Sequential decision plans, benthic macroinvertebrates, and biological monitoring programs

Abstract

A common obstacle to the inclusion of benthic macroinvertebrates in water quality monitoring programs is that numerous sample units must be examined in order to distinguish between impacted and unimpacted conditions, which can add significantly to the total cost of a monitoring program. Sequential decision plans can be used to reduce this cost because the number of sample units needed to classify a site as impacted or unimpacted is reduced by an average of 50%. A plan is created using definitions of unimpacted and impacted conditions, a description of the mathematical distribution of the data, and definitions of acceptable risks of type I and II errors. The applicability of using sequential decision plans and benthic macroinvertebrates in water quality monitoring programs is illustrated with several examples (e.g., identifying moderate and extreme changes in species richness in response to acid mine drainage; assessing the impact of a crude oil contamination on the density of two benthic populations; monitoring the effect of geothermal effluents on species diversity). These examples use data conforming to the negative binomial, Poisson, and normal distributions and define impact as changes in population density, species richness, or species diversity based on empirical data or the economic feasibility of the sequential decision plan. All mathematical formulae and intermediate values are provided for the step-by-step calculation of each sequential decision plan.