A Direct‐Gradient Multivariate Index of Biotic Condition

Abstract

Multimetric indexes constructed by summing metric scores have been criticized despite many of their merits. A leading criticism is the potential for investigator bias involved in metric selection and scoring. Often there is a large number of competing metrics equally well correlated with environmental stressors, requiring a judgment call by the investigator to select the most suitable metrics to include in the index and how to score them. Data‐driven procedures for multimetric index formulation published during the last decade have reduced this limitation, yet apprehension remains. Multivariate approaches that select metrics with statistical algorithms may reduce the level of investigator bias and alleviate a weakness of multimetric indexes. We investigated the suitability of a direct‐gradient multivariate procedure to derive an index of biotic condition for fish assemblages in oxbow lakes in the Lower Mississippi Alluvial Valley. Although this multivariate procedure also requires that the investigator identify a set of suitable metrics potentially associated with a set of environmental stressors, it is different from multimetric procedures because it limits investigator judgment in selecting a subset of biotic metrics to include in the index and because it produces metric weights suitable for computation of index scores. The procedure, applied to a sample of 35 competing biotic metrics measured at 50 oxbow lakes distributed over a wide geographical region in the Lower Mississippi Alluvial Valley, selected 11 metrics that adequately indexed the biotic condition of five test lakes. Because the multivariate index includes only metrics that explain the maximum variability in the stressor variables rather than a balanced set of metrics chosen to reflect various fish assemblage attributes, it is fundamentally different from multimetric indexes of biotic integrity with advantages and disadvantages. As such, it provides an alternative to multimetric procedures.

Development of a Multimetric Index for Fish Assemblages in a Cold Tailwater in Tennessee

“…The index development procedure that we used adhered to general recommendations (e.g., Roset et al 2007;Stoddard et al 2008) but relied on a multivariate technique to select metrics from a large list of candidates. Benefits of using a multivariate approach such as the one employed here include objectivity, repeatability, and an explicit link between environmental stressors and biotic responses (Miranda et al 2012). Perhaps the most important benefit of using a multivariate procedure, especially for a radically altered system, is the ability to develop a statistically Table 5 for a full description of the metrics (y-axis labels).…”

“…Multivariate statistics have been applied as an exploratory tool for comparing trends in environmental disturbances to macroinvertebrate assemblages (Gerritsen 1995;Fore et al 1996;Reynoldson et al 1997) and fish assemblages (Angradi et al 2009;Flinders et al 2009;Doll 2011). Recently, multivariate procedures have also been used to directly identify sets of suitable metrics for multimetric indices (Hallett et al 2012;Miranda et al 2012). Multimetric indices that lack a benchmark reference condition can be useful for describing changes in assemblages, but they are not highly transferable to other systems and should not be used to define biotic integrity Jennings et al 1995;Hawkins et al 2010).…”

Developing conceptual models for assessing benefits and impacts of USACE activities on freshwater mussel communities

“…2014). Additionally, multivariate approaches can reduce investigator bias during the selection of important environmental parameters (Miranda, Aycock, and Killgore 2012). USACE-Chicago District has an ongoing effort to develop a fish (ichthyofaunal) qualitative index for the Chicago Region that is based on a priori best professional scoring of each fish species according to their level of tolerance of anthropogenic activities coupled with their habitat specificity.…”