You are here:
THE EVALUATION OF METHODS FOR CREATING DEFENSIBLE, REPEATABLE, OBJECTIVE AND ACCURATE TOLERANCE VALUES
Citation:
BLOCKSOM, K. A. AND L. WINTERS. THE EVALUATION OF METHODS FOR CREATING DEFENSIBLE, REPEATABLE, OBJECTIVE AND ACCURATE TOLERANCE VALUES. U.S. Environmental Protection Agency, Washington, DC, EPA/600/R-06/045 (NTIS PB2007-102023), 2006.
Impact/Purpose:
The goal of this research is to develop methods and indicators that are useful for evaluating the condition of aquatic communities, for assessing the restoration of aquatic communities in response to mitigation and best management practices, and for determining the exposure of aquatic communities to different classes of stressors (i.e., pesticides, sedimentation, habitat alteration).
Description:
In the field of bioassessment, tolerance has traditionally referred to the degree to which organisms can withstand environmental degradation. This concept has been around for many years and its use is widespread. In numerous cases, tolerance values (TVs) have been assigned to individual taxa or groups of taxa to represent their tolerance to pollution. The TVs are then often combined into metrics which describe characteristics of aquatic communities. Perhaps the most familiar example is the Hilsenhoff Biotic Index (HBI) (Hilsenhoff, 1977), an index that has been incorporated into many bioassessment programs. The HBI is typically very useful in distinguishing among sites of higher and lower water quality. To calculate the HBI, each environmental agency or organization typically uses its own set of tolerance values. However, the origins of these values, and rationales for their selection, are often obscure and unverifiable. Available methods for deriving TVs more objectively vary substantially in approach and complexity. Therefore, this study conducts systematic comparisons of existing lists of macroinvertebrate TVs and their resulting HBI scores. It also compares several objective TV derivation approaches, as well as bioassessment metrics derived from each, to determine their repeatability and sensitivity to disturbance.
Several existing lists of macroinvertebrate TVs were assembled into a single database for the purpose of direct comparison. The values for a given taxon varied widely from list to list, but this variation was somewhat muted when TVs were incorporated into HBI scores, particularly at the genus level. The process of developing TVs by a particular organization seems to vary as well and often depends on professional judgment. Although such processes may provide effective tools for bioassessment, they are not repeatable. To compare objective TV development approaches, data from a single stream study in the mid-Atlantic highlands were divided randomly into calibration and validation sets, and each approach was carried out on both data sets. All but one approach consisted of first defining a disturbance gradient and then using one of two procedures to calculate the TV. One approach used EPT richness to describe the disturbance level, another used a principal components analysis (PCA), and a third used PCA with generalized additive modeling (GAM) to more precisely model the relationship between the probability of taxon presence and disturbance. A fourth approach relied on the observed frequency of a taxon compared to its expected frequency using predictive modeling. For the EPT, PCA, and GAM approaches, two procedures each for calculating TVs were examined, one based on a single value from the gradient defined and the other a weighted average. The TVs generated were compared between the calibration and validation data sets, as were three tolerance-based macroinvertebrate metrics. In addition, the ability of these metrics to distinguish between reference and impaired sites was evaluated for each method.
All of the approaches evaluated exhibited some degree of repeatability but varied with respect to the type of data and the degree of statistical experience or training required. The EPT approach was the least defensible approach because of its circular nature. Weighted procedures for calculating TVs from the defined disturbance gradient resulted in higher repeatability. Among the metrics evaluated, intolerant taxa richness was most useful overall because it distinguished most strongly between reference and impaired and was repeatable across most approaches.