Comments on the paper: on repeated parameter sampling in Monte Carlo simulations

Variability enters ecological systems at many different levels, from the differences between organisms and their individual life histories, to differences in the species composition of communities and changes in the external environment. Incorporating this variability into models, or accounting for it in analyses, is an increasingly important part of ecological modelling. One of the (many) ways variability or uncertainty enters models of ecological systems is in the assignment of values to physiological or interactive parameters. Introducing a technique known as stochastic parameters, Kremer [Ecol. Model. 18 (1983), 187–207] modelled parameters as time-variant, with their values being periodically chosen from prescribed distributions. Such an approach aims to model real-time variability in the ecological systems studied, but differs from traditional Monte Carlo simulations in which parameters are assumed time-invariant and stochasticity enters to model uncertainty in their measurement. Annan [Ecol. Model. 97 (1997) 111–115] examines the distinction between these two methods and proposes that suitably ‘corrected’ Monte Carlo simulations can replicate (and, therefore, supplant) those using stochastic parameters. In this paper, this proposal is examined and the fundamental assumptions which frame it are questioned. While Annanʹs simple analysis sheds light on one facet of Kremerʹs results, it does not justify his general conclusions.