Using laboratory populations to predict responses in nature: a case study using Anuran larvae with recommendations for improving estimates of effect sizes
University of New Brunswick
Experiments are carried out in laboratories, mesocosms, in situ field enclosures, and in whole ecosystems (field studies), but the intent is almost always to make inferences to natural systems. Limited environmental control and difficulty achieving appropriate replication in whole-ecosystem experiments make laboratory studies appealing alternatives, and studies in artificial venues have provided some of the most influential contributions to ecology in the past century. However, experiments in artificial venues have been questioned for their ability to predict responses of populations in nature. These discussions of the relevance of small-scale experiments have, almost exclusively, focused on how the reduction in complexity of venues at smaller spatial scales might limit extrapolation to more complex natural systems. There has been no attention given to whether animals reared in controlled (e.g. laboratory) conditions are actually representative of populations in nature. My research involves meta-analysis of tadpole population demographics across experimental venues, as well as laboratory and field experiments comparing wood frog (Lithobates sylvaticus) survival, growth and development, and size variation in response to intra-specific competitive interactions. Results demonstrate that mortality is consistently higher in natural than artificial populations of tadpoles, and that this results in a homogenizing effect on the population. This is most obviously explained by a reduction in the force of natural competitive interactions, but we also demonstrate that smaller individuals are more actively removed from natural populations. This has implications for studies of basic competitive interactions, and also influences estimates of toxicological experiments.