Thursday, June 4, 2015
Barneche and Allen 2015, Embracing general theory and taxon-level idiosyncrasies to explain nutrient recycling
This cake is to celebrate a paper by Barneche and Allen, entitled "Embracing general theory and taxon-level idiosyncrasies to explain nutrient recycling". It is being presented by Diego Barneche.
The value of mathematical theory in ecology is controversial; most ecologists are found in either side of a dichotomy: while some argue that general theory is the optimal way to advance mechanistic understanding in the discipline, others claim that general simple theory does not capture most of the variation in living systems. In our recent commentary piece (Barneche & Allen 2015), we argue that ecologists should embrace both views, and that this duality can be reconciled by using a combination of general mathematical theory and advanced statistical techniques, following a sequence of steps described below. These steps are exemplified using a recent study that uses a combination of predictions from the metabolic theory (MT) and ecological stoichiometry (ES) to explain body mass scaling of nutrient recycling rates in marine animals (Allgeier et al. 2015).
The first step entails recognising the scope of a given theory and explicitly declaring what assumptions are necessary to reach a given set of predictions. Allgeier et al. (2015) show that nutrient recycling rates follow 3/4-power body mass scaling, as should be predicted from MT-ES. Our study provides the theoretical rationale and the assumptions necessary to yield such prediction. In our view, this framework is essential to the understanding of ecological processes because false predictions serve as indications that one or more assumptions are violated/wrong – thus providing avenues on how to move forward on the understanding of ecological processes, either by modifying existing theory or by developing a new one. Although this assumption-prediction-testing-falsify cycle is the very core of the scientific method, it has been largely neglected in ecology.
The second step involves using statistical methods such as mixed effects models, which allow for the estimation of overall (i.e. mean) trends while accounting for deviations attributable to other variables not included in the theory. Many recent studies have made use of these mixed models to test for general theory while accounting for differences attributable to taxonomy (i.e. random effects) – there is to recognise that taxa might deviate from each other and from the overall trend. While taxonomy is not a true predictor in the sense that it characterises a process, it provides clues to what traits are responsible for differences among taxa, thus providing ideas on to how we should move forward.
The third and final step requires a careful exploration of how much variation is explained by different predictors, and what are the magnitudes of each one of them. For example, Allgeier et al. (2015) provides compelling evidence that despite substantial taxon-level idiosyncrasies (characterised as random effects in their LMM), body size is still the strongest predictor of nutrient recycling rates in marine animals. In our commentary, we notice that body size spans many orders of magnitude, while the other fixed-effect predictor (nutrient body content) spans less than one order of magnitude. However, the effect of body nutrient is stronger than that of body size, indicating that, for example, a doubling in body nutrient content has a greater effect in nutrient recycling rates than an equivalent increase in body mass.
We hope that with this roadmap, ecologists will increasingly embrace general theory and ecological idiosyncrasies in one single framework, which should help advance our understanding of ecological patterns and processes.