 |
Monday, May 26, 2008 - 5:00 PM
99
Using SIMPACTED data to establish power and error in a RCA assessment method
Trefor B. Reynoldson, Acadia University, Wolfville, NS B2R 4P6, Canada and J. L. Bailey, Department of Biology, The University of Western Ontario, London, ON N6A 5B7, Canada.
An important but ignored issue surrounding RCA methods is the assessment of power and error. Benthic macroinvertebrate and habitat data were collected for 158 reference stream sites and 84 test stream sites throughout the Yukon River basin. From the reference site data an RCA bioassessment model was developed using classification and discriminant function methods. To determine the power and error of test site assessment reference site benthic invertebrate data were subjected to three levels of disturbance using simulated impacts and these "simpacted" sites were assessed with BEAST assessment methods. We determined Type I (number reference sites outside an ellipse) and Type 2 error (number simpacted sites inside an ellipse) at three probability levels. Results showed that the nature of the reference assemblage affected both error rates, for example among four reference communities Type 2 error rates using a 90% probability ellipse ranged from 6 – 85% at the intermediate disturbance level. This shows that is critical to both match sites to appropriate reference communities and that knowledge of the power of the assessment be understood and may be an inherent attribute of the system.
Web Page: RCA, error, power
See more of Bioassessment - Statistical Issues
See more of Contributed Sessions
See more of The Nabs 56th Annual Meeting (25-28 May, 2008)
See more of Contributed Sessions
See more of The Nabs 56th Annual Meeting (25-28 May, 2008)