Statistical reconstruction of quantitative food webs using tracer addition experiments

Background/Question/Methods

Understanding how nutrients flow through food webs is central in ecosystem ecology. Tracer addition experiments are a powerful tool to reconstruct nutrient flows among ecosystem compartments. In essence, these experiments add an isotopically marked nutrient into an ecosystem and track its fate across compartments through time. As this type of data becomes increasingly available, there is a push for studies comparing the results of isotope tracer experiments across ecological conditions and ecosystems. Yet to date there was no formal statistical framework available for the analyses of such experiments and, therefore, their comparison. Such framework needs to be able to provide flux estimates and associated errors that account for (1) the interdependence of parameter estimates among all compartments, (2) the uncertainty in the diet composition of omnivores, and (3) the over-enrichment of compartments that feed selectively.

Results/Conclusions

Here we develop such a framework using Bayesian Hidden Markov Models and apply it to the analysis of 15N tracer addition experiments in montane streams of the Caribbean island of Trinidad. Through this case study, we illustrate how to (1) estimate uptake, fluxes, and turnover rates and associated uncertainty, (2) compare alternative models of food web structure, (3) propagate the error when estimating derived parameters, and (4) statistically compare parameter estimates from different streams. We also present its implementation as R package `isotracer`. Our method not only makes possible rigorous statistical estimation of nutrient fluxes and ecosystem properties but also represents a valuable tool for the design of future experiments.