Species’ phenotypic traits are widely thought to (and often do) predict their abundances (e.g. Harpole & Tilman 2006). The most abundant species at a given site should be those with traits that make them well-adapted to the biotic and abiotic conditions at that site. Species with traits poorly matched to the local biotic and abiotic environment should be rare or absent. Much research takes this argument for granted, and focuses on issues such as identifying the right traits (or surrogates for the right traits).
I too used to take this argument for granted, but I don’t anymore, thanks to a conversation I had a few years ago with Graham Bell. We were discussing neutral theory in community ecology, and I argued to Graham that trait-abundance correlations disproved neutral theory. In a neutral world, there’s no reason to expect species with particular traits to be predictably more or less abundant.
Graham replied with a good question: If some species are better adapted than others to local biotic and abiotic conditions, how can you have multiple species coexisting at all? Why doesn’t the best-adapted species just exclude all the others?
Like a good community ecologist, I replied, because there are also niche differences that generate negative frequency dependence, thereby stabilizing coexistence.
To which Graham responded with another good question: But if species’ per-capita growth rates are frequency dependent, why should we expect their traits to predict their abundances? After all, frequency dependence means that a species’ per-capita growth rate depends on the relative abundances, and thus on the traits, of all the other species, not just on its own traits.
I didn’t have a good answer to that second question, and I still don’t. I’m sure there are conditions under which phenotypic traits predict the abundances of stably coexisting species, but I don’t know what those conditions are. This seems important to know, in order to properly interpret the many field studies of trait-abundance (and trait-environment) correlations. I’m going to play around with some simple models to see if I can get a better handle on this.