Extrapolating Demography: Approach With Caution

We often want to measure how populations will respond to an environmental change (such as global warming), the introduction of an invasive species, or a management intervention (such as the release of a bio-control agent). However, demographic data is time consuming and expensive to collect. Many individuals have to be tagged and their fates recorded over time. We need at least three years of data, but for some species, like long lived trees and species that don’t reproduce very often, we need many more years of data. This means that it is not practical to measure every population of every species. Instead, we can only measure a few populations of a few species and then assume that the population dynamics are representative of those in larger regions and sets of species. But how good is this assumption?

decay_model_nofixed_noself

We used a global database of demographic models (specifically matrix models), called COMPADRE Plant Matrix Database, to directly test if population performance can be extrapolated, and if so, whether geographically close populations or those of closely related species, are better to extrapolate from. Matrix population models provide an ideal means to test how transferable population performance is across a wide suite of regions, life histories, and taxa within the plant kingdom. To date, matrix population models have been developed for over 1300 plant species across the globe. Matrix models summarise life histories ranging from simple to complex in a standard format. This allows the direct comparison of ecologically and biologically meaningful demographic metrics across populations, species and years.

We show that some important metrics of population performance, can be extrapolated. We could explain 25–40% of the variation in damping ratio, elasticities and temporal variation in population growth rate using these metrics from nearby populations, and to a much lesser extent, closely related species. The explanatory power of the geographic predictor term suggests that something about mid to small-scale environment is predictive of demography. However there a two important caveats. The first is that asymptotic population growth rate (the mostly commonly used metric of population performance) could not be extrapolated very well between populations. The second is that even with the largest geo-located dataset of demographic studies available we can only justify extrapolating important aspects of demography at limited scales (approx. >20 km), especially compared to the scales that threats such as species invasions and climate change occur at. Thus, demographic information should only be extrapolated with caution, and the initial assumption should be that any demographic results we obtain are applicable to the population they were measured for and those in the immediate environment. Capturing demography at scales relevant to landscape level threats will require more geographically extensive sampling.

For the full paper: SR Coutts, R Salguero‐Gómez, AM Csergő, Buckley YM (2016) Extrapolating demography with climate, proximity and phylogeny: approach with caution. Ecology Letters. doi: 10.1111/ele.12691

Click here for a non-paywalled version