fundamental unit, a debate on what Tansley described as the “necessity of first determining empirically our natural units”. In early years, it involved tension between Clements' and Gleason’s view of the plant community, either as a “superorganism” or simply as a conglomeration of co-occurring species. This latter, Gleasonian view won out, signaling a move towards the species-oriented approach that dominates community ecology today. In later years, there was a push to view the individual—not the species—as the fundamental unit, championed by people like Dan Simberloff. However, though this view has had some influence, it has never been mainstream.

 The basis of these debates about the basic unit is simple: do similarities matter more than differences? Recently, the argument that intraspecific differences are important and that community ecology should consider individuals has become much stronger. In “Why intraspecific variation matters in community ecology”, Bolnick et al. suggest that a species-level view of community ecology is an incomplete one, and that we should be aware of making simplifying assumptions about intraspecific variation (e.g. that it is minimal and species-level means are appropriate). Bolnick et al. state their hypothesis clearly:
  “… many models of species’ interactions implicitly assume that all conspecific individuals are effectively interchangeable. In this paper we argue that this assumption is misleading and that intraspecific trait variation can substantially alter ecological dynamics.” 
 To that end, the paper does an excellent job of identifying the key mechanisms by which intraspecific variation might be expected to alter ecological dynamics (especially as summarized in the paper's Table 1). Some of these mechanisms might be fairly ubiquitous. For example, when there are nonlinear relationships between trait values and interaction strengths, Jansen’s Inequality means that the amount of intraspecific variation around the species mean will alter the strength of that interaction. The mechanisms discussed make a convincing argument that intraspecific variation can alter ecological interactions and evolutionary dynamics.

However, a move to individual level ecology has many practical implications*: for example, it would require that we move beyond using average species-level demographic rates, dispersal abilities, and interaction strengths, since these miss important intraspecific variation; that phylogenetic trees be built to the level of the individual, requiring additional genetic information; and that perhaps fundamental changes be made to current coexistence theory. Possibly this would mean many more hours of fieldwork, more complex theory, and much more explanatory power is required. On the other hand, it could mean breakthroughs in how we understand longstanding ecological problems like ecosystem functioning, species diversity and coexistence, or trophic web structure.

For that reason, the fact that Bolnick et al. doesn’t demonstrate very clearly the gains or breakthroughs that could result from including intraspecific differences is a bit of a disappointment. Will we find that increasingly smaller amounts of variation are explained as we divide our units increasingly smaller? Or is the key to explaining community-level interactions found at the individual scale? Most of the examples in this paper are too simplistic to be useful, and for understandable reasons of space, there is little review of the literature (though they cite a number of important papers). That’s really too bad, since there are some subfields that have focused on intraspecific differences (for example, the ecosystem functioning literature), and their findings would contribute to the question of what makes intraspecific differences so promising for community ecologists. Despite that, when the mechanisms presented in Bolnick et al. are considered in combination with papers such as Crutsinger et al. 2006, Clark et al 2010, Albert et al. 2011, and Schindler et al. 2010 (just as a few examples), there is some tantalizing evidence suggesting that intraspecific variation can and does matter.

 *Although no doubt similar concerns about workload have accompanied any shift in approach throughout ecology's history. And certainly most shifts in ecological approach (spatial, phylogenetic, etc) only occur once the necessary methodological infrastructure was in place.


  1. Thanks for posting about this and for giving some of the historical background. I am just breaking into this literature and I am interested in how the individual perspective has grown and developed.

    It seems like accounting for individuals isn't too much of a logistical leap. To get species averages we measure many individuals anyways. This thrust is just accounting for the variation that we washed under the table before. In that sense, I see it as a question of whether or not we can learn about communities by ignoring intraspecific variation.

  2. Thanks Aaron, great comment. The history of ecology tells us a lot about why we are where we are today. I think some of the paradigms we follow in ecology have to do with priority effects and influential trends. However, logistical barriers are also important, and although it's true that sometimes we have and average out data from individuals anyways, it's also true that many ecologists rely on data that includes information only on what species were present in a community (and maybe in what abundances). Some subfields may find it logistically easier to incorporate individual variation than others.

  3. Thanks for the excellent post. This is definitely an useful approach but there's also an inherent field trade-off. Clark's work, which I really admire, has the advantage of all being based on an experimental forest he's been working in for almost 2 decades. He also has detailed paper in ecologial monographs ( . On top of the years he's spent studying the trees, he also has a study species that is easy to measure. Imagine trying to do this with birds. If you are studying bird communites, think about how much more work there is to measure traits (mist netting etc..) and the number of people needed vs just going into a forest with some binoculars. By studying species as opposed to individuals, we are putting a premium on sample size, spatial extent, etc because you can study a much larger area if you're just counting with binoculars. On top that, it becomes difficult to quantify what the true meaning of individual level variation means. I study ponds, so lets say instead of just counting mosquito larva, I want to study individual variation. I measure pigmentation for sake of example. Now I have my counts and lots of trait level measures, but how do I know what matters ecologially? Maybe pigmentation contributes to predator avoidance, or maybe it reduces mortality due to UV light, or maybe its just neutral variation. Now to make sense of that individual level variation I need multiple experiments per trait per species to understand what I'm measuring. Clark's work (with which I'm most familiar) is largely based on his hierarchical models, looking at the variance terms for random effects models. He can only speculate as to what that variation truly means. I loved Clark's work when it first came out and find it fascinating, but I think its important to consider how he was able to reach his conclusions and realize that it can only be applied to certain datasets. Really I think the great thing about this is that it brings home the importance of detailed long term datasets. As we enter an age more and more dominated by ecoinformatics and big data, small plots with detailed information can be just as useful as large open data sets (Like BCI on steroids).

  4. I wonder about an implicit all-else-equal assumption. If individual variants are taken to be the cause of ecological differences, then all else should be assumed to be equal. If, however, one variant is found in one ecosystem and another in a different, how to we tell it's not the differences in environmental factors favouring the different variants in different circumstances?

    In other words, if variant A prevails in habitat A and variant B in habitat B, maybe habitat A selects for variant A and vice verse.

  5. Joachim, I agree that the question of causation (e.g. "Is intraspecific variation caused by the environment or is intraspecific variation meaningful for coexistence.") is very important. It's no different from early studies of interspecific variation and coexistence, where the role of causation was sometimes ignored (a criticism of Diamond's assembly rules). Hopefully we've learned from those old criticisms and current research reflects that.