Profiting from prior information in Bayesian analyses of ecological data

Abstract

SummaryMost ecological studies include prior information only implicitly, usually in their design or the discussion of results. In this study, two examples demonstrate that using Bayesian statistics to incorporate basic ecological principles and prior data can be very cost‐effective for increasing confidence in ecological research.The first example is based on examining the effects of an experimental manipulation of the habitat of mulgaraDasycercus cristicauda, a marsupial of inland Australia. The second example is based on observational mark–recapture data to estimate the annual survival of the European dipperCinclus cinclus, a passerine in France.In the mulgara example, the prior information obtained from an observational study increased confidence that there was an adverse effect of experimental habitat manipulation on the species. The results suggested that the capture rate of mulgara was reduced to approximately one‐quarter by reduction of vegetation cover.In the European dipper example, prior information based on the body mass of the species and estimates of annual survival of other European passerines was shown to be worth between 1 and 5 years of mark–recapture field data.Synthesis and applications. Body mass can be used to predict annual survival of European passerines and other animals. Results of observational studies can provide prior information in experimental studies of impacts of habitat change. By using Bayesian methods, such prior information, if represented in a coherent and logical way, can be cost‐effective for adding certainty to ecological studies.