The Journal of Ecology just published our four-year study of community dynamics of alpine plants. The study is part of a multi-journal special feature on linking traits and demography.

Predicting community assembly is one of the key goals of modern ecology – it is necessary for theory-driven land management, climate change forecasting, policy-making for invasive species, and so on. But the goal has been very elusive, because assembly often seems to be unpredictable or random. This study asks whether community assembly can be better predicted. It starts from the premise that assembly is the outcome of demographic processes, and these processes are in turn the outcome of interactions between organisms’ phenotypes (functional traits) and their environment. The hypothesis is that if these traits and environment were known accurately enough, accurate community dynamics could be obtained.

To get at this question, I set up a monitoring study in the Gunnison National Forest in Colorado – an alpine screefield system that is under snow for the majority of the year.

For the past five years, we have been monitoring the demography of this community – every individual plant (including seedlings) is mapped, and annually has all major fitness components measured – growth, survival, recruitment, fecundity. There are almost twenty species and about a thousand individuals per year on average, so it is not a small undertaking.

In parallel, we map microenvironment (belowground and aboveground) at sub-meter spatial scales. Below you can see some of the different soils that occur across the site. We also measure above- and below-ground trait profiles for individuals of each species of varying sizes.

The question then is whether this information is sufficient to predict the vital rates of the species in the community. Using a series of mixed models, we showed that each vital rate is predictable, with explained variation between 50-100%.

Interestingly, almost all of the predictive power comes not from the main effects of each variable (e.g., that taller plants grow faster) but rather from interactions between variables (e.g., that taller plants reproduce more, but only when surrounded by neighbors and in more finely textured soils). This suggests that predictability is achievable, but only with highly detailed knowledge of traits and microenvironment together.

These analyses provide some cautious optimism for more predictive community ecology – and also open up a range of further questions for the role of species interactions, the role of external climate change, and so on. We have lots more work to do – especially around size-structured modeling of population growth rates – and long-term monitoring efforts to continue. For now, the study hopefully shows that some progress at the interface of functional ecology and demography is possible.

You can download a PDF copy, read at the journal, or download the complete demography/trait/environment dataset (2014-2017) at Dryad.

The 2018 census was just finished a few days ago. This year we saw very high mortality of some species (like this lupine) due to two years of drought and low snowpack. As we continue data collection over more years, the value of the data for community ecology theory – and for monitoring of climate change impacts – will continue to grow.