Enhancing the Performance and Interpretation of Freshwater Biological Indices: An Application in Arid Zone Streams

Document Type

Article

Journal/Book Title/Conference

Ecological Indicators

Volume

36

Publication Date

1-1-2014

Keywords

Arid environments, bioassessment, biological indices, community assembly, environmental monitoring, habitat isolation, meta-communities, models, incomplete sampling, evenness, stream ecosystems

First Page

470

Last Page

482

Abstract

We assessed the performance of biological indices developed for invertebrate assemblages occurring in arid zone streams: a multimetric index (MMI) and an O/E index of taxonomic completeness. Our overall goal was to advance our understanding of the factors that affect performance and interpretation of biological indices. Our specific objectives were to (1) develop biological indices that are insensitive to natural environmental gradients, (2) develop a general method to determine if the biological potential of an assessed site is adequately represented by the population of reference sites, (3) develop a robust method to select metrics for inclusion in MMIs that ensures maximum independence of metrics, and (4) determine if a fundamental sample property (the evenness of taxa counts within a sample) affects index performance. Random Forest modeling revealed that both individual metrics and taxa composition were strongly associated with natural environmental heterogeneity, which meant both the MMI and O/E index needed to be based on site-specific expectations. We produced a precise, responsive, and ecologically robust MMI by using principal components analysis to identify 7 statistically independent metrics from a list of 31 candidate assemblage-level metrics. However, the O/E index we developed was relatively imprecise compared with O/E indices developed for other regions. This imprecision may be the consequence of low predictability in local taxa composition associated with the relatively high spatial isolation of aquatic habitats within arid regions. We were also able to assess the likelihood that the biological potential of assessed sites were adequately characterized by the population of reference sites by developing and applying a multivariate, nearest-neighbor test that determined if an assessed site occurred within the environmental space of the reference site network. This approach is robust and applicable to all biological indices. We also demonstrate that the evenness of taxa counts within a sample is positively related to estimates of sample taxa richness and thus the scores of both indices. The relationship between richness and sample evenness can potentially compromise inferences regarding biological condition, and post hoc adjustments for the effects of evenness on index scores might be desirable. Further improvements in the performance and interpretation of biological indices will require simultaneous consideration of the effects of incomplete sampling on characterization of biological assemblages and the physical and biological factors that influence community assembly.

Link to Full Text

Share

COinS