| |
Presented at the NABS Annual meeting, Vancouver, British Columbia, 2004
in Survey Design 2
Developing indicators for use in probabilistic survey designs in Great Lakes coastal ecosystems.
C. Richards1, N. Danz2, L.B. Johnson2, T.P. Hollenhorst2, V.J. Brady2, D.H. Brenneman2, J.J.H. Ciborowski3, and T.R. Hrabik4. 1Sea Grant College Program, University of Minnesota Duluth, Duluth, MN, USA, 55812, 2Natural Resources Research Institute, University of Minnesota Duluth, Duluth, MN, USA 55811, 3Department of Biological Sciences, University of Windsor, Windsor, ON, CANADA, N9B 3P4, 4Department of Biology, University of Minnesota Duluth, Duluth, MN, USA, 55812
Probabilistic surveys of ecosystem condition require indicators that both estimate ecological condition and suggest plausible causes of ecosystem degradation. To link potential indicators with stress, the sample should be distributed across stress gradients of interest at appropriate scales. We developed a sampling framework to relate stress to ecological response for fish, benthic invertebrates, and water quality in Great Lakes wetlands. We used a GIS to compile over 200 variables representing six types of anthropogenic stress: agriculture, atmospheric deposition, land cover, human populations, point source pollution, and shoreline modification. We divided the Great lakes coast into 762 units with geomorphic criteria and calculated stresses for each unit. Principal components analysis (PCA) was used to remove redundancy within the stress categories and cluster analysis was used with pc scores to create groups of coastal units having similar stress profiles. Random selection from the clusters provided an unbiased sample of sites that spanned a range of stresses . Water quality parameters corresponded well to PCA axes representing land use gradients. Several fish community metrics (species richness, % exotics) corresponded well to the amount and availability of sand. A land use indicator (% development) was strongly related to fish species diversity.
|
|
