Message in a bottle? Utility and limitations of recent ecological bottle experiments

Many classic tests of ecological theory have involved populations and communities maintained for many generations in the laboratory under tightly controlled conditions. In spite of this, such “bottle experiments” now play only a minor role within the larger field of ecology, and their relevance to natural populations and communities is regarded with suspicion by many field ecologists. Here, we compare and critique several recent bottle experiments, which were designed to test open questions in ecological theory that could never feasibly be addressed in natural communities. Judging from this set of experiments, we suspect that it will be difficult to relate the qualitative results of bottle experiments to natural populations and communities. What we learn from these experiments depends heavily on the relationship between theoretical models and experimental design. If the demography of organisms is completely under experimental control, bottle experiments can teach us about the possible range of population dynamics, but not about what regulates dynamics in natural populations. Furthermore, if experimental results are not linked to a mechanistic model, we can support or refute broad generalizations, but there is no direct way to relate bottle experiments to natural communities. Consequently, we argue that the most informative bottle experiments must incorporate both mechanistic models and unmanipulated demography; such bottle experiments can generate new ideas and future directions for both empirical and theoretical research.