Detecting density dependence

A number of papers have appeared in the literature recently in which the performance of tests of density dependence in population time series data were compared. We believe that the search for an 'optimal' test for density dependence is unproductive and the practice of assessing competing tests via computer simulation flawed. It is unlikely that a single test will have uniformly superior power characteristics under all conditions. The difficulty with simulation studies is the lack of communality between them. We have observed differences in simulation results that have as much to do with differences in model parameterization, simulation run length, characterisation of the stochastic component, and methods of analysis of the results as they do with any inherent difference in the performance of the tests for density dependence being compared. We advocate the use of Bulmer's first test as a de facio standard because it makes minimal assumptions about the response-generating mechanism, has reasonable power characteristics, and is easy to implement. It is a test of serial correlation in the log abundance data and consequently is affected by the presence of autocorrelated errors. However, this should not be interpreted as a Failing of the test, but rather a caveat to its use.