Detection of human-induced stress in streams

Running water habitats are among the most precious, yet most threatened, ecosystems on earth. Hence, there is a need for reliable methods for detecting the effects of pollution on this valuable commodity. This thesis examined a number of different approaches commonly used in bioassessment of stream ecosystems for their ability to detect ecological change. In particular, focus was on assessing the utility of single metric, multimetric and multivariate approaches using macroinvertebrate communities, with organic pollution and acidity as stress gradients. For detecting the effects of organic pollution, two metrics (the DJ index and the CA scores) representing two different approaches (multimetric and the multivariate) were found to be reliable tools for detecting the effects of stress in streams of southern Sweden. These methods were sensitive (high coefficients of variation) and had high precision (low error) to the stress gradient. Notwithstanding, single metric approaches might also be used when the knowledge of multimetric or multivariate methods is not available. The saprobic index was one such metric that showed promise as a tool for detecting human-induced change in streams. However, many of the single metrics tested had low precision (high error) and hence should be used with caution. Regarding the effects of acid stress, one index (out of ten) was highly correlated between spring and autumn, and was thus deemed to be a reliable metric for use with autumn samples to assess even spring conditions. This multimetric index was developed for use in southern Sweden and is included in Swedish EPAs Environmental Quality Criteria. A ‘new’ weighted averaging method was examined and found to be useful for assessing the acidity of streams. However, more research is needed to develop the method as well as a user-friendly platform. Both multimetric and multivariate (the weighted averaging model) methods were shown to be reliable tools for assessing the effects of acid stress on stream macroinvertebrate assemblages. Moreover, the results of this thesis support the use of multimetric and multivariate approaches in bioassessment. If possible, a combination of these methods is recommended in assessing the effects of ecological change. Single metric indices could be used, however a deeper ecological understanding is needed to evaluate the inherent errors associated with these metrics. For the future, user-friendlier methods should be developed so that multimetric and multivariate approaches can be more widely used by water managers