Detecting ecosystem level disturbances

Journal Title
Journal ISSN
Volume Title
Ecosystem ecologists often face the challenge of detecting perturbations and predicting their consequences. My dissertation research explores the importance of variance to the detection of perturbations in lake and stream ecosystems in three different contexts. First, I test the sensitivity of two commonly used assessment designs (the Before-After­Control-Impact-Paired Series (BACIPS) and the Beyond BACI) to detect a perturbation in a set of stream ecosystems given the empirical variance structure of the invertebrate metrics. Environmental assessments commonly use benthic invertebrates to assess water quality. Both assessment designs are poor at detecting changes with the post-impact disturbance effect size I calculated. However, in my simulation results, the BACIPS design outperformed the Beyond BACI. This is likely due to high temporal variance in the invertebrates. To consistently detect a disturbance, the designs require very low variance in the metrics, or a large effect size. Secondly, I examine the nutrient limitation of the headwaters of the McLeod River and the implications of that limitation for the future of the streams given potential development plans for the watershed. The headwaters are affected by past mining disturbances and there is ongoing interest in developing an open-pit coal mine in the area. My experiments show evidence of nitrogen limitation, which is important since coal mining activity adds nitrogen to watersheds. Thirdly, I assess the level of data resolution required to answer qualitative and quantitative questions about primary production in a set of lakes. Using an information­theoretic approach, I test the ability of 40 models containing variable levels of spatial and temporal complexity to 1) fit the relationship between light and primary productivity in a set of 14 British Columbia (Canada) lakes that were part of a large-scale fertilization experiment, and 2) predict annual primary production. The top-ranked model partitioned the data by lake, year, fertilization status, season and sampling station. When I predicted primary production estimates for my top three models and three commonly used models from the limnological literature, I found that model selection is important for quantitative production predictions, but not necessarily for qualitative assessment of nutrient limitation.
Bibliography: p. 158-188
Irvine, R. L. (2004). Detecting ecosystem level disturbances (Unpublished doctoral thesis). University of Calgary, Calgary, AB. doi:10.11575/PRISM/20444