Monte Carlo Simulations of Benthic Macroinvertebrate Populations: Estimates Using Random, Stratified, and Gradient Sampling

Abstract

Listen

Translate article

Accuracy, precision, and cost are major concerns in optimizing a sampling strategy to estimate population sizes of benthic macroinvertebrates. We used Monte Carlo simulations to compare precision and accuracy for various efforts (= cost) to predict density of cohorts and total populations of the water but Aphelocheirus aestivalis (Fabr.) from 3 German rivers. We compared density predictions of this hemipteran made from simultaneously collected physical variables to density predictions made from random sampling. The joint preference factor (JPF) method predicted Aphelocheirus densities from combined preference gradients for depth, velocity, and substratum size. It clearly yielded the most powerful predictions of all physical approaches tested. However, the JPF method sometimes predicted bug density better or worse than random sampling, depending on 1) the mean density of a cohort or total population, 2) the degree of spatial patchiness (contagion), 3) the affordable sample number to predict from, and 4) whether all or only dominant size components of the substratum were considered. We concluded that benthic macroinvertebrate sampling should be optimized using a flexible strategy, depending on the study objectives and the budget. This approach can save considerable time for the given error level required in predictions, and/or can yield a considerable reduction of the error of predictions (up to 50%) with a comparable effort. Optimization of sampling strategies in studies of lotic benthos is appropriate, logical, and timely, because large expenditures for stream management may be based on data provided by lotic ecologists.